[PATCH 1/3] Make several more BFD globals thread-local

[Tom Tromey]

>>>>> "Tom" == Tom Tromey <tom@tromey.com> writes:

Alan> Oops, sorry.  I didn't see any uses of bfd_print_error in the sources
Alan> and didn't check its history.

Tom> Thanks, I appreciate the update.  I sent the rewrite of gdb's error
Tom> handler to gdb-patches today.

Tom> I also have a patch to handle thread-safety with the per-xvec messages.
Tom> I'm going test that with the bug you mentioned before submitting it.

Quick update on this...

I sent another note about my inability to access the test.

While looking into this, though, I couldn't reproduce the exact race
mentioned in the bug.  However, I still believe it is there.  And, my
testing shows another race, this time between a call to
bfd_cache_close_all on one thread, and bfd_check_format_matches on
another.  I plan to rework my patch to account for this as well, I think
by having bfd_cache_close_all skip any BFDs currently being
format-matched.

Tom