From: David Gibson <david@gibson.dropbear.id.au>
To: gwshan@linux.vnet.ibm.com, alex.williamson@redhat.com, aik@ozlabs.ru
Cc: lvivier@redhat.com, thuth@redhat.com, mdroth@linux.vnet.ibm.com,
qemu-devel@nongnu.org, qemu-ppc@nongnu.org,
David Gibson <david@gibson.dropbear.id.au>
Subject: [Qemu-devel] [RFC PATCH 11/14] spapr_pci: Allow EEH on spapr-pci-host-bridge
Date: Sat, 19 Sep 2015 17:18:34 +1000 [thread overview]
Message-ID: <1442647117-2726-12-git-send-email-david@gibson.dropbear.id.au> (raw)
In-Reply-To: <1442647117-2726-1-git-send-email-david@gibson.dropbear.id.au>
The pseries machine type has two variants of the PCI Host Bridge device:
spapr-pci-host-bridge and spapr-pci-vfio-host-bridge. Originally, only the
latter could support VFIO devices, but we've now extended the VFIO code so
that they both can.
However, it's still only spapr-pci-vfio-host-bridge which supports the PAPR
Enhanced Error Handling (EEH) interfaces. The reason is that we don't yet
have a way to determine the correct VFIOGroup for EEH operations on the
"plain" host bridge.
Handling this in general is problematic due to some limitations in the
current code, and some bugs in the kernel EEH interfaces. However, it's
easy enough to do for the unambiguous case: where there is only one VFIO
group used on the whole host bridge i.e. one Partitionable Endpoint (PE).
This re-implements spapr_phb_check_vfio_group() in terms of the host
bridge's Partitionable Endpoints, allowing EEH on any host bridge in the
unambiguous case. This is enough to make spapr-pci-host-bridge support
EEH as well as spapr-pci-vfio-host-bridge, since the latter only supported
devices from one IOMMU group anyway (although it didn't properly
enforce that).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
---
hw/ppc/spapr_pci.c | 36 ++++++++++++++++++++++++++++++++++++
hw/ppc/spapr_pci_vfio.c | 18 ------------------
include/hw/pci-host/spapr.h | 1 -
3 files changed, 36 insertions(+), 19 deletions(-)
diff --git a/hw/ppc/spapr_pci.c b/hw/ppc/spapr_pci.c
index 81ad3ae..5614b45 100644
--- a/hw/ppc/spapr_pci.c
+++ b/hw/ppc/spapr_pci.c
@@ -41,6 +41,8 @@
#include "hw/ppc/spapr_drc.h"
#include "sysemu/device_tree.h"
+#include "hw/vfio/vfio-common.h"
+
/* Copied from the kernel arch/powerpc/platforms/pseries/msi.c */
#define RTAS_QUERY_FN 0
#define RTAS_CHANGE_FN 1
@@ -446,6 +448,40 @@ static sPAPRPHBGuestPE *spapr_phb_pe_by_device(sPAPRPHBState *phb,
return pe;
}
+static int spapr_phb_check_vfio_group(sPAPRPHBState *phb, VFIOGroup **gpp)
+{
+ sPAPRPHBGuestPE *pe;
+
+ if (QLIST_EMPTY(&phb->pe_list)) {
+ /* No EEH capable devices on this PHB */
+ return RTAS_OUT_PARAM_ERROR;
+ }
+
+ /* Limitations in both qemu and the kernel mean that, for now, EEH
+ * won't work if there are devices from more than one PE
+ * (i.e. IOMMU group) on the same PHB */
+ pe = QLIST_FIRST(&phb->pe_list);
+ if (QLIST_NEXT(pe, list)) {
+ error_report("spapr-pci-host-bridge: EEH attempted on PHB with multiple"
+ " IOMMU groups");
+ return RTAS_OUT_HW_ERROR;
+ }
+
+ if (!object_dynamic_cast(OBJECT(phb), "spapr-pci-vfio-host-bridge")) {
+ sPAPRPHBVFIOState *svphb = SPAPR_PCI_VFIO_HOST_BRIDGE(phb);
+ /* FIXME: this is an abstraction violation */
+ if (pe->group->groupid != svphb->iommugroupid) {
+ error_report("spapr-pci-vfio-host-bridge: Bad IOMMU group");
+ return RTAS_OUT_HW_ERROR;
+ }
+ }
+
+ if (gpp) {
+ *gpp = pe->group;
+ }
+ return RTAS_OUT_SUCCESS;
+}
+
static void rtas_ibm_set_eeh_option(PowerPCCPU *cpu,
sPAPRMachineState *spapr,
uint32_t token, uint32_t nargs,
diff --git a/hw/ppc/spapr_pci_vfio.c b/hw/ppc/spapr_pci_vfio.c
index b61923c..a08292a 100644
--- a/hw/ppc/spapr_pci_vfio.c
+++ b/hw/ppc/spapr_pci_vfio.c
@@ -24,29 +24,11 @@
#include "hw/vfio/vfio.h"
#include "hw/vfio/vfio-eeh.h"
-#include "hw/vfio/vfio-common.h"
-
static Property spapr_phb_vfio_properties[] = {
