PR target/49780

[pf3gnuchains/gcc-fork.git] / gcc / config / i386 / sync.md

diff --git a/gcc/config/i386/sync.md b/gcc/config/i386/sync.md
index feea161..20378d0 100644 (file)
--- a/gcc/config/i386/sync.md
+++ b/gcc/config/i386/sync.md
@@ -1,5 +1,5 @@
 ;; GCC machine description for i386 synchronization instructions.
-;; Copyright (C) 2005, 2006, 2007, 2008, 2009
+;; Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2011
 ;; Free Software Foundation, Inc.
 ;;
 ;; This file is part of GCC.
@@ -47,7 +47,7 @@
        (unspec:BLK [(match_dup 0)] UNSPEC_MFENCE))
    (clobber (reg:CC FLAGS_REG))]
   "!(TARGET_64BIT || TARGET_SSE2)"
-  "lock{%;| }or{l}\t{$0, (%%esp)|DWORD PTR [esp], 0}"
+  "lock{%;} or{l}\t{$0, (%%esp)|DWORD PTR [esp], 0}"
   [(set_attr "memory" "unknown")])
 
 ;; ??? It would be possible to use cmpxchg8b on pentium for DImode
@@ -115,7 +115,7 @@
             [(match_dup 1) (match_dup 2) (match_dup 3)] UNSPECV_CMPXCHG)
           (match_dup 2)))]
   "TARGET_CMPXCHG"
-  "lock{%;| }cmpxchg{<imodesuffix>}\t{%3, %1|%1, %3}")
+  "lock{%;} cmpxchg{<imodesuffix>}\t{%3, %1|%1, %3}")
 
 (define_insn "sync_double_compare_and_swap<mode>"
   [(set (match_operand:DCASMODE 0 "register_operand" "=A")
@@ -134,7 +134,7 @@
            UNSPECV_CMPXCHG)
           (match_dup 2)))]
   ""
-  "lock{%;| }cmpxchg<doublemodesuffix>b\t%1")
+  "lock{%;} cmpxchg<doublemodesuffix>b\t%1")
 
 ;; Theoretically we'd like to use constraint "r" (any reg) for operand
 ;; 3, but that includes ecx.  If operand 3 and 4 are the same (like when
@@ -162,18 +162,75 @@
            UNSPECV_CMPXCHG)
          (match_dup 2)))]
   "!TARGET_64BIT && TARGET_CMPXCHG8B && flag_pic"
-  "xchg{l}\t%%ebx, %3\;lock{%;| }cmpxchg8b\t%1\;xchg{l}\t%%ebx, %3")
+  "xchg{l}\t%%ebx, %3\;lock{%;} cmpxchg8b\t%1\;xchg{l}\t%%ebx, %3")
 
