Hi,
I've been trying out the 2.6.0test*-kernels for some time, and it
would seem that Intel E1000 hangs my machine even on little traffic
(200k/s) and most certainly with 20M/s. Not only with test11bk6, but
also with test11 and I would say test9 too, however I didn't
specifically test out this part as there were some other stability
issues. Test11 seems to be stable except for For..
2.4-series was stable. My experience however is that 2.4.18 was the most
stable 2.4-series kernel ever.
One hang was seen while serving data over http, after around ~600
megabytes of data at ~200k/s. I've been avoiding transferring data out
of the machine as it seems otherwise quite stable..
The machine is set to reboot after oops or panic in 60 seconds; this
did not happen. Nor did the server get any kernel log
messages. Magic-keys did not respond.
My setup is a Dual Athlon MP 1600+ with Tyan Tiger S2466N motherboard,
512M memory. The other 64bit device (in the 64bit PCI slot) is 3ware
7000-series raid-controller, which seems to perform locally nicely
without problems.
My kernel configuration, contents of
/proc/{interrupts,ioports,dma,iomem}, list of modules loaded and the
output of lspci -v are available at:
http://www.modeemi.cs.tut.fi/~flux/e1000/
Kernel is compiled with gcc 2.95.4 (debian):
Linux 2.6.0-test11-bk6-sched-lirc-dyton20 (root@dyton) (gcc 2.95.4 20011002 ) #5 2CPU [dyton]
(the problem appeared without nick's scheduling patch (which btw
increased the resposiveness of the system under medium video capture
load considerably) and without the lirc-patch too)
--
_____________________________________________________________________
/ __// /__ ____ __ http://www.modeemi.fi/~flux/\ \
/ /_ / // // /\ \/ / \ /
/_/ /_/ \___/ /_/\_\@modeemi.fi \/
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
