Craig Topper

Does hwprobe report that unaligned accesses are fast on this CPU?

> > > Does hwprobe report that unaligned accesses are fast on this CPU? > > > > > > @cyyself, you seem to be actively maintaining kernel patches for...

> > hwprobe documentation does not say whether the unaligned access check is for scalar or scalar+vector. Based on this result, the implementation is only checking scalar. So I think...

> However, I don't think we do have anything to notify compiler that hw supports misaligned access. -mno-strict-align should tell TTI that misaligned accesses are fine. I'll double check.

Infinite loops are undefined behavior in C++ so LLVM happily deletes them. You can disable this behavior by passing `-fno-finite-loops` to clang. ARM's documentation suggests putting an empty volatile inline...

LLVM requires `#pragma STDC FENV_ACCESS ON` for this to work. CC: @andykaylor

> [@topperc](https://github.com/topperc) Isn't our existing check flexible enough? > > [simde/simde/simde-arch.h](https://github.com/simd-everywhere/simde/blob/9130a9ff26fbc1b68ab64e7c1c1d344f2bdf3ac9/simde/simde-arch.h#L553-L555) > > Lines 553 to 555 in [9130a9f](/simd-everywhere/simde/commit/9130a9ff26fbc1b68ab64e7c1c1d344f2bdf3ac9) > > #if defined(__riscv_zvlsseg) || defined(__riscv_v) > # define SIMDE_ARCH_RISCV_ZVLSSEG 1...