cvs commit: src/sys/dev/em if_em.c

Pav Lucistnik pav at FreeBSD.org
Tue Aug 22 19:07:01 UTC 2006


Eric Anderson píše v út 22. 08. 2006 v 13:58 -0500:
> On 08/22/06 13:53, Gleb Smirnoff wrote:
> > On Tue, Aug 22, 2006 at 10:35:19AM -0500, Eric Anderson wrote:
> > E> >E> >P> yongari     2006-08-22 02:32:48 UTC
> > E> >E> >P> 
> > E> >E> >P>   FreeBSD src repository
> > E> >E> >P> 
> > E> >E> >P>   Modified files:
> > E> >E> >P>     sys/dev/em           if_em.c 
> > E> >E> >P>   Log:
> > E> >E> >P>   It seems that em(4) misses Tx completion interrupts under certain
> > E> >E> >P>   conditions. The cause of missing Tx completion interrupts comes 
> > E> >from
> > E> >E> >P>   Tx interrupt moderation mechanism(delayed interrupts) or chipset 
> > E> >bug.
> > E> >E> >P>   If Tx interrupt moderation mechanism is the cause of false 
> > E> >watchdog
> > E> >E> >P>   timeout error we should have to fix all device drivers that have 
> > E> >Tx
> > E> >E> >P>   interrupt moderation capability. We may need more investigation
> > E> >E> >P>   for this issue. Anyway, the fix is the same for both cases.
> > E> >E> >P>   
> > E> >E> >P>   This should fix occasional watchdog timeout errors seen on a few
> > E> >E> >P>   systems.
> > E> >E> >P>   
> > E> >E> >P>   Reported by:    -net, Patrick M. Hausen < hausen AT punkt DOT de >
> > E> >E> >P>   Tested by:      Patrick M. Hausen < hausen AT punkt DOT de >
> > E> >E> >
> > E> >E> >This look like a workaround, not a fix the root of the problem. Several
> > E> >E> >people on net said that this problem disappears if debug.mpsafenet=0.
> > E> >E> >So I think there is a problem in FreeBSD or driver, not in chip.
> > E> >E> 
> > E> >E> And it also worked perfectly for a very very long time until 6.x tree. 
> > E> >E> I went from 5-STABLE to 6-STABLE and started seeing it a lot ( a few 
> > E> >E> times per day) on a couple servers.
> > E> >
> > E> >In 5-STABLE the watchdog events were not logged, but the watchdog was
> > E> >processed.
> > E> 
> > E> But my interface never seemed to disappear (I page if the interface goes 
> > E> down), and I'd still notice a link up/down, correct?
> > 
> > In 5-STABLE the interface up/down events were not logged, either.
> > 
> 
> 
> Ok.  Well, I just recalled that on a separate system (my workstation 
> actually), I went from 6.1-R to 6-STABLE as of about 8 days ago, and it 
> now exhibits the problem too, which hasn't happened before.

Yes, there was an MFC of new em driver to 6-STABLE after 6.1-RELEASE.
I was badly hit by it too. But it appears only on some revisions of the
silicon... 

-- 
Pav Lucistnik <pav at oook.cz>
              <pav at FreeBSD.org>

Like 'Do Notte Buye Betamacks.' That was a prediction for 1972.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: Toto je =?UTF-8?Q?digit=C3=A1ln=C4=9B?=
	=?ISO-8859-1?Q?_podepsan=E1?= =?UTF-8?Q?_=C4=8D=C3=A1st?=
	=?ISO-8859-1?Q?_zpr=E1vy?=
Url : http://lists.freebsd.org/pipermail/cvs-src/attachments/20060822/8fa50e6d/attachment.pgp


More information about the cvs-src mailing list