[PATCH] D56679: [InstCombine] Don't undo 0 - (X * Y) canonicalization when combining subs.

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri added a comment.

We are sure that we want to break the cycle here, and not at the other end?
**This** results in the optimal IR?



================
Comment at: test/Transforms/InstCombine/mul.ll:505-507
+ at X = global i32 5
+
+define i64 @test_mul_canonicalize_neg_is_not_undone(i64 %L1) {
----------------
Can you please add an explanation **here**?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D56679/new/

https://reviews.llvm.org/D56679