forked from OSchip/llvm-project
cc88445a91
I looked at canonicalizing in the other direction, but that causes many potential regressions and infinite loops because we already (possibly wrongly) canonicalize "trunc X to i1" into an and+icmp. This has a data layout restriction to avoid creating illegal mask instructions, but we could remove that if we can show that the backend can undo this when needed. The motivating example from issue #56119 is modeled by the PhaseOrdering test. |
||
|---|---|---|
| .. | ||
| AArch64 | ||
| ARM | ||
| X86 | ||
| 2010-03-22-empty-baseclass.ll | ||
| PR6627.ll | ||
| assume-explosion.ll | ||
| basic.ll | ||
| bitfield-bittests.ll | ||
| cmp-logic.ll | ||
| d83507-knowledge-retention-bug.ll | ||
| dce-after-argument-promotion.ll | ||
| deletion-of-loops-that-became-side-effect-free.ll | ||
| expect.ll | ||
| fast-basictest.ll | ||
| fast-reassociate-gvn.ll | ||
| gdce.ll | ||
| globalaa-retained.ll | ||
| inlining-alignment-assumptions.ll | ||
| instcombine-sroa-inttoptr.ll | ||
| lifetime-sanitizer.ll | ||
| loop-rotation-vs-common-code-hoisting.ll | ||
| lto-licm.ll | ||
| min-max-abs-cse.ll | ||
| minmax.ll | ||
| openmp-opt-module.ll | ||
| partialord-ule.ll | ||
| pr32544.ll | ||
| pr36760.ll | ||
| pr39116.ll | ||
| pr39282.ll | ||
| pr40750.ll | ||
| pr44461-br-to-switch-rotate.ll | ||
| pr45682.ll | ||
| pr45687.ll | ||
| reassociate-after-unroll.ll | ||
| reassociate-gvn-bdce.ll | ||
| reassociate-instcombine.ll | ||
| recompute-globalsaa.ll | ||
| rotate.ll | ||
| scev-custom-dl.ll | ||
| scev.ll | ||
| simplifycfg-options.ll | ||
| simplifycfg-switch-lowering-vs-correlatedpropagation.ll | ||
| two-shifts-by-sext.ll | ||
| unsigned-multiply-overflow-check.ll | ||
| vector-select.ll | ||
| vector-trunc-inseltpoison.ll | ||
| vector-trunc.ll | ||