Allow backtraces to cross thread boundaries #4093

actuallyatoaster · 2024-12-13T02:52:19Z

When a thread is spawned we store the ID of the spawning thread as well as a backtrace to the spawn location. When there's an error we can traverse upwards, displaying a backtrace up to the main thread.

For instance, with the following program:

extern "C" {
    fn unknown();
}

fn bad() {
    unsafe { unknown() }
}

fn main() {
    std::thread::spawn(bad);
}

Whereas previously we would see:

error: unsupported operation: can't call foreign function `unknown` on OS `linux`
 --> tests/temp.rs:6:14
  |
6 |     unsafe { unknown() }
  |              ^^^^^^^^^ can't call foreign function `unknown` on OS `linux`
  |
  = help: if this is a basic API commonly used on this target, please report an issue with Miri
  = help: however, note that Miri does not aim to support every FFI function out there; for instance, we will not support APIs for things such as GUIs, scripting languages, or databases
  = note: BACKTRACE on thread `unnamed-1`:
  = note: inside `bad` at tests/temp.rs:6:14: 6:23

note: some details are omitted, run with `MIRIFLAGS=-Zmiri-backtrace=full` for a verbose backtrace

error: aborting due to 1 previous error

We now have:

error: unsupported operation: can't call foreign function `unknown` on OS `linux`
  --> tests/temp.rs:6:14
   |
6  |     unsafe { unknown() }
   |              ^^^^^^^^^ can't call foreign function `unknown` on OS `linux`
   |
   = help: if this is a basic API commonly used on this target, please report an issue with Miri
   = help: however, note that Miri does not aim to support every FFI function out there; for instance, we will not support APIs for things such as GUIs, scripting languages, or databases
   = note: BACKTRACE on thread `unnamed-1`:
   = note: inside `bad` at tests/temp.rs:6:14: 6:23
   = note: thread `unnamed-1` was spawned by thread `main`
   = note: inside `std::sys::pal::unix::thread::Thread::new` at /home/reimu/.rustup/toolchains/miri/lib/rustlib/src/rust/library/std/src/sys/pal/unix/thread.rs:84:19: 84:86
   = note: inside `std::thread::Builder::spawn_unchecked_::<'_, fn() {bad}, ()>` at /home/reimu/.rustup/toolchains/miri/lib/rustlib/src/rust/library/std/src/thread/mod.rs:600:30: 600:64
   = note: inside `std::thread::Builder::spawn_unchecked::<fn() {bad}, ()>` at /home/reimu/.rustup/toolchains/miri/lib/rustlib/src/rust/library/std/src/thread/mod.rs:467:32: 467:62
   = note: inside `std::thread::Builder::spawn::<fn() {bad}, ()>` at /home/reimu/.rustup/toolchains/miri/lib/rustlib/src/rust/library/std/src/thread/mod.rs:400:18: 400:41
   = note: inside `std::thread::spawn::<fn() {bad}, ()>` at /home/reimu/.rustup/toolchains/miri/lib/rustlib/src/rust/library/std/src/thread/mod.rs:730:5: 730:28
note: inside `main`
  --> tests/temp.rs:10:5
   |
10 |     std::thread::spawn(bad);
   |     ^^^^^^^^^^^^^^^^^^^^^^^

note: some details are omitted, run with `MIRIFLAGS=-Zmiri-backtrace=full` for a verbose backtrace

error: aborting due to 1 previous error

src/diagnostics.rs

oli-obk · 2024-12-13T06:34:31Z

You'll need to bless the windows tests, too. It should suffice to pass --target with a windows target, no windows host needed

RalfJung · 2024-12-14T09:39:32Z

I have some nits.

RalfJung · 2024-12-14T09:39:21Z

src/concurrency/thread.rs

@@ -235,6 +235,9 @@ pub struct Thread<'tcx> {
    /// The join status.
    join_status: ThreadJoinStatus,

+    // ThreadId that spawned this thread and backtrace to where this thread was spawned


Suggested change

// ThreadId that spawned this thread and backtrace to where this thread was spawned

/// ThreadId that spawned this thread and backtrace to where this thread was spawned.

RalfJung · 2024-12-14T09:43:10Z

src/diagnostics.rs

+                // No span for non-local frames except the first frame (which is the error site).
+                if is_local && idx > 0 {
+                    err.subdiagnostic(frame_info.as_note(machine.tcx));


This comment is not right, the first frame is not the error site.

RalfJung · 2024-12-14T09:44:00Z

src/diagnostics.rs

+            for (idx, frame_info) in pruned_stacktrace.iter().enumerate() {
+                let is_local = machine.is_local(frame_info);
+                // No span for non-local frames except the first frame (which is the error site).
+                if is_local && idx > 0 {
+                    err.subdiagnostic(frame_info.as_note(machine.tcx));
+                } else {
+                    let sm = sess.source_map();
+                    let span = sm.span_to_embeddable_string(frame_info.span);
+                    err.note(format!("{frame_info} at {span}"));
+                }
+            }


This is duplicating existing logic. Can you move this into a shared helper fn?

