Don't use f64 shims for f32 cmath functions on non 32-bit x86 MSVC #68033
Conversation
These shims are only needed on 32-bit x86. Additionally since https://reviews.llvm.org/rL268875 LLVM handles adding the shims itself for the intrinsics.
|
r? @dtolnay (rust_highfive has picked a reviewer for you, use r? to override) |
rust-highfive
added
the
S-waiting-on-review
|
One hesitation I have is that on for example x86_64 MSVC some of these functions give slightly different results which look like they're actually less accurate. That makes this technically a breaking change but I don't know if we have strong stability or accuracy guarantees for these functions. |
dtolnay
changed the title
There was a problem hiding this comment.
Looks good to me. Thanks!
It looks like the existing workaround was added in #26601 before the LLVM fix to support Windows XP. At some point we should drop XP support and remove the remaining shims... https://internals.rust-lang.org/t/consider-dropping-support-for-windows-xp/8745
Regarding accuracy, the comment in cmath.rs acknowledges that the f64 promoted operations do not exactly match the f32 operations but were only "accurate enough for now".
FYI @alexcrichton
|
@bors r+ |
|
📌 Commit 084217a has been approved by |
bors
added
S-waiting-on-bors
Don't use f64 shims for f32 cmath functions on non 32-bit x86 MSVC These shims are only needed on 32-bit x86. Additionally since https://reviews.llvm.org/rL268875 LLVM handles adding the shims itself for the intrinsics.
Don't use f64 shims for f32 cmath functions on non 32-bit x86 MSVC These shims are only needed on 32-bit x86. Additionally since https://reviews.llvm.org/rL268875 LLVM handles adding the shims itself for the intrinsics.
Rollup of 5 pull requests Successful merges: - #68033 (Don't use f64 shims for f32 cmath functions on non 32-bit x86 MSVC) - #68244 (Enable leak sanitizer test case) - #68255 (Remove unused auxiliary file that was replaced with rust_test_helpers) - #68263 (rustdoc: HTML escape codeblocks which fail syntax highlighting) - #68274 (remove dead code) Failed merges: r? @ghost
Just to be clear, these shims were added to support the
The shims are still needed even in 32-bit x86 Windows 10 so we won't be able to remove them unfortunately. |
These shims are only needed on 32-bit x86. Additionally since https://reviews.llvm.org/rL268875 LLVM handles adding the shims itself for the intrinsics.