RFA: ia64 portion of libunwind patch
Wed Dec 10 20:58:00 GMT 2003
David Mosberger wrote:
>>>>>>On Thu, 04 Dec 2003 18:57:49 -0500, "J. Johnston" <firstname.lastname@example.org> said:
> Jeff> A questions regarding the .so name issue you mentioned. We
> Jeff> are already grabbing the function names from UNW_OBJ macro
> Jeff> from the generic libunwind.h header. I think we could
> Jeff> generate the libunwind.so name similarly using the UNW_TARGET.
> Jeff> Any problems with this strategy? (any scenarios where this
> Jeff> value doesn't match the extension used by the libunwind
> Jeff> library?)
> No, that sounds fine to me. The part that I don't understand is that
> at the moment it seems that only one libunwind-$TARGET.so can be
> loaded. With a multi-target-capable gdb, that would obviously not be
> sufficient, as you'd want to load, say, libunwind-ia64.so.1 for ia64
> and libunwind-x86.so.1 for x86. But it's mostly a theoretical issue
> at this point.
I have created the patch to use UNW_TARGET.
Yes, the current implementation doesn't support multiple gdb targets, but it is
set up to make such a task relatively straightforward in the future (i.e. simply
put the function handles off of the target vector and pass in the target name).
Kevin, ok to commit?
-- Jeff J.
2003-12-10 Jeff Johnston <email@example.com>
* libunwind-frame.c: If LIBUNWIND_SO not defined, define it to be
(libunwind_target_lib): New static char string.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the Gdb-patches