Will Howard willz06jw at gmail.com wrote:
#1 How do you know if Valgrind is working properly (AKA everything is patched and working).
#2 How do you decifer the output to start looking for bugs.
Here's some example output (cleaned up a bit by my perl script, http://kegel.com/wine/valgrind/valgrind-split-pl.txt) is at: http://kegel.com/wine/valgrind/logs-2008-03-03/vg-ws2_32_sock.txt
That particular output file says
Syscall param socketcall.sendmsg(msg.msg_iov[i]) points to uninitialised byte(s) at (within /lib/ld-2.6.1.so) by WSASendTo (socket.c:2674) by WS_sendto (socket.c:2827) by test_UDP (sock.c:1055) by func_sock (sock.c:1985) by run_test (test.h:406) by main (test.h:455)
This means there may be a bug somewhere near some of the lines mentioned. Sadly, only filename are given, not directories, so you kind of have to guess whether the files in question are in the directory ws2_32 or ws2_32/tests (or somewhere else).
The thing to do is to read and understand the test with the issue.
The course of action will then be obvious :-)
Did I mention you have to really understand the test? - Dan