|
From: Jeremy F. <je...@go...> - 2005-03-12 08:22:53
|
The address 0xB1285EB0 is within the Valgrind part of the address space,
which suggests that somehow it leaked into the client state. What's
strange is that you appear to be using --pointercheck=yes (which is the
default, and you don't seem to be turning it off), so I wouldn't expect
to see that particular kind of SIGSEGV (I'd expect to see GPF rather
than bad permissions).
Hm, come to think of it, it is possible that --pointercheck isn't
completely implemented for FP instructions.
If possible, could you grab a copy of /proc/<pid>/maps for that process,
so I can see/guess what's at 0xB1285EB0.
Have you tried other tools; it would be interesting to know if it only
happens under memcheck, or if addrcheck/none also cause it.
Is it possible to isolate this in a smaller piece of code?
J
|