From: Avihai Horon <avihaih@nvidia.com>
To: Joao Martins <joao.m.martins@oracle.com>, qemu-devel@nongnu.org
Cc: Yi Liu <yi.l.liu@intel.com>, Eric Auger <eric.auger@redhat.com>,
Zhenzhong Duan <zhenzhong.duan@intel.com>,
Alex Williamson <alex.williamson@redhat.com>,
Cedric Le Goater <clg@redhat.com>,
Paolo Bonzini <pbonzini@redhat.com>,
"Daniel P . Berrange" <berrange@redhat.com>,
Eduardo Habkost <eduardo@habkost.net>,
Eric Blake <eblake@redhat.com>,
Markus Armbruster <armbru@redhat.com>,
Jason Gunthorpe <jgg@nvidia.com>
Subject: Re: [PATCH RFCv2 4/8] vfio/iommufd: Implement VFIOIOMMUClass::set_dirty_tracking support
Date: Mon, 19 Feb 2024 11:17:40 +0200 [thread overview]
Message-ID: <9bd9ab52-6875-4cdd-9372-0b71856a5ae9@nvidia.com> (raw)
In-Reply-To: <20240212135643.5858-5-joao.m.martins@oracle.com>
Hi Joao,
On 12/02/2024 15:56, Joao Martins wrote:
> External email: Use caution opening links or attachments
>
>
> ioctl(iommufd, IOMMU_HWPT_SET_DIRTY_TRACKING, arg) is the UAPI that
> enables or disables dirty page tracking.
>
> It is called on the whole list of iommu domains it is are tracking,
> and on failure it rolls it back.
>
> Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
> ---
> backends/iommufd.c | 19 +++++++++++++++++++
> backends/trace-events | 1 +
> hw/vfio/common.c | 7 ++++++-
> hw/vfio/iommufd.c | 28 ++++++++++++++++++++++++++++
> include/sysemu/iommufd.h | 3 +++
> 5 files changed, 57 insertions(+), 1 deletion(-)
>
> diff --git a/backends/iommufd.c b/backends/iommufd.c
> index 2970135af4b9..954de61c2da0 100644
> --- a/backends/iommufd.c
> +++ b/backends/iommufd.c
> @@ -240,6 +240,25 @@ int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
> return !ret ? 0 : -errno;
> }
>
> +int iommufd_backend_set_dirty_tracking(IOMMUFDBackend *be, uint32_t hwpt_id,
> + bool start)
> +{
> + int ret;
> + struct iommu_hwpt_set_dirty_tracking set_dirty = {
> + .size = sizeof(set_dirty),
> + .hwpt_id = hwpt_id,
> + .flags = !start ? 0 : IOMMU_HWPT_DIRTY_TRACKING_ENABLE,
> + };
> +
> + ret = ioctl(be->fd, IOMMU_HWPT_SET_DIRTY_TRACKING, &set_dirty);
> + trace_iommufd_backend_set_dirty(be->fd, hwpt_id, start, ret);
> + if (ret) {
> + error_report("IOMMU_HWPT_SET_DIRTY_TRACKING failed: %s",
> + strerror(errno));
> + }
> + return !ret ? 0 : -errno;
> +}
> +
> static const TypeInfo iommufd_backend_info = {
> .name = TYPE_IOMMUFD_BACKEND,
> .parent = TYPE_OBJECT,
> diff --git a/backends/trace-events b/backends/trace-events
> index f83a276a4253..feba2baca5f7 100644
> --- a/backends/trace-events
> +++ b/backends/trace-events
> @@ -16,3 +16,4 @@ iommufd_backend_unmap_dma(int iommufd, uint32_t ioas, uint64_t iova, uint64_t si
> iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id, uint32_t pt_id, uint32_t flags, uint32_t hwpt_type, uint32_t len, uint64_t data_ptr, uint32_t out_hwpt_id, int ret) " iommufd=%d dev_id=%u pt_id=%u flags=0x%x hwpt_type=%u len=%u data_ptr=0x%"PRIx64" out_hwpt=%u (%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_set_dirty(int iommufd, uint32_t hwpt_id, bool start, int ret) " iommufd=%d hwpt=%d enable=%d (%d)"
s/hwpt=%d/hwpt=%u
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index d8fc7077f839..a940c0b6ede8 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -190,7 +190,7 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainerBase *bcontainer)
> QLIST_FOREACH(vbasedev, &bcontainer->device_list, container_next) {
> VFIOMigration *migration = vbasedev->migration;
>
> - if (!migration) {
> + if (!migration && !vbasedev->iommufd_dev.iommufd) {
> return false;
> }
>
> @@ -199,6 +199,11 @@ static bool vfio_devices_all_dirty_tracking(VFIOContainerBase *bcontainer)
> vfio_device_state_is_precopy(vbasedev))) {
> return false;
> }
> +
> + if (vbasedev->iommufd_dev.iommufd &&
> + !bcontainer->dirty_pages_supported) {
> + return false;
> + }
Why do we need this and the above?
