Always add LC_BUILD_VERSION for metadata object files #114114

keith · 2023-07-27T03:39:14Z

As of Xcode 15 Apple's linker has become a bit more strict about the warnings it produces. One of those new warnings requires all valid Mach-O object files in an archive to have a LC_BUILD_VERSION load command:

ld: warning: no platform load command found in 'ARCHIVE[arm64][2106](lib.rmeta)', assuming: iOS-simulator

This was already being done for Mac Catalyst so this change expands this logic to include it for all Apple platforms. I filed this behavior change as FB12546320 and was told it was the new intentional behavior.

rustbot · 2023-07-27T03:39:22Z

r? @cjgillot

(rustbot has picked a reviewer for you, use r? to override)

rustbot · 2023-07-27T03:39:24Z

These commits modify compiler targets.
(See the Target Tier Policy.)

cjgillot · 2023-07-28T18:17:10Z

I won't have time to review this before a week.
r? compiler

wesleywiser

Changes seem fine to me but I don't have much knowledge of Apple platforms. @thomcc I think you might have more context here. Does this seem ok to you?

BlackHoleFox

I am not the reviewer here, or @thomcc, but I do have some questions/suggestions as someone who has worked in these files a lot.

BlackHoleFox · 2023-08-21T19:30:41Z

compiler/rustc_target/src/spec/apple_base.rs

@@ -274,6 +320,11 @@ fn ios_deployment_target() -> (u32, u32) {
    from_set_deployment_target("IPHONEOS_DEPLOYMENT_TARGET").unwrap_or((7, 0))
 }

+fn mac_catalyst_deployment_target() -> (u32, u32) {
+    // If you are looking for the default deployment target, prefer `rustc --print deployment-target`.
+    from_set_deployment_target("IPHONEOS_DEPLOYMENT_TARGET").unwrap_or((14, 0))


Where did 14.0 come from here? Was it just because that's the current assumed target in the other spec files?

IMO (not the reviewer here) its fine to do this but in some followup the hardcoded versions in the current catalyst target specs should get nuked. This would match what users would expect now that were reading this variable. Definitely doesn't need to be done here though I think.

I pulled it from the original implementation of this behavior that only applied to catalyst https://github.com/rust-lang/rust/pull/114114/files#diff-6bbf6edd930f26d3b180eb5c250b0e48d8f3c5eb474e3274909ef6ae6f0d1e61L343

I agree having it configurable makes sense, at least in this case if you set IPHONEOS_DEPLOYMENT_TARGET it would win though

I missed the lines removed in the diff, oops. That answers the immediate question then.

I agree having it configurable makes sense, at least in this case if you set IPHONEOS_DEPLOYMENT_TARGET it would win though

Do you want to take that to a followup PR or shall I? Fixing the mess that's Catalyst has been on my list forever.

I can send that one after!

ah it looks like using the triple for this case 78bbe57#diff-f2aa7189f140501f7d0d283ef503e1980a9449ef0292b48b650f6e0c87bb8d6b complicates things slightly

iiuc it should just be making another of the <target_here>_lld_platform_version functions to use for this and other LLVM needs and have that source from the deployment target function. Might have missed a complication with the catalyst target though.

Ok sounds good I can try it after this one

It seems like to make this work I would need to change add_pre_link_args to take a String or something else since ours would be dynamically computed here, would that be a reasonable change?

compiler/rustc_target/src/spec/apple_base.rs

thomcc · 2023-08-21T20:30:51Z

@thomcc I think you might have more context here. Does this seem ok to you?

I missed this ping until now, but my opinion would be to defer to @BlackHoleFox's review, which seems quite thorough and accurate.

As of Xcode 15 Apple's linker has become a bit more strict about the warnings it produces. One of those new warnings requires all valid Mach-O object files in an archive to have a LC_BUILD_VERSION load command: ``` ld: warning: no platform load command found in 'ARCHIVE[arm64][2106](lib.rmeta)', assuming: iOS-simulator ``` This was already being done for Mac Catalyst so this change expands this logic to include it for all Apple platforms. I filed this behavior change as FB12546320 and was told it was the new intentional behavior.

BlackHoleFox

@wesleywiser this looks good to r+ from me now.

wesleywiser · 2023-08-29T16:06:18Z

@bors r+ rollup=never

(in case we need to bisect to this PR later)

bors · 2023-08-29T16:06:53Z

📌 Commit 2939e85 has been approved by wesleywiser

It is now in the queue for this repository.

bors · 2023-08-29T21:17:17Z

⌛ Testing commit 2939e85 with merge 84a9f4c...

bors · 2023-08-29T23:02:38Z

☀️ Test successful - checks-actions
Approved by: wesleywiser
Pushing 84a9f4c to master...

keith · 2023-08-29T23:34:35Z

thanks all!

rust-timer · 2023-08-30T06:08:26Z

Finished benchmarking commit (84a9f4c): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.2%	[2.2%, 2.2%]	1
Improvements ✅ (primary)	-0.6%	[-0.7%, -0.5%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.6%	[-0.7%, -0.5%]	2

Cycles

This benchmark run did not return any relevant results for this metric.

