* [PATCH v3 00/37] vfio: Adopt iommufd
@ 2023-10-26 10:30 Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
` (37 more replies)
0 siblings, 38 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
Hi,
Thanks all for giving guides and comments on previous series, here is
the v3 of pure iommufd support part.
Based on Cédric's suggestion, this series includes an effort to remove
spapr code from container.c, now all spapr functions are moved to spapr.c
or spapr_pci_vfio.c, but there are still a few trival check on
VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce many
callbacks and duplicate code just to remove them. Some functions are moved
to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
spapr_pci_vfio.c is arch specific, or else we need to introduce stub
functions to those spapr functions moved.
PATCH 1-5: Move spapr functions to spapr*.c
PATCH 6-18: Abstract out base container
PATCH 19-22: Introduce sparpr container and its specific interface
PATCH 23: Add --enable/--disable-iommufd config support
PATCH 24: Introduce iommufd object
PATCH 25-29: add IOMMUFD container and cdev support
PATCH 30-37: fd passing for IOMMUFD object and cdev
We have done wide test with different combinations, e.g:
- PCI device were tested
- FD passing and hot reset with some trick.
- device hotplug test with legacy and iommufd backends
- with or without vIOMMU for legacy and iommufd backends
- divices linked to different iommufds
- VFIO migration with a E800 net card(no dirty sync support) passthrough
- platform, ccw and ap were only compile-tested due to environment limit
Given some iommufd kernel limitations, the iommufd backend is
not yet fully on par with the legacy backend w.r.t. features like:
- p2p mappings (you will see related error traces)
- dirty page sync
- and etc.
qemu code: https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
Based on vfio-next, commit id: fd0e1c8bc1
--------------------------------------------------------------------------
Below are some background and graph about the design:
With the introduction of iommufd, the Linux kernel provides a generic
interface for userspace drivers to propagate their DMA mappings to kernel
for assigned devices. This series does the porting of the VFIO devices
onto the /dev/iommu uapi and let it coexist with the legacy implementation.
At QEMU level, interactions with the /dev/iommu are abstracted by a new
iommufd object (compiled in with the CONFIG_IOMMUFD option).
Any QEMU device (e.g. vfio device) wishing to use /dev/iommu must be
linked with an iommufd object. In this series, the vfio-pci device is
granted with such capability (other VFIO devices are not yet ready):
It gets a new optional parameter named iommufd which allows to pass
an iommufd object:
-object iommufd,id=iommufd0
-device vfio-pci,host=0000:02:00.0,iommufd=iommufd0
Note the /dev/iommu and vfio cdev can be externally opened by a
management layer. In such a case the fd is passed:
-object iommufd,id=iommufd0,fd=22
-device vfio-pci,iommufd=iommufd0,fd=23
If the fd parameter is not passed, the fd is opened by QEMU.
See https://www.mail-archive.com/qemu-devel@nongnu.org/msg937155.html
for detailed discuss on this requirement.
If no iommufd option is passed to the vfio-pci device, iommufd is not
used and the end-user gets the behavior based on the legacy vfio iommu
interfaces:
-device vfio-pci,host=0000:02:00.0
While the legacy kernel interface is group-centric, the new iommufd
interface is device-centric, relying on device fd and iommufd.
To support both interfaces in the QEMU VFIO device we reworked the vfio
container abstraction so that the generic VFIO code can use either
backend.
The VFIOContainer object becomes a base object derived into
a) the legacy VFIO container and
b) the new iommufd based container.
The base object implements generic code such as code related to
memory_listener and address space management whereas the derived
objects implement callbacks specific to either BE, legacy and
iommufd. Indeed each backend has its own way to setup secure context
and dma management interface. The below diagram shows how it looks
like with both BEs.
VFIO AddressSpace/Memory
+-------+ +----------+ +-----+ +-----+
| pci | | platform | | ap | | ccw |
+---+---+ +----+-----+ +--+--+ +--+--+ +----------------------+
| | | | | AddressSpace |
| | | | +------------+---------+
+---V-----------V-----------V--------V----+ /
| VFIOAddressSpace | <------------+
| | | MemoryListener
| VFIOContainer list |
+-------+----------------------------+----+
| |
| |
+-------V------+ +--------V----------+
| iommufd | | vfio legacy |
| container | | container |
+-------+------+ +--------+----------+
| |
| /dev/iommu | /dev/vfio/vfio
| /dev/vfio/devices/vfioX | /dev/vfio/$group_id
Userspace | |
============+============================+===========================
Kernel | device fd |
+---------------+ | group/container fd
| (BIND_IOMMUFD | | (SET_CONTAINER/SET_IOMMU)
| ATTACH_IOAS) | | device fd
| | |
| +-------V------------V-----------------+
iommufd | | vfio |
(map/unmap | +---------+--------------------+-------+
ioas_copy) | | | map/unmap
| | |
+------V------+ +-----V------+ +------V--------+
| iommfd core | | device | | vfio iommu |
+-------------+ +------------+ +---------------+
[Secure Context setup]
- iommufd BE: uses device fd and iommufd to setup secure context
(bind_iommufd, attach_ioas)
- vfio legacy BE: uses group fd and container fd to setup secure context
(set_container, set_iommu)
[Device access]
- iommufd BE: device fd is opened through /dev/vfio/devices/vfioX
- vfio legacy BE: device fd is retrieved from group fd ioctl
[DMA Mapping flow]
1. VFIOAddressSpace receives MemoryRegion add/del via MemoryListener
2. VFIO populates DMA map/unmap via the container BEs
*) iommufd BE: uses iommufd
*) vfio legacy BE: uses container fd
Changelog:
v3:
- Rename base container as VFIOContainerBase and legacy container as container (Cédric)
- Drop VFIO_IOMMU_BACKEND_OPS class and use struct instead (Cédric)
- Cleanup container.c by introducing spapr backend and move spapr code out (Cédric)
- Introduce vfio_iommu_spapr_ops (Cédric)
- Add doc of iommufd in qom.json and have iommufd member sorted (Markus)
- patch19 and patch21 are splitted to two parts to facilitate review
v2:
- patch "vfio: Add base container" in v1 is split into patch1-15 per Cédric
- add fd passing to platform/ap/ccw vfio device
- add (uintptr_t) cast in iommufd_backend_map_dma() per Cédric
- rename char_dev.h to chardev_open.h for same naming scheme per Daniel
- add full copyright per Daniel and Jason
Note changelog below are from full IOMMUFD series:
v1:
- Alloc hwpt instead of using auto hwpt
- elaborate iommufd code per Nicolin
- consolidate two patches and drop as.c
- typo error fix and function rename
rfcv4:
- rebase on top of v8.0.3
- Add one patch from Yi which is about vfio device add in kvm
- Remove IOAS_COPY optimization and focus on functions in this patchset
- Fix wrong name issue reported and fix suggested by Matthew
- Fix compilation issue reported and fix sugggsted by Nicolin
- Use query_dirty_bitmap callback to replace get_dirty_bitmap for better
granularity
- Add dev_iter_next() callback to avoid adding so many callback
at container scope, add VFIODevice.hwpt to support that
- Restore all functions back to common from container whenever possible,
mainly migration and reset related functions
- Add --enable/disable-iommufd config option, enabled by default in linux
- Remove VFIODevice.hwpt_next as it's redundant with VFIODevice.next
- Adapt new VFIO_DEVICE_PCI_HOT_RESET uAPI for IOMMUFD backed device
- vfio_kvm_device_add/del_group call vfio_kvm_device_add/del_fd to remove
redundant code
- Add FD passing support for vfio device backed by IOMMUFD
- Fix hot unplug resource leak issue in vfio_legacy_detach_device()
- Fix FD leak in vfio_get_devicefd()
rfcv3:
- rebase on top of v7.2.0
- Fix the compilation with CONFIG_IOMMUFD unset by using true classes for
VFIO backends
- Fix use after free in error path, reported by Alister
- Split common.c in several steps to ease the review
rfcv2:
- remove the first three patches of rfcv1
- add open cdev helper suggested by Jason
- remove the QOMification of the VFIOContainer and simply use standard ops
(David)
- add "-object iommufd" suggested by Alex
Thanks
Zhenzhong
Eric Auger (11):
vfio/container: Switch to dma_map|unmap API
vfio/common: Move giommu_list in base container
vfio/container: Move space field to base container
vfio/container: Switch to IOMMU BE
set_dirty_page_tracking/query_dirty_bitmap API
vfio/container: Convert functions to base container
vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base
container
vfio/container: Move listener to base container
vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size to base
container
vfio/container: Implement attach/detach_device
backends/iommufd: Introduce the iommufd object
vfio/pci: Allow the selection of a given iommu backend
Yi Liu (2):
util/char_dev: Add open_cdev()
vfio/iommufd: Implement the iommufd backend
Zhenzhong Duan (24):
vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c
vfio/container: Move vfio_container_add/del_section_window into
spapr.c
vfio/container: Move spapr specific init/deinit into spapr.c
vfio/spapr: Make vfio_spapr_create/remove_window static
vfio/common: Move vfio_host_win_add/del into spapr.c
vfio: Introduce base object for VFIOContainer and targetted interface
vfio/container: Introduce a empty VFIOIOMMUOps
vfio/container: Move per container device list in base container
vfio/container: Move iova_ranges to base container
vfio/spapr: Introduce spapr backend and target interface
vfio/spapr: switch to spapr IOMMU BE add/del_section_window
vfio/spapr: Move prereg_listener into spapr container
vfio/spapr: Move hostwin_list into spapr container
Add iommufd configure option
vfio/iommufd: Switch to manual hwpt allocation
vfio/iommufd: Add support for iova_ranges
vfio/iommufd: Bypass EEH if iommufd backend
vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info
vfio/pci: Adapt vfio pci hot reset support with iommufd BE
vfio/pci: Make vfio cdev pre-openable by passing a file handle
vfio: Allow the selection of a given iommu backend for platform ap and
ccw
vfio/platform: Make vfio cdev pre-openable by passing a file handle
vfio/ap: Make vfio cdev pre-openable by passing a file handle
vfio/ccw: Make vfio cdev pre-openable by passing a file handle
MAINTAINERS | 13 +
meson.build | 6 +
qapi/qom.json | 20 +
include/hw/vfio/vfio-common.h | 120 ++----
include/hw/vfio/vfio-container-base.h | 126 ++++++
include/hw/vfio/vfio-platform.h | 1 +
include/hw/vfio/vfio.h | 7 -
include/qemu/chardev_open.h | 16 +
include/sysemu/iommufd.h | 46 +++
backends/iommufd-stub.c | 59 +++
backends/iommufd.c | 268 ++++++++++++
hw/ppc/spapr_pci_vfio.c | 100 ++++-
hw/vfio/ap.c | 38 +-
hw/vfio/ccw.c | 40 +-
hw/vfio/common.c | 323 +++++++--------
hw/vfio/container-base.c | 131 ++++++
hw/vfio/container.c | 434 ++++----------------
hw/vfio/helpers.c | 34 +-
hw/vfio/iommufd.c | 562 ++++++++++++++++++++++++++
hw/vfio/pci.c | 253 ++++++++++--
hw/vfio/platform.c | 45 ++-
hw/vfio/spapr.c | 293 ++++++++++++--
util/chardev_open.c | 81 ++++
backends/Kconfig | 4 +
backends/meson.build | 5 +
backends/trace-events | 12 +
hw/vfio/meson.build | 4 +
hw/vfio/trace-events | 17 +-
meson_options.txt | 2 +
qemu-options.hx | 13 +
scripts/meson-buildoptions.sh | 3 +
util/meson.build | 1 +
32 files changed, 2400 insertions(+), 677 deletions(-)
create mode 100644 include/hw/vfio/vfio-container-base.h
delete mode 100644 include/hw/vfio/vfio.h
create mode 100644 include/qemu/chardev_open.h
create mode 100644 include/sysemu/iommufd.h
create mode 100644 backends/iommufd-stub.c
create mode 100644 backends/iommufd.c
create mode 100644 hw/vfio/container-base.c
create mode 100644 hw/vfio/iommufd.c
create mode 100644 util/chardev_open.c
--
2.34.1
^ permalink raw reply [flat|nested] 95+ messages in thread
* [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 14:23 ` Eric Farman
2023-10-27 9:19 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c Zhenzhong Duan
` (36 subsequent siblings)
37 siblings, 2 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
Thomas Huth, Tony Krowiak, Halil Pasic, Jason Herne, Eric Farman,
Matthew Rosato, open list:sPAPR (pseries),
open list:S390 general arch...
With vfio_eeh_as_ok/vfio_eeh_as_op moved and made static,
vfio.h becomes empty and is deleted.
No functional changes intended.
Suggested-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio.h | 7 ---
hw/ppc/spapr_pci_vfio.c | 100 +++++++++++++++++++++++++++++++++++++++-
hw/vfio/ap.c | 1 -
hw/vfio/ccw.c | 1 -
hw/vfio/common.c | 1 -
hw/vfio/container.c | 98 ---------------------------------------
hw/vfio/helpers.c | 1 -
7 files changed, 99 insertions(+), 110 deletions(-)
delete mode 100644 include/hw/vfio/vfio.h
diff --git a/include/hw/vfio/vfio.h b/include/hw/vfio/vfio.h
deleted file mode 100644
index 86248f5436..0000000000
--- a/include/hw/vfio/vfio.h
+++ /dev/null
@@ -1,7 +0,0 @@
-#ifndef HW_VFIO_H
-#define HW_VFIO_H
-
-bool vfio_eeh_as_ok(AddressSpace *as);
-int vfio_eeh_as_op(AddressSpace *as, uint32_t op);
-
-#endif
diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
index 9016720547..f283f7e38d 100644
--- a/hw/ppc/spapr_pci_vfio.c
+++ b/hw/ppc/spapr_pci_vfio.c
@@ -18,14 +18,112 @@
*/
#include "qemu/osdep.h"
+#include <sys/ioctl.h>
#include <linux/vfio.h>
#include "hw/ppc/spapr.h"
#include "hw/pci-host/spapr.h"
#include "hw/pci/msix.h"
#include "hw/pci/pci_device.h"
-#include "hw/vfio/vfio.h"
+#include "hw/vfio/vfio-common.h"
#include "qemu/error-report.h"
+/*
+ * Interfaces for IBM EEH (Enhanced Error Handling)
+ */
+static bool vfio_eeh_container_ok(VFIOContainer *container)
+{
+ /*
+ * As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
+ * implementation is broken if there are multiple groups in a
+ * container. The hardware works in units of Partitionable
+ * Endpoints (== IOMMU groups) and the EEH operations naively
+ * iterate across all groups in the container, without any logic
+ * to make sure the groups have their state synchronized. For
+ * certain operations (ENABLE) that might be ok, until an error
+ * occurs, but for others (GET_STATE) it's clearly broken.
+ */
+
+ /*
+ * XXX Once fixed kernels exist, test for them here
+ */
+
+ if (QLIST_EMPTY(&container->group_list)) {
+ return false;
+ }
+
+ if (QLIST_NEXT(QLIST_FIRST(&container->group_list), container_next)) {
+ return false;
+ }
+
+ return true;
+}
+
+static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
+{
+ struct vfio_eeh_pe_op pe_op = {
+ .argsz = sizeof(pe_op),
+ .op = op,
+ };
+ int ret;
+
+ if (!vfio_eeh_container_ok(container)) {
+ error_report("vfio/eeh: EEH_PE_OP 0x%x: "
+ "kernel requires a container with exactly one group", op);
+ return -EPERM;
+ }
+
+ ret = ioctl(container->fd, VFIO_EEH_PE_OP, &pe_op);
+ if (ret < 0) {
+ error_report("vfio/eeh: EEH_PE_OP 0x%x failed: %m", op);
+ return -errno;
+ }
+
+ return ret;
+}
+
+static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
+{
+ VFIOAddressSpace *space = vfio_get_address_space(as);
+ VFIOContainer *container = NULL;
+
+ if (QLIST_EMPTY(&space->containers)) {
+ /* No containers to act on */
+ goto out;
+ }
+
+ container = QLIST_FIRST(&space->containers);
+
+ if (QLIST_NEXT(container, next)) {
+ /*
+ * We don't yet have logic to synchronize EEH state across
+ * multiple containers
+ */
+ container = NULL;
+ goto out;
+ }
+
+out:
+ vfio_put_address_space(space);
+ return container;
+}
+
+static bool vfio_eeh_as_ok(AddressSpace *as)
+{
+ VFIOContainer *container = vfio_eeh_as_container(as);
+
+ return (container != NULL) && vfio_eeh_container_ok(container);
+}
+
+static int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
+{
+ VFIOContainer *container = vfio_eeh_as_container(as);
+
+ if (!container) {
+ return -ENODEV;
+ }
+ return vfio_eeh_container_op(container, op);
+}
+
bool spapr_phb_eeh_available(SpaprPhbState *sphb)
{
return vfio_eeh_as_ok(&sphb->iommu_as);
diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
index 5f257bffb9..bbf69ff55a 100644
--- a/hw/vfio/ap.c
+++ b/hw/vfio/ap.c
@@ -14,7 +14,6 @@
#include <linux/vfio.h>
#include <sys/ioctl.h>
#include "qapi/error.h"
-#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-common.h"
#include "hw/s390x/ap-device.h"
#include "qemu/error-report.h"
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index 6623ae237b..d857bb8d0f 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -20,7 +20,6 @@
#include <sys/ioctl.h>
#include "qapi/error.h"
-#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-common.h"
#include "hw/s390x/s390-ccw.h"
#include "hw/s390x/vfio-ccw.h"
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 9c5c6433f2..e72055e752 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -26,7 +26,6 @@
#include <linux/vfio.h>
#include "hw/vfio/vfio-common.h"
-#include "hw/vfio/vfio.h"
#include "hw/vfio/pci.h"
#include "exec/address-spaces.h"
#include "exec/memory.h"
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index fc88222377..83c0f05bba 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -26,7 +26,6 @@
#include <linux/vfio.h>
#include "hw/vfio/vfio-common.h"
-#include "hw/vfio/vfio.h"
#include "exec/address-spaces.h"
#include "exec/memory.h"
#include "exec/ram_addr.h"
@@ -1011,103 +1010,6 @@ static void vfio_put_base_device(VFIODevice *vbasedev)
close(vbasedev->fd);
}
-/*
- * Interfaces for IBM EEH (Enhanced Error Handling)
- */
-static bool vfio_eeh_container_ok(VFIOContainer *container)
-{
- /*
- * As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
- * implementation is broken if there are multiple groups in a
- * container. The hardware works in units of Partitionable
- * Endpoints (== IOMMU groups) and the EEH operations naively
- * iterate across all groups in the container, without any logic
- * to make sure the groups have their state synchronized. For
- * certain operations (ENABLE) that might be ok, until an error
- * occurs, but for others (GET_STATE) it's clearly broken.
- */
-
- /*
- * XXX Once fixed kernels exist, test for them here
- */
-
- if (QLIST_EMPTY(&container->group_list)) {
- return false;
- }
-
- if (QLIST_NEXT(QLIST_FIRST(&container->group_list), container_next)) {
- return false;
- }
-
- return true;
-}
-
-static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
-{
- struct vfio_eeh_pe_op pe_op = {
- .argsz = sizeof(pe_op),
- .op = op,
- };
- int ret;
-
- if (!vfio_eeh_container_ok(container)) {
- error_report("vfio/eeh: EEH_PE_OP 0x%x: "
- "kernel requires a container with exactly one group", op);
- return -EPERM;
- }
-
- ret = ioctl(container->fd, VFIO_EEH_PE_OP, &pe_op);
- if (ret < 0) {
- error_report("vfio/eeh: EEH_PE_OP 0x%x failed: %m", op);
- return -errno;
- }
-
- return ret;
-}
-
-static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
-{
- VFIOAddressSpace *space = vfio_get_address_space(as);
- VFIOContainer *container = NULL;
-
- if (QLIST_EMPTY(&space->containers)) {
- /* No containers to act on */
- goto out;
- }
-
- container = QLIST_FIRST(&space->containers);
-
- if (QLIST_NEXT(container, next)) {
- /*
- * We don't yet have logic to synchronize EEH state across
- * multiple containers
- */
- container = NULL;
- goto out;
- }
-
-out:
- vfio_put_address_space(space);
- return container;
-}
-
-bool vfio_eeh_as_ok(AddressSpace *as)
-{
- VFIOContainer *container = vfio_eeh_as_container(as);
-
- return (container != NULL) && vfio_eeh_container_ok(container);
-}
-
-int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
-{
- VFIOContainer *container = vfio_eeh_as_container(as);
-
- if (!container) {
- return -ENODEV;
- }
- return vfio_eeh_container_op(container, op);
-}
-
static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
{
char *tmp, group_path[PATH_MAX], *group_name;
diff --git a/hw/vfio/helpers.c b/hw/vfio/helpers.c
index 7e5da21b31..168847e7c5 100644
--- a/hw/vfio/helpers.c
+++ b/hw/vfio/helpers.c
@@ -23,7 +23,6 @@
#include <sys/ioctl.h>
#include "hw/vfio/vfio-common.h"
-#include "hw/vfio/vfio.h"
#include "hw/hw.h"
#include "trace.h"
#include "qapi/error.h"
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 9:19 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit " Zhenzhong Duan
` (35 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
vfio_container_add/del_section_window are spapr specific functions,
so move them into spapr.c to make container.c cleaner.
No functional changes intended.
Suggested-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/container.c | 90 ---------------------------------------------
hw/vfio/spapr.c | 90 +++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 90 insertions(+), 90 deletions(-)
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 83c0f05bba..7a3f005d1b 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -20,9 +20,6 @@
#include "qemu/osdep.h"
#include <sys/ioctl.h>
-#ifdef CONFIG_KVM
-#include <linux/kvm.h>
-#endif
#include <linux/vfio.h>
#include "hw/vfio/vfio-common.h"
@@ -32,7 +29,6 @@
#include "hw/hw.h"
#include "qemu/error-report.h"
#include "qemu/range.h"
-#include "sysemu/kvm.h"
#include "sysemu/reset.h"
#include "trace.h"
#include "qapi/error.h"
@@ -204,92 +200,6 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
return -errno;
}
-int vfio_container_add_section_window(VFIOContainer *container,
- MemoryRegionSection *section,
- Error **errp)
-{
- VFIOHostDMAWindow *hostwin;
- hwaddr pgsize = 0;
- int ret;
-
- if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
- return 0;
- }
-
- /* For now intersections are not allowed, we may relax this later */
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
- if (ranges_overlap(hostwin->min_iova,
- hostwin->max_iova - hostwin->min_iova + 1,
- section->offset_within_address_space,
- int128_get64(section->size))) {
- error_setg(errp,
- "region [0x%"PRIx64",0x%"PRIx64"] overlaps with existing"
- "host DMA window [0x%"PRIx64",0x%"PRIx64"]",
- section->offset_within_address_space,
- section->offset_within_address_space +
- int128_get64(section->size) - 1,
- hostwin->min_iova, hostwin->max_iova);
- return -EINVAL;
- }
- }
-
- ret = vfio_spapr_create_window(container, section, &pgsize);
- if (ret) {
- error_setg_errno(errp, -ret, "Failed to create SPAPR window");
- return ret;
- }
-
- vfio_host_win_add(container, section->offset_within_address_space,
- section->offset_within_address_space +
- int128_get64(section->size) - 1, pgsize);
-#ifdef CONFIG_KVM
- if (kvm_enabled()) {
- VFIOGroup *group;
- IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
- struct kvm_vfio_spapr_tce param;
- struct kvm_device_attr attr = {
- .group = KVM_DEV_VFIO_GROUP,
- .attr = KVM_DEV_VFIO_GROUP_SET_SPAPR_TCE,
- .addr = (uint64_t)(unsigned long)¶m,
- };
-
- if (!memory_region_iommu_get_attr(iommu_mr, IOMMU_ATTR_SPAPR_TCE_FD,
- ¶m.tablefd)) {
- QLIST_FOREACH(group, &container->group_list, container_next) {
- param.groupfd = group->fd;
- if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
- error_setg_errno(errp, errno,
- "vfio: failed GROUP_SET_SPAPR_TCE for "
- "KVM VFIO device %d and group fd %d",
- param.tablefd, param.groupfd);
- return -errno;
- }
- trace_vfio_spapr_group_attach(param.groupfd, param.tablefd);
- }
- }
- }
-#endif
- return 0;
-}
-
-void vfio_container_del_section_window(VFIOContainer *container,
- MemoryRegionSection *section)
-{
- if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
- return;
- }
-
- vfio_spapr_remove_window(container,
- section->offset_within_address_space);
- if (vfio_host_win_del(container,
- section->offset_within_address_space,
- section->offset_within_address_space +
- int128_get64(section->size) - 1) < 0) {
- hw_error("%s: Cannot delete missing window at %"HWADDR_PRIx,
- __func__, section->offset_within_address_space);
- }
-}
-
int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
{
int ret;
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 9ec1e95f6d..9a7517c042 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -11,6 +11,10 @@
#include "qemu/osdep.h"
#include <sys/ioctl.h>
#include <linux/vfio.h>
+#ifdef CONFIG_KVM
+#include <linux/kvm.h>
+#endif
+#include "sysemu/kvm.h"
#include "hw/vfio/vfio-common.h"
#include "hw/hw.h"
@@ -253,3 +257,89 @@ int vfio_spapr_remove_window(VFIOContainer *container,
return 0;
}
+
+int vfio_container_add_section_window(VFIOContainer *container,
+ MemoryRegionSection *section,
+ Error **errp)
+{
+ VFIOHostDMAWindow *hostwin;
+ hwaddr pgsize = 0;
+ int ret;
+
+ if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
+ return 0;
+ }
+
+ /* For now intersections are not allowed, we may relax this later */
+ QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ if (ranges_overlap(hostwin->min_iova,
+ hostwin->max_iova - hostwin->min_iova + 1,
+ section->offset_within_address_space,
+ int128_get64(section->size))) {
+ error_setg(errp,
+ "region [0x%"PRIx64",0x%"PRIx64"] overlaps with existing"
+ "host DMA window [0x%"PRIx64",0x%"PRIx64"]",
+ section->offset_within_address_space,
+ section->offset_within_address_space +
+ int128_get64(section->size) - 1,
+ hostwin->min_iova, hostwin->max_iova);
+ return -EINVAL;
+ }
+ }
+
+ ret = vfio_spapr_create_window(container, section, &pgsize);
+ if (ret) {
+ error_setg_errno(errp, -ret, "Failed to create SPAPR window");
+ return ret;
+ }
+
+ vfio_host_win_add(container, section->offset_within_address_space,
+ section->offset_within_address_space +
+ int128_get64(section->size) - 1, pgsize);
+#ifdef CONFIG_KVM
+ if (kvm_enabled()) {
+ VFIOGroup *group;
+ IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
+ struct kvm_vfio_spapr_tce param;
+ struct kvm_device_attr attr = {
+ .group = KVM_DEV_VFIO_GROUP,
+ .attr = KVM_DEV_VFIO_GROUP_SET_SPAPR_TCE,
+ .addr = (uint64_t)(unsigned long)¶m,
+ };
+
+ if (!memory_region_iommu_get_attr(iommu_mr, IOMMU_ATTR_SPAPR_TCE_FD,
+ ¶m.tablefd)) {
+ QLIST_FOREACH(group, &container->group_list, container_next) {
+ param.groupfd = group->fd;
+ if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
+ error_setg_errno(errp, errno,
+ "vfio: failed GROUP_SET_SPAPR_TCE for "
+ "KVM VFIO device %d and group fd %d",
+ param.tablefd, param.groupfd);
+ return -errno;
+ }
+ trace_vfio_spapr_group_attach(param.groupfd, param.tablefd);
+ }
+ }
+ }
+#endif
+ return 0;
+}
+
+void vfio_container_del_section_window(VFIOContainer *container,
+ MemoryRegionSection *section)
+{
+ if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
+ return;
+ }
+
+ vfio_spapr_remove_window(container,
+ section->offset_within_address_space);
+ if (vfio_host_win_del(container,
+ section->offset_within_address_space,
+ section->offset_within_address_space +
+ int128_get64(section->size) - 1) < 0) {
+ hw_error("%s: Cannot delete missing window at %"HWADDR_PRIx,
+ __func__, section->offset_within_address_space);
+ }
+}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit into spapr.c
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 9:27 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static Zhenzhong Duan
` (34 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
Move spapr specific init/deinit code into spapr.c and wrap
them with vfio_spapr_container_init/deinit, this way footprint
of spapr is further reduced, vfio_prereg_listener could also
be made static.
vfio_listener_release is unnecessary when prereg_listener is
moved out, so have it removed.
No functional changes intended.
Suggested-by: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 4 +-
hw/vfio/container.c | 82 +++++------------------------------
hw/vfio/spapr.c | 80 +++++++++++++++++++++++++++++++++-
3 files changed, 94 insertions(+), 72 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 0c3d390e8b..8d1d4f0a89 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -225,11 +225,14 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size);
+/* SPAPR specific */
int vfio_container_add_section_window(VFIOContainer *container,
MemoryRegionSection *section,
Error **errp);
void vfio_container_del_section_window(VFIOContainer *container,
MemoryRegionSection *section);
+bool vfio_spapr_container_init(VFIOContainer *container, Error **errp);
+void vfio_spapr_container_deinit(VFIOContainer *container);
void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
@@ -289,7 +292,6 @@ vfio_get_device_info_cap(struct vfio_device_info *info, uint16_t id);
struct vfio_info_cap_header *
vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
#endif
-extern const MemoryListener vfio_prereg_listener;
int vfio_spapr_create_window(VFIOContainer *container,
MemoryRegionSection *section,
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 7a3f005d1b..204b244b11 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -264,14 +264,6 @@ int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
return ret;
}
-static void vfio_listener_release(VFIOContainer *container)
-{
- memory_listener_unregister(&container->listener);
- if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
- memory_listener_unregister(&container->prereg_listener);
- }
-}
-
static struct vfio_info_cap_header *
vfio_get_iommu_type1_info_cap(struct vfio_iommu_type1_info *info, uint16_t id)
{
@@ -612,69 +604,11 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
case VFIO_SPAPR_TCE_v2_IOMMU:
case VFIO_SPAPR_TCE_IOMMU:
{
- struct vfio_iommu_spapr_tce_info info;
- bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
-
- /*
- * The host kernel code implementing VFIO_IOMMU_DISABLE is called
- * when container fd is closed so we do not call it explicitly
- * in this file.
- */
- if (!v2) {
- ret = ioctl(fd, VFIO_IOMMU_ENABLE);
- if (ret) {
- error_setg_errno(errp, errno, "failed to enable container");
- ret = -errno;
- goto enable_discards_exit;
- }
- } else {
- container->prereg_listener = vfio_prereg_listener;
-
- memory_listener_register(&container->prereg_listener,
- &address_space_memory);
- if (container->error) {
- memory_listener_unregister(&container->prereg_listener);
- ret = -1;
- error_propagate_prepend(errp, container->error,
- "RAM memory listener initialization failed: ");
- goto enable_discards_exit;
- }
- }
-
- info.argsz = sizeof(info);
- ret = ioctl(fd, VFIO_IOMMU_SPAPR_TCE_GET_INFO, &info);
+ ret = vfio_spapr_container_init(container, errp);
if (ret) {
- error_setg_errno(errp, errno,
- "VFIO_IOMMU_SPAPR_TCE_GET_INFO failed");
- ret = -errno;
- if (v2) {
- memory_listener_unregister(&container->prereg_listener);
- }
goto enable_discards_exit;
}
-
- if (v2) {
- container->pgsizes = info.ddw.pgsizes;
- /*
- * There is a default window in just created container.
- * To make region_add/del simpler, we better remove this
- * window now and let those iommu_listener callbacks
- * create/remove them when needed.
- */
- ret = vfio_spapr_remove_window(container, info.dma32_window_start);
- if (ret) {
- error_setg_errno(errp, -ret,
- "failed to remove existing window");
- goto enable_discards_exit;
- }
- } else {
- /* The default table uses 4K pages */
- container->pgsizes = 0x1000;
- vfio_host_win_add(container, info.dma32_window_start,
- info.dma32_window_start +
- info.dma32_window_size - 1,
- 0x1000);
- }
+ break;
}
}
@@ -704,7 +638,11 @@ listener_release_exit:
QLIST_REMOVE(group, container_next);
QLIST_REMOVE(container, next);
vfio_kvm_device_del_group(group);
- vfio_listener_release(container);
+ memory_listener_unregister(&container->listener);
+ if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
+ container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
+ vfio_spapr_container_deinit(container);
+ }
enable_discards_exit:
vfio_ram_block_discard_disable(container, false);
@@ -734,7 +672,11 @@ static void vfio_disconnect_container(VFIOGroup *group)
* group.
*/
if (QLIST_EMPTY(&container->group_list)) {
- vfio_listener_release(container);
+ memory_listener_unregister(&container->listener);
+ if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
+ container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
+ vfio_spapr_container_deinit(container);
+ }
}
if (ioctl(group->fd, VFIO_GROUP_UNSET_CONTAINER, &container->fd)) {
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 9a7517c042..fd0613dfde 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -15,6 +15,7 @@
#include <linux/kvm.h>
#endif
#include "sysemu/kvm.h"
+#include "exec/address-spaces.h"
#include "hw/vfio/vfio-common.h"
#include "hw/hw.h"
@@ -139,7 +140,7 @@ static void vfio_prereg_listener_region_del(MemoryListener *listener,
trace_vfio_prereg_unregister(reg.vaddr, reg.size, ret ? -errno : 0);
}
-const MemoryListener vfio_prereg_listener = {
+static const MemoryListener vfio_prereg_listener = {
.name = "vfio-pre-reg",
.region_add = vfio_prereg_listener_region_add,
.region_del = vfio_prereg_listener_region_del,
@@ -343,3 +344,80 @@ void vfio_container_del_section_window(VFIOContainer *container,
__func__, section->offset_within_address_space);
}
}
+
+bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
+{
+
+ struct vfio_iommu_spapr_tce_info info;
+ bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
+ int ret, fd = container->fd;
+
+ /*
+ * The host kernel code implementing VFIO_IOMMU_DISABLE is called
+ * when container fd is closed so we do not call it explicitly
+ * in this file.
+ */
+ if (!v2) {
+ ret = ioctl(fd, VFIO_IOMMU_ENABLE);
+ if (ret) {
+ error_setg_errno(errp, errno, "failed to enable container");
+ return -errno;
+ }
+ } else {
+ container->prereg_listener = vfio_prereg_listener;
+
+ memory_listener_register(&container->prereg_listener,
+ &address_space_memory);
+ if (container->error) {
+ ret = -1;
+ error_propagate_prepend(errp, container->error,
+ "RAM memory listener initialization failed: ");
+ goto listener_unregister_exit;
+ }
+ }
+
+ info.argsz = sizeof(info);
+ ret = ioctl(fd, VFIO_IOMMU_SPAPR_TCE_GET_INFO, &info);
+ if (ret) {
+ error_setg_errno(errp, errno,
+ "VFIO_IOMMU_SPAPR_TCE_GET_INFO failed");
+ ret = -errno;
+ goto listener_unregister_exit;
+ }
+
+ if (v2) {
+ container->pgsizes = info.ddw.pgsizes;
+ /*
+ * There is a default window in just created container.
+ * To make region_add/del simpler, we better remove this
+ * window now and let those iommu_listener callbacks
+ * create/remove them when needed.
+ */
+ ret = vfio_spapr_remove_window(container, info.dma32_window_start);
+ if (ret) {
+ error_setg_errno(errp, -ret,
+ "failed to remove existing window");
+ goto listener_unregister_exit;
+ }
+ } else {
+ /* The default table uses 4K pages */
+ container->pgsizes = 0x1000;
+ vfio_host_win_add(container, info.dma32_window_start,
+ info.dma32_window_start +
+ info.dma32_window_size - 1,
+ 0x1000);
+ }
+
+listener_unregister_exit:
+ if (v2) {
+ memory_listener_unregister(&container->prereg_listener);
+ }
+ return ret;
+}
+
+void vfio_spapr_container_deinit(VFIOContainer *container)
+{
+ if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
+ memory_listener_unregister(&container->prereg_listener);
+ }
+}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (2 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 9:27 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c Zhenzhong Duan
` (33 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
vfio_spapr_create_window calls vfio_spapr_remove_window,
With reoder of definition of the two, we can make
vfio_spapr_create/remove_window static.
No functional changes intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 6 -----
hw/vfio/spapr.c | 48 +++++++++++++++++------------------
2 files changed, 24 insertions(+), 30 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 8d1d4f0a89..f3174b3c5c 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -293,12 +293,6 @@ struct vfio_info_cap_header *
vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
#endif
-int vfio_spapr_create_window(VFIOContainer *container,
- MemoryRegionSection *section,
- hwaddr *pgsize);
-int vfio_spapr_remove_window(VFIOContainer *container,
- hwaddr offset_within_address_space);
-
bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
void vfio_migration_exit(VFIODevice *vbasedev);
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index fd0613dfde..43adbdb7b3 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -146,9 +146,30 @@ static const MemoryListener vfio_prereg_listener = {
.region_del = vfio_prereg_listener_region_del,
};
-int vfio_spapr_create_window(VFIOContainer *container,
- MemoryRegionSection *section,
- hwaddr *pgsize)
+static int vfio_spapr_remove_window(VFIOContainer *container,
+ hwaddr offset_within_address_space)
+{
+ struct vfio_iommu_spapr_tce_remove remove = {
+ .argsz = sizeof(remove),
+ .start_addr = offset_within_address_space,
+ };
+ int ret;
+
+ ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
+ if (ret) {
+ error_report("Failed to remove window at %"PRIx64,
+ (uint64_t)remove.start_addr);
+ return -errno;
+ }
+
+ trace_vfio_spapr_remove_window(offset_within_address_space);
+
+ return 0;
+}
+
+static int vfio_spapr_create_window(VFIOContainer *container,
+ MemoryRegionSection *section,
+ hwaddr *pgsize)
{
int ret = 0;
IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
@@ -238,27 +259,6 @@ int vfio_spapr_create_window(VFIOContainer *container,
return 0;
}
-int vfio_spapr_remove_window(VFIOContainer *container,
- hwaddr offset_within_address_space)
-{
- struct vfio_iommu_spapr_tce_remove remove = {
- .argsz = sizeof(remove),
- .start_addr = offset_within_address_space,
- };
- int ret;
-
- ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
- if (ret) {
- error_report("Failed to remove window at %"PRIx64,
- (uint64_t)remove.start_addr);
- return -errno;
- }
-
- trace_vfio_spapr_remove_window(offset_within_address_space);
-
- return 0;
-}
-
int vfio_container_add_section_window(VFIOContainer *container,
MemoryRegionSection *section,
Error **errp)
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (3 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 9:30 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
` (32 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
Only spapr supports a customed host window list, other vfio driver
assume 64bit host window. So remove the check in listener callback
and move vfio_host_win_add/del into spapr.c and make static.
Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 5 ---
hw/vfio/common.c | 64 -----------------------------------
hw/vfio/container.c | 16 ---------
hw/vfio/spapr.c | 38 +++++++++++++++++++++
4 files changed, 38 insertions(+), 85 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index f3174b3c5c..b9c7a7e588 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -207,11 +207,6 @@ typedef struct {
hwaddr pages;
} VFIOBitmap;
-void vfio_host_win_add(VFIOContainer *container,
- hwaddr min_iova, hwaddr max_iova,
- uint64_t iova_pgsizes);
-int vfio_host_win_del(VFIOContainer *container, hwaddr min_iova,
- hwaddr max_iova);
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOContainer *container);
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index e72055e752..0ebf4d9256 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -245,44 +245,6 @@ bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
return true;
}
-void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
- hwaddr max_iova, uint64_t iova_pgsizes)
-{
- VFIOHostDMAWindow *hostwin;
-
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
- if (ranges_overlap(hostwin->min_iova,
- hostwin->max_iova - hostwin->min_iova + 1,
- min_iova,
- max_iova - min_iova + 1)) {
- hw_error("%s: Overlapped IOMMU are not enabled", __func__);
- }
- }
-
- hostwin = g_malloc0(sizeof(*hostwin));
-
- hostwin->min_iova = min_iova;
- hostwin->max_iova = max_iova;
- hostwin->iova_pgsizes = iova_pgsizes;
- QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
-}
-
-int vfio_host_win_del(VFIOContainer *container,
- hwaddr min_iova, hwaddr max_iova)
-{
- VFIOHostDMAWindow *hostwin;
-
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
- if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
- QLIST_REMOVE(hostwin, hostwin_next);
- g_free(hostwin);
- return 0;
- }
- }
-
- return -1;
-}
-
static bool vfio_listener_skipped_section(MemoryRegionSection *section)
{
return (!memory_region_is_ram(section->mr) &&
@@ -531,22 +493,6 @@ static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
g_free(vrdl);
}
-static VFIOHostDMAWindow *vfio_find_hostwin(VFIOContainer *container,
- hwaddr iova, hwaddr end)
-{
- VFIOHostDMAWindow *hostwin;
- bool hostwin_found = false;
-
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
- if (hostwin->min_iova <= iova && end <= hostwin->max_iova) {
- hostwin_found = true;
- break;
- }
- }
-
- return hostwin_found ? hostwin : NULL;
-}
-
static bool vfio_known_safe_misalignment(MemoryRegionSection *section)
{
MemoryRegion *mr = section->mr;
@@ -647,13 +593,6 @@ static void vfio_listener_region_add(MemoryListener *listener,
goto fail;
}
- hostwin = vfio_find_hostwin(container, iova, end);
- if (!hostwin) {
- error_setg(&err, "Container %p can't map guest IOVA region"
- " 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx, container, iova, end);
- goto fail;
- }
-
memory_region_ref(section->mr);
if (memory_region_is_iommu(section->mr)) {
@@ -835,9 +774,6 @@ static void vfio_listener_region_del(MemoryListener *listener,
hwaddr pgmask;
VFIOHostDMAWindow *hostwin;
- hostwin = vfio_find_hostwin(container, iova, end);
- assert(hostwin); /* or region_add() would have failed */
-
pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
try_unmap = !((iova & pgmask) || (int128_get64(llsize) & pgmask));
} else if (memory_region_has_ram_discard_manager(section->mr)) {
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 204b244b11..242010036a 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -551,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->dma_max_mappings = 0;
container->iova_ranges = NULL;
QLIST_INIT(&container->giommu_list);
- QLIST_INIT(&container->hostwin_list);
QLIST_INIT(&container->vrdl_list);
ret = vfio_init_container(container, group->fd, errp);
@@ -591,14 +590,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
vfio_get_iommu_info_migration(container, info);
g_free(info);
-
- /*
- * FIXME: We should parse VFIO_IOMMU_TYPE1_INFO_CAP_IOVA_RANGE
- * information to get the actual window extent rather than assume
- * a 64-bit IOVA address space.
- */
- vfio_host_win_add(container, 0, (hwaddr)-1, container->pgsizes);
-
break;
}
case VFIO_SPAPR_TCE_v2_IOMMU:
@@ -687,7 +678,6 @@ static void vfio_disconnect_container(VFIOGroup *group)
if (QLIST_EMPTY(&container->group_list)) {
VFIOAddressSpace *space = container->space;
VFIOGuestIOMMU *giommu, *tmp;
- VFIOHostDMAWindow *hostwin, *next;
QLIST_REMOVE(container, next);
@@ -698,12 +688,6 @@ static void vfio_disconnect_container(VFIOGroup *group)
g_free(giommu);
}
- QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
- next) {
- QLIST_REMOVE(hostwin, hostwin_next);
- g_free(hostwin);
- }
-
trace_vfio_disconnect_container(container->fd);
close(container->fd);
vfio_free_container(container);
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 43adbdb7b3..3495737ab2 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -146,6 +146,44 @@ static const MemoryListener vfio_prereg_listener = {
.region_del = vfio_prereg_listener_region_del,
};
+static void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
+ hwaddr max_iova, uint64_t iova_pgsizes)
+{
+ VFIOHostDMAWindow *hostwin;
+
+ QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ if (ranges_overlap(hostwin->min_iova,
+ hostwin->max_iova - hostwin->min_iova + 1,
+ min_iova,
+ max_iova - min_iova + 1)) {
+ hw_error("%s: Overlapped IOMMU are not enabled", __func__);
+ }
+ }
+
+ hostwin = g_malloc0(sizeof(*hostwin));
+
+ hostwin->min_iova = min_iova;
+ hostwin->max_iova = max_iova;
+ hostwin->iova_pgsizes = iova_pgsizes;
+ QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
+}
+
+static int vfio_host_win_del(VFIOContainer *container,
+ hwaddr min_iova, hwaddr max_iova)
+{
+ VFIOHostDMAWindow *hostwin;
+
+ QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
+ QLIST_REMOVE(hostwin, hostwin_next);
+ g_free(hostwin);
+ return 0;
+ }
+ }
+
+ return -1;
+}
+
static int vfio_spapr_remove_window(VFIOContainer *container,
hwaddr offset_within_address_space)
{
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (4 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:02 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps Zhenzhong Duan
` (31 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Yi Sun
Introduce a dumb VFIOContainer base object and its targetted interface.
This is willingly not a QOM object because we don't want it to be
visible from the user interface. The VFIOContainer will be smoothly
populated in subsequent patches as well as interfaces.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 8 +---
include/hw/vfio/vfio-container-base.h | 64 +++++++++++++++++++++++++++
2 files changed, 66 insertions(+), 6 deletions(-)
create mode 100644 include/hw/vfio/vfio-container-base.h
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index b9c7a7e588..d8f293cb57 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -30,6 +30,7 @@
#include <linux/vfio.h>
#endif
#include "sysemu/sysemu.h"
+#include "hw/vfio/vfio-container-base.h"
#define VFIO_MSG_PREFIX "vfio %s: "
@@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
struct VFIOGroup;
typedef struct VFIOContainer {
+ VFIOContainerBase bcontainer;
VFIOAddressSpace *space;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener listener;
@@ -201,12 +203,6 @@ typedef struct VFIODisplay {
} dmabuf;
} VFIODisplay;
-typedef struct {
- unsigned long *bitmap;
- hwaddr size;
- hwaddr pages;
-} VFIOBitmap;
-
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOContainer *container);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
new file mode 100644
index 0000000000..5becbd51a7
--- /dev/null
+++ b/include/hw/vfio/vfio-container-base.h
@@ -0,0 +1,64 @@
+/*
+ * VFIO BASE CONTAINER
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
+#define HW_VFIO_VFIO_BASE_CONTAINER_H
+
+#include "exec/memory.h"
+#ifndef CONFIG_USER_ONLY
+#include "exec/hwaddr.h"
+#endif
+
+typedef struct VFIODevice VFIODevice;
+typedef struct VFIOIOMMUOps VFIOIOMMUOps;
+
+typedef struct {
+ unsigned long *bitmap;
+ hwaddr size;
+ hwaddr pages;
+} VFIOBitmap;
+
+/*
+ * This is the base object for vfio container backends
+ */
+typedef struct VFIOContainerBase {
+ const VFIOIOMMUOps *ops;
+} VFIOContainerBase;
+
+struct VFIOIOMMUOps {
+ /* basic feature */
+ int (*dma_map)(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly);
+ int (*dma_unmap)(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb);
+ int (*attach_device)(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp);
+ void (*detach_device)(VFIODevice *vbasedev);
+ /* migration feature */
+ int (*set_dirty_page_tracking)(VFIOContainerBase *bcontainer, bool start);
+ int (*query_dirty_bitmap)(VFIOContainerBase *bcontainer, VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size);
+};
+#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (5 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:20 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
` (30 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
This empty VFIOIOMMUOps named vfio_legacy_ops will hold all general
IOMMU ops of legacy container.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 2 +-
hw/vfio/container.c | 5 +++++
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index d8f293cb57..8ded5cd8e4 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -255,7 +255,7 @@ typedef QLIST_HEAD(VFIOGroupList, VFIOGroup) VFIOGroupList;
typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
extern VFIOGroupList vfio_group_list;
extern VFIODeviceList vfio_device_list;
-
+extern const VFIOIOMMUOps vfio_legacy_ops;
extern const MemoryListener vfio_memory_listener;
extern int vfio_kvm_device_fd;
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 242010036a..4bc43ddfa4 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -472,6 +472,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
Error **errp)
{
VFIOContainer *container;
+ VFIOContainerBase *bcontainer;
int ret, fd;
VFIOAddressSpace *space;
@@ -552,6 +553,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->iova_ranges = NULL;
QLIST_INIT(&container->giommu_list);
QLIST_INIT(&container->vrdl_list);
+ bcontainer = &container->bcontainer;
+ bcontainer->ops = &vfio_legacy_ops;
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -933,3 +936,5 @@ void vfio_detach_device(VFIODevice *vbasedev)
vfio_put_base_device(vbasedev);
vfio_put_group(group);
}
+
+const VFIOIOMMUOps vfio_legacy_ops;
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (6 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:26 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 09/37] vfio/common: Move giommu_list in base container Zhenzhong Duan
` (29 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 4 ---
include/hw/vfio/vfio-container-base.h | 7 ++++
hw/vfio/common.c | 45 +++++++++++++-----------
hw/vfio/container-base.c | 49 +++++++++++++++++++++++++++
hw/vfio/container.c | 22 ++++++++----
hw/vfio/meson.build | 1 +
hw/vfio/trace-events | 2 +-
7 files changed, 98 insertions(+), 32 deletions(-)
create mode 100644 hw/vfio/container-base.c
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 8ded5cd8e4..97056224f4 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -208,10 +208,6 @@ void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOContainer *container);
/* container->fd */
-int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
- ram_addr_t size, IOMMUTLBEntry *iotlb);
-int vfio_dma_map(VFIOContainer *container, hwaddr iova,
- ram_addr_t size, void *vaddr, bool readonly);
int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
hwaddr iova, hwaddr size);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 5becbd51a7..077e638ee8 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -45,6 +45,13 @@ typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
} VFIOContainerBase;
+int vfio_container_dma_map(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly);
+int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb);
+
struct VFIOIOMMUOps {
/* basic feature */
int (*dma_map)(VFIOContainerBase *bcontainer,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 0ebf4d9256..141f2b54a4 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -292,7 +292,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
- VFIOContainer *container = giommu->container;
+ VFIOContainerBase *bcontainer = &giommu->container->bcontainer;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
void *vaddr;
int ret;
@@ -322,21 +322,22 @@ static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
* of vaddr will always be there, even if the memory object is
* destroyed and its backing memory munmap-ed.
*/
- ret = vfio_dma_map(container, iova,
- iotlb->addr_mask + 1, vaddr,
- read_only);
+ ret = vfio_container_dma_map(bcontainer, iova,
+ iotlb->addr_mask + 1, vaddr,
+ read_only);
if (ret) {
- error_report("vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
- container, iova,
+ bcontainer, iova,
iotlb->addr_mask + 1, vaddr, ret, strerror(-ret));
}
} else {
- ret = vfio_dma_unmap(container, iova, iotlb->addr_mask + 1, iotlb);
+ ret = vfio_container_dma_unmap(bcontainer, iova,
+ iotlb->addr_mask + 1, iotlb);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova,
+ bcontainer, iova,
iotlb->addr_mask + 1, ret, strerror(-ret));
vfio_set_migration_error(ret);
}
@@ -355,9 +356,10 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
int ret;
/* Unmap with a single call. */
- ret = vfio_dma_unmap(vrdl->container, iova, size , NULL);
+ ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
+ iova, size , NULL);
if (ret) {
- error_report("%s: vfio_dma_unmap() failed: %s", __func__,
+ error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
strerror(-ret));
}
}
@@ -385,8 +387,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
section->offset_within_address_space;
vaddr = memory_region_get_ram_ptr(section->mr) + start;
- ret = vfio_dma_map(vrdl->container, iova, next - start,
- vaddr, section->readonly);
+ ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
+ next - start, vaddr, section->readonly);
if (ret) {
/* Rollback */
vfio_ram_discard_notify_discard(rdl, section);
@@ -685,10 +687,11 @@ static void vfio_listener_region_add(MemoryListener *listener,
}
}
- ret = vfio_dma_map(container, iova, int128_get64(llsize),
- vaddr, section->readonly);
+ ret = vfio_container_dma_map(&container->bcontainer,
+ iova, int128_get64(llsize), vaddr,
+ section->readonly);
if (ret) {
- error_setg(&err, "vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
+ error_setg(&err, "vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
container, iova, int128_get64(llsize), vaddr, ret,
strerror(-ret));
@@ -786,18 +789,20 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (int128_eq(llsize, int128_2_64())) {
/* The unmap ioctl doesn't accept a full 64-bit span. */
llsize = int128_rshift(llsize, 1);
- ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
+ ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ int128_get64(llsize), NULL);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
container, iova, int128_get64(llsize), ret,
strerror(-ret));
}
iova += int128_get64(llsize);
}
- ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
+ ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ int128_get64(llsize), NULL);
if (ret) {
- error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
+ error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
container, iova, int128_get64(llsize), ret,
strerror(-ret));
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
new file mode 100644
index 0000000000..9db8b89b2f
--- /dev/null
+++ b/hw/vfio/container-base.c
@@ -0,0 +1,49 @@
+/*
+ * VFIO BASE CONTAINER
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "qapi/error.h"
+#include "qemu/error-report.h"
+#include "hw/vfio/vfio-container-base.h"
+
+int vfio_container_dma_map(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ void *vaddr, bool readonly)
+{
+ if (!bcontainer->ops->dma_map) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->dma_map(bcontainer, iova, size, vaddr, readonly);
+}
+
+int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb)
+{
+ if (!bcontainer->ops->dma_unmap) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
+}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 4bc43ddfa4..c04df26323 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -115,9 +115,11 @@ unmap_exit:
/*
* DMA - Mapping and unmapping for the "type1" IOMMU interface used on x86
*/
-int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
- ram_addr_t size, IOMMUTLBEntry *iotlb)
+static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
+ ram_addr_t size, IOMMUTLBEntry *iotlb)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
struct vfio_iommu_type1_dma_unmap unmap = {
.argsz = sizeof(unmap),
.flags = 0,
@@ -151,7 +153,7 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
*/
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
- trace_vfio_dma_unmap_overflow_workaround();
+ trace_vfio_legacy_dma_unmap_overflow_workaround();
unmap.size -= 1ULL << ctz64(container->pgsizes);
continue;
}
@@ -170,9 +172,11 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
return 0;
}
-int vfio_dma_map(VFIOContainer *container, hwaddr iova,
- ram_addr_t size, void *vaddr, bool readonly)
+static int vfio_legacy_dma_map(VFIOContainerBase *bcontainer, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
struct vfio_iommu_type1_dma_map map = {
.argsz = sizeof(map),
.flags = VFIO_DMA_MAP_FLAG_READ,
@@ -191,7 +195,8 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
* the VGA ROM space.
*/
if (ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0 ||
- (errno == EBUSY && vfio_dma_unmap(container, iova, size, NULL) == 0 &&
+ (errno == EBUSY &&
+ vfio_legacy_dma_unmap(bcontainer, iova, size, NULL) == 0 &&
ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0)) {
return 0;
}
@@ -937,4 +942,7 @@ void vfio_detach_device(VFIODevice *vbasedev)
vfio_put_group(group);
}
-const VFIOIOMMUOps vfio_legacy_ops;
+const VFIOIOMMUOps vfio_legacy_ops = {
+ .dma_map = vfio_legacy_dma_map,
+ .dma_unmap = vfio_legacy_dma_unmap,
+};
diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
index 2a6912c940..eb6ce6229d 100644
--- a/hw/vfio/meson.build
+++ b/hw/vfio/meson.build
@@ -2,6 +2,7 @@ vfio_ss = ss.source_set()
vfio_ss.add(files(
'helpers.c',
'common.c',
+ 'container-base.c',
'container.c',
'spapr.c',
'migration.c',
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 0eb2387cf2..9f7fedee98 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -116,7 +116,7 @@ vfio_region_unmap(const char *name, unsigned long offset, unsigned long end) "Re
vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Device %s region %d: %d sparse mmap entries"
vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
-vfio_dma_unmap_overflow_workaround(void) ""
+vfio_legacy_dma_unmap_overflow_workaround(void) ""
vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 09/37] vfio/common: Move giommu_list in base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (7 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:38 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 10/37] vfio/container: Move space field to " Zhenzhong Duan
` (28 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Move the giommu_list field in the base object and store the
base container in the VFIOGuestIOMMU.
We introduce vfio_container_init/destroy helper on the base
container.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
[ clg: context changes ]
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 9 ---------
include/hw/vfio/vfio-container-base.h | 13 +++++++++++++
hw/vfio/common.c | 17 +++++++++++------
hw/vfio/container-base.c | 18 ++++++++++++++++++
hw/vfio/container.c | 13 +++----------
5 files changed, 45 insertions(+), 25 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 97056224f4..fcb4003a21 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -95,7 +95,6 @@ typedef struct VFIOContainer {
uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
unsigned int dma_max_mappings;
- QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
@@ -104,14 +103,6 @@ typedef struct VFIOContainer {
GList *iova_ranges;
} VFIOContainer;
-typedef struct VFIOGuestIOMMU {
- VFIOContainer *container;
- IOMMUMemoryRegion *iommu_mr;
- hwaddr iommu_offset;
- IOMMUNotifier n;
- QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
-} VFIOGuestIOMMU;
-
typedef struct VFIORamDiscardListener {
VFIOContainer *container;
MemoryRegion *mr;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 077e638ee8..71e1e4324e 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -43,8 +43,17 @@ typedef struct {
*/
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
+ QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
} VFIOContainerBase;
+typedef struct VFIOGuestIOMMU {
+ VFIOContainerBase *bcontainer;
+ IOMMUMemoryRegion *iommu_mr;
+ hwaddr iommu_offset;
+ IOMMUNotifier n;
+ QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
+} VFIOGuestIOMMU;
+
int vfio_container_dma_map(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
void *vaddr, bool readonly);
@@ -52,6 +61,10 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+void vfio_container_init(VFIOContainerBase *bcontainer,
+ const VFIOIOMMUOps *ops);
+void vfio_container_destroy(VFIOContainerBase *bcontainer);
+
struct VFIOIOMMUOps {
/* basic feature */
int (*dma_map)(VFIOContainerBase *bcontainer,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 141f2b54a4..4f130ad87c 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -292,7 +292,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
{
VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
- VFIOContainerBase *bcontainer = &giommu->container->bcontainer;
+ VFIOContainerBase *bcontainer = giommu->bcontainer;
hwaddr iova = iotlb->iova + giommu->iommu_offset;
void *vaddr;
int ret;
@@ -569,6 +569,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOContainerBase *bcontainer = &container->bcontainer;
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -613,7 +614,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
giommu->iommu_mr = iommu_mr;
giommu->iommu_offset = section->offset_within_address_space -
section->offset_within_region;
- giommu->container = container;
+ giommu->bcontainer = bcontainer;
llend = int128_add(int128_make64(section->offset_within_region),
section->size);
llend = int128_sub(llend, int128_one());
@@ -648,7 +649,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
g_free(giommu);
goto fail;
}
- QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
+ QLIST_INSERT_HEAD(&bcontainer->giommu_list, giommu, giommu_next);
memory_region_iommu_replay(giommu->iommu_mr, &giommu->n);
return;
@@ -733,6 +734,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOContainerBase *bcontainer = &container->bcontainer;
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -745,7 +747,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (memory_region_is_iommu(section->mr)) {
VFIOGuestIOMMU *giommu;
- QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
+ QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
giommu->n.start == section->offset_within_region) {
memory_region_unregister_iommu_notifier(section->mr,
@@ -1208,7 +1210,9 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
vfio_giommu_dirty_notifier *gdn = container_of(n,
vfio_giommu_dirty_notifier, n);
VFIOGuestIOMMU *giommu = gdn->giommu;
- VFIOContainer *container = giommu->container;
+ VFIOContainerBase *bcontainer = giommu->bcontainer;
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
hwaddr iova = iotlb->iova + giommu->iommu_offset;
ram_addr_t translated_addr;
int ret = -EINVAL;
@@ -1286,12 +1290,13 @@ static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
static int vfio_sync_dirty_bitmap(VFIOContainer *container,
MemoryRegionSection *section)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
ram_addr_t ram_addr;
if (memory_region_is_iommu(section->mr)) {
VFIOGuestIOMMU *giommu;
- QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
+ QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
giommu->n.start == section->offset_within_region) {
Int128 llend;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 9db8b89b2f..6be7ce65ad 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -47,3 +47,21 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+
+void vfio_container_init(VFIOContainerBase *bcontainer, const VFIOIOMMUOps *ops)
+{
+ bcontainer->ops = ops;
+ QLIST_INIT(&bcontainer->giommu_list);
+}
+
+void vfio_container_destroy(VFIOContainerBase *bcontainer)
+{
+ VFIOGuestIOMMU *giommu, *tmp;
+
+ QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
+ memory_region_unregister_iommu_notifier(
+ MEMORY_REGION(giommu->iommu_mr), &giommu->n);
+ QLIST_REMOVE(giommu, giommu_next);
+ g_free(giommu);
+ }
+}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index c04df26323..b2b6e05287 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -556,10 +556,9 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->dirty_pages_supported = false;
container->dma_max_mappings = 0;
container->iova_ranges = NULL;
- QLIST_INIT(&container->giommu_list);
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
- bcontainer->ops = &vfio_legacy_ops;
+ vfio_container_init(bcontainer, &vfio_legacy_ops);
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -661,6 +660,7 @@ put_space_exit:
static void vfio_disconnect_container(VFIOGroup *group)
{
VFIOContainer *container = group->container;
+ VFIOContainerBase *bcontainer = &container->bcontainer;
QLIST_REMOVE(group, container_next);
group->container = NULL;
@@ -685,16 +685,9 @@ static void vfio_disconnect_container(VFIOGroup *group)
if (QLIST_EMPTY(&container->group_list)) {
VFIOAddressSpace *space = container->space;
- VFIOGuestIOMMU *giommu, *tmp;
QLIST_REMOVE(container, next);
-
- QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next, tmp) {
- memory_region_unregister_iommu_notifier(
- MEMORY_REGION(giommu->iommu_mr), &giommu->n);
- QLIST_REMOVE(giommu, giommu_next);
- g_free(giommu);
- }
+ vfio_container_destroy(bcontainer);
trace_vfio_disconnect_container(container->fd);
close(container->fd);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 10/37] vfio/container: Move space field to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (8 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 09/37] vfio/common: Move giommu_list in base container Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:46 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
` (27 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
Move the space field to the base object. Also the VFIOAddressSpace
now contains a list of base containers.
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
[ clg: context changes ]
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 8 --------
include/hw/vfio/vfio-container-base.h | 9 +++++++++
hw/ppc/spapr_pci_vfio.c | 10 +++++-----
hw/vfio/common.c | 4 ++--
hw/vfio/container-base.c | 6 +++++-
hw/vfio/container.c | 18 +++++++++---------
6 files changed, 30 insertions(+), 25 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index fcb4003a21..857d2b8076 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -73,17 +73,10 @@ typedef struct VFIOMigration {
bool initial_data_sent;
} VFIOMigration;
-typedef struct VFIOAddressSpace {
- AddressSpace *as;
- QLIST_HEAD(, VFIOContainer) containers;
- QLIST_ENTRY(VFIOAddressSpace) list;
-} VFIOAddressSpace;
-
struct VFIOGroup;
typedef struct VFIOContainer {
VFIOContainerBase bcontainer;
- VFIOAddressSpace *space;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener listener;
MemoryListener prereg_listener;
@@ -98,7 +91,6 @@ typedef struct VFIOContainer {
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
- QLIST_ENTRY(VFIOContainer) next;
QLIST_HEAD(, VFIODevice) device_list;
GList *iova_ranges;
} VFIOContainer;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 71e1e4324e..a5fef3e6a8 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -38,12 +38,20 @@ typedef struct {
hwaddr pages;
} VFIOBitmap;
+typedef struct VFIOAddressSpace {
+ AddressSpace *as;
+ QLIST_HEAD(, VFIOContainerBase) containers;
+ QLIST_ENTRY(VFIOAddressSpace) list;
+} VFIOAddressSpace;
+
/*
* This is the base object for vfio container backends
*/
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
+ VFIOAddressSpace *space;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
+ QLIST_ENTRY(VFIOContainerBase) next;
} VFIOContainerBase;
typedef struct VFIOGuestIOMMU {
@@ -62,6 +70,7 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
IOMMUTLBEntry *iotlb);
void vfio_container_init(VFIOContainerBase *bcontainer,
+ VFIOAddressSpace *space,
const VFIOIOMMUOps *ops);
void vfio_container_destroy(VFIOContainerBase *bcontainer);
diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
index f283f7e38d..d1d07bec46 100644
--- a/hw/ppc/spapr_pci_vfio.c
+++ b/hw/ppc/spapr_pci_vfio.c
@@ -84,27 +84,27 @@ static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
{
VFIOAddressSpace *space = vfio_get_address_space(as);
- VFIOContainer *container = NULL;
+ VFIOContainerBase *bcontainer = NULL;
if (QLIST_EMPTY(&space->containers)) {
/* No containers to act on */
goto out;
}
- container = QLIST_FIRST(&space->containers);
+ bcontainer = QLIST_FIRST(&space->containers);
- if (QLIST_NEXT(container, next)) {
+ if (QLIST_NEXT(bcontainer, next)) {
/*
* We don't yet have logic to synchronize EEH state across
* multiple containers
*/
- container = NULL;
+ bcontainer = NULL;
goto out;
}
out:
vfio_put_address_space(space);
- return container;
+ return container_of(bcontainer, VFIOContainer, bcontainer);
}
static bool vfio_eeh_as_ok(AddressSpace *as)
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 4f130ad87c..f87a0dcec3 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
bool vfio_viommu_preset(VFIODevice *vbasedev)
{
- return vbasedev->container->space->as != &address_space_memory;
+ return vbasedev->container->bcontainer.space->as != &address_space_memory;
}
static void vfio_set_migration_error(int err)
@@ -924,7 +924,7 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
dirty.container = container;
memory_listener_register(&dirty.listener,
- container->space->as);
+ container->bcontainer.space->as);
*ranges = dirty.ranges;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 6be7ce65ad..99b2957d7b 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -48,9 +48,11 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
-void vfio_container_init(VFIOContainerBase *bcontainer, const VFIOIOMMUOps *ops)
+void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
+ const VFIOIOMMUOps *ops)
{
bcontainer->ops = ops;
+ bcontainer->space = space;
QLIST_INIT(&bcontainer->giommu_list);
}
@@ -58,6 +60,8 @@ void vfio_container_destroy(VFIOContainerBase *bcontainer)
{
VFIOGuestIOMMU *giommu, *tmp;
+ QLIST_REMOVE(bcontainer, next);
+
QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
memory_region_unregister_iommu_notifier(
MEMORY_REGION(giommu->iommu_mr), &giommu->n);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index b2b6e05287..761310fa51 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -514,7 +514,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
* details once we know which type of IOMMU we are using.
