[PATCH] D154717: [LV] Check if ops can safely be truncated in computeMinimumValueSizes.

[Nikita Popov via Phabricator via llvm-commits]

nikic added inline comments.


================
Comment at: llvm/lib/Analysis/VectorUtils.cpp:678
+              if (I.trunc(MinBW).zext(I.getBitWidth()) == I)
+                return false;
+            }
----------------
Hm, is this special case for constants correct? Consider a variant on the test case that does something like `%l = shl i32 %f, 24`. Then only the low 8 bits of `%f` would be demanded, and `24` can be truncated to 8 bits, but it still changes the result: Now the shift would return poison (while previously the shl + trunc would produce zero).


================
Comment at: llvm/lib/Analysis/VectorUtils.cpp:680
+            }
+            uint64_t BW = bit_width(DB.getDemandedBits(&U).getZExtValue());
+            return bit_ceil(BW) > MinBW;
----------------



Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D154717/new/

https://reviews.llvm.org/D154717