Bug 89472 – Web Inspector: other section of the native memory chart sometimes contain negative value

• Status: RESOLVED
• Resolution: INVALID
• Component: Web Inspector (Deprecated)

[Yury Semikhatsky]

See attached screenshot.

Steps to reproduce:

1. Open attached file.
2. Open the native memory bar chart.
3. Refresh inspected page several times.

[Yury Semikhatsky]

Created attachment 148336 [details]
Screenshot

[Yury Semikhatsky]

Created attachment 148337 [details]
Test page

[Alexei Filippov]

Seems to be a v8 issue. I filed a bug http://code.google.com/p/v8/issues/detail?id=2191
But we can kinda workaround it.

[Alexei Filippov]

Created attachment 148782 [details]
Patch

[Yury Semikhatsky]

Comment on attachment 148782 [details]
Patch

This is a new feature so let's get backend to report correct size instead.

[Alexei Filippov]

I think it might still be an issue on our side. Assuming that sizes of different components are taken a [slightly] different time moments, it might happen that process memory size were reported before a big allocation attributed to a component occurred and its new size reported.

[Yury Semikhatsky]

(In reply to comment #6)
> I think it might still be an issue on our side. Assuming that sizes of different components are taken a [slightly] different time moments, it might happen that process memory size were reported before a big allocation attributed to a component occurred and its new size reported.

We take memory snapshot synchronously and there shouldn't be such big differences and even if we observe ones how can we tell if it the total reported size is wrong or a sum of partitions having incorrect size is incorrect?

[Alexei Filippov]

(In reply to comment #7)
> (In reply to comment #6)
> > I think it might still be an issue on our side. Assuming that sizes of different components are taken a [slightly] different time moments, it might happen that process memory size were reported before a big allocation attributed to a component occurred and its new size reported.
> 
> We take memory snapshot synchronously and there shouldn't be such big differences and even if we observe ones how can we tell if it the total reported size is wrong or a sum of partitions having incorrect size is incorrect?
Yes, we do it synchronously, but is the renderer process single threaded?
Regarding what component is incorrect. It's not really incorrect, it's rather a bit outdated (if we're talking about the scenario I described above), so reporting either is ok. Taking into account I cannot determine which component size is updated it's easier to just update the total.

[Brian Burg]

Closing as invalid, as this bug pertains to the old inspector UI and/or its tests.
Please file a new bug (https://www.webkit.org/new-inspector-bug) if the bug/feature/issue is still relevant to WebKit trunk.