[Valgrind-users] impossible occuring during exit--libc-related functions at fault?

[Michael R. <M.R...@F5...>]

I have a program that has its own definitions of many libc functions (and links libc statically).  When it exits, the impossible happens.  The impossible happening looks like this:

==6609== Can't extend stack to 0x55FD780 during signal delivery for thread 1:
==6609==   no stack segment
==6609==
==6609== Process terminating with default action of signal 11 (SIGSEGV): dumping core
==6609==  Access not within mapped region at address 0x55FD780
==6609==    at 0xC5AAFE: mco_obj_pos_2 (mcoobj.c:833)
--6609-- VALGRIND INTERNAL ERROR: Valgrind received a signal 7 (SIGBUS) - exiting
--6609-- si_code=2;  Faulting address: 0x4A05005;  sp: 0x63DE03D8

valgrind: the 'impossible' happened:
   Killed by fatal signal
==6609==    at 0x38015DD5: ??? (m_syscall.c:150)
==6609==    by 0x3804E09F: myvprintf_int64 (m_debuglog.c:596)
==6609==    by 0x0: ???

sched status:
  running_tid=1

Thread 1: status = VgTs_Runnable
==6609==    at 0xC5AAFE: mco_obj_pos_2 (mcoobj.c:833)

This is valgrind 3.3.1 and Linux 2.4.21.  Any ideas?  In particular, I'm wondering whether it is known to be problematic for programs run under valgrind to define many of the libc functions themselves and/or statically link libc?