RalfJung · 2024-12-14T09:45:11Z

tests/fail-dep/concurrency/libc_pthread_mutex_deadlock.stderr

+   = note: inside `std::sys::pal::PLATFORM::thread::Thread::new` at RUSTLIB/std/src/sys/pal/PLATFORM/thread.rs:LL:CC
+   = note: inside `std::thread::Builder::spawn_unchecked_::<'_, {closure@tests/fail-dep/concurrency/libc_pthread_mutex_deadlock.rs:LL:CC}, ()>` at RUSTLIB/std/src/thread/mod.rs:LL:CC
+   = note: inside `std::thread::Builder::spawn_unchecked::<{closure@tests/fail-dep/concurrency/libc_pthread_mutex_deadlock.rs:LL:CC}, ()>` at RUSTLIB/std/src/thread/mod.rs:LL:CC
+   = note: inside `std::thread::Builder::spawn::<{closure@tests/fail-dep/concurrency/libc_pthread_mutex_deadlock.rs:LL:CC}, ()>` at RUSTLIB/std/src/thread/mod.rs:LL:CC
+   = note: inside `std::thread::spawn::<{closure@tests/fail-dep/concurrency/libc_pthread_mutex_deadlock.rs:LL:CC}, ()>` at RUSTLIB/std/src/thread/mod.rs:LL:CC


It's not pretty to show this for every backtrace. We spent a bunch of effort making backtraces nice, i.e. they usually stop at main and don't go all the way to the start lang item. We should do something similar here to avoid drowning the user in noise.

I could imagine two options:

we mark these frames as #[caller_location] in std (we have done this for a bunch of other functions to make backtraces prettier)

when pruning the stack for another thread, we also remove all non-local frames at the top

I am leaning towards the first option since the second option might prune too much when the thread is spawned deep inside some other library.

I have posted rust-lang/rust#134560 to add the #[caller_location] for this.

RalfJung · 2024-12-20T14:01:08Z

@rustbot author
Please post @rustbot ready when the PR is ready for the next round of review.

mri: add track_caller to thread spawning methods for better backtraces This is in preparation for rust-lang/miri#4093

Rollup merge of rust-lang#134560 - RalfJung:miri-thread-spawn, r=jhpratt mri: add track_caller to thread spawning methods for better backtraces This is in preparation for rust-lang/miri#4093

Include thread spawn in backtrace

6ea9883

oli-obk reviewed Dec 13, 2024

View reviewed changes

src/diagnostics.rs Show resolved Hide resolved

bless windows tests

db8436f

oli-obk added this pull request to the merge queue Dec 14, 2024

RalfJung removed this pull request from the merge queue due to a manual request Dec 14, 2024

RalfJung reviewed Dec 14, 2024

View reviewed changes

rustbot added the S-waiting-on-author Status: Waiting for the PR author to address review comments label Dec 20, 2024

RalfJung mentioned this pull request Dec 20, 2024

mri: add track_caller to thread spawning methods for better backtraces rust-lang/rust#134560

Merged

jieyouxu added a commit to jieyouxu/rust that referenced this pull request Dec 20, 2024

Rollup merge of rust-lang#134560 - RalfJung:miri-thread-spawn, r=jhpratt

94ae08f

mri: add track_caller to thread spawning methods for better backtraces This is in preparation for rust-lang/miri#4093

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Allow backtraces to cross thread boundaries #4093

Allow backtraces to cross thread boundaries #4093

actuallyatoaster commented Dec 13, 2024

oli-obk commented Dec 13, 2024

RalfJung commented Dec 14, 2024

RalfJung Dec 14, 2024

RalfJung Dec 14, 2024

RalfJung Dec 14, 2024

RalfJung Dec 14, 2024

RalfJung Dec 20, 2024

RalfJung commented Dec 20, 2024

	// ThreadId that spawned this thread and backtrace to where this thread was spawned
	/// ThreadId that spawned this thread and backtrace to where this thread was spawned.

Allow backtraces to cross thread boundaries #4093

Are you sure you want to change the base?

Allow backtraces to cross thread boundaries #4093

Conversation

actuallyatoaster commented Dec 13, 2024

oli-obk commented Dec 13, 2024

RalfJung commented Dec 14, 2024

RalfJung Dec 14, 2024

Choose a reason for hiding this comment

RalfJung Dec 14, 2024

Choose a reason for hiding this comment

RalfJung Dec 14, 2024

Choose a reason for hiding this comment

RalfJung Dec 14, 2024

Choose a reason for hiding this comment

RalfJung Dec 20, 2024

Choose a reason for hiding this comment

RalfJung commented Dec 20, 2024