[llvm] r197520 - Assert that the last operand is actually EFLAGS

Andrew Trick atrick at apple.com
Tue Dec 17 14:56:32 PST 2013


LGTM. Thanks!
-Andy

On Dec 17, 2013, at 12:28 PM, Duncan P. N. Exon Smith <dexonsmith at apple.com> wrote:

> Author: dexonsmith
> Date: Tue Dec 17 14:28:21 2013
> New Revision: 197520
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=197520&view=rev
> Log:
> Assert that the last operand is actually EFLAGS
> 
> This is another follow-up to r197503, after a post-commit review by
> Andy.
> 
> <rdar://problem/15627766>
> 
> Modified:
>    llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> 
> Modified: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86ISelLowering.cpp?rev=197520&r1=197519&r2=197520&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp (original)
> +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp Tue Dec 17 14:28:21 2013
> @@ -15338,6 +15338,12 @@ X86TargetLowering::EmitVAStartSaveXMMReg
>     MBB->addSuccessor(EndMBB);
>   }
> 
> +  // Make sure the last operand is EFLAGS, which gets clobbered by the branch
> +  // that was just emitted, but clearly shouldn't be "saved".
> +  assert((MI->getNumOperands() <= 3 ||
> +          !MI->getOperand(MI->getNumOperands() - 1).isReg() ||
> +          MI->getOperand(MI->getNumOperands() - 1).getReg() == X86::EFLAGS)
> +         && "Expected last argument to be EFLAGS");
>   unsigned MOVOpc = Subtarget->hasFp256() ? X86::VMOVAPSmr : X86::MOVAPSmr;
>   // In the XMM save block, save all the XMM argument registers.
>   for (int i = 3, e = MI->getNumOperands() - 1; i != e; ++i) {
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits




More information about the llvm-commits mailing list