*/
- QLIST_FOREACH(container, &space->containers, next) {
+ QLIST_FOREACH(bcontainer, &space->containers, next) {
+ container = container_of(bcontainer, VFIOContainer, bcontainer);
if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
ret = vfio_ram_block_discard_disable(container, true);
if (ret) {
@@ -550,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
container = g_malloc0(sizeof(*container));
- container->space = space;
container->fd = fd;
container->error = NULL;
container->dirty_pages_supported = false;
@@ -558,7 +558,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->iova_ranges = NULL;
QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
- vfio_container_init(bcontainer, &vfio_legacy_ops);
+ vfio_container_init(bcontainer, space, &vfio_legacy_ops);
ret = vfio_init_container(container, group->fd, errp);
if (ret) {
@@ -613,14 +613,15 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
vfio_kvm_device_add_group(group);
QLIST_INIT(&container->group_list);
- QLIST_INSERT_HEAD(&space->containers, container, next);
+ QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
group->container = container;
QLIST_INSERT_HEAD(&container->group_list, group, container_next);
container->listener = vfio_memory_listener;
- memory_listener_register(&container->listener, container->space->as);
+ memory_listener_register(&container->listener,
+ container->bcontainer.space->as);
if (container->error) {
ret = -1;
@@ -634,7 +635,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
return 0;
listener_release_exit:
QLIST_REMOVE(group, container_next);
- QLIST_REMOVE(container, next);
+ QLIST_REMOVE(bcontainer, next);
vfio_kvm_device_del_group(group);
memory_listener_unregister(&container->listener);
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
@@ -684,9 +685,8 @@ static void vfio_disconnect_container(VFIOGroup *group)
}
if (QLIST_EMPTY(&container->group_list)) {
- VFIOAddressSpace *space = container->space;
+ VFIOAddressSpace *space = container->bcontainer.space;
- QLIST_REMOVE(container, next);
vfio_container_destroy(bcontainer);
trace_vfio_disconnect_container(container->fd);
@@ -706,7 +706,7 @@ static VFIOGroup *vfio_get_group(int groupid, AddressSpace *as, Error **errp)
QLIST_FOREACH(group, &vfio_group_list, next) {
if (group->groupid == groupid) {
/* Found it. Now is it already in the right context? */
- if (group->container->space->as == as) {
+ if (group->container->bcontainer.space->as == as) {
return group;
} else {
error_setg(errp, "group %d used in multiple address spaces",
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (9 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 10/37] vfio/container: Move space field to " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 14:53 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 12/37] vfio/container: Move per container device list in base container Zhenzhong Duan
` (26 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
dirty_pages_supported field is also moved to the base container
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 6 ------
include/hw/vfio/vfio-container-base.h | 6 ++++++
hw/vfio/common.c | 12 ++++++++----
hw/vfio/container-base.c | 23 +++++++++++++++++++++++
hw/vfio/container.c | 21 ++++++++++++++-------
5 files changed, 51 insertions(+), 17 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 857d2b8076..d053c61872 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -83,7 +83,6 @@ typedef struct VFIOContainer {
unsigned iommu_type;
Error *error;
bool initialized;
- bool dirty_pages_supported;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
@@ -190,11 +189,6 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
bool vfio_devices_all_running_and_saving(VFIOContainer *container);
-/* container->fd */
-int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
-int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
- hwaddr iova, hwaddr size);
-
/* SPAPR specific */
int vfio_container_add_section_window(VFIOContainer *container,
MemoryRegionSection *section,
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index a5fef3e6a8..ea8436a064 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -50,6 +50,7 @@ typedef struct VFIOAddressSpace {
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
VFIOAddressSpace *space;
+ bool dirty_pages_supported;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_ENTRY(VFIOContainerBase) next;
} VFIOContainerBase;
@@ -68,6 +69,11 @@ int vfio_container_dma_map(VFIOContainerBase *bcontainer,
int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
+ bool start);
+int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size);
void vfio_container_init(VFIOContainerBase *bcontainer,
VFIOAddressSpace *space,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index f87a0dcec3..7d9b87fc67 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1081,7 +1081,8 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
if (vfio_devices_all_device_dirty_tracking(container)) {
ret = vfio_devices_dma_logging_start(container);
} else {
- ret = vfio_set_dirty_page_tracking(container, true);
+ ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
+ true);
}
if (ret) {
@@ -1099,7 +1100,8 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
if (vfio_devices_all_device_dirty_tracking(container)) {
vfio_devices_dma_logging_stop(container);
} else {
- ret = vfio_set_dirty_page_tracking(container, false);
+ ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
+ false);
}
if (ret) {
@@ -1167,7 +1169,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
VFIOBitmap vbmap;
int ret;
- if (!container->dirty_pages_supported && !all_device_dirty_tracking) {
+ if (!container->bcontainer.dirty_pages_supported &&
+ !all_device_dirty_tracking) {
cpu_physical_memory_set_dirty_range(ram_addr, size,
tcg_enabled() ? DIRTY_CLIENTS_ALL :
DIRTY_CLIENTS_NOCODE);
@@ -1182,7 +1185,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
if (all_device_dirty_tracking) {
ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
} else {
- ret = vfio_query_dirty_bitmap(container, &vbmap, iova, size);
+ ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
+ iova, size);
}
if (ret) {
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 99b2957d7b..a7cf517dd2 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -48,11 +48,34 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
+ bool start)
+{
+ /* Fallback to all pages dirty if dirty page sync isn't supported */
+ if (!bcontainer->ops->set_dirty_page_tracking) {
+ return 0;
+ }
+
+ return bcontainer->ops->set_dirty_page_tracking(bcontainer, start);
+}
+
+int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size)
+{
+ if (!bcontainer->ops->query_dirty_bitmap) {
+ return -EINVAL;
+ }
+
+ return bcontainer->ops->query_dirty_bitmap(bcontainer, vbmap, iova, size);
+}
+
void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
const VFIOIOMMUOps *ops)
{
bcontainer->ops = ops;
bcontainer->space = space;
+ bcontainer->dirty_pages_supported = false;
QLIST_INIT(&bcontainer->giommu_list);
}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 761310fa51..6f02ca133e 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -131,7 +131,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
if (!vfio_devices_all_device_dirty_tracking(container) &&
- container->dirty_pages_supported) {
+ container->bcontainer.dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -205,14 +205,17 @@ static int vfio_legacy_dma_map(VFIOContainerBase *bcontainer, hwaddr iova,
return -errno;
}
-int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
+static int vfio_legacy_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
+ bool start)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
int ret;
struct vfio_iommu_type1_dirty_bitmap dirty = {
.argsz = sizeof(dirty),
};
- if (!container->dirty_pages_supported) {
+ if (!bcontainer->dirty_pages_supported) {
return 0;
}
@@ -232,9 +235,12 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
return ret;
}
-int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
- hwaddr iova, hwaddr size)
+static int vfio_legacy_query_dirty_bitmap(VFIOContainerBase *bcontainer,
+ VFIOBitmap *vbmap,
+ hwaddr iova, hwaddr size)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
struct vfio_iommu_type1_dirty_bitmap *dbitmap;
struct vfio_iommu_type1_dirty_bitmap_get *range;
int ret;
@@ -461,7 +467,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
* qemu_real_host_page_size to mark those dirty.
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
- container->dirty_pages_supported = true;
+ container->bcontainer.dirty_pages_supported = true;
container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
container->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
@@ -553,7 +559,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
container->error = NULL;
- container->dirty_pages_supported = false;
container->dma_max_mappings = 0;
container->iova_ranges = NULL;
QLIST_INIT(&container->vrdl_list);
@@ -938,4 +943,6 @@ void vfio_detach_device(VFIODevice *vbasedev)
const VFIOIOMMUOps vfio_legacy_ops = {
.dma_map = vfio_legacy_dma_map,
.dma_unmap = vfio_legacy_dma_unmap,
+ .set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
+ .query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 12/37] vfio/container: Move per container device list in base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (10 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 15:03 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 13/37] vfio/container: Convert functions to " Zhenzhong Duan
` (25 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
VFIO Device is also changed to point to base container instead of
legacy container.
No fucntional change intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
[ clg: context changes ]
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 3 +--
include/hw/vfio/vfio-container-base.h | 1 +
hw/vfio/common.c | 23 +++++++++++++++--------
hw/vfio/container.c | 12 ++++++------
4 files changed, 23 insertions(+), 16 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index d053c61872..f7a84dc8d0 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -90,7 +90,6 @@ typedef struct VFIOContainer {
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
- QLIST_HEAD(, VFIODevice) device_list;
GList *iova_ranges;
} VFIOContainer;
@@ -118,7 +117,7 @@ typedef struct VFIODevice {
QLIST_ENTRY(VFIODevice) container_next;
QLIST_ENTRY(VFIODevice) global_next;
struct VFIOGroup *group;
- VFIOContainer *container;
+ VFIOContainerBase *bcontainer;
char *sysfsdev;
char *name;
DeviceState *dev;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index ea8436a064..f1de1ef120 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -53,6 +53,7 @@ typedef struct VFIOContainerBase {
bool dirty_pages_supported;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
QLIST_ENTRY(VFIOContainerBase) next;
+ QLIST_HEAD(, VFIODevice) device_list;
} VFIOContainerBase;
typedef struct VFIOGuestIOMMU {
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 7d9b87fc67..f8475348ad 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
bool vfio_viommu_preset(VFIODevice *vbasedev)
{
- return vbasedev->container->bcontainer.space->as != &address_space_memory;
+ return vbasedev->bcontainer->space->as != &address_space_memory;
}
static void vfio_set_migration_error(int err)
@@ -179,6 +179,7 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
MigrationState *ms = migrate_get_current();
@@ -187,7 +188,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
return false;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
VFIOMigration *migration = vbasedev->migration;
if (!migration) {
@@ -205,9 +206,10 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (!vbasedev->dirty_pages_supported) {
return false;
}
@@ -222,13 +224,14 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
*/
bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
if (!migration_is_active(migrate_get_current())) {
return false;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
VFIOMigration *migration = vbasedev->migration;
if (!migration) {
@@ -835,12 +838,13 @@ static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
VFIOContainer *container)
{
VFIOPCIDevice *pcidev;
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
Object *owner;
owner = memory_region_owner(section->mr);
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (vbasedev->type != VFIO_DEVICE_TYPE_PCI) {
continue;
}
@@ -941,13 +945,14 @@ static void vfio_devices_dma_logging_stop(VFIOContainer *container)
uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
sizeof(uint64_t))] = {};
struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
feature->argsz = sizeof(buf);
feature->flags = VFIO_DEVICE_FEATURE_SET |
VFIO_DEVICE_FEATURE_DMA_LOGGING_STOP;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (!vbasedev->dirty_tracking) {
continue;
}
@@ -1038,6 +1043,7 @@ static int vfio_devices_dma_logging_start(VFIOContainer *container)
{
struct vfio_device_feature *feature;
VFIODirtyRanges ranges;
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret = 0;
@@ -1048,7 +1054,7 @@ static int vfio_devices_dma_logging_start(VFIOContainer *container)
return -errno;
}
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
if (vbasedev->dirty_tracking) {
continue;
}
@@ -1141,10 +1147,11 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret;
- QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
+ QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
ret = vfio_device_dma_logging_report(vbasedev, iova, size,
vbmap->bitmap);
if (ret) {
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 6f02ca133e..3228dd2109 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -889,7 +889,7 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
int groupid = vfio_device_groupid(vbasedev, errp);
VFIODevice *vbasedev_iter;
VFIOGroup *group;
- VFIOContainer *container;
+ VFIOContainerBase *bcontainer;
int ret;
if (groupid < 0) {
@@ -916,9 +916,9 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
return ret;
}
- container = group->container;
- vbasedev->container = container;
- QLIST_INSERT_HEAD(&container->device_list, vbasedev, container_next);
+ bcontainer = &group->container->bcontainer;
+ vbasedev->bcontainer = bcontainer;
+ QLIST_INSERT_HEAD(&bcontainer->device_list, vbasedev, container_next);
QLIST_INSERT_HEAD(&vfio_device_list, vbasedev, global_next);
return ret;
@@ -928,13 +928,13 @@ void vfio_detach_device(VFIODevice *vbasedev)
{
VFIOGroup *group = vbasedev->group;
- if (!vbasedev->container) {
+ if (!vbasedev->bcontainer) {
return;
}
QLIST_REMOVE(vbasedev, global_next);
QLIST_REMOVE(vbasedev, container_next);
- vbasedev->container = NULL;
+ vbasedev->bcontainer = NULL;
trace_vfio_detach_device(vbasedev->name, group->groupid);
vfio_put_base_device(vbasedev);
vfio_put_group(group);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 13/37] vfio/container: Convert functions to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (11 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 12/37] vfio/container: Move per container device list in base container Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 15:03 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
` (24 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
In the prospect to get rid of VFIOContainer refs
in common.c lets convert misc functions to use the base
container object instead:
vfio_devices_all_dirty_tracking
vfio_devices_all_device_dirty_tracking
vfio_devices_all_running_and_mig_active
vfio_devices_query_dirty_bitmap
vfio_get_dirty_bitmap
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 9 ++++----
hw/vfio/common.c | 42 +++++++++++++++--------------------
hw/vfio/container.c | 6 ++---
hw/vfio/trace-events | 2 +-
4 files changed, 26 insertions(+), 33 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index f7a84dc8d0..fb3c7aea8f 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -186,7 +186,6 @@ typedef struct VFIODisplay {
VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
-bool vfio_devices_all_running_and_saving(VFIOContainer *container);
/* SPAPR specific */
int vfio_container_add_section_window(VFIOContainer *container,
@@ -260,11 +259,11 @@ bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
void vfio_migration_exit(VFIODevice *vbasedev);
int vfio_bitmap_alloc(VFIOBitmap *vbmap, hwaddr size);
-bool vfio_devices_all_running_and_mig_active(VFIOContainer *container);
-bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container);
-int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
+bool vfio_devices_all_running_and_mig_active(VFIOContainerBase *bcontainer);
+bool vfio_devices_all_device_dirty_tracking(VFIOContainerBase *bcontainer);
+int vfio_devices_query_dirty_bitmap(VFIOContainerBase *bcontainer,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size);
-int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOContainerBase *bcontainer, uint64_t iova,
uint64_t size, ram_addr_t ram_addr);
#endif /* HW_VFIO_VFIO_COMMON_H */
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index f8475348ad..91411d9844 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -177,9 +177,8 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
migration->device_state == VFIO_DEVICE_STATE_PRE_COPY_P2P;
}
-static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
+static bool vfio_devices_all_dirty_tracking(VFIOContainerBase *bcontainer)
{
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
MigrationState *ms = migrate_get_current();
@@ -204,9 +203,8 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
return true;
}
-bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
+bool vfio_devices_all_device_dirty_tracking(VFIOContainerBase *bcontainer)
{
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
@@ -222,9 +220,8 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
* Check if all VFIO devices are running and migration is active, which is
* essentially equivalent to the migration being in pre-copy phase.
*/
-bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
+bool vfio_devices_all_running_and_mig_active(VFIOContainerBase *bcontainer)
{
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
if (!migration_is_active(migrate_get_current())) {
@@ -1084,7 +1081,7 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
VFIOContainer *container = container_of(listener, VFIOContainer, listener);
int ret;
- if (vfio_devices_all_device_dirty_tracking(container)) {
+ if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
ret = vfio_devices_dma_logging_start(container);
} else {
ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
@@ -1103,7 +1100,7 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
VFIOContainer *container = container_of(listener, VFIOContainer, listener);
int ret = 0;
- if (vfio_devices_all_device_dirty_tracking(container)) {
+ if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
vfio_devices_dma_logging_stop(container);
} else {
ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
@@ -1143,11 +1140,10 @@ static int vfio_device_dma_logging_report(VFIODevice *vbasedev, hwaddr iova,
return 0;
}
-int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
+int vfio_devices_query_dirty_bitmap(VFIOContainerBase *bcontainer,
VFIOBitmap *vbmap, hwaddr iova,
hwaddr size)
{
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret;
@@ -1167,17 +1163,16 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
return 0;
}
-int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
+int vfio_get_dirty_bitmap(VFIOContainerBase *bcontainer, uint64_t iova,
uint64_t size, ram_addr_t ram_addr)
{
bool all_device_dirty_tracking =
- vfio_devices_all_device_dirty_tracking(container);
+ vfio_devices_all_device_dirty_tracking(bcontainer);
uint64_t dirty_pages;
VFIOBitmap vbmap;
int ret;
- if (!container->bcontainer.dirty_pages_supported &&
- !all_device_dirty_tracking) {
+ if (!bcontainer->dirty_pages_supported && !all_device_dirty_tracking) {
cpu_physical_memory_set_dirty_range(ram_addr, size,
tcg_enabled() ? DIRTY_CLIENTS_ALL :
DIRTY_CLIENTS_NOCODE);
@@ -1190,10 +1185,9 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
}
if (all_device_dirty_tracking) {
- ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
+ ret = vfio_devices_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
} else {
- ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
- iova, size);
+ ret = vfio_container_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
}
if (ret) {
@@ -1203,8 +1197,7 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
dirty_pages = cpu_physical_memory_set_dirty_lebitmap(vbmap.bitmap, ram_addr,
vbmap.pages);
- trace_vfio_get_dirty_bitmap(container->fd, iova, size, vbmap.size,
- ram_addr, dirty_pages);
+ trace_vfio_get_dirty_bitmap(iova, size, vbmap.size, ram_addr, dirty_pages);
out:
g_free(vbmap.bitmap);
@@ -1238,8 +1231,8 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
rcu_read_lock();
if (vfio_get_xlat_addr(iotlb, NULL, &translated_addr, NULL)) {
- ret = vfio_get_dirty_bitmap(container, iova, iotlb->addr_mask + 1,
- translated_addr);
+ ret = vfio_get_dirty_bitmap(&container->bcontainer, iova,
+ iotlb->addr_mask + 1, translated_addr);
if (ret) {
error_report("vfio_iommu_map_dirty_notify(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
@@ -1268,7 +1261,8 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
* Sync the whole mapped region (spanning multiple individual mappings)
* in one go.
*/
- return vfio_get_dirty_bitmap(vrdl->container, iova, size, ram_addr);
+ return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
+ ram_addr);
}
static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
@@ -1337,7 +1331,7 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
ram_addr = memory_region_get_ram_addr(section->mr) +
section->offset_within_region;
- return vfio_get_dirty_bitmap(container,
+ return vfio_get_dirty_bitmap(&container->bcontainer,
REAL_HOST_PAGE_ALIGN(section->offset_within_address_space),
int128_get64(section->size), ram_addr);
}
@@ -1352,7 +1346,7 @@ static void vfio_listener_log_sync(MemoryListener *listener,
return;
}
- if (vfio_devices_all_dirty_tracking(container)) {
+ if (vfio_devices_all_dirty_tracking(&container->bcontainer)) {
ret = vfio_sync_dirty_bitmap(container, section);
if (ret) {
error_report("vfio: Failed to sync dirty bitmap, err: %d (%s)", ret,
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 3228dd2109..8d5b408e86 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -129,8 +129,8 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
bool need_dirty_sync = false;
int ret;
- if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
- if (!vfio_devices_all_device_dirty_tracking(container) &&
+ if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
+ if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
container->bcontainer.dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -162,7 +162,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
}
if (need_dirty_sync) {
- ret = vfio_get_dirty_bitmap(container, iova, size,
+ ret = vfio_get_dirty_bitmap(bcontainer, iova, size,
iotlb->translated_addr);
if (ret) {
return ret;
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 9f7fedee98..08a1f9dfa4 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -117,7 +117,7 @@ vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Devic
vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
vfio_legacy_dma_unmap_overflow_workaround(void) ""
-vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
+vfio_get_dirty_bitmap(uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
# platform.c
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (12 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 13/37] vfio/container: Convert functions to " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 15:52 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 15/37] vfio/container: Move listener " Zhenzhong Duan
` (23 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
Move vrdl_list, pgsizes and dma_max_mappings to the base
container object
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
[ clg: context changes ]
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 13 -------
include/hw/vfio/vfio-container-base.h | 13 +++++++
hw/vfio/common.c | 49 ++++++++++++++-------------
hw/vfio/container-base.c | 12 +++++++
hw/vfio/container.c | 12 +++----
hw/vfio/spapr.c | 18 +++++++---
6 files changed, 68 insertions(+), 49 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index fb3c7aea8f..65ae2d76cf 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -85,24 +85,11 @@ typedef struct VFIOContainer {
bool initialized;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
- unsigned long pgsizes;
- unsigned int dma_max_mappings;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
- QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
GList *iova_ranges;
} VFIOContainer;
-typedef struct VFIORamDiscardListener {
- VFIOContainer *container;
- MemoryRegion *mr;
- hwaddr offset_within_address_space;
- hwaddr size;
- uint64_t granularity;
- RamDiscardListener listener;
- QLIST_ENTRY(VFIORamDiscardListener) next;
-} VFIORamDiscardListener;
-
typedef struct VFIOHostDMAWindow {
hwaddr min_iova;
hwaddr max_iova;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index f1de1ef120..849c8b34b2 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -50,8 +50,11 @@ typedef struct VFIOAddressSpace {
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
VFIOAddressSpace *space;
+ unsigned long pgsizes;
+ unsigned int dma_max_mappings;
bool dirty_pages_supported;
QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
+ QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
QLIST_ENTRY(VFIOContainerBase) next;
QLIST_HEAD(, VFIODevice) device_list;
} VFIOContainerBase;
@@ -64,6 +67,16 @@ typedef struct VFIOGuestIOMMU {
QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
} VFIOGuestIOMMU;
+typedef struct VFIORamDiscardListener {
+ VFIOContainerBase *bcontainer;
+ MemoryRegion *mr;
+ hwaddr offset_within_address_space;
+ hwaddr size;
+ uint64_t granularity;
+ RamDiscardListener listener;
+ QLIST_ENTRY(VFIORamDiscardListener) next;
+} VFIORamDiscardListener;
+
int vfio_container_dma_map(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
void *vaddr, bool readonly);
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 91411d9844..9b34e7e0f8 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -351,13 +351,13 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
{
VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
listener);
+ VFIOContainerBase *bcontainer = vrdl->bcontainer;
const hwaddr size = int128_get64(section->size);
const hwaddr iova = section->offset_within_address_space;
int ret;
/* Unmap with a single call. */
- ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
- iova, size , NULL);
+ ret = vfio_container_dma_unmap(bcontainer, iova, size , NULL);
if (ret) {
error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
strerror(-ret));
@@ -369,6 +369,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
{
VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
listener);
+ VFIOContainerBase *bcontainer = vrdl->bcontainer;
const hwaddr end = section->offset_within_region +
int128_get64(section->size);
hwaddr start, next, iova;
@@ -387,8 +388,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
section->offset_within_address_space;
vaddr = memory_region_get_ram_ptr(section->mr) + start;
- ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
- next - start, vaddr, section->readonly);
+ ret = vfio_container_dma_map(bcontainer, iova, next - start,
+ vaddr, section->readonly);
if (ret) {
/* Rollback */
vfio_ram_discard_notify_discard(rdl, section);
@@ -398,7 +399,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
return 0;
}
-static void vfio_register_ram_discard_listener(VFIOContainer *container,
+static void vfio_register_ram_discard_listener(VFIOContainerBase *bcontainer,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
@@ -411,7 +412,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
g_assert(QEMU_IS_ALIGNED(int128_get64(section->size), TARGET_PAGE_SIZE));
vrdl = g_new0(VFIORamDiscardListener, 1);
- vrdl->container = container;
+ vrdl->bcontainer = bcontainer;
vrdl->mr = section->mr;
vrdl->offset_within_address_space = section->offset_within_address_space;
vrdl->size = int128_get64(section->size);
@@ -419,14 +420,14 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
section->mr);
g_assert(vrdl->granularity && is_power_of_2(vrdl->granularity));
- g_assert(container->pgsizes &&
- vrdl->granularity >= 1ULL << ctz64(container->pgsizes));
+ g_assert(bcontainer->pgsizes &&
+ vrdl->granularity >= 1ULL << ctz64(bcontainer->pgsizes));
ram_discard_listener_init(&vrdl->listener,
vfio_ram_discard_notify_populate,
vfio_ram_discard_notify_discard, true);
ram_discard_manager_register_listener(rdm, &vrdl->listener, section);
- QLIST_INSERT_HEAD(&container->vrdl_list, vrdl, next);
+ QLIST_INSERT_HEAD(&bcontainer->vrdl_list, vrdl, next);
/*
* Sanity-check if we have a theoretically problematic setup where we could
@@ -441,7 +442,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
* number of sections in the address space we could have over time,
* also consuming DMA mappings.
*/
- if (container->dma_max_mappings) {
+ if (bcontainer->dma_max_mappings) {
unsigned int vrdl_count = 0, vrdl_mappings = 0, max_memslots = 512;
#ifdef CONFIG_KVM
@@ -450,7 +451,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
}
#endif
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
hwaddr start, end;
start = QEMU_ALIGN_DOWN(vrdl->offset_within_address_space,
@@ -462,23 +463,23 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
}
if (vrdl_mappings + max_memslots - vrdl_count >
- container->dma_max_mappings) {
+ bcontainer->dma_max_mappings) {
warn_report("%s: possibly running out of DMA mappings. E.g., try"
" increasing the 'block-size' of virtio-mem devies."
" Maximum possible DMA mappings: %d, Maximum possible"
- " memslots: %d", __func__, container->dma_max_mappings,
+ " memslots: %d", __func__, bcontainer->dma_max_mappings,
max_memslots);
}
}
}
-static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
+static void vfio_unregister_ram_discard_listener(VFIOContainerBase *bcontainer,
MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
VFIORamDiscardListener *vrdl = NULL;
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
if (vrdl->mr == section->mr &&
vrdl->offset_within_address_space ==
section->offset_within_address_space) {
@@ -627,7 +628,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
iommu_idx);
ret = memory_region_iommu_set_page_size_mask(giommu->iommu_mr,
- container->pgsizes,
+ bcontainer->pgsizes,
&err);
if (ret) {
g_free(giommu);
@@ -663,7 +664,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
* about changes.
*/
if (memory_region_has_ram_discard_manager(section->mr)) {
- vfio_register_ram_discard_listener(container, section);
+ vfio_register_ram_discard_listener(bcontainer, section);
return;
}
@@ -782,7 +783,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
try_unmap = !((iova & pgmask) || (int128_get64(llsize) & pgmask));
} else if (memory_region_has_ram_discard_manager(section->mr)) {
- vfio_unregister_ram_discard_listener(container, section);
+ vfio_unregister_ram_discard_listener(bcontainer, section);
/* Unregistering will trigger an unmap. */
try_unmap = false;
}
@@ -1261,17 +1262,17 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
* Sync the whole mapped region (spanning multiple individual mappings)
* in one go.
*/
- return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
- ram_addr);
+ return vfio_get_dirty_bitmap(vrdl->bcontainer, iova, size, ram_addr);
}
-static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
- MemoryRegionSection *section)
+static int
+vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section)
{
RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
VFIORamDiscardListener *vrdl = NULL;
- QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
+ QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
if (vrdl->mr == section->mr &&
vrdl->offset_within_address_space ==
section->offset_within_address_space) {
@@ -1325,7 +1326,7 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
}
return 0;
} else if (memory_region_has_ram_discard_manager(section->mr)) {
- return vfio_sync_ram_discard_listener_dirty_bitmap(container, section);
+ return vfio_sync_ram_discard_listener_dirty_bitmap(bcontainer, section);
}
ram_addr = memory_region_get_ram_addr(section->mr) +
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index a7cf517dd2..568f891841 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -76,15 +76,27 @@ void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
bcontainer->ops = ops;
bcontainer->space = space;
bcontainer->dirty_pages_supported = false;
+ bcontainer->dma_max_mappings = 0;
QLIST_INIT(&bcontainer->giommu_list);
+ QLIST_INIT(&bcontainer->vrdl_list);
}
void vfio_container_destroy(VFIOContainerBase *bcontainer)
{
+ VFIORamDiscardListener *vrdl, *vrdl_tmp;
VFIOGuestIOMMU *giommu, *tmp;
QLIST_REMOVE(bcontainer, next);
+ QLIST_FOREACH_SAFE(vrdl, &bcontainer->vrdl_list, next, vrdl_tmp) {
+ RamDiscardManager *rdm;
+
+ rdm = memory_region_get_ram_discard_manager(vrdl->mr);
+ ram_discard_manager_unregister_listener(rdm, &vrdl->listener);
+ QLIST_REMOVE(vrdl, next);
+ g_free(vrdl);
+ }
+
QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
memory_region_unregister_iommu_notifier(
MEMORY_REGION(giommu->iommu_mr), &giommu->n);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 8d5b408e86..0e265ffa67 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -154,7 +154,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
trace_vfio_legacy_dma_unmap_overflow_workaround();
- unmap.size -= 1ULL << ctz64(container->pgsizes);
+ unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
continue;
}
error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
@@ -559,9 +559,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
container->error = NULL;
- container->dma_max_mappings = 0;
container->iova_ranges = NULL;
- QLIST_INIT(&container->vrdl_list);
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, &vfio_legacy_ops);
@@ -589,13 +587,13 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
- container->pgsizes = info->iova_pgsizes;
+ container->bcontainer.pgsizes = info->iova_pgsizes;
} else {
- container->pgsizes = qemu_real_host_page_size();
+ container->bcontainer.pgsizes = qemu_real_host_page_size();
}
- if (!vfio_get_info_dma_avail(info, &container->dma_max_mappings)) {
- container->dma_max_mappings = 65535;
+ if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
+ container->bcontainer.dma_max_mappings = 65535;
}
vfio_get_info_iova_range(info, container);
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 3495737ab2..dbc4c24052 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -223,13 +223,13 @@ static int vfio_spapr_create_window(VFIOContainer *container,
if (pagesize > rampagesize) {
pagesize = rampagesize;
}
- pgmask = container->pgsizes & (pagesize | (pagesize - 1));
+ pgmask = container->bcontainer.pgsizes & (pagesize | (pagesize - 1));
pagesize = pgmask ? (1ULL << (63 - clz64(pgmask))) : 0;
if (!pagesize) {
error_report("Host doesn't support page size 0x%"PRIx64
", the supported mask is 0x%lx",
memory_region_iommu_get_min_page_size(iommu_mr),
- container->pgsizes);
+ container->bcontainer.pgsizes);
return -EINVAL;
}
@@ -385,7 +385,7 @@ void vfio_container_del_section_window(VFIOContainer *container,
bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
{
-
+ VFIOContainerBase *bcontainer = &container->bcontainer;
struct vfio_iommu_spapr_tce_info info;
bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
int ret, fd = container->fd;
@@ -424,7 +424,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
}
if (v2) {
- container->pgsizes = info.ddw.pgsizes;
+ bcontainer->pgsizes = info.ddw.pgsizes;
/*
* There is a default window in just created container.
* To make region_add/del simpler, we better remove this
@@ -439,7 +439,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
}
} else {
/* The default table uses 4K pages */
- container->pgsizes = 0x1000;
+ bcontainer->pgsizes = 0x1000;
vfio_host_win_add(container, info.dma32_window_start,
info.dma32_window_start +
info.dma32_window_size - 1,
@@ -455,7 +455,15 @@ listener_unregister_exit:
void vfio_spapr_container_deinit(VFIOContainer *container)
{
+ VFIOHostDMAWindow *hostwin, *next;
+
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
memory_listener_unregister(&container->prereg_listener);
}
+ QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
+ next) {
+ QLIST_REMOVE(hostwin, hostwin_next);
+ g_free(hostwin);
+ }
+
}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 15/37] vfio/container: Move listener to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (13 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
` (22 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
From: Eric Auger <eric.auger@redhat.com>
Move listener to base container. Also error and initialized fields
are moved at the same time.
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
[ clg: context changes ]
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 3 -
include/hw/vfio/vfio-container-base.h | 3 +
hw/vfio/common.c | 110 +++++++++++++-------------
hw/vfio/container-base.c | 1 +
hw/vfio/container.c | 23 +++---
hw/vfio/spapr.c | 11 +--
6 files changed, 76 insertions(+), 75 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 65ae2d76cf..56452018a9 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -78,11 +78,8 @@ struct VFIOGroup;
typedef struct VFIOContainer {
VFIOContainerBase bcontainer;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
- MemoryListener listener;
MemoryListener prereg_listener;
unsigned iommu_type;
- Error *error;
- bool initialized;
uint64_t dirty_pgsizes;
uint64_t max_dirty_bitmap_size;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 849c8b34b2..89642e6b45 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -50,6 +50,9 @@ typedef struct VFIOAddressSpace {
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
VFIOAddressSpace *space;
+ MemoryListener listener;
+ Error *error;
+ bool initialized;
unsigned long pgsizes;
unsigned int dma_max_mappings;
bool dirty_pages_supported;
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 9b34e7e0f8..52872f515a 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -541,7 +541,7 @@ static bool vfio_listener_valid_section(MemoryRegionSection *section,
return true;
}
-static bool vfio_get_section_iova_range(VFIOContainer *container,
+static bool vfio_get_section_iova_range(VFIOContainerBase *bcontainer,
MemoryRegionSection *section,
hwaddr *out_iova, hwaddr *out_end,
Int128 *out_llend)
@@ -569,8 +569,10 @@ static bool vfio_get_section_iova_range(VFIOContainer *container,
static void vfio_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
- VFIOContainerBase *bcontainer = &container->bcontainer;
+ VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
+ listener);
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -582,7 +584,8 @@ static void vfio_listener_region_add(MemoryListener *listener,
return;
}
- if (!vfio_get_section_iova_range(container, section, &iova, &end, &llend)) {
+ if (!vfio_get_section_iova_range(bcontainer, section, &iova, &end,
+ &llend)) {
if (memory_region_is_ram_device(section->mr)) {
trace_vfio_listener_region_add_no_dma_map(
memory_region_name(section->mr),
@@ -689,13 +692,12 @@ static void vfio_listener_region_add(MemoryListener *listener,
}
}
- ret = vfio_container_dma_map(&container->bcontainer,
- iova, int128_get64(llsize), vaddr,
- section->readonly);
+ ret = vfio_container_dma_map(bcontainer, iova, int128_get64(llsize),
+ vaddr, section->readonly);
if (ret) {
error_setg(&err, "vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx", %p) = %d (%s)",
- container, iova, int128_get64(llsize), vaddr, ret,
+ bcontainer, iova, int128_get64(llsize), vaddr, ret,
strerror(-ret));
if (memory_region_is_ram_device(section->mr)) {
/* Allow unexpected mappings not to be fatal for RAM devices */
@@ -717,9 +719,9 @@ fail:
* can gracefully fail. Runtime, there's not much we can do other
* than throw a hardware error.
*/
- if (!container->initialized) {
- if (!container->error) {
- error_propagate_prepend(&container->error, err,
+ if (!bcontainer->initialized) {
+ if (!bcontainer->error) {
+ error_propagate_prepend(&bcontainer->error, err,
"Region %s: ",
memory_region_name(section->mr));
} else {
@@ -734,8 +736,10 @@ fail:
static void vfio_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
- VFIOContainerBase *bcontainer = &container->bcontainer;
+ VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
+ listener);
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -768,7 +772,8 @@ static void vfio_listener_region_del(MemoryListener *listener,
*/
}
- if (!vfio_get_section_iova_range(container, section, &iova, &end, &llend)) {
+ if (!vfio_get_section_iova_range(bcontainer, section, &iova, &end,
+ &llend)) {
return;
}
@@ -792,22 +797,22 @@ static void vfio_listener_region_del(MemoryListener *listener,
if (int128_eq(llsize, int128_2_64())) {
/* The unmap ioctl doesn't accept a full 64-bit span. */
llsize = int128_rshift(llsize, 1);
- ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ ret = vfio_container_dma_unmap(bcontainer, iova,
int128_get64(llsize), NULL);
if (ret) {
error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, int128_get64(llsize), ret,
+ bcontainer, iova, int128_get64(llsize), ret,
strerror(-ret));
}
iova += int128_get64(llsize);
}
- ret = vfio_container_dma_unmap(&container->bcontainer, iova,
+ ret = vfio_container_dma_unmap(bcontainer, iova,
int128_get64(llsize), NULL);
if (ret) {
error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, int128_get64(llsize), ret,
+ bcontainer, iova, int128_get64(llsize), ret,
strerror(-ret));
}
}
@@ -827,16 +832,15 @@ typedef struct VFIODirtyRanges {
} VFIODirtyRanges;
typedef struct VFIODirtyRangesListener {
- VFIOContainer *container;
+ VFIOContainerBase *bcontainer;
VFIODirtyRanges ranges;
MemoryListener listener;
} VFIODirtyRangesListener;
static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
- VFIOContainer *container)
+ VFIOContainerBase *bcontainer)
{
VFIOPCIDevice *pcidev;
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
Object *owner;
@@ -865,7 +869,7 @@ static void vfio_dirty_tracking_update(MemoryListener *listener,
hwaddr iova, end, *min, *max;
if (!vfio_listener_valid_section(section, "tracking_update") ||
- !vfio_get_section_iova_range(dirty->container, section,
+ !vfio_get_section_iova_range(dirty->bcontainer, section,
&iova, &end, NULL)) {
return;
}
@@ -889,7 +893,7 @@ static void vfio_dirty_tracking_update(MemoryListener *listener,
* The alternative would be an IOVATree but that has a much bigger runtime
* overhead and unnecessary complexity.
*/
- if (vfio_section_is_vfio_pci(section, dirty->container) &&
+ if (vfio_section_is_vfio_pci(section, dirty->bcontainer) &&
iova >= UINT32_MAX) {
min = &range->minpci64;
max = &range->maxpci64;
@@ -913,7 +917,7 @@ static const MemoryListener vfio_dirty_tracking_listener = {
.region_add = vfio_dirty_tracking_update,
};
-static void vfio_dirty_tracking_init(VFIOContainer *container,
+static void vfio_dirty_tracking_init(VFIOContainerBase *bcontainer,
VFIODirtyRanges *ranges)
{
VFIODirtyRangesListener dirty;
@@ -923,10 +927,10 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
dirty.ranges.min64 = UINT64_MAX;
dirty.ranges.minpci64 = UINT64_MAX;
dirty.listener = vfio_dirty_tracking_listener;
- dirty.container = container;
+ dirty.bcontainer = bcontainer;
memory_listener_register(&dirty.listener,
- container->bcontainer.space->as);
+ bcontainer->space->as);
*ranges = dirty.ranges;
@@ -938,12 +942,11 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
memory_listener_unregister(&dirty.listener);
}
-static void vfio_devices_dma_logging_stop(VFIOContainer *container)
+static void vfio_devices_dma_logging_stop(VFIOContainerBase *bcontainer)
{
uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
sizeof(uint64_t))] = {};
struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
feature->argsz = sizeof(buf);
@@ -964,7 +967,7 @@ static void vfio_devices_dma_logging_stop(VFIOContainer *container)
}
static struct vfio_device_feature *
-vfio_device_feature_dma_logging_start_create(VFIOContainer *container,
+vfio_device_feature_dma_logging_start_create(VFIOContainerBase *bcontainer,
VFIODirtyRanges *tracking)
{
struct vfio_device_feature *feature;
@@ -1037,16 +1040,15 @@ static void vfio_device_feature_dma_logging_start_destroy(
g_free(feature);
}
-static int vfio_devices_dma_logging_start(VFIOContainer *container)
+static int vfio_devices_dma_logging_start(VFIOContainerBase *bcontainer)
{
struct vfio_device_feature *feature;
VFIODirtyRanges ranges;
- VFIOContainerBase *bcontainer = &container->bcontainer;
VFIODevice *vbasedev;
int ret = 0;
- vfio_dirty_tracking_init(container, &ranges);
- feature = vfio_device_feature_dma_logging_start_create(container,
+ vfio_dirty_tracking_init(bcontainer, &ranges);
+ feature = vfio_device_feature_dma_logging_start_create(bcontainer,
&ranges);
if (!feature) {
return -errno;
@@ -1069,7 +1071,7 @@ static int vfio_devices_dma_logging_start(VFIOContainer *container)
out:
if (ret) {
- vfio_devices_dma_logging_stop(container);
+ vfio_devices_dma_logging_stop(bcontainer);
}
vfio_device_feature_dma_logging_start_destroy(feature);
@@ -1079,14 +1081,14 @@ out:
static void vfio_listener_log_global_start(MemoryListener *listener)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
+ listener);
int ret;
- if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
- ret = vfio_devices_dma_logging_start(container);
+ if (vfio_devices_all_device_dirty_tracking(bcontainer)) {
+ ret = vfio_devices_dma_logging_start(bcontainer);
} else {
- ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
- true);
+ ret = vfio_container_set_dirty_page_tracking(bcontainer, true);
}
if (ret) {
@@ -1098,14 +1100,14 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
static void vfio_listener_log_global_stop(MemoryListener *listener)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
+ listener);
int ret = 0;
- if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
- vfio_devices_dma_logging_stop(container);
+ if (vfio_devices_all_device_dirty_tracking(bcontainer)) {
+ vfio_devices_dma_logging_stop(bcontainer);
} else {
- ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
- false);
+ ret = vfio_container_set_dirty_page_tracking(bcontainer, false);
}
if (ret) {
@@ -1216,8 +1218,6 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
vfio_giommu_dirty_notifier, n);
VFIOGuestIOMMU *giommu = gdn->giommu;
VFIOContainerBase *bcontainer = giommu->bcontainer;
- VFIOContainer *container = container_of(bcontainer, VFIOContainer,
- bcontainer);
hwaddr iova = iotlb->iova + giommu->iommu_offset;
ram_addr_t translated_addr;
int ret = -EINVAL;
@@ -1232,12 +1232,12 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
rcu_read_lock();
if (vfio_get_xlat_addr(iotlb, NULL, &translated_addr, NULL)) {
- ret = vfio_get_dirty_bitmap(&container->bcontainer, iova,
- iotlb->addr_mask + 1, translated_addr);
+ ret = vfio_get_dirty_bitmap(bcontainer, iova, iotlb->addr_mask + 1,
+ translated_addr);
if (ret) {
error_report("vfio_iommu_map_dirty_notify(%p, 0x%"HWADDR_PRIx", "
"0x%"HWADDR_PRIx") = %d (%s)",
- container, iova, iotlb->addr_mask + 1, ret,
+ bcontainer, iova, iotlb->addr_mask + 1, ret,
strerror(-ret));
}
}
@@ -1293,10 +1293,9 @@ vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainerBase *bcontainer,
&vrdl);
}
-static int vfio_sync_dirty_bitmap(VFIOContainer *container,
+static int vfio_sync_dirty_bitmap(VFIOContainerBase *bcontainer,
MemoryRegionSection *section)
{
- VFIOContainerBase *bcontainer = &container->bcontainer;
ram_addr_t ram_addr;
if (memory_region_is_iommu(section->mr)) {
@@ -1332,7 +1331,7 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
ram_addr = memory_region_get_ram_addr(section->mr) +
section->offset_within_region;
- return vfio_get_dirty_bitmap(&container->bcontainer,
+ return vfio_get_dirty_bitmap(bcontainer,
REAL_HOST_PAGE_ALIGN(section->offset_within_address_space),
int128_get64(section->size), ram_addr);
}
@@ -1340,15 +1339,16 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
static void vfio_listener_log_sync(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer, listener);
+ VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
+ listener);
int ret;
if (vfio_listener_skipped_section(section)) {
return;
}
- if (vfio_devices_all_dirty_tracking(&container->bcontainer)) {
- ret = vfio_sync_dirty_bitmap(container, section);
+ if (vfio_devices_all_dirty_tracking(bcontainer)) {
+ ret = vfio_sync_dirty_bitmap(bcontainer, section);
if (ret) {
error_report("vfio: Failed to sync dirty bitmap, err: %d (%s)", ret,
strerror(-ret));
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 568f891841..b57fdede91 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -75,6 +75,7 @@ void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
{
bcontainer->ops = ops;
bcontainer->space = space;
+ bcontainer->error = NULL;
bcontainer->dirty_pages_supported = false;
bcontainer->dma_max_mappings = 0;
QLIST_INIT(&bcontainer->giommu_list);
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 0e265ffa67..b8f36f56d2 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -453,6 +453,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
{
struct vfio_info_cap_header *hdr;
struct vfio_iommu_type1_info_cap_migration *cap_mig;
+ VFIOContainerBase *bcontainer = &container->bcontainer;
hdr = vfio_get_iommu_info_cap(info, VFIO_IOMMU_TYPE1_INFO_CAP_MIGRATION);
if (!hdr) {
@@ -467,7 +468,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
* qemu_real_host_page_size to mark those dirty.
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
- container->bcontainer.dirty_pages_supported = true;
+ bcontainer->dirty_pages_supported = true;
container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
container->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
@@ -558,7 +559,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
- container->error = NULL;
container->iova_ranges = NULL;
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, &vfio_legacy_ops);
@@ -587,9 +587,9 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
}
if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
- container->bcontainer.pgsizes = info->iova_pgsizes;
+ bcontainer->pgsizes = info->iova_pgsizes;
} else {
- container->bcontainer.pgsizes = qemu_real_host_page_size();
+ bcontainer->pgsizes = qemu_real_host_page_size();
}
if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
@@ -621,26 +621,25 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
group->container = container;
QLIST_INSERT_HEAD(&container->group_list, group, container_next);
- container->listener = vfio_memory_listener;
+ bcontainer->listener = vfio_memory_listener;
- memory_listener_register(&container->listener,
- container->bcontainer.space->as);
+ memory_listener_register(&bcontainer->listener, bcontainer->space->as);
- if (container->error) {
+ if (bcontainer->error) {
ret = -1;
- error_propagate_prepend(errp, container->error,
+ error_propagate_prepend(errp, bcontainer->error,
"memory listener initialization failed: ");
goto listener_release_exit;
}
- container->initialized = true;
+ bcontainer->initialized = true;
return 0;
listener_release_exit:
QLIST_REMOVE(group, container_next);
QLIST_REMOVE(bcontainer, next);
vfio_kvm_device_del_group(group);
- memory_listener_unregister(&container->listener);
+ memory_listener_unregister(&bcontainer->listener);
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
vfio_spapr_container_deinit(container);
@@ -675,7 +674,7 @@ static void vfio_disconnect_container(VFIOGroup *group)
* group.
*/
if (QLIST_EMPTY(&container->group_list)) {
- memory_listener_unregister(&container->listener);
+ memory_listener_unregister(&bcontainer->listener);
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
vfio_spapr_container_deinit(container);
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index dbc4c24052..5786377317 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -46,6 +46,7 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
{
VFIOContainer *container = container_of(listener, VFIOContainer,
prereg_listener);
+ VFIOContainerBase *bcontainer = &container->bcontainer;
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
int ret;
@@ -88,9 +89,9 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
* can gracefully fail. Runtime, there's not much we can do other
* than throw a hardware error.
*/
- if (!container->initialized) {
- if (!container->error) {
- error_setg_errno(&container->error, -ret,
+ if (!bcontainer->initialized) {
+ if (!bcontainer->error) {
+ error_setg_errno(&bcontainer->error, -ret,
"Memory registering failed");
}
} else {
@@ -406,9 +407,9 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
memory_listener_register(&container->prereg_listener,
&address_space_memory);
- if (container->error) {
+ if (bcontainer->error) {
ret = -1;
- error_propagate_prepend(errp, container->error,
+ error_propagate_prepend(errp, bcontainer->error,
"RAM memory listener initialization failed: ");
goto listener_unregister_exit;
}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (14 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 15/37] vfio/container: Move listener " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 16:01 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 17/37] vfio/container: Move iova_ranges " Zhenzhong Duan
` (21 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No functional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 2 --
include/hw/vfio/vfio-container-base.h | 2 ++
hw/vfio/container.c | 11 ++++++-----
3 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 56452018a9..423ab2436c 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -80,8 +80,6 @@ typedef struct VFIOContainer {
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
MemoryListener prereg_listener;
unsigned iommu_type;
- uint64_t dirty_pgsizes;
- uint64_t max_dirty_bitmap_size;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
GList *iova_ranges;
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 89642e6b45..526d23acfd 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -53,6 +53,8 @@ typedef struct VFIOContainerBase {
MemoryListener listener;
Error *error;
bool initialized;
+ uint64_t dirty_pgsizes;
+ uint64_t max_dirty_bitmap_size;
unsigned long pgsizes;
unsigned int dma_max_mappings;
bool dirty_pages_supported;
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index b8f36f56d2..68dc6d240f 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -64,6 +64,7 @@ static int vfio_dma_unmap_bitmap(VFIOContainer *container,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb)
{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
struct vfio_iommu_type1_dma_unmap *unmap;
struct vfio_bitmap *bitmap;
VFIOBitmap vbmap;
@@ -91,7 +92,7 @@ static int vfio_dma_unmap_bitmap(VFIOContainer *container,
bitmap->size = vbmap.size;
bitmap->data = (__u64 *)vbmap.bitmap;
- if (vbmap.size > container->max_dirty_bitmap_size) {
+ if (vbmap.size > bcontainer->max_dirty_bitmap_size) {
error_report("UNMAP: Size of bitmap too big 0x%"PRIx64, vbmap.size);
ret = -E2BIG;
goto unmap_exit;
@@ -131,7 +132,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
- container->bcontainer.dirty_pages_supported) {
+ bcontainer->dirty_pages_supported) {
return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
}
@@ -154,7 +155,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
container->iommu_type == VFIO_TYPE1v2_IOMMU) {
trace_vfio_legacy_dma_unmap_overflow_workaround();
- unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
+ unmap.size -= 1ULL << ctz64(bcontainer->pgsizes);
continue;
}
error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
@@ -469,8 +470,8 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
*/
if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
bcontainer->dirty_pages_supported = true;
- container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
- container->dirty_pgsizes = cap_mig->pgsize_bitmap;
+ bcontainer->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
+ bcontainer->dirty_pgsizes = cap_mig->pgsize_bitmap;
}
}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 17/37] vfio/container: Move iova_ranges to base container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (15 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 18/37] vfio/container: Implement attach/detach_device Zhenzhong Duan
` (20 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
Meanwhile remove the helper function vfio_free_container as it
only calls g_free now.
No functional change intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 1 -
include/hw/vfio/vfio-container-base.h | 1 +
hw/vfio/common.c | 5 +++--
hw/vfio/container-base.c | 3 +++
hw/vfio/container.c | 19 ++++++-------------
5 files changed, 13 insertions(+), 16 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 423ab2436c..938f75e70c 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -82,7 +82,6 @@ typedef struct VFIOContainer {
unsigned iommu_type;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
- GList *iova_ranges;
} VFIOContainer;
typedef struct VFIOHostDMAWindow {
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 526d23acfd..2ffafb0d58 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -62,6 +62,7 @@ typedef struct VFIOContainerBase {
QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
QLIST_ENTRY(VFIOContainerBase) next;
QLIST_HEAD(, VFIODevice) device_list;
+ GList *iova_ranges;
} VFIOContainerBase;
typedef struct VFIOGuestIOMMU {
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 52872f515a..d62c815d7f 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -638,9 +638,10 @@ static void vfio_listener_region_add(MemoryListener *listener,
goto fail;
}
- if (container->iova_ranges) {
+ if (bcontainer->iova_ranges) {
ret = memory_region_iommu_set_iova_ranges(giommu->iommu_mr,
- container->iova_ranges, &err);
+ bcontainer->iova_ranges,
+ &err);
if (ret) {
g_free(giommu);
goto fail;
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index b57fdede91..6363ed3552 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -78,6 +78,7 @@ void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
bcontainer->error = NULL;
bcontainer->dirty_pages_supported = false;
bcontainer->dma_max_mappings = 0;
+ bcontainer->iova_ranges = NULL;
QLIST_INIT(&bcontainer->giommu_list);
QLIST_INIT(&bcontainer->vrdl_list);
}
@@ -104,4 +105,6 @@ void vfio_container_destroy(VFIOContainerBase *bcontainer)
QLIST_REMOVE(giommu, giommu_next);
g_free(giommu);
}
+
+ g_list_free_full(bcontainer->iova_ranges, g_free);
}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 68dc6d240f..36c34683ad 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -308,7 +308,7 @@ bool vfio_get_info_dma_avail(struct vfio_iommu_type1_info *info,
}
static bool vfio_get_info_iova_range(struct vfio_iommu_type1_info *info,
- VFIOContainer *container)
+ VFIOContainerBase *bcontainer)
{
struct vfio_info_cap_header *hdr;
struct vfio_iommu_type1_info_cap_iova_range *cap;
@@ -326,8 +326,8 @@ static bool vfio_get_info_iova_range(struct vfio_iommu_type1_info *info,
range_set_bounds(range, cap->iova_ranges[i].start,
cap->iova_ranges[i].end);
- container->iova_ranges =
- range_list_insert(container->iova_ranges, range);
+ bcontainer->iova_ranges =
+ range_list_insert(bcontainer->iova_ranges, range);
}
return true;
@@ -475,12 +475,6 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
}
}
-static void vfio_free_container(VFIOContainer *container)
-{
- g_list_free_full(container->iova_ranges, g_free);
- g_free(container);
-}
-
static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
Error **errp)
{
@@ -560,7 +554,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container = g_malloc0(sizeof(*container));
container->fd = fd;
- container->iova_ranges = NULL;
bcontainer = &container->bcontainer;
vfio_container_init(bcontainer, space, &vfio_legacy_ops);
@@ -597,7 +590,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
container->bcontainer.dma_max_mappings = 65535;
}
- vfio_get_info_iova_range(info, container);
+ vfio_get_info_iova_range(info, bcontainer);
vfio_get_iommu_info_migration(container, info);
g_free(info);
@@ -650,7 +643,7 @@ enable_discards_exit:
vfio_ram_block_discard_disable(container, false);
free_container_exit:
- vfio_free_container(container);
+ g_free(container);
close_fd_exit:
close(fd);
@@ -694,7 +687,7 @@ static void vfio_disconnect_container(VFIOGroup *group)
trace_vfio_disconnect_container(container->fd);
close(container->fd);
- vfio_free_container(container);
+ g_free(container);
vfio_put_address_space(space);
}
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 18/37] vfio/container: Implement attach/detach_device
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (16 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 17/37] vfio/container: Move iova_ranges " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 16:04 ` Cédric Le Goater
2023-10-27 16:06 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface Zhenzhong Duan
` (19 subsequent siblings)
37 siblings, 2 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
No fucntional change intended.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
hw/vfio/common.c | 16 ++++++++++++++++
hw/vfio/container.c | 12 +++++-------
2 files changed, 21 insertions(+), 7 deletions(-)
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index d62c815d7f..64565b4ae9 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1500,3 +1500,19 @@ retry:
return info;
}
+
+int vfio_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
+{
+ const VFIOIOMMUOps *ops = &vfio_legacy_ops;
+
+ return ops->attach_device(name, vbasedev, as, errp);
+}
+
+void vfio_detach_device(VFIODevice *vbasedev)
+{
+ if (!vbasedev->bcontainer) {
+ return;
+ }
+ vbasedev->bcontainer->ops->detach_device(vbasedev);
+}
diff --git a/hw/vfio/container.c b/hw/vfio/container.c
index 36c34683ad..c8ff0f2037 100644
--- a/hw/vfio/container.c
+++ b/hw/vfio/container.c
@@ -874,8 +874,8 @@ static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
* @name and @vbasedev->name are likely to be different depending
* on the type of the device, hence the need for passing @name
*/
-int vfio_attach_device(char *name, VFIODevice *vbasedev,
- AddressSpace *as, Error **errp)
+static int vfio_legacy_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
{
int groupid = vfio_device_groupid(vbasedev, errp);
VFIODevice *vbasedev_iter;
@@ -915,14 +915,10 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
return ret;
}
-void vfio_detach_device(VFIODevice *vbasedev)
+static void vfio_legacy_detach_device(VFIODevice *vbasedev)
{
VFIOGroup *group = vbasedev->group;
- if (!vbasedev->bcontainer) {
- return;
- }
-
QLIST_REMOVE(vbasedev, global_next);
QLIST_REMOVE(vbasedev, container_next);
vbasedev->bcontainer = NULL;
@@ -934,6 +930,8 @@ void vfio_detach_device(VFIODevice *vbasedev)
const VFIOIOMMUOps vfio_legacy_ops = {
.dma_map = vfio_legacy_dma_map,
.dma_unmap = vfio_legacy_dma_unmap,
+ .attach_device = vfio_legacy_attach_device,
+ .detach_device = vfio_legacy_detach_device,
.set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
.query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (17 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 18/37] vfio/container: Implement attach/detach_device Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-27 16:04 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 20/37] vfio/spapr: switch to spapr IOMMU BE add/del_section_window Zhenzhong Duan
` (18 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
Introduce an empry spapr backend which will hold spapr specific
content, currently only prereg_listener and hostwin_list.
