Skip to content

bootstrap: don't LTO C dependencies on aarch64 - #161689

Merged
rust-bors[bot] merged 1 commit into
rust-lang:mainfrom
RalfJung:aarch64-no-lto
Aug 25, 2026
Merged

bootstrap: don't LTO C dependencies on aarch64#161689
rust-bors[bot] merged 1 commit into
rust-lang:mainfrom
RalfJung:aarch64-no-lto

Conversation

@RalfJung

@RalfJung RalfJung commented Aug 24, 2026

Copy link
Copy Markdown
Member

Since #161535 apparently may take a while, let's land something to un-break Miri: we know that on aarch64-linux, shipping pure-LTO object files is a bad idea as we can't assume the user has an LTO-capable linker.

r? @Kobzol
Cc @weihanglo

@rustbot rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-bootstrap Relevant to the bootstrap subteam: Rust's build system (x.py and src/bootstrap) labels Aug 24, 2026
@rustbot

rustbot commented Aug 24, 2026

Copy link
Copy Markdown
Collaborator

Kobzol is not on the review rotation at the moment.
They may take a while to respond.

@RalfJung
RalfJung force-pushed the aarch64-no-lto branch 3 times, most recently from 6a1a15f to 4bb86a4 Compare August 24, 2026 18:29
@RalfJung

RalfJung commented Aug 24, 2026

Copy link
Copy Markdown
Member Author

@bors try jobs=dist-aarch64*

@rust-bors

This comment has been minimized.

Sign in to view
rust-bors Bot pushed a commit that referenced this pull request Aug 24, 2026
@bors
Auto merge of #161689 - RalfJung:aarch64-no-lto, r=<try>
826c9d3 
bootstrap: don't LTO C dependencies on aarch64


try-job: dist-aarch64*
@RalfJung RalfJung mentioned this pull request Aug 24, 2026
Closed
weihanglo
weihanglo approved these changes Aug 24, 2026
View reviewed changes

@weihanglo weihanglo left a comment •
edited by rustbot
Loading

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, seems reasonable.

View changes since this review

@mati865

mati865 commented Aug 24, 2026

Copy link
Copy Markdown
Member

What is so special about AArch64? I guess we don't ship LLD there, so turning the condition to allow only x86_64 (if Linux?) could make more sense.

@weihanglo

weihanglo commented Aug 24, 2026

Copy link
Copy Markdown
Member

What is so special about AArch64? I guess we don't ship LLD there, so turning the condition to allow only x86_64 (if Linux?) could make more sense.

We also ships bincode-only rlib for Apple targets, and happy with that I believe.

@Kobzol

Kobzol commented Aug 24, 2026

Copy link
Copy Markdown
Member

We only saw breakage on aarch64 Linux so far, we're unsure why as of yet. As a hotfix, this seems fine.

You can r=me once the try build comes out green and ideally if you can confirm that its artifacts unbork miri.

@weihanglo

weihanglo commented Aug 24, 2026

Copy link
Copy Markdown
Member

We only saw breakage on aarch64 Linux so far, we're unsure why as of yet. As a hotfix, this seems fine.

I thought the why being that the miri aarch64 job didn't use lld but some GCC linker and not LLVM-bincode compatible, no?

#t-compiler > Analysing broken jemalloc rlib

@RalfJung

RalfJung commented Aug 24, 2026 •
edited
Loading

Copy link
Copy Markdown
Member Author

What is so special about AArch64? I guess we don't ship LLD there,

Nothing, but I want to unblock Miri with a minimal change here. I wasted many hours over multiple days going for a fix that makes sense, I ran out of patience.

so turning the condition to allow only x86_64 (if Linux?) could make more sense.

That would be a much less minimal change. Feel free to make a PR that further tunes this condition after this one lands and we're finally back to normal operation.

The alternative is to revert #161260 as clearly it was landed without us fully grasping the consequences and decisions made in that PR.

@rust-bors

rust-bors Bot commented Aug 24, 2026

Copy link
Copy Markdown
Contributor

☀️ Try build successful (CI)
Build commit: 826c9d3 (826c9d3d41a873ace77fd08d3e213a7aa18d27db)
Base parent: 0a3fa2a (0a3fa2af35783dcd1b4206a85fe7811297eea0bf)

@RalfJung

RalfJung commented Aug 24, 2026

Copy link
Copy Markdown
Member Author

Miri looking good.
@bors r=Kobzol p=1

@rust-bors

rust-bors Bot commented Aug 24, 2026

Copy link
Copy Markdown
Contributor

📌 Commit b1f7d8e has been approved by Kobzol

It is now in the queue for this repository.

