[PATCH] D25991: [PPC] Peephole to remove extra fcmp that checks for NaN

[Ehsan Amiri via llvm-commits]

amehsan added inline comments.


================
Comment at: lib/Target/PowerPC/PPCInstrInfo.cpp:1619
   // comparison with zero.
   if (OpC == PPC::FCMPUS || OpC == PPC::FCMPUD)
+    return optimizeFPCmpInstr(CmpInstr, SrcReg, SrcReg2, Mask, Value, MRI);
----------------
amehsan wrote:
> nemanjai wrote:
> > This may be a naive question since I don't really know under which conditions we emit XSCMPUDP vs. FCMPUD, but can this not be done with XSCMPUDP as well? Or perhaps there is no need for this.
> That's a good question.  The problem is PPCInstrInfo::analyzeCompare() does not return true for FCMPUD so from Peephole opt point of view XSCMPUD is not a compare opcode. Since there might be some finer points that has to be take  into account, I opened a [[ https://llvm.org/bugs/show_bug.cgi?id=30805 | PR for this ]], rather than fixing it here.
Typo in the comment: PPCInstrInfo::analyzeCompare() does not return true for **XSCMPUDP **


https://reviews.llvm.org/D25991