This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Avoiding MIPS mthi/mflo and mtlo/mfhi hazards
At Wed, 5 May 2004 08:40:46 +0000 (UTC), "Richard Sandiford" wrote:
> 2004-05-05 Eric Christopher <echristo@redhat.com>
>
> * config/mips/mips.md: Update the madd define_split for new mflo/mfhi
> representation.
I find that even with this, the sb1-elf build chokes building libgcc.
Looks like there's a corresponding msub define_split.
I made the following, and with it sb1-elf builds and
mipsisa64-elf tests same for c/c++ testsuites before/after.
(It's based on Eric's patch for the madd define_split. I could
believe it's correct, I didn't see any obvious differences in the RTL
pattern that mean that it should be different. But I don't know. I
didn't want to check it in until I found out for sure. 8-)
chris
--
2004-05-05 Chris Demetriou <cgd@broadcom.com>
* config/mips/mips.md: Update the msub define_split for new mflo/mfhi
representation.
Index: config/mips/mips.md
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config/mips/mips.md,v
retrieving revision 1.240
diff -u -p -r1.240 mips.md
--- config/mips/mips.md 5 May 2004 08:37:30 -0000 1.240
+++ config/mips/mips.md 5 May 2004 22:39:33 -0000
@@ -1883,7 +1883,7 @@
(clobber (match_dup 4))
(clobber (match_dup 5))
(clobber (match_dup 6))])
- (set (match_dup 0) (match_dup 1))]
+ (set (match_dup 0) (unspec:SI [(match_dup 5) (match_dup 4)] UNSPEC_MFHILO))]
"")
(define_insn "*muls"