[patch/rfc] Always define PC_IN_CALL_DUMMY et.al.
Andrew Cagney
ac131313@redhat.com
Tue Nov 26 17:07:00 GMT 2002
> Hello,
>
> The attached tries to ensure that all targets always specify PC_IN_CALL_DUMMY, USE_GENERIC_DUMMY_FRAMES and CALL_DUMMY_LOCATION.
>
> This is pretty straight forward, just define them in the non multi-arch, non generic dummy frame (hp/pa, mn10200, z8k). The only tricky one is the sparc (which is a bit of a mess) a but I'm pretty sure I've got it right.
>
> I'll look to commit this one in a few days.
>
> (This is a first step in eliminating some of the generic dummy frame methods. Namely PC_IN_CALL_DUMMY and USE_GENERIC_DUMMY_FRAMES).
>
> (joel, note the hp/pa code)
> Andrew
I've checked this in. Also tested on a Solaris 2.7 machine.
Before/after test results were pretty bad:
=== gdb Summary ===
# of expected passes 7803
# of unexpected failures 250
# of unexpected successes 2
# of expected failures 65
# of unresolved testcases 18
# of untested testcases 3
# of unsupported tests 2
but identical.
Andrew
> 2002-11-24 Andrew Cagney <ac131313@redhat.com>
>
> * config/h8500/tm-h8500.h (CALL_DUMMY_LOCATION): Define as ON_STACK.
> (USE_GENERIC_DUMMY_FRAMES): Define as zero.
> (PC_IN_CALL_DUMMY): Define as pc_in_call_dummy_on_stack.
> * config/mn10200/tm-mn10200.h (PC_IN_CALL_DUMMY): Define as
> pc_in_call_dummy_at_entry_point.
> * config/pa/tm-hppa.h (CALL_DUMMY_LOCATION): Define as ON_STACK.
> (USE_GENERIC_DUMMY_FRAMES): Define as zero.
> (PC_IN_CALL_DUMMY): Define as pc_in_call_dummy_on_stack.
> * config/pa/tm-hppa64.h (CALL_DUMMY_LOCATION): Delete undefine.
> * config/sparc/tm-sparc.h (PC_IN_CALL_DUMMY): Define as
> pc_in_call_dummy_on_stack.
> * config/sparc/tm-sp64.h (PC_IN_CALL_DUMMY): Redefine as
> pc_in_call_dummy_at_entry_point and pc_in_call_dummy_on_stack.
> * config/z8k/tm-z8k.h (CALL_DUMMY_LOCATION): Define as ON_STACK.
> (USE_GENERIC_DUMMY_FRAMES): Define as zero.
> (PC_IN_CALL_DUMMY): Defile as pc_in_call_dummy_on_stack.
More information about the Gdb-patches
mailing list