Bodo Stroesser wrote:
stub_segv_handler misses the "push ebp" at the beginning. As you
do the normally corresponding "pop eax" explicitly, I think stack
pointer is wrong on call of sigreturn.
I have no idea, what makes happen this. Maybe it depends on compiler
version?
I hope, the attached patch fixes the problem. The patch is tested in
my 2.6.12-rc4 + skas0, where I didn't see the problem. It still works
fine for me.
Rob, could you please test whether the patch fixes the problem for you?
Bodo
#
# From: Bodo Stroesser <[EMAIL PROTECTED]>
#
# It's wrong to pop a fixed number of words from stack before
# calling sigreturn, as the number depends on what code is generated
# by the compiler for the start of stub_segv_handler().
# What we need is esp containing the address of sigcontext. So we
# explicitly load that pointer into esp.
#
# Signed-off-by: Bodo Stroesser <[EMAIL PROTECTED]>
Index: linux-2.6.12-rc4/arch/um/sys-i386/stub_segv.c
===================================================================
--- linux-2.6.12-rc4.orig/arch/um/sys-i386/stub_segv.c 2005-07-14 13:15:53.364847980 +0200
+++ linux-2.6.12-rc4/arch/um/sys-i386/stub_segv.c 2005-07-14 13:36:02.980002269 +0200
@@ -22,10 +22,10 @@ stub_segv_handler(int sig)
__asm__("movl %0, %%eax ; int $0x80": : "g" (__NR_getpid));
__asm__("movl %%eax, %%ebx ; movl %0, %%eax ; movl %1, %%ecx ;"
"int $0x80": : "g" (__NR_kill), "g" (SIGUSR1));
- /* Pop the frame pointer and return address since we need to leave
+ /* Load pointer to sigcontext into esp, since we need to leave
* the stack in its original form when we do the sigreturn here, by
* hand.
*/
- __asm__("popl %%eax ; popl %%eax ; popl %%eax ; movl %0, %%eax ; "
- "int $0x80" : : "g" (__NR_sigreturn));
+ __asm__("mov %0,%%esp ; movl %1, %%eax ; "
+ "int $0x80" : : "a" (sc), "g" (__NR_sigreturn));
}