Conversation
rust-highfive
added
the
S-waiting-on-review
|
My thoughts haven't really changed on this over time: |
|
I don't know how much I'll try to run perf to see if it's even worth pushing for. It could be changed from blacklisting Assuming the perf looks good it should be |
|
Indeed! |
|
On Ryzen platform running openSUSE Tumbleweed (Linux 5.0.6, glibc 2.29) this results in slightly red noise: Rust perf seems to be running on Debian Stretch with significantly older packages so it may yield different results but you can go ahead and close this if you want. |
|
Ok, sounds like we may not be quite ready for this yet! |
|
My guess is that if you want to see improvements, you need to use bare |
Issue rust-lang#25088 has been part of `thread_local!` for quite some time now. Historical attempts have been made to add `#[inline]` to `__getit` in rust-lang#43931, rust-lang#50252, and rust-lang#59720, but these attempts ended up not landing at the time due to segfaults on Windows. In the interim though with `const`-initialized thread locals AFAIK this is the only remaining bug which is why you might want to use `#[thread_local]` over `thread_local!`. As a result I figured it was time to resubmit this and see how it fares on CI and if I can help debugging any issues that crop up. Closes rust-lang#25088
…ss-crate, r=Mark-Simulacrum std: Attempt again to inline thread-local-init across crates Issue rust-lang#25088 has been part of `thread_local!` for quite some time now. Historical attempts have been made to add `#[inline]` to `__getit` in rust-lang#43931, rust-lang#50252, and rust-lang#59720, but these attempts ended up not landing at the time due to segfaults on Windows. In the interim though with `const`-initialized thread locals AFAIK this is the only remaining bug which is why you might want to use `#[thread_local]` over `thread_local!`. As a result I figured it was time to resubmit this and see how it fares on CI and if I can help debugging any issues that crop up. Closes rust-lang#25088
4 participants
Trying to land #50252 (originally part of #43931) for platforms where it doesn't fail.
cc #25088
r? @eddyb