From: Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
To: svt-core@virtuozzo.com
Cc: andrey.drobyshev@virtuozzo.com
Subject: [PATCH QEMU HCI-8.1 v3 00/28] vfio: support qemu-update with passed through GPU's and vGPU's #VSTOR-103860
Date: Fri, 16 Oct 2026 18:25:58 +0300 [thread overview]
Message-ID: <20261016-vfio_v1-v3-0-47487cb5a8e5@virtuozzo.com> (raw)
v2 -> v3:
Fixes of the issues found by Sashiko AI review.
- Patch 1 (NEW) ("VZ: migration/cpr-exec: fix Error leak in preserve_ram()")
- Patch 2 (NEW) ("migration/cpr-exec: restore RAM write access if exec fails"):
RAM was left read-only after a failed exec
- Patch 3 (NEW) ("vfio/region: create BAR dma-bufs with O_CLOEXEC"): every
cpr-exec leaked the BAR dma-buf fds into the new QEMU
- Patch 11 -> 14 ("migration: yank: Move register instance earlier"): don't
abort on a duplicate yank instance when migrate is issued on a target
which has postcopy advised; such a migrate is refused instead
- Patch 17 -> 20 ("migration/cpr: add cpr_is_in_progress()"): false for
snapshots, which run with the mode parameter still set
- Patch 18 -> 21 ("vfio/migration: don't migrate device state during CPR"):
all sites, incl. switchover ack, are gated on cpr_is_in_progress();
multifd_transfer is reset at cleanup, so a stale value doesn't let the
multifd save thread run during CPR
- Patch 19 -> 22 ("vfio/listener: skip dirty page tracking during CPR"):
gated on cpr_is_in_progress() instead of the mode alone
- Patch 21 -> 24 ("VZ: migration/cpr-exec: abort cpr-exec if memory can't be
preserved"): single failure path, restores RAM write access too
- Patch 22 -> 25 ("VZ: migration/cpr-exec: take the vhost-net devices back if
exec fails"): only the devices which were actually released are taken
back
- Patch 26 (NEW) ("VZ: migration/cpr-exec: don't exec if a vhost-net device
can't be released"): previously the exec went ahead with a device
still owned by the old QEMU
- Patch 23 -> 27 ("VZ: vfio/region: preserve BAR mappings across cpr-exec"):
the cpr-transfer blocker is registered at device attach and reported
through errp; missing saved mappings are mmapped anew instead of being
an error; the saved mappings are named by device, region and index
v2: http://lore.virtuozzo.com/20261013-vfio_v1-v2-0-433455c0a294@virtuozzo.com
---
Akihiko Odaki (1):
vfio/pci: Grow buffer in vfio_pci_host_match()
Andrey Drobyshev (14):
VZ: migration/cpr-exec: fix Error leak in preserve_ram()
migration/cpr-exec: restore RAM write access if exec fails
vfio/region: create BAR dma-bufs with O_CLOEXEC
kvm: tolerate ioctls after kvm_close()
kvm: fix coalesced MMIO flush racing with kvm_close()
migration/cpr: add cpr_is_in_progress()
vfio/migration: don't migrate device state during CPR
vfio/listener: skip dirty page tracking during CPR
VZ: vfio/cpr: don't close KVM on cpr-exec
VZ: migration/cpr-exec: abort cpr-exec if memory can't be preserved
VZ: migration/cpr-exec: take the vhost-net devices back if exec fails
VZ: migration/cpr-exec: don't exec if a vhost-net device can't be released
VZ: vfio/region: preserve BAR mappings across cpr-exec #VSTOR-103860
VZ: migration/cpr-exec: allow cpr-exec with VFIO devices #VSTOR-103860
Arun Menon (1):
migration: Fix regression of passing error_fatal into vmstate_load_state()
Cédric Le Goater (2):
vfio/pci: Reject invalid MSI-X Table and PBA BIR values
vfio/pci: Fix information leak in vfio_rom_read()
Fabiano Rosas (3):
migration: yank: Move register instance earlier
migration: Free cpr-transfer MigrationAddress along with gsource
migration: Move CPR HUP watch to cpr-transfer.c
Igor Mammedov (1):
q35: Fix migration of SMRAM state
Mario Casquero (1):
vfio/pci: Initialize rom_read_failed in vfio_pci_load_rom()
Markus Armbruster (1):
migration: Plug memory leaks after migrate_set_error()
Peter Xu (1):
migration: Fix crash on second migration when cancel early
Tycho Andersen (AMD) (1):
hw/vfio: Fix liveness check in vfio_connect_kvm_msi_virq()
Vladimir Sementsov-Ogievskiy (1):
migration: vmstate_save_state_v: fix double error_setg
Yanfei Xu (1):
migration: ensure APIC is loaded prior to VFIO PCI devices
accel/kvm/kvm-all.c | 23 ++++++++++
hw/display/virtio-gpu.c | 22 +++++++---
hw/i386/kvm/apic.c | 6 ++-
hw/intc/apic_common.c | 1 +
hw/net/vhost_net.c | 41 ++++++++++++++++--
hw/pci-host/q35.c | 25 +++++------
hw/pci/pci.c | 15 ++++++-
hw/s390x/virtio-ccw.c | 17 +++++++-
hw/scsi/spapr_vscsi.c | 10 ++++-
hw/vfio/cpr.c | 11 +++--
hw/vfio/device.c | 29 ++++++++++---
hw/vfio/listener.c | 6 +++
hw/vfio/migration-multifd.c | 2 +
hw/vfio/migration.c | 38 ++++++++++++++++-
hw/vfio/pci.c | 25 +++++++----
hw/vfio/region.c | 47 +++++++++++++++++++-
hw/virtio/virtio-mmio.c | 15 ++++++-
hw/virtio/virtio-pci.c | 15 ++++++-
hw/virtio/virtio.c | 10 +++--
include/hw/vfio/vfio-cpr.h | 1 +
include/hw/vfio/vfio-device.h | 2 -
include/hw/virtio/vhost.h | 2 +
include/migration/cpr.h | 7 +++
include/migration/vmstate.h | 1 +
include/net/vhost_net.h | 3 +-
migration/cpr-exec.c | 40 ++++++++++++++---
migration/cpr-transfer.c | 33 +++++++++++++++
