[PATCH] D68857: [X86] Add strict fp support for operations of X87 instructions

Mon Nov 25 20:56:52 PST 2019

craig.topper added a comment.

In D68857#1759727 <https://reviews.llvm.org/D68857#1759727>, @LiuChen3 wrote:

> Rebase.
>  STRICT_FP_ROUND has been set as custom so I remove setting it as legal. When entering the custom processing function, if the operand type is legal,the function behaves the same as setOperationAction(ISD:STRICT_FP_ROUND, TYPE, legal)
>  Modify the test case. Since we do not disable sse, we can pass float-pointer value by register instead of passing by pointer.

STRICT_FP_ROUND should only be Custom on a 64-bit target. This probably only works because LegalizeDAG bailed out of ExpandNode due to strictfp mutation being disabled. And there is no libcall support for STRICT_FP_ROUND so LegalizeDAG gave up and just left the node alone.

================
Comment at: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:5227
+    // X87 instructions has enabled these strict fp operation.
+    if (!Subtarget->hasSSE1() && Subtarget->hasX87())
+      break;
----------------
craig.topper wrote:
> This isn't the right check for the fp80 operations. They're independent of sse.
For FP_ROUND isn't the input type f80 not the result type?

================
Comment at: llvm/lib/Target/X86/X86ISelDAGToDAG.cpp:5235
     // nodes.
     CurDAG->mutateStrictFPToFP(Node);
     break;
----------------
I think this code should also call TLI.isStrictFPEnabled() before doing the mutate.

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D68857/new/

https://reviews.llvm.org/D68857