maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

My previous commit had an incomplete message, it should have been: 

make the 'fp return in ST(0)' optimization smart enough to
look through token factor nodes.  THis allows us to compile 
testcases like CodeGen/X86/fp-stack-retcopy.ll into:

_carg:
	subl	$12, %esp
	call	L_foo$stub
	fstpl	(%esp)
	fldl	(%esp)
	addl	$12, %esp
	ret

instead of:

_carg:
	subl	$28, %esp
	call	L_foo$stub
	fstpl	16(%esp)
	movsd	16(%esp), %xmm0
	movsd	%xmm0, 8(%esp)
	fldl	8(%esp)
	addl	$28, %esp
	ret

Still not optimal, but much better and this is a trivial patch.  Fixing 
the rest requires invasive surgery that is is not llvm 2.2 material.

llvm-svn: 46054
This commit is contained in:
Chris Lattner 2008-01-16 05:56:59 +00:00
parent ea001f1db7
commit 8f7cec859e
2 changed files with 13 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -767,7 +767,7 @@ SDOperand X86TargetLowering::LowerRET(SDOperand Op, SelectionDAG &DAG) {
// If this is a load into a scalarsse value, don't store the loaded value
// back to the stack, only to reload it: just replace the scalar-sse load.
if (ISD::isNON_EXTLoad(Value.Val) &&
Chain.reachesChainWithoutSideEffects(Value.getOperand(0))) {
Chain.reachesChainWithoutSideEffects(Value.getOperand(0))) {
Chain = Value.getOperand(0);
MemLoc = Value.getOperand(1);
} else {

12
llvm/test/CodeGen/X86/fp-stack-retcopy.ll Normal file
View File

@ -0,0 +1,12 @@
; This should not copy the result of foo into an xmm register.
; RUN: llvm-as < %s | llc -march=x86 -mcpu=yonah -mtriple=i686-apple-darwin9 | not grep xmm
; rdar://5689903
declare double @foo()
define double @carg({ double, double }* byval %z) nounwind {
entry:
%tmp5 = tail call double @foo() nounwind ; <double> [#uses=1]
ret double %tmp5
}