[PATCH] D21395: Fix for Bug 28144

[Mehdi AMINI via llvm-commits]

mehdi_amini added inline comments.

================
Comment at: test/CodeGen/X86/ipra-reg-usage.ll:6
@@ -5,3 +5,3 @@
 define preserve_allcc void @foo()#0 {
-; CHECK: foo Clobbered Registers: EFLAGS YMM0 YMM1 YMM2 YMM3 YMM4 YMM5 YMM6 YMM7 YMM8 YMM9 YMM10 YMM11 YMM12 YMM13 YMM14 YMM15 ZMM0 ZMM1 ZMM2 ZMM3 ZMM4 ZMM5 ZMM6 ZMM7 ZMM8 ZMM9 ZMM10 ZMM11 ZMM12 ZMM13 ZMM14 ZMM15
+; CHECK: foo Clobbered Registers: EFLAGS FS CR10 K2 YMM0 YMM1 YMM2 YMM3 YMM4 YMM5 YMM6 YMM7 YMM8 YMM9 YMM10 YMM11 YMM12 YMM13 YMM14 YMM15 ZMM0 ZMM1 ZMM2 ZMM3 ZMM4 ZMM5 ZMM6 ZMM7 ZMM8 ZMM9 ZMM10 ZMM11 ZMM12 ZMM13 ZMM14 ZMM15
   call void @bar1()
----------------
vivekvpandya wrote:
> mehdi_amini wrote:
> > The list sounds more reasonable now. I think you should have sanitized one-by-one in the first place.
> I am really confused with this result, because locally I added a print method to see which regs are clobbered in mask and I verified the output. But it seems I need to dig more about this this indicates perhaps with 2 loop approach I have missed something 
If you're confused with this result, you shouldn't ask us to commit this patch, but should first make sure we all have a good understanding of what's going on here.
Unfortunately I won't have much time to dig in here.


http://reviews.llvm.org/D21395