[regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Salvatore Bonaccorso]

Hi all,

Roland Schwarzkopf reported to the Debian mailing list a problem which
he encountered once updating in Debian from 5.10.244 to 5.10.247. The
report is quoted below and found in
https://lists.debian.org/debian-kernel/2025/12/msg00223.html

Roland did bisect the changes between 5.10.244 to 5.10.247 and found
that the issue is introduced with 1550f3673972 ("net: rtnetlink: add
bulk delete support flag") which is the backport to the 5.10.y series.

On Thu, Dec 18, 2025 at 02:59:55PM +0100, Roland Schwarzkopf wrote:
> Hi Salvatore,
> 
> On 12/17/25 20:28, Salvatore Bonaccorso wrote:
> > Hi Roland,
> > 
> > I'm CC'ing Ben Hutchings directly as well as he takes care of the
> > Debian LTS kernel updates. Idellly we make this as well a proper bug
> > for easier tracking.
> > 
> > On Wed, Dec 17, 2025 at 01:35:54PM +0100, Roland Schwarzkopf wrote:
> > > Hi there,
> > > 
> > > after upgrading to the latest kernel on Debian 11
> > > (linux-image-5.10.0-37-amd64) I have an issue using libvirt with qemu/kvm
> > > virtual machines and macvtap networking. When a machine is shut down,
> > > libvirt can not delete the corresponding macvtap device. Thus, starting the
> > > machine again is not possible. After manually removing the macvtap device
> > > using `ip link delete` the vm can be started again.
> > > 
> > > In the journal the following message is shown:
> > > 
> > > Dec 17 13:19:27 iblis libvirtd[535]: error destroying network device macvtap0: Operation not supported
> > > 
> > > After downgrading the kernel to linux-image-5.10.0-36-amd64, the problem
> > > disappears. I tested this on a fresh minimal install of Debian 11 - to
> > > exclude that anything else on my production machines is causing this issue.
> > > 
> > > Since the older kernel does not have this issue, I assume this is related to
> > > the kernel and not to libvirt?
> > > 
> > > I tried to check for bug reports of the kernel package, but the bug tracker
> > > finds no reports and even states that the package does not exist (I used the
> > > "Bug reports" link on
> > > https://packages.debian.org/bullseye/linux-image-5.10.0-37-amd64). This left
> > > me a bit puzzled. Since I don't have experience with the debian bug
> > > reporting process, I had no other idea than writing to this list.
> > You would need to search for inhttps://bugs.debian.org/src:linux ,
> > but that said I'm not aware of any bug reports in that direction.
> > 
> > Would you be in the position of bisecting the problem as you can say
> > that 5.10.244 is good and 5.10.247 is bad and regressed? If you can do
> > that that would involve compiling a couple of kernels to narrow down
> > where the problem is introduced:
> > 
> >      git clone --single-branch -b linux-5.10.yhttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
> >      cd linux-stable
> >      git checkout v5.10.244
> >      cp /boot/config-$(uname -r) .config
> >      yes '' | make localmodconfig
> >      make savedefconfig
> >      mv defconfig arch/x86/configs/my_defconfig
> > 
> >      # test 5.10.244 to ensure this is "good"
> >      make my_defconfig
> >      make -j $(nproc) bindeb-pkg
> >      ... install the resulting .deb package and confirm it successfully boots / problem does not exist
> > 
> >      # test 5.10.247 to ensure this is "bad"
> >      git checkout v5.10.247
> >      make my_defconfig
> >      make -j $(nproc) bindeb-pkg
> >      ... install the resulting .deb package and confirm it fails to boot / problem exists
> > 
> > With that confirmed, the bisection can start:
> > 
> >      git bisect start
> >      git bisect good v5.10.244
> >      git bisect bad v5.10.247
> > 
> > In each bisection step git checks out a state between the oldest
> > known-bad and the newest known-good commit. In each step test using:
> > 
> >      make my_defconfig
> >      make -j $(nproc) bindeb-pkg
> >      ... install, try to boot / verify if problem exists
> > 
> > and if the problem is hit run:
> > 
> >      git bisect bad
> > 
> > and if the problem doesn't trigger run:
> > 
> >      git bisect good
> > 
> > . Please pay attention to always select the just built kernel for
> > booting, it won't always be the default kernel picked up by grub.
> > 
> > Iterate until git announces to have identified the first bad commit.
> > 
> > Then provide the output of
> > 
> >      git bisect log
> > 
> > In the course of the bisection you might have to uninstall previous
> > kernels again to not exhaust the disk space in /boot. Also in the end
> > uninstall all self-built kernels again.
> 
> I just did my first bisection \o/ (sorry)
> 
> Here are the results:
> 
> git bisect start
> # bad: [f964b940099f9982d723d4c77988d4b0dda9c165] Linux 5.10.247
> git bisect bad f964b940099f9982d723d4c77988d4b0dda9c165
> # good: [863b76df7d1e327979946a2d3893479c3275bfa4] Linux 5.10.244
> git bisect good f52ee6ea810273e527a5d319e5f400be8c8424c1
> # good: [dc9fdb7586b90e33c766eac52b6f3d1c9ec365a1] net: usb: lan78xx: Add error handling to lan78xx_init_mac_address
> git bisect good dc9fdb7586b90e33c766eac52b6f3d1c9ec365a1
> # bad: [2272d5757ce5d3fb416d9f2497b015678eb85c0d] phy: cadence: cdns-dphy: Enable lower resolutions in dphy
> git bisect bad 2272d5757ce5d3fb416d9f2497b015678eb85c0d
> # bad: [547539f08b9e3629ce68479889813e58c8087e70] ALSA: usb-audio: fix control pipe direction
> git bisect bad 547539f08b9e3629ce68479889813e58c8087e70
> # bad: [3509c748e79435d09e730673c8c100b7f0ebc87c] most: usb: hdm_probe: Fix calling put_device() before device initialization
> git bisect bad 3509c748e79435d09e730673c8c100b7f0ebc87c
> # bad: [a6ebcafc2f5ff7f0d1ce0c6dc38ac09a16a56ec0] net: add ndo_fdb_del_bulk
> git bisect bad a6ebcafc2f5ff7f0d1ce0c6dc38ac09a16a56ec0
> # good: [b8a72692aa42b7dcd179a96b90bc2763ac74576a] hfsplus: fix KMSAN uninit-value issue in __hfsplus_ext_cache_extent()
> git bisect good b8a72692aa42b7dcd179a96b90bc2763ac74576a
> # good: [2b42a595863556b394bd702d46f4a9d0d2985aaa] m68k: bitops: Fix find_*_bit() signatures
> git bisect good 2b42a595863556b394bd702d46f4a9d0d2985aaa
> # good: [9d9f7d71d46cff3491a443a3cf452cecf87d51ef] net: rtnetlink: use BIT for flag values
> git bisect good 9d9f7d71d46cff3491a443a3cf452cecf87d51ef
> # bad: [1550f3673972c5cfba714135f8bf26784e6f2b0f] net: rtnetlink: add bulk delete support flag
> git bisect bad 1550f3673972c5cfba714135f8bf26784e6f2b0f
> # good: [c8879afa24169e504f78c9ca43a4d0d7397049eb] net: netlink: add NLM_F_BULK delete request modifier
> git bisect good c8879afa24169e504f78c9ca43a4d0d7397049eb
> # first bad commit: [1550f3673972c5cfba714135f8bf26784e6f2b0f] net: rtnetlink: add bulk delete support flag
> 
> Is there anything else I can do to help?

