rust/compiler/rustc_codegen_llvm
Guillaume Gomez 344b6a1668
Rollup merge of #130630 - taiki-e:s390x-clobber-abi, r=Amanieu
Support clobber_abi and vector/access registers (clobber-only) in s390x inline assembly

This supports `clobber_abi` which is one of the requirements of stabilization mentioned in #93335.

This also supports vector registers (as `vreg`) and access registers (as `areg`) as clobber-only, which need to support clobbering of them to implement clobber_abi.

Refs:
- "1.2.1.1. Register Preservation Rules" section in ELF Application Binary Interface s390x Supplement, Version 1.6.1 (lzsabi_s390x.pdf in https://github.com/IBM/s390x-abi/releases/tag/v1.6.1)
- Register definition in LLVM:
  - Vector registers https://github.com/llvm/llvm-project/blob/llvmorg-19.1.0/llvm/lib/Target/SystemZ/SystemZRegisterInfo.td#L249
  - Access registers https://github.com/llvm/llvm-project/blob/llvmorg-19.1.0/llvm/lib/Target/SystemZ/SystemZRegisterInfo.td#L332

I have three questions:
- ~~ELF Application Binary Interface s390x Supplement says that `cc` (condition code, bits 18-19 of PSW) is "Volatile".
  However, we do not have a register class for `cc` and instead mark `cc` as clobbered unless `preserves_flags` is specified (https://github.com/rust-lang/rust/pull/111331).
  Therefore, in the current implementation, if both `preserves_flags` and `clobber_abi` are specified, `cc` is not marked as clobbered. Is this okay? Or even if `preserves_flags` is used, should `cc` be marked as clobbered if `clobber_abi` is used?~~ UPDATE: resolved https://github.com/rust-lang/rust/pull/130630#issuecomment-2367923121
- ~~ELF Application Binary Interface s390x Supplement says that `pm` (program mask, bits 20-23 of PSW) is "Cleared".
  There does not appear to be any registers associated with this in either [LLVM](https://github.com/llvm/llvm-project/blob/llvmorg-19.1.0/llvm/lib/Target/SystemZ/SystemZRegisterInfo.td) or [GCC](33ccc1314d/gcc/config/s390/s390.h (L407-L431)), so at this point I don't see any way other than to just ignore it. Is this okay as-is?~~ UPDATE: resolved https://github.com/rust-lang/rust/pull/130630#issuecomment-2367923121
- Is "areg" a good name for register class name for access registers? It may be a bit confusing between that and `reg_addr`, which uses the “a” constraint (https://github.com/rust-lang/rust/pull/119431)...

Note:

- GCC seems to [recognize only `a0` and `a1`](33ccc1314d/gcc/config/s390/s390.h (L428-L429)), and using `a[2-15]` [causes errors](https://godbolt.org/z/a46vx8jjn).
  Given that cg_gcc has a similar problem with other architecture (https://github.com/rust-lang/rustc_codegen_gcc/issues/485), I don't feel this is a blocker for this PR, but it is worth mentioning here.
- `vreg` should be able to accept `#[repr(simd)]` types as input if the `vector` target feature added in https://github.com/rust-lang/rust/pull/127506 is enabled, but core_arch has no s390x vector type and both `#[repr(simd)]` and `core::simd` are unstable, so I have not implemented it in this PR. EDIT: And supporting it is probably more complex than doing the equivalent on other architectures... https://github.com/rust-lang/rust/pull/88245#issuecomment-905559591

cc `@uweigand`

r? `@Amanieu`

`@rustbot` label +O-SystemZ
2024-10-01 17:32:07 +02:00
..
src Rollup merge of #130630 - taiki-e:s390x-clobber-abi, r=Amanieu 2024-10-01 17:32:07 +02:00
Cargo.toml Update some dependency versions that allow better licensing 2024-08-18 13:59:27 -05:00
messages.ftl Always use ar_archive_writer for import libs 2024-08-17 19:10:46 +00:00
README.md mv compiler to compiler/ 2020-08-30 18:45:07 +03:00

The codegen crate contains the code to convert from MIR into LLVM IR, and then from LLVM IR into machine code. In general it contains code that runs towards the end of the compilation process.

For more information about how codegen works, see the rustc dev guide.