Mark ___asan_globals_registered as an exported symbol for LTO

[anforowicz]

Export a weak symbol defined by AddressSanitizer instrumentation.
Previously, when using LTO, the symbol would get internalized and eliminated.

Fixes #113404.

---

FWIW, let me list similar PRs from the past + who reviewed them:

• Export weak symbols used by MemorySanitizer #68410 (fixing __msan_keep_going and __msan_track_origins; reviewed by @varkor)
• Add codegen test for PGO instrumentation. #60038 and Add basic PGO support. #48346 (fixing __llvm_profile_raw_version and __llvm_profile_filename; reviewed by @alexcrichton)

[rustbot]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @TaKO8Ki (or someone else) soon.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

• @rustbot author: the review is finished, PR author should check the comments and take action accordingly
• @rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

[tmiasko]

Thanks.

r? @tmiasko @bors r+

[bors]

📌 Commit 73b5842 has been approved by tmiasko

It is now in the queue for this repository.

[matthiaskrgr]

@bors r-
failed in #114751 (comment)

[anforowicz]

Thanks for reporting the test failure in #114751 (comment). The problem seems obvious in retrospect - on arm-android bot the LLVM-IR says:

@___asan_globals_registered = common hidden global i32 0

and the test expects almost the same line, but with i64 instead of i32.

Let me relax the test expectations and ping you when the PR has been updated...

[anforowicz]

@tmiasko and/or @matthiaskrgr - could you PTAL again?

[tmiasko]

@bors r+ rollup=never

[bors]

📌 Commit 633b4551832e83342e59abcf52d891f4ec82e299 has been approved by tmiasko

It is now in the queue for this repository.

[bors]

⌛ Testing commit 633b4551832e83342e59abcf52d891f4ec82e299 with merge 254538cb341f6aecb004662528685be25202831d...

[bors]

💔 Test failed - checks-actions

[tmiasko]

@bors r+ rollup=never

[bors]

📌 Commit bf9820e6fef8f8c06604b5673f36e33d94ed4c25 has been approved by tmiasko

It is now in the queue for this repository.

[bors]

⌛ Testing commit bf9820e6fef8f8c06604b5673f36e33d94ed4c25 with merge a9721fbd74af81a9f912673e5138da0847c8c9b3...

[bors]

💔 Test failed - checks-actions

[rustbot]

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

[anforowicz]

@tmiasko - can you PTAL again? This probably needs a quick re-review (to look at the new pub fn binary_format() in impl TargetOptions ), but maybe this is otherwise ready for another try via @bors r+ rollup=never. WDYT?

[bors]

@anforowicz: 🔑 Insufficient privileges: Not in reviewers

[tmiasko]

In this approach we would also have to replicate remaining conditions that control whether ___asan_globals_registered is used, including ShouldUseMachOGlobalsSection and !getUniqueModuleId(&M).empty(). The latter seems especially problematic since it depends on the contents of the module.

At the moment I don't have alternative suggestions, short of disabling UseGlobalGC.

[anforowicz]

@rustbot author

[bors]

☔ The latest upstream changes (presumably #114946) made this pull request unmergeable. Please resolve the merge conflicts.

[anforowicz]

This PR has been made obsolete by #114946