IIUC, vfio_devices_all_dirty_tracking() is used to check if this is a
"proper time" to issue a dirty page sync (e.g., if migration is active,
if we are in pre-copy and dirty tracking during pre-copy is enabled).
If it's a "proper time" to do dirty page sync, even if
bcontainer->dirty_pages_supported is false, we should still issue a
dirty sync which will mark all dirty.
> }
> return true;
> }
> diff --git a/hw/vfio/iommufd.c b/hw/vfio/iommufd.c
> index edacb6d72748..361e659288fd 100644
> --- a/hw/vfio/iommufd.c
> +++ b/hw/vfio/iommufd.c
> @@ -25,6 +25,7 @@
> #include "qemu/cutils.h"
> #include "qemu/chardev_open.h"
> #include "pci.h"
> +#include "migration/migration.h"
This is redundant.
Thanks.
>
> static int iommufd_cdev_map(const VFIOContainerBase *bcontainer, hwaddr iova,
> ram_addr_t size, void *vaddr, bool readonly)
> @@ -115,6 +116,32 @@ static void iommufd_cdev_unbind_and_disconnect(VFIODevice *vbasedev)
> iommufd_backend_disconnect(vbasedev->iommufd_dev.iommufd);
> }
>
> +static int iommufd_set_dirty_page_tracking(const VFIOContainerBase *bcontainer,
> + bool start)
> +{
> + const VFIOIOMMUFDContainer *container =
> + container_of(bcontainer, VFIOIOMMUFDContainer, bcontainer);
> + int ret;
> + VFIOIOASHwpt *hwpt;
> +
> + QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
> + ret = iommufd_backend_set_dirty_tracking(container->be,
> + hwpt->hwpt_id, start);
> + if (ret) {
> + goto err;
> + }
> + }
> +
> + return 0;
> +
> +err:
> + QLIST_FOREACH(hwpt, &container->hwpt_list, next) {
> + iommufd_backend_set_dirty_tracking(container->be,
> + hwpt->hwpt_id, !start);
> + }
> + return ret;
> +}
> +
> static int iommufd_cdev_getfd(const char *sysfs_path, Error **errp)
> {
> long int ret = -ENOTTY;
> @@ -737,6 +764,7 @@ static void vfio_iommu_iommufd_class_init(ObjectClass *klass, void *data)
> vioc->detach_device = iommufd_cdev_detach;
> vioc->pci_hot_reset = iommufd_cdev_pci_hot_reset;
> vioc->host_iommu_device_init = vfio_cdev_host_iommu_device_init;
> + vioc->set_dirty_page_tracking = iommufd_set_dirty_page_tracking;
> };
>
> static const TypeInfo types[] = {
> diff --git a/include/sysemu/iommufd.h b/include/sysemu/iommufd.h
> index 1966b75caae2..562c189dd92c 100644
> --- a/include/sysemu/iommufd.h
> +++ b/include/sysemu/iommufd.h
> @@ -53,4 +53,7 @@ int iommufd_backend_alloc_hwpt(int iommufd, uint32_t dev_id,
> uint32_t pt_id, uint32_t flags,
> uint32_t data_type, uint32_t data_len,
> void *data_ptr, uint32_t *out_hwpt);
> +int iommufd_backend_set_dirty_tracking(IOMMUFDBackend *be, uint32_t hwpt_id,
> + bool start);
> +
> #endif
> --
> 2.39.3
