-
Notifications
You must be signed in to change notification settings - Fork 12.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove some stack frames from .async
calls
#66398
Conversation
The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
r? @rkruppe (rust_highfive has picked a reviewer for you, use r? to override) |
📌 Commit 3fe7cfc has been approved by |
Remove some stack frames from `.async` calls The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
Remove some stack frames from `.async` calls The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
Remove some stack frames from `.async` calls The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
Remove some stack frames from `.async` calls The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
Rollup of 12 pull requests Successful merges: - #65557 (rename Error::iter_chain() and remove Error::iter_sources()) - #66013 (Avoid hashing the key twice in `get_query()`.) - #66306 (Remove cannot mutate statics in initializer of another static error) - #66338 (Update mdbook.) - #66388 (Do not ICE on recovery from unmet associated type bound obligation) - #66390 (Fix ICE when trying to suggest `Type<>` instead of `Type()`) - #66391 (Do not ICE in `if` without `else` in `async fn`) - #66394 (Fix two OOM issues related to `ConstProp`) - #66398 (Remove some stack frames from `.async` calls) - #66410 (miri: helper methods for max values of machine's usize/isize) - #66418 (Link to tracking issue in HIR const-check error code description) - #66419 (Don't warn labels beginning with `_` on unused_labels lint) Failed merges: r? @ghost
Remove some stack frames from `.async` calls The `Context` argument is currently smuggled through TLS for async-generated futures. The current infrastructure is closure-based, and results in an extra 6 stack frames when .awaiting an async-generated future! ``` 12: foo::async_b::{{closure}} at src/main.rs:10 13: <std::future::GenFuture<T> as core::future::future::Future>::poll::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 14: std::future::set_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:79 15: <std::future::GenFuture<T> as core::future::future::Future>::poll at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:43 16: std::future::poll_with_tls_context::{{closure}} at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 17: std::future::get_task_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:111 18: std::future::poll_with_tls_context at /rustc/4560ea788cb760f0a34127156c78e2552949f734/src/libstd/future.rs:121 19: foo::async_a::{{closure}} at src/main.rs:6 ``` While the long (medium?) term solution is to remove the use of TLS entirely, we can improve things a bit in the meantime. In particular, this commit does 2 things: 1. `get_task_context` has been inlined into `poll_with_tls_context`, removing 2 frames (16 and 17 above). 2. `set_task_context` now returns a guard type that resets the TLS rather than taking a closure, removing 2 frames (13 and 14 above). We can also remove frame 18 by removing `poll_with_tls_context` in favor of a `get_task_context` function which returns a guard, but that requires adjusting the code generated for .await, so I've left that off for now.
Rollup of 12 pull requests Successful merges: - #65557 (rename Error::iter_chain() and remove Error::iter_sources()) - #66013 (Avoid hashing the key twice in `get_query()`.) - #66306 (Remove cannot mutate statics in initializer of another static error) - #66338 (Update mdbook.) - #66388 (Do not ICE on recovery from unmet associated type bound obligation) - #66390 (Fix ICE when trying to suggest `Type<>` instead of `Type()`) - #66391 (Do not ICE in `if` without `else` in `async fn`) - #66398 (Remove some stack frames from `.async` calls) - #66410 (miri: helper methods for max values of machine's usize/isize) - #66418 (Link to tracking issue in HIR const-check error code description) - #66419 (Don't warn labels beginning with `_` on unused_labels lint) - #66428 (Cleanup unused function) Failed merges: r? @ghost
@sfackler, was the visibility change to |
Yes, it was intentional. The TLS interface is not a public API and will be removed entirely in the future. |
31: Fix build failure on latest nightly r=taiki-e a=taiki-e This fixes breakage due to rust-lang/rust#66398. Fixes #30 Co-authored-by: Taiki Endo <[email protected]>
The
Context
argument is currently smuggled through TLS forasync-generated futures. The current infrastructure is closure-based,
and results in an extra 6 stack frames when .awaiting an async-generated
future!
While the long (medium?) term solution is to remove the use of TLS
entirely, we can improve things a bit in the meantime. In particular,
this commit does 2 things:
get_task_context
has been inlined intopoll_with_tls_context
,removing 2 frames (16 and 17 above).
set_task_context
now returns a guard type that resets the TLSrather than taking a closure, removing 2 frames (13 and 14 above).
We can also remove frame 18 by removing
poll_with_tls_context
in favorof a
get_task_context
function which returns a guard, but thatrequires adjusting the code generated for .await, so I've left that off
for now.