Define a generic macro for -mno-unaligned-access

[wzssyqa]

Arm32/Arm64/LoongArch/MIPSr6/RISCV may disable unaligned memory access with option
-mno-unaligned-access

Currently Arm32/Arm64 predefine __ARM_FEATURE_UNALIGNED, while LoongArch, RISC-V and MIPSr6 not yet.

Maybe it's time for use to define a more generic predefined macro for this case.

Will __UNALIGN_ACCESS_DISABLED__ be OK?

For GCC, see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111555

[llvmbot]

@llvm/issue-subscribers-clang-driver

Author: YunQiang Su (wzssyqa)

Arm32/Arm64/LoongArch/MIPSr6 may disable unaligned memory access with option `-mno-unaligned-access`

Currently Arm32/Arm64 predefine __ARM_FEATURE_UNALIGNED, while LoongArch and MIPSr6 not yet.

Maybe it's time for use to define a more generic predefined macro for this case.

Will __UNALIGN_ACCESS_DISABLED__ be OK?

For GCC, see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111555

[wzssyqa]

@MaskRay any idea?

[MaskRay]

An architecture may make fine-grained choices for unaligned access (scalar vs vector, etc; RISC-V has recently got cleanup in #73971)). I think this is an area where targets may diverge large enough that a generic macro may not make sense for GCC and Clang.

[MaskRay]

On the GCC side, Richard Biener seems to reject the idea as well: https://gcc.gnu.org/pipermail/gcc-patches/2024-March/647920.html

[wzssyqa]

On the GCC side, Richard Biener seems to reject the idea as well: https://gcc.gnu.org/pipermail/gcc-patches/2024-March/647920.html

Yes. So I closed this ticket.