Also introduce and instantiate a spapr specific target interface,
currently only has add/del_window callbacks.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 8 ++++++++
include/hw/vfio/vfio-container-base.h | 2 ++
hw/vfio/spapr.c | 8 ++++++++
3 files changed, 18 insertions(+)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 938f75e70c..a74e60e677 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -169,6 +169,14 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
void vfio_put_address_space(VFIOAddressSpace *space);
/* SPAPR specific */
+typedef struct VFIOIOMMUSpaprOps {
+ int (*add_window)(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp);
+ void (*del_window)(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section);
+} VFIOIOMMUSpaprOps;
+
int vfio_container_add_section_window(VFIOContainer *container,
MemoryRegionSection *section,
Error **errp);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 2ffafb0d58..1e1854d24f 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -31,6 +31,7 @@
typedef struct VFIODevice VFIODevice;
typedef struct VFIOIOMMUOps VFIOIOMMUOps;
+typedef struct VFIOIOMMUSpaprOps VFIOIOMMUSpaprOps;
typedef struct {
unsigned long *bitmap;
@@ -49,6 +50,7 @@ typedef struct VFIOAddressSpace {
*/
typedef struct VFIOContainerBase {
const VFIOIOMMUOps *ops;
+ const VFIOIOMMUSpaprOps *spapr_ops;
VFIOAddressSpace *space;
MemoryListener listener;
Error *error;
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 5786377317..3739004151 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -24,6 +24,10 @@
#include "qapi/error.h"
#include "trace.h"
+typedef struct VFIOSpaprContainer {
+ VFIOContainer container;
+} VFIOSpaprContainer;
+
static bool vfio_prereg_listener_skipped_section(MemoryRegionSection *section)
{
if (memory_region_is_iommu(section->mr)) {
@@ -384,6 +388,8 @@ void vfio_container_del_section_window(VFIOContainer *container,
}
}
+const VFIOIOMMUSpaprOps vfio_iommu_spapr_ops;
+
bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
{
VFIOContainerBase *bcontainer = &container->bcontainer;
@@ -447,6 +453,8 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
0x1000);
}
+ bcontainer->spapr_ops = &vfio_iommu_spapr_ops;
+
listener_unregister_exit:
if (v2) {
memory_listener_unregister(&container->prereg_listener);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 20/37] vfio/spapr: switch to spapr IOMMU BE add/del_section_window
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (18 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 21/37] vfio/spapr: Move prereg_listener into spapr container Zhenzhong Duan
` (17 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
No fucntional change intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 5 -----
include/hw/vfio/vfio-container-base.h | 5 +++++
hw/vfio/common.c | 8 ++------
hw/vfio/container-base.c | 23 ++++++++++++++++++++++-
hw/vfio/spapr.c | 22 ++++++++++++++++------
5 files changed, 45 insertions(+), 18 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index a74e60e677..1489b49d71 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -177,11 +177,6 @@ typedef struct VFIOIOMMUSpaprOps {
MemoryRegionSection *section);
} VFIOIOMMUSpaprOps;
-int vfio_container_add_section_window(VFIOContainer *container,
- MemoryRegionSection *section,
- Error **errp);
-void vfio_container_del_section_window(VFIOContainer *container,
- MemoryRegionSection *section);
bool vfio_spapr_container_init(VFIOContainer *container, Error **errp);
void vfio_spapr_container_deinit(VFIOContainer *container);
diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
index 1e1854d24f..88463bcf24 100644
--- a/include/hw/vfio/vfio-container-base.h
+++ b/include/hw/vfio/vfio-container-base.h
@@ -91,6 +91,11 @@ int vfio_container_dma_map(VFIOContainerBase *bcontainer,
int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
IOMMUTLBEntry *iotlb);
+int vfio_container_add_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp);
+void vfio_container_del_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section);
int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
bool start);
int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 64565b4ae9..41a5609e33 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -571,8 +571,6 @@ static void vfio_listener_region_add(MemoryListener *listener,
{
VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
listener);
- VFIOContainer *container = container_of(bcontainer, VFIOContainer,
- bcontainer);
hwaddr iova, end;
Int128 llend, llsize;
void *vaddr;
@@ -596,7 +594,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
return;
}
- if (vfio_container_add_section_window(container, section, &err)) {
+ if (vfio_container_add_section_window(bcontainer, section, &err)) {
goto fail;
}
@@ -739,8 +737,6 @@ static void vfio_listener_region_del(MemoryListener *listener,
{
VFIOContainerBase *bcontainer = container_of(listener, VFIOContainerBase,
listener);
- VFIOContainer *container = container_of(bcontainer, VFIOContainer,
- bcontainer);
hwaddr iova, end;
Int128 llend, llsize;
int ret;
@@ -820,7 +816,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
memory_region_unref(section->mr);
- vfio_container_del_section_window(container, section);
+ vfio_container_del_section_window(bcontainer, section);
}
typedef struct VFIODirtyRanges {
diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
index 6363ed3552..2efb26cfe4 100644
--- a/hw/vfio/container-base.c
+++ b/hw/vfio/container-base.c
@@ -24,7 +24,7 @@
#include "qemu/osdep.h"
#include "qapi/error.h"
#include "qemu/error-report.h"
-#include "hw/vfio/vfio-container-base.h"
+#include "hw/vfio/vfio-common.h"
int vfio_container_dma_map(VFIOContainerBase *bcontainer,
hwaddr iova, ram_addr_t size,
@@ -48,6 +48,27 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
}
+int vfio_container_add_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp)
+{
+ if (!bcontainer->spapr_ops || !bcontainer->spapr_ops->add_window) {
+ return 0;
+ }
+
+ return bcontainer->spapr_ops->add_window(bcontainer, section, errp);
+}
+
+void vfio_container_del_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section)
+{
+ if (!bcontainer->spapr_ops || !bcontainer->spapr_ops->del_window) {
+ return;
+ }
+
+ return bcontainer->spapr_ops->del_window(bcontainer, section);
+}
+
int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
bool start)
{
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 3739004151..43e32e544b 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -302,10 +302,13 @@ static int vfio_spapr_create_window(VFIOContainer *container,
return 0;
}
-int vfio_container_add_section_window(VFIOContainer *container,
- MemoryRegionSection *section,
- Error **errp)
+static int
+vfio_spapr_container_add_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section,
+ Error **errp)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
VFIOHostDMAWindow *hostwin;
hwaddr pgsize = 0;
int ret;
@@ -370,9 +373,13 @@ int vfio_container_add_section_window(VFIOContainer *container,
return 0;
}
-void vfio_container_del_section_window(VFIOContainer *container,
- MemoryRegionSection *section)
+static void
+vfio_spapr_container_del_section_window(VFIOContainerBase *bcontainer,
+ MemoryRegionSection *section)
{
+ VFIOContainer *container = container_of(bcontainer, VFIOContainer,
+ bcontainer);
+
if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
return;
}
@@ -388,7 +395,10 @@ void vfio_container_del_section_window(VFIOContainer *container,
}
}
-const VFIOIOMMUSpaprOps vfio_iommu_spapr_ops;
+const VFIOIOMMUSpaprOps vfio_iommu_spapr_ops = {
+ .add_window = vfio_spapr_container_add_section_window,
+ .del_window = vfio_spapr_container_del_section_window,
+};
bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
{
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 21/37] vfio/spapr: Move prereg_listener into spapr container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (19 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 20/37] vfio/spapr: switch to spapr IOMMU BE add/del_section_window Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 22/37] vfio/spapr: Move hostwin_list " Zhenzhong Duan
` (16 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
No functional changes intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 1 -
hw/vfio/spapr.c | 24 ++++++++++++++++--------
2 files changed, 16 insertions(+), 9 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 1489b49d71..913e9f5771 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -78,7 +78,6 @@ struct VFIOGroup;
typedef struct VFIOContainer {
VFIOContainerBase bcontainer;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
- MemoryListener prereg_listener;
unsigned iommu_type;
QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 43e32e544b..566a79906c 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -26,6 +26,7 @@
typedef struct VFIOSpaprContainer {
VFIOContainer container;
+ MemoryListener prereg_listener;
} VFIOSpaprContainer;
static bool vfio_prereg_listener_skipped_section(MemoryRegionSection *section)
@@ -48,8 +49,9 @@ static void *vfio_prereg_gpa_to_vaddr(MemoryRegionSection *section, hwaddr gpa)
static void vfio_prereg_listener_region_add(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer,
- prereg_listener);
+ VFIOSpaprContainer *scontainer = container_of(listener, VFIOSpaprContainer,
+ prereg_listener);
+ VFIOContainer *container = &scontainer->container;
VFIOContainerBase *bcontainer = &container->bcontainer;
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
@@ -107,8 +109,9 @@ static void vfio_prereg_listener_region_add(MemoryListener *listener,
static void vfio_prereg_listener_region_del(MemoryListener *listener,
MemoryRegionSection *section)
{
- VFIOContainer *container = container_of(listener, VFIOContainer,
- prereg_listener);
+ VFIOSpaprContainer *scontainer = container_of(listener, VFIOSpaprContainer,
+ prereg_listener);
+ VFIOContainer *container = &scontainer->container;
const hwaddr gpa = section->offset_within_address_space;
hwaddr end;
int ret;
@@ -403,6 +406,8 @@ const VFIOIOMMUSpaprOps vfio_iommu_spapr_ops = {
bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
{
VFIOContainerBase *bcontainer = &container->bcontainer;
+ VFIOSpaprContainer *scontainer = container_of(container, VFIOSpaprContainer,
+ container);
struct vfio_iommu_spapr_tce_info info;
bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
int ret, fd = container->fd;
@@ -419,9 +424,9 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
return -errno;
}
} else {
- container->prereg_listener = vfio_prereg_listener;
+ scontainer->prereg_listener = vfio_prereg_listener;
- memory_listener_register(&container->prereg_listener,
+ memory_listener_register(&scontainer->prereg_listener,
&address_space_memory);
if (bcontainer->error) {
ret = -1;
@@ -467,7 +472,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
listener_unregister_exit:
if (v2) {
- memory_listener_unregister(&container->prereg_listener);
+ memory_listener_unregister(&scontainer->prereg_listener);
}
return ret;
}
@@ -477,7 +482,10 @@ void vfio_spapr_container_deinit(VFIOContainer *container)
VFIOHostDMAWindow *hostwin, *next;
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
- memory_listener_unregister(&container->prereg_listener);
+ VFIOSpaprContainer *scontainer = container_of(container,
+ VFIOSpaprContainer,
+ container);
+ memory_listener_unregister(&scontainer->prereg_listener);
}
QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
next) {
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 22/37] vfio/spapr: Move hostwin_list into spapr container
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (20 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 21/37] vfio/spapr: Move prereg_listener into spapr container Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 23/37] Add iommufd configure option Zhenzhong Duan
` (15 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
No functional changes intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
include/hw/vfio/vfio-common.h | 1 -
hw/vfio/spapr.c | 30 +++++++++++++++++-------------
2 files changed, 17 insertions(+), 14 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 913e9f5771..4f46af1ef4 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -79,7 +79,6 @@ typedef struct VFIOContainer {
VFIOContainerBase bcontainer;
int fd; /* /dev/vfio/vfio, empowered by the attached groups */
unsigned iommu_type;
- QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
QLIST_HEAD(, VFIOGroup) group_list;
} VFIOContainer;
diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
index 566a79906c..70b59e858c 100644
--- a/hw/vfio/spapr.c
+++ b/hw/vfio/spapr.c
@@ -27,6 +27,7 @@
typedef struct VFIOSpaprContainer {
VFIOContainer container;
MemoryListener prereg_listener;
+ QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
} VFIOSpaprContainer;
static bool vfio_prereg_listener_skipped_section(MemoryRegionSection *section)
@@ -154,12 +155,12 @@ static const MemoryListener vfio_prereg_listener = {
.region_del = vfio_prereg_listener_region_del,
};
-static void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
+static void vfio_host_win_add(VFIOSpaprContainer *scontainer, hwaddr min_iova,
hwaddr max_iova, uint64_t iova_pgsizes)
{
VFIOHostDMAWindow *hostwin;
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &scontainer->hostwin_list, hostwin_next) {
if (ranges_overlap(hostwin->min_iova,
hostwin->max_iova - hostwin->min_iova + 1,
min_iova,
@@ -173,15 +174,15 @@ static void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
hostwin->min_iova = min_iova;
hostwin->max_iova = max_iova;
hostwin->iova_pgsizes = iova_pgsizes;
- QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
+ QLIST_INSERT_HEAD(&scontainer->hostwin_list, hostwin, hostwin_next);
}
-static int vfio_host_win_del(VFIOContainer *container,
+static int vfio_host_win_del(VFIOSpaprContainer *scontainer,
hwaddr min_iova, hwaddr max_iova)
{
VFIOHostDMAWindow *hostwin;
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &scontainer->hostwin_list, hostwin_next) {
if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
QLIST_REMOVE(hostwin, hostwin_next);
g_free(hostwin);
@@ -312,6 +313,8 @@ vfio_spapr_container_add_section_window(VFIOContainerBase *bcontainer,
{
VFIOContainer *container = container_of(bcontainer, VFIOContainer,
bcontainer);
+ VFIOSpaprContainer *scontainer = container_of(container, VFIOSpaprContainer,
+ container);
VFIOHostDMAWindow *hostwin;
hwaddr pgsize = 0;
int ret;
@@ -321,7 +324,7 @@ vfio_spapr_container_add_section_window(VFIOContainerBase *bcontainer,
}
/* For now intersections are not allowed, we may relax this later */
- QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
+ QLIST_FOREACH(hostwin, &scontainer->hostwin_list, hostwin_next) {
if (ranges_overlap(hostwin->min_iova,
hostwin->max_iova - hostwin->min_iova + 1,
section->offset_within_address_space,
@@ -343,7 +346,7 @@ vfio_spapr_container_add_section_window(VFIOContainerBase *bcontainer,
return ret;
}
- vfio_host_win_add(container, section->offset_within_address_space,
+ vfio_host_win_add(scontainer, section->offset_within_address_space,
section->offset_within_address_space +
int128_get64(section->size) - 1, pgsize);
#ifdef CONFIG_KVM
@@ -382,6 +385,8 @@ vfio_spapr_container_del_section_window(VFIOContainerBase *bcontainer,
{
VFIOContainer *container = container_of(bcontainer, VFIOContainer,
bcontainer);
+ VFIOSpaprContainer *scontainer = container_of(container, VFIOSpaprContainer,
+ container);
if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
return;
@@ -389,7 +394,7 @@ vfio_spapr_container_del_section_window(VFIOContainerBase *bcontainer,
vfio_spapr_remove_window(container,
section->offset_within_address_space);
- if (vfio_host_win_del(container,
+ if (vfio_host_win_del(scontainer,
section->offset_within_address_space,
section->offset_within_address_space +
int128_get64(section->size) - 1) < 0) {
@@ -462,7 +467,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
} else {
/* The default table uses 4K pages */
bcontainer->pgsizes = 0x1000;
- vfio_host_win_add(container, info.dma32_window_start,
+ vfio_host_win_add(scontainer, info.dma32_window_start,
info.dma32_window_start +
info.dma32_window_size - 1,
0x1000);
@@ -479,15 +484,14 @@ listener_unregister_exit:
void vfio_spapr_container_deinit(VFIOContainer *container)
{
+ VFIOSpaprContainer *scontainer = container_of(container, VFIOSpaprContainer,
+ container);
VFIOHostDMAWindow *hostwin, *next;
if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
- VFIOSpaprContainer *scontainer = container_of(container,
- VFIOSpaprContainer,
- container);
memory_listener_unregister(&scontainer->prereg_listener);
}
- QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
+ QLIST_FOREACH_SAFE(hostwin, &scontainer->hostwin_list, hostwin_next,
next) {
QLIST_REMOVE(hostwin, hostwin_next);
g_free(hostwin);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 23/37] Add iommufd configure option
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (21 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 22/37] vfio/spapr: Move hostwin_list " Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-31 8:58 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
` (14 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Paolo Bonzini, Marc-André Lureau,
Daniel P. Berrangé, Thomas Huth, Philippe Mathieu-Daudé
This adds "--enable-iommufd/--disable-iommufd" to enable or disable
iommufd support, enabled by default.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
meson.build | 6 ++++++
meson_options.txt | 2 ++
scripts/meson-buildoptions.sh | 3 +++
3 files changed, 11 insertions(+)
diff --git a/meson.build b/meson.build
index 4961c82a6b..a8b0edf7b1 100644
--- a/meson.build
+++ b/meson.build
@@ -560,6 +560,10 @@ have_tpm = get_option('tpm') \
.require(targetos != 'windows', error_message: 'TPM emulation only available on POSIX systems') \
.allowed()
+have_iommufd = get_option('iommufd') \
+ .require(targetos == 'linux', error_message: 'iommufd is supported only on Linux') \
+ .allowed()
+
# vhost
have_vhost_user = get_option('vhost_user') \
.disable_auto_if(targetos != 'linux') \
@@ -2133,6 +2137,7 @@ if get_option('tcg').allowed()
endif
config_host_data.set('CONFIG_TPM', have_tpm)
config_host_data.set('CONFIG_TSAN', get_option('tsan'))
+config_host_data.set('CONFIG_IOMMUFD', have_iommufd)
config_host_data.set('CONFIG_USB_LIBUSB', libusb.found())
config_host_data.set('CONFIG_VDE', vde.found())
config_host_data.set('CONFIG_VHOST_NET', have_vhost_net)
@@ -4074,6 +4079,7 @@ summary_info += {'vhost-user-crypto support': have_vhost_user_crypto}
summary_info += {'vhost-user-blk server support': have_vhost_user_blk_server}
summary_info += {'vhost-vdpa support': have_vhost_vdpa}
summary_info += {'build guest agent': have_ga}
+summary_info += {'iommufd support': have_iommufd}
summary(summary_info, bool_yn: true, section: 'Configurable features')
# Compilation information
diff --git a/meson_options.txt b/meson_options.txt
index 3c7398f3c6..91bb958cae 100644
--- a/meson_options.txt
+++ b/meson_options.txt
@@ -109,6 +109,8 @@ option('dbus_display', type: 'feature', value: 'auto',
description: '-display dbus support')
option('tpm', type : 'feature', value : 'auto',
description: 'TPM support')
+option('iommufd', type : 'feature', value : 'auto',
+ description: 'iommufd support')
# Do not enable it by default even for Mingw32, because it doesn't
# work on Wine.
diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
index 7ca4b77eae..1effc46f7d 100644
--- a/scripts/meson-buildoptions.sh
+++ b/scripts/meson-buildoptions.sh
@@ -125,6 +125,7 @@ meson_options_help() {
printf "%s\n" ' guest-agent-msi Build MSI package for the QEMU Guest Agent'
printf "%s\n" ' hvf HVF acceleration support'
printf "%s\n" ' iconv Font glyph conversion support'
+ printf "%s\n" ' iommufd iommufd support'
printf "%s\n" ' jack JACK sound support'
printf "%s\n" ' keyring Linux keyring support'
printf "%s\n" ' kvm KVM acceleration support'
@@ -342,6 +343,8 @@ _meson_option_parse() {
--enable-install-blobs) printf "%s" -Dinstall_blobs=true ;;
--disable-install-blobs) printf "%s" -Dinstall_blobs=false ;;
--interp-prefix=*) quote_sh "-Dinterp_prefix=$2" ;;
+ --enable-iommufd) printf "%s" -Diommufd=enabled ;;
+ --disable-iommufd) printf "%s" -Diommufd=disabled ;;
--enable-jack) printf "%s" -Djack=enabled ;;
--disable-jack) printf "%s" -Djack=disabled ;;
--enable-keyring) printf "%s" -Dkeyring=enabled ;;
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (22 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 23/37] Add iommufd configure option Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 13:27 ` Markus Armbruster
2023-10-26 10:30 ` [PATCH v3 25/37] util/char_dev: Add open_cdev() Zhenzhong Duan
` (13 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Paolo Bonzini, Daniel P. Berrangé,
Eduardo Habkost, Eric Blake, Markus Armbruster
From: Eric Auger <eric.auger@redhat.com>
Introduce an iommufd object which allows the interaction
with the host /dev/iommu device.
The /dev/iommu can have been already pre-opened outside of qemu,
in which case the fd can be passed directly along with the
iommufd object:
This allows the iommufd object to be shared accross several
subsystems (VFIO, VDPA, ...). For example, libvirt would open
the /dev/iommu once.
If no fd is passed along with the iommufd object, the /dev/iommu
is opened by the qemu code.
The CONFIG_IOMMUFD option must be set to compile this new object.
Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
MAINTAINERS | 7 +
qapi/qom.json | 20 +++
include/sysemu/iommufd.h | 46 +++++++
backends/iommufd-stub.c | 59 +++++++++
backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
backends/Kconfig | 4 +
backends/meson.build | 5 +
backends/trace-events | 12 ++
qemu-options.hx | 13 ++
9 files changed, 434 insertions(+)
create mode 100644 include/sysemu/iommufd.h
create mode 100644 backends/iommufd-stub.c
create mode 100644 backends/iommufd.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 7f9912baa0..7aa57ab16f 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2109,6 +2109,13 @@ F: hw/vfio/ap.c
F: docs/system/s390x/vfio-ap.rst
L: qemu-s390x@nongnu.org
+iommufd
+M: Yi Liu <yi.l.liu@intel.com>
+M: Eric Auger <eric.auger@redhat.com>
+S: Supported
+F: backends/iommufd.c
+F: include/sysemu/iommufd.h
+
vhost
M: Michael S. Tsirkin <mst@redhat.com>
S: Supported
diff --git a/qapi/qom.json b/qapi/qom.json
index c53ef978ff..ef0b50f107 100644
--- a/qapi/qom.json
+++ b/qapi/qom.json
@@ -794,6 +794,22 @@
{ 'struct': 'VfioUserServerProperties',
'data': { 'socket': 'SocketAddress', 'device': 'str' } }
+##
+# @IOMMUFDProperties:
+#
+# Properties for iommufd objects.
+#
+# @fd: file descriptor name previously passed via 'getfd' command,
+# which represents a pre-opened /dev/iommu. This allows the
+# iommufd object to be shared accross several subsystems
+# (VFIO, VDPA, ...) and file descriptor to be shared with
+# other process, e.g: DPDK.
+#
+# Since: 8.2
+##
+{ 'struct': 'IOMMUFDProperties',
+ 'data': { '*fd': 'str' } }
+
##
# @RngProperties:
#
@@ -934,6 +950,8 @@
'input-barrier',
{ 'name': 'input-linux',
'if': 'CONFIG_LINUX' },
+ { 'name': 'iommufd',
+ 'if': 'CONFIG_IOMMUFD' },
'iothread',
'main-loop',
{ 'name': 'memory-backend-epc',
@@ -1003,6 +1021,8 @@
'input-barrier': 'InputBarrierProperties',
'input-linux': { 'type': 'InputLinuxProperties',
'if': 'CONFIG_LINUX' },
+ 'iommufd': { 'type': 'IOMMUFDProperties',
+ 'if': 'CONFIG_IOMMUFD' },
'iothread': 'IothreadProperties',
'main-loop': 'MainLoopProperties',
'memory-backend-epc': { 'type': 'MemoryBackendEpcProperties',
diff --git a/include/sysemu/iommufd.h b/include/sysemu/iommufd.h
new file mode 100644
index 0000000000..f0e5c7eeb8
--- /dev/null
+++ b/include/sysemu/iommufd.h
@@ -0,0 +1,46 @@
+#ifndef SYSEMU_IOMMUFD_H
+#define SYSEMU_IOMMUFD_H
+
+#include "qom/object.h"
+#include "qemu/thread.h"
+#include "exec/hwaddr.h"
+#include "exec/cpu-common.h"
+
+#define TYPE_IOMMUFD_BACKEND "iommufd"
+OBJECT_DECLARE_TYPE(IOMMUFDBackend, IOMMUFDBackendClass,
+ IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND(obj) \
+ OBJECT_CHECK(IOMMUFDBackend, (obj), TYPE_IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND_GET_CLASS(obj) \
+ OBJECT_GET_CLASS(IOMMUFDBackendClass, (obj), TYPE_IOMMUFD_BACKEND)
+#define IOMMUFD_BACKEND_CLASS(klass) \
+ OBJECT_CLASS_CHECK(IOMMUFDBackendClass, (klass), TYPE_IOMMUFD_BACKEND)
+struct IOMMUFDBackendClass {
+ ObjectClass parent_class;
+};
+
+struct IOMMUFDBackend {
+ Object parent;
+
+ /*< protected >*/
+ int fd; /* /dev/iommu file descriptor */
+ bool owned; /* is the /dev/iommu opened internally */
+ QemuMutex lock;
+ uint32_t users;
+
+ /*< public >*/
+};
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp);
+void iommufd_backend_disconnect(IOMMUFDBackend *be);
+
+int iommufd_backend_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id);
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id);
+void iommufd_backend_free_id(int fd, uint32_t id);
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly);
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size);
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt);
+#endif
diff --git a/backends/iommufd-stub.c b/backends/iommufd-stub.c
new file mode 100644
index 0000000000..02ac844c17
--- /dev/null
+++ b/backends/iommufd-stub.c
@@ -0,0 +1,59 @@
+/*
+ * iommufd container backend stub
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "sysemu/iommufd.h"
+#include "qemu/error-report.h"
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp)
+{
+ return 0;
+}
+void iommufd_backend_disconnect(IOMMUFDBackend *be)
+{
+}
+void iommufd_backend_free_id(int fd, uint32_t id)
+{
+}
+int iommufd_backend_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id)
+{
+ return 0;
+}
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id)
+{
+}
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ return 0;
+}
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size)
+{
+ return 0;
+}
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt)
+{
+ return 0;
+}
diff --git a/backends/iommufd.c b/backends/iommufd.c
new file mode 100644
index 0000000000..3f0ed37847
--- /dev/null
+++ b/backends/iommufd.c
@@ -0,0 +1,268 @@
+/*
+ * iommufd container backend
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include "sysemu/iommufd.h"
+#include "qapi/error.h"
+#include "qapi/qmp/qerror.h"
+#include "qemu/module.h"
+#include "qom/object_interfaces.h"
+#include "qemu/error-report.h"
+#include "monitor/monitor.h"
+#include "trace.h"
+#include <sys/ioctl.h>
+#include <linux/iommufd.h>
+
+static void iommufd_backend_init(Object *obj)
+{
+ IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+
+ be->fd = -1;
+ be->users = 0;
+ be->owned = true;
+ qemu_mutex_init(&be->lock);
+}
+
+static void iommufd_backend_finalize(Object *obj)
+{
+ IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+
+ if (be->owned) {
+ close(be->fd);
+ be->fd = -1;
+ }
+}
+
+static void iommufd_backend_set_fd(Object *obj, const char *str, Error **errp)
+{
+ IOMMUFDBackend *be = IOMMUFD_BACKEND(obj);
+ int fd = -1;
+
+ fd = monitor_fd_param(monitor_cur(), str, errp);
+ if (fd == -1) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ qemu_mutex_lock(&be->lock);
+ be->fd = fd;
+ be->owned = false;
+ qemu_mutex_unlock(&be->lock);
+ trace_iommu_backend_set_fd(be->fd);
+}
+
+static void iommufd_backend_class_init(ObjectClass *oc, void *data)
+{
+ object_class_property_add_str(oc, "fd", NULL, iommufd_backend_set_fd);
+}
+
+int iommufd_backend_connect(IOMMUFDBackend *be, Error **errp)
+{
+ int fd, ret = 0;
+
+ qemu_mutex_lock(&be->lock);
+ if (be->users == UINT32_MAX) {
+ error_setg(errp, "too many connections");
+ ret = -E2BIG;
+ goto out;
+ }
+ if (be->owned && !be->users) {
+ fd = qemu_open_old("/dev/iommu", O_RDWR);
+ if (fd < 0) {
+ error_setg_errno(errp, errno, "/dev/iommu opening failed");
+ ret = fd;
+ goto out;
+ }
+ be->fd = fd;
+ }
+ be->users++;
+out:
+ trace_iommufd_backend_connect(be->fd, be->owned,
+ be->users, ret);
+ qemu_mutex_unlock(&be->lock);
+ return ret;
+}
+
+void iommufd_backend_disconnect(IOMMUFDBackend *be)
+{
+ qemu_mutex_lock(&be->lock);
+ if (!be->users) {
+ goto out;
+ }
+ be->users--;
+ if (!be->users && be->owned) {
+ close(be->fd);
+ be->fd = -1;
+ }
+out:
+ trace_iommufd_backend_disconnect(be->fd, be->users);
+ qemu_mutex_unlock(&be->lock);
+}
+
+static int iommufd_backend_alloc_ioas(int fd, uint32_t *ioas_id)
+{
+ int ret;
+ struct iommu_ioas_alloc alloc_data = {
+ .size = sizeof(alloc_data),
+ .flags = 0,
+ };
+
+ ret = ioctl(fd, IOMMU_IOAS_ALLOC, &alloc_data);
+ if (ret) {
+ error_report("Failed to allocate ioas %m");
+ }
+
+ *ioas_id = alloc_data.out_ioas_id;
+ trace_iommufd_backend_alloc_ioas(fd, *ioas_id, ret);
+
+ return ret;
+}
+
+void iommufd_backend_free_id(int fd, uint32_t id)
+{
+ int ret;
+ struct iommu_destroy des = {
+ .size = sizeof(des),
+ .id = id,
+ };
+
+ ret = ioctl(fd, IOMMU_DESTROY, &des);
+ trace_iommufd_backend_free_id(fd, id, ret);
+ if (ret) {
+ error_report("Failed to free id: %u %m", id);
+ }
+}
+
+int iommufd_backend_get_ioas(IOMMUFDBackend *be, uint32_t *ioas_id)
+{
+ int ret;
+
+ ret = iommufd_backend_alloc_ioas(be->fd, ioas_id);
+ trace_iommufd_backend_get_ioas(be->fd, *ioas_id, ret);
+ return ret;
+}
+
+void iommufd_backend_put_ioas(IOMMUFDBackend *be, uint32_t ioas_id)
+{
+ iommufd_backend_free_id(be->fd, ioas_id);
+ trace_iommufd_backend_put_ioas(be->fd, ioas_id);
+}
+
+int iommufd_backend_map_dma(IOMMUFDBackend *be, uint32_t ioas_id, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ int ret;
+ struct iommu_ioas_map map = {
+ .size = sizeof(map),
+ .flags = IOMMU_IOAS_MAP_READABLE |
+ IOMMU_IOAS_MAP_FIXED_IOVA,
+ .ioas_id = ioas_id,
+ .__reserved = 0,
+ .user_va = (uintptr_t)vaddr,
+ .iova = iova,
+ .length = size,
+ };
+
+ if (!readonly) {
+ map.flags |= IOMMU_IOAS_MAP_WRITEABLE;
+ }
+
+ ret = ioctl(be->fd, IOMMU_IOAS_MAP, &map);
+ trace_iommufd_backend_map_dma(be->fd, ioas_id, iova, size,
+ vaddr, readonly, ret);
+ if (ret) {
+ error_report("IOMMU_IOAS_MAP failed: %m");
+ }
+ return !ret ? 0 : -errno;
+}
+
+int iommufd_backend_unmap_dma(IOMMUFDBackend *be, uint32_t ioas_id,
+ hwaddr iova, ram_addr_t size)
+{
+ int ret;
+ struct iommu_ioas_unmap unmap = {
+ .size = sizeof(unmap),
+ .ioas_id = ioas_id,
+ .iova = iova,
+ .length = size,
+ };
+
+ ret = ioctl(be->fd, IOMMU_IOAS_UNMAP, &unmap);
+ trace_iommufd_backend_unmap_dma(be->fd, ioas_id, iova, size, ret);
+ /*
+ * TODO: IOMMUFD doesn't support mapping PCI BARs for now.
+ * It's not a problem if there is no p2p dma, relax it here
+ * and avoid many noisy trigger from vIOMMU side.
+ */
+ if (ret && errno == ENOENT) {
+ ret = 0;
+ }
+ if (ret) {
+ error_report("IOMMU_IOAS_UNMAP failed: %m");
+ }
+ return !ret ? 0 : -errno;
+}
+
+int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
+ uint32_t pt_id, uint32_t *out_hwpt)
+{
+ int ret;
+ struct iommu_hwpt_alloc alloc_hwpt = {
+ .size = sizeof(struct iommu_hwpt_alloc),
+ .flags = 0,
+ .dev_id = dev_id,
+ .pt_id = pt_id,
+ .__reserved = 0,
+ };
+
+ ret = ioctl(iommufd, IOMMU_HWPT_ALLOC, &alloc_hwpt);
+ trace_iommufd_backend_alloc_hwpt(iommufd, dev_id, pt_id,
+ alloc_hwpt.out_hwpt_id, ret);
+
+ if (ret) {
+ error_report("IOMMU_HWPT_ALLOC failed: %m");
+ } else {
+ *out_hwpt = alloc_hwpt.out_hwpt_id;
+ }
+ return !ret ? 0 : -errno;
+}
+
+static const TypeInfo iommufd_backend_info = {
+ .name = TYPE_IOMMUFD_BACKEND,
+ .parent = TYPE_OBJECT,
+ .instance_size = sizeof(IOMMUFDBackend),
+ .instance_init = iommufd_backend_init,
+ .instance_finalize = iommufd_backend_finalize,
+ .class_size = sizeof(IOMMUFDBackendClass),
+ .class_init = iommufd_backend_class_init,
+ .interfaces = (InterfaceInfo[]) {
+ { TYPE_USER_CREATABLE },
+ { }
+ }
+};
+
+static void register_types(void)
+{
+ type_register_static(&iommufd_backend_info);
+}
+
+type_init(register_types);
diff --git a/backends/Kconfig b/backends/Kconfig
index f35abc1609..2cb23f62fa 100644
--- a/backends/Kconfig
+++ b/backends/Kconfig
@@ -1 +1,5 @@
source tpm/Kconfig
+
+config IOMMUFD
+ bool
+ depends on VFIO
diff --git a/backends/meson.build b/backends/meson.build
index 914c7c4afb..05ac57ff15 100644
--- a/backends/meson.build
+++ b/backends/meson.build
@@ -20,6 +20,11 @@ if have_vhost_user
system_ss.add(when: 'CONFIG_VIRTIO', if_true: files('vhost-user.c'))
endif
system_ss.add(when: 'CONFIG_VIRTIO_CRYPTO', if_true: files('cryptodev-vhost.c'))
+if have_iommufd
+ system_ss.add(files('iommufd.c'))
+else
+ system_ss.add(files('iommufd-stub.c'))
+endif
if have_vhost_user_crypto
system_ss.add(when: 'CONFIG_VIRTIO_CRYPTO', if_true: files('cryptodev-vhost-user.c'))
endif
diff --git a/backends/trace-events b/backends/trace-events
index 652eb76a57..e5f828bca2 100644
--- a/backends/trace-events
+++ b/backends/trace-events
@@ -5,3 +5,15 @@ dbus_vmstate_pre_save(void)
dbus_vmstate_post_load(int version_id) "version_id: %d"
dbus_vmstate_loading(const char *id) "id: %s"
dbus_vmstate_saving(const char *id) "id: %s"
+
+# iommufd.c
+iommufd_backend_connect(int fd, bool owned, uint32_t users, int ret) "fd=%d owned=%d users=%d (%d)"
+iommufd_backend_disconnect(int fd, uint32_t users) "fd=%d users=%d"
+iommu_backend_set_fd(int fd) "pre-opened /dev/iommu fd=%d"
+iommufd_backend_get_ioas(int iommufd, uint32_t ioas, int ret) " iommufd=%d ioas=%d (%d)"
+iommufd_backend_put_ioas(int iommufd, uint32_t ioas) " iommufd=%d ioas=%d"
+iommufd_backend_map_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, void *vaddr, bool readonly, int ret) " iommufd=%d ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" addr=%p readonly=%d (%d)"
+iommufd_backend_unmap_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t size, int ret) " iommufd=%d ioas=%d iova=0x%"PRIx64" size=0x%"PRIx64" (%d)"
+iommufd_backend_alloc_ioas(int iommufd, uint32_t ioas, int ret) " iommufd=%d ioas=%d (%d)"
+iommufd_backend_free_id(int iommufd, uint32_t id, int ret) " iommufd=%d id=%d (%d)"
+iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id, uint32_t pt_id, uint32_t out_hwpt_id, int ret) " iommufd=%d dev_id=%u pt_id=%u out_hwpt=%u (%d)"
diff --git a/qemu-options.hx b/qemu-options.hx
index e26230bac5..ddfaddf8ce 100644
--- a/qemu-options.hx
+++ b/qemu-options.hx
@@ -5210,6 +5210,19 @@ SRST
The ``share`` boolean option is on by default with memfd.
+#ifdef CONFIG_IOMMUFD
+ ``-object iommufd,id=id[,fd=fd]``
+ Creates an iommufd backend which allows control of DMA mapping
+ through the /dev/iommu device.
+
+ The ``id`` parameter is a unique ID which frontends (such as
+ vfio-pci of vdpa) will use to connect with the iommufd backend.
+
+ The ``fd`` parameter is an optional pre-opened file descriptor
+ resulting from /dev/iommu opening. Usually the iommufd is shared
+ across all subsystems, bringing the benefit of centralized
+ reference counting.
+#endif
``-object rng-builtin,id=id``
Creates a random number generator backend which obtains entropy
from QEMU builtin functions. The ``id`` parameter is a unique ID
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 25/37] util/char_dev: Add open_cdev()
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (23 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-30 14:53 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 26/37] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
` (12 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Yi Liu <yi.l.liu@intel.com>
/dev/vfio/devices/vfioX may not exist. In that case it is still possible
to open /dev/char/$major:$minor instead. Add helper function to abstract
the cdev open.
Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
MAINTAINERS | 6 +++
include/qemu/chardev_open.h | 16 ++++++++
util/chardev_open.c | 81 +++++++++++++++++++++++++++++++++++++
util/meson.build | 1 +
4 files changed, 104 insertions(+)
create mode 100644 include/qemu/chardev_open.h
create mode 100644 util/chardev_open.c
diff --git a/MAINTAINERS b/MAINTAINERS
index 7aa57ab16f..123e999bee 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3447,6 +3447,12 @@ S: Maintained
F: include/qemu/iova-tree.h
F: util/iova-tree.c
+cdev Open
+M: Yi Liu <yi.l.liu@intel.com>
+S: Maintained
+F: include/qemu/chardev_open.h
+F: util/chardev_open.c
+
elf2dmp
M: Viktor Prutyanov <viktor.prutyanov@phystech.edu>
S: Maintained
diff --git a/include/qemu/chardev_open.h b/include/qemu/chardev_open.h
new file mode 100644
index 0000000000..6580d351c6
--- /dev/null
+++ b/include/qemu/chardev_open.h
@@ -0,0 +1,16 @@
+/*
+ * QEMU Chardev Helper
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2. See
+ * the COPYING file in the top-level directory.
+ */
+
+#ifndef QEMU_CHARDEV_HELPERS_H
+#define QEMU_CHARDEV_HELPERS_H
+
+int open_cdev(const char *devpath, dev_t cdev);
+#endif
diff --git a/util/chardev_open.c b/util/chardev_open.c
new file mode 100644
index 0000000000..005d2b81bd
--- /dev/null
+++ b/util/chardev_open.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) 2019, Mellanox Technologies. All rights reserved.
+ * Copyright (C) 2023 Intel Corporation.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ *
+ * Copied from
+ * https://github.com/linux-rdma/rdma-core/blob/master/util/open_cdev.c
+ *
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/chardev_open.h"
+
+static int open_cdev_internal(const char *path, dev_t cdev)
+{
+ struct stat st;
+ int fd;
+
+ fd = qemu_open_old(path, O_RDWR);
+ if (fd == -1) {
+ return -1;
+ }
+ if (fstat(fd, &st) || !S_ISCHR(st.st_mode) ||
+ (cdev != 0 && st.st_rdev != cdev)) {
+ close(fd);
+ return -1;
+ }
+ return fd;
+}
+
+static int open_cdev_robust(dev_t cdev)
+{
+ g_autofree char *devpath;
+
+ /*
+ * This assumes that udev is being used and is creating the /dev/char/
+ * symlinks.
+ */
+ devpath = g_strdup_printf("/dev/char/%u:%u", major(cdev), minor(cdev));
+ return open_cdev_internal(devpath, cdev);
+}
+
+int open_cdev(const char *devpath, dev_t cdev)
+{
+ int fd;
+
+ fd = open_cdev_internal(devpath, cdev);
+ if (fd == -1 && cdev != 0) {
+ return open_cdev_robust(cdev);
+ }
+ return fd;
+}
diff --git a/util/meson.build b/util/meson.build
index eb677b40c2..eda0b06062 100644
--- a/util/meson.build
+++ b/util/meson.build
@@ -107,6 +107,7 @@ if have_block
util_ss.add(files('filemonitor-stub.c'))
endif
util_ss.add(when: 'CONFIG_LINUX', if_true: files('vfio-helpers.c'))
+ util_ss.add(when: 'CONFIG_LINUX', if_true: files('chardev_open.c'))
endif
if cpu == 'aarch64'
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 26/37] vfio/iommufd: Implement the iommufd backend
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (24 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 25/37] util/char_dev: Add open_cdev() Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation Zhenzhong Duan
` (11 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Yi Liu <yi.l.liu@intel.com>
Add the iommufd backend. The IOMMUFD container class is implemented
based on the new /dev/iommu user API. This backend obviously depends
on CONFIG_IOMMUFD.
So far, the iommufd backend doesn't support dirty page sync yet due
to missing support in the host kernel.
Co-authored-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 23 ++
hw/vfio/common.c | 19 +-
hw/vfio/iommufd.c | 460 ++++++++++++++++++++++++++++++++++
hw/vfio/meson.build | 3 +
hw/vfio/trace-events | 12 +
5 files changed, 513 insertions(+), 4 deletions(-)
create mode 100644 hw/vfio/iommufd.c
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 4f46af1ef4..95afee7243 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -89,6 +89,23 @@ typedef struct VFIOHostDMAWindow {
QLIST_ENTRY(VFIOHostDMAWindow) hostwin_next;
} VFIOHostDMAWindow;
+#ifdef CONFIG_IOMMUFD
+typedef struct VFIOIOASHwpt {
+ uint32_t hwpt_id;
+ QLIST_HEAD(, VFIODevice) device_list;
+ QLIST_ENTRY(VFIOIOASHwpt) next;
+} VFIOIOASHwpt;
+
+typedef struct IOMMUFDBackend IOMMUFDBackend;
+
+typedef struct VFIOIOMMUFDContainer {
+ VFIOContainerBase bcontainer;
+ IOMMUFDBackend *be;
+ uint32_t ioas_id;
+ QLIST_HEAD(, VFIOIOASHwpt) hwpt_list;
+} VFIOIOMMUFDContainer;
+#endif
+
typedef struct VFIODeviceOps VFIODeviceOps;
typedef struct VFIODevice {
@@ -116,6 +133,11 @@ typedef struct VFIODevice {
OnOffAuto pre_copy_dirty_page_tracking;
bool dirty_pages_supported;
bool dirty_tracking;
+#ifdef CONFIG_IOMMUFD
+ int devid;
+ VFIOIOASHwpt *hwpt;
+ IOMMUFDBackend *iommufd;
+#endif
} VFIODevice;
struct VFIODeviceOps {
@@ -209,6 +231,7 @@ typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
extern VFIOGroupList vfio_group_list;
extern VFIODeviceList vfio_device_list;
extern const VFIOIOMMUOps vfio_legacy_ops;
+extern const VFIOIOMMUOps vfio_iommufd_ops;
extern const MemoryListener vfio_memory_listener;
extern int vfio_kvm_device_fd;
diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 41a5609e33..6f0534d832 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -1464,10 +1464,13 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as)
void vfio_put_address_space(VFIOAddressSpace *space)
{
- if (QLIST_EMPTY(&space->containers)) {
- QLIST_REMOVE(space, list);
- g_free(space);
+ if (!QLIST_EMPTY(&space->containers)) {
+ return;
}
+
+ QLIST_REMOVE(space, list);
+ g_free(space);
+
if (QLIST_EMPTY(&vfio_address_spaces)) {
qemu_unregister_reset(vfio_reset_handler, NULL);
}
@@ -1500,8 +1503,16 @@ retry:
int vfio_attach_device(char *name, VFIODevice *vbasedev,
AddressSpace *as, Error **errp)
{
- const VFIOIOMMUOps *ops = &vfio_legacy_ops;
+ const VFIOIOMMUOps *ops;
+#ifdef CONFIG_IOMMUFD
+ if (vbasedev->iommufd) {
+ ops = &vfio_iommufd_ops;
+ } else
+#endif
+ {
+ ops = &vfio_legacy_ops;
+ }
return ops->attach_device(name, vbasedev, as, errp);
}
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
new file mode 100644
index 0000000000..aee64d63f3
--- /dev/null
+++ b/hw/vfio/iommufd.c
@@ -0,0 +1,460 @@
+/*
+ * iommufd container backend
+ *
+ * Copyright (C) 2023 Intel Corporation.
+ * Copyright Red Hat, Inc. 2023
+ *
+ * Authors: Yi Liu <yi.l.liu@intel.com>
+ * Eric Auger <eric.auger@redhat.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "qemu/osdep.h"
+#include <sys/ioctl.h>
+#include <linux/vfio.h>
+#include <linux/iommufd.h>
+
+#include "hw/vfio/vfio-common.h"
+#include "qemu/error-report.h"
+#include "trace.h"
+#include "qapi/error.h"
+#include "sysemu/iommufd.h"
+#include "hw/qdev-core.h"
+#include "sysemu/reset.h"
+#include "qemu/cutils.h"
+#include "qemu/chardev_open.h"
+
+static int iommufd_map(VFIOContainerBase *bcontainer, hwaddr iova,
+ ram_addr_t size, void *vaddr, bool readonly)
+{
+ VFIOIOMMUFDContainer *container =
+ container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+
+ return iommufd_backend_map_dma(container->be,
+ container->ioas_id,
+ iova, size, vaddr, readonly);
+}
+
+static int iommufd_unmap(VFIOContainerBase *bcontainer,
+ hwaddr iova, ram_addr_t size,
+ IOMMUTLBEntry *iotlb)
+{
+ VFIOIOMMUFDContainer *container =
+ container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+
+ /* TODO: Handle dma_unmap_bitmap with iotlb args (migration) */
+ return iommufd_backend_unmap_dma(container->be,
+ container->ioas_id, iova, size);
+}
+
+static void vfio_kvm_device_add_device(VFIODevice *vbasedev)
+{
+ Error *err = NULL;
+
+ if (vfio_kvm_device_add_fd(vbasedev->fd, &err)) {
+ error_report_err(err);
+ }
+}
+
+static void vfio_kvm_device_del_device(VFIODevice *vbasedev)
+{
+ Error *err = NULL;
+
+ if (vfio_kvm_device_del_fd(vbasedev->fd, &err)) {
+ error_report_err(err);
+ }
+}
+
+static int iommufd_connect_and_bind(VFIODevice *vbasedev, Error **errp)
+{
+ IOMMUFDBackend *iommufd = vbasedev->iommufd;
+ struct vfio_device_bind_iommufd bind = {
+ .argsz = sizeof(bind),
+ .flags = 0,
+ };
+ int ret;
+
+ ret = iommufd_backend_connect(iommufd, errp);
+ if (ret) {
+ return ret;
+ }
+
+ /*
+ * Add device to kvm-vfio to be prepared for the tracking
+ * in KVM. Especially for some emulated devices, it requires
+ * to have kvm information in the device open.
+ */
+ vfio_kvm_device_add_device(vbasedev);
+
+ /* Bind device to iommufd */
+ bind.iommufd = iommufd->fd;
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_BIND_IOMMUFD, &bind);
+ if (ret) {
+ error_setg_errno(errp, errno, "error bind device fd=%d to iommufd=%d",
+ vbasedev->fd, bind.iommufd);
+ goto err_bind;
+ }
+
+ vbasedev->devid = bind.out_devid;
+ trace_vfio_iommufd_bind_device(bind.iommufd, vbasedev->name,
+ vbasedev->fd, vbasedev->devid);
+ return ret;
+err_bind:
+ vfio_kvm_device_del_device(vbasedev);
+ iommufd_backend_disconnect(iommufd);
+ return ret;
+}
+
+static void iommufd_unbind_and_disconnect(VFIODevice *vbasedev)
+{
+ /* Unbind is automatically conducted when device fd is closed */
+ vfio_kvm_device_del_device(vbasedev);
+ iommufd_backend_disconnect(vbasedev->iommufd);
+}
+
+static int vfio_get_devicefd(const char *sysfs_path, Error **errp)
+{
+ long int ret = -ENOTTY;
+ char *path, *vfio_dev_path = NULL, *vfio_path = NULL;
+ DIR *dir = NULL;
+ struct dirent *dent;
+ gchar *contents;
+ struct stat st;
+ gsize length;
+ int major, minor;
+ dev_t vfio_devt;
+
+ path = g_strdup_printf("%s/vfio-dev", sysfs_path);
+ if (stat(path, &st) < 0) {
+ error_setg_errno(errp, errno, "no such host device");
+ goto out_free_path;
+ }
+
+ dir = opendir(path);
+ if (!dir) {
+ error_setg_errno(errp, errno, "couldn't open dirrectory %s", path);
+ goto out_free_path;
+ }
+
+ while ((dent = readdir(dir))) {
+ if (!strncmp(dent->d_name, "vfio", 4)) {
+ vfio_dev_path = g_strdup_printf("%s/%s/dev", path, dent->d_name);
+ break;
+ }
+ }
+
+ if (!vfio_dev_path) {
+ error_setg(errp, "failed to find vfio-dev/vfioX/dev");
+ goto out_close_dir;
+ }
+
+ if (!g_file_get_contents(vfio_dev_path, &contents, &length, NULL)) {
+ error_setg(errp, "failed to load \"%s\"", vfio_dev_path);
+ goto out_free_dev_path;
+ }
+
+ if (sscanf(contents, "%d:%d", &major, &minor) != 2) {
+ error_setg(errp, "failed to get major:minor for \"%s\"", vfio_dev_path);
+ goto out_free_dev_path;
+ }
+ g_free(contents);
+ vfio_devt = makedev(major, minor);
+
+ vfio_path = g_strdup_printf("/dev/vfio/devices/%s", dent->d_name);
+ ret = open_cdev(vfio_path, vfio_devt);
+ if (ret < 0) {
+ error_setg(errp, "Failed to open %s", vfio_path);
+ }
+
+ trace_vfio_iommufd_get_devicefd(vfio_path, ret);
+ g_free(vfio_path);
+
+out_free_dev_path:
+ g_free(vfio_dev_path);
+out_close_dir:
+ closedir(dir);
+out_free_path:
+ if (*errp) {
+ error_prepend(errp, VFIO_MSG_PREFIX, path);
+ }
+ g_free(path);
+
+ return ret;
+}
+
+static VFIOIOASHwpt *vfio_container_get_hwpt(VFIOIOMMUFDContainer *container,
+ uint32_t hwpt_id)
+{
+ VFIOIOASHwpt *hwpt;
+
+ QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
+ if (hwpt->hwpt_id == hwpt_id) {
+ return hwpt;
+ }
+ }
+
+ hwpt = g_malloc0(sizeof(*hwpt));
+
+ hwpt->hwpt_id = hwpt_id;
+ QLIST_INIT(&hwpt->device_list);
+ QLIST_INSERT_HEAD(&container->hwpt_list, hwpt, next);
+
+ return hwpt;
+}
+
+static void vfio_container_put_hwpt(IOMMUFDBackend *be, VFIOIOASHwpt *hwpt)
+{
+ QLIST_REMOVE(hwpt, next);
+ g_free(hwpt);
+}
+
+static int vfio_device_attach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container,
+ Error **errp)
+{
+ int ret, iommufd = vbasedev->iommufd->fd;
+ VFIOIOASHwpt *hwpt;
+ struct vfio_device_attach_iommufd_pt attach_data = {
+ .argsz = sizeof(attach_data),
+ .flags = 0,
+ .pt_id = container->ioas_id,
+ };
+
+ /* Attach device to an ioas within iommufd */
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_ATTACH_IOMMUFD_PT, &attach_data);
+ if (ret) {
+ error_setg_errno(errp, errno,
+ "[iommufd=%d] error attach %s (%d) to ioasid=%d",
+ container->be->fd, vbasedev->name, vbasedev->fd,
+ attach_data.pt_id);
+ return ret;
+
+ }
+ hwpt = vfio_container_get_hwpt(container, attach_data.pt_id);
+
+ QLIST_INSERT_HEAD(&hwpt->device_list, vbasedev, next);
+ vbasedev->hwpt = hwpt;
+
+ trace_vfio_iommufd_attach_device(iommufd, vbasedev->name, vbasedev->fd,
+ container->ioas_id, attach_data.pt_id);
+ return ret;
+}
+
+static void vfio_device_detach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container)
+{
+ VFIOIOASHwpt *hwpt = vbasedev->hwpt;
+ struct vfio_device_attach_iommufd_pt detach_data = {
+ .argsz = sizeof(detach_data),
+ .flags = 0,
+ };
+
+ if (ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data)) {
+ error_report("detach %s from ioas id=%d failed: %m", vbasedev->name,
+ container->ioas_id);
+ }
+
+ QLIST_REMOVE(vbasedev, next);
+ vbasedev->hwpt = NULL;
+ if (QLIST_EMPTY(&hwpt->device_list)) {
+ vfio_container_put_hwpt(vbasedev->iommufd, hwpt);
+ }
+
+ trace_vfio_iommufd_detach_device(container->be->fd, vbasedev->name,
+ container->ioas_id);
+}
+
+static void vfio_iommufd_container_destroy(VFIOIOMMUFDContainer *container)
+{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
+
+ if (!QLIST_EMPTY(&container->hwpt_list)) {
+ return;
+ }
+ memory_listener_unregister(&bcontainer->listener);
+ vfio_container_destroy(bcontainer);
+ iommufd_backend_put_ioas(container->be, container->ioas_id);
+ g_free(container);
+}
+
+static int vfio_ram_block_discard_disable(bool state)
+{
+ /*
+ * We support coordinated discarding of RAM via the RamDiscardManager.
+ */
+ return ram_block_uncoordinated_discard_disable(state);
+}
+
+static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
+ AddressSpace *as, Error **errp)
+{
+ VFIOContainerBase *bcontainer;
+ VFIOIOMMUFDContainer *container;
+ VFIOAddressSpace *space;
+ struct vfio_device_info dev_info = { .argsz = sizeof(dev_info) };
+ int ret, devfd;
+ uint32_t ioas_id;
+ Error *err = NULL;
+
+ devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+
+ ret = iommufd_connect_and_bind(vbasedev, errp);
+ if (ret) {
+ goto err_connect_bind;
+ }
+
+ space = vfio_get_address_space(as);
+
+ /* try to attach to an existing container in this space */
+ QLIST_FOREACH(bcontainer, &space->containers, next) {
+ container = container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+ if (bcontainer->ops != &vfio_iommufd_ops ||
+ vbasedev->iommufd != container->be) {
+ continue;
+ }
+ if (vfio_device_attach_container(vbasedev, container, &err)) {
+ const char *msg = error_get_pretty(err);
+
+ trace_vfio_iommufd_fail_attach_existing_container(msg);
+ error_free(err);
+ err = NULL;
+ } else {
+ ret = vfio_ram_block_discard_disable(true);
+ if (ret) {
+ error_setg(errp,
+ "Cannot set discarding of RAM broken (%d)", ret);
+ goto err_discard_disable;
+ }
+ goto found_container;
+ }
+ }
+
+ /* Need to allocate a new dedicated container */
+ ret = iommufd_backend_get_ioas(vbasedev->iommufd, &ioas_id);
+ if (ret < 0) {
+ error_setg_errno(errp, errno, "Failed to alloc ioas");
+ goto err_get_ioas;
+ }
+
+ trace_vfio_iommufd_alloc_ioas(vbasedev->iommufd->fd, ioas_id);
+
+ container = g_malloc0(sizeof(*container));
+ container->be = vbasedev->iommufd;
+ container->ioas_id = ioas_id;
+ QLIST_INIT(&container->hwpt_list);
+
+ bcontainer = &container->bcontainer;
+ vfio_container_init(bcontainer, space, &vfio_iommufd_ops);
+ QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
+
+ ret = vfio_device_attach_container(vbasedev, container, errp);
+ if (ret) {
+ goto err_attach_container;
+ }
+
+ ret = vfio_ram_block_discard_disable(true);
+ if (ret) {
+ goto err_discard_disable;
+ }
+
+ bcontainer->pgsizes = qemu_real_host_page_size();
+
+ bcontainer->listener = vfio_memory_listener;
+ memory_listener_register(&bcontainer->listener, bcontainer->space->as);
+
+ if (bcontainer->error) {
+ ret = -1;
+ error_propagate_prepend(errp, bcontainer->error,
+ "memory listener initialization failed: ");
+ goto err_listener_register;
+ }
+
+ bcontainer->initialized = true;
+
+found_container:
+ ret = ioctl(devfd, VFIO_DEVICE_GET_INFO, &dev_info);
+ if (ret) {
+ error_setg_errno(errp, errno, "error getting device info");
+ goto err_listener_register;
+ }
+
+ /*
+ * TODO: examine RAM_BLOCK_DISCARD stuff, should we do group level
+ * for discarding incompatibility check as well?
+ */
+ if (vbasedev->ram_block_discard_allowed) {
+ vfio_ram_block_discard_disable(false);
+ }
+
+ vbasedev->group = 0;
+ vbasedev->num_irqs = dev_info.num_irqs;
+ vbasedev->num_regions = dev_info.num_regions;
+ vbasedev->flags = dev_info.flags;
+ vbasedev->reset_works = !!(dev_info.flags & VFIO_DEVICE_FLAGS_RESET);
+ vbasedev->bcontainer = bcontainer;
+ QLIST_INSERT_HEAD(&bcontainer->device_list, vbasedev, container_next);
+ QLIST_INSERT_HEAD(&vfio_device_list, vbasedev, global_next);
+
+ trace_vfio_iommufd_device_info(vbasedev->name, devfd, vbasedev->num_irqs,
+ vbasedev->num_regions, vbasedev->flags);
+ return 0;
+
+err_listener_register:
+ vfio_ram_block_discard_disable(false);
+err_discard_disable:
+ vfio_device_detach_container(vbasedev, container);
+err_attach_container:
+ vfio_iommufd_container_destroy(container);
+err_get_ioas:
+ vfio_put_address_space(space);
+ iommufd_unbind_and_disconnect(vbasedev);
+err_connect_bind:
+ close(vbasedev->fd);
+ return ret;
+}
+
+static void iommufd_detach_device(VFIODevice *vbasedev)
+{
+ VFIOContainerBase *bcontainer = vbasedev->bcontainer;
+ VFIOIOMMUFDContainer *container;
+ VFIOAddressSpace *space = bcontainer->space;
+
+ QLIST_REMOVE(vbasedev, global_next);
+ QLIST_REMOVE(vbasedev, container_next);
+ vbasedev->bcontainer = NULL;
+
+ if (!vbasedev->ram_block_discard_allowed) {
+ vfio_ram_block_discard_disable(false);
+ }
+
+ container = container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
+ vfio_device_detach_container(vbasedev, container);
+ vfio_iommufd_container_destroy(container);
+ vfio_put_address_space(space);
+
+ iommufd_unbind_and_disconnect(vbasedev);
+ close(vbasedev->fd);
+}
+
+const VFIOIOMMUOps vfio_iommufd_ops = {
+ .dma_map = iommufd_map,
+ .dma_unmap = iommufd_unmap,
+ .attach_device = iommufd_attach_device,
+ .detach_device = iommufd_detach_device,
+};
diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
index eb6ce6229d..9cae2c9e21 100644
--- a/hw/vfio/meson.build
+++ b/hw/vfio/meson.build
@@ -7,6 +7,9 @@ vfio_ss.add(files(
'spapr.c',
'migration.c',
))
+if have_iommufd
+ vfio_ss.add(files('iommufd.c'))
+endif
vfio_ss.add(when: 'CONFIG_VFIO_PCI', if_true: files(
'display.c',
'pci-quirks.c',
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 08a1f9dfa4..9b180cf77c 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -164,3 +164,15 @@ vfio_state_pending_estimate(const char *name, uint64_t precopy, uint64_t postcop
vfio_state_pending_exact(const char *name, uint64_t precopy, uint64_t postcopy, uint64_t stopcopy_size, uint64_t precopy_init_size, uint64_t precopy_dirty_size) " (%s) precopy 0x%"PRIx64" postcopy 0x%"PRIx64" stopcopy size 0x%"PRIx64" precopy initial size 0x%"PRIx64" precopy dirty size 0x%"PRIx64
vfio_vmstate_change(const char *name, int running, const char *reason, const char *dev_state) " (%s) running %d reason %s device state %s"
vfio_vmstate_change_prepare(const char *name, int running, const char *reason, const char *dev_state) " (%s) running %d reason %s device state %s"
+
+#iommufd.c
+
+vfio_iommufd_get_devicefd(const char *dev, int devfd) " %s (fd=%d)"
+vfio_iommufd_bind_device(int iommufd, const char *name, int devfd, int devid) " [iommufd=%d] Successfully bound device %s (fd=%d): output devid=%d"
+vfio_iommufd_fail_attach_existing_hwpt(const char *msg) " %s"
+vfio_iommufd_attach_device(int iommufd, const char *name, int devfd, int ioasid, int hwptid) " [iommufd=%d] Successfully attached device %s (%d) to ioasid=%d: output hwptd=%d"
+vfio_iommufd_detach_device(int iommufd, const char *name, int ioasid) " [iommufd=%d] Detached %s from ioasid=%d"
+vfio_iommufd_alloc_ioas(int iommufd, int ioas_id) " [iommufd=%d] new IOMMUFD container with ioasid=%d"
+vfio_iommufd_device_info(char *name, int devfd, int num_irqs, int num_regions, int flags) " %s (%d) num_irqs=%d num_regions=%d flags=%d"
+vfio_iommufd_fail_attach_existing_container(const char *msg) " %s"
+vfio_iommufd_container_reset(char *name) " Successfully reset %s"
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (25 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 26/37] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-30 13:52 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 28/37] vfio/iommufd: Add support for iova_ranges Zhenzhong Duan
` (10 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
IOMMUFD supports auto allocated hwpt and manually allocated one.
Manually allocated hwpt has benefit that its life cycle is under
user's control, so it could be used as stage 2 page table by nested
feature in the future.
Introduce two helpers __vfio_device_attach/detach_hwpt to facilitate
this change.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/iommufd.c | 89 +++++++++++++++++++++++++++++++++++++----------
1 file changed, 70 insertions(+), 19 deletions(-)
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index aee64d63f3..c1daaf1c39 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -217,38 +217,91 @@ static VFIOIOASHwpt *vfio_container_get_hwpt(VFIOIOMMUFDContainer *container,
static void vfio_container_put_hwpt(IOMMUFDBackend *be, VFIOIOASHwpt *hwpt)
{
QLIST_REMOVE(hwpt, next);
+ iommufd_backend_free_id(be->fd, hwpt->hwpt_id);
g_free(hwpt);
}
-static int vfio_device_attach_container(VFIODevice *vbasedev,
- VFIOIOMMUFDContainer *container,
- Error **errp)
+static int __vfio_device_attach_hwpt(VFIODevice *vbasedev, uint32_t hwpt_id,
+ Error **errp)
{
- int ret, iommufd = vbasedev->iommufd->fd;
- VFIOIOASHwpt *hwpt;
struct vfio_device_attach_iommufd_pt attach_data = {
.argsz = sizeof(attach_data),
.flags = 0,
- .pt_id = container->ioas_id,
+ .pt_id = hwpt_id,
};
+ int ret;
- /* Attach device to an ioas within iommufd */
ret = ioctl(vbasedev->fd, VFIO_DEVICE_ATTACH_IOMMUFD_PT, &attach_data);
if (ret) {
error_setg_errno(errp, errno,
- "[iommufd=%d] error attach %s (%d) to ioasid=%d",
- container->be->fd, vbasedev->name, vbasedev->fd,
- attach_data.pt_id);
+ "[iommufd=%d] error attach %s (%d) to hwpt_id=%d",
+ vbasedev->iommufd->fd, vbasedev->name, vbasedev->fd,
+ hwpt_id);
+ }
+ return ret;
+}
+
+static int __vfio_device_detach_hwpt(VFIODevice *vbasedev, Error **errp)
+{
+ struct vfio_device_detach_iommufd_pt detach_data = {
+ .argsz = sizeof(detach_data),
+ .flags = 0,
+ };
+ int ret;
+
+ ret = ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data);
+ if (ret) {
+ error_setg_errno(errp, errno, "detach %s from ioas failed",
+ vbasedev->name);
+ }
+ return ret;
+}
+
+static int vfio_device_attach_container(VFIODevice *vbasedev,
+ VFIOIOMMUFDContainer *container,
+ Error **errp)
+{
+ int ret, iommufd = vbasedev->iommufd->fd;
+ VFIOIOASHwpt *hwpt;
+ uint32_t hwpt_id;
+ Error *err = NULL;
+
+ /* try to attach to an existing hwpt in this container */
+ QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
+ ret = __vfio_device_attach_hwpt(vbasedev, hwpt->hwpt_id, &err);
+ if (ret) {
+ const char *msg = error_get_pretty(err);
+
+ trace_vfio_iommufd_fail_attach_existing_hwpt(msg);
+ error_free(err);
+ err = NULL;
+ } else {
+ goto found_hwpt;
+ }
+ }
+
+ ret = iommufd_backend_alloc_hwpt(iommufd, vbasedev->devid,
+ container->ioas_id, &hwpt_id);
+
+ if (ret) {
+ error_setg_errno(errp, errno, "error alloc shadow hwpt");
return ret;
+ }
+ /* Attach device to an hwpt within iommufd */
+ ret = __vfio_device_attach_hwpt(vbasedev, hwpt_id, errp);
+ if (ret) {
+ iommufd_backend_free_id(iommufd, hwpt_id);
+ return ret;
}
- hwpt = vfio_container_get_hwpt(container, attach_data.pt_id);
+ hwpt = vfio_container_get_hwpt(container, hwpt_id);
+found_hwpt:
QLIST_INSERT_HEAD(&hwpt->device_list, vbasedev, next);
vbasedev->hwpt = hwpt;
trace_vfio_iommufd_attach_device(iommufd, vbasedev->name, vbasedev->fd,
- container->ioas_id, attach_data.pt_id);
+ container->ioas_id, hwpt->hwpt_id);
return ret;
}
@@ -256,14 +309,12 @@ static void vfio_device_detach_container(VFIODevice *vbasedev,
VFIOIOMMUFDContainer *container)
{
VFIOIOASHwpt *hwpt = vbasedev->hwpt;
- struct vfio_device_attach_iommufd_pt detach_data = {
- .argsz = sizeof(detach_data),
- .flags = 0,
- };
+ Error *err = NULL;
+ int ret;
- if (ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data)) {
- error_report("detach %s from ioas id=%d failed: %m", vbasedev->name,
- container->ioas_id);
+ ret = __vfio_device_detach_hwpt(vbasedev, &err);
+ if (ret) {
+ error_report_err(err);
}
QLIST_REMOVE(vbasedev, next);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 28/37] vfio/iommufd: Add support for iova_ranges
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (26 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend Zhenzhong Duan
` (9 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
Some vIOMMU such as virtio-iommu use iova ranges from host side to
setup reserved ranges for passthrough device, so that guest will not
use an iova range beyond host support.
Use an uAPI of IOMMUFD to get iova ranges of host side and pass to
vIOMMU just like the legacy backend.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/iommufd.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 47 insertions(+)
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index c1daaf1c39..18a09d7f5a 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -348,6 +348,52 @@ static int vfio_ram_block_discard_disable(bool state)
return ram_block_uncoordinated_discard_disable(state);
}
+static int vfio_get_info_iova_range(VFIOIOMMUFDContainer *container,
+ uint32_t ioas_id)
+{
+ VFIOContainerBase *bcontainer = &container->bcontainer;
+ struct iommu_ioas_iova_ranges *info;
+ struct iommu_iova_range *iova_ranges;
+ int ret, sz, fd = container->be->fd;
+
+ info = g_malloc0(sizeof(*info));
+ info->size = sizeof(*info);
+ info->ioas_id = ioas_id;
+
+ ret = ioctl(fd, IOMMU_IOAS_IOVA_RANGES, info);
+ if (ret && errno != EMSGSIZE) {
+ goto error;
+ }
+
+ sz = info->num_iovas * sizeof(struct iommu_iova_range);
+ info = g_realloc(info, sizeof(*info) + sz);
+ info->allowed_iovas = (uint64_t)(info + 1);
+
+ ret = ioctl(fd, IOMMU_IOAS_IOVA_RANGES, info);
+ if (ret) {
+ goto error;
+ }
+
+ iova_ranges = (struct iommu_iova_range *)info->allowed_iovas;
+
+ for (int i = 0; i < info->num_iovas; i++) {
+ Range *range = g_new(Range, 1);
+
+ range_set_bounds(range, iova_ranges[i].start, iova_ranges[i].last);
+ bcontainer->iova_ranges =
+ range_list_insert(bcontainer->iova_ranges, range);
+ }
+
+ g_free(info);
+ return 0;
+
+error:
+ ret = -errno;
+ g_free(info);
+ error_report("vfio/iommufd: Cannot get iova ranges: %m");
+ return ret;
+}
+
static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
AddressSpace *as, Error **errp)
{
@@ -425,6 +471,7 @@ static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
}
bcontainer->pgsizes = qemu_real_host_page_size();
+ vfio_get_info_iova_range(container, ioas_id);
bcontainer->listener = vfio_memory_listener;
memory_listener_register(&bcontainer->listener, bcontainer->space->as);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (27 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 28/37] vfio/iommufd: Add support for iova_ranges Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-30 13:56 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info Zhenzhong Duan
` (8 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
IBM EEH is only supported by legacy backend currently, bypass it
for IOMMUFD backend.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/ppc/spapr_pci_vfio.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
index d1d07bec46..a2518838a1 100644
--- a/hw/ppc/spapr_pci_vfio.c
+++ b/hw/ppc/spapr_pci_vfio.c
@@ -93,10 +93,10 @@ static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
bcontainer = QLIST_FIRST(&space->containers);
- if (QLIST_NEXT(bcontainer, next)) {
+ if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != &vfio_legacy_ops) {
/*
* We don't yet have logic to synchronize EEH state across
- * multiple containers
+ * multiple containers, iommufd isn't supported too.
*/
bcontainer = NULL;
goto out;
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (28 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-30 13:59 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
` (7 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
This helper will be used by both legacy and iommufd backends.
