[GIT PULL 00/24] Various virtio-gpu/dmabuf related fixes

Marc-André Lureau posted 24 patches 2 weeks, 2 days ago
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20260910-virtio-gpu-cleanup-v1-0-ccb8927a671e@redhat.com
Maintainers: "Michael S. Tsirkin" <mst@redhat.com>, Stefano Garzarella <sgarzare@redhat.com>, "Marc-André Lureau" <marcandre.lureau@redhat.com>, "Alex Bennée" <alex.bennee@linaro.org>, Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>, Dmitry Osipenko <dmitry.osipenko@collabora.com>, Alex Williamson <alex@shazbot.org>, "Cédric Le Goater" <clg@redhat.com>
There is a newer version of this series
contrib/vhost-user-gpu/vhost-user-gpu.c |  28 ++----
contrib/vhost-user-gpu/virgl.c          |  12 +++
hw/display/vhost-user-gpu.c             |   1 -
hw/display/virtio-gpu-rutabaga.c        |  38 ++++----
hw/display/virtio-gpu-udmabuf-stubs.c   |   2 +-
hw/display/virtio-gpu-udmabuf.c         |  89 +++++++------------
hw/display/virtio-gpu-virgl.c           | 135 ++++++++++++++++++-----------
hw/display/virtio-gpu.c                 | 148 +++++++++++++++++++++-----------
hw/vfio/display.c                       |   1 -
include/hw/virtio/virtio-gpu.h          |  31 ++++---
include/ui/dmabuf.h                     |   7 --
include/ui/egl-helpers.h                |   4 +-
include/ui/gtk.h                        |  14 ++-
ui/console.c                            |   3 +
ui/dbus-listener.c                      |  69 ++++++++++-----
ui/dmabuf.c                             |  65 ++------------
ui/egl-headless.c                       |  17 ++++
ui/egl-helpers.c                        |  15 ++--
ui/gtk-egl.c                            |  55 +++++++-----
ui/gtk-gl-area.c                        |  33 ++++---
ui/gtk.c                                |  43 ++++++----
ui/sdl2-gl.c                            |   7 ++
ui/spice-display.c                      |   2 +
23 files changed, 450 insertions(+), 369 deletions(-)
[GIT PULL 00/24] Various virtio-gpu/dmabuf related fixes
Posted by Marc-André Lureau 2 weeks, 2 days ago
The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:

  Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)

are available in the Git repository at:

  https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v1

for you to fetch changes up to d26343c12d84227263b8d2fc89bac7632a98e00d:

  contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 17:29:47 +0400)

----------------------------------------------------------------
Various virtio-gpu/dmabuf related fixes

Hi,

While working on CVE-2026-66020 (which had a few revision on list), I
kept finding several places where virtio-gpu scanout and dmabuf
ownership was unclear.

The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
unmapped its memory. Following that path also found stale dmabuf
references, duplicated scanout bookkeeping, and inconsistent cleanup
between display backends.

Those changes touch a lot of area and I have done basic testing with the
various backends. I would not recommend backporting them all to stable.
The first 2 patches address CVE-2026-66020 in the general case and
should be safe to backport.

thanks

To: qemu-devel@nongnu.org
Cc: Alex Bennée <alex.bennee@linaro.org>
Cc: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Stefano Garzarella <sgarzare@redhat.com>
Cc: Alex Williamson <alex@shazbot.org>
Cc: Cédric Le Goater <clg@redhat.com>
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>

---
Changes in v5:
- EDITME: describe what is new in this series revision.
- EDITME: use bulletpoints and terse descriptions.
- Link to v4: https://lore.kernel.org/qemu-devel/20260910-virtio-gpu-cleanup-v4-0-7d623714bd71@redhat.com

Changes in v4:
- replaced "ui: make GL context current before releasing DMA-BUF
  textures" with more complete "ui: make GL context current in display
  backend callbacks"
- rebase, drop eof line leftover, collect rb trailer
- Link to v3: https://lore.kernel.org/qemu-devel/20260909-virtio-gpu-cleanup-v3-0-99266590e79d@redhat.com

Changes in v3:
- updated "virtio-gpu: fix use-after-free on blob cursor", now "clear res->blob on mapping cleanup"
- always disable GL when clearning scanout in "virtio-gpu: consolidate disabling scanout"
- drop "vhost-user-gpu: release DMA-BUFs on scanout disable" for now
- rebased, collect rb trailers
- Link to v2: https://lore.kernel.org/qemu-devel/20260904-virtio-gpu-cleanup-v2-0-15b5ae6b0023@redhat.com

