[llvm-branch-commits] [llvm] [AMDGPU][GlobalISel] Combine (sext (trunc (sext_in_reg x))) (PR #131312)

[Jay Foad via llvm-branch-commits]

================
@@ -258,6 +258,14 @@ def sext_trunc_sextload : GICombineRule<
          [{ return Helper.matchSextTruncSextLoad(*${d}); }]),
   (apply [{ Helper.applySextTruncSextLoad(*${d}); }])>;
 
+def sext_trunc_sextinreg : GICombineRule<
+  (defs root:$dst),
+  (match (G_SEXT_INREG $sir, $src, $width),
+         (G_TRUNC $trunc, $sir),
+         (G_SEXT $dst, $trunc),
+         [{ return (MRI.getType(${trunc}.getReg()).getScalarSizeInBits() >= ${width}.getImm()); }]),
----------------
jayfoad wrote:

That seems like a weird design. Surely all the checking of conditions should be done in the "match" part? I didn't even know that an "apply" function was allowed to fail.

https://github.com/llvm/llvm-project/pull/131312