No functional changes intended.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/pci.c | 54 +++++++++++++++++++++++++++++++++++----------------
1 file changed, 37 insertions(+), 17 deletions(-)
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 5cbc771e55..c17e1f4376 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -2459,22 +2459,13 @@ static bool vfio_pci_host_match(PCIHostDeviceAddress *addr, const char *name)
return (strcmp(tmp, name) == 0);
}
-static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+static int vfio_pci_get_pci_hot_reset_info(VFIOPCIDevice *vdev,
+ struct vfio_pci_hot_reset_info **info_p)
{
- VFIOGroup *group;
struct vfio_pci_hot_reset_info *info;
- struct vfio_pci_dependent_device *devices;
- struct vfio_pci_hot_reset *reset;
- int32_t *fds;
- int ret, i, count;
- bool multi = false;
+ int ret, count;
- trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
-
- if (!single) {
- vfio_pci_pre_reset(vdev);
- }
- vdev->vbasedev.needs_reset = false;
+ assert(info_p && !*info_p);
info = g_malloc0(sizeof(*info));
info->argsz = sizeof(*info);
@@ -2482,24 +2473,53 @@ static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
if (ret && errno != ENOSPC) {
ret = -errno;
+ g_free(info);
if (!vdev->has_pm_reset) {
error_report("vfio: Cannot reset device %s, "
"no available reset mechanism.", vdev->vbasedev.name);
}
- goto out_single;
+ return ret;
}
count = info->count;
- info = g_realloc(info, sizeof(*info) + (count * sizeof(*devices)));
- info->argsz = sizeof(*info) + (count * sizeof(*devices));
- devices = &info->devices[0];
+ info = g_realloc(info, sizeof(*info) + (count * sizeof(info->devices[0])));
+ info->argsz = sizeof(*info) + (count * sizeof(info->devices[0]));
ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
if (ret) {
ret = -errno;
+ g_free(info);
error_report("vfio: hot reset info failed: %m");
+ return ret;
+ }
+
+ *info_p = info;
+ return 0;
+}
+
+static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+{
+ VFIOGroup *group;
+ struct vfio_pci_hot_reset_info *info = NULL;
+ struct vfio_pci_dependent_device *devices;
+ struct vfio_pci_hot_reset *reset;
+ int32_t *fds;
+ int ret, i, count;
+ bool multi = false;
+
+ trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
+
+ if (!single) {
+ vfio_pci_pre_reset(vdev);
+ }
+ vdev->vbasedev.needs_reset = false;
+
+ ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
+
+ if (ret) {
goto out_single;
}
+ devices = &info->devices[0];
trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (29 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-30 14:04 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 32/37] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
` (6 subsequent siblings)
37 siblings, 1 reply; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
As pci hot reset path need to reference pci specific functions
and data structures, adding container level callback functions
for legacy and iommufd BE and referencing those pci specific
func/data is no better than implementing reset support with
iommufd BE directly in pci.c
This way we can also share the common bus reset and system reset
path for both BEs.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
---
hw/vfio/pci.c | 156 ++++++++++++++++++++++++++++++++++++++++++-
hw/vfio/trace-events | 1 +
2 files changed, 156 insertions(+), 1 deletion(-)
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index c17e1f4376..d7a41c8def 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -42,6 +42,7 @@
#include "qapi/error.h"
#include "migration/blocker.h"
#include "migration/qemu-file.h"
+#include "linux/iommufd.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -2497,7 +2498,7 @@ static int vfio_pci_get_pci_hot_reset_info(VFIOPCIDevice *vdev,
return 0;
}
-static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+static int vfio_pci_hot_reset_legacy(VFIOPCIDevice *vdev, bool single)
{
VFIOGroup *group;
struct vfio_pci_hot_reset_info *info = NULL;
@@ -2661,6 +2662,159 @@ out_single:
return ret;
}
+#ifdef CONFIG_IOMMUFD
+static VFIODevice *vfio_pci_find_by_iommufd_devid(__u32 devid)
+{
+ VFIODevice *vbasedev_iter;
+
+ QLIST_FOREACH(vbasedev_iter, &vfio_device_list, global_next) {
+ if (vbasedev_iter->bcontainer->ops != &vfio_iommufd_ops) {
+ continue;
+ }
+ if (devid == vbasedev_iter->devid) {
+ return vbasedev_iter;
+ }
+ }
+ return NULL;
+}
+
+static int vfio_pci_hot_reset_iommufd(VFIOPCIDevice *vdev, bool single)
+{
+ struct vfio_pci_hot_reset_info *info = NULL;
+ struct vfio_pci_dependent_device *devices;
+ struct vfio_pci_hot_reset *reset;
+ int ret, i;
+ bool multi = false;
+
+ trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
+
+ if (!single) {
+ vfio_pci_pre_reset(vdev);
+ }
+ vdev->vbasedev.needs_reset = false;
+
+ ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
+
+ if (ret) {
+ goto out_single;
+ }
+
+ assert(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID);
+
+ devices = &info->devices[0];
+
+ if (!(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID_OWNED)) {
+ if (!vdev->has_pm_reset) {
+ for (i = 0; i < info->count; i++) {
+ if (devices[i].devid == VFIO_PCI_DEVID_NOT_OWNED) {
+ error_report("vfio: Cannot reset device %s, "
+ "depends on device %04x:%02x:%02x.%x "
+ "which is not owned.",
+ vdev->vbasedev.name, devices[i].segment,
+ devices[i].bus, PCI_SLOT(devices[i].devfn),
+ PCI_FUNC(devices[i].devfn));
+ }
+ }
+ }
+ ret = -EPERM;
+ goto out_single;
+ }
+
+ trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
+
+ for (i = 0; i < info->count; i++) {
+ VFIOPCIDevice *tmp;
+ VFIODevice *vbasedev_iter;
+
+ trace_vfio_pci_hot_reset_dep_devices_iommufd(devices[i].segment,
+ devices[i].bus,
+ PCI_SLOT(devices[i].devfn),
+ PCI_FUNC(devices[i].devfn),
+ devices[i].devid);
+
+ /*
+ * If a VFIO cdev device is resettable, all the dependent devices
+ * are either bound to same iommufd or within same iommu_groups as
+ * one of the iommufd bound devices.
+ */
+ assert(devices[i].devid != VFIO_PCI_DEVID_NOT_OWNED);
+
+ if (devices[i].devid == vdev->vbasedev.devid ||
+ devices[i].devid == VFIO_PCI_DEVID_OWNED) {
+ continue;
+ }
+
+ vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
+ if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
+ vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
+ continue;
+ }
+ tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
+ if (single) {
+ ret = -EINVAL;
+ goto out_single;
+ }
+ vfio_pci_pre_reset(tmp);
+ tmp->vbasedev.needs_reset = false;
+ multi = true;
+ }
+
+ if (!single && !multi) {
+ ret = -EINVAL;
+ goto out_single;
+ }
+
+ /* Use zero length array for hot reset with iommufd backend */
+ reset = g_malloc0(sizeof(*reset));
+ reset->argsz = sizeof(*reset);
+
+ /* Bus reset! */
+ ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_PCI_HOT_RESET, reset);
+ g_free(reset);
+
+ trace_vfio_pci_hot_reset_result(vdev->vbasedev.name,
+ ret ? strerror(errno) : "Success");
+
+ /* Re-enable INTx on affected devices */
+ for (i = 0; i < info->count; i++) {
+ VFIOPCIDevice *tmp;
+ VFIODevice *vbasedev_iter;
+
+ if (devices[i].devid == vdev->vbasedev.devid ||
+ devices[i].devid == VFIO_PCI_DEVID_OWNED) {
+ continue;
+ }
+
+ vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
+ if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
+ vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
+ continue;
+ }
+ tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
+ vfio_pci_post_reset(tmp);
+ }
+out_single:
+ if (!single) {
+ vfio_pci_post_reset(vdev);
+ }
+ g_free(info);
+
+ return ret;
+}
+#endif
+
+static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
+{
+#ifdef CONFIG_IOMMUFD
+ if (vdev->vbasedev.iommufd) {
+ return vfio_pci_hot_reset_iommufd(vdev, single);
+ } else
+#endif
+ {
+ return vfio_pci_hot_reset_legacy(vdev, single);
+ }
+}
+
/*
* We want to differentiate hot reset of multiple in-use devices vs hot reset
* of a single in-use device. VFIO_DEVICE_RESET will already handle the case
diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
index 9b180cf77c..71c5840636 100644
--- a/hw/vfio/trace-events
+++ b/hw/vfio/trace-events
@@ -34,6 +34,7 @@ vfio_check_af_flr(const char *name) "%s Supports FLR via AF cap"
vfio_pci_hot_reset(const char *name, const char *type) " (%s) %s"
vfio_pci_hot_reset_has_dep_devices(const char *name) "%s: hot reset dependent devices:"
vfio_pci_hot_reset_dep_devices(int domain, int bus, int slot, int function, int group_id) "\t%04x:%02x:%02x.%x group %d"
+vfio_pci_hot_reset_dep_devices_iommufd(int domain, int bus, int slot, int function, int dev_id) "\t%04x:%02x:%02x.%x devid %d"
vfio_pci_hot_reset_result(const char *name, const char *result) "%s hot reset: %s"
vfio_populate_device_config(const char *name, unsigned long size, unsigned long offset, unsigned long flags) "Device %s config:\n size: 0x%lx, offset: 0x%lx, flags: 0x%lx"
vfio_populate_device_get_irq_info_failure(const char *errstr) "VFIO_DEVICE_GET_IRQ_INFO failure: %s"
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 32/37] vfio/pci: Allow the selection of a given iommu backend
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (30 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
@ 2023-10-26 10:30 ` Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 33/37] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
` (5 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:30 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
From: Eric Auger <eric.auger@redhat.com>
Now we support two types of iommu backends, let's add the capability
to select one of them. This depends on whether an iommufd object has
been linked with the vfio-pci device:
if the user wants to use the legacy backend, it shall not
link the vfio-pci device with any iommufd object:
-device vfio-pci,host=0000:02:00.0
This is called the legacy mode/backend.
If the user wants to use the iommufd backend (/dev/iommu) it
shall pass an iommufd object id in the vfio-pci device options:
-object iommufd,id=iommufd0
-device vfio-pci,host=0000:02:00.0,iommufd=iommufd0
Suggested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
hw/vfio/pci.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index d7a41c8def..386c08576a 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -43,6 +43,7 @@
#include "migration/blocker.h"
#include "migration/qemu-file.h"
#include "linux/iommufd.h"
+#include "sysemu/iommufd.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -3710,6 +3711,10 @@ static Property vfio_pci_dev_properties[] = {
* DEFINE_PROP_STRING("vfiofd", VFIOPCIDevice, vfiofd_name),
* DEFINE_PROP_STRING("vfiogroupfd, VFIOPCIDevice, vfiogroupfd_name),
*/
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 33/37] vfio/pci: Make vfio cdev pre-openable by passing a file handle
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (31 preceding siblings ...)
2023-10-26 10:30 ` [PATCH v3 32/37] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
@ 2023-10-26 10:31 ` Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 34/37] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
` (4 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu. This way qemu never needs
to have privilege to open a VFIO or iommu cdev node.
Together with the earlier support of pre-opening /dev/iommu device,
now we have full support of passing a vfio device to unprivileged
qemu by management tool. This mode is no more considered for the
legacy backend. So let's remove the "TODO" comment.
Add a helper function vfio_device_get_name() to check fd and get
device name, it will also be used by other vfio devices.
There is no easy way to check if a device is mdev with FD passing,
so fail the x-balloon-allowed check unconditionally in this case.
There is also no easy way to get BDF as name with FD passing, so
we fake a name by VFIO_FD[fd].
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-common.h | 1 +
hw/vfio/helpers.c | 33 +++++++++++++++++++++++++++++
hw/vfio/iommufd.c | 12 +++++++----
hw/vfio/pci.c | 40 ++++++++++++++++++++++++-----------
4 files changed, 70 insertions(+), 16 deletions(-)
diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
index 95afee7243..5c00699290 100644
--- a/include/hw/vfio/vfio-common.h
+++ b/include/hw/vfio/vfio-common.h
@@ -258,6 +258,7 @@ struct vfio_info_cap_header *
vfio_get_device_info_cap(struct vfio_device_info *info, uint16_t id);
struct vfio_info_cap_header *
vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp);
#endif
bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
diff --git a/hw/vfio/helpers.c b/hw/vfio/helpers.c
index 168847e7c5..044dbbc501 100644
--- a/hw/vfio/helpers.c
+++ b/hw/vfio/helpers.c
@@ -609,3 +609,36 @@ bool vfio_has_region_cap(VFIODevice *vbasedev, int region, uint16_t cap_type)
return ret;
}
+
+int vfio_device_get_name(VFIODevice *vbasedev, Error **errp)
+{
+ struct stat st;
+
+ if (vbasedev->fd < 0) {
+ if (stat(vbasedev->sysfsdev, &st) < 0) {
+ error_setg_errno(errp, errno, "no such host device");
+ error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ return -errno;
+ }
+ /* User may specify a name, e.g: VFIO platform device */
+ if (!vbasedev->name) {
+ vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+ }
+ }
+#ifdef CONFIG_IOMMUFD
+ else {
+ if (!vbasedev->iommufd) {
+ error_setg(errp, "Use FD passing only with iommufd backend");
+ return -EINVAL;
+ }
+ /*
+ * Give a name with fd so any function printing out vbasedev->name
+ * will not break.
+ */
+ if (!vbasedev->name) {
+ vbasedev->name = g_strdup_printf("VFIO_FD%d", vbasedev->fd);
+ }
+ }
+#endif
+ return 0;
+}
diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
index 18a09d7f5a..20df807d7c 100644
--- a/hw/vfio/iommufd.c
+++ b/hw/vfio/iommufd.c
@@ -405,11 +405,15 @@ static int iommufd_attach_device(char *name, VFIODevice *vbasedev,
uint32_t ioas_id;
Error *err = NULL;
- devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
- if (devfd < 0) {
- return devfd;
+ if (vbasedev->fd < 0) {
+ devfd = vfio_get_devicefd(vbasedev->sysfsdev, errp);
+ if (devfd < 0) {
+ return devfd;
+ }
+ vbasedev->fd = devfd;
+ } else {
+ devfd = vbasedev->fd;
}
- vbasedev->fd = devfd;
ret = iommufd_connect_and_bind(vbasedev, errp);
if (ret) {
diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
index 386c08576a..32a4e8beb0 100644
--- a/hw/vfio/pci.c
+++ b/hw/vfio/pci.c
@@ -44,6 +44,7 @@
#include "migration/qemu-file.h"
#include "linux/iommufd.h"
#include "sysemu/iommufd.h"
+#include "monitor/monitor.h"
#define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
@@ -3267,18 +3268,23 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
VFIODevice *vbasedev = &vdev->vbasedev;
char *tmp, *subsys;
Error *err = NULL;
- struct stat st;
int i, ret;
bool is_mdev;
char uuid[UUID_FMT_LEN];
char *name;
- if (!vbasedev->sysfsdev) {
+ if (vbasedev->fd < 0 && !vbasedev->sysfsdev) {
if (!(~vdev->host.domain || ~vdev->host.bus ||
~vdev->host.slot || ~vdev->host.function)) {
error_setg(errp, "No provided host device");
+#ifdef CONFIG_IOMMUFD
+ error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F, "
+ "-device vfio-pci,sysfsdev=PATH_TO_DEVICE "
+ "or -device vfio-pci,fd=DEVICE_FD\n");
+#else
error_append_hint(errp, "Use -device vfio-pci,host=DDDD:BB:DD.F "
"or -device vfio-pci,sysfsdev=PATH_TO_DEVICE\n");
+#endif
return;
}
vbasedev->sysfsdev =
@@ -3287,13 +3293,9 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
vdev->host.slot, vdev->host.function);
}
- if (stat(vbasedev->sysfsdev, &st) < 0) {
- error_setg_errno(errp, errno, "no such host device");
- error_prepend(errp, VFIO_MSG_PREFIX, vbasedev->sysfsdev);
+ if (vfio_device_get_name(vbasedev, errp)) {
return;
}
-
- vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
vbasedev->ops = &vfio_pci_ops;
vbasedev->type = VFIO_DEVICE_TYPE_PCI;
vbasedev->dev = DEVICE(vdev);
@@ -3653,6 +3655,7 @@ static void vfio_instance_init(Object *obj)
vdev->host.bus = ~0U;
vdev->host.slot = ~0U;
vdev->host.function = ~0U;
+ vdev->vbasedev.fd = -1;
vdev->nv_gpudirect_clique = 0xFF;
@@ -3706,11 +3709,6 @@ static Property vfio_pci_dev_properties[] = {
qdev_prop_nv_gpudirect_clique, uint8_t),
DEFINE_PROP_OFF_AUTO_PCIBAR("x-msix-relocation", VFIOPCIDevice, msix_relo,
OFF_AUTOPCIBAR_OFF),
- /*
- * TODO - support passed fds... is this necessary?
- * DEFINE_PROP_STRING("vfiofd", VFIOPCIDevice, vfiofd_name),
- * DEFINE_PROP_STRING("vfiogroupfd, VFIOPCIDevice, vfiogroupfd_name),
- */
#ifdef CONFIG_IOMMUFD
DEFINE_PROP_LINK("iommufd", VFIOPCIDevice, vbasedev.iommufd,
TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
@@ -3718,6 +3716,21 @@ static Property vfio_pci_dev_properties[] = {
DEFINE_PROP_END_OF_LIST(),
};
+#ifdef CONFIG_IOMMUFD
+static void vfio_pci_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOPCIDevice *vdev = VFIO_PCI(obj);
+ int fd = -1;
+
+ fd = monitor_fd_param(monitor_cur(), str, errp);
+ if (fd == -1) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ vdev->vbasedev.fd = fd;
+}
+#endif
+
static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
@@ -3725,6 +3738,9 @@ static void vfio_pci_dev_class_init(ObjectClass *klass, void *data)
dc->reset = vfio_pci_reset;
device_class_set_props(dc, vfio_pci_dev_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_pci_set_fd);
+#endif
dc->desc = "VFIO-based PCI device assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
pdc->realize = vfio_realize;
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 34/37] vfio: Allow the selection of a given iommu backend for platform ap and ccw
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (32 preceding siblings ...)
2023-10-26 10:31 ` [PATCH v3 33/37] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
@ 2023-10-26 10:31 ` Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 35/37] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
` (3 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Thomas Huth, Tony Krowiak, Halil Pasic,
Jason Herne, Eric Farman, Matthew Rosato,
open list:S390 general arch...
Previously we added support to select iommu backend for vfio pci
device. Now we added others, E.g: platform, ap and ccw.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
include/hw/vfio/vfio-platform.h | 1 +
hw/vfio/ap.c | 5 +++++
hw/vfio/ccw.c | 5 +++++
hw/vfio/platform.c | 4 ++++
4 files changed, 15 insertions(+)
diff --git a/include/hw/vfio/vfio-platform.h b/include/hw/vfio/vfio-platform.h
index c414c3dffc..f57f4276f2 100644
--- a/include/hw/vfio/vfio-platform.h
+++ b/include/hw/vfio/vfio-platform.h
@@ -18,6 +18,7 @@
#include "hw/sysbus.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "qemu/event_notifier.h"
#include "qemu/queue.h"
#include "qom/object.h"
diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
index bbf69ff55a..6a4186ccd3 100644
--- a/hw/vfio/ap.c
+++ b/hw/vfio/ap.c
@@ -15,6 +15,7 @@
#include <sys/ioctl.h>
#include "qapi/error.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "hw/s390x/ap-device.h"
#include "qemu/error-report.h"
#include "qemu/event_notifier.h"
@@ -204,6 +205,10 @@ static void vfio_ap_unrealize(DeviceState *dev)
static Property vfio_ap_properties[] = {
DEFINE_PROP_STRING("sysfsdev", VFIOAPDevice, vdev.sysfsdev),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOAPDevice, vdev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index d857bb8d0f..7695ede0fc 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -21,6 +21,7 @@
#include "qapi/error.h"
#include "hw/vfio/vfio-common.h"
+#include "sysemu/iommufd.h"
#include "hw/s390x/s390-ccw.h"
#include "hw/s390x/vfio-ccw.h"
#include "hw/qdev-properties.h"
@@ -677,6 +678,10 @@ static void vfio_ccw_unrealize(DeviceState *dev)
static Property vfio_ccw_properties[] = {
DEFINE_PROP_STRING("sysfsdev", VFIOCCWDevice, vdev.sysfsdev),
DEFINE_PROP_BOOL("force-orb-pfch", VFIOCCWDevice, force_orb_pfch, false),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOCCWDevice, vdev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c
index 8e3d4ac458..a1c25e0337 100644
--- a/hw/vfio/platform.c
+++ b/hw/vfio/platform.c
@@ -649,6 +649,10 @@ static Property vfio_platform_dev_properties[] = {
DEFINE_PROP_UINT32("mmap-timeout-ms", VFIOPlatformDevice,
mmap_timeout, 1100),
DEFINE_PROP_BOOL("x-irqfd", VFIOPlatformDevice, irqfd_allowed, true),
+#ifdef CONFIG_IOMMUFD
+ DEFINE_PROP_LINK("iommufd", VFIOPlatformDevice, vbasedev.iommufd,
+ TYPE_IOMMUFD_BACKEND, IOMMUFDBackend *),
+#endif
DEFINE_PROP_END_OF_LIST(),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 35/37] vfio/platform: Make vfio cdev pre-openable by passing a file handle
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (33 preceding siblings ...)
2023-10-26 10:31 ` [PATCH v3 34/37] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
@ 2023-10-26 10:31 ` Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 36/37] vfio/ap: " Zhenzhong Duan
` (2 subsequent siblings)
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan
This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu. This way qemu never needs
to have privilege to open a VFIO or iommu cdev node.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
hw/vfio/platform.c | 41 +++++++++++++++++++++++++++++++++--------
1 file changed, 33 insertions(+), 8 deletions(-)
diff --git a/hw/vfio/platform.c b/hw/vfio/platform.c
index a1c25e0337..aa0b2b9583 100644
--- a/hw/vfio/platform.c
+++ b/hw/vfio/platform.c
@@ -35,6 +35,7 @@
#include "hw/platform-bus.h"
#include "hw/qdev-properties.h"
#include "sysemu/kvm.h"
+#include "monitor/monitor.h"
/*
* Functions used whatever the injection method
@@ -529,14 +530,13 @@ static VFIODeviceOps vfio_platform_ops = {
*/
static int vfio_base_device_init(VFIODevice *vbasedev, Error **errp)
{
- struct stat st;
int ret;
- /* @sysfsdev takes precedence over @host */
- if (vbasedev->sysfsdev) {
+ /* @fd takes precedence over @sysfsdev which takes precedence over @host */
+ if (vbasedev->fd < 0 && vbasedev->sysfsdev) {
g_free(vbasedev->name);
vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
- } else {
+ } else if (vbasedev->fd < 0) {
if (!vbasedev->name || strchr(vbasedev->name, '/')) {
error_setg(errp, "wrong host device name");
return -EINVAL;
@@ -546,10 +546,9 @@ static int vfio_base_device_init(VFIODevice *vbasedev, Error **errp)
vbasedev->name);
}
- if (stat(vbasedev->sysfsdev, &st) < 0) {
- error_setg_errno(errp, errno,
- "failed to get the sysfs host device file status");
- return -errno;
+ ret = vfio_device_get_name(vbasedev, errp);
+ if (ret) {
+ return ret;
}
ret = vfio_attach_device(vbasedev->name, vbasedev,
@@ -656,6 +655,28 @@ static Property vfio_platform_dev_properties[] = {
DEFINE_PROP_END_OF_LIST(),
};
+static void vfio_platform_instance_init(Object *obj)
+{
+ VFIOPlatformDevice *vdev = VFIO_PLATFORM_DEVICE(obj);
+
+ vdev->vbasedev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_platform_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOPlatformDevice *vdev = VFIO_PLATFORM_DEVICE(obj);
+ int fd = -1;
+
+ fd = monitor_fd_param(monitor_cur(), str, errp);
+ if (fd == -1) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ vdev->vbasedev.fd = fd;
+}
+#endif
+
static void vfio_platform_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
@@ -663,6 +684,9 @@ static void vfio_platform_class_init(ObjectClass *klass, void *data)
dc->realize = vfio_platform_realize;
device_class_set_props(dc, vfio_platform_dev_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_platform_set_fd);
+#endif
dc->vmsd = &vfio_platform_vmstate;
dc->desc = "VFIO-based platform device assignment";
sbc->connect_irq_notifier = vfio_start_irqfd_injection;
@@ -675,6 +699,7 @@ static const TypeInfo vfio_platform_dev_info = {
.name = TYPE_VFIO_PLATFORM,
.parent = TYPE_SYS_BUS_DEVICE,
.instance_size = sizeof(VFIOPlatformDevice),
+ .instance_init = vfio_platform_instance_init,
.class_init = vfio_platform_class_init,
.class_size = sizeof(VFIOPlatformDeviceClass),
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 36/37] vfio/ap: Make vfio cdev pre-openable by passing a file handle
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (34 preceding siblings ...)
2023-10-26 10:31 ` [PATCH v3 35/37] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
@ 2023-10-26 10:31 ` Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 37/37] vfio/ccw: " Zhenzhong Duan
2023-10-26 13:27 ` [PATCH v3 00/37] vfio: Adopt iommufd Markus Armbruster
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Thomas Huth, Tony Krowiak, Halil Pasic,
Jason Herne, open list:S390 general arch...
This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu. This way qemu never needs
to have privilege to open a VFIO or iommu cdev node.
Opportunisticly, remove some unnecessory double-cast.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
hw/vfio/ap.c | 32 +++++++++++++++++++++++++++++++-
1 file changed, 31 insertions(+), 1 deletion(-)
diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
index 6a4186ccd3..0a810f8b88 100644
--- a/hw/vfio/ap.c
+++ b/hw/vfio/ap.c
@@ -29,6 +29,7 @@
#include "hw/s390x/ap-bridge.h"
#include "exec/address-spaces.h"
#include "qom/object.h"
+#include "monitor/monitor.h"
#define TYPE_VFIO_AP_DEVICE "vfio-ap"
@@ -159,7 +160,10 @@ static void vfio_ap_realize(DeviceState *dev, Error **errp)
VFIOAPDevice *vapdev = VFIO_AP_DEVICE(dev);
VFIODevice *vbasedev = &vapdev->vdev;
- vbasedev->name = g_path_get_basename(vbasedev->sysfsdev);
+ if (vfio_device_get_name(vbasedev, errp)) {
+ return;
+ }
+
vbasedev->ops = &vfio_ap_ops;
vbasedev->type = VFIO_DEVICE_TYPE_AP;
vbasedev->dev = dev;
@@ -229,11 +233,36 @@ static const VMStateDescription vfio_ap_vmstate = {
.unmigratable = 1,
};
+static void vfio_ap_instance_init(Object *obj)
+{
+ VFIOAPDevice *vapdev = VFIO_AP_DEVICE(obj);
+
+ vapdev->vdev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_ap_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOAPDevice *vapdev = VFIO_AP_DEVICE(obj);
+ int fd = -1;
+
+ fd = monitor_fd_param(monitor_cur(), str, errp);
+ if (fd == -1) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ vapdev->vdev.fd = fd;
+}
+#endif
+
static void vfio_ap_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
device_class_set_props(dc, vfio_ap_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_ap_set_fd);
+#endif
dc->vmsd = &vfio_ap_vmstate;
dc->desc = "VFIO-based AP device assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
@@ -248,6 +277,7 @@ static const TypeInfo vfio_ap_info = {
.name = TYPE_VFIO_AP_DEVICE,
.parent = TYPE_AP_DEVICE,
.instance_size = sizeof(VFIOAPDevice),
+ .instance_init = vfio_ap_instance_init,
.class_init = vfio_ap_class_init,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* [PATCH v3 37/37] vfio/ccw: Make vfio cdev pre-openable by passing a file handle
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (35 preceding siblings ...)
2023-10-26 10:31 ` [PATCH v3 36/37] vfio/ap: " Zhenzhong Duan
@ 2023-10-26 10:31 ` Zhenzhong Duan
2023-10-26 13:27 ` [PATCH v3 00/37] vfio: Adopt iommufd Markus Armbruster
37 siblings, 0 replies; 95+ messages in thread
From: Zhenzhong Duan @ 2023-10-26 10:31 UTC (permalink / raw)
To: qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Zhenzhong Duan, Eric Farman, Matthew Rosato, Thomas Huth,
open list:vfio-ccw
This gives management tools like libvirt a chance to open the vfio
cdev with privilege and pass FD to qemu. This way qemu never needs
to have privilege to open a VFIO or iommu cdev node.
Opportunisticly, remove a redundant definition of TYPE_VFIO_CCW.
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Signed-off-by: Cédric Le Goater <clg@redhat.com>
---
hw/vfio/ccw.c | 34 +++++++++++++++++++++++++++++++---
1 file changed, 31 insertions(+), 3 deletions(-)
diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
index 7695ede0fc..a674bd8d6d 100644
--- a/hw/vfio/ccw.c
+++ b/hw/vfio/ccw.c
@@ -30,6 +30,7 @@
#include "qemu/error-report.h"
#include "qemu/main-loop.h"
#include "qemu/module.h"
+#include "monitor/monitor.h"
struct VFIOCCWDevice {
S390CCWDevice cdev;
@@ -589,11 +590,12 @@ static void vfio_ccw_realize(DeviceState *dev, Error **errp)
}
}
+ if (vfio_device_get_name(vbasedev, errp)) {
+ return;
+ }
+
vbasedev->ops = &vfio_ccw_ops;
vbasedev->type = VFIO_DEVICE_TYPE_CCW;
- vbasedev->name = g_strdup_printf("%x.%x.%04x", vcdev->cdev.hostid.cssid,
- vcdev->cdev.hostid.ssid,
- vcdev->cdev.hostid.devid);
vbasedev->dev = dev;
/*
@@ -690,12 +692,37 @@ static const VMStateDescription vfio_ccw_vmstate = {
.unmigratable = 1,
};
+static void vfio_ccw_instance_init(Object *obj)
+{
+ VFIOCCWDevice *vcdev = VFIO_CCW(obj);
+
+ vcdev->vdev.fd = -1;
+}
+
+#ifdef CONFIG_IOMMUFD
+static void vfio_ccw_set_fd(Object *obj, const char *str, Error **errp)
+{
+ VFIOCCWDevice *vcdev = VFIO_CCW(obj);
+ int fd = -1;
+
+ fd = monitor_fd_param(monitor_cur(), str, errp);
+ if (fd == -1) {
+ error_prepend(errp, "Could not parse remote object fd %s:", str);
+ return;
+ }
+ vcdev->vdev.fd = fd;
+}
+#endif
+
static void vfio_ccw_class_init(ObjectClass *klass, void *data)
{
DeviceClass *dc = DEVICE_CLASS(klass);
S390CCWDeviceClass *cdc = S390_CCW_DEVICE_CLASS(klass);
device_class_set_props(dc, vfio_ccw_properties);
+#ifdef CONFIG_IOMMUFD
+ object_class_property_add_str(klass, "fd", NULL, vfio_ccw_set_fd);
+#endif
dc->vmsd = &vfio_ccw_vmstate;
dc->desc = "VFIO-based subchannel assignment";
set_bit(DEVICE_CATEGORY_MISC, dc->categories);
@@ -713,6 +740,7 @@ static const TypeInfo vfio_ccw_info = {
.name = TYPE_VFIO_CCW,
.parent = TYPE_S390_CCW,
.instance_size = sizeof(VFIOCCWDevice),
+ .instance_init = vfio_ccw_instance_init,
.class_init = vfio_ccw_class_init,
};
--
2.34.1
^ permalink raw reply related [flat|nested] 95+ messages in thread
* Re: [PATCH v3 00/37] vfio: Adopt iommufd
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
` (36 preceding siblings ...)
2023-10-26 10:31 ` [PATCH v3 37/37] vfio/ccw: " Zhenzhong Duan
@ 2023-10-26 13:27 ` Markus Armbruster
2023-10-27 6:17 ` Duan, Zhenzhong
37 siblings, 1 reply; 95+ messages in thread
From: Markus Armbruster @ 2023-10-26 13:27 UTC (permalink / raw)
To: Zhenzhong Duan
Cc: qemu-devel, alex.williamson, clg, jgg, nicolinc, joao.m.martins,
eric.auger, peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun,
chao.p.peng
Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
> Hi,
>
> Thanks all for giving guides and comments on previous series, here is
> the v3 of pure iommufd support part.
>
> Based on Cédric's suggestion, this series includes an effort to remove
> spapr code from container.c, now all spapr functions are moved to spapr.c
> or spapr_pci_vfio.c, but there are still a few trival check on
> VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce many
> callbacks and duplicate code just to remove them. Some functions are moved
> to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
> spapr_pci_vfio.c is arch specific, or else we need to introduce stub
> functions to those spapr functions moved.
[...]
> qemu code: https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
> Based on vfio-next, commit id: fd0e1c8bc1
I fetched this, and get several compile errors with Clang (but not with
GCC):
FAILED: libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o
clang -m64 -mcx16 -Ilibqemu-x86_64-softmmu.fa.p -I. -I.. -Itarget/i386 -I../target/i386 -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/pixman-1 -I/usr/include/capstone -I/usr/include/spice-server -I/usr/include/spice-1 -I/usr/include/cacard -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -I/usr/include/sysprof-4 -I/usr/include/nss3 -I/usr/include/nspr4 -I/usr/include/PCSC -fcolor-diagnostics -Wall -Winvalid-pch -Werror -std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem linux-headers -iquote . -iquote /work/armbru/qemu -iquote /work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -iquote /work/armbru/qemu/host/include/generic -iquote /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -fsanitize=fuzzer-no-link -fPIE -isystem../linux-headers -isystemlinux-headers -DNEED_CPU_H '-DCONFIG_TARGET="x86_64-softmmu-config-target.h"' '-DCONFIG_DEVICES="x86_64-softmmu-config-devices.h"' -MD -MQ libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o -MF libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o.d -o libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o -c ../hw/vfio/common.c
../hw/vfio/common.c:682:40: error: variable 'hostwin' is uninitialized when used here [-Werror,-Wuninitialized]
hwaddr pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
^~~~~~~
../hw/vfio/common.c:578:31: note: initialize the variable 'hostwin' to silence this warning
VFIOHostDMAWindow *hostwin;
^
= NULL
../hw/vfio/common.c:785:33: error: variable 'hostwin' is uninitialized when used here [-Werror,-Wuninitialized]
pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
^~~~~~~
../hw/vfio/common.c:783:35: note: initialize the variable 'hostwin' to silence this warning
VFIOHostDMAWindow *hostwin;
^
= NULL
2 errors generated.
FAILED: tests/unit/test-resv-mem.p/test-resv-mem.c.o
clang -m64 -mcx16 -Itests/unit/test-resv-mem.p -Itests/unit -I../tests/unit -I. -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -I/usr/include/sysprof-4 -fcolor-diagnostics -Wall -Winvalid-pch -Werror -std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem linux-headers -iquote . -iquote /work/armbru/qemu -iquote /work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -iquote /work/armbru/qemu/host/include/generic -iquote /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -fsanitize=fuzzer-no-link -fPIE -MD -MQ tests/unit/test-resv-mem.p/test-resv-mem.c.o -MF tests/unit/test-resv-mem.p/test-resv-mem.c.o.d -o tests/unit/test-resv-mem.p/test-resv-mem.c.o -c ../tests/unit/test-resv-mem.c
../tests/unit/test-resv-mem.c:42:9: error: variable 'i' set but not used [-Werror,-Wunused-but-set-variable]
int i = 0;
^
1 error generated.
Delete @i, please.
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
2023-10-26 10:30 ` [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
@ 2023-10-26 13:27 ` Markus Armbruster
2023-10-27 7:50 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Markus Armbruster @ 2023-10-26 13:27 UTC (permalink / raw)
To: Zhenzhong Duan
Cc: qemu-devel, alex.williamson, clg, jgg, nicolinc, joao.m.martins,
eric.auger, peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun,
chao.p.peng, Paolo Bonzini, Daniel P. Berrangé,
Eduardo Habkost, Eric Blake
Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
> From: Eric Auger <eric.auger@redhat.com>
>
> Introduce an iommufd object which allows the interaction
> with the host /dev/iommu device.
>
> The /dev/iommu can have been already pre-opened outside of qemu,
> in which case the fd can be passed directly along with the
> iommufd object:
>
> This allows the iommufd object to be shared accross several
> subsystems (VFIO, VDPA, ...). For example, libvirt would open
> the /dev/iommu once.
>
> If no fd is passed along with the iommufd object, the /dev/iommu
> is opened by the qemu code.
>
> The CONFIG_IOMMUFD option must be set to compile this new object.
>
> Suggested-by: Alex Williamson <alex.williamson@redhat.com>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> MAINTAINERS | 7 +
> qapi/qom.json | 20 +++
> include/sysemu/iommufd.h | 46 +++++++
> backends/iommufd-stub.c | 59 +++++++++
> backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
> backends/Kconfig | 4 +
> backends/meson.build | 5 +
> backends/trace-events | 12 ++
> qemu-options.hx | 13 ++
> 9 files changed, 434 insertions(+)
> create mode 100644 include/sysemu/iommufd.h
> create mode 100644 backends/iommufd-stub.c
> create mode 100644 backends/iommufd.c
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 7f9912baa0..7aa57ab16f 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -2109,6 +2109,13 @@ F: hw/vfio/ap.c
> F: docs/system/s390x/vfio-ap.rst
> L: qemu-s390x@nongnu.org
>
> +iommufd
> +M: Yi Liu <yi.l.liu@intel.com>
> +M: Eric Auger <eric.auger@redhat.com>
> +S: Supported
> +F: backends/iommufd.c
> +F: include/sysemu/iommufd.h
> +
> vhost
> M: Michael S. Tsirkin <mst@redhat.com>
> S: Supported
> diff --git a/qapi/qom.json b/qapi/qom.json
> index c53ef978ff..ef0b50f107 100644
> --- a/qapi/qom.json
> +++ b/qapi/qom.json
> @@ -794,6 +794,22 @@
> { 'struct': 'VfioUserServerProperties',
> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>
> +##
> +# @IOMMUFDProperties:
> +#
> +# Properties for iommufd objects.
> +#
> +# @fd: file descriptor name previously passed via 'getfd' command,
> +# which represents a pre-opened /dev/iommu. This allows the
Two spaces between sentences for consistency, please.
> +# iommufd object to be shared accross several subsystems
> +# (VFIO, VDPA, ...) and file descriptor to be shared with
Comma before "and file".
Either "the file descriptor", or "file descriptors".
> +# other process, e.g: DPDK.
e.g.
Alternatively "such as DPDK."
> +#
> +# Since: 8.2
> +##
> +{ 'struct': 'IOMMUFDProperties',
> + 'data': { '*fd': 'str' } }
@fd is optional. How does the iommufd object behave when @fd is absent?
> +
> ##
> # @RngProperties:
> #
> @@ -934,6 +950,8 @@
##
# @ObjectType:
None of the enum members are documented. I'm not asking you to fix that
now.
#
# Features:
#
# @unstable: Member @x-remote-object is experimental.
#
# Since: 6.0
##
{ 'enum': 'ObjectType',
'data': [
[...]
> 'input-barrier',
> { 'name': 'input-linux',
> 'if': 'CONFIG_LINUX' },
> + { 'name': 'iommufd',
> + 'if': 'CONFIG_IOMMUFD' },
Should struct IOMMUFDProperties also have 'if': 'CONFIG_IOMMUFD'?
> 'iothread',
> 'main-loop',
> { 'name': 'memory-backend-epc',
> @@ -1003,6 +1021,8 @@
> 'input-barrier': 'InputBarrierProperties',
> 'input-linux': { 'type': 'InputLinuxProperties',
> 'if': 'CONFIG_LINUX' },
> + 'iommufd': { 'type': 'IOMMUFDProperties',
> + 'if': 'CONFIG_IOMMUFD' },
> 'iothread': 'IothreadProperties',
> 'main-loop': 'MainLoopProperties',
> 'memory-backend-epc': { 'type': 'MemoryBackendEpcProperties',
[...]
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
@ 2023-10-26 14:23 ` Eric Farman
2023-10-27 9:19 ` Cédric Le Goater
1 sibling, 0 replies; 95+ messages in thread
From: Eric Farman @ 2023-10-26 14:23 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, clg, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, Cédric Le Goater,
David Gibson, Harsh Prateek Bora, Thomas Huth, Tony Krowiak,
Halil Pasic, Jason Herne, Matthew Rosato,
open list:sPAPR (pseries), open list:S390 general arch...
On Thu, 2023-10-26 at 18:30 +0800, Zhenzhong Duan wrote:
> With vfio_eeh_as_ok/vfio_eeh_as_op moved and made static,
> vfio.h becomes empty and is deleted.
>
> No functional changes intended.
>
> Suggested-by: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio.h | 7 ---
> hw/ppc/spapr_pci_vfio.c | 100
> +++++++++++++++++++++++++++++++++++++++-
> hw/vfio/ap.c | 1 -
> hw/vfio/ccw.c | 1 -
> hw/vfio/common.c | 1 -
> hw/vfio/container.c | 98 --------------------------------------
> -
> hw/vfio/helpers.c | 1 -
> 7 files changed, 99 insertions(+), 110 deletions(-)
> delete mode 100644 include/hw/vfio/vfio.h
Can't speak to the code movement, but removing the header file from
those places where it's not used is fine.
Acked-by: Eric Farman <farman@linux.ibm.com>
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 00/37] vfio: Adopt iommufd
2023-10-26 13:27 ` [PATCH v3 00/37] vfio: Adopt iommufd Markus Armbruster
@ 2023-10-27 6:17 ` Duan, Zhenzhong
2023-10-27 7:45 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-27 6:17 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, clg@redhat.com,
jgg@nvidia.com, nicolinc@nvidia.com, Martins, Joao,
eric.auger@redhat.com, peterx@redhat.com, jasowang@redhat.com,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P
Hi Markus,
>-----Original Message-----
>From: Markus Armbruster <armbru@redhat.com>
>Sent: Thursday, October 26, 2023 9:27 PM
>Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>
>Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>
>> Hi,
>>
>> Thanks all for giving guides and comments on previous series, here is
>> the v3 of pure iommufd support part.
>>
>> Based on Cédric's suggestion, this series includes an effort to remove
>> spapr code from container.c, now all spapr functions are moved to spapr.c
>> or spapr_pci_vfio.c, but there are still a few trival check on
>> VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce many
>> callbacks and duplicate code just to remove them. Some functions are moved
>> to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
>> spapr_pci_vfio.c is arch specific, or else we need to introduce stub
>> functions to those spapr functions moved.
>
>[...]
>
>> qemu code:
>https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
>> Based on vfio-next, commit id: fd0e1c8bc1
>
>I fetched this, and get several compile errors with Clang (but not with
>GCC):
>
> FAILED: libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o
> clang -m64 -mcx16 -Ilibqemu-x86_64-softmmu.fa.p -I. -I.. -Itarget/i386 -
>I../target/i386 -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/pixman-1 -
>I/usr/include/capstone -I/usr/include/spice-server -I/usr/include/spice-1 -
>I/usr/include/cacard -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>I/usr/include/sysprof-4 -I/usr/include/nss3 -I/usr/include/nspr4 -
>I/usr/include/PCSC -fcolor-diagnostics -Wall -Winvalid-pch -Werror -std=gnu11 -
>O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-prototypes -
>Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -
>Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body
>-Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-
>attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-
>negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-
>at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>/work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -
>iquote /work/armbru/qemu/host/include/generic -iquote
>/work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>fsanitize=fuzzer-no-link -fPIE -isystem../linux-headers -isystemlinux-headers -
>DNEED_CPU_H '-DCONFIG_TARGET="x86_64-softmmu-config-target.h"' '-
>DCONFIG_DEVICES="x86_64-softmmu-config-devices.h"' -MD -MQ libqemu-
>x86_64-softmmu.fa.p/hw_vfio_common.c.o -MF libqemu-x86_64-
>softmmu.fa.p/hw_vfio_common.c.o.d -o libqemu-x86_64-
>softmmu.fa.p/hw_vfio_common.c.o -c ../hw/vfio/common.c
> ../hw/vfio/common.c:682:40: error: variable 'hostwin' is uninitialized when
>used here [-Werror,-Wuninitialized]
> hwaddr pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
> ^~~~~~~
> ../hw/vfio/common.c:578:31: note: initialize the variable 'hostwin' to silence
>this warning
> VFIOHostDMAWindow *hostwin;
> ^
> = NULL
> ../hw/vfio/common.c:785:33: error: variable 'hostwin' is uninitialized when
>used here [-Werror,-Wuninitialized]
> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
> ^~~~~~~
> ../hw/vfio/common.c:783:35: note: initialize the variable 'hostwin' to silence
>this warning
> VFIOHostDMAWindow *hostwin;
> ^
> = NULL
> 2 errors generated.
> FAILED: tests/unit/test-resv-mem.p/test-resv-mem.c.o
> clang -m64 -mcx16 -Itests/unit/test-resv-mem.p -Itests/unit -I../tests/unit -I. -
>Iqapi -Itrace -Iui -Iui/shader -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>I/usr/include/sysprof-4 -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-
>prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -
>Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -
>Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -
>Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-dirs -
>Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-
>at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>/work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -
>iquote /work/armbru/qemu/host/include/generic -iquote
>/work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>fsanitize=fuzzer-no-link -fPIE -MD -MQ tests/unit/test-resv-mem.p/test-resv-
>mem.c.o -MF tests/unit/test-resv-mem.p/test-resv-mem.c.o.d -o tests/unit/test-
>resv-mem.p/test-resv-mem.c.o -c ../tests/unit/test-resv-mem.c
> ../tests/unit/test-resv-mem.c:42:9: error: variable 'i' set but not used [-
>Werror,-Wunused-but-set-variable]
> int i = 0;
> ^
> 1 error generated.
>
>Delete @i, please.
Thanks for your report, I'll fix it and build with both compilers next time.
About second error, I'll send a separate patch as it's unrelated to this series.
BRs.
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 00/37] vfio: Adopt iommufd
2023-10-27 6:17 ` Duan, Zhenzhong
@ 2023-10-27 7:45 ` Cédric Le Goater
2023-10-27 8:16 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 7:45 UTC (permalink / raw)
To: Duan, Zhenzhong, Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, jgg@nvidia.com,
nicolinc@nvidia.com, Martins, Joao, eric.auger@redhat.com,
peterx@redhat.com, jasowang@redhat.com, Tian, Kevin, Liu, Yi L,
Sun, Yi Y, Peng, Chao P
On 10/27/23 08:17, Duan, Zhenzhong wrote:
> Hi Markus,
>
>> -----Original Message-----
>> From: Markus Armbruster <armbru@redhat.com>
>> Sent: Thursday, October 26, 2023 9:27 PM
>> Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>>
>> Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>>
>>> Hi,
>>>
>>> Thanks all for giving guides and comments on previous series, here is
>>> the v3 of pure iommufd support part.
>>>
>>> Based on Cédric's suggestion, this series includes an effort to remove
>>> spapr code from container.c, now all spapr functions are moved to spapr.c
>>> or spapr_pci_vfio.c, but there are still a few trival check on
>>> VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce many
>>> callbacks and duplicate code just to remove them. Some functions are moved
>>> to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
>>> spapr_pci_vfio.c is arch specific, or else we need to introduce stub
>>> functions to those spapr functions moved.
>>
>> [...]
>>
>>> qemu code:
>> https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
>>> Based on vfio-next, commit id: fd0e1c8bc1
>>
>> I fetched this, and get several compile errors with Clang (but not with
>> GCC):
>>
>> FAILED: libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o
>> clang -m64 -mcx16 -Ilibqemu-x86_64-softmmu.fa.p -I. -I.. -Itarget/i386 -
>> I../target/i386 -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/pixman-1 -
>> I/usr/include/capstone -I/usr/include/spice-server -I/usr/include/spice-1 -
>> I/usr/include/cacard -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>> I/usr/include/sysprof-4 -I/usr/include/nss3 -I/usr/include/nspr4 -
>> I/usr/include/PCSC -fcolor-diagnostics -Wall -Winvalid-pch -Werror -std=gnu11 -
>> O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-prototypes -
>> Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -
>> Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body
>> -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-
>> attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-
>> negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-
>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>> /work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -
>> iquote /work/armbru/qemu/host/include/generic -iquote
>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>> fsanitize=fuzzer-no-link -fPIE -isystem../linux-headers -isystemlinux-headers -
>> DNEED_CPU_H '-DCONFIG_TARGET="x86_64-softmmu-config-target.h"' '-
>> DCONFIG_DEVICES="x86_64-softmmu-config-devices.h"' -MD -MQ libqemu-
>> x86_64-softmmu.fa.p/hw_vfio_common.c.o -MF libqemu-x86_64-
>> softmmu.fa.p/hw_vfio_common.c.o.d -o libqemu-x86_64-
>> softmmu.fa.p/hw_vfio_common.c.o -c ../hw/vfio/common.c
>> ../hw/vfio/common.c:682:40: error: variable 'hostwin' is uninitialized when
>> used here [-Werror,-Wuninitialized]
>> hwaddr pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>> ^~~~~~~
>> ../hw/vfio/common.c:578:31: note: initialize the variable 'hostwin' to silence
>> this warning
>> VFIOHostDMAWindow *hostwin;
>> ^
>> = NULL
>> ../hw/vfio/common.c:785:33: error: variable 'hostwin' is uninitialized when
>> used here [-Werror,-Wuninitialized]
>> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>> ^~~~~~~
>> ../hw/vfio/common.c:783:35: note: initialize the variable 'hostwin' to silence
>> this warning
>> VFIOHostDMAWindow *hostwin;
>> ^
>> = NULL
>> 2 errors generated.
>> FAILED: tests/unit/test-resv-mem.p/test-resv-mem.c.o
>> clang -m64 -mcx16 -Itests/unit/test-resv-mem.p -Itests/unit -I../tests/unit -I. -
>> Iqapi -Itrace -Iui -Iui/shader -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>> I/usr/include/sysprof-4 -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>> std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-
>> prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -
>> Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -
>> Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -
>> Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-dirs -
>> Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-not-
>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>> /work/armbru/qemu/include -iquote /work/armbru/qemu/host/include/x86_64 -
>> iquote /work/armbru/qemu/host/include/generic -iquote
>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>> fsanitize=fuzzer-no-link -fPIE -MD -MQ tests/unit/test-resv-mem.p/test-resv-
>> mem.c.o -MF tests/unit/test-resv-mem.p/test-resv-mem.c.o.d -o tests/unit/test-
>> resv-mem.p/test-resv-mem.c.o -c ../tests/unit/test-resv-mem.c
>> ../tests/unit/test-resv-mem.c:42:9: error: variable 'i' set but not used [-
>> Werror,-Wunused-but-set-variable]
>> int i = 0;
>> ^
>> 1 error generated.
>>
>> Delete @i, please.
>
> Thanks for your report, I'll fix it and build with both compilers next time.
a compile on a fedora rawhide also reports :
../hw/vfio/iommufd.c: In function ‘vfio_get_info_iova_range’:
../hw/vfio/iommufd.c:370:27: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
370 | info->allowed_iovas = (uint64_t)(info + 1);
| ^
../hw/vfio/iommufd.c:377:19: error: cast to pointer from integer of different size [-Werror=int-to-pointer-cast]
377 | iova_ranges = (struct iommu_iova_range *)info->allowed_iovas;
| ^
cc1: all warnings being treated as errors
> About second error, I'll send a separate patch as it's unrelated to this series.
Since this is a big series, I suggest that you take this patch in your build
environment,
diff --git a/meson.build b/meson.build
index 72a57288a026325d5ff753131c037e99f6f35c1a..c946cbef5b29e23475dc4cf345655e0466cbfade 100644
--- a/meson.build
+++ b/meson.build
@@ -462,6 +462,9 @@ warn_flags = [
'-Wno-tautological-type-limit-compare',
'-Wno-psabi',
'-Wno-gnu-variable-sized-type-not-at-end',
+ '-Wshadow=local',
+ '-Wno-error=shadow=local',
+ '-Wno-error=shadow=compatible-local',
]
if targetos != 'darwin'
Thanks,
C.
^ permalink raw reply related [flat|nested] 95+ messages in thread
* RE: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
2023-10-26 13:27 ` Markus Armbruster
@ 2023-10-27 7:50 ` Duan, Zhenzhong
2023-10-27 8:30 ` Markus Armbruster
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-27 7:50 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, clg@redhat.com,
jgg@nvidia.com, nicolinc@nvidia.com, Martins, Joao,
eric.auger@redhat.com, peterx@redhat.com, jasowang@redhat.com,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Paolo Bonzini,
Daniel P.Berrangé, Eduardo Habkost, Eric Blake
>-----Original Message-----
>From: Markus Armbruster <armbru@redhat.com>
>Sent: Thursday, October 26, 2023 9:28 PM
>Subject: Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
>
>Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> Introduce an iommufd object which allows the interaction
>> with the host /dev/iommu device.
>>
>> The /dev/iommu can have been already pre-opened outside of qemu,
>> in which case the fd can be passed directly along with the
>> iommufd object:
>>
>> This allows the iommufd object to be shared accross several
>> subsystems (VFIO, VDPA, ...). For example, libvirt would open
>> the /dev/iommu once.
>>
>> If no fd is passed along with the iommufd object, the /dev/iommu
>> is opened by the qemu code.
>>
>> The CONFIG_IOMMUFD option must be set to compile this new object.
>>
>> Suggested-by: Alex Williamson <alex.williamson@redhat.com>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>> ---
>> MAINTAINERS | 7 +
>> qapi/qom.json | 20 +++
>> include/sysemu/iommufd.h | 46 +++++++
>> backends/iommufd-stub.c | 59 +++++++++
>> backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
>> backends/Kconfig | 4 +
>> backends/meson.build | 5 +
>> backends/trace-events | 12 ++
>> qemu-options.hx | 13 ++
>> 9 files changed, 434 insertions(+)
>> create mode 100644 include/sysemu/iommufd.h
>> create mode 100644 backends/iommufd-stub.c
>> create mode 100644 backends/iommufd.c
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 7f9912baa0..7aa57ab16f 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -2109,6 +2109,13 @@ F: hw/vfio/ap.c
>> F: docs/system/s390x/vfio-ap.rst
>> L: qemu-s390x@nongnu.org
>>
>> +iommufd
>> +M: Yi Liu <yi.l.liu@intel.com>
>> +M: Eric Auger <eric.auger@redhat.com>
>> +S: Supported
>> +F: backends/iommufd.c
>> +F: include/sysemu/iommufd.h
>> +
>> vhost
>> M: Michael S. Tsirkin <mst@redhat.com>
>> S: Supported
>> diff --git a/qapi/qom.json b/qapi/qom.json
>> index c53ef978ff..ef0b50f107 100644
>> --- a/qapi/qom.json
>> +++ b/qapi/qom.json
>> @@ -794,6 +794,22 @@
>> { 'struct': 'VfioUserServerProperties',
>> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>>
>> +##
>> +# @IOMMUFDProperties:
>> +#
>> +# Properties for iommufd objects.
>> +#
>> +# @fd: file descriptor name previously passed via 'getfd' command,
>> +# which represents a pre-opened /dev/iommu. This allows the
>
>Two spaces between sentences for consistency, please.
Presume you mean " 'data': { '*fd': 'str' } }" line, not above line.
>
>> +# iommufd object to be shared accross several subsystems
>> +# (VFIO, VDPA, ...) and file descriptor to be shared with
>
>Comma before "and file".
>
>Either "the file descriptor", or "file descriptors".
>
>> +# other process, e.g: DPDK.
>
>e.g.
>
>Alternatively "such as DPDK."
Will fix.
>
>> +#
>> +# Since: 8.2
>> +##
>> +{ 'struct': 'IOMMUFDProperties',
>> + 'data': { '*fd': 'str' } }
>
>@fd is optional. How does the iommufd object behave when @fd is absent?
If no fd is passed along with the iommufd object, the /dev/iommu
is opened by the qemu code. Let me know if this also needs to be documented.
>
>> +
>> ##
>> # @RngProperties:
>> #
>> @@ -934,6 +950,8 @@
> ##
> # @ObjectType:
>
>None of the enum members are documented. I'm not asking you to fix that
>now.
>
> #
> # Features:
> #
> # @unstable: Member @x-remote-object is experimental.
> #
> # Since: 6.0
> ##
> { 'enum': 'ObjectType',
> 'data': [
> [...]
>> 'input-barrier',
>> { 'name': 'input-linux',
>> 'if': 'CONFIG_LINUX' },
>> + { 'name': 'iommufd',
>> + 'if': 'CONFIG_IOMMUFD' },
>
>Should struct IOMMUFDProperties also have 'if': 'CONFIG_IOMMUFD'?
Yes, thanks for point out, will fix.
BRs.
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 00/37] vfio: Adopt iommufd
2023-10-27 7:45 ` Cédric Le Goater
@ 2023-10-27 8:16 ` Duan, Zhenzhong
2023-10-27 13:43 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-27 8:16 UTC (permalink / raw)
To: Cédric Le Goater, Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, jgg@nvidia.com,
nicolinc@nvidia.com, Martins, Joao, eric.auger@redhat.com,
peterx@redhat.com, jasowang@redhat.com, Tian, Kevin, Liu, Yi L,
Sun, Yi Y, Peng, Chao P
Hi Cédric,
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 3:46 PM
>Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>
>On 10/27/23 08:17, Duan, Zhenzhong wrote:
>> Hi Markus,
>>
>>> -----Original Message-----
>>> From: Markus Armbruster <armbru@redhat.com>
>>> Sent: Thursday, October 26, 2023 9:27 PM
>>> Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>>>
>>> Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>>>
>>>> Hi,
>>>>
>>>> Thanks all for giving guides and comments on previous series, here is
>>>> the v3 of pure iommufd support part.
>>>>
>>>> Based on Cédric's suggestion, this series includes an effort to remove
>>>> spapr code from container.c, now all spapr functions are moved to spapr.c
>>>> or spapr_pci_vfio.c, but there are still a few trival check on
>>>> VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce
>many
>>>> callbacks and duplicate code just to remove them. Some functions are moved
>>>> to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
>>>> spapr_pci_vfio.c is arch specific, or else we need to introduce stub
>>>> functions to those spapr functions moved.
>>>
>>> [...]
>>>
>>>> qemu code:
>>> https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
>>>> Based on vfio-next, commit id: fd0e1c8bc1
>>>
>>> I fetched this, and get several compile errors with Clang (but not with
>>> GCC):
>>>
>>> FAILED: libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o
>>> clang -m64 -mcx16 -Ilibqemu-x86_64-softmmu.fa.p -I. -I.. -Itarget/i386 -
>>> I../target/i386 -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/pixman-1 -
>>> I/usr/include/capstone -I/usr/include/spice-server -I/usr/include/spice-1 -
>>> I/usr/include/cacard -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>>> I/usr/include/sysprof-4 -I/usr/include/nss3 -I/usr/include/nspr4 -
>>> I/usr/include/PCSC -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>std=gnu11 -
>>> O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-
>prototypes -
>>> Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -
>>> Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-
>body
>>> -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-
>>> attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-
>>> negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-
>not-
>>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>>> /work/armbru/qemu/include -iquote
>/work/armbru/qemu/host/include/x86_64 -
>>> iquote /work/armbru/qemu/host/include/generic -iquote
>>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>>> fsanitize=fuzzer-no-link -fPIE -isystem../linux-headers -isystemlinux-headers -
>>> DNEED_CPU_H '-DCONFIG_TARGET="x86_64-softmmu-config-target.h"' '-
>>> DCONFIG_DEVICES="x86_64-softmmu-config-devices.h"' -MD -MQ libqemu-
>>> x86_64-softmmu.fa.p/hw_vfio_common.c.o -MF libqemu-x86_64-
>>> softmmu.fa.p/hw_vfio_common.c.o.d -o libqemu-x86_64-
>>> softmmu.fa.p/hw_vfio_common.c.o -c ../hw/vfio/common.c
>>> ../hw/vfio/common.c:682:40: error: variable 'hostwin' is uninitialized when
>>> used here [-Werror,-Wuninitialized]
>>> hwaddr pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>>> ^~~~~~~
>>> ../hw/vfio/common.c:578:31: note: initialize the variable 'hostwin' to
>silence
>>> this warning
>>> VFIOHostDMAWindow *hostwin;
>>> ^
>>> = NULL
>>> ../hw/vfio/common.c:785:33: error: variable 'hostwin' is uninitialized when
>>> used here [-Werror,-Wuninitialized]
>>> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>>> ^~~~~~~
>>> ../hw/vfio/common.c:783:35: note: initialize the variable 'hostwin' to
>silence
>>> this warning
>>> VFIOHostDMAWindow *hostwin;
>>> ^
>>> = NULL
>>> 2 errors generated.
>>> FAILED: tests/unit/test-resv-mem.p/test-resv-mem.c.o
>>> clang -m64 -mcx16 -Itests/unit/test-resv-mem.p -Itests/unit -I../tests/unit -I.
>-
>>> Iqapi -Itrace -Iui -Iui/shader -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include
>-
>>> I/usr/include/sysprof-4 -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>>> std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -
>Wmissing-
>>> prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -
>>> Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-
>qualifiers -
>>> Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -
>>> Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-
>dirs -
>>> Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -
>Wno-
>>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-
>not-
>>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>>> /work/armbru/qemu/include -iquote
>/work/armbru/qemu/host/include/x86_64 -
>>> iquote /work/armbru/qemu/host/include/generic -iquote
>>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>>> fsanitize=fuzzer-no-link -fPIE -MD -MQ tests/unit/test-resv-mem.p/test-resv-
>>> mem.c.o -MF tests/unit/test-resv-mem.p/test-resv-mem.c.o.d -o
>tests/unit/test-
>>> resv-mem.p/test-resv-mem.c.o -c ../tests/unit/test-resv-mem.c
>>> ../tests/unit/test-resv-mem.c:42:9: error: variable 'i' set but not used [-
>>> Werror,-Wunused-but-set-variable]
>>> int i = 0;
>>> ^
>>> 1 error generated.
>>>
>>> Delete @i, please.
>>
>> Thanks for your report, I'll fix it and build with both compilers next time.
>
>a compile on a fedora rawhide also reports :
>
>../hw/vfio/iommufd.c: In function ‘vfio_get_info_iova_range’:
>../hw/vfio/iommufd.c:370:27: error: cast from pointer to integer of different size
>[-Werror=pointer-to-int-cast]
> 370 | info->allowed_iovas = (uint64_t)(info + 1);
> | ^
>../hw/vfio/iommufd.c:377:19: error: cast to pointer from integer of different size
>[-Werror=int-to-pointer-cast]
> 377 | iova_ranges = (struct iommu_iova_range *)info->allowed_iovas;
> | ^
>cc1: all warnings being treated as errors
I failed to understand this warning, is your env a 32bit fedora?
>
>
>> About second error, I'll send a separate patch as it's unrelated to this series.
>
>
>Since this is a big series, I suggest that you take this patch in your build
>environment,
Got it, will do, thanks for your suggestions.
BRs.
Zhenzhong
>
>
>diff --git a/meson.build b/meson.build
>index
>72a57288a026325d5ff753131c037e99f6f35c1a..c946cbef5b29e23475dc4cf3456
>55e0466cbfade 100644
>--- a/meson.build
>+++ b/meson.build
>@@ -462,6 +462,9 @@ warn_flags = [
> '-Wno-tautological-type-limit-compare',
> '-Wno-psabi',
> '-Wno-gnu-variable-sized-type-not-at-end',
>+ '-Wshadow=local',
>+ '-Wno-error=shadow=local',
>+ '-Wno-error=shadow=compatible-local',
> ]
>
> if targetos != 'darwin'
>
>
>Thanks,
>
>C.
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
2023-10-27 7:50 ` Duan, Zhenzhong
@ 2023-10-27 8:30 ` Markus Armbruster
2023-10-27 9:41 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Markus Armbruster @ 2023-10-27 8:30 UTC (permalink / raw)
To: Duan, Zhenzhong
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, clg@redhat.com,
jgg@nvidia.com, nicolinc@nvidia.com, Martins, Joao,
eric.auger@redhat.com, peterx@redhat.com, jasowang@redhat.com,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Paolo Bonzini,
Daniel P.Berrangé, Eduardo Habkost, Eric Blake
"Duan, Zhenzhong" <zhenzhong.duan@intel.com> writes:
>>-----Original Message-----
>> From: Markus Armbruster <armbru@redhat.com>
>> Sent: Thursday, October 26, 2023 9:28 PM
>> Subject: Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
>>
>> Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>>
>>> From: Eric Auger <eric.auger@redhat.com>
>>>
>>> Introduce an iommufd object which allows the interaction
>>> with the host /dev/iommu device.
>>>
>>> The /dev/iommu can have been already pre-opened outside of qemu,
>>> in which case the fd can be passed directly along with the
>>> iommufd object:
>>>
>>> This allows the iommufd object to be shared accross several
>>> subsystems (VFIO, VDPA, ...). For example, libvirt would open
>>> the /dev/iommu once.
>>>
>>> If no fd is passed along with the iommufd object, the /dev/iommu
>>> is opened by the qemu code.
>>>
>>> The CONFIG_IOMMUFD option must be set to compile this new object.
>>>
>>> Suggested-by: Alex Williamson <alex.williamson@redhat.com>
>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>>> ---
>>> MAINTAINERS | 7 +
>>> qapi/qom.json | 20 +++
>>> include/sysemu/iommufd.h | 46 +++++++
>>> backends/iommufd-stub.c | 59 +++++++++
>>> backends/iommufd.c | 268 +++++++++++++++++++++++++++++++++++++++
>>> backends/Kconfig | 4 +
>>> backends/meson.build | 5 +
>>> backends/trace-events | 12 ++
>>> qemu-options.hx | 13 ++
>>> 9 files changed, 434 insertions(+)
>>> create mode 100644 include/sysemu/iommufd.h
>>> create mode 100644 backends/iommufd-stub.c
>>> create mode 100644 backends/iommufd.c
>>>
>>> diff --git a/MAINTAINERS b/MAINTAINERS
>>> index 7f9912baa0..7aa57ab16f 100644
>>> --- a/MAINTAINERS
>>> +++ b/MAINTAINERS
>>> @@ -2109,6 +2109,13 @@ F: hw/vfio/ap.c
>>> F: docs/system/s390x/vfio-ap.rst
>>> L: qemu-s390x@nongnu.org
>>>
>>> +iommufd
>>> +M: Yi Liu <yi.l.liu@intel.com>
>>> +M: Eric Auger <eric.auger@redhat.com>
>>> +S: Supported
>>> +F: backends/iommufd.c
>>> +F: include/sysemu/iommufd.h
>>> +
>>> vhost
>>> M: Michael S. Tsirkin <mst@redhat.com>
>>> S: Supported
>>> diff --git a/qapi/qom.json b/qapi/qom.json
>>> index c53ef978ff..ef0b50f107 100644
>>> --- a/qapi/qom.json
>>> +++ b/qapi/qom.json
>>> @@ -794,6 +794,22 @@
>>> { 'struct': 'VfioUserServerProperties',
>>> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>>>
>>> +##
>>> +# @IOMMUFDProperties:
>>> +#
>>> +# Properties for iommufd objects.
>>> +#
>>> +# @fd: file descriptor name previously passed via 'getfd' command,
>>> +# which represents a pre-opened /dev/iommu. This allows the
>>
>> Two spaces between sentences for consistency, please.
>
> Presume you mean " 'data': { '*fd': 'str' } }" line, not above line.
I'd like you to format the doc comment like this:
# @fd: file descriptor name previously passed via 'getfd' command,
# which represents a pre-opened /dev/iommu. This allows the
Note the two spaces after the period in "/dev/iommu. This allows".
>>
>>> +# iommufd object to be shared accross several subsystems
>>> +# (VFIO, VDPA, ...) and file descriptor to be shared with
>>
>> Comma before "and file".
>>
>> Either "the file descriptor", or "file descriptors".
>>
>>> +# other process, e.g: DPDK.
>>
>> e.g.
>>
>> Alternatively "such as DPDK."
>
> Will fix.
>
>>
>>> +#
>>> +# Since: 8.2
>>> +##
>>> +{ 'struct': 'IOMMUFDProperties',
>>> + 'data': { '*fd': 'str' } }
>>
>> @fd is optional. How does the iommufd object behave when @fd is absent?
>
> If no fd is passed along with the iommufd object, the /dev/iommu
> is opened by the qemu code. Let me know if this also needs to be documented.
When something is optional, we pretty much always need to document
behavior when it's absent.
Ideally, behavior when absent is identical to behavior when present with
a certain default value. Then documenting the default value suffices.
We commonly do this like (default: VALUE).