Changes in v2: addresses Akihiko review
- reorder to have CVE fix first, easily back-portable (but also more
  limited fix)
- include virtio_gpu_{simple,virgl}_resource_new() patch, and other sent
  separately
- move draw_submitted from QemuDmaBuf to VirtualGfxConsole
- contrib/vhost-user-gpu: disable scanouts on resource unref
- drop rfc, collect rb, adjust commit messages
- Link to v1: https://lore.kernel.org/qemu-devel/20260825-virtio-gpu-cleanup-v1-0-d3bb612b9975@redhat.com

----------------------------------------------------------------
Marc-André Lureau (24):
      virtio-gpu: disable blob scanouts on mapping cleanup
      virtio-gpu: clear res->blob on mapping cleanup
      hw/display/virtio-gpu: introduce virtio_gpu_{simple,virgl}_resource_new()
      virtio-gpu/virgl: free BHs and timers on renderer reset
      hw/display/virtio-gpu: set share_handle for udmabuf blob resources
      ui/gtk: move GL fence tracking from QemuDmaBuf to VirtualGfxConsole
      ui/gtk: move draw_submitted from QemuDmaBuf to VirtualGfxConsole
      ui/dbus: disable scanout iff dmabuf is current
      virtio-gpu: store the scanout DMABUF in virtio_gpu_scanout
      ui/console: disable GL scanout when dmabuf is the active one
      ui/sdl2: clear guest_fb.dmabuf on release
      ui/dmabuf: own and close fds on free
      ui: make GL context current in display backend callbacks
      virtio-gpu: release the dmabuf when a scanout is disabled
      contrib/vhost-user-gpu: drop scanout_bitmask bookkeeping
      virtio-gpu-rutabaga: fix scanout handling
      virtio-gpu: drop scanout_bitmask bookkeeping
      virtio-gpu: release dmabuf when switching to a surface
      virtio-gpu/virgl: release dmabuf when a texture replaces a scanout
      virtio-gpu: consolidate disabling scanout
      virtio-gpu: minor refactoring
      virtio-gpu/virgl: disable scanouts on resource unref
      virtio-gpu: extract virtio_gpu_disable_scanout_for_resource helper
      contrib/vhost-user-gpu: disable scanouts on resource unref

 contrib/vhost-user-gpu/vhost-user-gpu.c |  28 ++----
 contrib/vhost-user-gpu/virgl.c          |  12 +++
 hw/display/vhost-user-gpu.c             |   1 -
 hw/display/virtio-gpu-rutabaga.c        |  38 ++++----
 hw/display/virtio-gpu-udmabuf-stubs.c   |   2 +-
 hw/display/virtio-gpu-udmabuf.c         |  89 +++++++------------
 hw/display/virtio-gpu-virgl.c           | 135 ++++++++++++++++++-----------
 hw/display/virtio-gpu.c                 | 148 +++++++++++++++++++++-----------
 hw/vfio/display.c                       |   1 -
 include/hw/virtio/virtio-gpu.h          |  31 ++++---
 include/ui/dmabuf.h                     |   7 --
 include/ui/egl-helpers.h                |   4 +-
 include/ui/gtk.h                        |  14 ++-
 ui/console.c                            |   3 +
 ui/dbus-listener.c                      |  69 ++++++++++-----
 ui/dmabuf.c                             |  65 ++------------
 ui/egl-headless.c                       |  17 ++++
 ui/egl-helpers.c                        |  15 ++--
 ui/gtk-egl.c                            |  55 +++++++-----
 ui/gtk-gl-area.c                        |  33 ++++---
 ui/gtk.c                                |  43 ++++++----
 ui/sdl2-gl.c                            |   7 ++
 ui/spice-display.c                      |   2 +
 23 files changed, 450 insertions(+), 369 deletions(-)


