Replaced mutex in constellation logging by a reentrant mutex. #12637

asajeffrey · 2016-07-28T20:36:46Z

The double-panic in #12553 may be caused by using a non-reentrant lock, which panics on reetry. This PR adds a reentrant lock type (slightly annoyingly, the implementation in std isn't exported) and uses it for logging. cc @jdm

./mach build -d does not report any errors
./mach test-tidy does not report any errors
These changes fix #12619.
These changes do not require tests because they are designed to remove a class of intermittents.

This change is

highfive · 2016-07-28T20:36:50Z

Warning

These commits modify unsafe code. Please review it carefully!

asajeffrey · 2016-07-28T21:46:58Z

The unit tests are the same ones that are used by the Rust reentrant lock implementation.

emilio · 2016-07-28T22:09:26Z

components/util/remutex.rs

@@ -0,0 +1,217 @@
+/* This Source Code Form is subject to the terms of the Mozilla Public


Is this true, is this source MPL?

Well I wrote it :)

Actually, you have a point, which is that the unit tests are Apache/MIT licensed, not MPL. I'll fix that.

Oh, I thought you copied it straight from libstd, but yeah, the tests still apply.

No, it's the same interface as the libstd version, but implemented on top of std, rather than using reentrant locks from the OS.

emilio · 2016-07-28T22:24:39Z

components/util/remutex.rs

+// TODO: can we use the thread-id crate for this?
+
+#[derive(Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash, Debug)]
+pub struct ThreadId(NonZero<usize>);


It would be nice to standardize this kind of things, we already have something like that in style/tid, would you mind filling a issue about this?

I would recommend using the thread-id crate instead since it does not make use of thread-local storage, which can be slow on some platforms.

Filed #12641.

@Amanieu: I'd like to get this landed, then think about whether we can improve it by using third-party crates. Adding new dependencies to Servo is fairly easy, but not completely trivial.

emilio · 2016-07-28T22:27:54Z

I don't think you need to ignore all the lints for the tests (see https://github.com/asajeffrey/servo/blob/3f2b55aa0b00d45c9bea873d55dc912298e17de0/python/tidy/servo_tidy/licenseck.py#L62).

LGTM, I'd like to review the mutex a bit more carefully, though in the meantime I guess we can give it a try run:

@bors-servo: try

bors-servo · 2016-07-28T22:27:57Z

⌛ Trying commit 3f2b55a with merge 1606cd4...


        Auto merge of #12637 - asajeffrey:constellation-use-reentrant-logging…

@jdm

…-mutex, r=<try> Replaced mutex in constellation logging by a reentrant mutex.  The double-panic in #12553 may be caused by using a non-reentrant lock, which panics on reetry. This PR adds a reentrant lock type (slightly annoyingly, the implementation in std isn't exported) and uses it for logging. cc @jdm ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes fix #12619. - [X] These changes do not require tests because they are designed to remove a class of intermittents.   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/12637)

bors-servo · 2016-07-28T22:36:45Z

💔 Test failed - mac-dev-unit

asajeffrey · 2016-07-28T23:06:13Z

http://build.servo.org/builders/mac-dev-unit/builds/2291 is failing because it's trying to build remutex (which uses unstable NonZero) for geckolib (which is stable).

emilio · 2016-07-28T23:36:26Z

@bors-servo: try

LGTM with squash, let's see what try thinks :P

bors-servo · 2016-07-28T23:36:29Z

⌛ Trying commit 22fe34c with merge 1e63e2f...


        Auto merge of #12637 - asajeffrey:constellation-use-reentrant-logging…

@jdm

…-mutex, r=<try> Replaced mutex in constellation logging by a reentrant mutex.  The double-panic in #12553 may be caused by using a non-reentrant lock, which panics on reetry. This PR adds a reentrant lock type (slightly annoyingly, the implementation in std isn't exported) and uses it for logging. cc @jdm ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes fix #12619. - [X] These changes do not require tests because they are designed to remove a class of intermittents.   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/12637)

bors-servo · 2016-07-29T01:33:59Z

💔 Test failed - mac-rel-wpt


        Replaced mutex in constellation logging by a reentrant mutex.

asajeffrey · 2016-07-29T14:45:16Z

Squashed.

asajeffrey · 2016-07-29T14:48:15Z

I think the failure of /XMLHttpRequest/send-usp.worker is nothing to do with this PR, just checking on IRC to see if anyone's seen this before.

asajeffrey · 2016-07-29T14:56:22Z

On IRC, @jdm says this is #12654. http://logs.glob.uno/?c=mozilla%23servo&s=29+Jul+2016&e=29+Jul+2016#c491848

@emilio is this r=you?

emilio · 2016-07-29T17:40:23Z

@bors-servo: r+

Yep, lgtm.

bors-servo · 2016-07-29T17:40:24Z

📌 Commit e3030d0 has been approved by emilio

bors-servo · 2016-07-29T17:49:09Z

⌛ Testing commit e3030d0 with merge fdb1e51...


