I have some very strange warnings that I just can't make heads or tails of. I
have been running a suite of apps through valgrind over the last few weeks, and
it has been very very helpful, except these last few warnings have me perplexed.
All of the apps use the same networking library, and the warnings are all in
calls to that library. The first strange thing is that only 2 of the apps are
getting the warnings, but they all use the library the same way.
The second strange thing is that the warning itself doesn't make sense to me
because it can't possibly be true...
==11124==
==11124== Conditional jump or move depends on uninitialised value(s)
==11124== at 0x406508E: ReadConditionEventProcess (process.c:1272)
==11124== by 0x40608F9: ConnReadToCondition (connio.c:2472)
==11124== by 0x80524C7: HandleConnection (dmc.c:2041)
==11124== by 0x40C134A: start_thread (in /lib/libpthread-2.4.so)
==11124== by 0x471A65D: clone (in /lib/libc-2.4.so)
==11124== Uninitialised value was created by a heap allocation
==11124== at 0x4021B38: malloc (vg_replace_malloc.c:207)
==11124== by 0x47FA04B: (within /usr/lib/libcrypto.so.0.9.8)
Below is process.c lines 1271 and 1272. Since line 1272 only has one variable
I am forced o assume that valgrind thinks that it wasn't initialised. The call
directly above it is setting it though. Am I missing something? How can that
possibly be an uninitialised value?
Before you ask this was built with no optimizations.
receiveCount = SocketReceive( ce->socket, receiveStart, ioBufferSpace, &(
ce->timeOut ), &blockedOn, &err );
if( 0 < receiveCount ) {
...
Any clues?
Micah
------------------------------------------------------------------------------
_______________________________________________
Valgrind-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/valgrind-users
