1374721 - Normalize moz-extension UUID URIs in diffs

Status: NEW

(Toolkit :: about:memory, enhancement)

Description

[Andrew McCreight [:mccr8] (PTO until 12-2)]

I have an entry like this in my about:memory report:

compartment(moz-extension://4934f883-3f08-4a47-ad69-2a6efffa9309/, [anonymous sandbox] (from: resource://gre/modules/ExtensionParent.jsm:366))

The UUID or whatever it is should be sanitized when doing a diff.

Comment 1

[Eric Rahm [:erahm]]

Is the UUID consistent b/w runs? If so I'm not sure we want to remove it as it makes it clear which add-on is responsible for the compartment. Or is it just some random runtime generated uuid?

Comment 2

[Andrew McCreight [:mccr8] (PTO until 12-2)]

No, it isn't consistent between runs. For instance this is from a diff of AWSY runs:

│  │  ├──-0.02 MB (-1.64%) ++ compartment(moz-extension://0bc35c98-864b-47cd-a172-274e8480b255/, [anonymous sandbox] (from: resource://gre/modules/ExtensionParent.jsm:366))
│  │  ├──0.02 MB (01.64%) ++ compartment(moz-extension://df251463-934e-4eea-b2e9-7ba577db2361/, [anonymous sandbox] (from: resource://gre/modules/ExtensionParent.jsm:366))

Comment 3

[Eric Rahm [:erahm]]

(In reply to Andrew McCreight [:mccr8] from comment #2)
> No, it isn't consistent between runs. For instance this is from a diff of
> AWSY runs:
> 

Oh they must be rebuilding an extension? Or does that repro locally too? It would be nice if we could make this optional. For AWSY it definitely makes sense, for normal user reports maybe not.

Comment 4

[Andrew McCreight [:mccr8] (PTO until 12-2)]

(In reply to Eric Rahm [:erahm] (please no mozreview requests) from comment #3)
> Oh they must be rebuilding an extension? Or does that repro locally too? It
> would be nice if we could make this optional. For AWSY it definitely makes
> sense, for normal user reports maybe not.

I believe it is an anti fingerprinting measure.

Comment 5

[Kris Maglione [:kmag]]

> (In reply to Eric Rahm [:erahm] (please no mozreview requests) from comment #3)
> Oh they must be rebuilding an extension? Or does that repro locally too?

We generate a random URL UUID for each installation of the extension. There's
a separate internal ID which remains the same, though. For the sake of
about:memory and AWSY, it might make sense to mangle the URLs to use the
internal ID instead.

(In reply to Andrew McCreight [:mccr8] from comment #4)
> I believe it is an anti fingerprinting measure.

It's something along those lines, yes. Although there are actually some
circumstances where it makes us more fingerprintable (i.e., when extensions
expose resources to web content in web-visible ways).