        Auto merge of #12637 - asajeffrey:constellation-use-reentrant-logging…

@jdm

…-mutex, r=emilio Replaced mutex in constellation logging by a reentrant mutex.  The double-panic in #12553 may be caused by using a non-reentrant lock, which panics on reetry. This PR adds a reentrant lock type (slightly annoyingly, the implementation in std isn't exported) and uses it for logging. cc @jdm ---  - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes fix #12619. - [X] These changes do not require tests because they are designed to remove a class of intermittents.   --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/12637)

bors-servo · 2016-07-29T18:53:11Z

☀️ Test successful - arm32, arm64, linux-dev, linux-rel, mac-dev-unit, mac-rel-css, mac-rel-wpt, windows-dev

highfive assigned emilio Jul 28, 2016

highfive added the S-awaiting-review label Jul 28, 2016

asajeffrey added A-constellation I-panic and removed S-awaiting-review labels Jul 28, 2016

highfive added the S-awaiting-review label Jul 28, 2016

asajeffrey added the S-needs-squash label Jul 28, 2016

emilio reviewed Jul 28, 2016
View changes

highfive added S-tests-failed and removed S-tests-failed labels Jul 28, 2016

highfive added the S-tests-failed label Jul 29, 2016

Replaced mutex in constellation logging by a reentrant mutex.

Loading status checks…

e3030d0

asajeffrey force-pushed the asajeffrey:constellation-use-reentrant-logging-mutex branch from 22fe34c to e3030d0 Jul 29, 2016

highfive removed the S-tests-failed label Jul 29, 2016

asajeffrey removed the S-needs-squash label Jul 29, 2016

highfive added S-awaiting-merge and removed S-awaiting-review labels Jul 29, 2016

bors-servo merged commit e3030d0 into servo:master Jul 29, 2016
3 checks passed

3 checks passed

continuous-integration/appveyor/pr AppVeyor build succeeded
Details

continuous-integration/travis-ci/pr The Travis CI build passed
Details

homu Test successful
Details

highfive removed the S-awaiting-merge label Jul 29, 2016

bors-servo mentioned this pull request Jul 29, 2016

Remove util::ipc module (use ipc_channel instead; fixes #12312) #12558

Closed

4 of 5 tasks complete

servo / servo

Replaced mutex in constellation logging by a reentrant mutex. #12637

Replaced mutex in constellation logging by a reentrant mutex. #12637

asajeffrey commented Jul 28, 2016 •

edited by larsbergstrom

highfive commented Jul 28, 2016

asajeffrey commented Jul 28, 2016

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

emilio commented Jul 28, 2016

bors-servo commented Jul 28, 2016

bors-servo commented Jul 28, 2016

asajeffrey commented Jul 28, 2016

emilio commented Jul 28, 2016

bors-servo commented Jul 28, 2016

bors-servo commented Jul 29, 2016

asajeffrey commented Jul 29, 2016 •

edited

asajeffrey commented Jul 29, 2016

asajeffrey commented Jul 29, 2016

emilio commented Jul 29, 2016

bors-servo commented Jul 29, 2016

bors-servo commented Jul 29, 2016

bors-servo commented Jul 29, 2016

		@@ -0,0 +1,217 @@
		/* This Source Code Form is subject to the terms of the Mozilla Public

servo / servo

Join GitHub today

Replaced mutex in constellation logging by a reentrant mutex. #12637

Replaced mutex in constellation logging by a reentrant mutex. #12637

Conversation

asajeffrey commented Jul 28, 2016 • edited by larsbergstrom

highfive commented Jul 28, 2016

asajeffrey commented Jul 28, 2016

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

emilio commented Jul 28, 2016

bors-servo commented Jul 28, 2016

bors-servo commented Jul 28, 2016

asajeffrey commented Jul 28, 2016

emilio commented Jul 28, 2016

bors-servo commented Jul 28, 2016

bors-servo commented Jul 29, 2016

asajeffrey commented Jul 29, 2016 • edited

asajeffrey commented Jul 29, 2016

asajeffrey commented Jul 29, 2016

emilio commented Jul 29, 2016

bors-servo commented Jul 29, 2016

bors-servo commented Jul 29, 2016

bors-servo commented Jul 29, 2016

asajeffrey commented Jul 28, 2016 •

edited by larsbergstrom

asajeffrey commented Jul 29, 2016 •

edited