Hello!

Attached patch reverts unwanted part of r216820 [1] movqi_internal
change to restore correct type calculation.

2015-04-02  Uros Bizjak  <ubiz...@gmail.com>

    Revert parts of r216820.
    * config/i386/i386.md (movqi_internal): Correct type calculation
    for alternatives 3 and 5.

Bootstrapped and regression tested on x86_64-linux-gnu {,-m32} and
committed to mainline SVN.

[1] https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=216820

Uros.
Index: config/i386/i386.md
===================================================================
--- config/i386/i386.md (revision 221826)
+++ config/i386/i386.md (working copy)
@@ -2533,9 +2533,7 @@
 }
   [(set_attr "isa" "*,*,*,*,*,*,*,*,*,*,avx512dq,avx512dq")
    (set (attr "type")
-     (cond [(eq_attr "alternative" "3,5")
-             (const_string "imovx")
-           (eq_attr "alternative" "7,8,9,10,11")
+     (cond [(eq_attr "alternative" "7,8,9,10,11")
              (const_string "mskmov")
            (and (eq_attr "alternative" "5")
                 (not (match_operand:QI 1 "aligned_operand")))
@@ -2546,6 +2544,8 @@
                 (ior (not (match_test "TARGET_PARTIAL_REG_STALL"))
                      (not (match_test "TARGET_QIMODE_MATH"))))
              (const_string "imov")
+           (eq_attr "alternative" "3,5")
+             (const_string "imovx")
            (and (match_test "TARGET_MOVX")
                 (eq_attr "alternative" "2"))
              (const_string "imovx")

Reply via email to