"Use slab_build_skb() instead" deprecation warning triggered by tg3

"Use slab_build_skb() instead" deprecation warning triggered by tg3

[Fiona Ebner]

Hi,
we've got a user report about the WARN_ONCE introduced by ce098da1497c
("skbuff: Introduce slab_build_skb()") [0]. The stack trace indicates
that the call comes from the tg3 module. While this is still kernel 6.2
and I can't verify that the issue is still there with newer kernels, I
don't see related changes in drivers/net/ethernet/broadcom/tg3.* after
ce098da1497c, so I thought I should let you know.

[0]: https://forum.proxmox.com/threads/132338/

Best Regards,
Fiona

Re: "Use slab_build_skb() instead" deprecation warning triggered by tg3

[Bagas Sanjaya]

[-- Attachment #1: Type: text/plain, Size: 856 bytes --]

On Fri, Aug 18, 2023 at 10:05:11AM +0200, Fiona Ebner wrote:
> Hi,
> we've got a user report about the WARN_ONCE introduced by ce098da1497c
> ("skbuff: Introduce slab_build_skb()") [0]. The stack trace indicates
> that the call comes from the tg3 module. While this is still kernel 6.2
> and I can't verify that the issue is still there with newer kernels, I
> don't see related changes in drivers/net/ethernet/broadcom/tg3.* after
> ce098da1497c, so I thought I should let you know.
> 

Thanks for the regression report. I'm adding it to regzbot:

#regzbot ^introduced: ce098da1497c6d
#regzbot link: https://forum.proxmox.com/threads/132338/

PS: The proxmox forum link (except full dmesg log pasted there) is in
German, so someone fluent in the language can be helpful here.

-- 
An old man doll... just what I always wanted! - Clara

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: "Use slab_build_skb() instead" deprecation warning triggered by tg3

[Fiona Ebner]

Am 18.08.23 um 13:12 schrieb Bagas Sanjaya:
> On Fri, Aug 18, 2023 at 10:05:11AM +0200, Fiona Ebner wrote:
>> Hi,
>> we've got a user report about the WARN_ONCE introduced by ce098da1497c
>> ("skbuff: Introduce slab_build_skb()") [0]. The stack trace indicates
>> that the call comes from the tg3 module. While this is still kernel 6.2
>> and I can't verify that the issue is still there with newer kernels, I
>> don't see related changes in drivers/net/ethernet/broadcom/tg3.* after
>> ce098da1497c, so I thought I should let you know.
>>
> 
> Thanks for the regression report. I'm adding it to regzbot:
> 
> #regzbot ^introduced: ce098da1497c6d
> #regzbot link: https://forum.proxmox.com/threads/132338/
> 
> PS: The proxmox forum link (except full dmesg log pasted there) is in
> German, so someone fluent in the language can be helpful here.
> 

At the moment, there is not much other text: The user says that they got
the trace after a recent upgrade, but cannot notice any real problems
and I'm telling them that it's a deprecation warning and that I reported
it upstream. If I should ask for some specific information, please let
me know :) But not sure if more information from the user is even
required here or if the fix will just be similar to the one in
8c495270845d ("bnx2x: use the right build_skb() helper").

Best Regards,
Fiona

Re: "Use slab_build_skb() instead" deprecation warning triggered by tg3

[Kees Cook]

On Fri, Aug 18, 2023 at 06:12:33PM +0700, Bagas Sanjaya wrote:
> On Fri, Aug 18, 2023 at 10:05:11AM +0200, Fiona Ebner wrote:
> > Hi,
> > we've got a user report about the WARN_ONCE introduced by ce098da1497c
> > ("skbuff: Introduce slab_build_skb()") [0]. The stack trace indicates
> > that the call comes from the tg3 module. While this is still kernel 6.2
> > and I can't verify that the issue is still there with newer kernels, I
> > don't see related changes in drivers/net/ethernet/broadcom/tg3.* after
> > ce098da1497c, so I thought I should let you know.
> > 
> 
> Thanks for the regression report. I'm adding it to regzbot:
> 
> #regzbot ^introduced: ce098da1497c6d
> #regzbot link: https://forum.proxmox.com/threads/132338/
> 
> PS: The proxmox forum link (except full dmesg log pasted there) is in
> German, so someone fluent in the language can be helpful here.

Since this doesn't cause a behavioral regression, should it be tracked
by regzbot? The WARN is serving as a reminder to the maintainer to
adjust the allocation method.

-- 
Kees Cook