Hi,
Sorry, no subject last time!
I've just seen the following assertions pop out of one of my servers
running 2.6.16.9 with grsecurity. I've searched the archives of LKML
and netdev and I've only found posts relating to 2.6.9, after which
some related bugs were fixed... It looks like these bugs are related
to e1000, which is the driver I'm using. The system was running 24
days before these appeared and it's still running absolutely fine.
May 16 09:15:12 baldrick kernel: [6442250.504000] KERNEL: assertion (!
sk->sk_forward_alloc) failed at net/core/stream.c (283)
May 16 09:15:12 baldrick kernel: [6442250.513000] KERNEL: assertion (!
sk->sk_forward_alloc) failed at net/ipv4/af_inet.c (150)
baldrick bootc # ethtool -k eth0
Offload parameters for eth0:
rx-checksumming: on
tx-checksumming: on
scatter-gather: on
tcp segmentation offload: on
Many thanks,
Chris
PS: I'm not subscribed to netdev.
--
Chris Boot
[email protected]
http://www.bootc.net/
On Tue, May 16, 2006 at 11:38:38AM +0100, Chris Boot ([email protected]) wrote:
> May 16 09:15:12 baldrick kernel: [6442250.504000] KERNEL: assertion (!
> sk->sk_forward_alloc) failed at net/core/stream.c (283)
> May 16 09:15:12 baldrick kernel: [6442250.513000] KERNEL: assertion (!
> sk->sk_forward_alloc) failed at net/ipv4/af_inet.c (150)
>
> tcp segmentation offload: on
This bug was fixed in .10 in stack and .12 in e1000 driver
(probably unrelated to your problem, if you do not have packet plit
option enabled).
> Many thanks,
> Chris
--
Evgeniy Polyakov
On 16 May 2006, at 11:51, Evgeniy Polyakov wrote:
> On Tue, May 16, 2006 at 11:38:38AM +0100, Chris Boot
> ([email protected]) wrote:
>> May 16 09:15:12 baldrick kernel: [6442250.504000] KERNEL:
>> assertion (!
>> sk->sk_forward_alloc) failed at net/core/stream.c (283)
>> May 16 09:15:12 baldrick kernel: [6442250.513000] KERNEL:
>> assertion (!
>> sk->sk_forward_alloc) failed at net/ipv4/af_inet.c (150)
>>
>> tcp segmentation offload: on
>
> This bug was fixed in .10 in stack and .12 in e1000 driver
> (probably unrelated to your problem, if you do not have packet plit
> option enabled).
Aha, right. Well, I've got an upgrade to .16 ready to go and
scheduled for Friday. What's interesting is that I never saw this
problem come up before, and I've gone through each major interation
of the kernel since 2.6.9...
Cheers,
Chris
--
Chris Boot
[email protected]
http://www.bootc.net/
Hi Chris,
here are some steps to narrow it down.
1. try the latest kernel first (2.6.16.16). This BUG should be fixed there.
2. try without grsecurity patch
3. if it still persists:
Please provide more information about your setup before
submitting a bug.
lspci -vvv and your .config would be the minimum
some ethtool outputs (ethtool -k ethX) would help here.
4. If you like to have it resolved very fast, please try git-bisect.
This can take a lot of time (several recompiles and reboots needed!).
Happy Bug hunting!
Regards
Ingo Oeser