Re: [GIT PULL 00/24] Various virtio-gpu/dmabuf related fixes
Posted by Peter Maydell 2 weeks, 2 days ago
On Thu, 10 Sept 2026 at 14:30, Marc-André Lureau
<marcandre.lureau@redhat.com> wrote:
>
> The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:
>
>   Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)
>
> are available in the Git repository at:
>
>   https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v1
>
> for you to fetch changes up to d26343c12d84227263b8d2fc89bac7632a98e00d:
>
>   contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 17:29:47 +0400)
>
> ----------------------------------------------------------------
> Various virtio-gpu/dmabuf related fixes
>
> Hi,
>
> While working on CVE-2026-66020 (which had a few revision on list), I
> kept finding several places where virtio-gpu scanout and dmabuf
> ownership was unclear.
>
> The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
> unmapped its memory. Following that path also found stale dmabuf
> references, duplicated scanout bookkeeping, and inconsistent cleanup
> between display backends.
>
> Those changes touch a lot of area and I have done basic testing with the
> various backends. I would not recommend backporting them all to stable.
> The first 2 patches address CVE-2026-66020 in the general case and
> should be safe to backport.
>
> thanks
>
> To: qemu-devel@nongnu.org
> Cc: Alex Bennée <alex.bennee@linaro.org>
> Cc: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> Cc: "Michael S. Tsirkin" <mst@redhat.com>
> Cc: Stefano Garzarella <sgarzare@redhat.com>
> Cc: Alex Williamson <alex@shazbot.org>
> Cc: Cédric Le Goater <clg@redhat.com>
> Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
>
> ---
> Changes in v5:
> - EDITME: describe what is new in this series revision.
> - EDITME: use bulletpoints and terse descriptions.
> - Link to v4: https://lore.kernel.org/qemu-devel/20260910-virtio-gpu-cleanup-v4-0-7d623714bd71@redhat.com

So is this different from the v4 on list? Please don't send pull
requests as if they were new revisions of individual patch series.
This is not the mechanism the project uses: pull requests should
be an accumulation of whatever patches the maintainer of a
subsystem has. Sometimes that will happen to be a single
patchset, but usually not.

thanks
-- PMM
Re: [GIT PULL 00/24] Various virtio-gpu/dmabuf related fixes
Posted by Marc-André Lureau 2 weeks, 2 days ago
Hi

On Thu, Sep 10, 2026 at 6:45 PM Peter Maydell <peter.maydell@linaro.org> wrote:
>
> On Thu, 10 Sept 2026 at 14:30, Marc-André Lureau
> <marcandre.lureau@redhat.com> wrote:
> >
> > The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:
> >
> >   Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)
> >
> > are available in the Git repository at:
> >
> >   https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v1
> >
> > for you to fetch changes up to d26343c12d84227263b8d2fc89bac7632a98e00d:
> >
> >   contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 17:29:47 +0400)
> >
> > ----------------------------------------------------------------
> > Various virtio-gpu/dmabuf related fixes
> >
> > Hi,
> >
> > While working on CVE-2026-66020 (which had a few revision on list), I
> > kept finding several places where virtio-gpu scanout and dmabuf
> > ownership was unclear.
> >
> > The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
> > unmapped its memory. Following that path also found stale dmabuf
> > references, duplicated scanout bookkeeping, and inconsistent cleanup
> > between display backends.
> >
> > Those changes touch a lot of area and I have done basic testing with the
> > various backends. I would not recommend backporting them all to stable.
> > The first 2 patches address CVE-2026-66020 in the general case and
> > should be safe to backport.
> >
> > thanks
> >
> > To: qemu-devel@nongnu.org
> > Cc: Alex Bennée <alex.bennee@linaro.org>
> > Cc: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> > Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> > Cc: "Michael S. Tsirkin" <mst@redhat.com>
> > Cc: Stefano Garzarella <sgarzare@redhat.com>
> > Cc: Alex Williamson <alex@shazbot.org>
> > Cc: Cédric Le Goater <clg@redhat.com>
> > Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> >
> > ---
> > Changes in v5:
> > - EDITME: describe what is new in this series revision.
> > - EDITME: use bulletpoints and terse descriptions.
> > - Link to v4: https://lore.kernel.org/qemu-devel/20260910-virtio-gpu-cleanup-v4-0-7d623714bd71@redhat.com
>
> So is this different from the v4 on list? Please don't send pull
> requests as if they were new revisions of individual patch series.
> This is not the mechanism the project uses: pull requests should
> be an accumulation of whatever patches the maintainer of a
> subsystem has. Sometimes that will happen to be a single
> patchset, but usually not.

My bad, I should have cleaned-up the series cover letter. There is no
difference with v4 beside trailer tags.
I didn't include other pending ui or related patches, I could, but I
would rather have smaller PR.

