Never inline C variadics, cold functions, functions with incompatible attributes ... #78966
Conversation
This file contains 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
rust-highfive
added
the
S-waiting-on-review
jyn514
added
A-mir-opt-inlining
|
@bors r+ very excited about the generator inlining. With this we'll be able to craft some MIR optimizations for async code |
|
📌 Commit f91a0ef has been approved by |
bors
added
S-waiting-on-bors
m-ou-se
added a commit
to m-ou-se/rust
that referenced
this pull request
Nov 12, 2020
Never inline C variadics, cold functions, functions with incompatible attributes ... ... and fix generator inlining. Closes rust-lang#67863. Closes rust-lang#78859.
|
@bors r- This failed in #78992 (comment): https://github.com/rust-lang-ci/rust/runs/1392022160 |
bors
added
S-waiting-on-author
The inliner looks if a sanitizer is enabled before considering `no_sanitize` attribute as possible source of incompatibility. The MIR inlining could happen in a crate with sanitizer disabled, but code generation in a crate with sanitizer enabled, thus the attribute would be incorrectly ignored. To avoid the issue never inline functions with different `no_sanitize` attributes.
The information about cold attribute is lost during inlining, Avoid the issue by never inlining cold functions.
The callee body is already transformed; the condition is always false.
tmiasko
force-pushed
the
inline-never
branch
from
f91a0ef
to
2a010dd
Compare
|
Ignored new generator test on wasm. |
|
@bors r+ |
|
📌 Commit 2a010dd has been approved by |
bors
added
S-waiting-on-bors
Dylan-DPC-zz
pushed a commit
to Dylan-DPC-zz/rust
that referenced
this pull request
Nov 13, 2020
Never inline C variadics, cold functions, functions with incompatible attributes ... ... and fix generator inlining. Closes rust-lang#67863. Closes rust-lang#78859.
Dylan-DPC-zz
pushed a commit
to Dylan-DPC-zz/rust
that referenced
this pull request
Nov 13, 2020
Never inline C variadics, cold functions, functions with incompatible attributes ... ... and fix generator inlining. Closes rust-lang#67863. Closes rust-lang#78859.
Dylan-DPC-zz
pushed a commit
to Dylan-DPC-zz/rust
that referenced
this pull request
Nov 14, 2020
Never inline C variadics, cold functions, functions with incompatible attributes ... ... and fix generator inlining. Closes rust-lang#67863. Closes rust-lang#78859.
bors
added a commit
to rust-lang-ci/rust
that referenced
this pull request
Nov 15, 2020
Rollup of 15 pull requests Successful merges: - rust-lang#78352 (Do not call `unwrap` with `signatures` option enabled) - rust-lang#78590 (refactor: removing alloc::collections::vec_deque ignore-tidy-filelength) - rust-lang#78848 (Bump minimal supported LLVM version to 9) - rust-lang#78856 (Explicitly checking for or-pattern before test) - rust-lang#78948 (test: add `()=()=()=...` to weird-exprs.rs) - rust-lang#78962 (Add a test for r# identifiers) - rust-lang#78963 (Added some unit tests as requested) - rust-lang#78966 (Never inline C variadics, cold functions, functions with incompatible attributes ...) - rust-lang#78968 (Include llvm-as in llvm-tools-preview component) - rust-lang#78969 (Normalize function type during validation) - rust-lang#78980 (Fix rustc_ast_pretty print_qpath resulting in invalid macro input) - rust-lang#78986 (Avoid installing external LLVM dylibs) - rust-lang#78988 (Fix an intrinsic invocation on threaded wasm) - rust-lang#78993 (rustc_target: Fix dash vs underscore mismatches in option names) - rust-lang#79013 (Clean up outdated `use_once_payload` pretty printer comment) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
|
Does this PR have the potential to break this workaround in hashbrown? rust-lang/hashbrown#209 |
|
@novacrazy Thanks, that is a great example of a pattern that patch here tries to enable. Currently in MIR the cold attribute can be only attached to a function. If a call to a cold function were to be inlined in MIR, the fact that this is a cold path is lost to the LLVM. The changes here disable inlining of cold functions at MIR level precisely to preserve this information. That being said, the MIR inlining is disabled by default. So unless |
Anecdotally, this happens when LLVM inlines cold functions too (which it does if you only have one call to it unless it's |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
... and fix generator inlining.
Closes #67863.
Closes #78859.