Description of problem: Linux guest running RH EL 3 Beta 1 crashed with kernel BUG at slab.c:1943! Red Hat Enterprise Linux release 2.9.5AS (Taroon) Kernel 2.4.21-1.1931.2.349.2.2.ent on an s390x linuxsjj login: kernel BUG at slab.c:1943! illegal operation: 0001 CPU: 0 Not tainted autofs ctc fsm af_packet ext3 jbd dasd_fba_mod dasd_eckd_mod dasd_mod Process more (pid: 889, task: 00000000031f0000, ksp: 00000000031f3ac8, f: 0) Krnl PSW : 0400000180000000 0000000000067732 kmem_cache_reap [kernel] 0x562 (2.4.21-1.1931.2.349.2.2.ent) Krnl GPRS: 0000000000000001 00000000031f3ab0 000000000000001b 0000000000000073 0000000000067730 0000000000395150 0000000000000000 0000000000000000 00000000000006f2 000000000000682e 00000000000009a8 0000000000a57b20 0000000000906000 00000000001b27e8 0000000000067730 00000000031f3c58 Krnl ACRS: 00000100 00184920 00000000 00000000 00000001 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 Krnl Code: 00 00 58 50 b0 38 e3 c0 c0 00 00 04 b9 16 00 15 b9 08 00 91 [<00000000000a53fa>] seq_read [kernel] 0x326 (0x31f3da0) [<000000000007c18a>] sys_read [kernel] 0xd2 (0x31f3e50) [<0000000000015722>] sysc_noemu [kernel] 0x10 (0x31f3ef0) Version-Release number of selected component (if applicable): Red Hat Enterprise Linux release 2.9.5AS (Taroon) Kernel 2.4.21-1.1931.2.349.2.2.ent on an s390x How reproducible: I bought up LINUXSJJ earlier and was doing various things. I issued a 'man grep' command and I just hung up (there were no messages on the VM console). I closed the putty session and reopened it, but was unable to reacquire a connection (network issue maybe?) When this happened I rebooted the linux system (IPL 200 CLEAR) and issued some more commands. Below is the last things I did before I hung again. (Note the Segementation fault). I then switched to the VM console and saw those messages. root@linuxsjj proc]# slabinfo -bash: slabinfo: command not found [root@linuxsjj proc]# more slabinfo Segmentation fault [root@linuxsjj proc]# ;s -bash: syntax error near unexpected token `;' [root@linuxsjj proc]# ls 1 370 734 770 bus cpuinfo driver ioports ksyms misc 10 4 743 772 chandev crypto execdomains irq loadavg modules 2 5 747 8 chpids dasd filesystems irq_count locks mounts 23 6 756 892 cio_ignore devices fs kcore mdstat net 3 7 769 9 cmdline dma iomem kmsg meminfo partiti [root@linuxsjj proc]# more slabinfo Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Please try a more recent kernel, available via RHN.
I do not remember how to access the "Sushi" RHN channel precisely, but the TAM should know (for BoA it's Frank).
The reporter will try replicating the problem on Beta 2 if time permits. Thank you
Emil, this is a very disturbing report, although you are the only one who reports this. What is the status? Does it reoccur? If yes, was the traceback the same? Also, add "-c 8" to the KLOGD_OPTIONS in /etc/sysconfig/syslog. This will print more to the console. I suspect something may be missing. The trust of this investigation currently is the implication of seq_file. Running "cat /etc/termcap| strace -o xxx more" shows no access to /proc whatsoever, so I do not see how seq_file can even be involved. This is suggestive. Also, man by default uses less(1), not more(1). Ergo: the error was not caused by "man grep", but instead by someone else who did "more /proc/something". We need Emil to identify what /proc file triggers the fault. Indiscriminate more in /proc is definitely unsupported, but I am willing to look into this case specifically if it can be reproduced with something reasonable, like /proc/PID/maps, or if a regular user can trip it.
Oh wait, disregard that comment (except that "more grep" wasn't at fault indeed). Emil wrote that it's /proc/slabinfo. This happens if a module was removed. Which module?
Closing due to lack of response.