I can send another PR if those extra lines in the cover letter are too
dirty. Let me know

thanks
Re: [GIT PULL 00/24] Various virtio-gpu/dmabuf related fixes
Posted by Peter Maydell 2 weeks, 2 days ago
On Thu, 10 Sept 2026 at 15:55, Marc-André Lureau
<marcandre.lureau@redhat.com> wrote:
>
> Hi
>
> On Thu, Sep 10, 2026 at 6:45 PM Peter Maydell <peter.maydell@linaro.org> wrote:
> >
> > On Thu, 10 Sept 2026 at 14:30, Marc-André Lureau
> > <marcandre.lureau@redhat.com> wrote:
> > >
> > > The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:
> > >
> > >   Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)
> > >
> > > are available in the Git repository at:
> > >
> > >   https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v1
> > >
> > > for you to fetch changes up to d26343c12d84227263b8d2fc89bac7632a98e00d:
> > >
> > >   contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 17:29:47 +0400)
> > >
> > > ----------------------------------------------------------------
> > > Various virtio-gpu/dmabuf related fixes
> > >
> > > Hi,
> > >
> > > While working on CVE-2026-66020 (which had a few revision on list), I
> > > kept finding several places where virtio-gpu scanout and dmabuf
> > > ownership was unclear.
> > >
> > > The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
> > > unmapped its memory. Following that path also found stale dmabuf
> > > references, duplicated scanout bookkeeping, and inconsistent cleanup
> > > between display backends.
> > >
> > > Those changes touch a lot of area and I have done basic testing with the
> > > various backends. I would not recommend backporting them all to stable.
> > > The first 2 patches address CVE-2026-66020 in the general case and
> > > should be safe to backport.
> > >
> > > thanks
> > >
> > > To: qemu-devel@nongnu.org
> > > Cc: Alex Bennée <alex.bennee@linaro.org>
> > > Cc: Akihiko Odaki <odaki@rsg.ci.i.u-tokyo.ac.jp>
> > > Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> > > Cc: "Michael S. Tsirkin" <mst@redhat.com>
> > > Cc: Stefano Garzarella <sgarzare@redhat.com>
> > > Cc: Alex Williamson <alex@shazbot.org>
> > > Cc: Cédric Le Goater <clg@redhat.com>
> > > Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> > >
> > > ---
> > > Changes in v5:
> > > - EDITME: describe what is new in this series revision.
> > > - EDITME: use bulletpoints and terse descriptions.
> > > - Link to v4: https://lore.kernel.org/qemu-devel/20260910-virtio-gpu-cleanup-v4-0-7d623714bd71@redhat.com
> >
> > So is this different from the v4 on list? Please don't send pull
> > requests as if they were new revisions of individual patch series.
> > This is not the mechanism the project uses: pull requests should
> > be an accumulation of whatever patches the maintainer of a
> > subsystem has. Sometimes that will happen to be a single
> > patchset, but usually not.
>
> My bad, I should have cleaned-up the series cover letter. There is no
> difference with v4 beside trailer tags.
> I didn't include other pending ui or related patches, I could, but I
> would rather have smaller PR.

Thanks for clarifying.

> I can send another PR if those extra lines in the cover letter are too
> dirty. Let me know

I think since the text is also in the commit message for the
merge commit and so it all appears in the git log it would be
worth cleaning that up and sending a fresh PR. (You can just
resend a new cover letter email with the new pullreq info,
you don't need to resend all the patchmails.)

thanks
-- PMM
[GIT PULL v2 00/24] Various virtio-gpu/dmabuf related fixes
Posted by marcandre.lureau@redhat.com 2 weeks, 2 days ago
From: Marc-André Lureau <marcandre.lureau@redhat.com>

The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:

  Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)

are available in the Git repository at:

  https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v2

for you to fetch changes up to d6e516555805b95a7a935c3fd74568dabf7796f2:

  contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 19:59:43 +0400)

----------------------------------------------------------------
Various virtio-gpu/dmabuf related fixes

While working on CVE-2026-66020 (which had a few revision on list), I
kept finding several places where virtio-gpu scanout and dmabuf
ownership was unclear.

The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
unmapped its memory. Following that path also found stale dmabuf
references, duplicated scanout bookkeeping, and inconsistent cleanup
between display backends.

Those changes touch a lot of area and I have done basic testing with the
various backends. I would not recommend backporting them all to stable.
The first 2 patches address CVE-2026-66020 in the general case and
should be safe to backport.