Is there soemthing missing?

Roland I think it would be helpful if you can test as well more recent
stable series versions to confirm if the issue is present there as
well or not, which might indicate a 5.10.y specific backporting
problem.

#regzbot introduced: 1550f3673972c5cfba714135f8bf26784e6f2b0f

Regards,
Salvatore

Re: [regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Thorsten Leemhuis]

On 12/18/25 20:50, Salvatore Bonaccorso wrote:
> 
> Is there soemthing missing?
> 
> Roland I think it would be helpful if you can test as well more recent
> stable series versions to confirm if the issue is present there as
> well or not, which might indicate a 5.10.y specific backporting
> problem.

FWIW, it (as usual) would be very important to know if this happens with
mainline as well, as that determines if it's a general problem or a
backporting problem -- and the outcome, strictly speaking, determines if
the developers of the change in question or the stable team are
responsible. If we are lucky that distinction does not matter, but if we
are unlucky neither camp might look into this as of now. I wrote more
about this at length here:
https://linux-regtracking.leemhuis.info/post/frequent-reasons-why-linux-kernel-bug-reports-are-ignored/

But yes, given that 5.10.y is quite old, it might be easier to test more
recent stable series first. But if they show the same problem, please
test mainline -- and ideally try if reverting is able to resolve the
problem.

