[2.1.129] scheduling problem

Marcus Meissner (msmeissn@stud.informatik.uni-erlangen.de)
Tue, 24 Nov 1998 21:58:20 +0100 (MET)


In article <19981123224753.A12524@revue.linux-kernel-kahuna.sdsu.edu>,
Sang Woo Kang <skang@kahuna.sdsu.edu> wrote:
...
>Problem 2:
>With linux-2.1.128 & 129, when I try to execute any program at a random point,
>the system almost become dead. Here's a chronological sequence of events:

I do experience the same problem. Today I had two near hangs.

Running at this time were X, some xterms, less, one make. A usual situation
which has worked for most of the last 100 2.1 developer kernels.

(No changed hardware. No changed configuration, except CONFIG_FB_MATROX
enabled (I have a Matrox Mystique). But the computer was displaying
the XF86_SVGA/Matrox Server, like it did in the last months with older
versions. No changed compiler, no changed libc, no changed tools.)

- The machine was up for half an hour.
- Most applications were still usable. (Mouse moved, Consoles switched,
xterm scrolled, xcpuinfo was still running).
Harddisks were still accessed by applications (including syslog, sync,
emergenyc umount, so no SCSI hang.)
- Applications did not terminate. I terminated the less, it cleared the xterm
but did not return to the shellprompt. Dito the other applications.

- xcpuinfo Load Meter slowly growed to 100% System Time.

- SysRq-showPc showed nearly everytime an EIP within 'schedule()'.

I suspect the scheduler is broken in some cases.
I will try to debug further, but this is rather difficult.

Marcus

-- 
   <URL:http://wwwcip.informatik.uni-erlangen.de/~msmeissn/>

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/