>
next prev parent reply other threads:[~2024-02-19 9:19 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-12 13:56 [PATCH RFCv2 0/8] vfio/iommufd: IOMMUFD Dirty Tracking Joao Martins
2024-02-12 13:56 ` [PATCH RFCv2 1/8] backends/iommufd: Introduce helper function iommufd_device_get_hw_capabilities() Joao Martins
2024-02-26 7:29 ` Duan, Zhenzhong
2024-02-26 10:10 ` Joao Martins
2024-02-27 4:04 ` Duan, Zhenzhong
2024-02-12 13:56 ` [PATCH RFCv2 2/8] vfio/iommufd: Introduce auto domain creation Joao Martins
2024-02-12 16:27 ` Jason Gunthorpe
2024-02-12 18:09 ` Joao Martins
2024-02-26 9:14 ` Duan, Zhenzhong
2024-02-13 12:01 ` Joao Martins
2024-02-19 8:58 ` Avihai Horon
2024-02-20 10:42 ` Joao Martins
2024-02-12 13:56 ` [PATCH RFCv2 3/8] vfio/iommufd: Probe and request hwpt dirty tracking capability Joao Martins
2024-02-19 9:03 ` Avihai Horon
2024-02-20 10:51 ` Joao Martins
2024-02-20 12:46 ` Avihai Horon
2024-02-12 13:56 ` [PATCH RFCv2 4/8] vfio/iommufd: Implement VFIOIOMMUClass::set_dirty_tracking support Joao Martins
2024-02-19 9:17 ` Avihai Horon [this message]
2024-02-12 13:56 ` [PATCH RFCv2 5/8] vfio/iommufd: Implement VFIOIOMMUClass::query_dirty_bitmap support Joao Martins
2024-02-19 9:30 ` Avihai Horon
2024-02-20 10:59 ` Joao Martins
2024-02-20 12:52 ` Avihai Horon
2024-02-20 13:17 ` Joao Martins
2024-02-12 13:56 ` [PATCH RFCv2 6/8] backends/iommufd: Add ability to disable hugepages Joao Martins
2024-02-12 16:59 ` Jason Gunthorpe
2024-02-12 17:17 ` Markus Armbruster
2024-02-12 19:16 ` Joao Martins
2024-02-19 10:05 ` Avihai Horon
2024-02-20 11:01 ` Joao Martins
2024-02-12 13:56 ` [PATCH RFCv2 7/8] vfio/migration: Don't block migration device dirty tracking is unsupported Joao Martins
2024-02-19 10:12 ` Avihai Horon
2024-02-20 11:05 ` Joao Martins
2024-02-12 13:56 ` [PATCH RFCv2 8/8] vfio/common: Allow disabling device dirty page tracking Joao Martins
2024-02-13 11:59 ` [PATCH RFCv2 0/8] vfio/iommufd: IOMMUFD Dirty Tracking Joao Martins
2024-02-14 15:40 ` Cédric Le Goater
2024-02-14 16:25 ` Joao Martins
2024-02-15 14:20 ` Eric Auger
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=9bd9ab52-6875-4cdd-9372-0b71856a5ae9@nvidia.com \
--to=avihaih@nvidia.com \
--cc=alex.williamson@redhat.com \
--cc=armbru@redhat.com \
--cc=berrange@redhat.com \
--cc=clg@redhat.com \
--cc=eblake@redhat.com \
--cc=eduardo@habkost.net \
--cc=eric.auger@redhat.com \
--cc=jgg@nvidia.com \
--cc=joao.m.martins@oracle.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=yi.l.liu@intel.com \
--cc=zhenzhong.duan@intel.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).