Skip to content

Windows thread-local keyless drop #90442

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

Merged
merged 2 commits into from
Nov 2, 2021
Merged
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 36 additions & 1 deletion library/std/src/sys/windows/thread_local_dtor.rs
Original file line number Diff line number Diff line change
@@ -1,4 +1,39 @@
//! Implements thread-local destructors that are not associated with any
//! particular data.

#![unstable(feature = "thread_local_internals", issue = "none")]
#![cfg(target_thread_local)]
use super::c;

// Using a per-thread list avoids the problems in synchronizing global state.
#[thread_local]
static mut DESTRUCTORS: Vec<(*mut u8, unsafe extern "C" fn(*mut u8))> = Vec::new();

pub unsafe fn register_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) {
DESTRUCTORS.push((t, dtor));
}

// See windows/thread_local_keys.rs for an explanation of this callback function.
// The short version is that all the function pointers in the `.CRT$XL*` array
// will be called whenever a thread or process starts or ends.

#[link_section = ".CRT$XLD"]
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This all seems pretty reasonable to me! Could the previously-registered "at thread exit" function get a new hook to call into this module though? I think it 'd probably be best to avoid these #[link_section] annotations if we can and have them consolidated into one place if possible.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, I've used a normal Rust function for running the new dtors and simply made the exit function call it. Is this what you had in mind?

#[doc(hidden)]
#[used]
pub static TLS_CALLBACK: unsafe extern "system" fn(c::LPVOID, c::DWORD, c::LPVOID) = tls_callback;

pub use crate::sys_common::thread_local_dtor::register_dtor_fallback as register_dtor;
unsafe extern "system" fn tls_callback(_: c::LPVOID, reason: c::DWORD, _: c::LPVOID) {
if reason == c::DLL_THREAD_DETACH || reason == c::DLL_PROCESS_DETACH {
// Drop all the destructors.
//
// Note: While this is potentially an infinite loop, it *should* be
// the case that this loop always terminates because we provide the
// guarantee that a TLS key cannot be set after it is flagged for
// destruction.
while let Some((ptr, dtor)) = DESTRUCTORS.pop() {
(dtor)(ptr);
}
// We're done so free the memory.
DESTRUCTORS.shrink_to_fit();
}
}