>>> +
>>> ##
>>> # @RngProperties:
>>> #
>>> @@ -934,6 +950,8 @@
>> ##
>> # @ObjectType:
>>
>> None of the enum members are documented. I'm not asking you to fix that
>> now.
>>
>> #
>> # Features:
>> #
>> # @unstable: Member @x-remote-object is experimental.
>> #
>> # Since: 6.0
>> ##
>> { 'enum': 'ObjectType',
>> 'data': [
>> [...]
>>> 'input-barrier',
>>> { 'name': 'input-linux',
>>> 'if': 'CONFIG_LINUX' },
>>> + { 'name': 'iommufd',
>>> + 'if': 'CONFIG_IOMMUFD' },
>>
>> Should struct IOMMUFDProperties also have 'if': 'CONFIG_IOMMUFD'?
>
> Yes, thanks for point out, will fix.
You're welcome!
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
2023-10-26 14:23 ` Eric Farman
@ 2023-10-27 9:19 ` Cédric Le Goater
1 sibling, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 9:19 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, Cédric Le Goater,
David Gibson, Harsh Prateek Bora, Thomas Huth, Tony Krowiak,
Halil Pasic, Jason Herne, Eric Farman, Matthew Rosato,
open list:sPAPR (pseries), open list:S390 general arch...
On 10/26/23 12:30, Zhenzhong Duan wrote:
> With vfio_eeh_as_ok/vfio_eeh_as_op moved and made static,
> vfio.h becomes empty and is deleted.
>
> No functional changes intended.
>
> Suggested-by: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio.h | 7 ---
> hw/ppc/spapr_pci_vfio.c | 100 +++++++++++++++++++++++++++++++++++++++-
> hw/vfio/ap.c | 1 -
> hw/vfio/ccw.c | 1 -
> hw/vfio/common.c | 1 -
> hw/vfio/container.c | 98 ---------------------------------------
> hw/vfio/helpers.c | 1 -
> 7 files changed, 99 insertions(+), 110 deletions(-)
> delete mode 100644 include/hw/vfio/vfio.h
>
> diff --git a/include/hw/vfio/vfio.h b/include/hw/vfio/vfio.h
> deleted file mode 100644
> index 86248f5436..0000000000
> --- a/include/hw/vfio/vfio.h
> +++ /dev/null
> @@ -1,7 +0,0 @@
> -#ifndef HW_VFIO_H
> -#define HW_VFIO_H
> -
> -bool vfio_eeh_as_ok(AddressSpace *as);
> -int vfio_eeh_as_op(AddressSpace *as, uint32_t op);
> -
> -#endif
> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
> index 9016720547..f283f7e38d 100644
> --- a/hw/ppc/spapr_pci_vfio.c
> +++ b/hw/ppc/spapr_pci_vfio.c
> @@ -18,14 +18,112 @@
> */
>
> #include "qemu/osdep.h"
> +#include <sys/ioctl.h>
> #include <linux/vfio.h>
> #include "hw/ppc/spapr.h"
> #include "hw/pci-host/spapr.h"
> #include "hw/pci/msix.h"
> #include "hw/pci/pci_device.h"
> -#include "hw/vfio/vfio.h"
> +#include "hw/vfio/vfio-common.h"
> #include "qemu/error-report.h"
>
> +/*
> + * Interfaces for IBM EEH (Enhanced Error Handling)
> + */
> +static bool vfio_eeh_container_ok(VFIOContainer *container)
> +{
> + /*
> + * As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
> + * implementation is broken if there are multiple groups in a
> + * container. The hardware works in units of Partitionable
> + * Endpoints (== IOMMU groups) and the EEH operations naively
> + * iterate across all groups in the container, without any logic
> + * to make sure the groups have their state synchronized. For
> + * certain operations (ENABLE) that might be ok, until an error
> + * occurs, but for others (GET_STATE) it's clearly broken.
> + */
> +
> + /*
> + * XXX Once fixed kernels exist, test for them here
> + */
> +
> + if (QLIST_EMPTY(&container->group_list)) {
> + return false;
> + }
> +
> + if (QLIST_NEXT(QLIST_FIRST(&container->group_list), container_next)) {
> + return false;
> + }
> +
> + return true;
> +}
> +
> +static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
> +{
> + struct vfio_eeh_pe_op pe_op = {
> + .argsz = sizeof(pe_op),
> + .op = op,
> + };
> + int ret;
> +
> + if (!vfio_eeh_container_ok(container)) {
> + error_report("vfio/eeh: EEH_PE_OP 0x%x: "
> + "kernel requires a container with exactly one group", op);
> + return -EPERM;
> + }
> +
> + ret = ioctl(container->fd, VFIO_EEH_PE_OP, &pe_op);
> + if (ret < 0) {
> + error_report("vfio/eeh: EEH_PE_OP 0x%x failed: %m", op);
> + return -errno;
> + }
> +
> + return ret;
> +}
> +
> +static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
> +{
> + VFIOAddressSpace *space = vfio_get_address_space(as);
> + VFIOContainer *container = NULL;
> +
> + if (QLIST_EMPTY(&space->containers)) {
> + /* No containers to act on */
> + goto out;
> + }
> +
> + container = QLIST_FIRST(&space->containers);
> +
> + if (QLIST_NEXT(container, next)) {
> + /*
> + * We don't yet have logic to synchronize EEH state across
> + * multiple containers
> + */
> + container = NULL;
> + goto out;
> + }
> +
> +out:
> + vfio_put_address_space(space);
> + return container;
> +}
> +
> +static bool vfio_eeh_as_ok(AddressSpace *as)
> +{
> + VFIOContainer *container = vfio_eeh_as_container(as);
> +
> + return (container != NULL) && vfio_eeh_container_ok(container);
> +}
> +
> +static int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
> +{
> + VFIOContainer *container = vfio_eeh_as_container(as);
> +
> + if (!container) {
> + return -ENODEV;
> + }
> + return vfio_eeh_container_op(container, op);
> +}
> +
> bool spapr_phb_eeh_available(SpaprPhbState *sphb)
> {
> return vfio_eeh_as_ok(&sphb->iommu_as);
> diff --git a/hw/vfio/ap.c b/hw/vfio/ap.c
> index 5f257bffb9..bbf69ff55a 100644
> --- a/hw/vfio/ap.c
> +++ b/hw/vfio/ap.c
> @@ -14,7 +14,6 @@
> #include <linux/vfio.h>
> #include <sys/ioctl.h>
> #include "qapi/error.h"
> -#include "hw/vfio/vfio.h"
> #include "hw/vfio/vfio-common.h"
> #include "hw/s390x/ap-device.h"
> #include "qemu/error-report.h"
> diff --git a/hw/vfio/ccw.c b/hw/vfio/ccw.c
> index 6623ae237b..d857bb8d0f 100644
> --- a/hw/vfio/ccw.c
> +++ b/hw/vfio/ccw.c
> @@ -20,7 +20,6 @@
> #include <sys/ioctl.h>
>
> #include "qapi/error.h"
> -#include "hw/vfio/vfio.h"
> #include "hw/vfio/vfio-common.h"
> #include "hw/s390x/s390-ccw.h"
> #include "hw/s390x/vfio-ccw.h"
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 9c5c6433f2..e72055e752 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -26,7 +26,6 @@
> #include <linux/vfio.h>
>
> #include "hw/vfio/vfio-common.h"
> -#include "hw/vfio/vfio.h"
> #include "hw/vfio/pci.h"
> #include "exec/address-spaces.h"
> #include "exec/memory.h"
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index fc88222377..83c0f05bba 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -26,7 +26,6 @@
> #include <linux/vfio.h>
>
> #include "hw/vfio/vfio-common.h"
> -#include "hw/vfio/vfio.h"
> #include "exec/address-spaces.h"
> #include "exec/memory.h"
> #include "exec/ram_addr.h"
> @@ -1011,103 +1010,6 @@ static void vfio_put_base_device(VFIODevice *vbasedev)
> close(vbasedev->fd);
> }
>
> -/*
> - * Interfaces for IBM EEH (Enhanced Error Handling)
> - */
> -static bool vfio_eeh_container_ok(VFIOContainer *container)
> -{
> - /*
> - * As of 2016-03-04 (linux-4.5) the host kernel EEH/VFIO
> - * implementation is broken if there are multiple groups in a
> - * container. The hardware works in units of Partitionable
> - * Endpoints (== IOMMU groups) and the EEH operations naively
> - * iterate across all groups in the container, without any logic
> - * to make sure the groups have their state synchronized. For
> - * certain operations (ENABLE) that might be ok, until an error
> - * occurs, but for others (GET_STATE) it's clearly broken.
> - */
> -
> - /*
> - * XXX Once fixed kernels exist, test for them here
> - */
> -
> - if (QLIST_EMPTY(&container->group_list)) {
> - return false;
> - }
> -
> - if (QLIST_NEXT(QLIST_FIRST(&container->group_list), container_next)) {
> - return false;
> - }
> -
> - return true;
> -}
> -
> -static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
> -{
> - struct vfio_eeh_pe_op pe_op = {
> - .argsz = sizeof(pe_op),
> - .op = op,
> - };
> - int ret;
> -
> - if (!vfio_eeh_container_ok(container)) {
> - error_report("vfio/eeh: EEH_PE_OP 0x%x: "
> - "kernel requires a container with exactly one group", op);
> - return -EPERM;
> - }
> -
> - ret = ioctl(container->fd, VFIO_EEH_PE_OP, &pe_op);
> - if (ret < 0) {
> - error_report("vfio/eeh: EEH_PE_OP 0x%x failed: %m", op);
> - return -errno;
> - }
> -
> - return ret;
> -}
> -
> -static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
> -{
> - VFIOAddressSpace *space = vfio_get_address_space(as);
> - VFIOContainer *container = NULL;
> -
> - if (QLIST_EMPTY(&space->containers)) {
> - /* No containers to act on */
> - goto out;
> - }
> -
> - container = QLIST_FIRST(&space->containers);
> -
> - if (QLIST_NEXT(container, next)) {
> - /*
> - * We don't yet have logic to synchronize EEH state across
> - * multiple containers
> - */
> - container = NULL;
> - goto out;
> - }
> -
> -out:
> - vfio_put_address_space(space);
> - return container;
> -}
> -
> -bool vfio_eeh_as_ok(AddressSpace *as)
> -{
> - VFIOContainer *container = vfio_eeh_as_container(as);
> -
> - return (container != NULL) && vfio_eeh_container_ok(container);
> -}
> -
> -int vfio_eeh_as_op(AddressSpace *as, uint32_t op)
> -{
> - VFIOContainer *container = vfio_eeh_as_container(as);
> -
> - if (!container) {
> - return -ENODEV;
> - }
> - return vfio_eeh_container_op(container, op);
> -}
> -
> static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
> {
> char *tmp, group_path[PATH_MAX], *group_name;
> diff --git a/hw/vfio/helpers.c b/hw/vfio/helpers.c
> index 7e5da21b31..168847e7c5 100644
> --- a/hw/vfio/helpers.c
> +++ b/hw/vfio/helpers.c
> @@ -23,7 +23,6 @@
> #include <sys/ioctl.h>
>
> #include "hw/vfio/vfio-common.h"
> -#include "hw/vfio/vfio.h"
> #include "hw/hw.h"
> #include "trace.h"
> #include "qapi/error.h"
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c
2023-10-26 10:30 ` [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c Zhenzhong Duan
@ 2023-10-27 9:19 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 9:19 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, David Gibson,
Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> vfio_container_add/del_section_window are spapr specific functions,
> so move them into spapr.c to make container.c cleaner.
>
> No functional changes intended.
>
> Suggested-by: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> hw/vfio/container.c | 90 ---------------------------------------------
> hw/vfio/spapr.c | 90 +++++++++++++++++++++++++++++++++++++++++++++
> 2 files changed, 90 insertions(+), 90 deletions(-)
>
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 83c0f05bba..7a3f005d1b 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -20,9 +20,6 @@
>
> #include "qemu/osdep.h"
> #include <sys/ioctl.h>
> -#ifdef CONFIG_KVM
> -#include <linux/kvm.h>
> -#endif
> #include <linux/vfio.h>
>
> #include "hw/vfio/vfio-common.h"
> @@ -32,7 +29,6 @@
> #include "hw/hw.h"
> #include "qemu/error-report.h"
> #include "qemu/range.h"
> -#include "sysemu/kvm.h"
> #include "sysemu/reset.h"
> #include "trace.h"
> #include "qapi/error.h"
> @@ -204,92 +200,6 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> return -errno;
> }
>
> -int vfio_container_add_section_window(VFIOContainer *container,
> - MemoryRegionSection *section,
> - Error **errp)
> -{
> - VFIOHostDMAWindow *hostwin;
> - hwaddr pgsize = 0;
> - int ret;
> -
> - if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
> - return 0;
> - }
> -
> - /* For now intersections are not allowed, we may relax this later */
> - QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> - if (ranges_overlap(hostwin->min_iova,
> - hostwin->max_iova - hostwin->min_iova + 1,
> - section->offset_within_address_space,
> - int128_get64(section->size))) {
> - error_setg(errp,
> - "region [0x%"PRIx64",0x%"PRIx64"] overlaps with existing"
> - "host DMA window [0x%"PRIx64",0x%"PRIx64"]",
> - section->offset_within_address_space,
> - section->offset_within_address_space +
> - int128_get64(section->size) - 1,
> - hostwin->min_iova, hostwin->max_iova);
> - return -EINVAL;
> - }
> - }
> -
> - ret = vfio_spapr_create_window(container, section, &pgsize);
> - if (ret) {
> - error_setg_errno(errp, -ret, "Failed to create SPAPR window");
> - return ret;
> - }
> -
> - vfio_host_win_add(container, section->offset_within_address_space,
> - section->offset_within_address_space +
> - int128_get64(section->size) - 1, pgsize);
> -#ifdef CONFIG_KVM
> - if (kvm_enabled()) {
> - VFIOGroup *group;
> - IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
> - struct kvm_vfio_spapr_tce param;
> - struct kvm_device_attr attr = {
> - .group = KVM_DEV_VFIO_GROUP,
> - .attr = KVM_DEV_VFIO_GROUP_SET_SPAPR_TCE,
> - .addr = (uint64_t)(unsigned long)¶m,
> - };
> -
> - if (!memory_region_iommu_get_attr(iommu_mr, IOMMU_ATTR_SPAPR_TCE_FD,
> - ¶m.tablefd)) {
> - QLIST_FOREACH(group, &container->group_list, container_next) {
> - param.groupfd = group->fd;
> - if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
> - error_setg_errno(errp, errno,
> - "vfio: failed GROUP_SET_SPAPR_TCE for "
> - "KVM VFIO device %d and group fd %d",
> - param.tablefd, param.groupfd);
> - return -errno;
> - }
> - trace_vfio_spapr_group_attach(param.groupfd, param.tablefd);
> - }
> - }
> - }
> -#endif
> - return 0;
> -}
> -
> -void vfio_container_del_section_window(VFIOContainer *container,
> - MemoryRegionSection *section)
> -{
> - if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
> - return;
> - }
> -
> - vfio_spapr_remove_window(container,
> - section->offset_within_address_space);
> - if (vfio_host_win_del(container,
> - section->offset_within_address_space,
> - section->offset_within_address_space +
> - int128_get64(section->size) - 1) < 0) {
> - hw_error("%s: Cannot delete missing window at %"HWADDR_PRIx,
> - __func__, section->offset_within_address_space);
> - }
> -}
> -
> int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
> {
> int ret;
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 9ec1e95f6d..9a7517c042 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -11,6 +11,10 @@
> #include "qemu/osdep.h"
> #include <sys/ioctl.h>
> #include <linux/vfio.h>
> +#ifdef CONFIG_KVM
> +#include <linux/kvm.h>
> +#endif
> +#include "sysemu/kvm.h"
>
> #include "hw/vfio/vfio-common.h"
> #include "hw/hw.h"
> @@ -253,3 +257,89 @@ int vfio_spapr_remove_window(VFIOContainer *container,
>
> return 0;
> }
> +
> +int vfio_container_add_section_window(VFIOContainer *container,
> + MemoryRegionSection *section,
> + Error **errp)
> +{
> + VFIOHostDMAWindow *hostwin;
> + hwaddr pgsize = 0;
> + int ret;
> +
> + if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
> + return 0;
> + }
> +
> + /* For now intersections are not allowed, we may relax this later */
> + QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> + if (ranges_overlap(hostwin->min_iova,
> + hostwin->max_iova - hostwin->min_iova + 1,
> + section->offset_within_address_space,
> + int128_get64(section->size))) {
> + error_setg(errp,
> + "region [0x%"PRIx64",0x%"PRIx64"] overlaps with existing"
> + "host DMA window [0x%"PRIx64",0x%"PRIx64"]",
> + section->offset_within_address_space,
> + section->offset_within_address_space +
> + int128_get64(section->size) - 1,
> + hostwin->min_iova, hostwin->max_iova);
> + return -EINVAL;
> + }
> + }
> +
> + ret = vfio_spapr_create_window(container, section, &pgsize);
> + if (ret) {
> + error_setg_errno(errp, -ret, "Failed to create SPAPR window");
> + return ret;
> + }
> +
> + vfio_host_win_add(container, section->offset_within_address_space,
> + section->offset_within_address_space +
> + int128_get64(section->size) - 1, pgsize);
> +#ifdef CONFIG_KVM
> + if (kvm_enabled()) {
> + VFIOGroup *group;
> + IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
> + struct kvm_vfio_spapr_tce param;
> + struct kvm_device_attr attr = {
> + .group = KVM_DEV_VFIO_GROUP,
> + .attr = KVM_DEV_VFIO_GROUP_SET_SPAPR_TCE,
> + .addr = (uint64_t)(unsigned long)¶m,
> + };
> +
> + if (!memory_region_iommu_get_attr(iommu_mr, IOMMU_ATTR_SPAPR_TCE_FD,
> + ¶m.tablefd)) {
> + QLIST_FOREACH(group, &container->group_list, container_next) {
> + param.groupfd = group->fd;
> + if (ioctl(vfio_kvm_device_fd, KVM_SET_DEVICE_ATTR, &attr)) {
> + error_setg_errno(errp, errno,
> + "vfio: failed GROUP_SET_SPAPR_TCE for "
> + "KVM VFIO device %d and group fd %d",
> + param.tablefd, param.groupfd);
> + return -errno;
> + }
> + trace_vfio_spapr_group_attach(param.groupfd, param.tablefd);
> + }
> + }
> + }
> +#endif
> + return 0;
> +}
> +
> +void vfio_container_del_section_window(VFIOContainer *container,
> + MemoryRegionSection *section)
> +{
> + if (container->iommu_type != VFIO_SPAPR_TCE_v2_IOMMU) {
> + return;
> + }
> +
> + vfio_spapr_remove_window(container,
> + section->offset_within_address_space);
> + if (vfio_host_win_del(container,
> + section->offset_within_address_space,
> + section->offset_within_address_space +
> + int128_get64(section->size) - 1) < 0) {
> + hw_error("%s: Cannot delete missing window at %"HWADDR_PRIx,
> + __func__, section->offset_within_address_space);
> + }
> +}
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit into spapr.c
2023-10-26 10:30 ` [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit " Zhenzhong Duan
@ 2023-10-27 9:27 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 9:27 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, David Gibson,
Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> Move spapr specific init/deinit code into spapr.c and wrap
> them with vfio_spapr_container_init/deinit, this way footprint
> of spapr is further reduced, vfio_prereg_listener could also
> be made static.
>
> vfio_listener_release is unnecessary when prereg_listener is
> moved out, so have it removed.
>
> No functional changes intended.
>
> Suggested-by: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 4 +-
> hw/vfio/container.c | 82 +++++------------------------------
> hw/vfio/spapr.c | 80 +++++++++++++++++++++++++++++++++-
> 3 files changed, 94 insertions(+), 72 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 0c3d390e8b..8d1d4f0a89 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -225,11 +225,14 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
> int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> hwaddr iova, hwaddr size);
>
> +/* SPAPR specific */
> int vfio_container_add_section_window(VFIOContainer *container,
> MemoryRegionSection *section,
> Error **errp);
> void vfio_container_del_section_window(VFIOContainer *container,
> MemoryRegionSection *section);
> +bool vfio_spapr_container_init(VFIOContainer *container, Error **errp);
> +void vfio_spapr_container_deinit(VFIOContainer *container);
>
> void vfio_disable_irqindex(VFIODevice *vbasedev, int index);
> void vfio_unmask_single_irqindex(VFIODevice *vbasedev, int index);
> @@ -289,7 +292,6 @@ vfio_get_device_info_cap(struct vfio_device_info *info, uint16_t id);
> struct vfio_info_cap_header *
> vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
> #endif
> -extern const MemoryListener vfio_prereg_listener;
>
> int vfio_spapr_create_window(VFIOContainer *container,
> MemoryRegionSection *section,
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 7a3f005d1b..204b244b11 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -264,14 +264,6 @@ int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> return ret;
> }
>
> -static void vfio_listener_release(VFIOContainer *container)
> -{
> - memory_listener_unregister(&container->listener);
> - if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
> - memory_listener_unregister(&container->prereg_listener);
> - }
> -}
> -
> static struct vfio_info_cap_header *
> vfio_get_iommu_type1_info_cap(struct vfio_iommu_type1_info *info, uint16_t id)
> {
> @@ -612,69 +604,11 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> case VFIO_SPAPR_TCE_v2_IOMMU:
> case VFIO_SPAPR_TCE_IOMMU:
> {
> - struct vfio_iommu_spapr_tce_info info;
> - bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
> -
> - /*
> - * The host kernel code implementing VFIO_IOMMU_DISABLE is called
> - * when container fd is closed so we do not call it explicitly
> - * in this file.
> - */
> - if (!v2) {
> - ret = ioctl(fd, VFIO_IOMMU_ENABLE);
> - if (ret) {
> - error_setg_errno(errp, errno, "failed to enable container");
> - ret = -errno;
> - goto enable_discards_exit;
> - }
> - } else {
> - container->prereg_listener = vfio_prereg_listener;
> -
> - memory_listener_register(&container->prereg_listener,
> - &address_space_memory);
> - if (container->error) {
> - memory_listener_unregister(&container->prereg_listener);
> - ret = -1;
> - error_propagate_prepend(errp, container->error,
> - "RAM memory listener initialization failed: ");
> - goto enable_discards_exit;
> - }
> - }
> -
> - info.argsz = sizeof(info);
> - ret = ioctl(fd, VFIO_IOMMU_SPAPR_TCE_GET_INFO, &info);
> + ret = vfio_spapr_container_init(container, errp);
> if (ret) {
> - error_setg_errno(errp, errno,
> - "VFIO_IOMMU_SPAPR_TCE_GET_INFO failed");
> - ret = -errno;
> - if (v2) {
> - memory_listener_unregister(&container->prereg_listener);
> - }
> goto enable_discards_exit;
> }
> -
> - if (v2) {
> - container->pgsizes = info.ddw.pgsizes;
> - /*
> - * There is a default window in just created container.
> - * To make region_add/del simpler, we better remove this
> - * window now and let those iommu_listener callbacks
> - * create/remove them when needed.
> - */
> - ret = vfio_spapr_remove_window(container, info.dma32_window_start);
> - if (ret) {
> - error_setg_errno(errp, -ret,
> - "failed to remove existing window");
> - goto enable_discards_exit;
> - }
> - } else {
> - /* The default table uses 4K pages */
> - container->pgsizes = 0x1000;
> - vfio_host_win_add(container, info.dma32_window_start,
> - info.dma32_window_start +
> - info.dma32_window_size - 1,
> - 0x1000);
> - }
> + break;
> }
> }
>
> @@ -704,7 +638,11 @@ listener_release_exit:
> QLIST_REMOVE(group, container_next);
> QLIST_REMOVE(container, next);
> vfio_kvm_device_del_group(group);
> - vfio_listener_release(container);
> + memory_listener_unregister(&container->listener);
> + if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
> + container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
> + vfio_spapr_container_deinit(container);
> + }
>
> enable_discards_exit:
> vfio_ram_block_discard_disable(container, false);
> @@ -734,7 +672,11 @@ static void vfio_disconnect_container(VFIOGroup *group)
> * group.
> */
> if (QLIST_EMPTY(&container->group_list)) {
> - vfio_listener_release(container);
> + memory_listener_unregister(&container->listener);
> + if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
> + container->iommu_type == VFIO_SPAPR_TCE_IOMMU) {
> + vfio_spapr_container_deinit(container);
> + }
> }
>
> if (ioctl(group->fd, VFIO_GROUP_UNSET_CONTAINER, &container->fd)) {
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 9a7517c042..fd0613dfde 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -15,6 +15,7 @@
> #include <linux/kvm.h>
> #endif
> #include "sysemu/kvm.h"
> +#include "exec/address-spaces.h"
>
> #include "hw/vfio/vfio-common.h"
> #include "hw/hw.h"
> @@ -139,7 +140,7 @@ static void vfio_prereg_listener_region_del(MemoryListener *listener,
> trace_vfio_prereg_unregister(reg.vaddr, reg.size, ret ? -errno : 0);
> }
>
> -const MemoryListener vfio_prereg_listener = {
> +static const MemoryListener vfio_prereg_listener = {
> .name = "vfio-pre-reg",
> .region_add = vfio_prereg_listener_region_add,
> .region_del = vfio_prereg_listener_region_del,
> @@ -343,3 +344,80 @@ void vfio_container_del_section_window(VFIOContainer *container,
> __func__, section->offset_within_address_space);
> }
> }
> +
> +bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> +{
> +
> + struct vfio_iommu_spapr_tce_info info;
> + bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
> + int ret, fd = container->fd;
> +
> + /*
> + * The host kernel code implementing VFIO_IOMMU_DISABLE is called
> + * when container fd is closed so we do not call it explicitly
> + * in this file.
> + */
> + if (!v2) {
> + ret = ioctl(fd, VFIO_IOMMU_ENABLE);
> + if (ret) {
> + error_setg_errno(errp, errno, "failed to enable container");
> + return -errno;
> + }
> + } else {
> + container->prereg_listener = vfio_prereg_listener;
> +
> + memory_listener_register(&container->prereg_listener,
> + &address_space_memory);
> + if (container->error) {
> + ret = -1;
> + error_propagate_prepend(errp, container->error,
> + "RAM memory listener initialization failed: ");
> + goto listener_unregister_exit;
> + }
> + }
> +
> + info.argsz = sizeof(info);
> + ret = ioctl(fd, VFIO_IOMMU_SPAPR_TCE_GET_INFO, &info);
> + if (ret) {
> + error_setg_errno(errp, errno,
> + "VFIO_IOMMU_SPAPR_TCE_GET_INFO failed");
> + ret = -errno;
> + goto listener_unregister_exit;
> + }
> +
> + if (v2) {
> + container->pgsizes = info.ddw.pgsizes;
> + /*
> + * There is a default window in just created container.
> + * To make region_add/del simpler, we better remove this
> + * window now and let those iommu_listener callbacks
> + * create/remove them when needed.
> + */
> + ret = vfio_spapr_remove_window(container, info.dma32_window_start);
> + if (ret) {
> + error_setg_errno(errp, -ret,
> + "failed to remove existing window");
> + goto listener_unregister_exit;
> + }
> + } else {
> + /* The default table uses 4K pages */
> + container->pgsizes = 0x1000;
> + vfio_host_win_add(container, info.dma32_window_start,
> + info.dma32_window_start +
> + info.dma32_window_size - 1,
> + 0x1000);
> + }
> +
> +listener_unregister_exit:
> + if (v2) {
> + memory_listener_unregister(&container->prereg_listener);
> + }
> + return ret;
> +}
> +
> +void vfio_spapr_container_deinit(VFIOContainer *container)
> +{
> + if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
> + memory_listener_unregister(&container->prereg_listener);
> + }
> +}
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static
2023-10-26 10:30 ` [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static Zhenzhong Duan
@ 2023-10-27 9:27 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 9:27 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, Cédric Le Goater,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> vfio_spapr_create_window calls vfio_spapr_remove_window,
> With reoder of definition of the two, we can make
> vfio_spapr_create/remove_window static.
>
> No functional changes intended.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 6 -----
> hw/vfio/spapr.c | 48 +++++++++++++++++------------------
> 2 files changed, 24 insertions(+), 30 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 8d1d4f0a89..f3174b3c5c 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -293,12 +293,6 @@ struct vfio_info_cap_header *
> vfio_get_cap(void *ptr, uint32_t cap_offset, uint16_t id);
> #endif
>
> -int vfio_spapr_create_window(VFIOContainer *container,
> - MemoryRegionSection *section,
> - hwaddr *pgsize);
> -int vfio_spapr_remove_window(VFIOContainer *container,
> - hwaddr offset_within_address_space);
> -
> bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
> void vfio_migration_exit(VFIODevice *vbasedev);
>
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index fd0613dfde..43adbdb7b3 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -146,9 +146,30 @@ static const MemoryListener vfio_prereg_listener = {
> .region_del = vfio_prereg_listener_region_del,
> };
>
> -int vfio_spapr_create_window(VFIOContainer *container,
> - MemoryRegionSection *section,
> - hwaddr *pgsize)
> +static int vfio_spapr_remove_window(VFIOContainer *container,
> + hwaddr offset_within_address_space)
> +{
> + struct vfio_iommu_spapr_tce_remove remove = {
> + .argsz = sizeof(remove),
> + .start_addr = offset_within_address_space,
> + };
> + int ret;
> +
> + ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
> + if (ret) {
> + error_report("Failed to remove window at %"PRIx64,
> + (uint64_t)remove.start_addr);
> + return -errno;
> + }
> +
> + trace_vfio_spapr_remove_window(offset_within_address_space);
> +
> + return 0;
> +}
> +
> +static int vfio_spapr_create_window(VFIOContainer *container,
> + MemoryRegionSection *section,
> + hwaddr *pgsize)
> {
> int ret = 0;
> IOMMUMemoryRegion *iommu_mr = IOMMU_MEMORY_REGION(section->mr);
> @@ -238,27 +259,6 @@ int vfio_spapr_create_window(VFIOContainer *container,
> return 0;
> }
>
> -int vfio_spapr_remove_window(VFIOContainer *container,
> - hwaddr offset_within_address_space)
> -{
> - struct vfio_iommu_spapr_tce_remove remove = {
> - .argsz = sizeof(remove),
> - .start_addr = offset_within_address_space,
> - };
> - int ret;
> -
> - ret = ioctl(container->fd, VFIO_IOMMU_SPAPR_TCE_REMOVE, &remove);
> - if (ret) {
> - error_report("Failed to remove window at %"PRIx64,
> - (uint64_t)remove.start_addr);
> - return -errno;
> - }
> -
> - trace_vfio_spapr_remove_window(offset_within_address_space);
> -
> - return 0;
> -}
> -
> int vfio_container_add_section_window(VFIOContainer *container,
> MemoryRegionSection *section,
> Error **errp)
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c
2023-10-26 10:30 ` [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c Zhenzhong Duan
@ 2023-10-27 9:30 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 9:30 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, David Gibson,
Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> Only spapr supports a customed host window list, other vfio driver
> assume 64bit host window. So remove the check in listener callback
> and move vfio_host_win_add/del into spapr.c and make static.
>
> Suggested-by: Alex Williamson <alex.williamson@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 5 ---
> hw/vfio/common.c | 64 -----------------------------------
> hw/vfio/container.c | 16 ---------
> hw/vfio/spapr.c | 38 +++++++++++++++++++++
> 4 files changed, 38 insertions(+), 85 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index f3174b3c5c..b9c7a7e588 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -207,11 +207,6 @@ typedef struct {
> hwaddr pages;
> } VFIOBitmap;
>
> -void vfio_host_win_add(VFIOContainer *container,
> - hwaddr min_iova, hwaddr max_iova,
> - uint64_t iova_pgsizes);
> -int vfio_host_win_del(VFIOContainer *container, hwaddr min_iova,
> - hwaddr max_iova);
> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index e72055e752..0ebf4d9256 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -245,44 +245,6 @@ bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
> return true;
> }
>
> -void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
> - hwaddr max_iova, uint64_t iova_pgsizes)
> -{
> - VFIOHostDMAWindow *hostwin;
> -
> - QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> - if (ranges_overlap(hostwin->min_iova,
> - hostwin->max_iova - hostwin->min_iova + 1,
> - min_iova,
> - max_iova - min_iova + 1)) {
> - hw_error("%s: Overlapped IOMMU are not enabled", __func__);
> - }
> - }
> -
> - hostwin = g_malloc0(sizeof(*hostwin));
> -
> - hostwin->min_iova = min_iova;
> - hostwin->max_iova = max_iova;
> - hostwin->iova_pgsizes = iova_pgsizes;
> - QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
> -}
> -
> -int vfio_host_win_del(VFIOContainer *container,
> - hwaddr min_iova, hwaddr max_iova)
> -{
> - VFIOHostDMAWindow *hostwin;
> -
> - QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> - if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
> - QLIST_REMOVE(hostwin, hostwin_next);
> - g_free(hostwin);
> - return 0;
> - }
> - }
> -
> - return -1;
> -}
> -
> static bool vfio_listener_skipped_section(MemoryRegionSection *section)
> {
> return (!memory_region_is_ram(section->mr) &&
> @@ -531,22 +493,6 @@ static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
> g_free(vrdl);
> }
>
> -static VFIOHostDMAWindow *vfio_find_hostwin(VFIOContainer *container,
> - hwaddr iova, hwaddr end)
> -{
> - VFIOHostDMAWindow *hostwin;
> - bool hostwin_found = false;
> -
> - QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> - if (hostwin->min_iova <= iova && end <= hostwin->max_iova) {
> - hostwin_found = true;
> - break;
> - }
> - }
> -
> - return hostwin_found ? hostwin : NULL;
> -}
> -
> static bool vfio_known_safe_misalignment(MemoryRegionSection *section)
> {
> MemoryRegion *mr = section->mr;
> @@ -647,13 +593,6 @@ static void vfio_listener_region_add(MemoryListener *listener,
> goto fail;
> }
>
> - hostwin = vfio_find_hostwin(container, iova, end);
> - if (!hostwin) {
> - error_setg(&err, "Container %p can't map guest IOVA region"
> - " 0x%"HWADDR_PRIx"..0x%"HWADDR_PRIx, container, iova, end);
> - goto fail;
> - }
> -
> memory_region_ref(section->mr);
>
> if (memory_region_is_iommu(section->mr)) {
> @@ -835,9 +774,6 @@ static void vfio_listener_region_del(MemoryListener *listener,
> hwaddr pgmask;
> VFIOHostDMAWindow *hostwin;
>
> - hostwin = vfio_find_hostwin(container, iova, end);
> - assert(hostwin); /* or region_add() would have failed */
> -
> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
> try_unmap = !((iova & pgmask) || (int128_get64(llsize) & pgmask));
> } else if (memory_region_has_ram_discard_manager(section->mr)) {
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 204b244b11..242010036a 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -551,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container->dma_max_mappings = 0;
> container->iova_ranges = NULL;
> QLIST_INIT(&container->giommu_list);
> - QLIST_INIT(&container->hostwin_list);
> QLIST_INIT(&container->vrdl_list);
>
> ret = vfio_init_container(container, group->fd, errp);
> @@ -591,14 +590,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
>
> vfio_get_iommu_info_migration(container, info);
> g_free(info);
> -
> - /*
> - * FIXME: We should parse VFIO_IOMMU_TYPE1_INFO_CAP_IOVA_RANGE
> - * information to get the actual window extent rather than assume
> - * a 64-bit IOVA address space.
> - */
> - vfio_host_win_add(container, 0, (hwaddr)-1, container->pgsizes);
> -
> break;
> }
> case VFIO_SPAPR_TCE_v2_IOMMU:
> @@ -687,7 +678,6 @@ static void vfio_disconnect_container(VFIOGroup *group)
> if (QLIST_EMPTY(&container->group_list)) {
> VFIOAddressSpace *space = container->space;
> VFIOGuestIOMMU *giommu, *tmp;
> - VFIOHostDMAWindow *hostwin, *next;
>
> QLIST_REMOVE(container, next);
>
> @@ -698,12 +688,6 @@ static void vfio_disconnect_container(VFIOGroup *group)
> g_free(giommu);
> }
>
> - QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
> - next) {
> - QLIST_REMOVE(hostwin, hostwin_next);
> - g_free(hostwin);
> - }
> -
> trace_vfio_disconnect_container(container->fd);
> close(container->fd);
> vfio_free_container(container);
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 43adbdb7b3..3495737ab2 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -146,6 +146,44 @@ static const MemoryListener vfio_prereg_listener = {
> .region_del = vfio_prereg_listener_region_del,
> };
>
> +static void vfio_host_win_add(VFIOContainer *container, hwaddr min_iova,
> + hwaddr max_iova, uint64_t iova_pgsizes)
> +{
> + VFIOHostDMAWindow *hostwin;
> +
> + QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> + if (ranges_overlap(hostwin->min_iova,
> + hostwin->max_iova - hostwin->min_iova + 1,
> + min_iova,
> + max_iova - min_iova + 1)) {
> + hw_error("%s: Overlapped IOMMU are not enabled", __func__);
> + }
> + }
> +
> + hostwin = g_malloc0(sizeof(*hostwin));
> +
> + hostwin->min_iova = min_iova;
> + hostwin->max_iova = max_iova;
> + hostwin->iova_pgsizes = iova_pgsizes;
> + QLIST_INSERT_HEAD(&container->hostwin_list, hostwin, hostwin_next);
> +}
> +
> +static int vfio_host_win_del(VFIOContainer *container,
> + hwaddr min_iova, hwaddr max_iova)
> +{
> + VFIOHostDMAWindow *hostwin;
> +
> + QLIST_FOREACH(hostwin, &container->hostwin_list, hostwin_next) {
> + if (hostwin->min_iova == min_iova && hostwin->max_iova == max_iova) {
> + QLIST_REMOVE(hostwin, hostwin_next);
> + g_free(hostwin);
> + return 0;
> + }
> + }
> +
> + return -1;
> +}
> +
> static int vfio_spapr_remove_window(VFIOContainer *container,
> hwaddr offset_within_address_space)
> {
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
2023-10-27 8:30 ` Markus Armbruster
@ 2023-10-27 9:41 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-27 9:41 UTC (permalink / raw)
To: Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, clg@redhat.com,
jgg@nvidia.com, nicolinc@nvidia.com, Martins, Joao,
eric.auger@redhat.com, peterx@redhat.com, jasowang@redhat.com,
Tian, Kevin, Liu, Yi L, Sun, Yi Y, Peng, Chao P, Paolo Bonzini,
Daniel P.Berrangé, Eduardo Habkost, Eric Blake
>-----Original Message-----
>From: Markus Armbruster <armbru@redhat.com>
>Sent: Friday, October 27, 2023 4:31 PM
>Subject: Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object
>
>"Duan, Zhenzhong" <zhenzhong.duan@intel.com> writes:
>
>>>-----Original Message-----
>>> From: Markus Armbruster <armbru@redhat.com>
>>> Sent: Thursday, October 26, 2023 9:28 PM
>>> Subject: Re: [PATCH v3 24/37] backends/iommufd: Introduce the iommufd
>object
>>>
>>> Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>>>
>>>> From: Eric Auger <eric.auger@redhat.com>
>>>>
>>>> Introduce an iommufd object which allows the interaction
>>>> with the host /dev/iommu device.
>>>>
>>>> The /dev/iommu can have been already pre-opened outside of qemu,
>>>> in which case the fd can be passed directly along with the
>>>> iommufd object:
>>>>
>>>> This allows the iommufd object to be shared accross several
>>>> subsystems (VFIO, VDPA, ...). For example, libvirt would open
>>>> the /dev/iommu once.
>>>>
>>>> If no fd is passed along with the iommufd object, the /dev/iommu
>>>> is opened by the qemu code.
>>>>
>>>> The CONFIG_IOMMUFD option must be set to compile this new object.
>>>>
>>>> Suggested-by: Alex Williamson <alex.williamson@redhat.com>
>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>>>> ---
>>>> MAINTAINERS | 7 +
>>>> qapi/qom.json | 20 +++
>>>> include/sysemu/iommufd.h | 46 +++++++
>>>> backends/iommufd-stub.c | 59 +++++++++
>>>> backends/iommufd.c | 268
>+++++++++++++++++++++++++++++++++++++++
>>>> backends/Kconfig | 4 +
>>>> backends/meson.build | 5 +
>>>> backends/trace-events | 12 ++
>>>> qemu-options.hx | 13 ++
>>>> 9 files changed, 434 insertions(+)
>>>> create mode 100644 include/sysemu/iommufd.h
>>>> create mode 100644 backends/iommufd-stub.c
>>>> create mode 100644 backends/iommufd.c
>>>>
>>>> diff --git a/MAINTAINERS b/MAINTAINERS
>>>> index 7f9912baa0..7aa57ab16f 100644
>>>> --- a/MAINTAINERS
>>>> +++ b/MAINTAINERS
>>>> @@ -2109,6 +2109,13 @@ F: hw/vfio/ap.c
>>>> F: docs/system/s390x/vfio-ap.rst
>>>> L: qemu-s390x@nongnu.org
>>>>
>>>> +iommufd
>>>> +M: Yi Liu <yi.l.liu@intel.com>
>>>> +M: Eric Auger <eric.auger@redhat.com>
>>>> +S: Supported
>>>> +F: backends/iommufd.c
>>>> +F: include/sysemu/iommufd.h
>>>> +
>>>> vhost
>>>> M: Michael S. Tsirkin <mst@redhat.com>
>>>> S: Supported
>>>> diff --git a/qapi/qom.json b/qapi/qom.json
>>>> index c53ef978ff..ef0b50f107 100644
>>>> --- a/qapi/qom.json
>>>> +++ b/qapi/qom.json
>>>> @@ -794,6 +794,22 @@
>>>> { 'struct': 'VfioUserServerProperties',
>>>> 'data': { 'socket': 'SocketAddress', 'device': 'str' } }
>>>>
>>>> +##
>>>> +# @IOMMUFDProperties:
>>>> +#
>>>> +# Properties for iommufd objects.
>>>> +#
>>>> +# @fd: file descriptor name previously passed via 'getfd' command,
>>>> +# which represents a pre-opened /dev/iommu. This allows the
>>>
>>> Two spaces between sentences for consistency, please.
>>
>> Presume you mean " 'data': { '*fd': 'str' } }" line, not above line.
>
>I'd like you to format the doc comment like this:
>
>
> # @fd: file descriptor name previously passed via 'getfd' command,
> # which represents a pre-opened /dev/iommu. This allows the
>
>Note the two spaces after the period in "/dev/iommu. This allows".
Understood.
>
>>>
>>>> +# iommufd object to be shared accross several subsystems
>>>> +# (VFIO, VDPA, ...) and file descriptor to be shared with
>>>
>>> Comma before "and file".
>>>
>>> Either "the file descriptor", or "file descriptors".
>>>
>>>> +# other process, e.g: DPDK.
>>>
>>> e.g.
>>>
>>> Alternatively "such as DPDK."
>>
>> Will fix.
>>
>>>
>>>> +#
>>>> +# Since: 8.2
>>>> +##
>>>> +{ 'struct': 'IOMMUFDProperties',
>>>> + 'data': { '*fd': 'str' } }
>>>
>>> @fd is optional. How does the iommufd object behave when @fd is absent?
>>
>> If no fd is passed along with the iommufd object, the /dev/iommu
>> is opened by the qemu code. Let me know if this also needs to be documented.
>
>When something is optional, we pretty much always need to document
>behavior when it's absent.
>
>Ideally, behavior when absent is identical to behavior when present with
>a certain default value. Then documenting the default value suffices.
>We commonly do this like (default: VALUE).
Got it.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 00/37] vfio: Adopt iommufd
2023-10-27 8:16 ` Duan, Zhenzhong
@ 2023-10-27 13:43 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 13:43 UTC (permalink / raw)
To: Duan, Zhenzhong, Markus Armbruster
Cc: qemu-devel@nongnu.org, alex.williamson@redhat.com, jgg@nvidia.com,
nicolinc@nvidia.com, Martins, Joao, eric.auger@redhat.com,
peterx@redhat.com, jasowang@redhat.com, Tian, Kevin, Liu, Yi L,
Sun, Yi Y, Peng, Chao P
On 10/27/23 10:16, Duan, Zhenzhong wrote:
> Hi Cédric,
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Friday, October 27, 2023 3:46 PM
>> Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>>
>> On 10/27/23 08:17, Duan, Zhenzhong wrote:
>>> Hi Markus,
>>>
>>>> -----Original Message-----
>>>> From: Markus Armbruster <armbru@redhat.com>
>>>> Sent: Thursday, October 26, 2023 9:27 PM
>>>> Subject: Re: [PATCH v3 00/37] vfio: Adopt iommufd
>>>>
>>>> Zhenzhong Duan <zhenzhong.duan@intel.com> writes:
>>>>
>>>>> Hi,
>>>>>
>>>>> Thanks all for giving guides and comments on previous series, here is
>>>>> the v3 of pure iommufd support part.
>>>>>
>>>>> Based on Cédric's suggestion, this series includes an effort to remove
>>>>> spapr code from container.c, now all spapr functions are moved to spapr.c
>>>>> or spapr_pci_vfio.c, but there are still a few trival check on
>>>>> VFIO_SPAPR_TCE_*_IOMMU which I am not sure if deserved to introduce
>> many
>>>>> callbacks and duplicate code just to remove them. Some functions are moved
>>>>> to spapr.c instead of spapr_pci_vfio.c to avoid compile issue because
>>>>> spapr_pci_vfio.c is arch specific, or else we need to introduce stub
>>>>> functions to those spapr functions moved.
>>>>
>>>> [...]
>>>>
>>>>> qemu code:
>>>> https://github.com/yiliu1765/qemu/commits/zhenzhong/iommufd_cdev_v3
>>>>> Based on vfio-next, commit id: fd0e1c8bc1
>>>>
>>>> I fetched this, and get several compile errors with Clang (but not with
>>>> GCC):
>>>>
>>>> FAILED: libqemu-x86_64-softmmu.fa.p/hw_vfio_common.c.o
>>>> clang -m64 -mcx16 -Ilibqemu-x86_64-softmmu.fa.p -I. -I.. -Itarget/i386 -
>>>> I../target/i386 -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/pixman-1 -
>>>> I/usr/include/capstone -I/usr/include/spice-server -I/usr/include/spice-1 -
>>>> I/usr/include/cacard -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include -
>>>> I/usr/include/sysprof-4 -I/usr/include/nss3 -I/usr/include/nspr4 -
>>>> I/usr/include/PCSC -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>> std=gnu11 -
>>>> O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -Wmissing-
>> prototypes -
>>>> Wstrict-prototypes -Wredundant-decls -Wold-style-definition -Wtype-limits -
>>>> Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-
>> body
>>>> -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wmissing-format-
>>>> attribute -Wno-initializer-overrides -Wno-missing-include-dirs -Wno-shift-
>>>> negative-value -Wno-string-plus-int -Wno-typedef-redefinition -Wno-
>>>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-
>> not-
>>>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>>>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>>>> /work/armbru/qemu/include -iquote
>> /work/armbru/qemu/host/include/x86_64 -
>>>> iquote /work/armbru/qemu/host/include/generic -iquote
>>>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>>>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>>>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>>>> fsanitize=fuzzer-no-link -fPIE -isystem../linux-headers -isystemlinux-headers -
>>>> DNEED_CPU_H '-DCONFIG_TARGET="x86_64-softmmu-config-target.h"' '-
>>>> DCONFIG_DEVICES="x86_64-softmmu-config-devices.h"' -MD -MQ libqemu-
>>>> x86_64-softmmu.fa.p/hw_vfio_common.c.o -MF libqemu-x86_64-
>>>> softmmu.fa.p/hw_vfio_common.c.o.d -o libqemu-x86_64-
>>>> softmmu.fa.p/hw_vfio_common.c.o -c ../hw/vfio/common.c
>>>> ../hw/vfio/common.c:682:40: error: variable 'hostwin' is uninitialized when
>>>> used here [-Werror,-Wuninitialized]
>>>> hwaddr pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>>>> ^~~~~~~
>>>> ../hw/vfio/common.c:578:31: note: initialize the variable 'hostwin' to
>> silence
>>>> this warning
>>>> VFIOHostDMAWindow *hostwin;
>>>> ^
>>>> = NULL
>>>> ../hw/vfio/common.c:785:33: error: variable 'hostwin' is uninitialized when
>>>> used here [-Werror,-Wuninitialized]
>>>> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
>>>> ^~~~~~~
>>>> ../hw/vfio/common.c:783:35: note: initialize the variable 'hostwin' to
>> silence
>>>> this warning
>>>> VFIOHostDMAWindow *hostwin;
>>>> ^
>>>> = NULL
>>>> 2 errors generated.
>>>> FAILED: tests/unit/test-resv-mem.p/test-resv-mem.c.o
>>>> clang -m64 -mcx16 -Itests/unit/test-resv-mem.p -Itests/unit -I../tests/unit -I.
>> -
>>>> Iqapi -Itrace -Iui -Iui/shader -I/usr/include/glib-2.0 -I/usr/lib64/glib-2.0/include
>> -
>>>> I/usr/include/sysprof-4 -fcolor-diagnostics -Wall -Winvalid-pch -Werror -
>>>> std=gnu11 -O2 -g -fstack-protector-strong -Wundef -Wwrite-strings -
>> Wmissing-
>>>> prototypes -Wstrict-prototypes -Wredundant-decls -Wold-style-definition -
>>>> Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-
>> qualifiers -
>>>> Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -
>>>> Wmissing-format-attribute -Wno-initializer-overrides -Wno-missing-include-
>> dirs -
>>>> Wno-shift-negative-value -Wno-string-plus-int -Wno-typedef-redefinition -
>> Wno-
>>>> tautological-type-limit-compare -Wno-psabi -Wno-gnu-variable-sized-type-
>> not-
>>>> at-end -Wthread-safety -isystem /work/armbru/qemu/linux-headers -isystem
>>>> linux-headers -iquote . -iquote /work/armbru/qemu -iquote
>>>> /work/armbru/qemu/include -iquote
>> /work/armbru/qemu/host/include/x86_64 -
>>>> iquote /work/armbru/qemu/host/include/generic -iquote
>>>> /work/armbru/qemu/tcg/i386 -pthread -D_GNU_SOURCE -
>>>> D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -fno-strict-aliasing -fno-
>>>> common -fwrapv -fsanitize-coverage-allowlist=instrumentation-filter -
>>>> fsanitize=fuzzer-no-link -fPIE -MD -MQ tests/unit/test-resv-mem.p/test-resv-
>>>> mem.c.o -MF tests/unit/test-resv-mem.p/test-resv-mem.c.o.d -o
>> tests/unit/test-
>>>> resv-mem.p/test-resv-mem.c.o -c ../tests/unit/test-resv-mem.c
>>>> ../tests/unit/test-resv-mem.c:42:9: error: variable 'i' set but not used [-
>>>> Werror,-Wunused-but-set-variable]
>>>> int i = 0;
>>>> ^
>>>> 1 error generated.
>>>>
>>>> Delete @i, please.
>>>
>>> Thanks for your report, I'll fix it and build with both compilers next time.
>>
>> a compile on a fedora rawhide also reports :
>>
>> ../hw/vfio/iommufd.c: In function ‘vfio_get_info_iova_range’:
>> ../hw/vfio/iommufd.c:370:27: error: cast from pointer to integer of different size
>> [-Werror=pointer-to-int-cast]
>> 370 | info->allowed_iovas = (uint64_t)(info + 1);
>> | ^
>> ../hw/vfio/iommufd.c:377:19: error: cast to pointer from integer of different size
>> [-Werror=int-to-pointer-cast]
>> 377 | iova_ranges = (struct iommu_iova_range *)info->allowed_iovas;
>> | ^
>> cc1: all warnings being treated as errors
>
> I failed to understand this warning, is your env a 32bit fedora?
it's not a rawhide but a 32bit indeed.
Thanks,
C.
>
>>
>>
>>> About second error, I'll send a separate patch as it's unrelated to this series.
>>
>>
>> Since this is a big series, I suggest that you take this patch in your build
>> environment,
>
> Got it, will do, thanks for your suggestions.
>
> BRs.
> Zhenzhong
>
>>
>>
>> diff --git a/meson.build b/meson.build
>> index
>> 72a57288a026325d5ff753131c037e99f6f35c1a..c946cbef5b29e23475dc4cf3456
>> 55e0466cbfade 100644
>> --- a/meson.build
>> +++ b/meson.build
>> @@ -462,6 +462,9 @@ warn_flags = [
>> '-Wno-tautological-type-limit-compare',
>> '-Wno-psabi',
>> '-Wno-gnu-variable-sized-type-not-at-end',
>> + '-Wshadow=local',
>> + '-Wno-error=shadow=local',
>> + '-Wno-error=shadow=compatible-local',
>> ]
>>
>> if targetos != 'darwin'
>>
>>
>> Thanks,
>>
>> C.
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-26 10:30 ` [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
@ 2023-10-27 14:02 ` Cédric Le Goater
2023-10-30 2:40 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:02 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> Introduce a dumb VFIOContainer base object and its targetted interface.
targeted
> This is willingly not a QOM object because we don't want it to be
> visible from the user interface. The VFIOContainer will be smoothly
> populated in subsequent patches as well as interfaces.
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 8 +---
> include/hw/vfio/vfio-container-base.h | 64 +++++++++++++++++++++++++++
> 2 files changed, 66 insertions(+), 6 deletions(-)
> create mode 100644 include/hw/vfio/vfio-container-base.h
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index b9c7a7e588..d8f293cb57 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -30,6 +30,7 @@
> #include <linux/vfio.h>
> #endif
> #include "sysemu/sysemu.h"
> +#include "hw/vfio/vfio-container-base.h"
>
> #define VFIO_MSG_PREFIX "vfio %s: "
>
> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
> struct VFIOGroup;
>
> typedef struct VFIOContainer {
> + VFIOContainerBase bcontainer;
> VFIOAddressSpace *space;
> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
> MemoryListener listener;
> @@ -201,12 +203,6 @@ typedef struct VFIODisplay {
> } dmabuf;
> } VFIODisplay;
>
> -typedef struct {
> - unsigned long *bitmap;
> - hwaddr size;
> - hwaddr pages;
> -} VFIOBitmap;
> -
> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> new file mode 100644
> index 0000000000..5becbd51a7
> --- /dev/null
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -0,0 +1,64 @@
> +/*
> + * VFIO BASE CONTAINER
> + *
> + * Copyright (C) 2023 Intel Corporation.
> + * Copyright Red Hat, Inc. 2023
> + *
> + * Authors: Yi Liu <yi.l.liu@intel.com>
> + * Eric Auger <eric.auger@redhat.com>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> +
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> +
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, see <http://www.gnu.org/licenses/>.
This should be enough :
SPDX-License-Identifier: GPL-2.0-or-later
> + */
> +
> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
HW_VFIO_VFIO_CONTAINER_BASE_H
> +
> +#include "exec/memory.h"
> +#ifndef CONFIG_USER_ONLY
> +#include "exec/hwaddr.h"
> +#endif
including "exec/memory.h" should be enough.
> +
> +typedef struct VFIODevice VFIODevice;
> +typedef struct VFIOIOMMUOps VFIOIOMMUOps;
> +
> +typedef struct {
> + unsigned long *bitmap;
> + hwaddr size;
> + hwaddr pages;
> +} VFIOBitmap;
> +
> +/*
> + * This is the base object for vfio container backends
> + */
> +typedef struct VFIOContainerBase {
> + const VFIOIOMMUOps *ops;
> +} VFIOContainerBase;
> +
> +struct VFIOIOMMUOps {
> + /* basic feature */
> + int (*dma_map)(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + void *vaddr, bool readonly);
> + int (*dma_unmap)(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + IOMMUTLBEntry *iotlb);
Could the VFIOContainerBase *parameter be const ?
> + int (*attach_device)(char *name, VFIODevice *vbasedev,
cont char *name ?
The rest looks good to me.
Thanks
C.
> + AddressSpace *as, Error **errp);
> + void (*detach_device)(VFIODevice *vbasedev);
> + /* migration feature */
> + int (*set_dirty_page_tracking)(VFIOContainerBase *bcontainer, bool start);
> + int (*query_dirty_bitmap)(VFIOContainerBase *bcontainer, VFIOBitmap *vbmap,
> + hwaddr iova, hwaddr size);
> +};
> +#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps
2023-10-26 10:30 ` [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps Zhenzhong Duan
@ 2023-10-27 14:20 ` Cédric Le Goater
2023-10-30 2:43 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:20 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> This empty VFIOIOMMUOps named vfio_legacy_ops will hold all general
> IOMMU ops of legacy container.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 2 +-
> hw/vfio/container.c | 5 +++++
> 2 files changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index d8f293cb57..8ded5cd8e4 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -255,7 +255,7 @@ typedef QLIST_HEAD(VFIOGroupList, VFIOGroup) VFIOGroupList;
> typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
> extern VFIOGroupList vfio_group_list;
> extern VFIODeviceList vfio_device_list;
> -
> +extern const VFIOIOMMUOps vfio_legacy_ops;
why does it need to be external ?
Thanks,
C.
> extern const MemoryListener vfio_memory_listener;
> extern int vfio_kvm_device_fd;
>
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 242010036a..4bc43ddfa4 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -472,6 +472,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> Error **errp)
> {
> VFIOContainer *container;
> + VFIOContainerBase *bcontainer;
> int ret, fd;
> VFIOAddressSpace *space;
>
> @@ -552,6 +553,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container->iova_ranges = NULL;
> QLIST_INIT(&container->giommu_list);
> QLIST_INIT(&container->vrdl_list);
> + bcontainer = &container->bcontainer;
> + bcontainer->ops = &vfio_legacy_ops;
>
> ret = vfio_init_container(container, group->fd, errp);
> if (ret) {
> @@ -933,3 +936,5 @@ void vfio_detach_device(VFIODevice *vbasedev)
> vfio_put_base_device(vbasedev);
> vfio_put_group(group);
> }
> +
> +const VFIOIOMMUOps vfio_legacy_ops;
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API
2023-10-26 10:30 ` [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
@ 2023-10-27 14:26 ` Cédric Le Goater
2023-10-30 2:48 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:26 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 4 ---
> include/hw/vfio/vfio-container-base.h | 7 ++++
> hw/vfio/common.c | 45 +++++++++++++-----------
> hw/vfio/container-base.c | 49 +++++++++++++++++++++++++++
> hw/vfio/container.c | 22 ++++++++----
> hw/vfio/meson.build | 1 +
> hw/vfio/trace-events | 2 +-
> 7 files changed, 98 insertions(+), 32 deletions(-)
> create mode 100644 hw/vfio/container-base.c
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 8ded5cd8e4..97056224f4 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -208,10 +208,6 @@ void vfio_put_address_space(VFIOAddressSpace *space);
> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>
> /* container->fd */
> -int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> - ram_addr_t size, IOMMUTLBEntry *iotlb);
> -int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> - ram_addr_t size, void *vaddr, bool readonly);
> int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
> int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> hwaddr iova, hwaddr size);
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 5becbd51a7..077e638ee8 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -45,6 +45,13 @@ typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> } VFIOContainerBase;
>
> +int vfio_container_dma_map(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + void *vaddr, bool readonly);
> +int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + IOMMUTLBEntry *iotlb);
> +
> struct VFIOIOMMUOps {
> /* basic feature */
> int (*dma_map)(VFIOContainerBase *bcontainer,
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 0ebf4d9256..141f2b54a4 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -292,7 +292,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
> static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> {
> VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
> - VFIOContainer *container = giommu->container;
> + VFIOContainerBase *bcontainer = &giommu->container->bcontainer;
> hwaddr iova = iotlb->iova + giommu->iommu_offset;
> void *vaddr;
> int ret;
> @@ -322,21 +322,22 @@ static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> * of vaddr will always be there, even if the memory object is
> * destroyed and its backing memory munmap-ed.
> */
> - ret = vfio_dma_map(container, iova,
> - iotlb->addr_mask + 1, vaddr,
> - read_only);
> + ret = vfio_container_dma_map(bcontainer, iova,
> + iotlb->addr_mask + 1, vaddr,
> + read_only);
> if (ret) {
> - error_report("vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
> + error_report("vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx", %p) = %d (%s)",
> - container, iova,
> + bcontainer, iova,
> iotlb->addr_mask + 1, vaddr, ret, strerror(-ret));
> }
> } else {
> - ret = vfio_dma_unmap(container, iova, iotlb->addr_mask + 1, iotlb);
> + ret = vfio_container_dma_unmap(bcontainer, iova,
> + iotlb->addr_mask + 1, iotlb);
> if (ret) {
> - error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> + error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx") = %d (%s)",
> - container, iova,
> + bcontainer, iova,
> iotlb->addr_mask + 1, ret, strerror(-ret));
> vfio_set_migration_error(ret);
> }
> @@ -355,9 +356,10 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
> int ret;
>
> /* Unmap with a single call. */
> - ret = vfio_dma_unmap(vrdl->container, iova, size , NULL);
> + ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
> + iova, size , NULL);
> if (ret) {
> - error_report("%s: vfio_dma_unmap() failed: %s", __func__,
> + error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
> strerror(-ret));
> }
> }
> @@ -385,8 +387,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
> section->offset_within_address_space;
> vaddr = memory_region_get_ram_ptr(section->mr) + start;
>
> - ret = vfio_dma_map(vrdl->container, iova, next - start,
> - vaddr, section->readonly);
> + ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
> + next - start, vaddr, section->readonly);
> if (ret) {
> /* Rollback */
> vfio_ram_discard_notify_discard(rdl, section);
> @@ -685,10 +687,11 @@ static void vfio_listener_region_add(MemoryListener *listener,
> }
> }
>
> - ret = vfio_dma_map(container, iova, int128_get64(llsize),
> - vaddr, section->readonly);
> + ret = vfio_container_dma_map(&container->bcontainer,
> + iova, int128_get64(llsize), vaddr,
> + section->readonly);
> if (ret) {
> - error_setg(&err, "vfio_dma_map(%p, 0x%"HWADDR_PRIx", "
> + error_setg(&err, "vfio_container_dma_map(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx", %p) = %d (%s)",
> container, iova, int128_get64(llsize), vaddr, ret,
> strerror(-ret));
> @@ -786,18 +789,20 @@ static void vfio_listener_region_del(MemoryListener *listener,
> if (int128_eq(llsize, int128_2_64())) {
> /* The unmap ioctl doesn't accept a full 64-bit span. */
> llsize = int128_rshift(llsize, 1);
> - ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
> + ret = vfio_container_dma_unmap(&container->bcontainer, iova,
> + int128_get64(llsize), NULL);
> if (ret) {
> - error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> + error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx") = %d (%s)",
> container, iova, int128_get64(llsize), ret,
> strerror(-ret));
> }
> iova += int128_get64(llsize);
> }
> - ret = vfio_dma_unmap(container, iova, int128_get64(llsize), NULL);
> + ret = vfio_container_dma_unmap(&container->bcontainer, iova,
> + int128_get64(llsize), NULL);
> if (ret) {
> - error_report("vfio_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> + error_report("vfio_container_dma_unmap(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx") = %d (%s)",
> container, iova, int128_get64(llsize), ret,
> strerror(-ret));
> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
> new file mode 100644
> index 0000000000..9db8b89b2f
> --- /dev/null
> +++ b/hw/vfio/container-base.c
> @@ -0,0 +1,49 @@
> +/*
> + * VFIO BASE CONTAINER
> + *
> + * Copyright (C) 2023 Intel Corporation.
> + * Copyright Red Hat, Inc. 2023
> + *
> + * Authors: Yi Liu <yi.l.liu@intel.com>
> + * Eric Auger <eric.auger@redhat.com>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> +
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> + * GNU General Public License for more details.
> +
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, see <http://www.gnu.org/licenses/>.
> + */
Please reduce the header with an SPDX identifier.
> +
> +#include "qemu/osdep.h"
> +#include "qapi/error.h"
> +#include "qemu/error-report.h"
> +#include "hw/vfio/vfio-container-base.h"
> +
> +int vfio_container_dma_map(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + void *vaddr, bool readonly)
> +{
> + if (!bcontainer->ops->dma_map) {
I think this could be an assert instead ?
> + return -EINVAL;
> + }
> +
> + return bcontainer->ops->dma_map(bcontainer, iova, size, vaddr, readonly);
> +}
> +
> +int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> + hwaddr iova, ram_addr_t size,
> + IOMMUTLBEntry *iotlb)
> +{
> + if (!bcontainer->ops->dma_unmap) {
> + return -EINVAL;
> + }
> +
> + return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
> +}
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 4bc43ddfa4..c04df26323 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -115,9 +115,11 @@ unmap_exit:
> /*
> * DMA - Mapping and unmapping for the "type1" IOMMU interface used on x86
> */
> -int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> - ram_addr_t size, IOMMUTLBEntry *iotlb)
> +static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
> + ram_addr_t size, IOMMUTLBEntry *iotlb)
> {
> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
> + bcontainer);
> struct vfio_iommu_type1_dma_unmap unmap = {
> .argsz = sizeof(unmap),
> .flags = 0,
> @@ -151,7 +153,7 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> */
> if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
> container->iommu_type == VFIO_TYPE1v2_IOMMU) {
> - trace_vfio_dma_unmap_overflow_workaround();
> + trace_vfio_legacy_dma_unmap_overflow_workaround();
> unmap.size -= 1ULL << ctz64(container->pgsizes);
> continue;
> }
> @@ -170,9 +172,11 @@ int vfio_dma_unmap(VFIOContainer *container, hwaddr iova,
> return 0;
> }
>
> -int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> - ram_addr_t size, void *vaddr, bool readonly)
> +static int vfio_legacy_dma_map(VFIOContainerBase *bcontainer, hwaddr iova,
> + ram_addr_t size, void *vaddr, bool readonly)
> {
> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
> + bcontainer);
> struct vfio_iommu_type1_dma_map map = {
> .argsz = sizeof(map),
> .flags = VFIO_DMA_MAP_FLAG_READ,
> @@ -191,7 +195,8 @@ int vfio_dma_map(VFIOContainer *container, hwaddr iova,
> * the VGA ROM space.
