Update note, with the SROA change, we now produce:

_pairtest:
        movl 8(%esp), %eax
        movl 4(%esp), %ecx
        movd %eax, %xmm0
        movd %ecx, %xmm1
        addss %xmm0, %xmm1
        movl 12(%esp), %eax
        movss %xmm1, (%eax)
        ret

instead of:

_pairtest:
        subl $12, %esp
        movl 20(%esp), %eax
        movl %eax, 4(%esp)
        movl 16(%esp), %eax
        movl %eax, (%esp)
        movss (%esp), %xmm0
        addss 4(%esp), %xmm0
        movl 24(%esp), %eax
        movss %xmm0, (%eax)
        addl $12, %esp
        ret

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@32427 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/README.txt b/lib/Target/X86/README.txt
index 4b9a9dd0299a7fab2017b0f250d6e7347a1c75c1..0667580a44419fa6d9f57fed22a2bf9276ca1c7c 100644 (file)
--- a/lib/Target/X86/README.txt
+++ b/lib/Target/X86/README.txt
@@ -430,16 +430,13 @@ void pairtest(pair P, float *FP) {
 We currently generate this code with llvmgcc4:
 
 _pairtest:
-        subl $12, %esp
-        movl 20(%esp), %eax
-        movl %eax, 4(%esp)
-        movl 16(%esp), %eax
-        movl %eax, (%esp)
-        movss (%esp), %xmm0
-        addss 4(%esp), %xmm0
-        movl 24(%esp), %eax
-        movss %xmm0, (%eax)
-        addl $12, %esp
+        movl 8(%esp), %eax
+        movl 4(%esp), %ecx
+        movd %eax, %xmm0
+        movd %ecx, %xmm1
+        addss %xmm0, %xmm1
+        movl 12(%esp), %eax
+        movss %xmm1, (%eax)
         ret
 
 we should be able to generate:
@@ -455,6 +452,10 @@ integer chunks.  It does this so that structs like {short,short} are passed in
 a single 32-bit integer stack slot.  We should handle the safe cases above much
 nicer, while still handling the hard cases.
 
+While true in general, in this specific case we could do better by promoting
+load int + bitcast to float -> load fload.  This basically needs alignment info,
+the code is already implemented (but disabled) in dag combine).
+
 //===---------------------------------------------------------------------===//
 
 Another instruction selector deficiency: