[X86] Enable call frame optimization ("mov to push") not only for optsize (PR26325)
The size savings are significant, and from what I can tell, both ICC and GCC do this.
Differential Revision: http://reviews.llvm.org/D18573
llvm-svn: 264966
diff --git a/llvm/test/CodeGen/X86/2006-11-12-CSRetCC.ll b/llvm/test/CodeGen/X86/2006-11-12-CSRetCC.ll
index b6a8fc0..9e1bf9e 100644
--- a/llvm/test/CodeGen/X86/2006-11-12-CSRetCC.ll
+++ b/llvm/test/CodeGen/X86/2006-11-12-CSRetCC.ll
@@ -6,7 +6,14 @@
define i32 @main() {
; CHECK-LABEL: main:
; CHECK-NOT: ret
-; CHECK: subl $4, %{{.*}}
+; CHECK: subl $12, %esp
+; CHECK: pushl
+; CHECK: pushl
+; CHECK: pushl
+; CHECK: pushl
+; CHECK: pushl
+; CHECK: calll cexp
+; CHECK: addl $28, %esp
; CHECK: ret
entry: