forked from OSchip/llvm-project
7fa503ef4a
And another step towards transformss not introducing inttoptr and/or ptrtoint casts that weren't there already. In this case, when load/store uses have conflicting types, instead of falling back to the iN, we can try to use allocated sub-type. As disscussed, this isn't the best idea overall (we shouldn't rely on allocated type), but it works fine as a temporary measure. I've measured, and @ `-O3` as of vanilla llvm test-suite + RawSpeed, this results in +0.05% more bitcasts, -5.51% less inttoptr and -1.05% less ptrtoint (at the end of middle-end opt pipeline) See https://bugs.llvm.org/show_bug.cgi?id=47592 Reviewed By: efriedma Differential Revision: https://reviews.llvm.org/D88788 |
||
|---|---|---|
| .. | ||
| PR16736.ll | ||
| PR26163.ll | ||
| big-endian-bitfield.ll | ||
| big-endian-dump.ll | ||
| bitfield.ll | ||
| call-site-info-output.ll | ||
| cfi-eof-prologue.mir | ||
| constant-dbgloc.ll | ||
| entry-value-multi-byte-expr.ll | ||
| float-args.ll | ||
| float-stack-arg.ll | ||
| header.ll | ||
| illegal-fragment.ll | ||
| line.test | ||
| lit.local.cfg | ||
| little-endian-dump.ll | ||
| lowerbdgdeclare_vla.ll | ||
| multiple-constant-uses-drops-dbgloc.ll | ||
| partial-subreg.ll | ||
| processes-relocations.ll | ||
| prologue_end.ll | ||
| s-super-register.ll | ||
| salvage-debug-info.ll | ||
| sdag-split-arg.ll | ||
| sdag-split-arg1.ll | ||
| selectiondag-deadcode.ll | ||
| single-constant-use-preserves-dbgloc.ll | ||
| split-complex.ll | ||
| sroa-complex.ll | ||
| tls.ll | ||