[PATCH] [X86][SSE] Vector integer/float conversion memory folding

[Sanjay Patel]

================
Comment at: test/CodeGen/X86/avx1-stack-reload-folding.ll:17-23
@@ +16,9 @@
+
+define <128 x float> @stack_fold_cvtdq2ps(<128 x i32> %a, <128 x i32> %b, <128 x i32> %c, <128 x i32> %d) #0 {
+  %1 = and <128 x i32> %a, %b
+  %2 = and <128 x i32> %c, %d
+  %3 = sitofp <128 x i32> %1 to <128 x float>
+  %4 = sitofp <128 x i32> %2 to <128 x float>
+  %5 = fadd <128 x float> %3, %4
+  ret <128 x float> %5
+
----------------
Is there a guideline regarding IR test cases with types that aren't specified by the ABI? 

Ie, is the lowering of 128-element vectors to legal x86 types stable? Would it be better to define a struct here?

http://reviews.llvm.org/D5981