RE: e1000 Invalid truesize

RE: e1000 Invalid truesize

[Ronciak, John]

Please follow what has been happening on netdev.  Dave Miller added this
BUG to 2.6.17-rc as something we (Intel LAD) pointed out to them.  Jeff
Garzik has, as of this morning, queued our e1000 patch for this.  He's
been a little behind.

If you want to know more about the problem, please read the entire
netdev discussion about this.

Cheers,
John
-----------------------------------------------------------
"Those who would give up essential Liberty, to purchase a little
temporary Safety, deserve neither Liberty nor Safety.", Benjamin
Franklin 1755 


> -----Original Message-----
> From: Jens Axboe [mailto:axboe@suse.de] 
> Sent: Wednesday, April 26, 2006 7:25 AM
> To: netdev@vger.kernel.org
> Cc: Ronciak, John; Brandeburg, Jesse; Kirsher, Jeffrey T
> Subject: BUG: e1000 Invalid truesize
> 
> 
> Hi,
> 
> My problematic e1000 card (that sporadically gets transmit 
> timeouts with
> a driver newer than 2.6.14) hits new problems in current git 
> after this
> commit:
> 
>     [NET]: Add skb->truesize assertion checking.
> 
> I thought I'd mention it, since it may help fix the real problem.
> Basically I get tons of:
> 
> SKB BUG: Invalid truesize (592) len=712, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=448, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=712, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=448, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=500, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=500, sizeof(sk_buff)=208
> SKB BUG: Invalid truesize (592) len=500, sizeof(sk_buff)=208
> [...]
> 
> and some net connections don't work well. Putting the 2.6.14 
> driver back
> into current git makes things work smoothly, as always.
> 
> (Please be sure to cc me, I'm not on netdev)
> 
> -- 
> Jens Axboe
> 

Re: e1000 Invalid truesize

[Jens Axboe]

On Wed, Apr 26 2006, Ronciak, John wrote:
> Please follow what has been happening on netdev.  Dave Miller added this
> BUG to 2.6.17-rc as something we (Intel LAD) pointed out to them.  Jeff
> Garzik has, as of this morning, queued our e1000 patch for this.  He's
> been a little behind.

Ah ok, I guess I should have searched there first. I was just grasping
at straws, hoping perhaps some new debug aid exposed my e1000 problem
and it finally could get fixed.

> If you want to know more about the problem, please read the entire
> netdev discussion about this.

Thanks.

-- 
Jens Axboe