Subject: Re: pthreads and pmax
To: Simon Burge <email@example.com>
From: Chris Johnson <firstname.lastname@example.org>
Date: 05/24/2001 18:03:59
> Simon Burge <email@example.com>
> Bob Lantz wrote:
> > I seem to recall that problems with pthreads, mySQL, etc. were
> > related to not having ll/sc or an atomic store syscall on non-R4000
> > mips NetBSD systems.
> The problems with the databases are usually TAS problems (ie, ll/sc, or
> lack of ).
> There are some patches for pthreads that mostly work, but threads that
> use floating point fail badly. I'll see if I can dig those patches up
> and have another look.
On our Linux 2.0/MIPS system (QED 52xx based) there were problems
on atomic memory because Linux was not guaranteeing that shared
memory segments were aligned at a large enough boundary to avoid
virtual aliasing on the cache index (16KB boundary).
There was also a problem with the glibc FP varargs implementation
that seemed to manifest most commonly on pthreads apps. I seem to
recall that the issue was a misaligned stack base in the subthreads
interacting with a rounding/masking assumption in varargs.
(The FP bug was pretty entertaining - GDB showed the calculations
were done correctly, but printf() couldn't output the result.)
We pushed both of these fixes to their respective code maintainers,
but I don't know if they were ever folded into updates.
After these fixes, pthreads and mySQL worked fine. YMMV.