@rust-bors rust-bors Bot 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 24, 2026
GuillaumeGomez added a commit to GuillaumeGomez/rust that referenced this pull request Aug 24, 2026
@GuillaumeGomez
Rollup merge of rust-lang#161689 - RalfJung:aarch64-no-lto, r=Kobzol
54745fa 
bootstrap: don't LTO C dependencies on aarch64

Since rust-lang#161535 apparently may take a while, let's land something to un-break Miri: we know that on aarch64-linux, shipping pure-LTO object files is a bad idea as we can't assume the user has an LTO-capable linker.

r? @Kobzol
Cc @weihanglo
@GuillaumeGomez GuillaumeGomez mentioned this pull request Aug 24, 2026
Closed
@rust-bors

rust-bors Bot commented Aug 24, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

⌛ Testing commit b1f7d8e with merge 107a2d7...

Workflow: https://github.com/rust-lang/rust/actions/runs/32785166387

rust-bors Bot pushed a commit that referenced this pull request Aug 24, 2026
@bors
Auto merge of #161689 - RalfJung:aarch64-no-lto, r=Kobzol
107a2d7 
bootstrap: don't LTO C dependencies on aarch64

Since #161535 apparently may take a while, let's land something to un-break Miri: we know that on aarch64-linux, shipping pure-LTO object files is a bad idea as we can't assume the user has an LTO-capable linker.

r? @Kobzol 
Cc @weihanglo
@GuillaumeGomez GuillaumeGomez mentioned this pull request Aug 24, 2026
Merged
@GuillaumeGomez

GuillaumeGomez commented Aug 24, 2026

Copy link
Copy Markdown
Member

Included in #161706.

@bors yield

@rust-bors

rust-bors Bot commented Aug 24, 2026

Copy link
Copy Markdown
Contributor

Auto build was cancelled. Cancelled workflows:

The next pull request likely to be tested is #161706.

rust-bors Bot pushed a commit that referenced this pull request Aug 24, 2026
@bors
Auto merge of #161706 - GuillaumeGomez:rollup-UfytneT, r=GuillaumeGomez
9bb55c8 
Rollup of 9 pull requests

Successful merges:

 - #161689 (bootstrap: don't LTO C dependencies on aarch64)
 - #161553 (rustdoc: fix issue preventing "read more" links from generating.)
 - #161670 (Pass LargeDataThreshold to LLVM module IR)
 - #146529 (rustdoc: Nuke `--passes=list` and defossilize the passes infrastructure)
 - #156009 (Check `redundant_explicit_links` against generated URLs)
 - #159583 (Add new `invalid_markdown_table` rustdoc lint)
 - #161098 (PassWrapper: handle LLVM 24 change in function types)
 - #161641 (Check for missing rustfmt in the stdarch intrinsic test step sooner)
 - #161661 (Don't treat slashes as path seps after drive letters in verbatim paths)

Failed merges:

 - #156749 (remove `box_patterns`)
@mati865

mati865 commented Aug 24, 2026

Copy link
Copy Markdown
Member

We only saw breakage on aarch64 Linux so far, we're unsure why as of yet. As a hotfix, this seems fine.

Probably because we don't LTO other Linux hosts?
x86_64 is the only Linux target where we enable LLD by default:

rust/src/bootstrap/src/core/config/toml/target.rs

Line 133 in e776960

pub fn default_linux_linker_overrides() -> HashMap<String, DefaultLinuxLinkerOverride> {

We also ships bincode-only rlib for Apple targets, and happy with that I believe.

I have no clue about Apple targets tooling, but we don't do LTO of the C/C++ code there. Searching for llvm.thin-lto shows only Aarch64 and x86_64 Linux jobs.

@rust-bors
rust-bors Bot merged commit 1b6b615 into rust-lang:main Aug 25, 2026
13 of 14 checks passed
@rustbot rustbot added this to the 1.100.0 milestone Aug 25, 2026
rust-bors Bot pushed a commit that referenced this pull request Aug 25, 2026
@bors
Unrolled build for #161689 in rollup 161706
89d9a83 
Rollup merge of #161689 - RalfJung:aarch64-no-lto, r=Kobzol

bootstrap: don't LTO C dependencies on aarch64

Since #161535 apparently may take a while, let's land something to un-break Miri: we know that on aarch64-linux, shipping pure-LTO object files is a bad idea as we can't assume the user has an LTO-capable linker.

r? @Kobzol
Cc @weihanglo
@RalfJung

RalfJung commented Aug 25, 2026

Copy link
Copy Markdown
Member Author

FWIW on apple we do something with LTO and C dependencies that causes warnings: #161561

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@weihanglo weihanglo weihanglo approved these changes

Assignees

@Kobzol Kobzol

Labels

S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. T-bootstrap Relevant to the bootstrap subteam: Rust's build system (x.py and src/bootstrap)

Projects

None yet

Milestone

1.100.0

Development

Successfully merging this pull request may close these issues.

6 participants

@RalfJung @rustbot @mati865 @weihanglo @Kobzol @GuillaumeGomez