bjoernager/rust
bjoernager/rust
1
Fork 0
Code Issues Pull requests Activity

Rollup merge of #131016 - madsmtm:no-sdk-version-in-object, r=jieyouxu

Browse source 
Apple: Do not specify an SDK version in `rlib` object files

This was added in https://github.com/rust-lang/rust/pull/114114, but is unnecessary, since it ends up being overwritten when linking anyhow, and it feels wrong to embed some arbitrary SDK version in here. The object files produced by LLVM also do not set this, and the tooling shows `n/a` when it's `0`, so it seems to genuinely be optional in object files.

I've also added a test for the different places the SDK version shows up, and documented a bit more in the code how SDK versions work.

See https://github.com/rust-lang/rust/issues/129432 for the bigger picture.

Tested with (excludes the same few targets as in https://github.com/rust-lang/rust/pull/130435):
```console
./x test tests/run-make/apple-sdk-version --target aarch64-apple-darwin,aarch64-apple-ios,aarch64-apple-ios-macabi,aarch64-apple-ios-sim,aarch64-apple-tvos,aarch64-apple-tvos-sim,aarch64-apple-visionos,aarch64-apple-visionos-sim,aarch64-apple-watchos,aarch64-apple-watchos-sim,arm64_32-apple-watchos,armv7k-apple-watchos,armv7s-apple-ios,x86_64-apple-darwin,x86_64-apple-ios,x86_64-apple-ios-macabi,x86_64-apple-tvos,x86_64-apple-watchos-sim,x86_64h-apple-darwin
IPHONEOS_DEPLOYMENT_TARGET=10.0 ./x test tests/run-make/apple-sdk-version --target=i386-apple-ios
```

CC `@BlackHoleFox,` you [originally commented on these values](https://github.com/rust-lang/rust/pull/114114#discussion_r1300599445).

`@rustbot` label O-apple
This commit is contained in:
Matthias Krüger 2024-10-02 17:10:43 +02:00 • committed by GitHub
commit 5fd6218d72
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
6 changed files with 145 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

48
compiler/rustc_codegen_ssa/src/back/link.rs
View file

@ -2959,11 +2959,12 @@ pub(crate) fn are_upstream_rust_objects_already_included(sess: &Session) -> bool
}
}
/// We need to communicate four things to the linker on Apple/Darwin targets:
/// We need to communicate five things to the linker on Apple/Darwin targets:
/// - The architecture.
/// - The operating system (and that it's an Apple platform).
/// - The deployment target.
/// - The environment / ABI.
/// - The deployment target.
/// - The SDK version.
fn add_apple_link_args(cmd: &mut dyn Linker, sess: &Session, flavor: LinkerFlavor) {
if !sess.target.is_like_osx {
return;
@ -3039,7 +3040,38 @@ fn add_apple_link_args(cmd: &mut dyn Linker, sess: &Session, flavor: LinkerFlavo
let (major, minor, patch) = current_apple_deployment_target(&sess.target);
let min_version = format!("{major}.{minor}.{patch}");
// Lie about the SDK version, we don't know it here
// The SDK version is used at runtime when compiling with a newer SDK / version of Xcode:
// - By dyld to give extra warnings and errors, see e.g.:
// <https://github.com/apple-oss-distributions/dyld/blob/dyld-1165.3/common/MachOFile.cpp#L3029>
// <https://github.com/apple-oss-distributions/dyld/blob/dyld-1165.3/common/MachOFile.cpp#L3738-L3857>
// - By system frameworks to change certain behaviour. For example, the default value of
// `-[NSView wantsBestResolutionOpenGLSurface]` is `YES` when the SDK version is >= 10.15.
// <https://developer.apple.com/documentation/appkit/nsview/1414938-wantsbestresolutionopenglsurface?language=objc>
//
// We do not currently know the actual SDK version though, so we have a few options:
// 1. Use the minimum version supported by rustc.
// 2. Use the same as the deployment target.
// 3. Use an arbitary recent version.
// 4. Omit the version.
//
// The first option is too low / too conservative, and means that users will not get the
// same behaviour from a binary compiled with rustc as with one compiled by clang.
//
// The second option is similarly conservative, and also wrong since if the user specified a
// higher deployment target than the SDK they're compiling/linking with, the runtime might
// make invalid assumptions about the capabilities of the binary.
//
// The third option requires that `rustc` is periodically kept up to date with Apple's SDK
// version, and is also wrong for similar reasons as above.
//
// The fourth option is bad because while `ld`, `otool`, `vtool` and such understand it to
// mean "absent" or `n/a`, dyld doesn't actually understand it, and will end up interpreting
// it as 0.0, which is again too low/conservative.
//
// Currently, we lie about the SDK version, and choose the second option.
//
// FIXME(madsmtm): Parse the SDK version from the SDK root instead.
// <https://github.com/rust-lang/rust/issues/129432>
let sdk_version = &*min_version;
// From the man page for ld64 (`man ld`):
@ -3053,11 +3085,13 @@ fn add_apple_link_args(cmd: &mut dyn Linker, sess: &Session, flavor: LinkerFlavo
cmd.link_args(&["-platform_version", platform_name, &*min_version, sdk_version]);
} else {
// cc == Cc::Yes
//
// We'd _like_ to use `-target` everywhere, since that can uniquely
// communicate all the required details, but that doesn't work on GCC,
// and since we don't know whether the `cc` compiler is Clang, GCC, or
// something else, we fall back to other options that also work on GCC
// when compiling for macOS.
// communicate all the required details except for the SDK version
// (which is read by Clang itself from the SDKROOT), but that doesn't
// work on GCC, and since we don't know whether the `cc` compiler is
// Clang, GCC, or something else, we fall back to other options that
// also work on GCC when compiling for macOS.
//
// Targets other than macOS are ill-supported by GCC (it doesn't even
// support e.g. `-miphoneos-version-min`), so in those cases we can