Ciao, Thorsten

Re: [regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Roland Schwarzkopf]

Hi all,

On 12/18/25 20:50, Salvatore Bonaccorso wrote:
> Hi all,
>
> Roland Schwarzkopf reported to the Debian mailing list a problem which
> he encountered once updating in Debian from 5.10.244 to 5.10.247. The
> report is quoted below and found in
> https://lists.debian.org/debian-kernel/2025/12/msg00223.html
>
> Roland did bisect the changes between 5.10.244 to 5.10.247 and found
> that the issue is introduced with 1550f3673972 ("net: rtnetlink: add
> bulk delete support flag") which is the backport to the 5.10.y series.
>
> On Thu, Dec 18, 2025 at 02:59:55PM +0100, Roland Schwarzkopf wrote:
>> Hi Salvatore,
>>
>> On 12/17/25 20:28, Salvatore Bonaccorso wrote:
>>> Hi Roland,
>>>
>>> I'm CC'ing Ben Hutchings directly as well as he takes care of the
>>> Debian LTS kernel updates. Idellly we make this as well a proper bug
>>> for easier tracking.
>>>
>>> On Wed, Dec 17, 2025 at 01:35:54PM +0100, Roland Schwarzkopf wrote:
>>>> Hi there,
>>>>
>>>> after upgrading to the latest kernel on Debian 11
>>>> (linux-image-5.10.0-37-amd64) I have an issue using libvirt with qemu/kvm
>>>> virtual machines and macvtap networking. When a machine is shut down,
>>>> libvirt can not delete the corresponding macvtap device. Thus, starting the
>>>> machine again is not possible. After manually removing the macvtap device
>>>> using `ip link delete` the vm can be started again.
>>>>
>>>> In the journal the following message is shown:
>>>>
>>>> Dec 17 13:19:27 iblis libvirtd[535]: error destroying network device macvtap0: Operation not supported
>>>>
>>>> After downgrading the kernel to linux-image-5.10.0-36-amd64, the problem
>>>> disappears. I tested this on a fresh minimal install of Debian 11 - to
>>>> exclude that anything else on my production machines is causing this issue.
>>>>
>>>> Since the older kernel does not have this issue, I assume this is related to
>>>> the kernel and not to libvirt?
>>>>
>>>> I tried to check for bug reports of the kernel package, but the bug tracker
>>>> finds no reports and even states that the package does not exist (I used the
>>>> "Bug reports" link on
>>>> https://packages.debian.org/bullseye/linux-image-5.10.0-37-amd64). This left
>>>> me a bit puzzled. Since I don't have experience with the debian bug
>>>> reporting process, I had no other idea than writing to this list.
>>> You would need to search for inhttps://bugs.debian.org/src:linux ,
>>> but that said I'm not aware of any bug reports in that direction.
>>>
>>> Would you be in the position of bisecting the problem as you can say
>>> that 5.10.244 is good and 5.10.247 is bad and regressed? If you can do
>>> that that would involve compiling a couple of kernels to narrow down
>>> where the problem is introduced:
>>>
>>>       git clone --single-branch -b linux-5.10.yhttps://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git
>>>       cd linux-stable
>>>       git checkout v5.10.244
>>>       cp /boot/config-$(uname -r) .config
>>>       yes '' | make localmodconfig
>>>       make savedefconfig
>>>       mv defconfig arch/x86/configs/my_defconfig
>>>
>>>       # test 5.10.244 to ensure this is "good"
>>>       make my_defconfig
>>>       make -j $(nproc) bindeb-pkg
>>>       ... install the resulting .deb package and confirm it successfully boots / problem does not exist
>>>
>>>       # test 5.10.247 to ensure this is "bad"
>>>       git checkout v5.10.247
>>>       make my_defconfig
>>>       make -j $(nproc) bindeb-pkg
