On Tue, 2009-03-03 at 18:55 -0600, Scott Emery wrote:
> It may be that the default optimization caused the pmcd
> core to be confusing. I have rebuilt pmcd with "-g -O0". I await
> a similar failure to test my hypothesis.
Could be. Also, I should point out that we have many many pmcd's
running here, for many years now, on both i386 and x86_64 and I've
never observed a pmcd core in all that time.
I do, however, know of one other person on an SGI Altix platform
who has had a similar issue (wouldn't let me take a closer look
though, even though they're here in Melbourne, and pretty sure
they never reported it to SGI customer support either).
So, it could be something SGI/SuSE-specific? (I'm on Debian and
RHEL4/5). Compiler version, optimisation level, libc version
... something like that, perhaps, could be root cause here. Ken
had a dazzling array of spectacular failures that were toolchain
related too, in the past, but that was on Ubuntu.
cheers.
--
Nathan
|