MIPS Linux has different values assigned to signals than
X86 or PPC Linux.   The result is that a SIGBUS on MIPS
(value 0xA) is reported as SIGUSR1, since that signal has
the that value on X86.  This causes confusion (obviously).
Telling gdb to handle SIGUSR1 will cause it to respond to
SIGBUS but ignore SIGUSR1 (which has the value 0x16 on MIPS).

I didn't find a bug report for this in Bugzilla.  I
have trouble believing that this problem has gone unnoticed.

The attached patch adds target_signal->internal_signal
translation, and vice versa, for MIPS Linux targets.  I
adapted the code from target_symbol_from_host() and
do_target_symbol_to_host()in gdb/common/signals.c.  Other
than being poor names, these functions and the wrappers
around them seem unnecessarily complicated.  In particular,
I left out EXE_BAD_* translations because they appear to
be darwin-specific and translations for REALTIME_LO/HI
seem to be obsolete.  ChangeLog-2006 mentions removing
these defines.

Which leads me to some questions:
   1 -- Is there a bug report for this somewhere?
   2 -- Is there any reason to implement EXE_BAD_* translations?
   3 -- Is the REALTIME_LO/HI translation obsolete cruft?
   4 -- Do the multiple layers of wrappers around target_
        signal_{to,from}_host in signals.c serve any purpose?

I'm still testing the patch.  I have a core file which now
displays the correct signal, but I have to create a test
program where I can test the translation in the other direction.

Any comments on the patch?

-- 
Michael Eager	 eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306  650-325-8077