>>>       ... install the resulting .deb package and confirm it fails to boot / problem exists
>>>
>>> With that confirmed, the bisection can start:
>>>
>>>       git bisect start
>>>       git bisect good v5.10.244
>>>       git bisect bad v5.10.247
>>>
>>> In each bisection step git checks out a state between the oldest
>>> known-bad and the newest known-good commit. In each step test using:
>>>
>>>       make my_defconfig
>>>       make -j $(nproc) bindeb-pkg
>>>       ... install, try to boot / verify if problem exists
>>>
>>> and if the problem is hit run:
>>>
>>>       git bisect bad
>>>
>>> and if the problem doesn't trigger run:
>>>
>>>       git bisect good
>>>
>>> . Please pay attention to always select the just built kernel for
>>> booting, it won't always be the default kernel picked up by grub.
>>>
>>> Iterate until git announces to have identified the first bad commit.
>>>
>>> Then provide the output of
>>>
>>>       git bisect log
>>>
>>> In the course of the bisection you might have to uninstall previous
>>> kernels again to not exhaust the disk space in /boot. Also in the end
>>> uninstall all self-built kernels again.
>> I just did my first bisection \o/ (sorry)
>>
>> Here are the results:
>>
>> git bisect start
>> # bad: [f964b940099f9982d723d4c77988d4b0dda9c165] Linux 5.10.247
>> git bisect bad f964b940099f9982d723d4c77988d4b0dda9c165
>> # good: [863b76df7d1e327979946a2d3893479c3275bfa4] Linux 5.10.244
>> git bisect good f52ee6ea810273e527a5d319e5f400be8c8424c1
>> # good: [dc9fdb7586b90e33c766eac52b6f3d1c9ec365a1] net: usb: lan78xx: Add error handling to lan78xx_init_mac_address
>> git bisect good dc9fdb7586b90e33c766eac52b6f3d1c9ec365a1
>> # bad: [2272d5757ce5d3fb416d9f2497b015678eb85c0d] phy: cadence: cdns-dphy: Enable lower resolutions in dphy
>> git bisect bad 2272d5757ce5d3fb416d9f2497b015678eb85c0d
>> # bad: [547539f08b9e3629ce68479889813e58c8087e70] ALSA: usb-audio: fix control pipe direction
>> git bisect bad 547539f08b9e3629ce68479889813e58c8087e70
>> # bad: [3509c748e79435d09e730673c8c100b7f0ebc87c] most: usb: hdm_probe: Fix calling put_device() before device initialization
>> git bisect bad 3509c748e79435d09e730673c8c100b7f0ebc87c
>> # bad: [a6ebcafc2f5ff7f0d1ce0c6dc38ac09a16a56ec0] net: add ndo_fdb_del_bulk
>> git bisect bad a6ebcafc2f5ff7f0d1ce0c6dc38ac09a16a56ec0
>> # good: [b8a72692aa42b7dcd179a96b90bc2763ac74576a] hfsplus: fix KMSAN uninit-value issue in __hfsplus_ext_cache_extent()
>> git bisect good b8a72692aa42b7dcd179a96b90bc2763ac74576a
>> # good: [2b42a595863556b394bd702d46f4a9d0d2985aaa] m68k: bitops: Fix find_*_bit() signatures
>> git bisect good 2b42a595863556b394bd702d46f4a9d0d2985aaa
>> # good: [9d9f7d71d46cff3491a443a3cf452cecf87d51ef] net: rtnetlink: use BIT for flag values
>> git bisect good 9d9f7d71d46cff3491a443a3cf452cecf87d51ef
>> # bad: [1550f3673972c5cfba714135f8bf26784e6f2b0f] net: rtnetlink: add bulk delete support flag
>> git bisect bad 1550f3673972c5cfba714135f8bf26784e6f2b0f
>> # good: [c8879afa24169e504f78c9ca43a4d0d7397049eb] net: netlink: add NLM_F_BULK delete request modifier
>> git bisect good c8879afa24169e504f78c9ca43a4d0d7397049eb
>> # first bad commit: [1550f3673972c5cfba714135f8bf26784e6f2b0f] net: rtnetlink: add bulk delete support flag
>>
>> Is there anything else I can do to help?
> Is there soemthing missing?
>
> Roland I think it would be helpful if you can test as well more recent
> stable series versions to confirm if the issue is present there as
> well or not, which might indicate a 5.10.y specific backporting
> problem.

