sync adding nounwind attribute with generating abort-on-panic shim #63884
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
|
r? @eddyb (rust_highfive has picked a reviewer for you, use r? to override) |
rust-highfive
added
the
S-waiting-on-review
RalfJung
commented
Aug 25, 2019
RalfJung
commented
Aug 25, 2019
Member
Author
There was a problem hiding this comment.
Do we really want to assume that an imported extern "Rust" { fn bar(); } does not unwind? I just expanded the comment here to describe already existing behavior.
Cc @gnzlbg
Contributor
There was a problem hiding this comment.
Do we really want to assume that an imported extern "Rust" { fn bar(); } does not unwind?
IIRC, "Rust", "platform-intrinsic", "rust-intrinsic", and "rustcall" are actually the same ABI, and it is ok for extern declarations of those to actually unwind. One such a declaration would be the integer add intrinsic that panics on integer overflow. We also sometimes import llvm intrinsics (via #[feature(link_llvm_intrinsics(]) using the "unadjusted" ABI. I'm not sure if that one should be able to unwind as well.
Member
Author
There was a problem hiding this comment.
My personal inclination is to just remove the foreign_item case here. But that decision should probably be made by someone who actually knows the surrounding code.
Member
Author
There was a problem hiding this comment.
In fact I had to remove that branch to make the codegen tests work as expected. Whatever that means.
RalfJung
commented
Aug 25, 2019
src/librustc/ty/context.rs
Outdated
Member
Author
There was a problem hiding this comment.
I had no idea what the right place would be for this method. Suggestions?
RalfJung
force-pushed
the
abort-on-panic
branch
from
0cf0812 to
0f6bbcb
Compare
Contributor
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
RalfJung
added
the
beta-nominated
Member
Author
Member
Author
|
Adjusted and expanded the failing test. I tried for a long time to also adjust and expand |
Centril
added
I-nominated
T-lang
Contributor
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
Member
Author
|
Hm, incremental thinlto tests are failing and I have no idea why, |
This comment has been minimized.
This comment has been minimized.
Member
Author
|
@Centril I am not entirely sure why you nominated this for T-lang? This is a bugfix, solving the problem that the comments in LLVM codegen are wrong (they say we add abort-on-panic shims but we don't). #62603 seems like the more interesting one from a T-lang perspective, that one actually changed program behavior. |
RalfJung
force-pushed
the
abort-on-panic
branch
from
ea60cf7 to
cef82c3
Compare
RalfJung
removed
the
beta-nominated
RalfJung
force-pushed
the
abort-on-panic
branch
from
cef82c3 to
2214932
Compare
Member
Author
|
I felt like this PR did a bit too much for backporting, so I moved the soundness-fix part into a separate PR: #63909. This PR here is now based on that. Here's the relative diff. |
Contributor
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
Contributor
Why is this nominated for beta ? AFAICT this only fixes a bug in a unstable Rust feature right ? |
Contributor
|
Does it remove the |
Member
Author
No, it (or rather, #63909) removes the This PR here is not nominated for backport any more.
#63909 does. This PR includes that one. That is already reflected in the PR message. |
RalfJung
added
S-blocked
Member
Author
|
Marking as blocked and removing nomination; that has been moved to #63909. |
Collaborator
|
☔ The latest upstream changes (presumably #64264) made this pull request unmergeable. Please resolve the merge conflicts. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, the code paths for deciding whether to add an abort-on-panic shim, and whether to add the
nounwindattribute, are separate. This unifies them. As a consequence, we now add anounwindattribute to#[unwind(aborts)]functions.The logic is also changed slightly to not take the call ABI into account any more. Fixes #63883.
Based on #63909, relative diff.