1949834 - Content analysis warn response cancels timeout for other parts

Status: RESOLVED FIXED

(Firefox :: Data Loss Prevention, defect)

Description

[David Parks [:handyman]]

Bug 1949390 cancelled timeout when we get a warn response -- but it cancels timeout for the entire user action. So if another part of the user action takes longer than the timeout period, it will not timeout. Instead of the way it was done, it should not cancel the timer but, instead, check if there are outstanding requests when the timer goes off (vs. all requests either have responses or are showing WARN dialogs). The way to make that determination is in flux (bug 1949166) but it would happen here.

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1949390

Comment 2

[David Parks [:handyman]]

Attachment: Bug 1949834: Make timeouts work with warn r=#dlp-reviewers!

We can't just stop the timer when we get a warn response, and we can't just
cancel all warn dialogs when the timeout timer triggers. This patch means
we only cancel the timer when all final responses are in (this implies
no outstanding warn responses). It also means that we only cancel warn
dialogs on timeout if some request in the user action is not resolved
(neither with a warn nor a final response).

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7b1541544409
Make timeouts work with warn r=dlp-reviewers,gstoll

Comment 6

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/7b1541544409

Comment 7

[Cosmin Sabou [:CosminS]]

https://hg.mozilla.org/mozilla-central/rev/7b1541544409

Comment 8

[Alexandru Trif, Desktop Test Engineering [:atrif]]

Hello! I verified that I can no longer reproduce bug 1950306 and bug 1950309 with Firefox 137.0a1 (20250227090415) on WIndows 10x64. Should we look for anything else here? Thank you!

Comment 9

[David Parks [:handyman]]

Thanks atrif. There aren't really any special cases to test for this, just the usual timeout behavior should do.

Comment 10

[Alexandru Trif, Desktop Test Engineering [:atrif]]

(In reply to David Parks [:handyman] from comment #9)

Thanks atrif. There aren't really any special cases to test for this, just the usual timeout behavior should do.

Thank you! I have created these test cases (C2944323/ C2944324/ C2944325) for the TimeoutResult policy. Can either of you please let me know if these are correct, or if there is anything else I should add? Thank you!

Comment 11

[Greg Stoll :gstoll]

Those look good to me - I'll let :handyman take a look as well. Thanks!

Comment 12

[David Parks [:handyman]]

LGTM!