DEFINE_PROP_INT32("iommu", sPAPRPHBVFIOState, iommugroupid, -1),
DEFINE_PROP_END_OF_LIST(),
};
-int spapr_phb_check_vfio_group(sPAPRPHBState *phb, VFIOGroup **gpp)
-{
- VFIOGroup *group;
-
- if (!object_dynamic_cast(OBJECT(phb), "spapr-pci-vfio-host-bridge")) {
- return RTAS_OUT_PARAM_ERROR;
- }
-
- /* FIXME: this is an abstraction violation */
- group = vfio_get_group(SPAPR_PCI_VFIO_HOST_BRIDGE(phb)->iommugroupid,
- &phb->iommu_as);
- if (gpp)
- *gpp = group;
- return RTAS_OUT_SUCCESS;
-}
-
static void spapr_phb_vfio_finish_realize(sPAPRPHBState *sphb, Error **errp)
{
sPAPRPHBVFIOState *svphb = SPAPR_PCI_VFIO_HOST_BRIDGE(sphb);
diff --git a/include/hw/pci-host/spapr.h b/include/hw/pci-host/spapr.h
index 535e5ef..8c8d187 100644
--- a/include/hw/pci-host/spapr.h
+++ b/include/hw/pci-host/spapr.h
@@ -138,7 +138,6 @@ void spapr_pci_msi_init(sPAPRMachineState *spapr, hwaddr addr);
void spapr_pci_rtas_init(void);
-int spapr_phb_check_vfio_group(sPAPRPHBState *phb, VFIOGroup **gpp);
sPAPRPHBState *spapr_pci_find_phb(sPAPRMachineState *spapr, uint64_t buid);
PCIDevice *spapr_pci_find_dev(sPAPRMachineState *spapr, uint64_t buid,
uint32_t config_addr);
--
2.4.3
next prev parent reply other threads:[~2015-09-19 7:18 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-09-19 7:18 [Qemu-devel] [RFC PATCH 00/14] Allow EEH on "normal" sPAPR PCI host bridge David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 01/14] vfio: Start adding VFIO/EEH interface David Gibson
2015-09-23 17:28 ` Alex Williamson
2015-09-24 1:11 ` David Gibson
2015-09-24 2:12 ` Alex Williamson
2015-09-24 4:09 ` David Gibson
2015-09-24 5:45 ` Thomas Huth
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 02/14] spapr_pci: Switch EEH to vfio_eeh_op() interface David Gibson
2015-09-23 17:28 ` Alex Williamson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 03/14] spapr_pci: Expose and generalize spapr_phb_check_vfio_group() David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 04/14] spapr_pci: Fold spapr_phb_vfio_eeh_configure() into spapr_pci code David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 05/14] spapr_pci: Fold spapr_phb_vfio_eeh_reset() " David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 06/14] spapr_pci: Fold spapr_phb_vfio_eeh_get_state() " David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 07/14] spapr_pci: Fold spapr_phb_vfio_eeh_set_option() " David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 08/14] spapr_pci: Fold spapr_phb_vfio_eeh_configure() " David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 09/14] vfio: Expose a VFIO PCI device's group for EEH David Gibson
2015-09-23 17:28 ` Alex Williamson
2015-09-24 1:16 ` David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 10/14] spapr_pci: Track guest Partitionable Endpoints David Gibson
2015-09-19 7:18 ` David Gibson [this message]
2015-09-23 17:28 ` [Qemu-devel] [RFC PATCH 11/14] spapr_pci: Allow EEH on spapr-pci-host-bridge Alex Williamson
2015-09-24 1:49 ` David Gibson
2015-09-24 2:19 ` Alex Williamson
2015-09-24 4:11 ` David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 12/14] spapr_pci: (Mostly) remove spapr-pci-vfio-host-bridge David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 13/14] spapr_pci: Remove finish_realize hook David Gibson
2015-09-19 7:18 ` [Qemu-devel] [RFC PATCH 14/14] vfio: Eliminate vfio_container_ioctl() David Gibson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1442647117-2726-12-git-send-email-david@gibson.dropbear.id.au \
--to=david@gibson.dropbear.id.au \
--cc=aik@ozlabs.ru \
--cc=alex.williamson@redhat.com \
--cc=gwshan@linux.vnet.ibm.com \
--cc=lvivier@redhat.com \
--cc=mdroth@linux.vnet.ibm.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=thuth@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).