-
Notifications
You must be signed in to change notification settings - Fork 53
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add aarch64-*-windows-gnu*
support
#65
Conversation
psm/build.rs
Outdated
@@ -34,6 +34,7 @@ fn find_assembly( | |||
Some(("src/arch/aarch_aapcs64.s", false)) | |||
} | |||
} | |||
("aarch64", _, "windows", _) => None, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This fails to include support for informational methods (stack direction, pointer.) https://github.com/rust-lang/stacker/blob/master/psm/src/arch/aarch_aapcs64.s should be contain a suitable implementation of these functions.
Why not generalize the match above to match gnu
too?
("aarch64", _, "windows", _) => {
if masm {
Some(("src/arch/aarch64_armasm.asm", false))
} else {
Some(("src/arch/aarch_aapcs64.s", false))
}
}
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OOTB it using this assembly with Clang fails with:
running: "aarch64-w64-mingw32-clang" "-O3" "-ffunction-sections" "-fdata-sections" "--target=aarch64-pc-windows-gnullvm" "-ffunction-sections" "-fdata-sections" "--target=aarch64-pc-windows-gnullvm" "-xassembler-with-cpp" "-DCFG_TARGET_OS_windows" "-DCFG_TARGET_ARCH_aarch64" "-DCFG_TARGET_ENV_gnu" "-o" "/home/mateusz/Projects/rust/build/x86_64-unknown-linux-gnu/stage0-rustc/aarch64-pc-windows-gnullvm/release/build/psm-e1f284177df4f4bb/out/src/arch/aarch_aapcs64.o" "-c" "src/arch/aarch_aapcs64.s"
cargo:warning=src/arch/aarch_aapcs64.s:24:7: error: expected absolute expression
cargo:warning=.type rust_psm_stack_direction,@function
cargo:warning= ^
cargo:warning=src/arch/aarch_aapcs64.s:31:1: error: unknown directive
cargo:warning=.size rust_psm_stack_direction,.rust_psm_stack_direction_end-rust_psm_stack_direction
cargo:warning=^
cargo:warning=src/arch/aarch_aapcs64.s:37:7: error: expected absolute expression
cargo:warning=.type rust_psm_stack_pointer,@function
cargo:warning= ^
cargo:warning=src/arch/aarch_aapcs64.s:44:1: error: unknown directive
cargo:warning=.size rust_psm_stack_pointer,.rust_psm_stack_pointer_end-rust_psm_stack_pointer
cargo:warning=^
cargo:warning=src/arch/aarch_aapcs64.s:50:7: error: expected absolute expression
cargo:warning=.type rust_psm_replace_stack,@function
cargo:warning= ^
cargo:warning=src/arch/aarch_aapcs64.s:58:1: error: unknown directive
cargo:warning=.size rust_psm_replace_stack,.rust_psm_replace_stack_end-rust_psm_replace_stack
cargo:warning=^
cargo:warning=src/arch/aarch_aapcs64.s:64:7: error: expected absolute expression
cargo:warning=.type rust_psm_on_stack,@function
cargo:warning= ^
cargo:warning=src/arch/aarch_aapcs64.s:84:1: error: unknown directive
cargo:warning=.size rust_psm_on_stack,.rust_psm_on_stack_end-rust_psm_on_stack
cargo:warning=^
Makes me wonder what assembler was used with msvc
toolchain when masm
is unavailable.
Managed to cross compile Rust with this change to aarch_aapcs64.s
:
#define GLOBL(fnname) .globl fnname
#define TYPE(fnname)
#define FUNCTION(fnname) fnname
#define SIZE(fnname,endlabel)
I don't have AArch64 Windows to test it though, should I ask somebody to test Rust with this change for me and update PR if it works?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hm, I wonder if this a limitation of the LLVM's built-in assembler.
I don't have AArch64 Windows to test it though, should I ask somebody to test Rust with this change for me and update PR if it works?
I don't anticipate such a change to affect the behaviour of this library during a “normal” operation, but I could see it having an effect on generation of unwind tables… Since we aren't looking to support actual segmented stacks in psm on windows without Fibers, I think it is fine to make the proposed change (especially because the .type
and .size
directives are applicable for COFF
output anyway)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Makes me wonder what assembler was used with msvc toolchain when masm is unavailable.
This configuration is largely untested ^^. The only non-msvc target actually tested is x86*windows-gnu
.
daea6d0
to
47839e7
Compare
aarch64-*-windows-gnu*
support (without assembly)aarch64-*-windows-gnu*
support
Thanks! |
it still fails with
|
Maybe a regression from f7f1e58? |
Rust PR for the target: rust-lang/rust#94872
First commit disables linking
psm_s
when it's absent since assembly is not compiled.