[llvm] [TLI] Add glibc 2.35 libmvec vector functions to the x86 table (PR #223817)

[via llvm-commits]

Andarwinux wrote:

> @tonuonu thanks for checking it on two machines, and for the controls. I've pushed a commit that takes your first option. It keeps the 46 new `_ZGVb` rows (24 symbols), which are valid on every x86-64 target, and drops the 46 new `_ZGVd` rows until #162239 is fixed. It doesn't touch the `_ZGVd` rows that were already in the table.
> 
> Both autogenerated tests are regenerated. In `replace-with-veclib-libmvec.ll` the `<4 x double>` and `<8 x float>` cases for the new functions are kept, now checking that they stay as intrinsics, so when the rows come back the diff there will show it. On an assertions build, the three changed tests and the other x86 tests that use `-vector-library=LIBMVEC` pass, and your `exp2((double)float)` case at `-march=x86-64` now calls `_ZGVbN2v_exp2`, with 0/64 wrong. I'll update the description to match.

please revert this.

> Given that this extends the wrong-code exposure, could you land the new `_ZGVb` rows first and defer the new `_ZGVd` rows until #162239 is addressed, or fix the target-feature selection before adding them? I'd prefer that over accepting the expansion based on the expectation that most users target AVX2+. Agreed that no separate AVX-512 issue is needed; let's keep #206273 open until the agreed fix lands.

If you targeted non-avx2+, you'll get miscompiled no matter what.

https://github.com/llvm/llvm-project/pull/223817