Reproducible nastiness in 2.0.3[2,3,4pre2] SMP

J.D. Bakker (bakker@thorgal.et.tudelft.nl)
Wed, 4 Mar 1998 19:01:42 +0100

Hi all,

None of the latest 2.0 kernels seem to work in SMP on our Quad PPro boxen.
To summarize:

We have
- AMI Goliath with four 200MHz 256k PPros
- KNE100TX on 100Mbps half duplex
- 3.2 GB IDE Quantum Fireball, firmware A0F.0800 with all hdparm options
off except readahead
- simple S3 video card, no X
- user home directories are mounted from another Linux box
- base RedHat 5.0 distribution, updated to all RH fixes as of today
(including glibc 2.0.6);
- upgraded to support 2.1.89-pre5 (this shouldn't make a difference, as
tools like modutils should be backward compatible, right ?)

I tried
- base 2.0.32 as included in RH 5.0
- clean 2.0.33
- 2.0.33 patched with Alan's 2.0.34pre2-patch
- both the 0.79 tulip as included in the kernel as the 0.83 tulip

Symptoms:
- (on all:) Too much work in interrupt
- (on 3[23]:) Aiee: scheduling in interrupt (no keyboard response, fsck all
disks)
- (on 34pre2:) CP: U #3Attempted flush tlb IPI when not AKP (=255)
(followed by:)
- IRQ DEADLOCK DETECTED BY CPU [012] (and locked again...)

2.1.89pre5 works great, even with IO-APIC, but I need stable TCP/IP for my
sims...

Extra data can be provided, of course.

Sincerely,

Jan-Derk Bakker.

--
Jan-Derk Bakker, bakker@mmc.et tudelft.nl


The lazy man's proverb:
    'There's no business like slow business !'






-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu