On Fri, Jul 29, 2011 at 02:00:16PM +0100, Andrew Haley wrote:

On 07/29/2011 01:50 PM, Andrew Haley wrote:

...

gcj still doesn't work because, for some reason I don't yet understand, backtraces are broken. When backtrace is called from a thread, it wanders off the end of the stack and segfaults. This doesn't happen with F13.

When I find out what's broken it, I'll let you know.

To compile the example, use

gcc thread.c -lpthread -fexceptions

FYI, this is extremely unlikely to be a gcc bug, given that trunk gcc works perfectly well on F13. It's probably a change to glibc.

From my understanding there is an issue with building glibc on ARM. The

following errors are given when linking a .so with -static-libgcc:

/usr/lib/gcc/armv5tel-redhat-linux-gnueabi/4.5.1/libgcc_eh.a(unwind-arm.o): In function `__gnu_Unwind_Backtrace': (.text+0xf28): undefined reference to `__stack_chk_guard'

No idea if this is related, but I think that the unwind-arm.o is the part that is responsible for generating backtraces. I'm not familiar with the internals of glibc nor gcc, so don't trust me too much on this ;)

For this issue with building glibc on ARM, I have opend Bug 726495: - https://bugzilla.redhat.com/show_bug.cgi?id=726495

HTH, Niels