No longer auto-inline f16 and f128 on well-supported platforms - #162883
Merged
Merged
Conversation
rustbot
added
S-waiting-on-author
In 5d81891 ("Always inline functions signatures containing `f16` or `f128`"), these types were changed to automatically inline so codegen wouldn't crash on poorly-supported platforms. We have since gained a cfg to reflect the type's codegen reliability. Update so we only check and auto-inline based on type if this config is set, which makes `f16` and `f128` act more like any other type on most platforms. We still can't remove this entirely since a lot of API in `std` wouldn't get inlined and would crash the few remaining poorly-supported backend+target combinations.
tgross35
force-pushed
the
f16-f128-autoinline
branch
from
September 17, 2026 05:16
3006b4d to
66f69fc
Compare
Member
Author
|
PR CI covers well-supported targets, checking some that still have broken f16/f128 @bors try jobs=llvm-21,dist-powerpc-linux,test-x86_64-mingw* |
This comment has been minimized.
This comment has been minimized.
rust-bors Bot
pushed a commit
that referenced
this pull request
Sep 17, 2026
No longer auto-inline `f16` and `f128` on well-supported platforms try-job: *llvm-21* try-job: dist-powerpc-linux try-job: test-x86_64-mingw*
tgross35
marked this pull request as ready for review
September 17, 2026 05:25
Collaborator
|
Some changes occurred to MIR optimizations cc @rust-lang/wg-mir-opt |
rustbot
added
S-waiting-on-review
Collaborator
|
r? @adwinwhite rustbot has assigned @adwinwhite. Use Why was this reviewer chosen?The reviewer was selected based on:
|
Member
Author
|
Assuming CI reflects my local tests, |
Contributor
Member
|
That @bors r+ |
Contributor
rust-bors
Bot
added
S-waiting-on-bors
rust-bors Bot
pushed a commit
that referenced
this pull request
Sep 26, 2026
Rollup of 13 pull requests Successful merges: - #160695 (Detect bad number of generics caused by bad derive) - #162883 (No longer auto-inline `f16` and `f128` on well-supported platforms) - #163133 ([rustdoc] Fix invalid jump to def link when `#[rustc_allow_incoherent_impl]` is involved) - #163215 (Fix suggestion for Option to bool with proper precedence handling) - #163266 (More deferred liveness cleanups) - #163274 (Support -Z merge-functions with gcc and add stack-protector asm tests) - #163290 (Merge `ForceWarning`/`Allow`/`Expect` into `Warning`) - #163298 (Don't emit sugg for pub-reexport extern std) - #163312 (Add rustdoc regression test for glob import of a crate that re-exports) - #163357 (Borrowck cleanups part trois) - #161710 (Stabilize `mem::conjure_zst`) - #163332 (Add some docs to `Global`) - #163362 (simplify ndk compiler test)
rust-bors Bot
pushed a commit
that referenced
this pull request
Sep 26, 2026
Rollup merge of #162883 - tgross35:f16-f128-autoinline, r=saethlin No longer auto-inline `f16` and `f128` on well-supported platforms In 5d81891 ("Always inline functions signatures containing `f16` or `f128`"), these types were changed to automatically inline so codegen wouldn't crash on poorly-supported platforms. We have since gained a cfg to reflect the type's codegen reliability. Update so we only check and auto-inline based on type if this config is set, which makes `f16` and `f128` act more like any other type on most platforms. We still can't remove this entirely since a lot of API in `std` wouldn't get inlined and would crash the few remaining poorly-supported backend+target combinations. Followup to #133050.
pull Bot
pushed a commit
to LeeeeeeM/miri
that referenced
this pull request
Sep 27, 2026
Rollup of 13 pull requests Successful merges: - rust-lang/rust#160695 (Detect bad number of generics caused by bad derive) - rust-lang/rust#162883 (No longer auto-inline `f16` and `f128` on well-supported platforms) - rust-lang/rust#163133 ([rustdoc] Fix invalid jump to def link when `#[rustc_allow_incoherent_impl]` is involved) - rust-lang/rust#163215 (Fix suggestion for Option to bool with proper precedence handling) - rust-lang/rust#163266 (More deferred liveness cleanups) - rust-lang/rust#163274 (Support -Z merge-functions with gcc and add stack-protector asm tests) - rust-lang/rust#163290 (Merge `ForceWarning`/`Allow`/`Expect` into `Warning`) - rust-lang/rust#163298 (Don't emit sugg for pub-reexport extern std) - rust-lang/rust#163312 (Add rustdoc regression test for glob import of a crate that re-exports) - rust-lang/rust#163357 (Borrowck cleanups part trois) - rust-lang/rust#161710 (Stabilize `mem::conjure_zst`) - rust-lang/rust#163332 (Add some docs to `Global`) - rust-lang/rust#163362 (simplify ndk compiler test)
tgross35
added a commit
to tgross35/compiler-builtins
that referenced
this pull request
Sep 28, 2026
`f16 and `f128` are no longer automatically inlined as of recent nightlies. This means `CastInto` methods may not be inlined, which is happening on Windows and causing an error from missing `__floattihf`. Mark them inline, which is probably good for performance too if this isn't happening automatically. Link: rust-lang/rust#162883
tgross35
added a commit
to tgross35/compiler-builtins
that referenced
this pull request
Sep 28, 2026
`f16 and `f128` are no longer automatically inlined as of recent nightlies. This means `CastInto` methods may not be inlined, which is happening on Windows and causing an error from missing `__floattihf`. Mark them inline, which is probably good for performance too if this isn't happening automatically. Link: rust-lang/rust#162883
tgross35
added a commit
to rust-lang/compiler-builtins
that referenced
this pull request
Sep 28, 2026
`f16 and `f128` are no longer automatically inlined as of recent nightlies. This means `CastInto` methods may not be inlined, which is happening on Windows and causing an error from missing `__floattihf`. Mark them inline, which is probably good for performance too if this isn't happening automatically. Link: rust-lang/rust#162883
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
In 5d81891 ("Always inline functions signatures containing
f16orf128"), these types were changed to automatically inline so codegen wouldn't crash on poorly-supported platforms. We have since gained a cfg to reflect the type's codegen reliability. Update so we only check and auto-inline based on type if this config is set, which makesf16andf128act more like any other type on most platforms.We still can't remove this entirely since a lot of API in
stdwouldn't get inlined and would crash the few remaining poorly-supported backend+target combinations.Followup to #133050.