[PATCH] D24154: [DAGCombiner] Fix incorrect sinking of a truncate into the operand of a shift.

[Andrea Di Biagio via llvm-commits]

andreadb added a comment.

In https://reviews.llvm.org/D24154#532078, @arsenm wrote:

> LGTM, but can you add a scalar case with the original test I added for this which hits the new limit of the full scalar bit width behavior instead of / 2


Thanks for the quick review.
Sure, I will add extra scalar test to reduce-trunc-shl.ll.
I can add tests that are similar to those that you have added in test AMDGPU/shift-i64-opts.ll. I will also add tests for the case where the shift count is equal to the scala size in bits.

Cheers,
Andrea


https://reviews.llvm.org/D24154