projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: d137528) | patch
[X86] combineAddOrSubToADCOrSBB - Fold ADD/SUB + (AND(SRL(X,Y),1) -> ADC/SBB+BT(X...
authorSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 21 Mar 2022 21:37:37 +0000 (21:37 +0000)
committerSimon Pilgrim <llvm-dev@redking.me.uk>
Mon, 21 Mar 2022 21:37:42 +0000 (21:37 +0000)
commit438ac282db97c584daf2d4d1a90e6b3d49ef9189
treefb807fe98529544ede91494280e09dfee423767ftree | snapshot
parentd13752851a4e134708f365a5b238e8ff375accaacommit | diff
[X86] combineAddOrSubToADCOrSBB - Fold ADD/SUB + (AND(SRL(X,Y),1) -> ADC/SBB+BT(X,Y) (REAPPLIED)

As suggested on PR35908, if we are adding/subtracting an extracted bit, attempt to use BT instead to fold the op and use a ADC/SBB op.

Reapply with extra type legality checks - LowerAndToBT was originally only used during lowering, now that it can occur earlier we might encounter illegal types that we can either promote to i32 or just bail.

Differential Revision: https://reviews.llvm.org/D122084
llvm/lib/Target/X86/X86ISelLowering.cpp diff | blob | history
llvm/test/CodeGen/X86/add-sub-bool.ll diff | blob | history
Domain: System / Toolchain;