[PATCH] D77448: [PowerPC] Canonicalize shuffles to match more single-instruction masks on LE

[Roland Froese via Phabricator via llvm-commits]

RolandF added inline comments.


================
Comment at: llvm/lib/Target/PowerPC/PPCISelLowering.cpp:9617
+      // the base pointer. This happens with (splat (s_to_v_permuted (ld))).
+      if (LD->getMemoryVT().getSizeInBits() <= (IsFourByte ? 32 : 64))
+        Offset = 0;
----------------
I don't think it is safe to use a load and splat for a smaller than word sized load.  The larger load might now cross a page boundary.


================
Comment at: llvm/lib/Target/PowerPC/PPCISelLowering.cpp:14077
+  EVT VT = OrigSToV.getValueType();
+  SDValue SToVPerm =
+      DAG.getNode(PPCISD::S_TO_V_PERMUTED, dl, VT, OrigSToV.getOperand(0));
----------------
This code is only used on the early, conditional returns, and not on the final return.  The function could be refactored such that the final return case is moved first, or the early returns could return the result of a function that generates the instruction, to prevent unused code.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D77448/new/

https://reviews.llvm.org/D77448