Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 10 Jun 2011 11:19:11 -0700
From:      "Vogel, Jack" <jack.vogel@intel.com>
To:        Jeremy Chadwick <freebsd@jdc.parodius.com>, Bartosz Stec <bartosz.stec@it4pro.pl>
Cc:        FreeBSD Stable <freebsd-stable@FreeBSD.org>, Andriy Gapon <avg@FreeBSD.org>
Subject:   RE: Panic during kernel booting on HP Proliant DL180G6 and latest STABLE
Message-ID:  <1DB50624F8348F48840F2E2CF6040A9D018ACA75BA@orsmsx508.amr.corp.intel.com>
In-Reply-To: <20110610093728.GA39453@icarus.home.lan>
References:  <4DC6A277.4030801@it4pro.pl> <4DC6E23B.2040207@it4pro.pl> <4DC81E22.5030806@it4pro.pl> <4DCC3844.6070008@it4pro.pl> <4DCD1672.4020405@FreeBSD.org> <4DF1DC26.5010600@it4pro.pl> <20110610093728.GA39453@icarus.home.lan>

next in thread | previous in thread | raw e-mail | index | archive | help
Whoops, ignore the last message, I saw your screen picture.

I have NEVER seen this particular message, I'm wondering if you're getting =
some corrupt bits or something.  Make sure you have a clean build of the ke=
rnel and driver.=20

Jack


-----Original Message-----
From: Jeremy Chadwick [mailto:freebsd@jdc.parodius.com]=20
Sent: Friday, June 10, 2011 2:37 AM
To: Bartosz Stec
Cc: Andriy Gapon; FreeBSD Stable; Vogel, Jack
Subject: Re: Panic during kernel booting on HP Proliant DL180G6 and latest =
STABLE

On Fri, Jun 10, 2011 at 10:56:06AM +0200, Bartosz Stec wrote:
> W dniu 2011-05-13 13:30, Andriy Gapon pisze:
> >on 12/05/2011 22:43 Bartosz Stec said the following:
> >>Allright, if anyone want to follow: http://www.freebsd.org/cgi/query-pr=
.cgi?pr=3D156974
> >>
> >I suspect that your best course here is to add some diagnostic printfs i=
n
> >igb_refresh_mbufs and in m_getzone to see what value is actually passed =
to
> >m_getjcl and why.
> >
> You're probably right, although I would need some guidance how to do
> it. Sorry :(
>=20
> Yesterday I rebuilt this machine with fresh sources to upgrade
> filesystem to ZFSv28 which was MFC'ed lately. After that server
> started to panic regardless of igb loaded as module or integrated
> into kernel.
>=20
> Finally I found that problem was caused by MSIX somehow.
> No more panic after
>=20
>    hw.pci.enable_msix=3D0
>=20
> was included in loader.conf.
>=20
> There's also no panic with GENERIC kernel.
>=20
> Should I really care abou MSIX?

Yes, you absolutely should.  The performance hit, depending on your
interrupt load, can be quite dramatic with this turned off.  Remember:
what you just turned off was MSI-X for the *entire system*, not just for
NICs.  Please read about it here:

http://en.wikipedia.org/wiki/Message_Signaled_Interrupts

CC'ing Jack Vogel here, because your issue pertains to igb(4).  Please
provide output from "pciconv -lvcb" that pertain to your igb NIC(s), as
well as relevant dmesg info that pertains to them (driver, etc.).  You
can XXX out MAC addresses if you're worried.

Jack, original thread starts here:

http://lists.freebsd.org/pipermail/freebsd-stable/2011-May/thread.html#6259=
6

--=20
| Jeremy Chadwick                                   jdc@parodius.com |
| Parodius Networking                       http://www.parodius.com/ |
| UNIX Systems Administrator                   Mountain View, CA, US |
| Making life hard for others since 1977.               PGP 4BD6C0CB |




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1DB50624F8348F48840F2E2CF6040A9D018ACA75BA>