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 630.72s -> 630.625s (-0.02%)
Artifact size: 316.01 MiB -> 315.99 MiB (-0.00%)

…r=jieyouxu Apple: Do not specify an SDK version in `rlib` object files This was added in rust-lang#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 rust-lang#129432 for the bigger picture. Tested with (excludes the same few targets as in rust-lang#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](rust-lang#114114 (comment)). `@rustbot` label O-apple

Rollup merge of rust-lang#131016 - madsmtm:no-sdk-version-in-object, r=jieyouxu Apple: Do not specify an SDK version in `rlib` object files This was added in rust-lang#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 rust-lang#129432 for the bigger picture. Tested with (excludes the same few targets as in rust-lang#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](rust-lang#114114 (comment)). `@rustbot` label O-apple

rustbot assigned cjgillot Jul 27, 2023

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Jul 27, 2023

rustbot assigned wesleywiser and unassigned cjgillot Jul 28, 2023

wesleywiser approved these changes Aug 7, 2023

View reviewed changes

BlackHoleFox suggested changes Aug 21, 2023

View reviewed changes

keith force-pushed the ks/always-add-lc_build_version-for-metadata-object-files branch from 8dceddb to d37fdc9 Compare August 21, 2023 20:32

keith added 2 commits August 21, 2023 13:32

Use target.abi instead of string matching llvm_target

f988cbb

Add comment about unused sdk versions

2939e85

BlackHoleFox approved these changes Aug 22, 2023

View reviewed changes

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 29, 2023

bors added the merged-by-bors This PR was explicitly merged by bors. label Aug 29, 2023

bors merged commit 84a9f4c into rust-lang:master Aug 29, 2023
11 checks passed

rustbot added this to the 1.74.0 milestone Aug 29, 2023

keith deleted the ks/always-add-lc_build_version-for-metadata-object-files branch August 29, 2023 23:34

dtolnay mentioned this pull request Oct 27, 2023

Declare rustc_target's dependency on object/macho #117259

Merged

derek-gerstmann mentioned this pull request Feb 16, 2024

[macos] XCode 15 issues warnings for object files when linking halide/Halide#7941

Closed

matthiaskrgr mentioned this pull request Mar 27, 2024

univariant: field #2 comes after unsized field #123134

Closed

madsmtm mentioned this pull request Aug 22, 2024

Apple deployment target, SDK root and SDK version #129432

Open

madsmtm mentioned this pull request Sep 29, 2024

Apple: Do not specify an SDK version in rlib object files #131016

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Always add LC_BUILD_VERSION for metadata object files #114114

Always add LC_BUILD_VERSION for metadata object files #114114

keith commented Jul 27, 2023

rustbot commented Jul 27, 2023

rustbot commented Jul 27, 2023

cjgillot commented Jul 28, 2023

wesleywiser left a comment

BlackHoleFox left a comment

BlackHoleFox Aug 21, 2023

keith Aug 21, 2023

BlackHoleFox Aug 21, 2023 •

edited

Loading

keith Aug 21, 2023

keith Aug 21, 2023

BlackHoleFox Aug 21, 2023 •

edited

Loading

keith Aug 22, 2023

keith Aug 30, 2023

thomcc commented Aug 21, 2023

BlackHoleFox left a comment

wesleywiser commented Aug 29, 2023

bors commented Aug 29, 2023

bors commented Aug 29, 2023

bors commented Aug 29, 2023

keith commented Aug 29, 2023

rust-timer commented Aug 30, 2023

Always add LC_BUILD_VERSION for metadata object files #114114

Always add LC_BUILD_VERSION for metadata object files #114114

Conversation

keith commented Jul 27, 2023

rustbot commented Jul 27, 2023

rustbot commented Jul 27, 2023

cjgillot commented Jul 28, 2023

wesleywiser left a comment

Choose a reason for hiding this comment

BlackHoleFox left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

BlackHoleFox Aug 21, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

BlackHoleFox Aug 21, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

thomcc commented Aug 21, 2023

BlackHoleFox left a comment

Choose a reason for hiding this comment

wesleywiser commented Aug 29, 2023

bors commented Aug 29, 2023

bors commented Aug 29, 2023

bors commented Aug 29, 2023

keith commented Aug 29, 2023

rust-timer commented Aug 30, 2023

Overall result: no relevant changes - no action needed

BlackHoleFox Aug 21, 2023 •

edited

Loading

BlackHoleFox Aug 21, 2023 •

edited

Loading