migration/cpr.c | 22 ++++++++++
migration/migration.c | 99 ++++++++++++++++++++++---------------------
migration/migration.h | 5 +++
migration/multifd.c | 2 +
migration/vmstate.c | 7 ++-
tests/qtest/q35-test.c | 6 ---
33 files changed, 464 insertions(+), 124 deletions(-)
---
base-commit: f3c1d9ef263edfab8d2a273078493bb766dcf8f8
change-id: 20261009-vfio_v1-a176cddd6b64
Best regards,
--
Andrey Drobyshev <andrey.drobyshev@virtuozzo.com>
next reply other threads:[~2026-10-16 15:26 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-16 15:25 Andrey Drobyshev [this message]
2026-10-16 15:25 ` [PATCH QEMU HCI-8.1 v3 01/28] VZ: migration/cpr-exec: fix Error leak in preserve_ram() Andrey Drobyshev
2026-10-16 15:28 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 02/28] migration/cpr-exec: restore RAM write access if exec fails Andrey Drobyshev
2026-10-16 15:28 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 03/28] vfio/region: create BAR dma-bufs with O_CLOEXEC Andrey Drobyshev
2026-10-16 15:30 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 04/28] migration: Plug memory leaks after migrate_set_error() Andrey Drobyshev
2026-10-16 15:29 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 05/28] hw/vfio: Fix liveness check in vfio_connect_kvm_msi_virq() Andrey Drobyshev
2026-10-16 15:30 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 06/28] vfio/pci: Reject invalid MSI-X Table and PBA BIR values Andrey Drobyshev
2026-10-16 15:30 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 07/28] vfio/pci: Fix information leak in vfio_rom_read() Andrey Drobyshev
2026-10-16 15:31 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 08/28] vfio/pci: Initialize rom_read_failed in vfio_pci_load_rom() Andrey Drobyshev
2026-10-16 15:32 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 09/28] vfio/pci: Grow buffer in vfio_pci_host_match() Andrey Drobyshev
2026-10-16 15:31 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 10/28] q35: Fix migration of SMRAM state Andrey Drobyshev
2026-10-16 15:33 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 11/28] migration: ensure APIC is loaded prior to VFIO PCI devices Andrey Drobyshev
2026-10-16 15:32 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 12/28] migration: Fix regression of passing error_fatal into vmstate_load_state() Andrey Drobyshev
2026-10-16 15:35 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 13/28] migration: vmstate_save_state_v: fix double error_setg Andrey Drobyshev
2026-10-16 15:38 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 14/28] migration: yank: Move register instance earlier Andrey Drobyshev
2026-10-16 15:41 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 15/28] migration: Free cpr-transfer MigrationAddress along with gsource Andrey Drobyshev
2026-10-16 15:39 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 16/28] migration: Move CPR HUP watch to cpr-transfer.c Andrey Drobyshev
2026-10-16 15:40 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 17/28] migration: Fix crash on second migration when cancel early Andrey Drobyshev
2026-10-16 15:44 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 18/28] kvm: tolerate ioctls after kvm_close() Andrey Drobyshev
2026-10-16 15:44 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 19/28] kvm: fix coalesced MMIO flush racing with kvm_close() Andrey Drobyshev
2026-10-16 15:46 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 20/28] migration/cpr: add cpr_is_in_progress() Andrey Drobyshev
2026-10-16 15:46 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 21/28] vfio/migration: don't migrate device state during CPR Andrey Drobyshev
2026-10-16 15:55 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 22/28] vfio/listener: skip dirty page tracking " Andrey Drobyshev
2026-10-16 15:50 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 23/28] VZ: vfio/cpr: don't close KVM on cpr-exec Andrey Drobyshev
2026-10-16 15:52 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 24/28] VZ: migration/cpr-exec: abort cpr-exec if memory can't be preserved Andrey Drobyshev
2026-10-16 15:55 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 25/28] VZ: migration/cpr-exec: take the vhost-net devices back if exec fails Andrey Drobyshev
2026-10-16 15:58 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 26/28] VZ: migration/cpr-exec: don't exec if a vhost-net device can't be released Andrey Drobyshev
2026-10-16 15:56 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 27/28] VZ: vfio/region: preserve BAR mappings across cpr-exec #VSTOR-103860 Andrey Drobyshev
2026-10-16 16:02 ` no-reply
2026-10-16 15:26 ` [PATCH QEMU HCI-8.1 v3 28/28] VZ: migration/cpr-exec: allow cpr-exec with VFIO devices #VSTOR-103860 Andrey Drobyshev
2026-10-16 15:59 ` no-reply
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=20261016-vfio_v1-v3-0-47487cb5a8e5@virtuozzo.com \
--to=andrey.drobyshev@virtuozzo.com \
--cc=svt-core@virtuozzo.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