I tested this on some newer kernel versions: stable 6.18.2 and 6.17.13 
as well as mainline 6.19-rc1. With all three kernel versions, a 
different problem occurs: I can't even start the virtual machine.

The relevant journal entry is:

Dec 19 17:25:41 iblis libvirtd[438]: error creating macvtap interface 
macvtap0@eno1 (08:00:27:25:16:0c): Operation not supported

I then searched in the commit log for the upstream commit of the one I 
found when bisecting the issue. It is a6cec0bcd342, which git describe 
says is v5.18-rc1-423-ga6cec0bcd342. Therefore I decided to also test 
the two longterm versions before and after that commit was introduced: 
5.15.197 and 6.1.159. With both kernel versions I found the same 
behavior as with the stable and mainline versions.

So on all newer kernel versions I tested behave identical, but different 
than the latest release in the 5.10.y branch.

I'm not sure if I caused this behaviour by making a mistake during 
building the kernels - I don't have much experience in that area. I used 
the same steps Salvatore gave me for bisecting the issue for the 
different versions, i.e.,

      cp /boot/config-$(uname -r) .config
      yes '' | make localmodconfig
      make savedefconfig
      mv defconfig arch/x86/configs/my_defconfig
      make my_defconfig
      make -j $(nproc) bindeb-pkg

Best regards,

Roland

>
> #regzbot introduced: 1550f3673972c5cfba714135f8bf26784e6f2b0f
>
> Regards,
> Salvatore

-- 
Dr. Roland Schwarzkopf <rschwarzkopf@mathematik.uni-marburg.de>
Dept. of Mathematics and Computer Science, IT Solutions
University of Marburg, Hans-Meerwein-Str. 6, D-35032 Marburg, Germany
Tel: +4964212821523, Fax: +4964212821573

Re: [regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 2036 bytes --]

Hi all,

On Fri, 2025-12-19 at 10:19 +0100, Thorsten Leemhuis wrote:
> On 12/18/25 20:50, Salvatore Bonaccorso wrote:
> > 
> > Is there soemthing missing?
> > 
> > Roland I think it would be helpful if you can test as well more recent
> > stable series versions to confirm if the issue is present there as
> > well or not, which might indicate a 5.10.y specific backporting
> > problem.
> 
> FWIW, it (as usual) would be very important to know if this happens with
> mainline as well, as that determines if it's a general problem or a
> backporting problem
[...]

The bug is this:

- libvirtd wrongly used to use NLM_F_CREATE (0x400) and NLM_F_EXCL
  (0x200) flags on an RTM_DELLINK operation.  These flags are only
  semantically valid for NEW-type operations.

- rtnetlink is rather lax about checking the flags on operations, so
  these unsupported flags had no effect.

- rtnetlink can now support NLM_F_BULK (0x200) on some DEL-type
  operations.  If the flag is used but is not valid for the specific
  operation then the operation now fails with EOPNOTSUPP.  Since
  NLM_F_EXCL == NLM_F_BULK and RTM_DELLINK does not support bulk
  operations, libvirtd now hits this error case.

I have not tested with mainline, but in principle the same issue should
occur with any other kernel version that has commit a6cec0bcd342 "net:
rtnetlink: add bulk delete support flag" together with an older version
of libvirt.

This was fixed in libvirt commit 1334002340b, which appears to have gone
into version 7.1.0, but Debian 11 "bullseye" has 7.0.0.

We can certainly fix the libvirt side of this in Debian, but this also
sounds like a case where the kernel should work around known buggy user-
space.  On the other hand, this has been upstream for over 3 years so
maybe it doesn't make sense now.

Please let me know whether I (or anyone) should try to implement a
workaround for this in the kernel.

Ben.

-- 
Ben Hutchings - Debian developer, member of kernel, installer and LTS
teams

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Thorsten Leemhuis]

@stable team and/or @net maintainers: this imho needs a judgement call
from your side. See below for details.

On 1/2/26 21:18, Ben Hutchings wrote:
> On Fri, 2025-12-19 at 10:19 +0100, Thorsten Leemhuis wrote:
>> On 12/18/25 20:50, Salvatore Bonaccorso wrote:
>>>
>>> Is there soemthing missing?
>>>
>>> Roland I think it would be helpful if you can test as well more recent
>>> stable series versions to confirm if the issue is present there as
>>> well or not, which might indicate a 5.10.y specific backporting
>>> problem.
>>
>> FWIW, it (as usual) would be very important to know if this happens with
>> mainline as well, as that determines if it's a general problem or a
>> backporting problem
> [...]
> 
> The bug is this:
> 
> - libvirtd wrongly used to use NLM_F_CREATE (0x400) and NLM_F_EXCL
>   (0x200) flags on an RTM_DELLINK operation.  These flags are only
>   semantically valid for NEW-type operations.
> 
> - rtnetlink is rather lax about checking the flags on operations, so
>   these unsupported flags had no effect.
> 
> - rtnetlink can now support NLM_F_BULK (0x200) on some DEL-type
>   operations.  If the flag is used but is not valid for the specific
>   operation then the operation now fails with EOPNOTSUPP.  Since
>   NLM_F_EXCL == NLM_F_BULK and RTM_DELLINK does not support bulk
>   operations, libvirtd now hits this error case.
> 
> I have not tested with mainline, but in principle the same issue should
> occur with any other kernel version that has commit a6cec0bcd342 "net:
> rtnetlink: add bulk delete support flag"

