-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Wed, Jun 15, 2005 at 04:31:16PM +0200, Michal Trojnara wrote:
Here is the whole backtrace:
...
#29 0x0804b4e0 in alloc_client_session () at client.c:92 #30 0x08054412 in create_client (ls=613270282, s=59548, arg=0x402d5440, cli=0x402d5440 <_IO_2_1_stderr_>) at sthreads.c:133 Previous frame inner to this frame (corrupt stack?)
Interesting, isn't it?
Yes, it seems that the stack is totally corrupted with all those 0x00000000s on it.
So the problem may not be in swapcontext() or ucontext at all. Maybe the stack gets corrupted just before the crash (swapcontext call) or 50 lines before that?
You should try to run the prog via gdb and set some breakpoint nearby (but before) swapcontext call and then step carefully and inspect what happens after each move, especially the contents of ctx->ctx and ready_head->ctx and all their members and sub-members.
....
OK I see in the mailing list that the bug is hunted already. It would be really hard to debug that problem. Thanks to https://listman.redhat.com/archives/phil-list/2003-January/msg00097.html