[llvm] [X86] combineINSERT_SUBVECTOR - attempt to widen mask predicate as scalar integer instead of inserting into a zero vXi1 vector (PR #192699)

[Alexander Kornienko via llvm-commits]

alexfh wrote:

> @alinas if you don't have the code repro yet, do you have the command line - target + -march -mcpu -Olevel in particular?

These are the potentially interesting bits of the command line of one of the compilations experiencing this timeout:
```
-cc1 -triple x86_64-grtev4-linux-gnu -O3 -emit-obj -fthinlto-index=... -disable-free -clear-ast-before-backend -disable-llvm-verifier -discard-value-names -function-alignment 5 -mrelocation-model static -fmerge-all-constants -mframe-pointer=all -relaxed-aliasing -fmath-errno -ffp-contract=off -fno-rounding-math -mconstructor-aliases -funwind-tables=2 -mcmodel=medium -mlarge-data-threshold=65536 -target-cpu haswell -target-feature +aes -target-feature +prfchw -tune-cpu haswell -fdebug-info-for-profiling -gno-call-site-info -gno-structor-decl-linkage-names -debug-info-kind=line-tables-only -dwarf-version=5 -debugger-tuning=gdb -mllvm -generate-type-units --compress-debug-sections=zstd -ffunction-sections -fdata-sections -fno-unique-section-names -funique-internal-linkage-names -fsplit-machine-functions -fprofile-sample-use=... -fprofile-remapping-file=... -vectorize-loops -vectorize-slp -mprefer-vector-width=128 -fshow-overloads=best -stats-file=/dev/null -target-feature -allow-light-256-bit -mllvm -disable-auto-upgrade-debug-info -mllvm --enable-detailed-function-properties -mllvm -alloc-token-fallback=9 -mllvm -enable-memprof-context-disambiguation -mllvm -optimize-hot-cold-new -mllvm -optimize-nobuiltin-hot-cold-new-new -mllvm -module-summary-max-indirect-edges=6 -mllvm -optimize-existing-hot-cold-new -mllvm -no-warn-sample-unused=true -mllvm -sample-profile-hot-inline-threshold=75000 -mllvm -enable-fs-discriminator=true -mllvm -disable-ra-fsprofile-loader=false -mllvm -disable-layout-fsprofile-loader=false -mllvm -improved-fs-discriminator=true -mllvm -disable-select-optimize=false -mllvm -pick-merged-source-locations=true -mllvm -salvage-stale-profile=true -mllvm -salvage-stale-profile-max-callsites=2000 -mllvm -enable-ext-tsp-block-placement=true -mllvm -ext-tsp-block-placement-max-blocks=5000 -mllvm -split-static-data=true -mllvm -memprof-annotate-static-data-prefix=true -mllvm -time-passes -fsanitize=alloc-token -fsanitize-annotate-debug-info=alloc-token -fno-sanitize-memory-param-retval -fno-sanitize-address-use-odr-indicator -fsanitize-alloc-token-fast-abi -falloc-token-max=2 -faddrsig -fdwarf2-cfi-asm
```

https://github.com/llvm/llvm-project/pull/192699