Re: [05/10] watchpoint.exp: Clean up "mips-idt-*" specific re-initialization.

[Pedro Alves <palves at redhat dot com>]

On 01/30/2012 08:14 PM, Stan Shebs wrote:
> On 1/26/12 6:19 AM, Pedro Alves wrote:
>> I don't know if there's still anyone who cares about mips-idt.  This cleans that
>> target's specific re-initialization hacks, by doing them close to where a test
>> starts running the program from scratch.
>>
>> If mips-idt is completely dead, we could instead get rid of all this stuff (it was
>> introduced in 1993), which exists in several tests.
>>
> 
> IDT still claims to be selling the MIPS chips, but it's not so clear about the boards that the mips-idt config targets.  Their site has board manuals dated as recently as 2006.

Thanks.

> Perhaps this would be a good candidate for official deprecation in 7.5 - I certainly wouldn't miss the funkiness that is remote-mips.c ... :-)

Indeed, though I think remote-mips.c still has users (but for mips-elf only).  Kevin brought it back from the dead not so long ago.

This made me go do some archaeology.  I noticed that:

- mips-idt really means mips-idt-ecoff.
- By gdb 4.12, mips-idt-ecoff selected the "idt" target -> config/mips/idt.mt
- By gdb 5.0,  mips-idt-ecoff selected the "embed" target -> config/mips/embed.mt  (which reads: # Target: Big-endian mips board, typically an IDT.)

So the idt target turned into a somewhat more generic embed target.

- In http://sourceware.org/ml/gdb-patches/2004-11/msg00300.html

2004-11-13  Andrew Cagney  <cagney@gnu.org>

	* regcache.h (deprecated_register_valid): Delete.
	(deprecated_registers): Delete.
	* regcache.c (deprecated_register_valid): Delete.
	(deprecated_registers): Delete.
	(deprecated_read_register_byte, _initialize_regcache)
	(deprecated_write_register_bytes, build_regcache): Update.
	* config/powerpc/ppc-sim.mt (TDEPFILES): Remove ppc-bdm.o and
	remote-sds.o.
	* config/powerpc/ppc-eabi.mt (TDEPFILES): Ditto.
	* config/mips/embed.mt (TDEPFILES): Remove remote-mips.o.

Andrew removes remote-mips.o from embed.mt.  This means that the boards in question,
which talk the MIPS remote protocol, have been undebuggeable/untesteable since this
commit, more than 9 years ago.  We currently only support, in configure.tgt:

mips*-*-elf)
        # Target: MIPS ELF
        gdb_target_obs="mips-tdep.o remote-mips.o"
        gdb_sim=../sim/mips/libsim.a
        ;;
mips*-*-*)
        # Target: MIPS
        gdb_target_obs="mips-tdep.o"
        gdb_sim=../sim/mips/libsim.a
        ;;


So I think we're pretty safe to zap any left over mips-idt bits immediately
without going through a deprecation phase.

-- 
Pedro Alves