> */
> if (ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0 ||
> - (errno == EBUSY && vfio_dma_unmap(container, iova, size, NULL) == 0 &&
> + (errno == EBUSY &&
> + vfio_legacy_dma_unmap(bcontainer, iova, size, NULL) == 0 &&
> ioctl(container->fd, VFIO_IOMMU_MAP_DMA, &map) == 0)) {
> return 0;
> }
> @@ -937,4 +942,7 @@ void vfio_detach_device(VFIODevice *vbasedev)
> vfio_put_group(group);
> }
>
> -const VFIOIOMMUOps vfio_legacy_ops;
> +const VFIOIOMMUOps vfio_legacy_ops = {
> + .dma_map = vfio_legacy_dma_map,
> + .dma_unmap = vfio_legacy_dma_unmap,
> +};
> diff --git a/hw/vfio/meson.build b/hw/vfio/meson.build
> index 2a6912c940..eb6ce6229d 100644
> --- a/hw/vfio/meson.build
> +++ b/hw/vfio/meson.build
> @@ -2,6 +2,7 @@ vfio_ss = ss.source_set()
> vfio_ss.add(files(
> 'helpers.c',
> 'common.c',
> + 'container-base.c',
> 'container.c',
> 'spapr.c',
> 'migration.c',
> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
> index 0eb2387cf2..9f7fedee98 100644
> --- a/hw/vfio/trace-events
> +++ b/hw/vfio/trace-events
> @@ -116,7 +116,7 @@ vfio_region_unmap(const char *name, unsigned long offset, unsigned long end) "Re
> vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Device %s region %d: %d sparse mmap entries"
> vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
> vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
> -vfio_dma_unmap_overflow_workaround(void) ""
> +vfio_legacy_dma_unmap_overflow_workaround(void) ""
> vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
> vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 09/37] vfio/common: Move giommu_list in base container
2023-10-26 10:30 ` [PATCH v3 09/37] vfio/common: Move giommu_list in base container Zhenzhong Duan
@ 2023-10-27 14:38 ` Cédric Le Goater
2023-10-30 2:48 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:38 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> Move the giommu_list field in the base object and store the
> base container in the VFIOGuestIOMMU.
>
> We introduce vfio_container_init/destroy helper on the base
> container.
This could have been a prereq patch.
>
> No fucntional change intended.
functional
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> [ clg: context changes ]
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
You can drop my Sob here and in all patches. The changes I did were
due to the recent 'iova_ranges' addition and the context was fuzzy.
Nothing major.
The rest looks good.
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 9 ---------
> include/hw/vfio/vfio-container-base.h | 13 +++++++++++++
> hw/vfio/common.c | 17 +++++++++++------
> hw/vfio/container-base.c | 18 ++++++++++++++++++
> hw/vfio/container.c | 13 +++----------
> 5 files changed, 45 insertions(+), 25 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 97056224f4..fcb4003a21 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -95,7 +95,6 @@ typedef struct VFIOContainer {
> uint64_t max_dirty_bitmap_size;
> unsigned long pgsizes;
> unsigned int dma_max_mappings;
> - QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> @@ -104,14 +103,6 @@ typedef struct VFIOContainer {
> GList *iova_ranges;
> } VFIOContainer;
>
> -typedef struct VFIOGuestIOMMU {
> - VFIOContainer *container;
> - IOMMUMemoryRegion *iommu_mr;
> - hwaddr iommu_offset;
> - IOMMUNotifier n;
> - QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
> -} VFIOGuestIOMMU;
> -
> typedef struct VFIORamDiscardListener {
> VFIOContainer *container;
> MemoryRegion *mr;
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 077e638ee8..71e1e4324e 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -43,8 +43,17 @@ typedef struct {
> */
> typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> + QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> } VFIOContainerBase;
>
> +typedef struct VFIOGuestIOMMU {
> + VFIOContainerBase *bcontainer;
> + IOMMUMemoryRegion *iommu_mr;
> + hwaddr iommu_offset;
> + IOMMUNotifier n;
> + QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
> +} VFIOGuestIOMMU;
> +
> int vfio_container_dma_map(VFIOContainerBase *bcontainer,
> hwaddr iova, ram_addr_t size,
> void *vaddr, bool readonly);
> @@ -52,6 +61,10 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> hwaddr iova, ram_addr_t size,
> IOMMUTLBEntry *iotlb);
>
> +void vfio_container_init(VFIOContainerBase *bcontainer,
> + const VFIOIOMMUOps *ops);
> +void vfio_container_destroy(VFIOContainerBase *bcontainer);
> +
> struct VFIOIOMMUOps {
> /* basic feature */
> int (*dma_map)(VFIOContainerBase *bcontainer,
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 141f2b54a4..4f130ad87c 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -292,7 +292,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb, void **vaddr,
> static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> {
> VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
> - VFIOContainerBase *bcontainer = &giommu->container->bcontainer;
> + VFIOContainerBase *bcontainer = giommu->bcontainer;
> hwaddr iova = iotlb->iova + giommu->iommu_offset;
> void *vaddr;
> int ret;
> @@ -569,6 +569,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> hwaddr iova, end;
> Int128 llend, llsize;
> void *vaddr;
> @@ -613,7 +614,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
> giommu->iommu_mr = iommu_mr;
> giommu->iommu_offset = section->offset_within_address_space -
> section->offset_within_region;
> - giommu->container = container;
> + giommu->bcontainer = bcontainer;
> llend = int128_add(int128_make64(section->offset_within_region),
> section->size);
> llend = int128_sub(llend, int128_one());
> @@ -648,7 +649,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
> g_free(giommu);
> goto fail;
> }
> - QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
> + QLIST_INSERT_HEAD(&bcontainer->giommu_list, giommu, giommu_next);
> memory_region_iommu_replay(giommu->iommu_mr, &giommu->n);
>
> return;
> @@ -733,6 +734,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
> MemoryRegionSection *section)
> {
> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> hwaddr iova, end;
> Int128 llend, llsize;
> int ret;
> @@ -745,7 +747,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
> if (memory_region_is_iommu(section->mr)) {
> VFIOGuestIOMMU *giommu;
>
> - QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
> + QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
> if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
> giommu->n.start == section->offset_within_region) {
> memory_region_unregister_iommu_notifier(section->mr,
> @@ -1208,7 +1210,9 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
> vfio_giommu_dirty_notifier *gdn = container_of(n,
> vfio_giommu_dirty_notifier, n);
> VFIOGuestIOMMU *giommu = gdn->giommu;
> - VFIOContainer *container = giommu->container;
> + VFIOContainerBase *bcontainer = giommu->bcontainer;
> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
> + bcontainer);
> hwaddr iova = iotlb->iova + giommu->iommu_offset;
> ram_addr_t translated_addr;
> int ret = -EINVAL;
> @@ -1286,12 +1290,13 @@ static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
> static int vfio_sync_dirty_bitmap(VFIOContainer *container,
> MemoryRegionSection *section)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> ram_addr_t ram_addr;
>
> if (memory_region_is_iommu(section->mr)) {
> VFIOGuestIOMMU *giommu;
>
> - QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
> + QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
> if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
> giommu->n.start == section->offset_within_region) {
> Int128 llend;
> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
> index 9db8b89b2f..6be7ce65ad 100644
> --- a/hw/vfio/container-base.c
> +++ b/hw/vfio/container-base.c
> @@ -47,3 +47,21 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
>
> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
> }
> +
> +void vfio_container_init(VFIOContainerBase *bcontainer, const VFIOIOMMUOps *ops)
> +{
> + bcontainer->ops = ops;
> + QLIST_INIT(&bcontainer->giommu_list);
> +}
> +
> +void vfio_container_destroy(VFIOContainerBase *bcontainer)
> +{
> + VFIOGuestIOMMU *giommu, *tmp;
> +
> + QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
> + memory_region_unregister_iommu_notifier(
> + MEMORY_REGION(giommu->iommu_mr), &giommu->n);
> + QLIST_REMOVE(giommu, giommu_next);
> + g_free(giommu);
> + }
> +}
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index c04df26323..b2b6e05287 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -556,10 +556,9 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container->dirty_pages_supported = false;
> container->dma_max_mappings = 0;
> container->iova_ranges = NULL;
> - QLIST_INIT(&container->giommu_list);
> QLIST_INIT(&container->vrdl_list);
> bcontainer = &container->bcontainer;
> - bcontainer->ops = &vfio_legacy_ops;
> + vfio_container_init(bcontainer, &vfio_legacy_ops);
>
> ret = vfio_init_container(container, group->fd, errp);
> if (ret) {
> @@ -661,6 +660,7 @@ put_space_exit:
> static void vfio_disconnect_container(VFIOGroup *group)
> {
> VFIOContainer *container = group->container;
> + VFIOContainerBase *bcontainer = &container->bcontainer;
>
> QLIST_REMOVE(group, container_next);
> group->container = NULL;
> @@ -685,16 +685,9 @@ static void vfio_disconnect_container(VFIOGroup *group)
>
> if (QLIST_EMPTY(&container->group_list)) {
> VFIOAddressSpace *space = container->space;
> - VFIOGuestIOMMU *giommu, *tmp;
>
> QLIST_REMOVE(container, next);
> -
> - QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next, tmp) {
> - memory_region_unregister_iommu_notifier(
> - MEMORY_REGION(giommu->iommu_mr), &giommu->n);
> - QLIST_REMOVE(giommu, giommu_next);
> - g_free(giommu);
> - }
> + vfio_container_destroy(bcontainer);
>
> trace_vfio_disconnect_container(container->fd);
> close(container->fd);
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 10/37] vfio/container: Move space field to base container
2023-10-26 10:30 ` [PATCH v3 10/37] vfio/container: Move space field to " Zhenzhong Duan
@ 2023-10-27 14:46 ` Cédric Le Goater
2023-10-30 2:51 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:46 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> Move the space field to the base object. Also the VFIOAddressSpace
> now contains a list of base containers.
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> [ clg: context changes ]
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> include/hw/vfio/vfio-common.h | 8 --------
> include/hw/vfio/vfio-container-base.h | 9 +++++++++
> hw/ppc/spapr_pci_vfio.c | 10 +++++-----
> hw/vfio/common.c | 4 ++--
> hw/vfio/container-base.c | 6 +++++-
> hw/vfio/container.c | 18 +++++++++---------
> 6 files changed, 30 insertions(+), 25 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index fcb4003a21..857d2b8076 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -73,17 +73,10 @@ typedef struct VFIOMigration {
> bool initial_data_sent;
> } VFIOMigration;
>
> -typedef struct VFIOAddressSpace {
> - AddressSpace *as;
> - QLIST_HEAD(, VFIOContainer) containers;
> - QLIST_ENTRY(VFIOAddressSpace) list;
> -} VFIOAddressSpace;
> -
> struct VFIOGroup;
>
> typedef struct VFIOContainer {
> VFIOContainerBase bcontainer;
> - VFIOAddressSpace *space;
> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
> MemoryListener listener;
> MemoryListener prereg_listener;
> @@ -98,7 +91,6 @@ typedef struct VFIOContainer {
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> - QLIST_ENTRY(VFIOContainer) next;
> QLIST_HEAD(, VFIODevice) device_list;
> GList *iova_ranges;
> } VFIOContainer;
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 71e1e4324e..a5fef3e6a8 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -38,12 +38,20 @@ typedef struct {
> hwaddr pages;
> } VFIOBitmap;
>
> +typedef struct VFIOAddressSpace {
> + AddressSpace *as;
> + QLIST_HEAD(, VFIOContainerBase) containers;
> + QLIST_ENTRY(VFIOAddressSpace) list;
> +} VFIOAddressSpace;
> +
> /*
> * This is the base object for vfio container backends
> */
> typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> + VFIOAddressSpace *space;
> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> + QLIST_ENTRY(VFIOContainerBase) next;
> } VFIOContainerBase;
>
> typedef struct VFIOGuestIOMMU {
> @@ -62,6 +70,7 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> IOMMUTLBEntry *iotlb);
>
> void vfio_container_init(VFIOContainerBase *bcontainer,
> + VFIOAddressSpace *space,
> const VFIOIOMMUOps *ops);
> void vfio_container_destroy(VFIOContainerBase *bcontainer);
>
> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
> index f283f7e38d..d1d07bec46 100644
> --- a/hw/ppc/spapr_pci_vfio.c
> +++ b/hw/ppc/spapr_pci_vfio.c
> @@ -84,27 +84,27 @@ static int vfio_eeh_container_op(VFIOContainer *container, uint32_t op)
> static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
> {
> VFIOAddressSpace *space = vfio_get_address_space(as);
> - VFIOContainer *container = NULL;
> + VFIOContainerBase *bcontainer = NULL;
>
> if (QLIST_EMPTY(&space->containers)) {
> /* No containers to act on */
> goto out;
> }
>
> - container = QLIST_FIRST(&space->containers);
> + bcontainer = QLIST_FIRST(&space->containers);
>
> - if (QLIST_NEXT(container, next)) {
> + if (QLIST_NEXT(bcontainer, next)) {
> /*
> * We don't yet have logic to synchronize EEH state across
> * multiple containers
> */
> - container = NULL;
> + bcontainer = NULL;
> goto out;
> }
>
> out:
> vfio_put_address_space(space);
> - return container;
> + return container_of(bcontainer, VFIOContainer, bcontainer);
> }
>
> static bool vfio_eeh_as_ok(AddressSpace *as)
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 4f130ad87c..f87a0dcec3 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
>
> bool vfio_viommu_preset(VFIODevice *vbasedev)
> {
> - return vbasedev->container->space->as != &address_space_memory;
> + return vbasedev->container->bcontainer.space->as != &address_space_memory;
> }
>
> static void vfio_set_migration_error(int err)
> @@ -924,7 +924,7 @@ static void vfio_dirty_tracking_init(VFIOContainer *container,
> dirty.container = container;
>
> memory_listener_register(&dirty.listener,
> - container->space->as);
> + container->bcontainer.space->as);
Could we introduce an helper returning :
container->bcontainer.space->as
The rest looks OK.
Thanks,
C.
>
> *ranges = dirty.ranges;
>
> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
> index 6be7ce65ad..99b2957d7b 100644
> --- a/hw/vfio/container-base.c
> +++ b/hw/vfio/container-base.c
> @@ -48,9 +48,11 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
> }
>
> -void vfio_container_init(VFIOContainerBase *bcontainer, const VFIOIOMMUOps *ops)
> +void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
> + const VFIOIOMMUOps *ops)
> {
> bcontainer->ops = ops;
> + bcontainer->space = space;
> QLIST_INIT(&bcontainer->giommu_list);
> }
>
> @@ -58,6 +60,8 @@ void vfio_container_destroy(VFIOContainerBase *bcontainer)
> {
> VFIOGuestIOMMU *giommu, *tmp;
>
> + QLIST_REMOVE(bcontainer, next);
> +
> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
> memory_region_unregister_iommu_notifier(
> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index b2b6e05287..761310fa51 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -514,7 +514,8 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> * details once we know which type of IOMMU we are using.
> */
>
> - QLIST_FOREACH(container, &space->containers, next) {
> + QLIST_FOREACH(bcontainer, &space->containers, next) {
> + container = container_of(bcontainer, VFIOContainer, bcontainer);
> if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
> ret = vfio_ram_block_discard_disable(container, true);
> if (ret) {
> @@ -550,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> }
>
> container = g_malloc0(sizeof(*container));
> - container->space = space;
> container->fd = fd;
> container->error = NULL;
> container->dirty_pages_supported = false;
> @@ -558,7 +558,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container->iova_ranges = NULL;
> QLIST_INIT(&container->vrdl_list);
> bcontainer = &container->bcontainer;
> - vfio_container_init(bcontainer, &vfio_legacy_ops);
> + vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>
> ret = vfio_init_container(container, group->fd, errp);
> if (ret) {
> @@ -613,14 +613,15 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> vfio_kvm_device_add_group(group);
>
> QLIST_INIT(&container->group_list);
> - QLIST_INSERT_HEAD(&space->containers, container, next);
> + QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
>
> group->container = container;
> QLIST_INSERT_HEAD(&container->group_list, group, container_next);
>
> container->listener = vfio_memory_listener;
>
> - memory_listener_register(&container->listener, container->space->as);
> + memory_listener_register(&container->listener,
> + container->bcontainer.space->as);
>
> if (container->error) {
> ret = -1;
> @@ -634,7 +635,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> return 0;
> listener_release_exit:
> QLIST_REMOVE(group, container_next);
> - QLIST_REMOVE(container, next);
> + QLIST_REMOVE(bcontainer, next);
> vfio_kvm_device_del_group(group);
> memory_listener_unregister(&container->listener);
> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
> @@ -684,9 +685,8 @@ static void vfio_disconnect_container(VFIOGroup *group)
> }
>
> if (QLIST_EMPTY(&container->group_list)) {
> - VFIOAddressSpace *space = container->space;
> + VFIOAddressSpace *space = container->bcontainer.space;
>
> - QLIST_REMOVE(container, next);
> vfio_container_destroy(bcontainer);
>
> trace_vfio_disconnect_container(container->fd);
> @@ -706,7 +706,7 @@ static VFIOGroup *vfio_get_group(int groupid, AddressSpace *as, Error **errp)
> QLIST_FOREACH(group, &vfio_group_list, next) {
> if (group->groupid == groupid) {
> /* Found it. Now is it already in the right context? */
> - if (group->container->space->as == as) {
> + if (group->container->bcontainer.space->as == as) {
> return group;
> } else {
> error_setg(errp, "group %d used in multiple address spaces",
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API
2023-10-26 10:30 ` [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
@ 2023-10-27 14:53 ` Cédric Le Goater
2023-10-30 2:53 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 14:53 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> dirty_pages_supported field is also moved to the base container
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> include/hw/vfio/vfio-common.h | 6 ------
> include/hw/vfio/vfio-container-base.h | 6 ++++++
> hw/vfio/common.c | 12 ++++++++----
> hw/vfio/container-base.c | 23 +++++++++++++++++++++++
> hw/vfio/container.c | 21 ++++++++++++++-------
> 5 files changed, 51 insertions(+), 17 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 857d2b8076..d053c61872 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -83,7 +83,6 @@ typedef struct VFIOContainer {
> unsigned iommu_type;
> Error *error;
> bool initialized;
> - bool dirty_pages_supported;
> uint64_t dirty_pgsizes;
> uint64_t max_dirty_bitmap_size;
> unsigned long pgsizes;
> @@ -190,11 +189,6 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>
> -/* container->fd */
> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> - hwaddr iova, hwaddr size);
> -
> /* SPAPR specific */
> int vfio_container_add_section_window(VFIOContainer *container,
> MemoryRegionSection *section,
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index a5fef3e6a8..ea8436a064 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -50,6 +50,7 @@ typedef struct VFIOAddressSpace {
> typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> VFIOAddressSpace *space;
> + bool dirty_pages_supported;
> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> QLIST_ENTRY(VFIOContainerBase) next;
> } VFIOContainerBase;
> @@ -68,6 +69,11 @@ int vfio_container_dma_map(VFIOContainerBase *bcontainer,
> int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> hwaddr iova, ram_addr_t size,
> IOMMUTLBEntry *iotlb);
> +int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
> + bool start);
> +int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
> + VFIOBitmap *vbmap,
> + hwaddr iova, hwaddr size);
>
> void vfio_container_init(VFIOContainerBase *bcontainer,
> VFIOAddressSpace *space,
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index f87a0dcec3..7d9b87fc67 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -1081,7 +1081,8 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
> if (vfio_devices_all_device_dirty_tracking(container)) {
> ret = vfio_devices_dma_logging_start(container);
> } else {
> - ret = vfio_set_dirty_page_tracking(container, true);
> + ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
> + true);
> }
>
> if (ret) {
> @@ -1099,7 +1100,8 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
> if (vfio_devices_all_device_dirty_tracking(container)) {
> vfio_devices_dma_logging_stop(container);
> } else {
> - ret = vfio_set_dirty_page_tracking(container, false);
> + ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
> + false);
> }
>
> if (ret) {
> @@ -1167,7 +1169,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> VFIOBitmap vbmap;
> int ret;
>
> - if (!container->dirty_pages_supported && !all_device_dirty_tracking) {
> + if (!container->bcontainer.dirty_pages_supported &&
> + !all_device_dirty_tracking) {
> cpu_physical_memory_set_dirty_range(ram_addr, size,
> tcg_enabled() ? DIRTY_CLIENTS_ALL :
> DIRTY_CLIENTS_NOCODE);
> @@ -1182,7 +1185,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> if (all_device_dirty_tracking) {
> ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
> } else {
> - ret = vfio_query_dirty_bitmap(container, &vbmap, iova, size);
> + ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
> + iova, size);
> }
>
> if (ret) {
> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
> index 99b2957d7b..a7cf517dd2 100644
> --- a/hw/vfio/container-base.c
> +++ b/hw/vfio/container-base.c
> @@ -48,11 +48,34 @@ int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
> }
>
> +int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
> + bool start)
> +{
> + /* Fallback to all pages dirty if dirty page sync isn't supported */
> + if (!bcontainer->ops->set_dirty_page_tracking) {
> + return 0;
> + }
I would start with an assert and relax the check later on, if needed and
in its own patch.
> + return bcontainer->ops->set_dirty_page_tracking(bcontainer, start);
> +}
> +
> +int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
> + VFIOBitmap *vbmap,
> + hwaddr iova, hwaddr size)
> +{
> + if (!bcontainer->ops->query_dirty_bitmap) {
> + return -EINVAL;
> + }
Same comment.
Thanks,
C.
> + return bcontainer->ops->query_dirty_bitmap(bcontainer, vbmap, iova, size);
> +}
> +
> void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
> const VFIOIOMMUOps *ops)
> {
> bcontainer->ops = ops;
> bcontainer->space = space;
> + bcontainer->dirty_pages_supported = false;
> QLIST_INIT(&bcontainer->giommu_list);
> }
>
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 761310fa51..6f02ca133e 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -131,7 +131,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
>
> if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
> if (!vfio_devices_all_device_dirty_tracking(container) &&
> - container->dirty_pages_supported) {
> + container->bcontainer.dirty_pages_supported) {
> return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
> }
>
> @@ -205,14 +205,17 @@ static int vfio_legacy_dma_map(VFIOContainerBase *bcontainer, hwaddr iova,
> return -errno;
> }
>
> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
> +static int vfio_legacy_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
> + bool start)
> {
> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
> + bcontainer);
> int ret;
> struct vfio_iommu_type1_dirty_bitmap dirty = {
> .argsz = sizeof(dirty),
> };
>
> - if (!container->dirty_pages_supported) {
> + if (!bcontainer->dirty_pages_supported) {
> return 0;
> }
>
> @@ -232,9 +235,12 @@ int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
> return ret;
> }
>
> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
> - hwaddr iova, hwaddr size)
> +static int vfio_legacy_query_dirty_bitmap(VFIOContainerBase *bcontainer,
> + VFIOBitmap *vbmap,
> + hwaddr iova, hwaddr size)
> {
> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
> + bcontainer);
> struct vfio_iommu_type1_dirty_bitmap *dbitmap;
> struct vfio_iommu_type1_dirty_bitmap_get *range;
> int ret;
> @@ -461,7 +467,7 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
> * qemu_real_host_page_size to mark those dirty.
> */
> if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
> - container->dirty_pages_supported = true;
> + container->bcontainer.dirty_pages_supported = true;
> container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
> container->dirty_pgsizes = cap_mig->pgsize_bitmap;
> }
> @@ -553,7 +559,6 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container = g_malloc0(sizeof(*container));
> container->fd = fd;
> container->error = NULL;
> - container->dirty_pages_supported = false;
> container->dma_max_mappings = 0;
> container->iova_ranges = NULL;
> QLIST_INIT(&container->vrdl_list);
> @@ -938,4 +943,6 @@ void vfio_detach_device(VFIODevice *vbasedev)
> const VFIOIOMMUOps vfio_legacy_ops = {
> .dma_map = vfio_legacy_dma_map,
> .dma_unmap = vfio_legacy_dma_unmap,
> + .set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
> + .query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
> };
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 12/37] vfio/container: Move per container device list in base container
2023-10-26 10:30 ` [PATCH v3 12/37] vfio/container: Move per container device list in base container Zhenzhong Duan
@ 2023-10-27 15:03 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 15:03 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> VFIO Device is also changed to point to base container instead of
> legacy container.
>
> No fucntional change intended.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> [ clg: context changes ]
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
Please drop my SoB.
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 3 +--
> include/hw/vfio/vfio-container-base.h | 1 +
> hw/vfio/common.c | 23 +++++++++++++++--------
> hw/vfio/container.c | 12 ++++++------
> 4 files changed, 23 insertions(+), 16 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index d053c61872..f7a84dc8d0 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -90,7 +90,6 @@ typedef struct VFIOContainer {
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> - QLIST_HEAD(, VFIODevice) device_list;
> GList *iova_ranges;
> } VFIOContainer;
>
> @@ -118,7 +117,7 @@ typedef struct VFIODevice {
> QLIST_ENTRY(VFIODevice) container_next;
> QLIST_ENTRY(VFIODevice) global_next;
> struct VFIOGroup *group;
> - VFIOContainer *container;
> + VFIOContainerBase *bcontainer;
> char *sysfsdev;
> char *name;
> DeviceState *dev;
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index ea8436a064..f1de1ef120 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -53,6 +53,7 @@ typedef struct VFIOContainerBase {
> bool dirty_pages_supported;
> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> QLIST_ENTRY(VFIOContainerBase) next;
> + QLIST_HEAD(, VFIODevice) device_list;
> } VFIOContainerBase;
>
> typedef struct VFIOGuestIOMMU {
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 7d9b87fc67..f8475348ad 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
>
> bool vfio_viommu_preset(VFIODevice *vbasedev)
> {
> - return vbasedev->container->bcontainer.space->as != &address_space_memory;
> + return vbasedev->bcontainer->space->as != &address_space_memory;
> }
>
> static void vfio_set_migration_error(int err)
> @@ -179,6 +179,7 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
>
> static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> MigrationState *ms = migrate_get_current();
>
> @@ -187,7 +188,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> return false;
> }
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> VFIOMigration *migration = vbasedev->migration;
>
> if (!migration) {
> @@ -205,9 +206,10 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
>
> bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> if (!vbasedev->dirty_pages_supported) {
> return false;
> }
> @@ -222,13 +224,14 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> */
> bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
>
> if (!migration_is_active(migrate_get_current())) {
> return false;
> }
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> VFIOMigration *migration = vbasedev->migration;
>
> if (!migration) {
> @@ -835,12 +838,13 @@ static bool vfio_section_is_vfio_pci(MemoryRegionSection *section,
> VFIOContainer *container)
> {
> VFIOPCIDevice *pcidev;
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> Object *owner;
>
> owner = memory_region_owner(section->mr);
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> if (vbasedev->type != VFIO_DEVICE_TYPE_PCI) {
> continue;
> }
> @@ -941,13 +945,14 @@ static void vfio_devices_dma_logging_stop(VFIOContainer *container)
> uint64_t buf[DIV_ROUND_UP(sizeof(struct vfio_device_feature),
> sizeof(uint64_t))] = {};
> struct vfio_device_feature *feature = (struct vfio_device_feature *)buf;
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
>
> feature->argsz = sizeof(buf);
> feature->flags = VFIO_DEVICE_FEATURE_SET |
> VFIO_DEVICE_FEATURE_DMA_LOGGING_STOP;
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> if (!vbasedev->dirty_tracking) {
> continue;
> }
> @@ -1038,6 +1043,7 @@ static int vfio_devices_dma_logging_start(VFIOContainer *container)
> {
> struct vfio_device_feature *feature;
> VFIODirtyRanges ranges;
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> int ret = 0;
>
> @@ -1048,7 +1054,7 @@ static int vfio_devices_dma_logging_start(VFIOContainer *container)
> return -errno;
> }
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> if (vbasedev->dirty_tracking) {
> continue;
> }
> @@ -1141,10 +1147,11 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> VFIOBitmap *vbmap, hwaddr iova,
> hwaddr size)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> int ret;
>
> - QLIST_FOREACH(vbasedev, &container->device_list, container_next) {
> + QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> ret = vfio_device_dma_logging_report(vbasedev, iova, size,
> vbmap->bitmap);
> if (ret) {
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 6f02ca133e..3228dd2109 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -889,7 +889,7 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
> int groupid = vfio_device_groupid(vbasedev, errp);
> VFIODevice *vbasedev_iter;
> VFIOGroup *group;
> - VFIOContainer *container;
> + VFIOContainerBase *bcontainer;
> int ret;
>
> if (groupid < 0) {
> @@ -916,9 +916,9 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
> return ret;
> }
>
> - container = group->container;
> - vbasedev->container = container;
> - QLIST_INSERT_HEAD(&container->device_list, vbasedev, container_next);
> + bcontainer = &group->container->bcontainer;
> + vbasedev->bcontainer = bcontainer;
> + QLIST_INSERT_HEAD(&bcontainer->device_list, vbasedev, container_next);
> QLIST_INSERT_HEAD(&vfio_device_list, vbasedev, global_next);
>
> return ret;
> @@ -928,13 +928,13 @@ void vfio_detach_device(VFIODevice *vbasedev)
> {
> VFIOGroup *group = vbasedev->group;
>
> - if (!vbasedev->container) {
> + if (!vbasedev->bcontainer) {
> return;
> }
>
> QLIST_REMOVE(vbasedev, global_next);
> QLIST_REMOVE(vbasedev, container_next);
> - vbasedev->container = NULL;
> + vbasedev->bcontainer = NULL;
> trace_vfio_detach_device(vbasedev->name, group->groupid);
> vfio_put_base_device(vbasedev);
> vfio_put_group(group);
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 13/37] vfio/container: Convert functions to base container
2023-10-26 10:30 ` [PATCH v3 13/37] vfio/container: Convert functions to " Zhenzhong Duan
@ 2023-10-27 15:03 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 15:03 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> In the prospect to get rid of VFIOContainer refs
> in common.c lets convert misc functions to use the base
> container object instead:
>
> vfio_devices_all_dirty_tracking
> vfio_devices_all_device_dirty_tracking
> vfio_devices_all_running_and_mig_active
> vfio_devices_query_dirty_bitmap
> vfio_get_dirty_bitmap
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
Please drop my Sob.
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 9 ++++----
> hw/vfio/common.c | 42 +++++++++++++++--------------------
> hw/vfio/container.c | 6 ++---
> hw/vfio/trace-events | 2 +-
> 4 files changed, 26 insertions(+), 33 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index f7a84dc8d0..fb3c7aea8f 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -186,7 +186,6 @@ typedef struct VFIODisplay {
>
> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
> -bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>
> /* SPAPR specific */
> int vfio_container_add_section_window(VFIOContainer *container,
> @@ -260,11 +259,11 @@ bool vfio_migration_realize(VFIODevice *vbasedev, Error **errp);
> void vfio_migration_exit(VFIODevice *vbasedev);
>
> int vfio_bitmap_alloc(VFIOBitmap *vbmap, hwaddr size);
> -bool vfio_devices_all_running_and_mig_active(VFIOContainer *container);
> -bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container);
> -int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> +bool vfio_devices_all_running_and_mig_active(VFIOContainerBase *bcontainer);
> +bool vfio_devices_all_device_dirty_tracking(VFIOContainerBase *bcontainer);
> +int vfio_devices_query_dirty_bitmap(VFIOContainerBase *bcontainer,
> VFIOBitmap *vbmap, hwaddr iova,
> hwaddr size);
> -int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> +int vfio_get_dirty_bitmap(VFIOContainerBase *bcontainer, uint64_t iova,
> uint64_t size, ram_addr_t ram_addr);
> #endif /* HW_VFIO_VFIO_COMMON_H */
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index f8475348ad..91411d9844 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -177,9 +177,8 @@ bool vfio_device_state_is_precopy(VFIODevice *vbasedev)
> migration->device_state == VFIO_DEVICE_STATE_PRE_COPY_P2P;
> }
>
> -static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> +static bool vfio_devices_all_dirty_tracking(VFIOContainerBase *bcontainer)
> {
> - VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> MigrationState *ms = migrate_get_current();
>
> @@ -204,9 +203,8 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainer *container)
> return true;
> }
>
> -bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> +bool vfio_devices_all_device_dirty_tracking(VFIOContainerBase *bcontainer)
> {
> - VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
>
> QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> @@ -222,9 +220,8 @@ bool vfio_devices_all_device_dirty_tracking(VFIOContainer *container)
> * Check if all VFIO devices are running and migration is active, which is
> * essentially equivalent to the migration being in pre-copy phase.
> */
> -bool vfio_devices_all_running_and_mig_active(VFIOContainer *container)
> +bool vfio_devices_all_running_and_mig_active(VFIOContainerBase *bcontainer)
> {
> - VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
>
> if (!migration_is_active(migrate_get_current())) {
> @@ -1084,7 +1081,7 @@ static void vfio_listener_log_global_start(MemoryListener *listener)
> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> int ret;
>
> - if (vfio_devices_all_device_dirty_tracking(container)) {
> + if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
> ret = vfio_devices_dma_logging_start(container);
> } else {
> ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
> @@ -1103,7 +1100,7 @@ static void vfio_listener_log_global_stop(MemoryListener *listener)
> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
> int ret = 0;
>
> - if (vfio_devices_all_device_dirty_tracking(container)) {
> + if (vfio_devices_all_device_dirty_tracking(&container->bcontainer)) {
> vfio_devices_dma_logging_stop(container);
> } else {
> ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
> @@ -1143,11 +1140,10 @@ static int vfio_device_dma_logging_report(VFIODevice *vbasedev, hwaddr iova,
> return 0;
> }
>
> -int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> +int vfio_devices_query_dirty_bitmap(VFIOContainerBase *bcontainer,
> VFIOBitmap *vbmap, hwaddr iova,
> hwaddr size)
> {
> - VFIOContainerBase *bcontainer = &container->bcontainer;
> VFIODevice *vbasedev;
> int ret;
>
> @@ -1167,17 +1163,16 @@ int vfio_devices_query_dirty_bitmap(VFIOContainer *container,
> return 0;
> }
>
> -int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> +int vfio_get_dirty_bitmap(VFIOContainerBase *bcontainer, uint64_t iova,
> uint64_t size, ram_addr_t ram_addr)
> {
> bool all_device_dirty_tracking =
> - vfio_devices_all_device_dirty_tracking(container);
> + vfio_devices_all_device_dirty_tracking(bcontainer);
> uint64_t dirty_pages;
> VFIOBitmap vbmap;
> int ret;
>
> - if (!container->bcontainer.dirty_pages_supported &&
> - !all_device_dirty_tracking) {
> + if (!bcontainer->dirty_pages_supported && !all_device_dirty_tracking) {
> cpu_physical_memory_set_dirty_range(ram_addr, size,
> tcg_enabled() ? DIRTY_CLIENTS_ALL :
> DIRTY_CLIENTS_NOCODE);
> @@ -1190,10 +1185,9 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> }
>
> if (all_device_dirty_tracking) {
> - ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
> + ret = vfio_devices_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
> } else {
> - ret = vfio_container_query_dirty_bitmap(&container->bcontainer, &vbmap,
> - iova, size);
> + ret = vfio_container_query_dirty_bitmap(bcontainer, &vbmap, iova, size);
> }
>
> if (ret) {
> @@ -1203,8 +1197,7 @@ int vfio_get_dirty_bitmap(VFIOContainer *container, uint64_t iova,
> dirty_pages = cpu_physical_memory_set_dirty_lebitmap(vbmap.bitmap, ram_addr,
> vbmap.pages);
>
> - trace_vfio_get_dirty_bitmap(container->fd, iova, size, vbmap.size,
> - ram_addr, dirty_pages);
> + trace_vfio_get_dirty_bitmap(iova, size, vbmap.size, ram_addr, dirty_pages);
> out:
> g_free(vbmap.bitmap);
>
> @@ -1238,8 +1231,8 @@ static void vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
>
> rcu_read_lock();
> if (vfio_get_xlat_addr(iotlb, NULL, &translated_addr, NULL)) {
> - ret = vfio_get_dirty_bitmap(container, iova, iotlb->addr_mask + 1,
> - translated_addr);
> + ret = vfio_get_dirty_bitmap(&container->bcontainer, iova,
> + iotlb->addr_mask + 1, translated_addr);
> if (ret) {
> error_report("vfio_iommu_map_dirty_notify(%p, 0x%"HWADDR_PRIx", "
> "0x%"HWADDR_PRIx") = %d (%s)",
> @@ -1268,7 +1261,8 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
> * Sync the whole mapped region (spanning multiple individual mappings)
> * in one go.
> */
> - return vfio_get_dirty_bitmap(vrdl->container, iova, size, ram_addr);
> + return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
> + ram_addr);
> }
>
> static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
> @@ -1337,7 +1331,7 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
> ram_addr = memory_region_get_ram_addr(section->mr) +
> section->offset_within_region;
>
> - return vfio_get_dirty_bitmap(container,
> + return vfio_get_dirty_bitmap(&container->bcontainer,
> REAL_HOST_PAGE_ALIGN(section->offset_within_address_space),
> int128_get64(section->size), ram_addr);
> }
> @@ -1352,7 +1346,7 @@ static void vfio_listener_log_sync(MemoryListener *listener,
> return;
> }
>
> - if (vfio_devices_all_dirty_tracking(container)) {
> + if (vfio_devices_all_dirty_tracking(&container->bcontainer)) {
> ret = vfio_sync_dirty_bitmap(container, section);
> if (ret) {
> error_report("vfio: Failed to sync dirty bitmap, err: %d (%s)", ret,
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 3228dd2109..8d5b408e86 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -129,8 +129,8 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
> bool need_dirty_sync = false;
> int ret;
>
> - if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
> - if (!vfio_devices_all_device_dirty_tracking(container) &&
> + if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
> + if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
> container->bcontainer.dirty_pages_supported) {
> return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
> }
> @@ -162,7 +162,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
> }
>
> if (need_dirty_sync) {
> - ret = vfio_get_dirty_bitmap(container, iova, size,
> + ret = vfio_get_dirty_bitmap(bcontainer, iova, size,
> iotlb->translated_addr);
> if (ret) {
> return ret;
> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
> index 9f7fedee98..08a1f9dfa4 100644
> --- a/hw/vfio/trace-events
> +++ b/hw/vfio/trace-events
> @@ -117,7 +117,7 @@ vfio_region_sparse_mmap_header(const char *name, int index, int nr_areas) "Devic
> vfio_region_sparse_mmap_entry(int i, unsigned long start, unsigned long end) "sparse entry %d [0x%lx - 0x%lx]"
> vfio_get_dev_region(const char *name, int index, uint32_t type, uint32_t subtype) "%s index %d, %08x/%08x"
> vfio_legacy_dma_unmap_overflow_workaround(void) ""
> -vfio_get_dirty_bitmap(int fd, uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "container fd=%d, iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
> +vfio_get_dirty_bitmap(uint64_t iova, uint64_t size, uint64_t bitmap_size, uint64_t start, uint64_t dirty_pages) "iova=0x%"PRIx64" size= 0x%"PRIx64" bitmap_size=0x%"PRIx64" start=0x%"PRIx64" dirty_pages=%"PRIu64
> vfio_iommu_map_dirty_notify(uint64_t iova_start, uint64_t iova_end) "iommu dirty @ 0x%"PRIx64" - 0x%"PRIx64
>
> # platform.c
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base container
2023-10-26 10:30 ` [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
@ 2023-10-27 15:52 ` Cédric Le Goater
2023-10-30 3:14 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 15:52 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun, Nicholas Piggin, Daniel Henrique Barboza, David Gibson,
Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> Move vrdl_list, pgsizes and dma_max_mappings to the base
> container object
>
> No functional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> [ clg: context changes ]
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> include/hw/vfio/vfio-common.h | 13 -------
> include/hw/vfio/vfio-container-base.h | 13 +++++++
> hw/vfio/common.c | 49 ++++++++++++++-------------
> hw/vfio/container-base.c | 12 +++++++
> hw/vfio/container.c | 12 +++----
> hw/vfio/spapr.c | 18 +++++++---
> 6 files changed, 68 insertions(+), 49 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index fb3c7aea8f..65ae2d76cf 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -85,24 +85,11 @@ typedef struct VFIOContainer {
> bool initialized;
> uint64_t dirty_pgsizes;
> uint64_t max_dirty_bitmap_size;
> - unsigned long pgsizes;
> - unsigned int dma_max_mappings;
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> - QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> GList *iova_ranges;
> } VFIOContainer;
>
> -typedef struct VFIORamDiscardListener {
> - VFIOContainer *container;
> - MemoryRegion *mr;
> - hwaddr offset_within_address_space;
> - hwaddr size;
> - uint64_t granularity;
> - RamDiscardListener listener;
> - QLIST_ENTRY(VFIORamDiscardListener) next;
> -} VFIORamDiscardListener;
> -
> typedef struct VFIOHostDMAWindow {
> hwaddr min_iova;
> hwaddr max_iova;
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index f1de1ef120..849c8b34b2 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -50,8 +50,11 @@ typedef struct VFIOAddressSpace {
> typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> VFIOAddressSpace *space;
> + unsigned long pgsizes;
> + unsigned int dma_max_mappings;
> bool dirty_pages_supported;
> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
> + QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
> QLIST_ENTRY(VFIOContainerBase) next;
> QLIST_HEAD(, VFIODevice) device_list;
> } VFIOContainerBase;
> @@ -64,6 +67,16 @@ typedef struct VFIOGuestIOMMU {
> QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
> } VFIOGuestIOMMU;
>
> +typedef struct VFIORamDiscardListener {
> + VFIOContainerBase *bcontainer;
> + MemoryRegion *mr;
> + hwaddr offset_within_address_space;
> + hwaddr size;
> + uint64_t granularity;
> + RamDiscardListener listener;
> + QLIST_ENTRY(VFIORamDiscardListener) next;
> +} VFIORamDiscardListener;
> +
> int vfio_container_dma_map(VFIOContainerBase *bcontainer,
> hwaddr iova, ram_addr_t size,
> void *vaddr, bool readonly);
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index 91411d9844..9b34e7e0f8 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -351,13 +351,13 @@ static void vfio_ram_discard_notify_discard(RamDiscardListener *rdl,
> {
> VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
> listener);
> + VFIOContainerBase *bcontainer = vrdl->bcontainer;
> const hwaddr size = int128_get64(section->size);
> const hwaddr iova = section->offset_within_address_space;
> int ret;
>
> /* Unmap with a single call. */
> - ret = vfio_container_dma_unmap(&vrdl->container->bcontainer,
> - iova, size , NULL);
> + ret = vfio_container_dma_unmap(bcontainer, iova, size , NULL);
> if (ret) {
> error_report("%s: vfio_container_dma_unmap() failed: %s", __func__,
> strerror(-ret));
> @@ -369,6 +369,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
> {
> VFIORamDiscardListener *vrdl = container_of(rdl, VFIORamDiscardListener,
> listener);
> + VFIOContainerBase *bcontainer = vrdl->bcontainer;
> const hwaddr end = section->offset_within_region +
> int128_get64(section->size);
> hwaddr start, next, iova;
> @@ -387,8 +388,8 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
> section->offset_within_address_space;
> vaddr = memory_region_get_ram_ptr(section->mr) + start;
>
> - ret = vfio_container_dma_map(&vrdl->container->bcontainer, iova,
> - next - start, vaddr, section->readonly);
> + ret = vfio_container_dma_map(bcontainer, iova, next - start,
> + vaddr, section->readonly);
> if (ret) {
> /* Rollback */
> vfio_ram_discard_notify_discard(rdl, section);
> @@ -398,7 +399,7 @@ static int vfio_ram_discard_notify_populate(RamDiscardListener *rdl,
> return 0;
> }
>
> -static void vfio_register_ram_discard_listener(VFIOContainer *container,
> +static void vfio_register_ram_discard_listener(VFIOContainerBase *bcontainer,
> MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> @@ -411,7 +412,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> g_assert(QEMU_IS_ALIGNED(int128_get64(section->size), TARGET_PAGE_SIZE));
>
> vrdl = g_new0(VFIORamDiscardListener, 1);
> - vrdl->container = container;
> + vrdl->bcontainer = bcontainer;
> vrdl->mr = section->mr;
> vrdl->offset_within_address_space = section->offset_within_address_space;
> vrdl->size = int128_get64(section->size);
> @@ -419,14 +420,14 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> section->mr);
>
> g_assert(vrdl->granularity && is_power_of_2(vrdl->granularity));
> - g_assert(container->pgsizes &&
> - vrdl->granularity >= 1ULL << ctz64(container->pgsizes));
> + g_assert(bcontainer->pgsizes &&
> + vrdl->granularity >= 1ULL << ctz64(bcontainer->pgsizes));
>
> ram_discard_listener_init(&vrdl->listener,
> vfio_ram_discard_notify_populate,
> vfio_ram_discard_notify_discard, true);
> ram_discard_manager_register_listener(rdm, &vrdl->listener, section);
> - QLIST_INSERT_HEAD(&container->vrdl_list, vrdl, next);
> + QLIST_INSERT_HEAD(&bcontainer->vrdl_list, vrdl, next);
>
> /*
> * Sanity-check if we have a theoretically problematic setup where we could
> @@ -441,7 +442,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> * number of sections in the address space we could have over time,
> * also consuming DMA mappings.
> */
> - if (container->dma_max_mappings) {
> + if (bcontainer->dma_max_mappings) {
> unsigned int vrdl_count = 0, vrdl_mappings = 0, max_memslots = 512;
>
> #ifdef CONFIG_KVM
> @@ -450,7 +451,7 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> }
> #endif
>
> - QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
> + QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
> hwaddr start, end;
>
> start = QEMU_ALIGN_DOWN(vrdl->offset_within_address_space,
> @@ -462,23 +463,23 @@ static void vfio_register_ram_discard_listener(VFIOContainer *container,
> }
>
> if (vrdl_mappings + max_memslots - vrdl_count >
> - container->dma_max_mappings) {
> + bcontainer->dma_max_mappings) {
> warn_report("%s: possibly running out of DMA mappings. E.g., try"
> " increasing the 'block-size' of virtio-mem devies."
> " Maximum possible DMA mappings: %d, Maximum possible"
> - " memslots: %d", __func__, container->dma_max_mappings,
> + " memslots: %d", __func__, bcontainer->dma_max_mappings,
> max_memslots);
> }
> }
> }
>
> -static void vfio_unregister_ram_discard_listener(VFIOContainer *container,
> +static void vfio_unregister_ram_discard_listener(VFIOContainerBase *bcontainer,
> MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> VFIORamDiscardListener *vrdl = NULL;
>
> - QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
> + QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
> if (vrdl->mr == section->mr &&
> vrdl->offset_within_address_space ==
> section->offset_within_address_space) {
> @@ -627,7 +628,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
> iommu_idx);
>
> ret = memory_region_iommu_set_page_size_mask(giommu->iommu_mr,
> - container->pgsizes,
> + bcontainer->pgsizes,
> &err);
> if (ret) {
> g_free(giommu);
> @@ -663,7 +664,7 @@ static void vfio_listener_region_add(MemoryListener *listener,
> * about changes.
> */
> if (memory_region_has_ram_discard_manager(section->mr)) {
> - vfio_register_ram_discard_listener(container, section);
> + vfio_register_ram_discard_listener(bcontainer, section);
> return;
> }
>
> @@ -782,7 +783,7 @@ static void vfio_listener_region_del(MemoryListener *listener,
> pgmask = (1ULL << ctz64(hostwin->iova_pgsizes)) - 1;
> try_unmap = !((iova & pgmask) || (int128_get64(llsize) & pgmask));
> } else if (memory_region_has_ram_discard_manager(section->mr)) {
> - vfio_unregister_ram_discard_listener(container, section);
> + vfio_unregister_ram_discard_listener(bcontainer, section);
> /* Unregistering will trigger an unmap. */
> try_unmap = false;
> }
> @@ -1261,17 +1262,17 @@ static int vfio_ram_discard_get_dirty_bitmap(MemoryRegionSection *section,
> * Sync the whole mapped region (spanning multiple individual mappings)
> * in one go.
> */
> - return vfio_get_dirty_bitmap(&vrdl->container->bcontainer, iova, size,
> - ram_addr);
> + return vfio_get_dirty_bitmap(vrdl->bcontainer, iova, size, ram_addr);
> }
>
> -static int vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
> - MemoryRegionSection *section)
> +static int
> +vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainerBase *bcontainer,
> + MemoryRegionSection *section)
> {
> RamDiscardManager *rdm = memory_region_get_ram_discard_manager(section->mr);
> VFIORamDiscardListener *vrdl = NULL;
>
> - QLIST_FOREACH(vrdl, &container->vrdl_list, next) {
> + QLIST_FOREACH(vrdl, &bcontainer->vrdl_list, next) {
> if (vrdl->mr == section->mr &&
> vrdl->offset_within_address_space ==
> section->offset_within_address_space) {
> @@ -1325,7 +1326,7 @@ static int vfio_sync_dirty_bitmap(VFIOContainer *container,
> }
> return 0;
> } else if (memory_region_has_ram_discard_manager(section->mr)) {
> - return vfio_sync_ram_discard_listener_dirty_bitmap(container, section);
> + return vfio_sync_ram_discard_listener_dirty_bitmap(bcontainer, section);
> }
>
> ram_addr = memory_region_get_ram_addr(section->mr) +
> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
> index a7cf517dd2..568f891841 100644
> --- a/hw/vfio/container-base.c
> +++ b/hw/vfio/container-base.c
> @@ -76,15 +76,27 @@ void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace *space,
> bcontainer->ops = ops;
> bcontainer->space = space;
> bcontainer->dirty_pages_supported = false;
> + bcontainer->dma_max_mappings = 0;
> QLIST_INIT(&bcontainer->giommu_list);
> + QLIST_INIT(&bcontainer->vrdl_list);
> }
>
> void vfio_container_destroy(VFIOContainerBase *bcontainer)
> {
> + VFIORamDiscardListener *vrdl, *vrdl_tmp;
> VFIOGuestIOMMU *giommu, *tmp;
>
> QLIST_REMOVE(bcontainer, next);
>
> + QLIST_FOREACH_SAFE(vrdl, &bcontainer->vrdl_list, next, vrdl_tmp) {
> + RamDiscardManager *rdm;
> +
> + rdm = memory_region_get_ram_discard_manager(vrdl->mr);
> + ram_discard_manager_unregister_listener(rdm, &vrdl->listener);
> + QLIST_REMOVE(vrdl, next);
> + g_free(vrdl);
> + }
Where was this done previously ? May be the vrdl list should be handled
separatly from pgsizes and dma_max_mappings.
> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next, tmp) {
> memory_region_unregister_iommu_notifier(
> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 8d5b408e86..0e265ffa67 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -154,7 +154,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
> if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
> container->iommu_type == VFIO_TYPE1v2_IOMMU) {
> trace_vfio_legacy_dma_unmap_overflow_workaround();
> - unmap.size -= 1ULL << ctz64(container->pgsizes);
> + unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
> continue;
> }
> error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
> @@ -559,9 +559,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> container = g_malloc0(sizeof(*container));
> container->fd = fd;
> container->error = NULL;
> - container->dma_max_mappings = 0;
> container->iova_ranges = NULL;
> - QLIST_INIT(&container->vrdl_list);
> bcontainer = &container->bcontainer;
> vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>
> @@ -589,13 +587,13 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
> }
>
> if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
> - container->pgsizes = info->iova_pgsizes;
> + container->bcontainer.pgsizes = info->iova_pgsizes;
> } else {
> - container->pgsizes = qemu_real_host_page_size();
> + container->bcontainer.pgsizes = qemu_real_host_page_size();
> }
>
> - if (!vfio_get_info_dma_avail(info, &container->dma_max_mappings)) {
> - container->dma_max_mappings = 65535;
> + if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
> + container->bcontainer.dma_max_mappings = 65535;
> }
>
> vfio_get_info_iova_range(info, container);
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 3495737ab2..dbc4c24052 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -223,13 +223,13 @@ static int vfio_spapr_create_window(VFIOContainer *container,
> if (pagesize > rampagesize) {
> pagesize = rampagesize;
> }
> - pgmask = container->pgsizes & (pagesize | (pagesize - 1));
> + pgmask = container->bcontainer.pgsizes & (pagesize | (pagesize - 1));
> pagesize = pgmask ? (1ULL << (63 - clz64(pgmask))) : 0;
> if (!pagesize) {
> error_report("Host doesn't support page size 0x%"PRIx64
> ", the supported mask is 0x%lx",
> memory_region_iommu_get_min_page_size(iommu_mr),
> - container->pgsizes);
> + container->bcontainer.pgsizes);
> return -EINVAL;
> }
>
> @@ -385,7 +385,7 @@ void vfio_container_del_section_window(VFIOContainer *container,
>
> bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> {
> -
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> struct vfio_iommu_spapr_tce_info info;
> bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
> int ret, fd = container->fd;
> @@ -424,7 +424,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> }
>
> if (v2) {
> - container->pgsizes = info.ddw.pgsizes;
> + bcontainer->pgsizes = info.ddw.pgsizes;
> /*
> * There is a default window in just created container.
> * To make region_add/del simpler, we better remove this
> @@ -439,7 +439,7 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> }
> } else {
> /* The default table uses 4K pages */
> - container->pgsizes = 0x1000;
> + bcontainer->pgsizes = 0x1000;
> vfio_host_win_add(container, info.dma32_window_start,
> info.dma32_window_start +
> info.dma32_window_size - 1,
> @@ -455,7 +455,15 @@ listener_unregister_exit:
>
> void vfio_spapr_container_deinit(VFIOContainer *container)
> {
> + VFIOHostDMAWindow *hostwin, *next;
> +
> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
> memory_listener_unregister(&container->prereg_listener);
> }
> + QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
> + next) {
> + QLIST_REMOVE(hostwin, hostwin_next);
> + g_free(hostwin);
> + }
> +
> }
I am sure this change belongs to this patch.
Thanks,
C.
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size to base container
2023-10-26 10:30 ` [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
@ 2023-10-27 16:01 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 16:01 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> No functional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
Drop the Sob please,
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Thanks,
C.
> ---
> include/hw/vfio/vfio-common.h | 2 --
> include/hw/vfio/vfio-container-base.h | 2 ++
> hw/vfio/container.c | 11 ++++++-----
> 3 files changed, 8 insertions(+), 7 deletions(-)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 56452018a9..423ab2436c 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -80,8 +80,6 @@ typedef struct VFIOContainer {
> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
> MemoryListener prereg_listener;
> unsigned iommu_type;
> - uint64_t dirty_pgsizes;
> - uint64_t max_dirty_bitmap_size;
> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
> QLIST_HEAD(, VFIOGroup) group_list;
> GList *iova_ranges;
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 89642e6b45..526d23acfd 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -53,6 +53,8 @@ typedef struct VFIOContainerBase {
> MemoryListener listener;
> Error *error;
> bool initialized;
> + uint64_t dirty_pgsizes;
> + uint64_t max_dirty_bitmap_size;
> unsigned long pgsizes;
> unsigned int dma_max_mappings;
> bool dirty_pages_supported;
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index b8f36f56d2..68dc6d240f 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -64,6 +64,7 @@ static int vfio_dma_unmap_bitmap(VFIOContainer *container,
> hwaddr iova, ram_addr_t size,
> IOMMUTLBEntry *iotlb)
> {
> + VFIOContainerBase *bcontainer = &container->bcontainer;
> struct vfio_iommu_type1_dma_unmap *unmap;
> struct vfio_bitmap *bitmap;
> VFIOBitmap vbmap;
> @@ -91,7 +92,7 @@ static int vfio_dma_unmap_bitmap(VFIOContainer *container,
> bitmap->size = vbmap.size;
> bitmap->data = (__u64 *)vbmap.bitmap;
>
> - if (vbmap.size > container->max_dirty_bitmap_size) {
> + if (vbmap.size > bcontainer->max_dirty_bitmap_size) {
> error_report("UNMAP: Size of bitmap too big 0x%"PRIx64, vbmap.size);
> ret = -E2BIG;
> goto unmap_exit;
> @@ -131,7 +132,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
>
> if (iotlb && vfio_devices_all_running_and_mig_active(bcontainer)) {
> if (!vfio_devices_all_device_dirty_tracking(bcontainer) &&
> - container->bcontainer.dirty_pages_supported) {
> + bcontainer->dirty_pages_supported) {
> return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
> }
>
> @@ -154,7 +155,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase *bcontainer, hwaddr iova,
> if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
> container->iommu_type == VFIO_TYPE1v2_IOMMU) {
> trace_vfio_legacy_dma_unmap_overflow_workaround();
> - unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
> + unmap.size -= 1ULL << ctz64(bcontainer->pgsizes);
> continue;
> }
> error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
> @@ -469,8 +470,8 @@ static void vfio_get_iommu_info_migration(VFIOContainer *container,
> */
> if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
> bcontainer->dirty_pages_supported = true;
> - container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
> - container->dirty_pgsizes = cap_mig->pgsize_bitmap;
> + bcontainer->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
> + bcontainer->dirty_pgsizes = cap_mig->pgsize_bitmap;
> }
> }
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 18/37] vfio/container: Implement attach/detach_device
2023-10-26 10:30 ` [PATCH v3 18/37] vfio/container: Implement attach/detach_device Zhenzhong Duan
@ 2023-10-27 16:04 ` Cédric Le Goater
2023-10-27 16:06 ` Cédric Le Goater
1 sibling, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 16:04 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> hw/vfio/common.c | 16 ++++++++++++++++
> hw/vfio/container.c | 12 +++++-------
> 2 files changed, 21 insertions(+), 7 deletions(-)
>
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index d62c815d7f..64565b4ae9 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -1500,3 +1500,19 @@ retry:
>
> return info;
> }
> +
> +int vfio_attach_device(char *name, VFIODevice *vbasedev,
> + AddressSpace *as, Error **errp)
> +{
> + const VFIOIOMMUOps *ops = &vfio_legacy_ops;
> +
> + return ops->attach_device(name, vbasedev, as, errp);
> +}
> +
> +void vfio_detach_device(VFIODevice *vbasedev)
> +{
> + if (!vbasedev->bcontainer) {
> + return;
> + }
> + vbasedev->bcontainer->ops->detach_device(vbasedev);
> +}
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 36c34683ad..c8ff0f2037 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -874,8 +874,8 @@ static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
> * @name and @vbasedev->name are likely to be different depending
> * on the type of the device, hence the need for passing @name
> */
> -int vfio_attach_device(char *name, VFIODevice *vbasedev,
> - AddressSpace *as, Error **errp)
> +static int vfio_legacy_attach_device(char *name, VFIODevice *vbasedev,
> + AddressSpace *as, Error **errp)
> {
> int groupid = vfio_device_groupid(vbasedev, errp);
> VFIODevice *vbasedev_iter;
> @@ -915,14 +915,10 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
> return ret;
> }
>
> -void vfio_detach_device(VFIODevice *vbasedev)
> +static void vfio_legacy_detach_device(VFIODevice *vbasedev)
> {
> VFIOGroup *group = vbasedev->group;
>
> - if (!vbasedev->bcontainer) {
> - return;
> - }
> -
> QLIST_REMOVE(vbasedev, global_next);
> QLIST_REMOVE(vbasedev, container_next);
> vbasedev->bcontainer = NULL;
> @@ -934,6 +930,8 @@ void vfio_detach_device(VFIODevice *vbasedev)
> const VFIOIOMMUOps vfio_legacy_ops = {
> .dma_map = vfio_legacy_dma_map,
> .dma_unmap = vfio_legacy_dma_unmap,
> + .attach_device = vfio_legacy_attach_device,
> + .detach_device = vfio_legacy_detach_device,
> .set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
> .query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
> };
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface
2023-10-26 10:30 ` [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface Zhenzhong Duan
@ 2023-10-27 16:04 ` Cédric Le Goater
2023-10-30 3:15 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 16:04 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, Cédric Le Goater,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> Introduce an empry spapr backend which will hold spapr specific
> content, currently only prereg_listener and hostwin_list.
>
> Also introduce and instantiate a spapr specific target interface,
> currently only has add/del_window callbacks.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> include/hw/vfio/vfio-common.h | 8 ++++++++
> include/hw/vfio/vfio-container-base.h | 2 ++
> hw/vfio/spapr.c | 8 ++++++++
> 3 files changed, 18 insertions(+)
>
> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
> index 938f75e70c..a74e60e677 100644
> --- a/include/hw/vfio/vfio-common.h
> +++ b/include/hw/vfio/vfio-common.h
> @@ -169,6 +169,14 @@ VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
> void vfio_put_address_space(VFIOAddressSpace *space);
>
> /* SPAPR specific */
> +typedef struct VFIOIOMMUSpaprOps {
> + int (*add_window)(VFIOContainerBase *bcontainer,
> + MemoryRegionSection *section,
> + Error **errp);
> + void (*del_window)(VFIOContainerBase *bcontainer,
> + MemoryRegionSection *section);
> +} VFIOIOMMUSpaprOps;
I don't think we need a new structure type.
If possible, these callbacks should be under VFIOIOMMUOps and only
defined in the VFIOIOMMUOps struct for spapr. This struct would a
copy of vfio_legacy_ops plus the two add/del_window handlers.
Thanks,
C.
> +
> int vfio_container_add_section_window(VFIOContainer *container,
> MemoryRegionSection *section,
> Error **errp);
> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-container-base.h
> index 2ffafb0d58..1e1854d24f 100644
> --- a/include/hw/vfio/vfio-container-base.h
> +++ b/include/hw/vfio/vfio-container-base.h
> @@ -31,6 +31,7 @@
>
> typedef struct VFIODevice VFIODevice;
> typedef struct VFIOIOMMUOps VFIOIOMMUOps;
> +typedef struct VFIOIOMMUSpaprOps VFIOIOMMUSpaprOps;
>
> typedef struct {
> unsigned long *bitmap;
> @@ -49,6 +50,7 @@ typedef struct VFIOAddressSpace {
> */
> typedef struct VFIOContainerBase {
> const VFIOIOMMUOps *ops;
> + const VFIOIOMMUSpaprOps *spapr_ops;
> VFIOAddressSpace *space;
> MemoryListener listener;
> Error *error;
> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
> index 5786377317..3739004151 100644
> --- a/hw/vfio/spapr.c
> +++ b/hw/vfio/spapr.c
> @@ -24,6 +24,10 @@
> #include "qapi/error.h"
> #include "trace.h"
>
> +typedef struct VFIOSpaprContainer {
> + VFIOContainer container;
> +} VFIOSpaprContainer;
> +
> static bool vfio_prereg_listener_skipped_section(MemoryRegionSection *section)
> {
> if (memory_region_is_iommu(section->mr)) {
> @@ -384,6 +388,8 @@ void vfio_container_del_section_window(VFIOContainer *container,
> }
> }
>
> +const VFIOIOMMUSpaprOps vfio_iommu_spapr_ops;
> +
> bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> {
> VFIOContainerBase *bcontainer = &container->bcontainer;
> @@ -447,6 +453,8 @@ bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
> 0x1000);
> }
>
> + bcontainer->spapr_ops = &vfio_iommu_spapr_ops;
> +
> listener_unregister_exit:
> if (v2) {
> memory_listener_unregister(&container->prereg_listener);
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 18/37] vfio/container: Implement attach/detach_device
2023-10-26 10:30 ` [PATCH v3 18/37] vfio/container: Implement attach/detach_device Zhenzhong Duan
2023-10-27 16:04 ` Cédric Le Goater
@ 2023-10-27 16:06 ` Cédric Le Goater
2023-10-30 3:20 ` Duan, Zhenzhong
1 sibling, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-27 16:06 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Yi Sun
Sorry, previous email was empty. Friday effect !
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Eric Auger <eric.auger@redhat.com>
>
> No fucntional change intended.
>
> Signed-off-by: Eric Auger <eric.auger@redhat.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> hw/vfio/common.c | 16 ++++++++++++++++
> hw/vfio/container.c | 12 +++++-------
> 2 files changed, 21 insertions(+), 7 deletions(-)
>
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index d62c815d7f..64565b4ae9 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -1500,3 +1500,19 @@ retry:
>
> return info;
> }
> +
> +int vfio_attach_device(char *name, VFIODevice *vbasedev,
> + AddressSpace *as, Error **errp)
> +{
> + const VFIOIOMMUOps *ops = &vfio_legacy_ops;
hmm, this looks wrong. please explain.
Thanks,
C.
> +
> + return ops->attach_device(name, vbasedev, as, errp);
> +}
> +
> +void vfio_detach_device(VFIODevice *vbasedev)
> +{
> + if (!vbasedev->bcontainer) {
> + return;
> + }
> + vbasedev->bcontainer->ops->detach_device(vbasedev);
> +}
> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
> index 36c34683ad..c8ff0f2037 100644
> --- a/hw/vfio/container.c
> +++ b/hw/vfio/container.c
> @@ -874,8 +874,8 @@ static int vfio_device_groupid(VFIODevice *vbasedev, Error **errp)
> * @name and @vbasedev->name are likely to be different depending
> * on the type of the device, hence the need for passing @name
> */
> -int vfio_attach_device(char *name, VFIODevice *vbasedev,
> - AddressSpace *as, Error **errp)
> +static int vfio_legacy_attach_device(char *name, VFIODevice *vbasedev,
> + AddressSpace *as, Error **errp)
> {
> int groupid = vfio_device_groupid(vbasedev, errp);
> VFIODevice *vbasedev_iter;
> @@ -915,14 +915,10 @@ int vfio_attach_device(char *name, VFIODevice *vbasedev,
> return ret;
> }
>
> -void vfio_detach_device(VFIODevice *vbasedev)
> +static void vfio_legacy_detach_device(VFIODevice *vbasedev)
> {
> VFIOGroup *group = vbasedev->group;
>
> - if (!vbasedev->bcontainer) {
> - return;
> - }
> -
> QLIST_REMOVE(vbasedev, global_next);
> QLIST_REMOVE(vbasedev, container_next);
> vbasedev->bcontainer = NULL;
> @@ -934,6 +930,8 @@ void vfio_detach_device(VFIODevice *vbasedev)
> const VFIOIOMMUOps vfio_legacy_ops = {
> .dma_map = vfio_legacy_dma_map,
> .dma_unmap = vfio_legacy_dma_unmap,
> + .attach_device = vfio_legacy_attach_device,
> + .detach_device = vfio_legacy_detach_device,
> .set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
> .query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
> };
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-27 14:02 ` Cédric Le Goater
@ 2023-10-30 2:40 ` Duan, Zhenzhong
2023-10-31 7:57 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:40 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:03 PM
>Subject: Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> Introduce a dumb VFIOContainer base object and its targetted interface.
