DragonFly kernel List (threaded) for 2006-12
Re: ifconfig on re0 panics Toshiba Satellite M115-S3094 with RealTek 8101E chipset
On Sun, Dec 10, 2006 at 11:07:08AM -0500, Joe Talbott wrote:
> On Sun, Dec 10, 2006 at 03:23:12PM +0800, Sepherosa Ziehau wrote:
> > On 12/10/06, Joe Talbott <firstname.lastname@example.org> wrote:
> > >I bought a Toshiba Satellite M115-S3094 which has a PCIE NIC that has
> > >a RealTek 8101E chip. When I try to ifconfig the interface to assign
> > >it an IP address my machine panics. I have posted some photos of the
> > >panic message and a db trace. I have also posted a verbose dmesg and
> > >pciconf output here:
> > >
> > >http://www.xenno.com/dfbsd/
> > >
> > >There are two directories 'panic1' and 'panic2'. Panic1 says "RAM
> > >parity error" and suggests hardware failure, yet this is the second
> > >laptop (exact same model) that I've tried and the panic persists
> > >(though I don't have pictures from the first laptop). I am running
> > >1.7.0-DEVELOPMENT from December 8, 2006.
> > Can you get a coredump, and post the back trace? That will be helpful.
> I was unable to get a coredump. Entering 'panic' at the db> prompt
> did not provide me with a core dump neither did 'call dodump'.
> After syncing with today's HEAD I don't get the panic, though the
> re(4) interface comes up I cannot pass data on the connection.
> I have posted the latest dmesg, pciconf, ifconfig and a tcpdump from
> the laptop as it tries to ping my desktop using the re(4) interface.
> The MAC addresses appear to be shifted the same as the first laptop as
> I reported on IRC.
> > >
> > >There are other issues as well that may or may not be related. I have
> > >a USB ethernet adaptor (aue) that auto negotiates to 10baseT/UTP on
> > >FreeBSD 6.1-RELEASE to negotiates to 100baseTX. The laptop has
> > mmm, I didn't quite understand here, do you mean that aue(4) will
> > operate at 10UTP on dfly?
> Yes, on dfly aue(4) operates at 10UTP on freebsd it operates at 100TX.
> I can force the interface to 100baseTX full-duplex and it works but
> slightly slower than 10UTP.
Possibly related to this is that if I delete my pkgsrc tree and 'cvs
up' it again I use up about 250 mbufs according to netstat -m. If I
immediately run 'cvs up' again on the freshly updated tree I start to
see my mbufs climb quite quickly. Which I believe lead to this error
output before my laptop crashed.
Dec 10 14:43:35 neptune kernel: aue0: no memory for rx list -- packet