Multiple meta platform categories when showing only content in new profiler

NEW
Unassigned

Status

()

Firefox
Developer Tools: Performance Tools (Profiler/Timeline)
P3
normal
3 years ago
2 years ago

People

(Reporter: jsantell, Unassigned)

Tracking

Firefox Tracking Flags

(Not tracked)

Details

Attachments

(2 attachments, 1 obsolete attachment)

Created attachment 8589976 [details]
Screen Shot 2015-04-08 at 4.54.26 PM.png

When recording a profile (via console.profile, although I don't think that matters), I saw two entries for "GC". After bug 1108843, these categories should always be combined if they are leaf nodes.
Created attachment 8589978 [details]
dupegc.json

Here's the profile that has duplicate GC entries.
Created attachment 8710769 [details]
extragc.json

It seems that the GC entry for "nsCycleCollector::forgetSkippable" is different than the GC entry for "js::Nursery::collect"
Attachment #8589978 - Attachment is obsolete: true
(In reply to Jordan Santell [:jsantell] [@jsantell] (Please needinfo) from comment #2)
> Created attachment 8710769 [details]
> extragc.json
> 
> It seems that the GC entry for "nsCycleCollector::forgetSkippable" is
> different than the GC entry for "js::Nursery::collect"

forgetSkippable should be CC not GC... is this not so?
I think this is because we create the tree from the samples before renaming to categories, so js::Nursery::collect and GCRuntime::collect would be different nodes in the tree, but they have the same category.
All the platform nodes should be flattened into the same category node -- GC is no different, we have many different Graphics, Network, etc frames with different function names, but combined to the same node (if on the same "depth" level)
(In reply to Jordan Santell [:jsantell] [@jsantell] (Please needinfo) from comment #5)
> All the platform nodes should be flattened into the same category node -- GC
> is no different, we have many different Graphics, Network, etc frames with
> different function names, but combined to the same node (if on the same
> "depth" level)

What if we have two sampled stacks:

1. JS->Pseudo(Gecko)->Psuedo(GC)
2. JS->Pseudo(GC)

Will these leaves be properly flattened to the same category node?
Triaging. Filter on ADRENOCORTICOTROPIC (yes).
OS: Mac OS X → Unspecified
Priority: -- → P3
Hardware: x86 → Unspecified
Version: 37 Branch → unspecified
You need to log in before you can comment on or make changes to this bug.