src: use RAII for mutexes and condition variables #7334
Conversation
bnoordhuis
added
c++
| @@ -470,21 +462,19 @@ void AgentImpl::OnRemoteDataIO(uv_stream_t* stream, | |||
| } | |||
| DisconnectAndDisposeIO(socket); | |||
| } | |||
| uv_cond_broadcast(&agent->pause_cond_); | |||
| agent->pause_cond_.Broadcast(scoped_lock); | |||
There was a problem hiding this comment.
@pavelfeldman Can you comment on whether the changes to this function look correct to you? ISTM it should be holding the lock when signalling the condition variable.
|
@bnoordhuis The next time you rebase this against master, could you include addaleax/node@d16e607? Sorry it took me so long to get #6635 landed. |
|
Rebased, with workaround for compilers that don't understand constexpr function pointers: https://ci.nodejs.org/job/node-test-pull-request/3024/ |
|
Green except for flaky test parallel/test-vm-timeout, tracked in #6727. |
|
nice. LGTM |
|
LGTM |
| inline ~ConditionVariableBase(); | ||
| inline void Broadcast(const ScopedLock&); | ||
| inline void Signal(const ScopedLock&); | ||
| inline void Wait(const ScopedLock& scoped_lock); |
There was a problem hiding this comment.
reason for adding the argument name scoped_lock in Wait() but not in Broadcast() or Signal()?
There was a problem hiding this comment.
Wait() uses it, the other two just take it to prove ownership (i.e., you can't signal or broadcast if you don't hold the lock.)
|
Thanks for answering the questions. Makes sense. LGTM. |
cpplint gets too easily confused by C++ constructs that look like function declarations but aren't. Furthermore, it's arguably a bad rule that conflicts with gcc's -Wunused-parameter flag. PR-URL: nodejs#7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: nodejs#7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
cpplint gets too easily confused by C++ constructs that look like function declarations but aren't. Furthermore, it's arguably a bad rule that conflicts with gcc's -Wunused-parameter flag. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
|
This touches the inspector. Marking as |
|
@bnoordhuis 781713d applies poorly on v6 but as far as I can tell the other commit is required to not have more conflicts on the inspector... If possible, separating out non-backportable work would be greatly appreciated by us releasers! |
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
|
@bnoordhuis should this be backported? |
|
@thealphanerd It's not necessary per se but it would be good. You'll have to drop the changes to src/inspector_agent.cc. |
|
#7715 - v4.x back-port |
|
@bnoordhuis this doesn't seem to apply cleanly to v6.x either, and #7715 has 213 commits. Could you backport this? |
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: nodejs#7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
|
For posterity, this landed in v6.x in commit 0593351. |
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
We will be introducing many more critical sections in the upcoming multi-isolate changes, so let's make manual synchronization a thing of the past. PR-URL: #7334 Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: James M Snell <[email protected]> Reviewed-By: Trevor Norris <[email protected]>
We will be introducing many more critical sections in the upcoming
multi-isolate changes, so let's make manual synchronization a thing
of the past.
CI: https://ci.nodejs.org/job/node-test-pull-request/3022/