1392272 - [Win] Optimize the way we resolve proxy settings

Status: RESOLVED FIXED

(Core :: Networking: HTTP, enhancement, P3)

Description

[Honza Bambas (:mayhemer)]

With the default prefs we do async proxy resolution for every channel.  Most users these days have Windows set to "no proxy" in the Windows internet settings.  We should detect changes by observing for either registry changes or doing some other magic and bypass proxy resolution completely when no proxy has been specified.

I can see 20+ ms (on a fast machine, in an opt build!) on EVERY channel.  This also includes main thread dispatch.

Not sure we could have this for 57, but would definitely be nice.

Loosely blocking CDP.

https://stackoverflow.com/questions/6192563/detect-windows-ie-proxy-settings-changes

Comment 1

[Honza Bambas (:mayhemer)]

One simple solution (some probability of regression prove..) could be to refresh the registry settings only when loading a top level document and not for sub-resources.

Comment 2

[Gary Chen [:xeonchen]]

Assign to myself, but maybe fix later, so keep necko-next.

Comment 3

[Firefox Bug Husbandry Bot]

Bulk change to priority: https://bugzilla.mozilla.org/show_bug.cgi?id=1399258

Comment 4

[Kershaw Chang [:kershaw]]

Attachment: Bug 1392272 - P1: [windows] Monitor system proxy changes, r=#necko

Comment 5

[Kershaw Chang [:kershaw]]

Attachment: Bug 1392272 - P2: [osx] Monitor system proxy changes, r=#necko

Depends on D127724

Comment 7

[Kershaw Chang [:kershaw]]

Add leave-open, since linux and android parts are not completed.

Comment 8

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/1c1c60a1f6ef
P1: [windows] Monitor system proxy changes, r=necko-reviewers,dragana
https://hg.mozilla.org/integration/autoland/rev/931db25d14fd
P2: [osx] Monitor system proxy changes, r=necko-reviewers,dragana

Comment 9

[Narcis Beleuzu [:NarcisB]]

https://hg.mozilla.org/mozilla-central/rev/1c1c60a1f6ef
https://hg.mozilla.org/mozilla-central/rev/931db25d14fd

Comment 10

[Mathew Hodson]

I think it would be better to close this bug since patches landed in Firefox 96 and open new bugs for Linux and Android.

Comment 11

[Kershaw Chang [:kershaw]]

(In reply to Mathew Hodson from comment #10)

I think it would be better to close this bug since patches landed in Firefox 96 and open new bugs for Linux and Android.

Note that the patches in this bug was reverted because of bug 1749501.
I'll file a new bug to track the necessary work.