----------------------------------------------------------------
Marc-André Lureau (24):
      virtio-gpu: disable blob scanouts on mapping cleanup
      virtio-gpu: clear res->blob on mapping cleanup
      hw/display/virtio-gpu: introduce virtio_gpu_{simple,virgl}_resource_new()
      virtio-gpu/virgl: free BHs and timers on renderer reset
      hw/display/virtio-gpu: set share_handle for udmabuf blob resources
      ui/gtk: move GL fence tracking from QemuDmaBuf to VirtualGfxConsole
      ui/gtk: move draw_submitted from QemuDmaBuf to VirtualGfxConsole
      ui/dbus: disable scanout iff dmabuf is current
      virtio-gpu: store the scanout DMABUF in virtio_gpu_scanout
      ui/console: disable GL scanout when dmabuf is the active one
      ui/sdl2: clear guest_fb.dmabuf on release
      ui/dmabuf: own and close fds on free
      ui: make GL context current in display backend callbacks
      virtio-gpu: release the dmabuf when a scanout is disabled
      contrib/vhost-user-gpu: drop scanout_bitmask bookkeeping
      virtio-gpu-rutabaga: fix scanout handling
      virtio-gpu: drop scanout_bitmask bookkeeping
      virtio-gpu: release dmabuf when switching to a surface
      virtio-gpu/virgl: release dmabuf when a texture replaces a scanout
      virtio-gpu: consolidate disabling scanout
      virtio-gpu: minor refactoring
      virtio-gpu/virgl: disable scanouts on resource unref
      virtio-gpu: extract virtio_gpu_disable_scanout_for_resource helper
      contrib/vhost-user-gpu: disable scanouts on resource unref

 contrib/vhost-user-gpu/vhost-user-gpu.c |  28 ++----
 contrib/vhost-user-gpu/virgl.c          |  12 +++
 hw/display/vhost-user-gpu.c             |   1 -
 hw/display/virtio-gpu-rutabaga.c        |  38 ++++----
 hw/display/virtio-gpu-udmabuf-stubs.c   |   2 +-
 hw/display/virtio-gpu-udmabuf.c         |  89 +++++++------------
 hw/display/virtio-gpu-virgl.c           | 135 ++++++++++++++++++-----------
 hw/display/virtio-gpu.c                 | 148 +++++++++++++++++++++-----------
 hw/vfio/display.c                       |   1 -
 include/hw/virtio/virtio-gpu.h          |  31 ++++---
 include/ui/dmabuf.h                     |   7 --
 include/ui/egl-helpers.h                |   4 +-
 include/ui/gtk.h                        |  14 ++-
 ui/console.c                            |   3 +
 ui/dbus-listener.c                      |  69 ++++++++++-----
 ui/dmabuf.c                             |  65 ++------------
 ui/egl-headless.c                       |  17 ++++
 ui/egl-helpers.c                        |  15 ++--
 ui/gtk-egl.c                            |  55 +++++++-----
 ui/gtk-gl-area.c                        |  33 ++++---
 ui/gtk.c                                |  43 ++++++----
 ui/sdl2-gl.c                            |   7 ++
 ui/spice-display.c                      |   2 +
 23 files changed, 450 insertions(+), 369 deletions(-)


Re: [GIT PULL v2 00/24] Various virtio-gpu/dmabuf related fixes
Posted by Richard Henderson 2 weeks, 1 day ago
On 9/10/26 06:02, marcandre.lureau@redhat.com wrote:
> From: Marc-André Lureau<marcandre.lureau@redhat.com>
> 
> The following changes since commit 257bf4f160c50ca8c4ebd603f519f5c786013fb7:
> 
>    Merge tag 'for-upstream' ofhttps://gitlab.com/bonzini/qemu into staging (2026-09-10 11:08:04 +0100)
> 
> are available in the Git repository at:
> 
>    https://gitlab.com/marcandre.lureau/qemu.git tags/virtio-gpu-pr-v2
> 
> for you to fetch changes up to d6e516555805b95a7a935c3fd74568dabf7796f2:
> 
>    contrib/vhost-user-gpu: disable scanouts on resource unref (2026-09-10 19:59:43 +0400)
> 
> ----------------------------------------------------------------
> Various virtio-gpu/dmabuf related fixes
> 
> While working on CVE-2026-66020 (which had a few revision on list), I
> kept finding several places where virtio-gpu scanout and dmabuf
> ownership was unclear.
> 
> The CVE comes from using a blob after RESOURCE_DETACH_BACKING has
> unmapped its memory. Following that path also found stale dmabuf
> references, duplicated scanout bookkeeping, and inconsistent cleanup
> between display backends.
> 
> Those changes touch a lot of area and I have done basic testing with the
> various backends. I would not recommend backporting them all to stable.
> The first 2 patches address CVE-2026-66020 in the general case and
> should be safe to backport.

Applied, thanks.

r~