Open Bug 1581792 Opened 7 years ago Updated 3 years ago

nsIThreadPool.ShutdownWithTimeout should be done in parallel, not serially

Categories

(Core :: XPCOM, enhancement)

enhancement

Tracking

()

People

(Reporter: valentin, Unassigned)

Details

! In D46002#1396790, @mayhemer wrote:
My idea was more like to have the timeout in the global shutdown logic I pointed to in the previous comment. We could simply tell all threads to shutdown and where not needed (for specific code logic expectations that needs to sync on the shutdown) just forget and continue. The nsThreadManager would then wait for ALL threads for a specific amount of time. What you are trying to do leads to wait for each particular thread/pool we attempt to shutdown serially. If we have 30 threads to kill and happen to wait for half of them each 5 seconds, it will take 5*15=75 seconds.

But that is for a followup, yes :)

Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.