Nikita Popov 2ec1d0f427 [InstCombine] Don't reassociate GEPs for loop invariance ...

Since D146813, LICM will reassociate GEPs to expose hoisting
opportunities itself. Don't perform this transform in InstCombine,
where it is fragile because it depends on an optional LoopInfo
analysis.

2023-04-18 12:17:07 +02:00

..

arm-ieee-vectorize.ll

…

arm-unroll.ll

…

gather-cost.ll

…

gcc-examples.ll

…

interleaved_cost.ll

…

lit.local.cfg

…

mve-gather-scatter-tailpred.ll

…

mve-icmpcost.ll

…

mve-interleaved-cost.ll

…

mve-known-trip-count.ll

…

mve-maskedldst.ll

…

mve-multiexit.ll

…

mve-predstorecost.ll

…

mve-qabs.ll

…

mve-recurrence.ll

…

mve-reduction-predselect.ll

…

mve-reduction-types.ll

…

mve-reductions.ll

[InstCombine] Don't reassociate GEPs for loop invariance

2023-04-18 12:17:07 +02:00

mve-saddsatcost.ll

…

mve-selectandorcost.ll

…

mve-shiftcost.ll

…

mve-vldn.ll

…

pointer_iv.ll

…

prefer-tail-loop-folding.ll

…

scalar-block-cost.ll

…

sphinx.ll

…

tail-fold-multiple-icmps.ll

…

tail-folding-allowed.ll

…

tail-folding-counting-down.ll

…

tail-folding-loop-hint.ll

…

tail-folding-not-allowed.ll

…

tail-folding-prefer-flag.ll

…

tail-folding-reduces-vf.ll

…

tail-folding-reductions-allowed.ll

…

tail-folding-scalar-epilogue-fallback.ll

…

vector_cast.ll

…

width-detect.ll

…