>
>targeted
Will fix.
>
>
>> This is willingly not a QOM object because we don't want it to be
>> visible from the user interface. The VFIOContainer will be smoothly
>> populated in subsequent patches as well as interfaces.
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>> include/hw/vfio/vfio-common.h | 8 +---
>> include/hw/vfio/vfio-container-base.h | 64 +++++++++++++++++++++++++++
>> 2 files changed, 66 insertions(+), 6 deletions(-)
>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index b9c7a7e588..d8f293cb57 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -30,6 +30,7 @@
>> #include <linux/vfio.h>
>> #endif
>> #include "sysemu/sysemu.h"
>> +#include "hw/vfio/vfio-container-base.h"
>>
>> #define VFIO_MSG_PREFIX "vfio %s: "
>>
>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>> struct VFIOGroup;
>>
>> typedef struct VFIOContainer {
>> + VFIOContainerBase bcontainer;
>> VFIOAddressSpace *space;
>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>> MemoryListener listener;
>> @@ -201,12 +203,6 @@ typedef struct VFIODisplay {
>> } dmabuf;
>> } VFIODisplay;
>>
>> -typedef struct {
>> - unsigned long *bitmap;
>> - hwaddr size;
>> - hwaddr pages;
>> -} VFIOBitmap;
>> -
>> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
>> void vfio_put_address_space(VFIOAddressSpace *space);
>> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> new file mode 100644
>> index 0000000000..5becbd51a7
>> --- /dev/null
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -0,0 +1,64 @@
>> +/*
>> + * VFIO BASE CONTAINER
>> + *
>> + * Copyright (C) 2023 Intel Corporation.
>> + * Copyright Red Hat, Inc. 2023
>> + *
>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>> + * Eric Auger <eric.auger@redhat.com>
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> +
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>> + * GNU General Public License for more details.
>> +
>> + * You should have received a copy of the GNU General Public License along
>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>
>This should be enough :
>
> SPDX-License-Identifier: GPL-2.0-or-later
Will do.
>
>> + */
>> +
>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>
>HW_VFIO_VFIO_CONTAINER_BASE_H
>
>> +
>> +#include "exec/memory.h"
>> +#ifndef CONFIG_USER_ONLY
>> +#include "exec/hwaddr.h"
>> +#endif
>
>including "exec/memory.h" should be enough.
Will do.
>
>
>> +
>> +typedef struct VFIODevice VFIODevice;
>> +typedef struct VFIOIOMMUOps VFIOIOMMUOps;
>> +
>> +typedef struct {
>> + unsigned long *bitmap;
>> + hwaddr size;
>> + hwaddr pages;
>> +} VFIOBitmap;
>> +
>> +/*
>> + * This is the base object for vfio container backends
>> + */
>> +typedef struct VFIOContainerBase {
>> + const VFIOIOMMUOps *ops;
>> +} VFIOContainerBase;
>> +
>> +struct VFIOIOMMUOps {
>> + /* basic feature */
>> + int (*dma_map)(VFIOContainerBase *bcontainer,
>> + hwaddr iova, ram_addr_t size,
>> + void *vaddr, bool readonly);
>> + int (*dma_unmap)(VFIOContainerBase *bcontainer,
>> + hwaddr iova, ram_addr_t size,
>> + IOMMUTLBEntry *iotlb);
>
>Could the VFIOContainerBase *parameter be const ?
Yes, VFIOContainerBase is not changed by dma_unmap and other
functions dma_unmap calls. I tried and found making it const here
would impact all functions it called with same parameter be const
in following patches which looks unrelated to the patch itself
to avoid compile error.
E.g. below functions are impacted,
vfio_devices_all_running_and_mig_active
vfio_devices_all_device_dirty_tracking
vfio_devices_query_dirty_bitmap
vfio_container_query_dirty_bitmap
vfio_legacy_query_dirty_bitmap
To make following patches cleaner to review, I would like to keep
current code except you or others have a strong opinion.
Another choice I can think of is to add const to all impact functions
in a separate patch.
>
>> + int (*attach_device)(char *name, VFIODevice *vbasedev,
>
>cont char *name ?
Yes, will fix.
Thanks
Zhenzhong
>
>The rest looks good to me.
>
>Thanks
>
>C.
>
>
>> + AddressSpace *as, Error **errp);
>> + void (*detach_device)(VFIODevice *vbasedev);
>> + /* migration feature */
>> + int (*set_dirty_page_tracking)(VFIOContainerBase *bcontainer, bool start);
>> + int (*query_dirty_bitmap)(VFIOContainerBase *bcontainer, VFIOBitmap
>*vbmap,
>> + hwaddr iova, hwaddr size);
>> +};
>> +#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps
2023-10-27 14:20 ` Cédric Le Goater
@ 2023-10-30 2:43 ` Duan, Zhenzhong
2023-10-31 8:21 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:43 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:21 PM
>Subject: Re: [PATCH v3 07/37] vfio/container: Introduce a empty
>VFIOIOMMUOps
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> This empty VFIOIOMMUOps named vfio_legacy_ops will hold all general
>> IOMMU ops of legacy container.
>>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>> include/hw/vfio/vfio-common.h | 2 +-
>> hw/vfio/container.c | 5 +++++
>> 2 files changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index d8f293cb57..8ded5cd8e4 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -255,7 +255,7 @@ typedef QLIST_HEAD(VFIOGroupList, VFIOGroup)
>VFIOGroupList;
>> typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
>> extern VFIOGroupList vfio_group_list;
>> extern VFIODeviceList vfio_device_list;
>> -
>> +extern const VFIOIOMMUOps vfio_legacy_ops;
>
>
>why does it need to be external ?
It is referenced by vfio_connect_container() and vfio_attach_device().
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API
2023-10-27 14:26 ` Cédric Le Goater
@ 2023-10-30 2:48 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:48 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:26 PM
>Subject: Re: [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
...
>> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
>> new file mode 100644
>> index 0000000000..9db8b89b2f
>> --- /dev/null
>> +++ b/hw/vfio/container-base.c
>> @@ -0,0 +1,49 @@
>> +/*
>> + * VFIO BASE CONTAINER
>> + *
>> + * Copyright (C) 2023 Intel Corporation.
>> + * Copyright Red Hat, Inc. 2023
>> + *
>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>> + * Eric Auger <eric.auger@redhat.com>
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License as published by
>> + * the Free Software Foundation; either version 2 of the License, or
>> + * (at your option) any later version.
>> +
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>> + * GNU General Public License for more details.
>> +
>> + * You should have received a copy of the GNU General Public License along
>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>> + */
>
>Please reduce the header with an SPDX identifier.
Will fix.
>
>> +
>> +#include "qemu/osdep.h"
>> +#include "qapi/error.h"
>> +#include "qemu/error-report.h"
>> +#include "hw/vfio/vfio-container-base.h"
>> +
>> +int vfio_container_dma_map(VFIOContainerBase *bcontainer,
>> + hwaddr iova, ram_addr_t size,
>> + void *vaddr, bool readonly)
>> +{
>> + if (!bcontainer->ops->dma_map) {
>
>I think this could be an assert instead ?
Good suggestion, will fix.
>
>> + return -EINVAL;
>> + }
>> +
>> + return bcontainer->ops->dma_map(bcontainer, iova, size, vaddr, readonly);
>> +}
>> +
>> +int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
>> + hwaddr iova, ram_addr_t size,
>> + IOMMUTLBEntry *iotlb)
>> +{
>> + if (!bcontainer->ops->dma_unmap) {
>> + return -EINVAL;
>> + }
and here.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 09/37] vfio/common: Move giommu_list in base container
2023-10-27 14:38 ` Cédric Le Goater
@ 2023-10-30 2:48 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:48 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:39 PM
>Subject: Re: [PATCH v3 09/37] vfio/common: Move giommu_list in base
>container
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> Move the giommu_list field in the base object and store the
>> base container in the VFIOGuestIOMMU.
>>
>> We introduce vfio_container_init/destroy helper on the base
>> container.
>
>This could have been a prereq patch.
Yes.
>
>>
>> No fucntional change intended.
>
>functional
Will fix.
>
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> [ clg: context changes ]
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>
>You can drop my Sob here and in all patches. The changes I did were
>due to the recent 'iova_ranges' addition and the context was fuzzy.
>Nothing major.
Got it, will do.
Thanks
Zhenzhong
>
>The rest looks good.
>
>Thanks,
>
>C.
>
>> ---
>> include/hw/vfio/vfio-common.h | 9 ---------
>> include/hw/vfio/vfio-container-base.h | 13 +++++++++++++
>> hw/vfio/common.c | 17 +++++++++++------
>> hw/vfio/container-base.c | 18 ++++++++++++++++++
>> hw/vfio/container.c | 13 +++----------
>> 5 files changed, 45 insertions(+), 25 deletions(-)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index 97056224f4..fcb4003a21 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -95,7 +95,6 @@ typedef struct VFIOContainer {
>> uint64_t max_dirty_bitmap_size;
>> unsigned long pgsizes;
>> unsigned int dma_max_mappings;
>> - QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
>> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
>> QLIST_HEAD(, VFIOGroup) group_list;
>> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
>> @@ -104,14 +103,6 @@ typedef struct VFIOContainer {
>> GList *iova_ranges;
>> } VFIOContainer;
>>
>> -typedef struct VFIOGuestIOMMU {
>> - VFIOContainer *container;
>> - IOMMUMemoryRegion *iommu_mr;
>> - hwaddr iommu_offset;
>> - IOMMUNotifier n;
>> - QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
>> -} VFIOGuestIOMMU;
>> -
>> typedef struct VFIORamDiscardListener {
>> VFIOContainer *container;
>> MemoryRegion *mr;
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> index 077e638ee8..71e1e4324e 100644
>> --- a/include/hw/vfio/vfio-container-base.h
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -43,8 +43,17 @@ typedef struct {
>> */
>> typedef struct VFIOContainerBase {
>> const VFIOIOMMUOps *ops;
>> + QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
>> } VFIOContainerBase;
>>
>> +typedef struct VFIOGuestIOMMU {
>> + VFIOContainerBase *bcontainer;
>> + IOMMUMemoryRegion *iommu_mr;
>> + hwaddr iommu_offset;
>> + IOMMUNotifier n;
>> + QLIST_ENTRY(VFIOGuestIOMMU) giommu_next;
>> +} VFIOGuestIOMMU;
>> +
>> int vfio_container_dma_map(VFIOContainerBase *bcontainer,
>> hwaddr iova, ram_addr_t size,
>> void *vaddr, bool readonly);
>> @@ -52,6 +61,10 @@ int vfio_container_dma_unmap(VFIOContainerBase
>*bcontainer,
>> hwaddr iova, ram_addr_t size,
>> IOMMUTLBEntry *iotlb);
>>
>> +void vfio_container_init(VFIOContainerBase *bcontainer,
>> + const VFIOIOMMUOps *ops);
>> +void vfio_container_destroy(VFIOContainerBase *bcontainer);
>> +
>> struct VFIOIOMMUOps {
>> /* basic feature */
>> int (*dma_map)(VFIOContainerBase *bcontainer,
>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>> index 141f2b54a4..4f130ad87c 100644
>> --- a/hw/vfio/common.c
>> +++ b/hw/vfio/common.c
>> @@ -292,7 +292,7 @@ static bool vfio_get_xlat_addr(IOMMUTLBEntry *iotlb,
>void **vaddr,
>> static void vfio_iommu_map_notify(IOMMUNotifier *n, IOMMUTLBEntry
>*iotlb)
>> {
>> VFIOGuestIOMMU *giommu = container_of(n, VFIOGuestIOMMU, n);
>> - VFIOContainerBase *bcontainer = &giommu->container->bcontainer;
>> + VFIOContainerBase *bcontainer = giommu->bcontainer;
>> hwaddr iova = iotlb->iova + giommu->iommu_offset;
>> void *vaddr;
>> int ret;
>> @@ -569,6 +569,7 @@ static void vfio_listener_region_add(MemoryListener
>*listener,
>> MemoryRegionSection *section)
>> {
>> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>> hwaddr iova, end;
>> Int128 llend, llsize;
>> void *vaddr;
>> @@ -613,7 +614,7 @@ static void vfio_listener_region_add(MemoryListener
>*listener,
>> giommu->iommu_mr = iommu_mr;
>> giommu->iommu_offset = section->offset_within_address_space -
>> section->offset_within_region;
>> - giommu->container = container;
>> + giommu->bcontainer = bcontainer;
>> llend = int128_add(int128_make64(section->offset_within_region),
>> section->size);
>> llend = int128_sub(llend, int128_one());
>> @@ -648,7 +649,7 @@ static void vfio_listener_region_add(MemoryListener
>*listener,
>> g_free(giommu);
>> goto fail;
>> }
>> - QLIST_INSERT_HEAD(&container->giommu_list, giommu, giommu_next);
>> + QLIST_INSERT_HEAD(&bcontainer->giommu_list, giommu, giommu_next);
>> memory_region_iommu_replay(giommu->iommu_mr, &giommu->n);
>>
>> return;
>> @@ -733,6 +734,7 @@ static void vfio_listener_region_del(MemoryListener
>*listener,
>> MemoryRegionSection *section)
>> {
>> VFIOContainer *container = container_of(listener, VFIOContainer, listener);
>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>> hwaddr iova, end;
>> Int128 llend, llsize;
>> int ret;
>> @@ -745,7 +747,7 @@ static void vfio_listener_region_del(MemoryListener
>*listener,
>> if (memory_region_is_iommu(section->mr)) {
>> VFIOGuestIOMMU *giommu;
>>
>> - QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
>> + QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
>> if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
>> giommu->n.start == section->offset_within_region) {
>> memory_region_unregister_iommu_notifier(section->mr,
>> @@ -1208,7 +1210,9 @@ static void
>vfio_iommu_map_dirty_notify(IOMMUNotifier *n, IOMMUTLBEntry *iotlb)
>> vfio_giommu_dirty_notifier *gdn = container_of(n,
>> vfio_giommu_dirty_notifier, n);
>> VFIOGuestIOMMU *giommu = gdn->giommu;
>> - VFIOContainer *container = giommu->container;
>> + VFIOContainerBase *bcontainer = giommu->bcontainer;
>> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
>> + bcontainer);
>> hwaddr iova = iotlb->iova + giommu->iommu_offset;
>> ram_addr_t translated_addr;
>> int ret = -EINVAL;
>> @@ -1286,12 +1290,13 @@ static int
>vfio_sync_ram_discard_listener_dirty_bitmap(VFIOContainer *container,
>> static int vfio_sync_dirty_bitmap(VFIOContainer *container,
>> MemoryRegionSection *section)
>> {
>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>> ram_addr_t ram_addr;
>>
>> if (memory_region_is_iommu(section->mr)) {
>> VFIOGuestIOMMU *giommu;
>>
>> - QLIST_FOREACH(giommu, &container->giommu_list, giommu_next) {
>> + QLIST_FOREACH(giommu, &bcontainer->giommu_list, giommu_next) {
>> if (MEMORY_REGION(giommu->iommu_mr) == section->mr &&
>> giommu->n.start == section->offset_within_region) {
>> Int128 llend;
>> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
>> index 9db8b89b2f..6be7ce65ad 100644
>> --- a/hw/vfio/container-base.c
>> +++ b/hw/vfio/container-base.c
>> @@ -47,3 +47,21 @@ int vfio_container_dma_unmap(VFIOContainerBase
>*bcontainer,
>>
>> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
>> }
>> +
>> +void vfio_container_init(VFIOContainerBase *bcontainer, const
>VFIOIOMMUOps *ops)
>> +{
>> + bcontainer->ops = ops;
>> + QLIST_INIT(&bcontainer->giommu_list);
>> +}
>> +
>> +void vfio_container_destroy(VFIOContainerBase *bcontainer)
>> +{
>> + VFIOGuestIOMMU *giommu, *tmp;
>> +
>> + QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next,
>tmp) {
>> + memory_region_unregister_iommu_notifier(
>> + MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>> + QLIST_REMOVE(giommu, giommu_next);
>> + g_free(giommu);
>> + }
>> +}
>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>> index c04df26323..b2b6e05287 100644
>> --- a/hw/vfio/container.c
>> +++ b/hw/vfio/container.c
>> @@ -556,10 +556,9 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> container->dirty_pages_supported = false;
>> container->dma_max_mappings = 0;
>> container->iova_ranges = NULL;
>> - QLIST_INIT(&container->giommu_list);
>> QLIST_INIT(&container->vrdl_list);
>> bcontainer = &container->bcontainer;
>> - bcontainer->ops = &vfio_legacy_ops;
>> + vfio_container_init(bcontainer, &vfio_legacy_ops);
>>
>> ret = vfio_init_container(container, group->fd, errp);
>> if (ret) {
>> @@ -661,6 +660,7 @@ put_space_exit:
>> static void vfio_disconnect_container(VFIOGroup *group)
>> {
>> VFIOContainer *container = group->container;
>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>>
>> QLIST_REMOVE(group, container_next);
>> group->container = NULL;
>> @@ -685,16 +685,9 @@ static void vfio_disconnect_container(VFIOGroup
>*group)
>>
>> if (QLIST_EMPTY(&container->group_list)) {
>> VFIOAddressSpace *space = container->space;
>> - VFIOGuestIOMMU *giommu, *tmp;
>>
>> QLIST_REMOVE(container, next);
>> -
>> - QLIST_FOREACH_SAFE(giommu, &container->giommu_list, giommu_next,
>tmp) {
>> - memory_region_unregister_iommu_notifier(
>> - MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>> - QLIST_REMOVE(giommu, giommu_next);
>> - g_free(giommu);
>> - }
>> + vfio_container_destroy(bcontainer);
>>
>> trace_vfio_disconnect_container(container->fd);
>> close(container->fd);
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 10/37] vfio/container: Move space field to base container
2023-10-27 14:46 ` Cédric Le Goater
@ 2023-10-30 2:51 ` Duan, Zhenzhong
2023-10-31 8:24 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:51 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:46 PM
>Subject: Re: [PATCH v3 10/37] vfio/container: Move space field to base container
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> Move the space field to the base object. Also the VFIOAddressSpace
>> now contains a list of base containers.
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> [ clg: context changes ]
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>> ---
>> include/hw/vfio/vfio-common.h | 8 --------
>> include/hw/vfio/vfio-container-base.h | 9 +++++++++
>> hw/ppc/spapr_pci_vfio.c | 10 +++++-----
>> hw/vfio/common.c | 4 ++--
>> hw/vfio/container-base.c | 6 +++++-
>> hw/vfio/container.c | 18 +++++++++---------
>> 6 files changed, 30 insertions(+), 25 deletions(-)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index fcb4003a21..857d2b8076 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -73,17 +73,10 @@ typedef struct VFIOMigration {
>> bool initial_data_sent;
>> } VFIOMigration;
>>
>> -typedef struct VFIOAddressSpace {
>> - AddressSpace *as;
>> - QLIST_HEAD(, VFIOContainer) containers;
>> - QLIST_ENTRY(VFIOAddressSpace) list;
>> -} VFIOAddressSpace;
>> -
>> struct VFIOGroup;
>>
>> typedef struct VFIOContainer {
>> VFIOContainerBase bcontainer;
>> - VFIOAddressSpace *space;
>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>> MemoryListener listener;
>> MemoryListener prereg_listener;
>> @@ -98,7 +91,6 @@ typedef struct VFIOContainer {
>> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
>> QLIST_HEAD(, VFIOGroup) group_list;
>> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
>> - QLIST_ENTRY(VFIOContainer) next;
>> QLIST_HEAD(, VFIODevice) device_list;
>> GList *iova_ranges;
>> } VFIOContainer;
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> index 71e1e4324e..a5fef3e6a8 100644
>> --- a/include/hw/vfio/vfio-container-base.h
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -38,12 +38,20 @@ typedef struct {
>> hwaddr pages;
>> } VFIOBitmap;
>>
>> +typedef struct VFIOAddressSpace {
>> + AddressSpace *as;
>> + QLIST_HEAD(, VFIOContainerBase) containers;
>> + QLIST_ENTRY(VFIOAddressSpace) list;
>> +} VFIOAddressSpace;
>> +
>> /*
>> * This is the base object for vfio container backends
>> */
>> typedef struct VFIOContainerBase {
>> const VFIOIOMMUOps *ops;
>> + VFIOAddressSpace *space;
>> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
>> + QLIST_ENTRY(VFIOContainerBase) next;
>> } VFIOContainerBase;
>>
>> typedef struct VFIOGuestIOMMU {
>> @@ -62,6 +70,7 @@ int vfio_container_dma_unmap(VFIOContainerBase
>*bcontainer,
>> IOMMUTLBEntry *iotlb);
>>
>> void vfio_container_init(VFIOContainerBase *bcontainer,
>> + VFIOAddressSpace *space,
>> const VFIOIOMMUOps *ops);
>> void vfio_container_destroy(VFIOContainerBase *bcontainer);
>>
>> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
>> index f283f7e38d..d1d07bec46 100644
>> --- a/hw/ppc/spapr_pci_vfio.c
>> +++ b/hw/ppc/spapr_pci_vfio.c
>> @@ -84,27 +84,27 @@ static int vfio_eeh_container_op(VFIOContainer
>*container, uint32_t op)
>> static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
>> {
>> VFIOAddressSpace *space = vfio_get_address_space(as);
>> - VFIOContainer *container = NULL;
>> + VFIOContainerBase *bcontainer = NULL;
>>
>> if (QLIST_EMPTY(&space->containers)) {
>> /* No containers to act on */
>> goto out;
>> }
>>
>> - container = QLIST_FIRST(&space->containers);
>> + bcontainer = QLIST_FIRST(&space->containers);
>>
>> - if (QLIST_NEXT(container, next)) {
>> + if (QLIST_NEXT(bcontainer, next)) {
>> /*
>> * We don't yet have logic to synchronize EEH state across
>> * multiple containers
>> */
>> - container = NULL;
>> + bcontainer = NULL;
>> goto out;
>> }
>>
>> out:
>> vfio_put_address_space(space);
>> - return container;
>> + return container_of(bcontainer, VFIOContainer, bcontainer);
>> }
>>
>> static bool vfio_eeh_as_ok(AddressSpace *as)
>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>> index 4f130ad87c..f87a0dcec3 100644
>> --- a/hw/vfio/common.c
>> +++ b/hw/vfio/common.c
>> @@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
>>
>> bool vfio_viommu_preset(VFIODevice *vbasedev)
>> {
>> - return vbasedev->container->space->as != &address_space_memory;
>> + return vbasedev->container->bcontainer.space->as !=
>&address_space_memory;
>> }
>>
>> static void vfio_set_migration_error(int err)
>> @@ -924,7 +924,7 @@ static void vfio_dirty_tracking_init(VFIOContainer
>*container,
>> dirty.container = container;
>>
>> memory_listener_register(&dirty.listener,
>> - container->space->as);
>> + container->bcontainer.space->as);
>
>Could we introduce an helper returning :
>
> container->bcontainer.space->as
Guess you mean this is too long? Then I would like to use
bcontainer->space->as except you have a strong opinion on that.
Thanks
Zhenzhong
>
>The rest looks OK.
>
>Thanks,
>
>C.
>
>>
>> *ranges = dirty.ranges;
>>
>> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
>> index 6be7ce65ad..99b2957d7b 100644
>> --- a/hw/vfio/container-base.c
>> +++ b/hw/vfio/container-base.c
>> @@ -48,9 +48,11 @@ int vfio_container_dma_unmap(VFIOContainerBase
>*bcontainer,
>> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
>> }
>>
>> -void vfio_container_init(VFIOContainerBase *bcontainer, const
>VFIOIOMMUOps *ops)
>> +void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace
>*space,
>> + const VFIOIOMMUOps *ops)
>> {
>> bcontainer->ops = ops;
>> + bcontainer->space = space;
>> QLIST_INIT(&bcontainer->giommu_list);
>> }
>>
>> @@ -58,6 +60,8 @@ void vfio_container_destroy(VFIOContainerBase
>*bcontainer)
>> {
>> VFIOGuestIOMMU *giommu, *tmp;
>>
>> + QLIST_REMOVE(bcontainer, next);
>> +
>> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next,
>tmp) {
>> memory_region_unregister_iommu_notifier(
>> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>> index b2b6e05287..761310fa51 100644
>> --- a/hw/vfio/container.c
>> +++ b/hw/vfio/container.c
>> @@ -514,7 +514,8 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> * details once we know which type of IOMMU we are using.
>> */
>>
>> - QLIST_FOREACH(container, &space->containers, next) {
>> + QLIST_FOREACH(bcontainer, &space->containers, next) {
>> + container = container_of(bcontainer, VFIOContainer, bcontainer);
>> if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
>> ret = vfio_ram_block_discard_disable(container, true);
>> if (ret) {
>> @@ -550,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> }
>>
>> container = g_malloc0(sizeof(*container));
>> - container->space = space;
>> container->fd = fd;
>> container->error = NULL;
>> container->dirty_pages_supported = false;
>> @@ -558,7 +558,7 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> container->iova_ranges = NULL;
>> QLIST_INIT(&container->vrdl_list);
>> bcontainer = &container->bcontainer;
>> - vfio_container_init(bcontainer, &vfio_legacy_ops);
>> + vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>>
>> ret = vfio_init_container(container, group->fd, errp);
>> if (ret) {
>> @@ -613,14 +613,15 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> vfio_kvm_device_add_group(group);
>>
>> QLIST_INIT(&container->group_list);
>> - QLIST_INSERT_HEAD(&space->containers, container, next);
>> + QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
>>
>> group->container = container;
>> QLIST_INSERT_HEAD(&container->group_list, group, container_next);
>>
>> container->listener = vfio_memory_listener;
>>
>> - memory_listener_register(&container->listener, container->space->as);
>> + memory_listener_register(&container->listener,
>> + container->bcontainer.space->as);
>>
>> if (container->error) {
>> ret = -1;
>> @@ -634,7 +635,7 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> return 0;
>> listener_release_exit:
>> QLIST_REMOVE(group, container_next);
>> - QLIST_REMOVE(container, next);
>> + QLIST_REMOVE(bcontainer, next);
>> vfio_kvm_device_del_group(group);
>> memory_listener_unregister(&container->listener);
>> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
>> @@ -684,9 +685,8 @@ static void vfio_disconnect_container(VFIOGroup
>*group)
>> }
>>
>> if (QLIST_EMPTY(&container->group_list)) {
>> - VFIOAddressSpace *space = container->space;
>> + VFIOAddressSpace *space = container->bcontainer.space;
>>
>> - QLIST_REMOVE(container, next);
>> vfio_container_destroy(bcontainer);
>>
>> trace_vfio_disconnect_container(container->fd);
>> @@ -706,7 +706,7 @@ static VFIOGroup *vfio_get_group(int groupid,
>AddressSpace *as, Error **errp)
>> QLIST_FOREACH(group, &vfio_group_list, next) {
>> if (group->groupid == groupid) {
>> /* Found it. Now is it already in the right context? */
>> - if (group->container->space->as == as) {
>> + if (group->container->bcontainer.space->as == as) {
>> return group;
>> } else {
>> error_setg(errp, "group %d used in multiple address spaces",
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API
2023-10-27 14:53 ` Cédric Le Goater
@ 2023-10-30 2:53 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 2:53 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 10:53 PM
>Subject: Re: [PATCH v3 11/37] vfio/container: Switch to IOMMU BE
>set_dirty_page_tracking/query_dirty_bitmap API
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> dirty_pages_supported field is also moved to the base container
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>> ---
>> include/hw/vfio/vfio-common.h | 6 ------
>> include/hw/vfio/vfio-container-base.h | 6 ++++++
>> hw/vfio/common.c | 12 ++++++++----
>> hw/vfio/container-base.c | 23 +++++++++++++++++++++++
>> hw/vfio/container.c | 21 ++++++++++++++-------
>> 5 files changed, 51 insertions(+), 17 deletions(-)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index 857d2b8076..d053c61872 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -83,7 +83,6 @@ typedef struct VFIOContainer {
>> unsigned iommu_type;
>> Error *error;
>> bool initialized;
>> - bool dirty_pages_supported;
>> uint64_t dirty_pgsizes;
>> uint64_t max_dirty_bitmap_size;
>> unsigned long pgsizes;
>> @@ -190,11 +189,6 @@ VFIOAddressSpace
>*vfio_get_address_space(AddressSpace *as);
>> void vfio_put_address_space(VFIOAddressSpace *space);
>> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>>
>> -/* container->fd */
>> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start);
>> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
>> - hwaddr iova, hwaddr size);
>> -
>> /* SPAPR specific */
>> int vfio_container_add_section_window(VFIOContainer *container,
>> MemoryRegionSection *section,
>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>container-base.h
>> index a5fef3e6a8..ea8436a064 100644
>> --- a/include/hw/vfio/vfio-container-base.h
>> +++ b/include/hw/vfio/vfio-container-base.h
>> @@ -50,6 +50,7 @@ typedef struct VFIOAddressSpace {
>> typedef struct VFIOContainerBase {
>> const VFIOIOMMUOps *ops;
>> VFIOAddressSpace *space;
>> + bool dirty_pages_supported;
>> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
>> QLIST_ENTRY(VFIOContainerBase) next;
>> } VFIOContainerBase;
>> @@ -68,6 +69,11 @@ int vfio_container_dma_map(VFIOContainerBase
>*bcontainer,
>> int vfio_container_dma_unmap(VFIOContainerBase *bcontainer,
>> hwaddr iova, ram_addr_t size,
>> IOMMUTLBEntry *iotlb);
>> +int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
>> + bool start);
>> +int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
>> + VFIOBitmap *vbmap,
>> + hwaddr iova, hwaddr size);
>>
>> void vfio_container_init(VFIOContainerBase *bcontainer,
>> VFIOAddressSpace *space,
>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>> index f87a0dcec3..7d9b87fc67 100644
>> --- a/hw/vfio/common.c
>> +++ b/hw/vfio/common.c
>> @@ -1081,7 +1081,8 @@ static void
>vfio_listener_log_global_start(MemoryListener *listener)
>> if (vfio_devices_all_device_dirty_tracking(container)) {
>> ret = vfio_devices_dma_logging_start(container);
>> } else {
>> - ret = vfio_set_dirty_page_tracking(container, true);
>> + ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
>> + true);
>> }
>>
>> if (ret) {
>> @@ -1099,7 +1100,8 @@ static void
>vfio_listener_log_global_stop(MemoryListener *listener)
>> if (vfio_devices_all_device_dirty_tracking(container)) {
>> vfio_devices_dma_logging_stop(container);
>> } else {
>> - ret = vfio_set_dirty_page_tracking(container, false);
>> + ret = vfio_container_set_dirty_page_tracking(&container->bcontainer,
>> + false);
>> }
>>
>> if (ret) {
>> @@ -1167,7 +1169,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container,
>uint64_t iova,
>> VFIOBitmap vbmap;
>> int ret;
>>
>> - if (!container->dirty_pages_supported && !all_device_dirty_tracking) {
>> + if (!container->bcontainer.dirty_pages_supported &&
>> + !all_device_dirty_tracking) {
>> cpu_physical_memory_set_dirty_range(ram_addr, size,
>> tcg_enabled() ? DIRTY_CLIENTS_ALL :
>> DIRTY_CLIENTS_NOCODE);
>> @@ -1182,7 +1185,8 @@ int vfio_get_dirty_bitmap(VFIOContainer *container,
>uint64_t iova,
>> if (all_device_dirty_tracking) {
>> ret = vfio_devices_query_dirty_bitmap(container, &vbmap, iova, size);
>> } else {
>> - ret = vfio_query_dirty_bitmap(container, &vbmap, iova, size);
>> + ret = vfio_container_query_dirty_bitmap(&container->bcontainer,
>&vbmap,
>> + iova, size);
>> }
>>
>> if (ret) {
>> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
>> index 99b2957d7b..a7cf517dd2 100644
>> --- a/hw/vfio/container-base.c
>> +++ b/hw/vfio/container-base.c
>> @@ -48,11 +48,34 @@ int vfio_container_dma_unmap(VFIOContainerBase
>*bcontainer,
>> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
>> }
>>
>> +int vfio_container_set_dirty_page_tracking(VFIOContainerBase *bcontainer,
>> + bool start)
>> +{
>> + /* Fallback to all pages dirty if dirty page sync isn't supported */
>> + if (!bcontainer->ops->set_dirty_page_tracking) {
>> + return 0;
>> + }
>
>
>I would start with an assert and relax the check later on, if needed and
>in its own patch.
Good suggestions! Will do.
>
>> + return bcontainer->ops->set_dirty_page_tracking(bcontainer, start);
>> +}
>> +
>> +int vfio_container_query_dirty_bitmap(VFIOContainerBase *bcontainer,
>> + VFIOBitmap *vbmap,
>> + hwaddr iova, hwaddr size)
>> +{
>> + if (!bcontainer->ops->query_dirty_bitmap) {
>> + return -EINVAL;
>> + }
>
>Same comment.
Yes, thanks
BRs.
Zhenzhong
>
>Thanks,
>
>C.
>
>> + return bcontainer->ops->query_dirty_bitmap(bcontainer, vbmap, iova, size);
>> +}
>> +
>> void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace
>*space,
>> const VFIOIOMMUOps *ops)
>> {
>> bcontainer->ops = ops;
>> bcontainer->space = space;
>> + bcontainer->dirty_pages_supported = false;
>> QLIST_INIT(&bcontainer->giommu_list);
>> }
>>
>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>> index 761310fa51..6f02ca133e 100644
>> --- a/hw/vfio/container.c
>> +++ b/hw/vfio/container.c
>> @@ -131,7 +131,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase
>*bcontainer, hwaddr iova,
>>
>> if (iotlb && vfio_devices_all_running_and_mig_active(container)) {
>> if (!vfio_devices_all_device_dirty_tracking(container) &&
>> - container->dirty_pages_supported) {
>> + container->bcontainer.dirty_pages_supported) {
>> return vfio_dma_unmap_bitmap(container, iova, size, iotlb);
>> }
>>
>> @@ -205,14 +205,17 @@ static int vfio_legacy_dma_map(VFIOContainerBase
>*bcontainer, hwaddr iova,
>> return -errno;
>> }
>>
>> -int vfio_set_dirty_page_tracking(VFIOContainer *container, bool start)
>> +static int vfio_legacy_set_dirty_page_tracking(VFIOContainerBase
>*bcontainer,
>> + bool start)
>> {
>> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
>> + bcontainer);
>> int ret;
>> struct vfio_iommu_type1_dirty_bitmap dirty = {
>> .argsz = sizeof(dirty),
>> };
>>
>> - if (!container->dirty_pages_supported) {
>> + if (!bcontainer->dirty_pages_supported) {
>> return 0;
>> }
>>
>> @@ -232,9 +235,12 @@ int vfio_set_dirty_page_tracking(VFIOContainer
>*container, bool start)
>> return ret;
>> }
>>
>> -int vfio_query_dirty_bitmap(VFIOContainer *container, VFIOBitmap *vbmap,
>> - hwaddr iova, hwaddr size)
>> +static int vfio_legacy_query_dirty_bitmap(VFIOContainerBase *bcontainer,
>> + VFIOBitmap *vbmap,
>> + hwaddr iova, hwaddr size)
>> {
>> + VFIOContainer *container = container_of(bcontainer, VFIOContainer,
>> + bcontainer);
>> struct vfio_iommu_type1_dirty_bitmap *dbitmap;
>> struct vfio_iommu_type1_dirty_bitmap_get *range;
>> int ret;
>> @@ -461,7 +467,7 @@ static void
>vfio_get_iommu_info_migration(VFIOContainer *container,
>> * qemu_real_host_page_size to mark those dirty.
>> */
>> if (cap_mig->pgsize_bitmap & qemu_real_host_page_size()) {
>> - container->dirty_pages_supported = true;
>> + container->bcontainer.dirty_pages_supported = true;
>> container->max_dirty_bitmap_size = cap_mig->max_dirty_bitmap_size;
>> container->dirty_pgsizes = cap_mig->pgsize_bitmap;
>> }
>> @@ -553,7 +559,6 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> container = g_malloc0(sizeof(*container));
>> container->fd = fd;
>> container->error = NULL;
>> - container->dirty_pages_supported = false;
>> container->dma_max_mappings = 0;
>> container->iova_ranges = NULL;
>> QLIST_INIT(&container->vrdl_list);
>> @@ -938,4 +943,6 @@ void vfio_detach_device(VFIODevice *vbasedev)
>> const VFIOIOMMUOps vfio_legacy_ops = {
>> .dma_map = vfio_legacy_dma_map,
>> .dma_unmap = vfio_legacy_dma_unmap,
>> + .set_dirty_page_tracking = vfio_legacy_set_dirty_page_tracking,
>> + .query_dirty_bitmap = vfio_legacy_query_dirty_bitmap,
>> };
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base container
2023-10-27 15:52 ` Cédric Le Goater
@ 2023-10-30 3:14 ` Duan, Zhenzhong
2023-10-31 8:26 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 3:14 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
Hi Cédric,
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Friday, October 27, 2023 11:52 PM
>Subject: Re: [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and
>dma_max_mappings to base container
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
...
>> void vfio_container_destroy(VFIOContainerBase *bcontainer)
>> {
>> + VFIORamDiscardListener *vrdl, *vrdl_tmp;
>> VFIOGuestIOMMU *giommu, *tmp;
>>
>> QLIST_REMOVE(bcontainer, next);
>>
>> + QLIST_FOREACH_SAFE(vrdl, &bcontainer->vrdl_list, next, vrdl_tmp) {
>> + RamDiscardManager *rdm;
>> +
>> + rdm = memory_region_get_ram_discard_manager(vrdl->mr);
>> + ram_discard_manager_unregister_listener(rdm, &vrdl->listener);
>> + QLIST_REMOVE(vrdl, next);
>> + g_free(vrdl);
>> + }
>
>Where was this done previously ?
Good catch! This should be removed.
>May be the vrdl list should be handled
>separatly from pgsizes and dma_max_mappings.
Good suggestion! Will do.
>
>> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next,
>tmp) {
>> memory_region_unregister_iommu_notifier(
>> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>> index 8d5b408e86..0e265ffa67 100644
>> --- a/hw/vfio/container.c
>> +++ b/hw/vfio/container.c
>> @@ -154,7 +154,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase
>*bcontainer, hwaddr iova,
>> if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
>> container->iommu_type == VFIO_TYPE1v2_IOMMU) {
>> trace_vfio_legacy_dma_unmap_overflow_workaround();
>> - unmap.size -= 1ULL << ctz64(container->pgsizes);
>> + unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
>> continue;
>> }
>> error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
>> @@ -559,9 +559,7 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> container = g_malloc0(sizeof(*container));
>> container->fd = fd;
>> container->error = NULL;
>> - container->dma_max_mappings = 0;
>> container->iova_ranges = NULL;
>> - QLIST_INIT(&container->vrdl_list);
>> bcontainer = &container->bcontainer;
>> vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>>
>> @@ -589,13 +587,13 @@ static int vfio_connect_container(VFIOGroup *group,
>AddressSpace *as,
>> }
>>
>> if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
>> - container->pgsizes = info->iova_pgsizes;
>> + container->bcontainer.pgsizes = info->iova_pgsizes;
>> } else {
>> - container->pgsizes = qemu_real_host_page_size();
>> + container->bcontainer.pgsizes = qemu_real_host_page_size();
>> }
>>
>> - if (!vfio_get_info_dma_avail(info, &container->dma_max_mappings)) {
>> - container->dma_max_mappings = 65535;
>> + if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
>> + container->bcontainer.dma_max_mappings = 65535;
>> }
>>
>> vfio_get_info_iova_range(info, container);
>> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
>> index 3495737ab2..dbc4c24052 100644
>> --- a/hw/vfio/spapr.c
>> +++ b/hw/vfio/spapr.c
>> @@ -223,13 +223,13 @@ static int vfio_spapr_create_window(VFIOContainer
>*container,
>> if (pagesize > rampagesize) {
>> pagesize = rampagesize;
>> }
>> - pgmask = container->pgsizes & (pagesize | (pagesize - 1));
>> + pgmask = container->bcontainer.pgsizes & (pagesize | (pagesize - 1));
>> pagesize = pgmask ? (1ULL << (63 - clz64(pgmask))) : 0;
>> if (!pagesize) {
>> error_report("Host doesn't support page size 0x%"PRIx64
>> ", the supported mask is 0x%lx",
>> memory_region_iommu_get_min_page_size(iommu_mr),
>> - container->pgsizes);
>> + container->bcontainer.pgsizes);
>> return -EINVAL;
>> }
>>
>> @@ -385,7 +385,7 @@ void
>vfio_container_del_section_window(VFIOContainer *container,
>>
>> bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
>> {
>> -
>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>> struct vfio_iommu_spapr_tce_info info;
>> bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
>> int ret, fd = container->fd;
>> @@ -424,7 +424,7 @@ bool vfio_spapr_container_init(VFIOContainer
>*container, Error **errp)
>> }
>>
>> if (v2) {
>> - container->pgsizes = info.ddw.pgsizes;
>> + bcontainer->pgsizes = info.ddw.pgsizes;
>> /*
>> * There is a default window in just created container.
>> * To make region_add/del simpler, we better remove this
>> @@ -439,7 +439,7 @@ bool vfio_spapr_container_init(VFIOContainer
>*container, Error **errp)
>> }
>> } else {
>> /* The default table uses 4K pages */
>> - container->pgsizes = 0x1000;
>> + bcontainer->pgsizes = 0x1000;
>> vfio_host_win_add(container, info.dma32_window_start,
>> info.dma32_window_start +
>> info.dma32_window_size - 1,
>> @@ -455,7 +455,15 @@ listener_unregister_exit:
>>
>> void vfio_spapr_container_deinit(VFIOContainer *container)
>> {
>> + VFIOHostDMAWindow *hostwin, *next;
>> +
>> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
>> memory_listener_unregister(&container->prereg_listener);
>> }
>> + QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
>> + next) {
>> + QLIST_REMOVE(hostwin, hostwin_next);
>> + g_free(hostwin);
>> + }
>> +
>> }
>
>I am sure this change belongs to this patch.
Good catch! Yes, I should move it into below patch.
"vfio/common: Move vfio_host_win_add/del into spapr.c"
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface
2023-10-27 16:04 ` Cédric Le Goater
@ 2023-10-30 3:15 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 3:15 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Saturday, October 28, 2023 12:05 AM
>Subject: Re: [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target
>interface
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> Introduce an empry spapr backend which will hold spapr specific
>> content, currently only prereg_listener and hostwin_list.
>>
>> Also introduce and instantiate a spapr specific target interface,
>> currently only has add/del_window callbacks.
>>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>> include/hw/vfio/vfio-common.h | 8 ++++++++
>> include/hw/vfio/vfio-container-base.h | 2 ++
>> hw/vfio/spapr.c | 8 ++++++++
>> 3 files changed, 18 insertions(+)
>>
>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>> index 938f75e70c..a74e60e677 100644
>> --- a/include/hw/vfio/vfio-common.h
>> +++ b/include/hw/vfio/vfio-common.h
>> @@ -169,6 +169,14 @@ VFIOAddressSpace
>*vfio_get_address_space(AddressSpace *as);
>> void vfio_put_address_space(VFIOAddressSpace *space);
>>
>> /* SPAPR specific */
>> +typedef struct VFIOIOMMUSpaprOps {
>> + int (*add_window)(VFIOContainerBase *bcontainer,
>> + MemoryRegionSection *section,
>> + Error **errp);
>> + void (*del_window)(VFIOContainerBase *bcontainer,
>> + MemoryRegionSection *section);
>> +} VFIOIOMMUSpaprOps;
>
>I don't think we need a new structure type.
>
>If possible, these callbacks should be under VFIOIOMMUOps and only
>defined in the VFIOIOMMUOps struct for spapr. This struct would a
>copy of vfio_legacy_ops plus the two add/del_window handlers.
Understood, will update.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 18/37] vfio/container: Implement attach/detach_device
2023-10-27 16:06 ` Cédric Le Goater
@ 2023-10-30 3:20 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-30 3:20 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Saturday, October 28, 2023 12:07 AM
>Subject: Re: [PATCH v3 18/37] vfio/container: Implement attach/detach_device
>
>Sorry, previous email was empty. Friday effect !
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Eric Auger <eric.auger@redhat.com>
>>
>> No fucntional change intended.
>>
>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>> ---
>> hw/vfio/common.c | 16 ++++++++++++++++
>> hw/vfio/container.c | 12 +++++-------
>> 2 files changed, 21 insertions(+), 7 deletions(-)
>>
>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>> index d62c815d7f..64565b4ae9 100644
>> --- a/hw/vfio/common.c
>> +++ b/hw/vfio/common.c
>> @@ -1500,3 +1500,19 @@ retry:
>>
>> return info;
>> }
>> +
>> +int vfio_attach_device(char *name, VFIODevice *vbasedev,
>> + AddressSpace *as, Error **errp)
>> +{
>> + const VFIOIOMMUOps *ops = &vfio_legacy_ops;
>
>hmm, this looks wrong. please explain.
The final shape will be:
int vfio_attach_device(char *name, VFIODevice *vbasedev,
AddressSpace *as, Error **errp)
{
const VFIOIOMMUOps *ops;
#ifdef CONFIG_IOMMUFD
if (vbasedev->iommufd) {
ops = &vfio_iommufd_ops;
} else
#endif
{
ops = &vfio_legacy_ops;
}
return ops->attach_device(name, vbasedev, as, errp);
}
Depending on if iommufd is selected, different ops will be chosen.
Then corresponding attach_device callback is called to attach to
iommufd or legacy container.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation
2023-10-26 10:30 ` [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation Zhenzhong Duan
@ 2023-10-30 13:52 ` Cédric Le Goater
2023-10-31 2:02 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-30 13:52 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> IOMMUFD supports auto allocated hwpt and manually allocated one.
> Manually allocated hwpt has benefit that its life cycle is under
> user's control, so it could be used as stage 2 page table by nested
> feature in the future.
Would an option be useful to switch from one mode to another ?
>
> Introduce two helpers __vfio_device_attach/detach_hwpt to facilitate
> this change.
I think merging this change with the previous patch makes sense.
It doesn't add much to keep it as a standalone patch unless we
want a feature toggle.
Thanks,
C.
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> hw/vfio/iommufd.c | 89 +++++++++++++++++++++++++++++++++++++----------
> 1 file changed, 70 insertions(+), 19 deletions(-)
>
> diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
> index aee64d63f3..c1daaf1c39 100644
> --- a/hw/vfio/iommufd.c
> +++ b/hw/vfio/iommufd.c
> @@ -217,38 +217,91 @@ static VFIOIOASHwpt *vfio_container_get_hwpt(VFIOIOMMUFDContainer *container,
> static void vfio_container_put_hwpt(IOMMUFDBackend *be, VFIOIOASHwpt *hwpt)
> {
> QLIST_REMOVE(hwpt, next);
> + iommufd_backend_free_id(be->fd, hwpt->hwpt_id);
> g_free(hwpt);
> }
>
> -static int vfio_device_attach_container(VFIODevice *vbasedev,
> - VFIOIOMMUFDContainer *container,
> - Error **errp)
> +static int __vfio_device_attach_hwpt(VFIODevice *vbasedev, uint32_t hwpt_id,
> + Error **errp)
> {
> - int ret, iommufd = vbasedev->iommufd->fd;
> - VFIOIOASHwpt *hwpt;
> struct vfio_device_attach_iommufd_pt attach_data = {
> .argsz = sizeof(attach_data),
> .flags = 0,
> - .pt_id = container->ioas_id,
> + .pt_id = hwpt_id,
> };
> + int ret;
>
> - /* Attach device to an ioas within iommufd */
> ret = ioctl(vbasedev->fd, VFIO_DEVICE_ATTACH_IOMMUFD_PT, &attach_data);
> if (ret) {
> error_setg_errno(errp, errno,
> - "[iommufd=%d] error attach %s (%d) to ioasid=%d",
> - container->be->fd, vbasedev->name, vbasedev->fd,
> - attach_data.pt_id);
> + "[iommufd=%d] error attach %s (%d) to hwpt_id=%d",
> + vbasedev->iommufd->fd, vbasedev->name, vbasedev->fd,
> + hwpt_id);
> + }
> + return ret;
> +}
> +
> +static int __vfio_device_detach_hwpt(VFIODevice *vbasedev, Error **errp)
> +{
> + struct vfio_device_detach_iommufd_pt detach_data = {
> + .argsz = sizeof(detach_data),
> + .flags = 0,
> + };
> + int ret;
> +
> + ret = ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data);
> + if (ret) {
> + error_setg_errno(errp, errno, "detach %s from ioas failed",
> + vbasedev->name);
> + }
> + return ret;
> +}
> +
> +static int vfio_device_attach_container(VFIODevice *vbasedev,
> + VFIOIOMMUFDContainer *container,
> + Error **errp)
> +{
> + int ret, iommufd = vbasedev->iommufd->fd;
> + VFIOIOASHwpt *hwpt;
> + uint32_t hwpt_id;
> + Error *err = NULL;
> +
> + /* try to attach to an existing hwpt in this container */
> + QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
> + ret = __vfio_device_attach_hwpt(vbasedev, hwpt->hwpt_id, &err);
> + if (ret) {
> + const char *msg = error_get_pretty(err);
> +
> + trace_vfio_iommufd_fail_attach_existing_hwpt(msg);
> + error_free(err);
> + err = NULL;
> + } else {
> + goto found_hwpt;
> + }
> + }
> +
> + ret = iommufd_backend_alloc_hwpt(iommufd, vbasedev->devid,
> + container->ioas_id, &hwpt_id);
> +
> + if (ret) {
> + error_setg_errno(errp, errno, "error alloc shadow hwpt");
> return ret;
> + }
>
> + /* Attach device to an hwpt within iommufd */
> + ret = __vfio_device_attach_hwpt(vbasedev, hwpt_id, errp);
> + if (ret) {
> + iommufd_backend_free_id(iommufd, hwpt_id);
> + return ret;
> }
> - hwpt = vfio_container_get_hwpt(container, attach_data.pt_id);
>
> + hwpt = vfio_container_get_hwpt(container, hwpt_id);
> +found_hwpt:
> QLIST_INSERT_HEAD(&hwpt->device_list, vbasedev, next);
> vbasedev->hwpt = hwpt;
>
> trace_vfio_iommufd_attach_device(iommufd, vbasedev->name, vbasedev->fd,
> - container->ioas_id, attach_data.pt_id);
> + container->ioas_id, hwpt->hwpt_id);
> return ret;
> }
>
> @@ -256,14 +309,12 @@ static void vfio_device_detach_container(VFIODevice *vbasedev,
> VFIOIOMMUFDContainer *container)
> {
> VFIOIOASHwpt *hwpt = vbasedev->hwpt;
> - struct vfio_device_attach_iommufd_pt detach_data = {
> - .argsz = sizeof(detach_data),
> - .flags = 0,
> - };
> + Error *err = NULL;
> + int ret;
>
> - if (ioctl(vbasedev->fd, VFIO_DEVICE_DETACH_IOMMUFD_PT, &detach_data)) {
> - error_report("detach %s from ioas id=%d failed: %m", vbasedev->name,
> - container->ioas_id);
> + ret = __vfio_device_detach_hwpt(vbasedev, &err);
> + if (ret) {
> + error_report_err(err);
> }
>
> QLIST_REMOVE(vbasedev, next);
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
2023-10-26 10:30 ` [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend Zhenzhong Duan
@ 2023-10-30 13:56 ` Cédric Le Goater
2023-10-31 2:26 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-30 13:56 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Nicholas Piggin, Daniel Henrique Barboza, Cédric Le Goater,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/26/23 12:30, Zhenzhong Duan wrote:
> IBM EEH is only supported by legacy backend currently, bypass it
> for IOMMUFD backend.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> hw/ppc/spapr_pci_vfio.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
> index d1d07bec46..a2518838a1 100644
> --- a/hw/ppc/spapr_pci_vfio.c
> +++ b/hw/ppc/spapr_pci_vfio.c
> @@ -93,10 +93,10 @@ static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
>
> bcontainer = QLIST_FIRST(&space->containers);
>
> - if (QLIST_NEXT(bcontainer, next)) {
> + if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != &vfio_legacy_ops) {
It's curious that a test on the VFIOIOMMUOps is needed so deep in
the software stack, and spapr should have its own VFIOIOMMUOps, which
de facto doesn't support iommufd.
Thanks,
C.
> /*
> * We don't yet have logic to synchronize EEH state across
> - * multiple containers
> + * multiple containers, iommufd isn't supported too.
> */
> bcontainer = NULL;
> goto out;
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info
2023-10-26 10:30 ` [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info Zhenzhong Duan
@ 2023-10-30 13:59 ` Cédric Le Goater
2023-10-31 2:30 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-30 13:59 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> This helper will be used by both legacy and iommufd backends.
>
> No functional changes intended.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
I think we can merge this change before hand.
Thanks,
C.
> ---
> hw/vfio/pci.c | 54 +++++++++++++++++++++++++++++++++++----------------
> 1 file changed, 37 insertions(+), 17 deletions(-)
>
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index 5cbc771e55..c17e1f4376 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -2459,22 +2459,13 @@ static bool vfio_pci_host_match(PCIHostDeviceAddress *addr, const char *name)
> return (strcmp(tmp, name) == 0);
> }
>
> -static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
> +static int vfio_pci_get_pci_hot_reset_info(VFIOPCIDevice *vdev,
> + struct vfio_pci_hot_reset_info **info_p)
> {
> - VFIOGroup *group;
> struct vfio_pci_hot_reset_info *info;
> - struct vfio_pci_dependent_device *devices;
> - struct vfio_pci_hot_reset *reset;
> - int32_t *fds;
> - int ret, i, count;
> - bool multi = false;
> + int ret, count;
>
> - trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
> -
> - if (!single) {
> - vfio_pci_pre_reset(vdev);
> - }
> - vdev->vbasedev.needs_reset = false;
> + assert(info_p && !*info_p);
>
> info = g_malloc0(sizeof(*info));
> info->argsz = sizeof(*info);
> @@ -2482,24 +2473,53 @@ static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
> ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
> if (ret && errno != ENOSPC) {
> ret = -errno;
> + g_free(info);
> if (!vdev->has_pm_reset) {
> error_report("vfio: Cannot reset device %s, "
> "no available reset mechanism.", vdev->vbasedev.name);
> }
> - goto out_single;
> + return ret;
> }
>
> count = info->count;
> - info = g_realloc(info, sizeof(*info) + (count * sizeof(*devices)));
> - info->argsz = sizeof(*info) + (count * sizeof(*devices));
> - devices = &info->devices[0];
> + info = g_realloc(info, sizeof(*info) + (count * sizeof(info->devices[0])));
> + info->argsz = sizeof(*info) + (count * sizeof(info->devices[0]));
>
> ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_GET_PCI_HOT_RESET_INFO, info);
> if (ret) {
> ret = -errno;
> + g_free(info);
> error_report("vfio: hot reset info failed: %m");
> + return ret;
> + }
> +
> + *info_p = info;
> + return 0;
> +}
> +
> +static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
> +{
> + VFIOGroup *group;
> + struct vfio_pci_hot_reset_info *info = NULL;
> + struct vfio_pci_dependent_device *devices;
> + struct vfio_pci_hot_reset *reset;
> + int32_t *fds;
> + int ret, i, count;
> + bool multi = false;
> +
> + trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
> +
> + if (!single) {
> + vfio_pci_pre_reset(vdev);
> + }
> + vdev->vbasedev.needs_reset = false;
> +
> + ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
> +
> + if (ret) {
> goto out_single;
> }
> + devices = &info->devices[0];
>
> trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE
2023-10-26 10:30 ` [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
@ 2023-10-30 14:04 ` Cédric Le Goater
2023-10-31 2:27 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-30 14:04 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> As pci hot reset path need to reference pci specific functions
> and data structures, adding container level callback functions
> for legacy and iommufd BE and referencing those pci specific
> func/data is no better than implementing reset support with
> iommufd BE directly in pci.c
yes but it includes a large section of IOMMUFD code in pci.c
which is ugly. Please make this an VFIOIOMMUOps handler instead.
Thanks,
C.
>
> This way we can also share the common bus reset and system reset
> path for both BEs.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> ---
> hw/vfio/pci.c | 156 ++++++++++++++++++++++++++++++++++++++++++-
> hw/vfio/trace-events | 1 +
> 2 files changed, 156 insertions(+), 1 deletion(-)
>
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index c17e1f4376..d7a41c8def 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -42,6 +42,7 @@
> #include "qapi/error.h"
> #include "migration/blocker.h"
> #include "migration/qemu-file.h"
> +#include "linux/iommufd.h"
>
> #define TYPE_VFIO_PCI_NOHOTPLUG "vfio-pci-nohotplug"
>
> @@ -2497,7 +2498,7 @@ static int vfio_pci_get_pci_hot_reset_info(VFIOPCIDevice *vdev,
> return 0;
> }
>
> -static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
> +static int vfio_pci_hot_reset_legacy(VFIOPCIDevice *vdev, bool single)
> {
> VFIOGroup *group;
> struct vfio_pci_hot_reset_info *info = NULL;
> @@ -2661,6 +2662,159 @@ out_single:
> return ret;
> }
>
> +#ifdef CONFIG_IOMMUFD
> +static VFIODevice *vfio_pci_find_by_iommufd_devid(__u32 devid)
> +{
> + VFIODevice *vbasedev_iter;
> +
> + QLIST_FOREACH(vbasedev_iter, &vfio_device_list, global_next) {
> + if (vbasedev_iter->bcontainer->ops != &vfio_iommufd_ops) {
> + continue;
> + }
> + if (devid == vbasedev_iter->devid) {
> + return vbasedev_iter;
> + }
> + }
> + return NULL;
> +}
> +
> +static int vfio_pci_hot_reset_iommufd(VFIOPCIDevice *vdev, bool single)
> +{
> + struct vfio_pci_hot_reset_info *info = NULL;
> + struct vfio_pci_dependent_device *devices;
> + struct vfio_pci_hot_reset *reset;
> + int ret, i;
> + bool multi = false;
> +
> + trace_vfio_pci_hot_reset(vdev->vbasedev.name, single ? "one" : "multi");
> +
> + if (!single) {
> + vfio_pci_pre_reset(vdev);
> + }
> + vdev->vbasedev.needs_reset = false;
> +
> + ret = vfio_pci_get_pci_hot_reset_info(vdev, &info);
> +
> + if (ret) {
> + goto out_single;
> + }
> +
> + assert(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID);
> +
> + devices = &info->devices[0];
> +
> + if (!(info->flags & VFIO_PCI_HOT_RESET_FLAG_DEV_ID_OWNED)) {
> + if (!vdev->has_pm_reset) {
> + for (i = 0; i < info->count; i++) {
> + if (devices[i].devid == VFIO_PCI_DEVID_NOT_OWNED) {
> + error_report("vfio: Cannot reset device %s, "
> + "depends on device %04x:%02x:%02x.%x "
> + "which is not owned.",
> + vdev->vbasedev.name, devices[i].segment,
> + devices[i].bus, PCI_SLOT(devices[i].devfn),
> + PCI_FUNC(devices[i].devfn));
> + }
> + }
> + }
> + ret = -EPERM;
> + goto out_single;
> + }
> +
> + trace_vfio_pci_hot_reset_has_dep_devices(vdev->vbasedev.name);
> +
> + for (i = 0; i < info->count; i++) {
> + VFIOPCIDevice *tmp;
> + VFIODevice *vbasedev_iter;
> +
> + trace_vfio_pci_hot_reset_dep_devices_iommufd(devices[i].segment,
> + devices[i].bus,
> + PCI_SLOT(devices[i].devfn),
> + PCI_FUNC(devices[i].devfn),
> + devices[i].devid);
> +
> + /*
> + * If a VFIO cdev device is resettable, all the dependent devices
> + * are either bound to same iommufd or within same iommu_groups as
> + * one of the iommufd bound devices.
> + */
> + assert(devices[i].devid != VFIO_PCI_DEVID_NOT_OWNED);
> +
> + if (devices[i].devid == vdev->vbasedev.devid ||
> + devices[i].devid == VFIO_PCI_DEVID_OWNED) {
> + continue;
> + }
> +
> + vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
> + if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
> + vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
> + continue;
> + }
> + tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
> + if (single) {
> + ret = -EINVAL;
> + goto out_single;
> + }
> + vfio_pci_pre_reset(tmp);
> + tmp->vbasedev.needs_reset = false;
> + multi = true;
> + }
> +
> + if (!single && !multi) {
> + ret = -EINVAL;
> + goto out_single;
> + }
> +
> + /* Use zero length array for hot reset with iommufd backend */
> + reset = g_malloc0(sizeof(*reset));
> + reset->argsz = sizeof(*reset);
> +
> + /* Bus reset! */
> + ret = ioctl(vdev->vbasedev.fd, VFIO_DEVICE_PCI_HOT_RESET, reset);
> + g_free(reset);
> +
> + trace_vfio_pci_hot_reset_result(vdev->vbasedev.name,
> + ret ? strerror(errno) : "Success");
> +
> + /* Re-enable INTx on affected devices */
> + for (i = 0; i < info->count; i++) {
> + VFIOPCIDevice *tmp;
> + VFIODevice *vbasedev_iter;
> +
> + if (devices[i].devid == vdev->vbasedev.devid ||
> + devices[i].devid == VFIO_PCI_DEVID_OWNED) {
> + continue;
> + }
> +
> + vbasedev_iter = vfio_pci_find_by_iommufd_devid(devices[i].devid);
> + if (!vbasedev_iter || !vbasedev_iter->dev->realized ||
> + vbasedev_iter->type != VFIO_DEVICE_TYPE_PCI) {
> + continue;
> + }
> + tmp = container_of(vbasedev_iter, VFIOPCIDevice, vbasedev);
> + vfio_pci_post_reset(tmp);
> + }
> +out_single:
> + if (!single) {
> + vfio_pci_post_reset(vdev);
> + }
> + g_free(info);
> +
> + return ret;
> +}
> +#endif
> +
> +static int vfio_pci_hot_reset(VFIOPCIDevice *vdev, bool single)
> +{
> +#ifdef CONFIG_IOMMUFD
> + if (vdev->vbasedev.iommufd) {
> + return vfio_pci_hot_reset_iommufd(vdev, single);
> + } else
> +#endif
> + {
> + return vfio_pci_hot_reset_legacy(vdev, single);
> + }
> +}> +
> /*
> * We want to differentiate hot reset of multiple in-use devices vs hot reset
> * of a single in-use device. VFIO_DEVICE_RESET will already handle the case
> diff --git a/hw/vfio/trace-events b/hw/vfio/trace-events
> index 9b180cf77c..71c5840636 100644
> --- a/hw/vfio/trace-events
> +++ b/hw/vfio/trace-events
> @@ -34,6 +34,7 @@ vfio_check_af_flr(const char *name) "%s Supports FLR via AF cap"
> vfio_pci_hot_reset(const char *name, const char *type) " (%s) %s"
> vfio_pci_hot_reset_has_dep_devices(const char *name) "%s: hot reset dependent devices:"
> vfio_pci_hot_reset_dep_devices(int domain, int bus, int slot, int function, int group_id) "\t%04x:%02x:%02x.%x group %d"
> +vfio_pci_hot_reset_dep_devices_iommufd(int domain, int bus, int slot, int function, int dev_id) "\t%04x:%02x:%02x.%x devid %d"
> vfio_pci_hot_reset_result(const char *name, const char *result) "%s hot reset: %s"
> vfio_populate_device_config(const char *name, unsigned long size, unsigned long offset, unsigned long flags) "Device %s config:\n size: 0x%lx, offset: 0x%lx, flags: 0x%lx"
> vfio_populate_device_get_irq_info_failure(const char *errstr) "VFIO_DEVICE_GET_IRQ_INFO failure: %s"
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 25/37] util/char_dev: Add open_cdev()
2023-10-26 10:30 ` [PATCH v3 25/37] util/char_dev: Add open_cdev() Zhenzhong Duan
@ 2023-10-30 14:53 ` Cédric Le Goater
2023-10-31 1:59 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-30 14:53 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng
On 10/26/23 12:30, Zhenzhong Duan wrote:
> From: Yi Liu <yi.l.liu@intel.com>
>
> /dev/vfio/devices/vfioX may not exist. In that case it is still possible
> to open /dev/char/$major:$minor instead. Add helper function to abstract
> the cdev open.
