This is the mail archive of the 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] MIPS: Optimize sync_compare_and_swap for const zero.

Richard Sandiford wrote:
David Daney <> writes:
This patch fixes a missed optimization problem with the sync_compare_and_swap insns I recently added.

For a constant zero compare value we can directly compare against $0, instead of loading zero into a register for the compare. This typically saves one instruction and reduces register usage.

Tested on mipsel-linux-gnu with bootstrap of all default languages yielding no regressions.

OK to commit?

2007-10-02 David Daney <>

* config/mips/ (sync_compare_and_swap<mode>): Handle compare
against constant zero.
* config/mips/mips.h (MIPS_COMPARE_AND_SWAP): Handle constant zero

Silly minor nit, but please use dJ rather than Jd for consistency with
other patterns.  OK with that change; no need to retest.

Thanks, this is what I committed:

Index: config/mips/
--- config/mips/	(revision 128962)
+++ config/mips/	(working copy)
@@ -4329,7 +4329,7 @@ (define_insn "sync_compare_and_swap<mode
   [(set (match_operand:GPR 0 "register_operand" "=&d,&d")
 	(match_operand:GPR 1 "memory_operand" "+R,R"))
    (set (match_dup 1)
-	(unspec_volatile:GPR [(match_operand:GPR 2 "register_operand" "d,d")
+	(unspec_volatile:GPR [(match_operand:GPR 2 "reg_or_0_operand" "dJ,dJ")
 			      (match_operand:GPR 3 "arith_operand" "I,d")]
Index: config/mips/mips.h
--- config/mips/mips.h	(revision 128962)
+++ config/mips/mips.h	(working copy)
@@ -2959,7 +2959,7 @@ while (0)
   "%(%<%[%|sync\n"				\
   "1:\tll" SUFFIX "\t%0,%1\n"			\
-  "\tbne\t%0,%2,2f\n"				\
+  "\tbne\t%0,%z2,2f\n"				\
   "\t" OP "\t%@,%3\n"				\
   "\tsc" SUFFIX "\t%@,%1\n"			\
   "\tbeq\t%@,%.,1b\n"				\

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