amdgcn: Fix subdf3 pattern

This patch fixes a typo in the subdf3 pattern that meant it had a
non-standard name and thus the compiler would emit a libcall rather than
the proper hardware instruction for DFmode subtraction.

Backport from mainline:

2021-01-13  Julian Brown  <julian@codesourcery.com>

gcc/
	* config/gcn/gcn-valu.md (subdf): Rename to...
	(subdf3): This.

(cherry picked from commit abb3993e49c04bd40e42f196f55785cc3fd81682)

diff --git a/gcc/config/gcn/gcn-valu.md b/gcc/config/gcn/gcn-valu.md
index e4d7f2a0f495f937cddf4498173d25e895b93fe5..34a5f35d4c79c937932d52a42e9e8bfc49c61f09 100644 (file)
--- a/gcc/config/gcn/gcn-valu.md
+++ b/gcc/config/gcn/gcn-valu.md
@@ -2185,7 +2185,7 @@
   [(set_attr "type" "vop3a")
    (set_attr "length" "8,8")])
 
-(define_insn "subdf"
+(define_insn "subdf3"
   [(set (match_operand:DF 0 "register_operand"  "=  v,   v")
        (minus:DF
          (match_operand:DF 1 "gcn_alu_operand" "vSvB,   v")