Serious do_IRQ bug..? Crash....

Chris Evans (chris@ferret.lmh.ox.ac.uk)
Wed, 22 May 1996 18:28:38 +0100 (BST)


I've been pushing my machine for the past day or two, by repeatedly
re-compiling, then `make clean'ing the kernel. This is on a P100 with SCSI
disks (aic7xxx) and 32Mb RAM, so it gets through them at a fair rate.

The kernel is pre2.0.6. With Becker's v0.25 of 3c59x.c driver. With
tagged queueing enabled in aic7xxx.c.

The first time, 108 compiles were managed before the machine died. The
second time, 30ish. I'm currently chasing the possibility of a hardware
fault (again!), and wouldn't post this unless the 2 crashes obtained had
an identical EIP and identical trace... not to mention very similar
register/stack dumps.

So here they are..

May 22 04:22:20 ferret kernel: CPU: 0
May 22 04:22:20 ferret kernel: EIP: 0010:[<0010c543>]
May 22 04:22:20 ferret kernel: EFLAGS: 00010002
May 22 04:22:20 ferret kernel: eax: 5f746573 ebx: 0018213b ecx:
083b735c edx: 00000000
May 22 04:22:20 ferret kernel: esi: 00000200 edi: 00d63fbc ebp:
083b73cc esp: 00d63f98
May 22 04:22:20 ferret kernel: ds: 0018 es: 0018 fs: 002b gs:
002b ss: 0018
May 22 04:22:20 ferret kernel: Process cc1 (pid: 3861, process nr: 46,
stackpage=00d63000)
May 22 04:22:20 ferret kernel: Stack: 00000200 6d656d5f 00d63fbc 00d63fbc
00000000 00000003 0010b2c5 00000200
May 22 04:22:20 ferret kernel: 00d63fbc 00000039 083b735c 083b735c
00000000 00000003 083b73cc 0000000a
May 22 04:22:20 ferret kernel: 0000002b 0000002b 0000002b 0000002b
fffffffe 080c47a0 00000023 00000293
May 22 04:22:20 ferret kernel: Call Trace: [<0010b2c5>]
May 22 04:22:20 ferret kernel: Code: ff d0 8b 5b 14 83 c4 0c 85 db 75 dd
5b 5e 5f c3 90 56 53 8b
May 22 04:22:20 ferret kernel: Aiee, killing interrupt handler

EIP is in 000000000010c50c T do_IRQ
Call trace calls functions: 000000000010b270 t fast_IRQ0_interrupt

The second crash...

May 22 13:43:50 ferret kernel: general protection: 0000
May 22 13:43:50 ferret kernel: CPU: 0
May 22 13:43:50 ferret kernel: EIP: 0010:[<0010c543>]
May 22 13:43:50 ferret kernel: EFLAGS: 00010002
May 22 13:43:50 ferret kernel: eax: 5f746573 ebx: 0018213b ecx:
00000000 edx: 00000000
May 22 13:43:50 ferret kernel: esi: 00000200 edi: 00b7dfbc ebp:
bfffc874 esp: 00b7df98
May 22 13:43:50 ferret kernel: ds: 0018 es: 0018 fs: 002b gs:
002b ss: 0018
May 22 13:43:50 ferret kernel: Process cc1 (pid: 27377, process nr: 38,
stackpage=00b7d000)
May 22 13:43:50 ferret kernel: Stack: 00000200 6d656d5f 00b7dfbc 00b7dfbc
bfffc82c bfffc92c 0010b2c5 00000200
May 22 13:43:50 ferret kernel: 00b7dfbc 40081f64 00000000 00000008
bfffc82c bfffc92c bfffc874 bfffc82c
May 22 13:43:50 ferret kernel: 0000002b 0000002b 0000002b 0000002b
fffffffe 40043cce 00000023 00000202
May 22 13:43:50 ferret kernel: Call Trace: [<0010b2c5>]
May 22 13:43:50 ferret kernel: Code: ff d0 8b 5b 14 83 c4 0c 85 db 75 dd
5b 5e 5f c3 90 56 53 8b
May 22 13:43:50 ferret kernel: Aiee, killing interrupt handler

Same EIP and call trace as above.

Hope this helps. If it looks reminiscent of dodgy hardware despite
similarity of crashes, please reply and say so since I can drag out the 24
hour on-site call out engineer again!!

cheers,
Chris.