Memory corruption with recent kernels?

[Jan Grant]

I've had problems with post-August kernels. Under load, builds are
dying. I see lots of these:

Sep 10 17:56:17 tribble /kernel: pid 60715 (make), uid 0: exited on signal 4 (core dumped)
Sep 10 17:56:21 tribble /kernel: pid 61489 (make), uid 0: exited on signal 11 (core dumped)
Sep 10 17:56:29 tribble /kernel: pid 63495 (sh), uid 0: exited on signal 4 (core dumped)
Sep 10 17:56:29 tribble /kernel: pid 63461 (make), uid 0: exited on signal 4 (core dumped)

... for example, this was during a run of portsdb -fUu.

Now, this surely looks like flaky memory to me. Unfortunately, I can
only reproduce this problem with recent kernels (messages above are from
a Sept-10 update); however, I _can_ reliably trigger these problems.
Rolling back to a Jul 29 kernel means these issues no longer show up.

I've had memtest slog away on this machine for 48 hours (not 100% proof,
of course, but some indication that it might be a software flaw).

Kernel's a GENERIC one running on a P4 with 512MB of memory.

Obviously this might just really be a dodgy bit or two that happen to
fall into the wrong spot under new kernels. I'll swap out the memory and
see if the problem re-occurs; I'm reporting here at the same time in
case anyone else is seeing this.

-- 
jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/
Tel +44(0)117 9287088 Fax +44 (0)117 9287112 http://ioctl.org/jan/
YKYBPTMRogueW... you try to move diagonally in vi.