This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: Relocation test fix for target=i686-mingw32 and host=i686-pc-linux
- From: Tom Tromey <tromey at redhat dot com>
- To: ali_anwar <ali_anwar at codesourcery dot com>
- Cc: <gdb-patches at sourceware dot org>
- Date: Wed, 10 Jul 2013 11:58:18 -0600
- Subject: Re: Relocation test fix for target=i686-mingw32 and host=i686-pc-linux
- References: <51DD381D dot 3070708 at codesourcery dot com>
>>>>> "Ali" == ali anwar <ali_anwar@codesourcery.com> writes:
Ali> Attached patch fixes a test case failure for target=i686-mingw32 and
Ali> host=i686-pc-linux.
Ali> Loading file to 0x00 is causing gdb to throw error,"A problem internal
Ali> to GDB has been detected, further debugging may prove unreliable."
Ali> Debugging revealed that call to coff_i386_reloc from
Ali> dwarf2_build_frame_info sets the section_htable to 0x00. Right after,
Ali> we try to get "got = bfd_get_section_by_name (unit->abfd, ".got");"
Ali> and end up with gdb error as hash table is no more valid at this
Ali> point.
Ali> Also please note that issue is not experienced with main, instead of
Ali> hack implemented in the test case's c file i.e. relocate.c.
I think it would be better to fix the crash.
Perhaps a change to the test suite is also needed; but that requires a
justification other than that it lets one work around a crash.
Tom