Hi All,

The following patch has been bootstrapped and regtested on powerpc64le-linux.

The existing smul<mode>3_highpart and umul<mode>3_highpart patterns
incorrectly defined the high-part multiply by shifting both operands
before multiplication. This does not match the semantics of the
instructions vmulhs<wd> and vmulhu<wd>, which perform a widened
multiplication and then return the high part.

This patch replaces the incorrect patterns with UNSPEC_VMULHS and
UNSPEC_VMULHU forms, and updates the predicate to use
altivec_register_operand, since these instructions accept only
Altivec registers.

2025-11-19  Jeevitha Palanisamy  <[email protected]>

gcc/
        PR target/122665
        * config/rs6000/vsx.md (UNSPEC_VMULHS, UNSPEC_VMULHU): New.
        (smul<mode>3_highpart): Use UNSPEC_VMULHS.
        (umul<mode>3_highpart): Use UNSPEC_VMULHU.

diff --git a/gcc/config/rs6000/vsx.md b/gcc/config/rs6000/vsx.md
index dd3573b8086..e66462f1c4b 100644
--- a/gcc/config/rs6000/vsx.md
+++ b/gcc/config/rs6000/vsx.md
@@ -362,6 +362,8 @@
    UNSPEC_REPLACE_UN
    UNSPEC_VDIVES
    UNSPEC_VDIVEU
+   UNSPEC_VMULHS
+   UNSPEC_VMULHU
    UNSPEC_VMSUMCUD
    UNSPEC_XXEVAL
    UNSPEC_XXSPLTIW
@@ -6546,25 +6548,19 @@
    (set_attr "size" "<bits>")])
 
 (define_insn "smul<mode>3_highpart"
-  [(set (match_operand:VIlong 0 "vsx_register_operand" "=v")
-       (mult:VIlong (ashiftrt
-                      (match_operand:VIlong 1 "vsx_register_operand" "v")
-                      (const_int 32))
-                    (ashiftrt
-                      (match_operand:VIlong 2 "vsx_register_operand" "v")
-                      (const_int 32))))]
+  [(set (match_operand:VIlong 0 "altivec_register_operand" "=v")
+        (unspec:VIlong [(match_operand:VIlong 1 "altivec_register_operand" "v")
+                       (match_operand:VIlong 2 "altivec_register_operand" "v")]
+                       UNSPEC_VMULHS))]
   "TARGET_POWER10"
   "vmulhs<wd> %0,%1,%2"
   [(set_attr "type" "veccomplex")])
 
 (define_insn "umul<mode>3_highpart"
-  [(set (match_operand:VIlong 0 "vsx_register_operand" "=v")
-       (us_mult:VIlong (ashiftrt
-                         (match_operand:VIlong 1 "vsx_register_operand" "v")
-                         (const_int 32))
-                       (ashiftrt
-                         (match_operand:VIlong 2 "vsx_register_operand" "v")
-                         (const_int 32))))]
+  [(set (match_operand:VIlong 0 "altivec_register_operand" "=v")
+         (unspec:VIlong [(match_operand:VIlong 1 "altivec_register_operand" 
"v")
+                        (match_operand:VIlong 2 "altivec_register_operand" 
"v")]
+                        UNSPEC_VMULHU))]
   "TARGET_POWER10"
   "vmulhu<wd> %0,%1,%2"
   [(set_attr "type" "veccomplex")])

Reply via email to