>
> Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> MAINTAINERS | 6 +++
> include/qemu/chardev_open.h | 16 ++++++++
> util/chardev_open.c | 81 +++++++++++++++++++++++++++++++++++++
> util/meson.build | 1 +
> 4 files changed, 104 insertions(+)
> create mode 100644 include/qemu/chardev_open.h
> create mode 100644 util/chardev_open.c
>
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 7aa57ab16f..123e999bee 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -3447,6 +3447,12 @@ S: Maintained
> F: include/qemu/iova-tree.h
> F: util/iova-tree.c
>
> +cdev Open
> +M: Yi Liu <yi.l.liu@intel.com>
> +S: Maintained
> +F: include/qemu/chardev_open.h
> +F: util/chardev_open.c
> +
> elf2dmp
> M: Viktor Prutyanov <viktor.prutyanov@phystech.edu>
> S: Maintained
> diff --git a/include/qemu/chardev_open.h b/include/qemu/chardev_open.h
> new file mode 100644
> index 0000000000..6580d351c6
> --- /dev/null
> +++ b/include/qemu/chardev_open.h
> @@ -0,0 +1,16 @@
> +/*
> + * QEMU Chardev Helper
> + *
> + * Copyright (C) 2023 Intel Corporation.
> + *
> + * Authors: Yi Liu <yi.l.liu@intel.com>
> + *
> + * This work is licensed under the terms of the GNU GPL, version 2. See
> + * the COPYING file in the top-level directory.
> + */
> +
> +#ifndef QEMU_CHARDEV_HELPERS_H
> +#define QEMU_CHARDEV_HELPERS_H
> +
> +int open_cdev(const char *devpath, dev_t cdev);
> +#endif
> diff --git a/util/chardev_open.c b/util/chardev_open.c
> new file mode 100644
> index 0000000000..005d2b81bd
> --- /dev/null
> +++ b/util/chardev_open.c
> @@ -0,0 +1,81 @@
> +/*
> + * Copyright (c) 2019, Mellanox Technologies. All rights reserved.
> + * Copyright (C) 2023 Intel Corporation.
> + *
> + * This software is available to you under a choice of one of two
> + * licenses. You may choose to be licensed under the terms of the GNU
> + * General Public License (GPL) Version 2, available from the file
> + * COPYING in the main directory of this source tree, or the
> + * OpenIB.org BSD license below:
> + *
> + * Redistribution and use in source and binary forms, with or
> + * without modification, are permitted provided that the following
> + * conditions are met:
> + *
> + * - Redistributions of source code must retain the above
> + * copyright notice, this list of conditions and the following
> + * disclaimer.
> + *
> + * - Redistributions in binary form must reproduce the above
> + * copyright notice, this list of conditions and the following
> + * disclaimer in the documentation and/or other materials
> + * provided with the distribution.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
> + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
> + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
> + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
> + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
> + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
> + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
> + * SOFTWARE.
> + *
> + * Authors: Yi Liu <yi.l.liu@intel.com>
> + *
> + * Copied from
> + * https://github.com/linux-rdma/rdma-core/blob/master/util/open_cdev.c
> + *
> + */
> +
> +#include "qemu/osdep.h"
> +#include "qemu/chardev_open.h"
> +
> +static int open_cdev_internal(const char *path, dev_t cdev)
> +{
> + struct stat st;
> + int fd;
> +
> + fd = qemu_open_old(path, O_RDWR);
> + if (fd == -1) {
> + return -1;
> + }
> + if (fstat(fd, &st) || !S_ISCHR(st.st_mode) ||
> + (cdev != 0 && st.st_rdev != cdev)) {
> + close(fd);
> + return -1;
> + }
> + return fd;
> +}
> +
> +static int open_cdev_robust(dev_t cdev)
> +{
> + g_autofree char *devpath;
Please initialize g_autofree variables ( = NULL)
Thanks,
C.
> +
> + /*
> + * This assumes that udev is being used and is creating the /dev/char/
> + * symlinks.
> + */
> + devpath = g_strdup_printf("/dev/char/%u:%u", major(cdev), minor(cdev));
> + return open_cdev_internal(devpath, cdev);
> +}
> +
> +int open_cdev(const char *devpath, dev_t cdev)
> +{
> + int fd;
> +
> + fd = open_cdev_internal(devpath, cdev);
> + if (fd == -1 && cdev != 0) {
> + return open_cdev_robust(cdev);
> + }
> + return fd;
> +}
> diff --git a/util/meson.build b/util/meson.build
> index eb677b40c2..eda0b06062 100644
> --- a/util/meson.build
> +++ b/util/meson.build
> @@ -107,6 +107,7 @@ if have_block
> util_ss.add(files('filemonitor-stub.c'))
> endif
> util_ss.add(when: 'CONFIG_LINUX', if_true: files('vfio-helpers.c'))
> + util_ss.add(when: 'CONFIG_LINUX', if_true: files('chardev_open.c'))
> endif
>
> if cpu == 'aarch64'
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 25/37] util/char_dev: Add open_cdev()
2023-10-30 14:53 ` Cédric Le Goater
@ 2023-10-31 1:59 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 1:59 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 30, 2023 10:54 PM
>Subject: Re: [PATCH v3 25/37] util/char_dev: Add open_cdev()
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> From: Yi Liu <yi.l.liu@intel.com>
>>
>> /dev/vfio/devices/vfioX may not exist. In that case it is still possible
>> to open /dev/char/$major:$minor instead. Add helper function to abstract
>> the cdev open.
>>
>> Suggested-by: Jason Gunthorpe <jgg@nvidia.com>
>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>> ---
>> MAINTAINERS | 6 +++
>> include/qemu/chardev_open.h | 16 ++++++++
>> util/chardev_open.c | 81 +++++++++++++++++++++++++++++++++++++
>> util/meson.build | 1 +
>> 4 files changed, 104 insertions(+)
>> create mode 100644 include/qemu/chardev_open.h
>> create mode 100644 util/chardev_open.c
>>
>> diff --git a/MAINTAINERS b/MAINTAINERS
>> index 7aa57ab16f..123e999bee 100644
>> --- a/MAINTAINERS
>> +++ b/MAINTAINERS
>> @@ -3447,6 +3447,12 @@ S: Maintained
>> F: include/qemu/iova-tree.h
>> F: util/iova-tree.c
>>
>> +cdev Open
>> +M: Yi Liu <yi.l.liu@intel.com>
>> +S: Maintained
>> +F: include/qemu/chardev_open.h
>> +F: util/chardev_open.c
>> +
>> elf2dmp
>> M: Viktor Prutyanov <viktor.prutyanov@phystech.edu>
>> S: Maintained
>> diff --git a/include/qemu/chardev_open.h b/include/qemu/chardev_open.h
>> new file mode 100644
>> index 0000000000..6580d351c6
>> --- /dev/null
>> +++ b/include/qemu/chardev_open.h
>> @@ -0,0 +1,16 @@
>> +/*
>> + * QEMU Chardev Helper
>> + *
>> + * Copyright (C) 2023 Intel Corporation.
>> + *
>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>> + *
>> + * This work is licensed under the terms of the GNU GPL, version 2. See
>> + * the COPYING file in the top-level directory.
>> + */
>> +
>> +#ifndef QEMU_CHARDEV_HELPERS_H
>> +#define QEMU_CHARDEV_HELPERS_H
>> +
>> +int open_cdev(const char *devpath, dev_t cdev);
>> +#endif
>> diff --git a/util/chardev_open.c b/util/chardev_open.c
>> new file mode 100644
>> index 0000000000..005d2b81bd
>> --- /dev/null
>> +++ b/util/chardev_open.c
>> @@ -0,0 +1,81 @@
>> +/*
>> + * Copyright (c) 2019, Mellanox Technologies. All rights reserved.
>> + * Copyright (C) 2023 Intel Corporation.
>> + *
>> + * This software is available to you under a choice of one of two
>> + * licenses. You may choose to be licensed under the terms of the GNU
>> + * General Public License (GPL) Version 2, available from the file
>> + * COPYING in the main directory of this source tree, or the
>> + * OpenIB.org BSD license below:
>> + *
>> + * Redistribution and use in source and binary forms, with or
>> + * without modification, are permitted provided that the following
>> + * conditions are met:
>> + *
>> + * - Redistributions of source code must retain the above
>> + * copyright notice, this list of conditions and the following
>> + * disclaimer.
>> + *
>> + * - Redistributions in binary form must reproduce the above
>> + * copyright notice, this list of conditions and the following
>> + * disclaimer in the documentation and/or other materials
>> + * provided with the distribution.
>> + *
>> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
>> + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
>OF
>> + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
>> + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
>HOLDERS
>> + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
>AN
>> + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR
>IN
>> + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
>THE
>> + * SOFTWARE.
>> + *
>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>> + *
>> + * Copied from
>> + * https://github.com/linux-rdma/rdma-core/blob/master/util/open_cdev.c
>> + *
>> + */
>> +
>> +#include "qemu/osdep.h"
>> +#include "qemu/chardev_open.h"
>> +
>> +static int open_cdev_internal(const char *path, dev_t cdev)
>> +{
>> + struct stat st;
>> + int fd;
>> +
>> + fd = qemu_open_old(path, O_RDWR);
>> + if (fd == -1) {
>> + return -1;
>> + }
>> + if (fstat(fd, &st) || !S_ISCHR(st.st_mode) ||
>> + (cdev != 0 && st.st_rdev != cdev)) {
>> + close(fd);
>> + return -1;
>> + }
>> + return fd;
>> +}
>> +
>> +static int open_cdev_robust(dev_t cdev)
>> +{
>> + g_autofree char *devpath;
>
>Please initialize g_autofree variables ( = NULL)
Thanks for point out, will fix.
BRs.
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation
2023-10-30 13:52 ` Cédric Le Goater
@ 2023-10-31 2:02 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 2:02 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 30, 2023 9:52 PM
>Subject: Re: [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> IOMMUFD supports auto allocated hwpt and manually allocated one.
>> Manually allocated hwpt has benefit that its life cycle is under
>> user's control, so it could be used as stage 2 page table by nested
>> feature in the future.
>
>Would an option be useful to switch from one mode to another ?
Looks unnecessary for me as we prefer manual allocation.
The purpose of splitting is to make review easier.
>
>>
>> Introduce two helpers __vfio_device_attach/detach_hwpt to facilitate
>> this change.
>
>I think merging this change with the previous patch makes sense.
>It doesn't add much to keep it as a standalone patch unless we
>want a feature toggle.
OK, will merge them in v4.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
2023-10-30 13:56 ` Cédric Le Goater
@ 2023-10-31 2:26 ` Duan, Zhenzhong
2023-10-31 9:01 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 2:26 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 30, 2023 9:57 PM
>Subject: Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> IBM EEH is only supported by legacy backend currently, bypass it
>> for IOMMUFD backend.
>>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>> ---
>> hw/ppc/spapr_pci_vfio.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
>> index d1d07bec46..a2518838a1 100644
>> --- a/hw/ppc/spapr_pci_vfio.c
>> +++ b/hw/ppc/spapr_pci_vfio.c
>> @@ -93,10 +93,10 @@ static VFIOContainer
>*vfio_eeh_as_container(AddressSpace *as)
>>
>> bcontainer = QLIST_FIRST(&space->containers);
>>
>> - if (QLIST_NEXT(bcontainer, next)) {
>> + if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != &vfio_legacy_ops) {
>
>It's curious that a test on the VFIOIOMMUOps is needed so deep in
>the software stack, and spapr should have its own VFIOIOMMUOps, which
>de facto doesn't support iommufd.
Yes, in this series, spapr shares same ops vfio_legacy_ops, in next series I should
check with vfio_iommu_spapr_ops.
The general vfio-pci device supports iommu property, if we pass a vfio device
with iommufd backend, I guess we will crash Qemu if there is no check here.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE
2023-10-30 14:04 ` Cédric Le Goater
@ 2023-10-31 2:27 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 2:27 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 30, 2023 10:05 PM
>Subject: Re: [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with
>iommufd BE
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> As pci hot reset path need to reference pci specific functions
>> and data structures, adding container level callback functions
>> for legacy and iommufd BE and referencing those pci specific
>> func/data is no better than implementing reset support with
>> iommufd BE directly in pci.c
>
>yes but it includes a large section of IOMMUFD code in pci.c
>which is ugly. Please make this an VFIOIOMMUOps handler instead.
OK, let me try it.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info
2023-10-30 13:59 ` Cédric Le Goater
@ 2023-10-31 2:30 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 2:30 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Monday, October 30, 2023 10:00 PM
>Subject: Re: [PATCH v3 30/37] vfio/pci: Extract out a helper
>vfio_pci_get_pci_hot_reset_info
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> This helper will be used by both legacy and iommufd backends.
>>
>> No functional changes intended.
>>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>
>I think we can merge this change before hand.
Yes, based on your suggestion in patch31, I will move the helper func
into hw/vfio/helpers.c and then you can pick it in v4.
Thanks
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-30 2:40 ` Duan, Zhenzhong
@ 2023-10-31 7:57 ` Cédric Le Goater
2023-10-31 8:31 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 7:57 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
On 10/30/23 03:40, Duan, Zhenzhong wrote:
>
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Friday, October 27, 2023 10:03 PM
>> Subject: Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and
>> targetted interface
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>
>> targeted
>
> Will fix.
>
>>
>>
>>> This is willingly not a QOM object because we don't want it to be
>>> visible from the user interface. The VFIOContainer will be smoothly
>>> populated in subsequent patches as well as interfaces.
>>>
>>> No fucntional change intended.
>>>
>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> ---
>>> include/hw/vfio/vfio-common.h | 8 +---
>>> include/hw/vfio/vfio-container-base.h | 64 +++++++++++++++++++++++++++
>>> 2 files changed, 66 insertions(+), 6 deletions(-)
>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>
>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>>> index b9c7a7e588..d8f293cb57 100644
>>> --- a/include/hw/vfio/vfio-common.h
>>> +++ b/include/hw/vfio/vfio-common.h
>>> @@ -30,6 +30,7 @@
>>> #include <linux/vfio.h>
>>> #endif
>>> #include "sysemu/sysemu.h"
>>> +#include "hw/vfio/vfio-container-base.h"
>>>
>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>
>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>> struct VFIOGroup;
>>>
>>> typedef struct VFIOContainer {
>>> + VFIOContainerBase bcontainer;
>>> VFIOAddressSpace *space;
>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>> MemoryListener listener;
>>> @@ -201,12 +203,6 @@ typedef struct VFIODisplay {
>>> } dmabuf;
>>> } VFIODisplay;
>>>
>>> -typedef struct {
>>> - unsigned long *bitmap;
>>> - hwaddr size;
>>> - hwaddr pages;
>>> -} VFIOBitmap;
>>> -
>>> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
>>> void vfio_put_address_space(VFIOAddressSpace *space);
>>> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>> container-base.h
>>> new file mode 100644
>>> index 0000000000..5becbd51a7
>>> --- /dev/null
>>> +++ b/include/hw/vfio/vfio-container-base.h
>>> @@ -0,0 +1,64 @@
>>> +/*
>>> + * VFIO BASE CONTAINER
>>> + *
>>> + * Copyright (C) 2023 Intel Corporation.
>>> + * Copyright Red Hat, Inc. 2023
>>> + *
>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>> + * Eric Auger <eric.auger@redhat.com>
>>> + *
>>> + * This program is free software; you can redistribute it and/or modify
>>> + * it under the terms of the GNU General Public License as published by
>>> + * the Free Software Foundation; either version 2 of the License, or
>>> + * (at your option) any later version.
>>> +
>>> + * This program is distributed in the hope that it will be useful,
>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>> + * GNU General Public License for more details.
>>> +
>>> + * You should have received a copy of the GNU General Public License along
>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>
>> This should be enough :
>>
>> SPDX-License-Identifier: GPL-2.0-or-later
>
> Will do.
>
>>
>>> + */
>>> +
>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>
>> HW_VFIO_VFIO_CONTAINER_BASE_H
>>
>>> +
>>> +#include "exec/memory.h"
>>> +#ifndef CONFIG_USER_ONLY
>>> +#include "exec/hwaddr.h"
>>> +#endif
>>
>> including "exec/memory.h" should be enough.
>
> Will do.
>
>>
>>
>>> +
>>> +typedef struct VFIODevice VFIODevice;
>>> +typedef struct VFIOIOMMUOps VFIOIOMMUOps;
>>> +
>>> +typedef struct {
>>> + unsigned long *bitmap;
>>> + hwaddr size;
>>> + hwaddr pages;
>>> +} VFIOBitmap;
>>> +
>>> +/*
>>> + * This is the base object for vfio container backends
>>> + */
>>> +typedef struct VFIOContainerBase {
>>> + const VFIOIOMMUOps *ops;
>>> +} VFIOContainerBase;
>>> +
>>> +struct VFIOIOMMUOps {
>>> + /* basic feature */
>>> + int (*dma_map)(VFIOContainerBase *bcontainer,
>>> + hwaddr iova, ram_addr_t size,
>>> + void *vaddr, bool readonly);
>>> + int (*dma_unmap)(VFIOContainerBase *bcontainer,
>>> + hwaddr iova, ram_addr_t size,
>>> + IOMMUTLBEntry *iotlb);
>>
>> Could the VFIOContainerBase *parameter be const ?
>
> Yes, VFIOContainerBase is not changed by dma_unmap and other
> functions dma_unmap calls. I tried and found making it const here
> would impact all functions it called with same parameter be const
> in following patches which looks unrelated to the patch itself
> to avoid compile error.
>
> E.g. below functions are impacted,
> vfio_devices_all_running_and_mig_active
> vfio_devices_all_device_dirty_tracking
> vfio_devices_query_dirty_bitmap
> vfio_container_query_dirty_bitmap
> vfio_legacy_query_dirty_bitmap>
> To make following patches cleaner to review, I would like to keep
> current code except you or others have a strong opinion.
> Another choice I can think of is to add const to all impact functions
> in a separate patch.
It would be good to have, for later.
On a related topic, I find the code difficult to read because it is
complex, of course, and dealing with a thick layer of software constructs
but also because it lacks a consistent naming scheme in the different
submodules. For instance, iommufd.c has various routine prefixes for
local routines. This is quite confusing. Same for :
hw/vfio/common.c
hw/vfio/helpers.c
hw/vfio/container.c
hw/vfio/container-base.c
This is not necessarily introduced by the changes of this series.
Something to improve for sure though.
Thanks,
C.
>
>>
>>> + int (*attach_device)(char *name, VFIODevice *vbasedev,
>>
>> cont char *name ?
>
> Yes, will fix.
>
> Thanks
> Zhenzhong
>
>>
>> The rest looks good to me.
>>
>> Thanks
>>
>> C.
>>
>>
>>> + AddressSpace *as, Error **errp);
>>> + void (*detach_device)(VFIODevice *vbasedev);
>>> + /* migration feature */
>>> + int (*set_dirty_page_tracking)(VFIOContainerBase *bcontainer, bool start);
>>> + int (*query_dirty_bitmap)(VFIOContainerBase *bcontainer, VFIOBitmap
>> *vbmap,
>>> + hwaddr iova, hwaddr size);
>>> +};
>>> +#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps
2023-10-30 2:43 ` Duan, Zhenzhong
@ 2023-10-31 8:21 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 8:21 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P
On 10/30/23 03:43, Duan, Zhenzhong wrote:
>
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Friday, October 27, 2023 10:21 PM
>> Subject: Re: [PATCH v3 07/37] vfio/container: Introduce a empty
>> VFIOIOMMUOps
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> This empty VFIOIOMMUOps named vfio_legacy_ops will hold all general
>>> IOMMU ops of legacy container.
>>>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> ---
>>> include/hw/vfio/vfio-common.h | 2 +-
>>> hw/vfio/container.c | 5 +++++
>>> 2 files changed, 6 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>>> index d8f293cb57..8ded5cd8e4 100644
>>> --- a/include/hw/vfio/vfio-common.h
>>> +++ b/include/hw/vfio/vfio-common.h
>>> @@ -255,7 +255,7 @@ typedef QLIST_HEAD(VFIOGroupList, VFIOGroup)
>> VFIOGroupList;
>>> typedef QLIST_HEAD(VFIODeviceList, VFIODevice) VFIODeviceList;
>>> extern VFIOGroupList vfio_group_list;
>>> extern VFIODeviceList vfio_device_list;
>>> -
>>> +extern const VFIOIOMMUOps vfio_legacy_ops;
>>
>>
>> why does it need to be external ?
>
> It is referenced by vfio_connect_container() and vfio_attach_device().
Yes. I realized that later on.
The backend is chosen when the device id attached :
int vfio_attach_device(char *name, VFIODevice *vbasedev,
AddressSpace *as, Error **errp)
{
const VFIOIOMMUOps *ops;
if (vbasedev->iommufd) {
ops = &vfio_iommufd_ops;
} else {
ops = &vfio_legacy_ops;
}
return ops->attach_device(name, vbasedev, as, errp);
}
To be noted that we don't need the backend ops but the attach_device()
handler only.
And then, the backend ops is assigned to the base container deeper
in the call stack with vfio_container_init(), which is a bit like a
chicken and egg problem to me.
vfio_legacy_ops.attach_device = vfio_legacy_attach_device()
vfio_get_group()
vfio_connect_container()
vfio_container_init(&vfio_legacy_ops)
vfio_iommufd_ops.attach_device = iommufd_attach_device()
vfio_container_init(&vfio_iommufd_ops)
vfio_legacy_attach_device() and iommufd_attach_device() are similar but
have different requirements. I don't see a good alternative. Unless we
introduce a QOM object wrapping the IOMMUFDBackend and the legacy one
to hold the VFIOIOMMUOps struct. Looks overkill.
Thanks,
C.
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 10/37] vfio/container: Move space field to base container
2023-10-30 2:51 ` Duan, Zhenzhong
@ 2023-10-31 8:24 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 8:24 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
On 10/30/23 03:51, Duan, Zhenzhong wrote:
>
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Friday, October 27, 2023 10:46 PM
>> Subject: Re: [PATCH v3 10/37] vfio/container: Move space field to base container
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> From: Eric Auger <eric.auger@redhat.com>
>>>
>>> Move the space field to the base object. Also the VFIOAddressSpace
>>> now contains a list of base containers.
>>>
>>> No fucntional change intended.
>>>
>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> [ clg: context changes ]
>>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>>> ---
>>> include/hw/vfio/vfio-common.h | 8 --------
>>> include/hw/vfio/vfio-container-base.h | 9 +++++++++
>>> hw/ppc/spapr_pci_vfio.c | 10 +++++-----
>>> hw/vfio/common.c | 4 ++--
>>> hw/vfio/container-base.c | 6 +++++-
>>> hw/vfio/container.c | 18 +++++++++---------
>>> 6 files changed, 30 insertions(+), 25 deletions(-)
>>>
>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-common.h
>>> index fcb4003a21..857d2b8076 100644
>>> --- a/include/hw/vfio/vfio-common.h
>>> +++ b/include/hw/vfio/vfio-common.h
>>> @@ -73,17 +73,10 @@ typedef struct VFIOMigration {
>>> bool initial_data_sent;
>>> } VFIOMigration;
>>>
>>> -typedef struct VFIOAddressSpace {
>>> - AddressSpace *as;
>>> - QLIST_HEAD(, VFIOContainer) containers;
>>> - QLIST_ENTRY(VFIOAddressSpace) list;
>>> -} VFIOAddressSpace;
>>> -
>>> struct VFIOGroup;
>>>
>>> typedef struct VFIOContainer {
>>> VFIOContainerBase bcontainer;
>>> - VFIOAddressSpace *space;
>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>> MemoryListener listener;
>>> MemoryListener prereg_listener;
>>> @@ -98,7 +91,6 @@ typedef struct VFIOContainer {
>>> QLIST_HEAD(, VFIOHostDMAWindow) hostwin_list;
>>> QLIST_HEAD(, VFIOGroup) group_list;
>>> QLIST_HEAD(, VFIORamDiscardListener) vrdl_list;
>>> - QLIST_ENTRY(VFIOContainer) next;
>>> QLIST_HEAD(, VFIODevice) device_list;
>>> GList *iova_ranges;
>>> } VFIOContainer;
>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>> container-base.h
>>> index 71e1e4324e..a5fef3e6a8 100644
>>> --- a/include/hw/vfio/vfio-container-base.h
>>> +++ b/include/hw/vfio/vfio-container-base.h
>>> @@ -38,12 +38,20 @@ typedef struct {
>>> hwaddr pages;
>>> } VFIOBitmap;
>>>
>>> +typedef struct VFIOAddressSpace {
>>> + AddressSpace *as;
>>> + QLIST_HEAD(, VFIOContainerBase) containers;
>>> + QLIST_ENTRY(VFIOAddressSpace) list;
>>> +} VFIOAddressSpace;
>>> +
>>> /*
>>> * This is the base object for vfio container backends
>>> */
>>> typedef struct VFIOContainerBase {
>>> const VFIOIOMMUOps *ops;
>>> + VFIOAddressSpace *space;
>>> QLIST_HEAD(, VFIOGuestIOMMU) giommu_list;
>>> + QLIST_ENTRY(VFIOContainerBase) next;
>>> } VFIOContainerBase;
>>>
>>> typedef struct VFIOGuestIOMMU {
>>> @@ -62,6 +70,7 @@ int vfio_container_dma_unmap(VFIOContainerBase
>> *bcontainer,
>>> IOMMUTLBEntry *iotlb);
>>>
>>> void vfio_container_init(VFIOContainerBase *bcontainer,
>>> + VFIOAddressSpace *space,
>>> const VFIOIOMMUOps *ops);
>>> void vfio_container_destroy(VFIOContainerBase *bcontainer);
>>>
>>> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
>>> index f283f7e38d..d1d07bec46 100644
>>> --- a/hw/ppc/spapr_pci_vfio.c
>>> +++ b/hw/ppc/spapr_pci_vfio.c
>>> @@ -84,27 +84,27 @@ static int vfio_eeh_container_op(VFIOContainer
>> *container, uint32_t op)
>>> static VFIOContainer *vfio_eeh_as_container(AddressSpace *as)
>>> {
>>> VFIOAddressSpace *space = vfio_get_address_space(as);
>>> - VFIOContainer *container = NULL;
>>> + VFIOContainerBase *bcontainer = NULL;
>>>
>>> if (QLIST_EMPTY(&space->containers)) {
>>> /* No containers to act on */
>>> goto out;
>>> }
>>>
>>> - container = QLIST_FIRST(&space->containers);
>>> + bcontainer = QLIST_FIRST(&space->containers);
>>>
>>> - if (QLIST_NEXT(container, next)) {
>>> + if (QLIST_NEXT(bcontainer, next)) {
>>> /*
>>> * We don't yet have logic to synchronize EEH state across
>>> * multiple containers
>>> */
>>> - container = NULL;
>>> + bcontainer = NULL;
>>> goto out;
>>> }
>>>
>>> out:
>>> vfio_put_address_space(space);
>>> - return container;
>>> + return container_of(bcontainer, VFIOContainer, bcontainer);
>>> }
>>>
>>> static bool vfio_eeh_as_ok(AddressSpace *as)
>>> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>>> index 4f130ad87c..f87a0dcec3 100644
>>> --- a/hw/vfio/common.c
>>> +++ b/hw/vfio/common.c
>>> @@ -145,7 +145,7 @@ void vfio_unblock_multiple_devices_migration(void)
>>>
>>> bool vfio_viommu_preset(VFIODevice *vbasedev)
>>> {
>>> - return vbasedev->container->space->as != &address_space_memory;
>>> + return vbasedev->container->bcontainer.space->as !=
>> &address_space_memory;
>>> }
>>>
>>> static void vfio_set_migration_error(int err)
>>> @@ -924,7 +924,7 @@ static void vfio_dirty_tracking_init(VFIOContainer
>> *container,
>>> dirty.container = container;
>>>
>>> memory_listener_register(&dirty.listener,
>>> - container->space->as);
>>> + container->bcontainer.space->as);
>>
>> Could we introduce an helper returning :
>>
>> container->bcontainer.space->as
>
> Guess you mean this is too long?
yes and I like helpers but it is a personal taste.
> Then I would like to use
> bcontainer->space->as except you have a strong opinion on that.
I don't mind.
Thanks,
C.
>
> Thanks
> Zhenzhong
>
>>
>> The rest looks OK.
>>
>> Thanks,
>>
>> C.
>>
>>>
>>> *ranges = dirty.ranges;
>>>
>>> diff --git a/hw/vfio/container-base.c b/hw/vfio/container-base.c
>>> index 6be7ce65ad..99b2957d7b 100644
>>> --- a/hw/vfio/container-base.c
>>> +++ b/hw/vfio/container-base.c
>>> @@ -48,9 +48,11 @@ int vfio_container_dma_unmap(VFIOContainerBase
>> *bcontainer,
>>> return bcontainer->ops->dma_unmap(bcontainer, iova, size, iotlb);
>>> }
>>>
>>> -void vfio_container_init(VFIOContainerBase *bcontainer, const
>> VFIOIOMMUOps *ops)
>>> +void vfio_container_init(VFIOContainerBase *bcontainer, VFIOAddressSpace
>> *space,
>>> + const VFIOIOMMUOps *ops)
>>> {
>>> bcontainer->ops = ops;
>>> + bcontainer->space = space;
>>> QLIST_INIT(&bcontainer->giommu_list);
>>> }
>>>
>>> @@ -58,6 +60,8 @@ void vfio_container_destroy(VFIOContainerBase
>> *bcontainer)
>>> {
>>> VFIOGuestIOMMU *giommu, *tmp;
>>>
>>> + QLIST_REMOVE(bcontainer, next);
>>> +
>>> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next,
>> tmp) {
>>> memory_region_unregister_iommu_notifier(
>>> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>>> index b2b6e05287..761310fa51 100644
>>> --- a/hw/vfio/container.c
>>> +++ b/hw/vfio/container.c
>>> @@ -514,7 +514,8 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> * details once we know which type of IOMMU we are using.
>>> */
>>>
>>> - QLIST_FOREACH(container, &space->containers, next) {
>>> + QLIST_FOREACH(bcontainer, &space->containers, next) {
>>> + container = container_of(bcontainer, VFIOContainer, bcontainer);
>>> if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
>>> ret = vfio_ram_block_discard_disable(container, true);
>>> if (ret) {
>>> @@ -550,7 +551,6 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> }
>>>
>>> container = g_malloc0(sizeof(*container));
>>> - container->space = space;
>>> container->fd = fd;
>>> container->error = NULL;
>>> container->dirty_pages_supported = false;
>>> @@ -558,7 +558,7 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> container->iova_ranges = NULL;
>>> QLIST_INIT(&container->vrdl_list);
>>> bcontainer = &container->bcontainer;
>>> - vfio_container_init(bcontainer, &vfio_legacy_ops);
>>> + vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>>>
>>> ret = vfio_init_container(container, group->fd, errp);
>>> if (ret) {
>>> @@ -613,14 +613,15 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> vfio_kvm_device_add_group(group);
>>>
>>> QLIST_INIT(&container->group_list);
>>> - QLIST_INSERT_HEAD(&space->containers, container, next);
>>> + QLIST_INSERT_HEAD(&space->containers, bcontainer, next);
>>>
>>> group->container = container;
>>> QLIST_INSERT_HEAD(&container->group_list, group, container_next);
>>>
>>> container->listener = vfio_memory_listener;
>>>
>>> - memory_listener_register(&container->listener, container->space->as);
>>> + memory_listener_register(&container->listener,
>>> + container->bcontainer.space->as);
>>>
>>> if (container->error) {
>>> ret = -1;
>>> @@ -634,7 +635,7 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> return 0;
>>> listener_release_exit:
>>> QLIST_REMOVE(group, container_next);
>>> - QLIST_REMOVE(container, next);
>>> + QLIST_REMOVE(bcontainer, next);
>>> vfio_kvm_device_del_group(group);
>>> memory_listener_unregister(&container->listener);
>>> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU ||
>>> @@ -684,9 +685,8 @@ static void vfio_disconnect_container(VFIOGroup
>> *group)
>>> }
>>>
>>> if (QLIST_EMPTY(&container->group_list)) {
>>> - VFIOAddressSpace *space = container->space;
>>> + VFIOAddressSpace *space = container->bcontainer.space;
>>>
>>> - QLIST_REMOVE(container, next);
>>> vfio_container_destroy(bcontainer);
>>>
>>> trace_vfio_disconnect_container(container->fd);
>>> @@ -706,7 +706,7 @@ static VFIOGroup *vfio_get_group(int groupid,
>> AddressSpace *as, Error **errp)
>>> QLIST_FOREACH(group, &vfio_group_list, next) {
>>> if (group->groupid == groupid) {
>>> /* Found it. Now is it already in the right context? */
>>> - if (group->container->space->as == as) {
>>> + if (group->container->bcontainer.space->as == as) {
>>> return group;
>>> } else {
>>> error_setg(errp, "group %d used in multiple address spaces",
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings to base container
2023-10-30 3:14 ` Duan, Zhenzhong
@ 2023-10-31 8:26 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 8:26 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun, Nicholas Piggin, Daniel Henrique Barboza,
David Gibson, Harsh Prateek Bora, open list:sPAPR (pseries)
On 10/30/23 04:14, Duan, Zhenzhong wrote:
> Hi Cédric,
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Friday, October 27, 2023 11:52 PM
>> Subject: Re: [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and
>> dma_max_mappings to base container
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> From: Eric Auger <eric.auger@redhat.com>
>>>
> ...
>
>>> void vfio_container_destroy(VFIOContainerBase *bcontainer)
>>> {
>>> + VFIORamDiscardListener *vrdl, *vrdl_tmp;
>>> VFIOGuestIOMMU *giommu, *tmp;
>>>
>>> QLIST_REMOVE(bcontainer, next);
>>>
>>> + QLIST_FOREACH_SAFE(vrdl, &bcontainer->vrdl_list, next, vrdl_tmp) {
>>> + RamDiscardManager *rdm;
>>> +
>>> + rdm = memory_region_get_ram_discard_manager(vrdl->mr);
>>> + ram_discard_manager_unregister_listener(rdm, &vrdl->listener);
>>> + QLIST_REMOVE(vrdl, next);
>>> + g_free(vrdl);
>>> + }
>>
>> Where was this done previously ?
>
> Good catch! This should be removed.
>
>> May be the vrdl list should be handled
>> separatly from pgsizes and dma_max_mappings.
>
> Good suggestion! Will do.
>
>>
>>> QLIST_FOREACH_SAFE(giommu, &bcontainer->giommu_list, giommu_next,
>> tmp) {
>>> memory_region_unregister_iommu_notifier(
>>> MEMORY_REGION(giommu->iommu_mr), &giommu->n);
>>> diff --git a/hw/vfio/container.c b/hw/vfio/container.c
>>> index 8d5b408e86..0e265ffa67 100644
>>> --- a/hw/vfio/container.c
>>> +++ b/hw/vfio/container.c
>>> @@ -154,7 +154,7 @@ static int vfio_legacy_dma_unmap(VFIOContainerBase
>> *bcontainer, hwaddr iova,
>>> if (errno == EINVAL && unmap.size && !(unmap.iova + unmap.size) &&
>>> container->iommu_type == VFIO_TYPE1v2_IOMMU) {
>>> trace_vfio_legacy_dma_unmap_overflow_workaround();
>>> - unmap.size -= 1ULL << ctz64(container->pgsizes);
>>> + unmap.size -= 1ULL << ctz64(container->bcontainer.pgsizes);
>>> continue;
>>> }
>>> error_report("VFIO_UNMAP_DMA failed: %s", strerror(errno));
>>> @@ -559,9 +559,7 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> container = g_malloc0(sizeof(*container));
>>> container->fd = fd;
>>> container->error = NULL;
>>> - container->dma_max_mappings = 0;
>>> container->iova_ranges = NULL;
>>> - QLIST_INIT(&container->vrdl_list);
>>> bcontainer = &container->bcontainer;
>>> vfio_container_init(bcontainer, space, &vfio_legacy_ops);
>>>
>>> @@ -589,13 +587,13 @@ static int vfio_connect_container(VFIOGroup *group,
>> AddressSpace *as,
>>> }
>>>
>>> if (info->flags & VFIO_IOMMU_INFO_PGSIZES) {
>>> - container->pgsizes = info->iova_pgsizes;
>>> + container->bcontainer.pgsizes = info->iova_pgsizes;
>>> } else {
>>> - container->pgsizes = qemu_real_host_page_size();
>>> + container->bcontainer.pgsizes = qemu_real_host_page_size();
>>> }
>>>
>>> - if (!vfio_get_info_dma_avail(info, &container->dma_max_mappings)) {
>>> - container->dma_max_mappings = 65535;
>>> + if (!vfio_get_info_dma_avail(info, &bcontainer->dma_max_mappings)) {
>>> + container->bcontainer.dma_max_mappings = 65535;
>>> }
>>>
>>> vfio_get_info_iova_range(info, container);
>>> diff --git a/hw/vfio/spapr.c b/hw/vfio/spapr.c
>>> index 3495737ab2..dbc4c24052 100644
>>> --- a/hw/vfio/spapr.c
>>> +++ b/hw/vfio/spapr.c
>>> @@ -223,13 +223,13 @@ static int vfio_spapr_create_window(VFIOContainer
>> *container,
>>> if (pagesize > rampagesize) {
>>> pagesize = rampagesize;
>>> }
>>> - pgmask = container->pgsizes & (pagesize | (pagesize - 1));
>>> + pgmask = container->bcontainer.pgsizes & (pagesize | (pagesize - 1));
>>> pagesize = pgmask ? (1ULL << (63 - clz64(pgmask))) : 0;
>>> if (!pagesize) {
>>> error_report("Host doesn't support page size 0x%"PRIx64
>>> ", the supported mask is 0x%lx",
>>> memory_region_iommu_get_min_page_size(iommu_mr),
>>> - container->pgsizes);
>>> + container->bcontainer.pgsizes);
>>> return -EINVAL;
>>> }
>>>
>>> @@ -385,7 +385,7 @@ void
>> vfio_container_del_section_window(VFIOContainer *container,
>>>
>>> bool vfio_spapr_container_init(VFIOContainer *container, Error **errp)
>>> {
>>> -
>>> + VFIOContainerBase *bcontainer = &container->bcontainer;
>>> struct vfio_iommu_spapr_tce_info info;
>>> bool v2 = container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU;
>>> int ret, fd = container->fd;
>>> @@ -424,7 +424,7 @@ bool vfio_spapr_container_init(VFIOContainer
>> *container, Error **errp)
>>> }
>>>
>>> if (v2) {
>>> - container->pgsizes = info.ddw.pgsizes;
>>> + bcontainer->pgsizes = info.ddw.pgsizes;
>>> /*
>>> * There is a default window in just created container.
>>> * To make region_add/del simpler, we better remove this
>>> @@ -439,7 +439,7 @@ bool vfio_spapr_container_init(VFIOContainer
>> *container, Error **errp)
>>> }
>>> } else {
>>> /* The default table uses 4K pages */
>>> - container->pgsizes = 0x1000;
>>> + bcontainer->pgsizes = 0x1000;
>>> vfio_host_win_add(container, info.dma32_window_start,
>>> info.dma32_window_start +
>>> info.dma32_window_size - 1,
>>> @@ -455,7 +455,15 @@ listener_unregister_exit:
>>>
>>> void vfio_spapr_container_deinit(VFIOContainer *container)
>>> {
>>> + VFIOHostDMAWindow *hostwin, *next;
>>> +
>>> if (container->iommu_type == VFIO_SPAPR_TCE_v2_IOMMU) {
>>> memory_listener_unregister(&container->prereg_listener);
>>> }
>>> + QLIST_FOREACH_SAFE(hostwin, &container->hostwin_list, hostwin_next,
>>> + next) {
>>> + QLIST_REMOVE(hostwin, hostwin_next);
>>> + g_free(hostwin);
>>> + }
>>> +
>>> }
>>
>> I am sure this change belongs to this patch.
"I am not sure" but you read my mind !
>
> Good catch! Yes, I should move it into below patch.
> "vfio/common: Move vfio_host_win_add/del into spapr.c"
yes.
Thanks,
C.
>
> Thanks
> Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface
2023-10-31 7:57 ` Cédric Le Goater
@ 2023-10-31 8:31 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 8:31 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Yi Sun
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Tuesday, October 31, 2023 3:58 PM
>Subject: Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and
>targetted interface
>
>On 10/30/23 03:40, Duan, Zhenzhong wrote:
>>
>>
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Friday, October 27, 2023 10:03 PM
>>> Subject: Re: [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer
>and
>>> targetted interface
>>>
>>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>>> Introduce a dumb VFIOContainer base object and its targetted interface.
>>>
>>> targeted
>>
>> Will fix.
>>
>>>
>>>
>>>> This is willingly not a QOM object because we don't want it to be
>>>> visible from the user interface. The VFIOContainer will be smoothly
>>>> populated in subsequent patches as well as interfaces.
>>>>
>>>> No fucntional change intended.
>>>>
>>>> Signed-off-by: Eric Auger <eric.auger@redhat.com>
>>>> Signed-off-by: Yi Liu <yi.l.liu@intel.com>
>>>> Signed-off-by: Yi Sun <yi.y.sun@linux.intel.com>
>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>> ---
>>>> include/hw/vfio/vfio-common.h | 8 +---
>>>> include/hw/vfio/vfio-container-base.h | 64
>+++++++++++++++++++++++++++
>>>> 2 files changed, 66 insertions(+), 6 deletions(-)
>>>> create mode 100644 include/hw/vfio/vfio-container-base.h
>>>>
>>>> diff --git a/include/hw/vfio/vfio-common.h b/include/hw/vfio/vfio-
>common.h
>>>> index b9c7a7e588..d8f293cb57 100644
>>>> --- a/include/hw/vfio/vfio-common.h
>>>> +++ b/include/hw/vfio/vfio-common.h
>>>> @@ -30,6 +30,7 @@
>>>> #include <linux/vfio.h>
>>>> #endif
>>>> #include "sysemu/sysemu.h"
>>>> +#include "hw/vfio/vfio-container-base.h"
>>>>
>>>> #define VFIO_MSG_PREFIX "vfio %s: "
>>>>
>>>> @@ -81,6 +82,7 @@ typedef struct VFIOAddressSpace {
>>>> struct VFIOGroup;
>>>>
>>>> typedef struct VFIOContainer {
>>>> + VFIOContainerBase bcontainer;
>>>> VFIOAddressSpace *space;
>>>> int fd; /* /dev/vfio/vfio, empowered by the attached groups */
>>>> MemoryListener listener;
>>>> @@ -201,12 +203,6 @@ typedef struct VFIODisplay {
>>>> } dmabuf;
>>>> } VFIODisplay;
>>>>
>>>> -typedef struct {
>>>> - unsigned long *bitmap;
>>>> - hwaddr size;
>>>> - hwaddr pages;
>>>> -} VFIOBitmap;
>>>> -
>>>> VFIOAddressSpace *vfio_get_address_space(AddressSpace *as);
>>>> void vfio_put_address_space(VFIOAddressSpace *space);
>>>> bool vfio_devices_all_running_and_saving(VFIOContainer *container);
>>>> diff --git a/include/hw/vfio/vfio-container-base.h b/include/hw/vfio/vfio-
>>> container-base.h
>>>> new file mode 100644
>>>> index 0000000000..5becbd51a7
>>>> --- /dev/null
>>>> +++ b/include/hw/vfio/vfio-container-base.h
>>>> @@ -0,0 +1,64 @@
>>>> +/*
>>>> + * VFIO BASE CONTAINER
>>>> + *
>>>> + * Copyright (C) 2023 Intel Corporation.
>>>> + * Copyright Red Hat, Inc. 2023
>>>> + *
>>>> + * Authors: Yi Liu <yi.l.liu@intel.com>
>>>> + * Eric Auger <eric.auger@redhat.com>
>>>> + *
>>>> + * This program is free software; you can redistribute it and/or modify
>>>> + * it under the terms of the GNU General Public License as published by
>>>> + * the Free Software Foundation; either version 2 of the License, or
>>>> + * (at your option) any later version.
>>>> +
>>>> + * This program is distributed in the hope that it will be useful,
>>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
>>>> + * GNU General Public License for more details.
>>>> +
>>>> + * You should have received a copy of the GNU General Public License along
>>>> + * with this program; if not, see <http://www.gnu.org/licenses/>.
>>>
>>> This should be enough :
>>>
>>> SPDX-License-Identifier: GPL-2.0-or-later
>>
>> Will do.
>>
>>>
>>>> + */
>>>> +
>>>> +#ifndef HW_VFIO_VFIO_BASE_CONTAINER_H
>>>> +#define HW_VFIO_VFIO_BASE_CONTAINER_H
>>>
>>> HW_VFIO_VFIO_CONTAINER_BASE_H
>>>
>>>> +
>>>> +#include "exec/memory.h"
>>>> +#ifndef CONFIG_USER_ONLY
>>>> +#include "exec/hwaddr.h"
>>>> +#endif
>>>
>>> including "exec/memory.h" should be enough.
>>
>> Will do.
>>
>>>
>>>
>>>> +
>>>> +typedef struct VFIODevice VFIODevice;
>>>> +typedef struct VFIOIOMMUOps VFIOIOMMUOps;
>>>> +
>>>> +typedef struct {
>>>> + unsigned long *bitmap;
>>>> + hwaddr size;
>>>> + hwaddr pages;
>>>> +} VFIOBitmap;
>>>> +
>>>> +/*
>>>> + * This is the base object for vfio container backends
>>>> + */
>>>> +typedef struct VFIOContainerBase {
>>>> + const VFIOIOMMUOps *ops;
>>>> +} VFIOContainerBase;
>>>> +
>>>> +struct VFIOIOMMUOps {
>>>> + /* basic feature */
>>>> + int (*dma_map)(VFIOContainerBase *bcontainer,
>>>> + hwaddr iova, ram_addr_t size,
>>>> + void *vaddr, bool readonly);
>>>> + int (*dma_unmap)(VFIOContainerBase *bcontainer,
>>>> + hwaddr iova, ram_addr_t size,
>>>> + IOMMUTLBEntry *iotlb);
>>>
>>> Could the VFIOContainerBase *parameter be const ?
>>
>> Yes, VFIOContainerBase is not changed by dma_unmap and other
>> functions dma_unmap calls. I tried and found making it const here
>> would impact all functions it called with same parameter be const
>> in following patches which looks unrelated to the patch itself
>> to avoid compile error.
>>
>> E.g. below functions are impacted,
>> vfio_devices_all_running_and_mig_active
>> vfio_devices_all_device_dirty_tracking
>> vfio_devices_query_dirty_bitmap
>> vfio_container_query_dirty_bitmap
>> vfio_legacy_query_dirty_bitmap>
>> To make following patches cleaner to review, I would like to keep
>> current code except you or others have a strong opinion.
>> Another choice I can think of is to add const to all impact functions
>> in a separate patch.
>
>It would be good to have, for later.
OK, I'll add a patch for const change in this series.
>
>On a related topic, I find the code difficult to read because it is
>complex, of course, and dealing with a thick layer of software constructs
>but also because it lacks a consistent naming scheme in the different
>submodules. For instance, iommufd.c has various routine prefixes for
>local routines. This is quite confusing. Same for :
>
> hw/vfio/common.c
> hw/vfio/helpers.c
> hw/vfio/container.c
> hw/vfio/container-base.c
I see, let me change all routines in iommufd.c to use iommufd_* prefix.
For container-base.c, vfio_container_* is the prefix.
For other files, I can make a prefix cleanup series in future as this series
focus on iommufd.
Thanks
Zhenzhong
>
>This is not necessarily introduced by the changes of this series.
>Something to improve for sure though.
>
>Thanks,
>
>C.
>
>
>>
>>>
>>>> + int (*attach_device)(char *name, VFIODevice *vbasedev,
>>>
>>> cont char *name ?
>>
>> Yes, will fix.
>>
>> Thanks
>> Zhenzhong
>>
>>>
>>> The rest looks good to me.
>>>
>>> Thanks
>>>
>>> C.
>>>
>>>
>>>> + AddressSpace *as, Error **errp);
>>>> + void (*detach_device)(VFIODevice *vbasedev);
>>>> + /* migration feature */
>>>> + int (*set_dirty_page_tracking)(VFIOContainerBase *bcontainer, bool
>start);
>>>> + int (*query_dirty_bitmap)(VFIOContainerBase *bcontainer, VFIOBitmap
>>> *vbmap,
>>>> + hwaddr iova, hwaddr size);
>>>> +};
>>>> +#endif /* HW_VFIO_VFIO_BASE_CONTAINER_H */
>>
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 23/37] Add iommufd configure option
2023-10-26 10:30 ` [PATCH v3 23/37] Add iommufd configure option Zhenzhong Duan
@ 2023-10-31 8:58 ` Cédric Le Goater
2023-10-31 10:52 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 8:58 UTC (permalink / raw)
To: Zhenzhong Duan, qemu-devel
Cc: alex.williamson, jgg, nicolinc, joao.m.martins, eric.auger,
peterx, jasowang, kevin.tian, yi.l.liu, yi.y.sun, chao.p.peng,
Paolo Bonzini, Marc-André Lureau, Daniel P. Berrangé,
Thomas Huth, Philippe Mathieu-Daudé
On 10/26/23 12:30, Zhenzhong Duan wrote:
> This adds "--enable-iommufd/--disable-iommufd" to enable or disable
> iommufd support, enabled by default.
It should be disabled by default for PPC.
Thanks,
C.
>
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
> Signed-off-by: Cédric Le Goater <clg@redhat.com>
> ---
> meson.build | 6 ++++++
> meson_options.txt | 2 ++
> scripts/meson-buildoptions.sh | 3 +++
> 3 files changed, 11 insertions(+)
>
> diff --git a/meson.build b/meson.build
> index 4961c82a6b..a8b0edf7b1 100644
> --- a/meson.build
> +++ b/meson.build
> @@ -560,6 +560,10 @@ have_tpm = get_option('tpm') \
> .require(targetos != 'windows', error_message: 'TPM emulation only available on POSIX systems') \
> .allowed()
>
> +have_iommufd = get_option('iommufd') \
> + .require(targetos == 'linux', error_message: 'iommufd is supported only on Linux') \
> + .allowed()
> +
> # vhost
> have_vhost_user = get_option('vhost_user') \
> .disable_auto_if(targetos != 'linux') \
> @@ -2133,6 +2137,7 @@ if get_option('tcg').allowed()
> endif
> config_host_data.set('CONFIG_TPM', have_tpm)
> config_host_data.set('CONFIG_TSAN', get_option('tsan'))
> +config_host_data.set('CONFIG_IOMMUFD', have_iommufd)
> config_host_data.set('CONFIG_USB_LIBUSB', libusb.found())
> config_host_data.set('CONFIG_VDE', vde.found())
> config_host_data.set('CONFIG_VHOST_NET', have_vhost_net)
> @@ -4074,6 +4079,7 @@ summary_info += {'vhost-user-crypto support': have_vhost_user_crypto}
> summary_info += {'vhost-user-blk server support': have_vhost_user_blk_server}
> summary_info += {'vhost-vdpa support': have_vhost_vdpa}
> summary_info += {'build guest agent': have_ga}
> +summary_info += {'iommufd support': have_iommufd}
> summary(summary_info, bool_yn: true, section: 'Configurable features')
>
> # Compilation information
> diff --git a/meson_options.txt b/meson_options.txt
> index 3c7398f3c6..91bb958cae 100644
> --- a/meson_options.txt
> +++ b/meson_options.txt
> @@ -109,6 +109,8 @@ option('dbus_display', type: 'feature', value: 'auto',
> description: '-display dbus support')
> option('tpm', type : 'feature', value : 'auto',
> description: 'TPM support')
> +option('iommufd', type : 'feature', value : 'auto',
> + description: 'iommufd support')
>
> # Do not enable it by default even for Mingw32, because it doesn't
> # work on Wine.
> diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
> index 7ca4b77eae..1effc46f7d 100644
> --- a/scripts/meson-buildoptions.sh
> +++ b/scripts/meson-buildoptions.sh
> @@ -125,6 +125,7 @@ meson_options_help() {
> printf "%s\n" ' guest-agent-msi Build MSI package for the QEMU Guest Agent'
> printf "%s\n" ' hvf HVF acceleration support'
> printf "%s\n" ' iconv Font glyph conversion support'
> + printf "%s\n" ' iommufd iommufd support'
> printf "%s\n" ' jack JACK sound support'
> printf "%s\n" ' keyring Linux keyring support'
> printf "%s\n" ' kvm KVM acceleration support'
> @@ -342,6 +343,8 @@ _meson_option_parse() {
> --enable-install-blobs) printf "%s" -Dinstall_blobs=true ;;
> --disable-install-blobs) printf "%s" -Dinstall_blobs=false ;;
> --interp-prefix=*) quote_sh "-Dinterp_prefix=$2" ;;
> + --enable-iommufd) printf "%s" -Diommufd=enabled ;;
> + --disable-iommufd) printf "%s" -Diommufd=disabled ;;
> --enable-jack) printf "%s" -Djack=enabled ;;
> --disable-jack) printf "%s" -Djack=disabled ;;
> --enable-keyring) printf "%s" -Dkeyring=enabled ;;
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
2023-10-31 2:26 ` Duan, Zhenzhong
@ 2023-10-31 9:01 ` Cédric Le Goater
2023-10-31 9:06 ` Duan, Zhenzhong
0 siblings, 1 reply; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 9:01 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
On 10/31/23 03:26, Duan, Zhenzhong wrote:
>
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Monday, October 30, 2023 9:57 PM
>> Subject: Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> IBM EEH is only supported by legacy backend currently, bypass it
>>> for IOMMUFD backend.
>>>
>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>> ---
>>> hw/ppc/spapr_pci_vfio.c | 4 ++--
>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
>>> index d1d07bec46..a2518838a1 100644
>>> --- a/hw/ppc/spapr_pci_vfio.c
>>> +++ b/hw/ppc/spapr_pci_vfio.c
>>> @@ -93,10 +93,10 @@ static VFIOContainer
>> *vfio_eeh_as_container(AddressSpace *as)
>>>
>>> bcontainer = QLIST_FIRST(&space->containers);
>>>
>>> - if (QLIST_NEXT(bcontainer, next)) {
>>> + if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != &vfio_legacy_ops) {
>>
>> It's curious that a test on the VFIOIOMMUOps is needed so deep in
>> the software stack, and spapr should have its own VFIOIOMMUOps, which
>> de facto doesn't support iommufd.
>
> Yes, in this series, spapr shares same ops vfio_legacy_ops, in next series I should
> check with vfio_iommu_spapr_ops.
Well, since PPC doesn't support IOMMUFD it should be tested before or compile
out as suggested on patch 23 "Add iommufd configure option"
Thanks,
C.
> The general vfio-pci device supports iommu property, if we pass a vfio device
> with iommufd backend, I guess we will crash Qemu if there is no check here.
>
> Thanks
> Zhenzhong
>
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
2023-10-31 9:01 ` Cédric Le Goater
@ 2023-10-31 9:06 ` Duan, Zhenzhong
0 siblings, 0 replies; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 9:06 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Nicholas Piggin, Daniel Henrique Barboza,
Cédric Le Goater, David Gibson, Harsh Prateek Bora,
open list:sPAPR (pseries)
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Tuesday, October 31, 2023 5:01 PM
>Subject: Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
>
>On 10/31/23 03:26, Duan, Zhenzhong wrote:
>>
>>
>>> -----Original Message-----
>>> From: Cédric Le Goater <clg@redhat.com>
>>> Sent: Monday, October 30, 2023 9:57 PM
>>> Subject: Re: [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend
>>>
>>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>>> IBM EEH is only supported by legacy backend currently, bypass it
>>>> for IOMMUFD backend.
>>>>
>>>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
>>>> ---
>>>> hw/ppc/spapr_pci_vfio.c | 4 ++--
>>>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
>>>> index d1d07bec46..a2518838a1 100644
>>>> --- a/hw/ppc/spapr_pci_vfio.c
>>>> +++ b/hw/ppc/spapr_pci_vfio.c
>>>> @@ -93,10 +93,10 @@ static VFIOContainer
>>> *vfio_eeh_as_container(AddressSpace *as)
>>>>
>>>> bcontainer = QLIST_FIRST(&space->containers);
>>>>
>>>> - if (QLIST_NEXT(bcontainer, next)) {
>>>> + if (QLIST_NEXT(bcontainer, next) || bcontainer->ops != &vfio_legacy_ops)
>{
>>>
>>> It's curious that a test on the VFIOIOMMUOps is needed so deep in
>>> the software stack, and spapr should have its own VFIOIOMMUOps, which
>>> de facto doesn't support iommufd.
>>
>> Yes, in this series, spapr shares same ops vfio_legacy_ops, in next series I
>should
>> check with vfio_iommu_spapr_ops.
>
>Well, since PPC doesn't support IOMMUFD it should be tested before or compile
>out as suggested on patch 23 "Add iommufd configure option"
Got it, I'll disabled it for PPC as you suggested.
Thanks
Zhenzhong
>
>Thanks,
>
>C.
>
>
>> The general vfio-pci device supports iommu property, if we pass a vfio device
>> with iommufd backend, I guess we will crash Qemu if there is no check here.
>>
>> Thanks
>> Zhenzhong
>>
^ permalink raw reply [flat|nested] 95+ messages in thread
* RE: [PATCH v3 23/37] Add iommufd configure option
2023-10-31 8:58 ` Cédric Le Goater
@ 2023-10-31 10:52 ` Duan, Zhenzhong
2023-10-31 11:25 ` Cédric Le Goater
0 siblings, 1 reply; 95+ messages in thread
From: Duan, Zhenzhong @ 2023-10-31 10:52 UTC (permalink / raw)
To: Cédric Le Goater, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Paolo Bonzini, Marc-André Lureau,
Daniel P. Berrangé, Thomas Huth, Philippe Mathieu-Daudé
Hi All,
>-----Original Message-----
>From: Cédric Le Goater <clg@redhat.com>
>Sent: Tuesday, October 31, 2023 4:59 PM
>Subject: Re: [PATCH v3 23/37] Add iommufd configure option
>
>On 10/26/23 12:30, Zhenzhong Duan wrote:
>> This adds "--enable-iommufd/--disable-iommufd" to enable or disable
>> iommufd support, enabled by default.
>
>It should be disabled by default for PPC.
I didn't find how to disable iommufd for PPC after digging in https://www.qemu.org/docs/master/devel/kconfig.html
Appreciate any guide or help! Thanks in advance.
BRs.
Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
* Re: [PATCH v3 23/37] Add iommufd configure option
2023-10-31 10:52 ` Duan, Zhenzhong
@ 2023-10-31 11:25 ` Cédric Le Goater
0 siblings, 0 replies; 95+ messages in thread
From: Cédric Le Goater @ 2023-10-31 11:25 UTC (permalink / raw)
To: Duan, Zhenzhong, qemu-devel@nongnu.org
Cc: alex.williamson@redhat.com, jgg@nvidia.com, nicolinc@nvidia.com,
Martins, Joao, eric.auger@redhat.com, peterx@redhat.com,
jasowang@redhat.com, Tian, Kevin, Liu, Yi L, Sun, Yi Y,
Peng, Chao P, Paolo Bonzini, Marc-André Lureau,
Daniel P. Berrangé, Thomas Huth, Philippe Mathieu-Daudé
On 10/31/23 11:52, Duan, Zhenzhong wrote:
> Hi All,
>
>> -----Original Message-----
>> From: Cédric Le Goater <clg@redhat.com>
>> Sent: Tuesday, October 31, 2023 4:59 PM
>> Subject: Re: [PATCH v3 23/37] Add iommufd configure option
>>
>> On 10/26/23 12:30, Zhenzhong Duan wrote:
>>> This adds "--enable-iommufd/--disable-iommufd" to enable or disable
>>> iommufd support, enabled by default.
>>
>> It should be disabled by default for PPC.
>
> I didn't find how to disable iommufd for PPC after digging in https://www.qemu.org/docs/master/devel/kconfig.html
>
> Appreciate any guide or help! Thanks in advance.
I believe the config approach should be changed to use a Kconfig
entry instead.
Thanks,
C.
>
> BRs.
> Zhenzhong
^ permalink raw reply [flat|nested] 95+ messages in thread
end of thread, other threads:[~2023-10-31 11:26 UTC | newest]
Thread overview: 95+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-10-26 10:30 [PATCH v3 00/37] vfio: Adopt iommufd Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 01/37] vfio/container: Move IBM EEH related functions into spapr_pci_vfio.c Zhenzhong Duan
2023-10-26 14:23 ` Eric Farman
2023-10-27 9:19 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 02/37] vfio/container: Move vfio_container_add/del_section_window into spapr.c Zhenzhong Duan
2023-10-27 9:19 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 03/37] vfio/container: Move spapr specific init/deinit " Zhenzhong Duan
2023-10-27 9:27 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 04/37] vfio/spapr: Make vfio_spapr_create/remove_window static Zhenzhong Duan
2023-10-27 9:27 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 05/37] vfio/common: Move vfio_host_win_add/del into spapr.c Zhenzhong Duan
2023-10-27 9:30 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 06/37] vfio: Introduce base object for VFIOContainer and targetted interface Zhenzhong Duan
2023-10-27 14:02 ` Cédric Le Goater
2023-10-30 2:40 ` Duan, Zhenzhong
2023-10-31 7:57 ` Cédric Le Goater
2023-10-31 8:31 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 07/37] vfio/container: Introduce a empty VFIOIOMMUOps Zhenzhong Duan
2023-10-27 14:20 ` Cédric Le Goater
2023-10-30 2:43 ` Duan, Zhenzhong
2023-10-31 8:21 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 08/37] vfio/container: Switch to dma_map|unmap API Zhenzhong Duan
2023-10-27 14:26 ` Cédric Le Goater
2023-10-30 2:48 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 09/37] vfio/common: Move giommu_list in base container Zhenzhong Duan
2023-10-27 14:38 ` Cédric Le Goater
2023-10-30 2:48 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 10/37] vfio/container: Move space field to " Zhenzhong Duan
2023-10-27 14:46 ` Cédric Le Goater
2023-10-30 2:51 ` Duan, Zhenzhong
2023-10-31 8:24 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 11/37] vfio/container: Switch to IOMMU BE set_dirty_page_tracking/query_dirty_bitmap API Zhenzhong Duan
2023-10-27 14:53 ` Cédric Le Goater
2023-10-30 2:53 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 12/37] vfio/container: Move per container device list in base container Zhenzhong Duan
2023-10-27 15:03 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 13/37] vfio/container: Convert functions to " Zhenzhong Duan
2023-10-27 15:03 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 14/37] vfio/container: Move vrdl_list, pgsizes and dma_max_mappings " Zhenzhong Duan
2023-10-27 15:52 ` Cédric Le Goater
2023-10-30 3:14 ` Duan, Zhenzhong
2023-10-31 8:26 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 15/37] vfio/container: Move listener " Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 16/37] vfio/container: Move dirty_pgsizes and max_dirty_bitmap_size " Zhenzhong Duan
2023-10-27 16:01 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 17/37] vfio/container: Move iova_ranges " Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 18/37] vfio/container: Implement attach/detach_device Zhenzhong Duan
2023-10-27 16:04 ` Cédric Le Goater
2023-10-27 16:06 ` Cédric Le Goater
2023-10-30 3:20 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 19/37] vfio/spapr: Introduce spapr backend and target interface Zhenzhong Duan
2023-10-27 16:04 ` Cédric Le Goater
2023-10-30 3:15 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 20/37] vfio/spapr: switch to spapr IOMMU BE add/del_section_window Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 21/37] vfio/spapr: Move prereg_listener into spapr container Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 22/37] vfio/spapr: Move hostwin_list " Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 23/37] Add iommufd configure option Zhenzhong Duan
2023-10-31 8:58 ` Cédric Le Goater
2023-10-31 10:52 ` Duan, Zhenzhong
2023-10-31 11:25 ` Cédric Le Goater
2023-10-26 10:30 ` [PATCH v3 24/37] backends/iommufd: Introduce the iommufd object Zhenzhong Duan
2023-10-26 13:27 ` Markus Armbruster
2023-10-27 7:50 ` Duan, Zhenzhong
2023-10-27 8:30 ` Markus Armbruster
2023-10-27 9:41 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 25/37] util/char_dev: Add open_cdev() Zhenzhong Duan
2023-10-30 14:53 ` Cédric Le Goater
2023-10-31 1:59 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 26/37] vfio/iommufd: Implement the iommufd backend Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 27/37] vfio/iommufd: Switch to manual hwpt allocation Zhenzhong Duan
2023-10-30 13:52 ` Cédric Le Goater
2023-10-31 2:02 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 28/37] vfio/iommufd: Add support for iova_ranges Zhenzhong Duan
2023-10-26 10:30 ` [PATCH v3 29/37] vfio/iommufd: Bypass EEH if iommufd backend Zhenzhong Duan
2023-10-30 13:56 ` Cédric Le Goater
2023-10-31 2:26 ` Duan, Zhenzhong
2023-10-31 9:01 ` Cédric Le Goater
2023-10-31 9:06 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 30/37] vfio/pci: Extract out a helper vfio_pci_get_pci_hot_reset_info Zhenzhong Duan
2023-10-30 13:59 ` Cédric Le Goater
2023-10-31 2:30 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 31/37] vfio/pci: Adapt vfio pci hot reset support with iommufd BE Zhenzhong Duan
2023-10-30 14:04 ` Cédric Le Goater
2023-10-31 2:27 ` Duan, Zhenzhong
2023-10-26 10:30 ` [PATCH v3 32/37] vfio/pci: Allow the selection of a given iommu backend Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 33/37] vfio/pci: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 34/37] vfio: Allow the selection of a given iommu backend for platform ap and ccw Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 35/37] vfio/platform: Make vfio cdev pre-openable by passing a file handle Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 36/37] vfio/ap: " Zhenzhong Duan
2023-10-26 10:31 ` [PATCH v3 37/37] vfio/ccw: " Zhenzhong Duan
2023-10-26 13:27 ` [PATCH v3 00/37] vfio: Adopt iommufd Markus Armbruster
2023-10-27 6:17 ` Duan, Zhenzhong
2023-10-27 7:45 ` Cédric Le Goater
2023-10-27 8:16 ` Duan, Zhenzhong
2023-10-27 13:43 ` Cédric Le Goater
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).