+;; For operand 2 nonmemory_operand predicate is used instead of
+;; register_operand to allow combiner to better optimize atomic
+;; additions of constants.
 (define_insn "sync_old_add<mode>"
   [(set (match_operand:SWI 0 "register_operand" "=<r>")
        (unspec_volatile:SWI
          [(match_operand:SWI 1 "memory_operand" "+m")] UNSPECV_XCHG))
    (set (match_dup 1)
        (plus:SWI (match_dup 1)
-                 (match_operand:SWI 2 "register_operand" "0")))
+                 (match_operand:SWI 2 "nonmemory_operand" "0")))
    (clobber (reg:CC FLAGS_REG))]
   "TARGET_XADD"
-  "lock{%;| }xadd{<imodesuffix>}\t{%0, %1|%1, %0}")
+  "lock{%;} xadd{<imodesuffix>}\t{%0, %1|%1, %0}")
+
+;; This peephole2 and following insn optimize
+;; __sync_fetch_and_add (x, -N) == N into just lock {add,sub,inc,dec}
+;; followed by testing of flags instead of lock xadd and comparisons.
+(define_peephole2
+  [(set (match_operand:SWI 0 "register_operand" "")
+       (match_operand:SWI 2 "const_int_operand" ""))
+   (parallel [(set (match_dup 0)
+                  (unspec_volatile:SWI
+                    [(match_operand:SWI 1 "memory_operand" "")] UNSPECV_XCHG))
+             (set (match_dup 1)
+                  (plus:SWI (match_dup 1)
+                            (match_dup 0)))
+             (clobber (reg:CC FLAGS_REG))])
+   (set (reg:CCZ FLAGS_REG)
+       (compare:CCZ (match_dup 0)
+                    (match_operand:SWI 3 "const_int_operand" "")))]
+  "peep2_reg_dead_p (3, operands[0])
+   && (unsigned HOST_WIDE_INT) INTVAL (operands[2])
+      == -(unsigned HOST_WIDE_INT) INTVAL (operands[3])
+   && !reg_overlap_mentioned_p (operands[0], operands[1])"
+  [(parallel [(set (reg:CCZ FLAGS_REG)
+                  (compare:CCZ (unspec_volatile:SWI [(match_dup 1)]
+                                                    UNSPECV_XCHG)
+                               (match_dup 3)))
+             (set (match_dup 1)
+                  (plus:SWI (match_dup 1)
+                            (match_dup 2)))])])
+
+(define_insn "*sync_old_add_cmp<mode>"
+  [(set (reg:CCZ FLAGS_REG)
+       (compare:CCZ (unspec_volatile:SWI
+                      [(match_operand:SWI 0 "memory_operand" "+m")]
+                      UNSPECV_XCHG)
+                    (match_operand:SWI 2 "const_int_operand" "i")))
+   (set (match_dup 0)
+       (plus:SWI (match_dup 0)
+                 (match_operand:SWI 1 "const_int_operand" "i")))]
+  "(unsigned HOST_WIDE_INT) INTVAL (operands[1])
+   == -(unsigned HOST_WIDE_INT) INTVAL (operands[2])"
+{
+  if (TARGET_USE_INCDEC)
+    {
+      if (operands[1] == const1_rtx)
+       return "lock{%;} inc{<imodesuffix>}\t%0";
+      if (operands[1] == constm1_rtx)
+       return "lock{%;} dec{<imodesuffix>}\t%0";
+    }
+
+  if (x86_maybe_negate_const_int (&operands[1], <MODE>mode))
+    return "lock{%;} sub{<imodesuffix>}\t{%1, %0|%0, %1}";
+
+  return "lock{%;} add{<imodesuffix>}\t{%1, %0|%0, %1}";
+})
 
 ;; Recall that xchg implicitly sets LOCK#, so adding it again wastes space.
 (define_insn "sync_lock_test_and_set<mode>"
@@ -197,15 +254,15 @@
   if (TARGET_USE_INCDEC)
     {
       if (operands[1] == const1_rtx)
-       return "lock{%;| }inc{<imodesuffix>}\t%0";
+       return "lock{%;} inc{<imodesuffix>}\t%0";
       if (operands[1] == constm1_rtx)
-       return "lock{%;| }dec{<imodesuffix>}\t%0";
+       return "lock{%;} dec{<imodesuffix>}\t%0";
     }
 
   if (x86_maybe_negate_const_int (&operands[1], <MODE>mode))
-    return "lock{%;| }sub{<imodesuffix>}\t{%1, %0|%0, %1}";
+    return "lock{%;} sub{<imodesuffix>}\t{%1, %0|%0, %1}";
 
-  return "lock{%;| }add{<imodesuffix>}\t{%1, %0|%0, %1}";
+  return "lock{%;} add{<imodesuffix>}\t{%1, %0|%0, %1}";
 })
 
 (define_insn "sync_sub<mode>"
@@ -220,12 +277,12 @@
   if (TARGET_USE_INCDEC)
     {
       if (operands[1] == const1_rtx)
-       return "lock{%;| }dec{<imodesuffix>}\t%0";
+       return "lock{%;} dec{<imodesuffix>}\t%0";
       if (operands[1] == constm1_rtx)
-       return "lock{%;| }inc{<imodesuffix>}\t%0";
+       return "lock{%;} inc{<imodesuffix>}\t%0";
     }
 
-  return "lock{%;| }sub{<imodesuffix>}\t{%1, %0|%0, %1}";
+  return "lock{%;} sub{<imodesuffix>}\t{%1, %0|%0, %1}";
 })
 
 (define_insn "sync_<code><mode>"
@@ -236,4 +293,4 @@
          UNSPECV_LOCK))
    (clobber (reg:CC FLAGS_REG))]
   ""
-  "lock{%;| }<logicprefix>{<imodesuffix>}\t{%1, %0|%0, %1}")
+  "lock{%;} <logic>{<imodesuffix>}\t{%1, %0|%0, %1}")