FWIW, merged for v5.19-rc1 and backported to v5.10.246 as 1550f3673972c5
End of October 2025 in parallel with 5b22f62724a0a0 ("net: rtnetlink:
fix module reference count leak issue in rtnetlink_rcv_msg") [v6.0-rc2],
which is a fix for the former.

> together with an older version of libvirt.
> 
> This was fixed in libvirt commit 1334002340b, which appears to have gone
> into version 7.1.0,

Could not find that commit when looking briefly, but that version was
released 2021-03-01.

> but Debian 11 "bullseye" has 7.0.0.
> 
> We can certainly fix the libvirt side of this in Debian, but this also
> sounds like a case where the kernel should work around known buggy user-
> space.  On the other hand, this has been upstream for over 3 years so
> maybe it doesn't make sense now.

Yeah, I tend to the latter as well (the @net maintainers can speak up if
the disagree). But we have one more middle-ground option here maybe the
@stable team could do: revert the backports of 1550f3673972c5 and
5b22f62724a0a0 from 5.10.y, unless they are strongly needed there.

> Please let me know whether I (or anyone) should try to implement a
> workaround for this in the kernel.

Ciao, Thorsten

Re: [regression 5.10.y] Libvirt can no longer delete macvtap devices after backport of a6cec0bcd342 ("net: rtnetlink: add bulk delete support flag") to 5.10.y series (Debian 11)

[Salvatore Bonaccorso]

Hi,

On Mon, Jan 05, 2026 at 01:30:59PM +0100, Thorsten Leemhuis wrote:
> @stable team and/or @net maintainers: this imho needs a judgement call
> from your side. See below for details.
> 
> On 1/2/26 21:18, Ben Hutchings wrote:
> > On Fri, 2025-12-19 at 10:19 +0100, Thorsten Leemhuis wrote:
> >> On 12/18/25 20:50, Salvatore Bonaccorso wrote:
> >>>
> >>> Is there soemthing missing?
> >>>
> >>> Roland I think it would be helpful if you can test as well more recent
> >>> stable series versions to confirm if the issue is present there as
> >>> well or not, which might indicate a 5.10.y specific backporting
> >>> problem.
> >>
> >> FWIW, it (as usual) would be very important to know if this happens with
> >> mainline as well, as that determines if it's a general problem or a
> >> backporting problem
> > [...]
> > 
> > The bug is this:
> > 
> > - libvirtd wrongly used to use NLM_F_CREATE (0x400) and NLM_F_EXCL
> >   (0x200) flags on an RTM_DELLINK operation.  These flags are only
> >   semantically valid for NEW-type operations.
> > 
> > - rtnetlink is rather lax about checking the flags on operations, so
> >   these unsupported flags had no effect.
> > 
> > - rtnetlink can now support NLM_F_BULK (0x200) on some DEL-type
> >   operations.  If the flag is used but is not valid for the specific
> >   operation then the operation now fails with EOPNOTSUPP.  Since
> >   NLM_F_EXCL == NLM_F_BULK and RTM_DELLINK does not support bulk
> >   operations, libvirtd now hits this error case.
> > 
> > I have not tested with mainline, but in principle the same issue should
> > occur with any other kernel version that has commit a6cec0bcd342 "net:
> > rtnetlink: add bulk delete support flag"
> 
> FWIW, merged for v5.19-rc1 and backported to v5.10.246 as 1550f3673972c5
> End of October 2025 in parallel with 5b22f62724a0a0 ("net: rtnetlink:
> fix module reference count leak issue in rtnetlink_rcv_msg") [v6.0-rc2],
> which is a fix for the former.
> 
> > together with an older version of libvirt.
> > 
> > This was fixed in libvirt commit 1334002340b, which appears to have gone
> > into version 7.1.0,
> 
> Could not find that commit when looking briefly, but that version was
> released 2021-03-01.

For reference it is this one I think:
https://gitlab.com/libvirt/libvirt/-/commit/81334002340be6bd3a1a34c6584a85fe25cc049c

Regards,
Salvatore