This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch] ppc64: move crt3.o makefile fragments, 64-bit call patterns



On Apr 17, 2006, at 6:39 PM, David Edelsohn wrote:


	I bootstrapped on powerpc64-linux with the patch and it does not
break, but I do not think that the patch is what you intended.

First, I think that you meant for the define_insn names to be
"call_*_sysv<mode>" not "...<P>", at least if you wanted different names
for 32-bit and 64-bit. The names never are referenced, so it does not
matter, but it's just confusing.


	Second, I think that you want to updates the sibcall_*_sysv
patterns as well.

O.K.


Some 64-bit testing of the Apple branch showed a problem in rs6000.c: rs6000_initialize_trampoline(); I replaced a hard-coded SImode with Pmode. While I was there, I removed a local variable named 'pmode' that ... looked odd.

Attachment: gcc.fsf.crt6.diffs.txt
Description: Text document


Bootstrapped & DejaGnu on Darwin/ppc32, no regressions.


2006-04-20 Stuart Hastings <stuart@apple.com>

        * config/rs6000/t-darwin (DARWIN_EXTRA_CRT_BUILD_CFLAGS): New.
        * config/t-darwin (crt3.o): Use it.
        * config/rs6000/rs6000.c (rs6000_initialize_trampoline): Remove
        'pmode', use Pmode instead of SImode for SYMBOL_REF.
        * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv,
        call_nonlocal_sysv, call_value_indirect_nonlocal_sysv,
        call_value_nonlocal_sysv): Suffix with the "P" macro for 32/64
        support.

O.K. for trunk?

stuart hastings
Apple Computer


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]