This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Re: V3 ABI stuff is broken
- To: Mark Kettenis <kettenis at wins dot uva dot nl>
- Subject: Re: V3 ABI stuff is broken
- From: Jim Blandy <jimb at zwingli dot cygnus dot com>
- Date: 01 Jun 2001 22:57:14 -0500
- Cc: gdb at sources dot redhat dot com
- References: <200105242031.f4OKVUO07451@delius.kettenis.local>
I think I've fixed this ("PATCH: relax restrictions on
per-architecture data registration"). If this problem isn't gone,
please post.
Mark Kettenis <kettenis@wins.uva.nl> writes:
> The new V3 ABI code in gnu-v3-abi.c uses a per-architecture
> data-pointer, but it looks like this mechanism is broken, at least for
> targets that are not yet multi-arch.
>
> If I understand things correcly in that case startup_gdbarch is used.
> However, the per-architecture data-pointer isn't allocated/initialized
> for startup_gdbarch. This leads to a lot of test failures for
> alpha-unknown-freebsd5.0 when using GCC 3.0.
>
> Should arch-utils.c:initialize_current_architecture() always call
> gdbarch_update_p(), even if the target is not multi-arch?