Stop abusing NS_FREE_PERMANENT_DATA for profile/coverage atexit hook requirements
Categories
(Firefox Build System :: General, task, P3)
Tracking
(Not tracked)
People
(Reporter: Gijs, Unassigned)
References
Details
(Keywords: in-triage)
We currently set NS_FREE_PERMANENT_DATA to true for pgo profiling (after bug 1196094) and code coverage (after bug 1460929). This is all to ensure we get atexit hooks. Ideally, we should use something more targeted to that problem.
Comment 1•5 years ago
|
||
Updated•5 years ago
|
Comment 2•5 years ago
|
||
Based on Comment Bug 1570619#c18, can we reprioritize this? We discussed this bug during platform triage today. Thanks.
Comment 3•5 years ago
|
||
+1 to Marcia's comment. This bug is flagged as something that would help a performance regression.
Updated•5 years ago
|
Comment 4•5 years ago
|
||
Mike is working on this.
Comment 5•4 years ago
|
||
I was working on this as a possible way to fix the perf regression in bug 1570619, but fixing this bug did not help the performance there. Do we want this changed even if there's no performance win?
Reporter | ||
Comment 6•4 years ago
|
||
(In reply to Michael Shal [:mshal] from comment #5)
I was working on this as a possible way to fix the perf regression in bug 1570619, but fixing this bug did not help the performance there. Do we want this changed even if there's no performance win?
I think so, yes, though I defer to :froydnj.
Comment 7•4 years ago
|
||
(In reply to :Gijs (he/him) from comment #6)
(In reply to Michael Shal [:mshal] from comment #5)
I was working on this as a possible way to fix the perf regression in bug 1570619, but fixing this bug did not help the performance there. Do we want this changed even if there's no performance win?
I think so, yes, though I defer to :froydnj.
I agree that we would like this changed, for separation of concerns if nothing else.
Updated•2 years ago
|
Description
•