Solaris 2.5.1 and 103640-32 patch
Looks like the only solution for now is to use back
the old libpthread.so.1 before the 103640-32 patch.
Seems to work for now. Thanks to those who've replied :-
"Martin Carpenter" <martincarpenter@hotmail.com>
Saroj Mahapatra <saroj@bear.com>
Michael P Campfield <campfiel@cs.utk.edu>
Mike Whitton <mike@eng.wayne.edu>
Thanks.
------------- BEGIN ORIGINAL POST -------------------
Hi,
I've recently patched my Solaris 2.5.1 on an UE450
with the 2.5.1-Recommended patches as of May 1 2000.
This patch includes the kernel and libpthread patch
103640-32.
The problem now is that programs linked with the
libpthread.so.1 library will now seg fault whenever
it is executed.
I've tried using GCC 2.95.1 and link with the pthread
library and this will cause the Seg Fault to happen.
i.e.
# gcc -o test test.c -lpthread
When I add another library to it (libsocket), the Seg
Fault does not happen!! i.e
# gcc -o test test.c -lpthread -lsocket
When I use Sun Workshop Compiler 5.0, this Seg Fault
does not happen.
Doing a truss on the buggy program will show me that it
Seg Faults at the point lwp_self() is called.i.e.
... <lotsa junk deleted> ..
...
..
lwp_create(0xEFFFF480, 0x0100, 0xEF40FE14) = 2
lwp_create(0xEFFFF480, 0x0100, 0xEF40FE14) = -276894496
lwp_mutex_lock(0xEF7ED760) = 0
sigaction(SIGWAITING, 0xEF666710, 0x00000000) = 0
lwp_self() = 1
Incurred fault #6, FLTBOUNDS %pc = 0xEF7CA154
siginfo: SIGSEGV SEGV_MAPERR addr=0xEF7FE000
Received signal #11, SIGSEGV [caught]
siginfo: SIGSEGV SEGV_MAPERR addr=0xEF7FE000
*** process killed ***
Any ideas on how to solve this problem? Appreciate it alot.
Thanks & Best regards,
John Wong
------------- END ORIGINAL POST -------------------
