[PATCH] D64662: [FPEnv] [PowerPC] Lower ppc_fp128 StrictFP Nodes to libcalls

[Andrew J Wock via Phabricator via llvm-commits]

ajwock marked an inline comment as done.
ajwock added inline comments.


================
Comment at: lib/CodeGen/SelectionDAG/LegalizeFloatTypes.cpp:1913
+  case ISD::FP_ROUND:         Res = ExpandFloatOp_FP_ROUND(N); break;
+  case ISD::STRICT_FP_ROUND:  Res = ExpandFloatOp_STRICT_FP_ROUND(N); break;
+  case ISD::FP_TO_SINT:       Res = ExpandFloatOp_FP_TO_SINT(N); break;
----------------
jsji wrote:
> Any reason that we are expanding `STRICT_FP_ROUND ` in `ExpandFloatOperand`, while all the others, especially `STRICT_FP_EXTEND` in `ExpandFloatResult`?
I put STRICT_FP_ROUND here for a rather simple-minded reason- FP_ROUND (non strict) is dealt with here as well.  

I couldn't find any documentation or commits on the reason that ExpandFloatOperand_FP_ROUND was here, but this particular addition was required to get clang with the https://reviews.llvm.org/D43142 forceconstrainedfp pass to successfully compile and pass the GNU Scientific Library's tests that involved the long long data type on a PowerPC server.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D64662/new/

https://reviews.llvm.org/D64662