Re: [patch rfc] Add NUM_REGS pseudo regs to MIPS

[Andrew Cagney <ac131313@redhat.com>]

Sigh.  Something in the backtrace is still amiss.


> #2  0x080ca201 in gdbarch_register_sim_regno (gdbarch=0x82fdc48, reg_nr=90)
>     at ../../combined/gdb/gdbarch.c:4078
> #3  0x08125173 in gdbsim_fetch_register (regno=90) at ../../combined/gdb/remote-sim.c:299
> #4  0x08093e0b in legacy_read_register_gen (regnum=90, 
>     myaddr=0xbffff140 " ñÿ¿HÜ/\bhñÿ¿î\211\f\bHÜ/\bZ")
>     at ../../combined/gdb/regcache.c:727

As far as I can tell, frame_register_unwind doesn't call 
legacy_read_register_gen.

> #5  0x08113891 in frame_register_unwind (frame=0x82e0f40, regnum=90, 
>     optimizedp=0xbffff0e8, lvalp=0xbffff0ec, addrp=0xbffff0f0, realnump=0xbffff014, 
>     bufferp=0xbffff140) at ../../combined/gdb/frame.c:534
> #6  0x080dbd82 in mips_get_saved_register (
>     raw_buffer=0xbffff140 " ñÿ¿HÜ/\bhñÿ¿î\211\f\bHÜ/\bZ", optimizedp=0xbffff0e8, 
>     addrp=0xbffff0f0, frame=0x82e0f40, regnum=90, lvalp=0xbffff0ec)
>     at ../../combined/gdb/mips-tdep.c:5603

I suspect some agressive tail-call optimization by GCC is involved.  Can 
you update your sources (so that we both have the same code) and then 
configure/build gdb using:

	CFLAGS="-O -g" .../configure ....

so that -O2 is disabled (or at least should be).

Andrew