1920697 - Firefox aborts requests running for more than 30s with status code (null)

Status: RESOLVED DUPLICATE of bug 1914309

(Core :: Networking, defect)

Description

[Simon Brami]

Steps to reproduce:

I ran a POST request to an HTTPS server, the request is expected to take more than 30 seconds

User agent:
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:130.0) Gecko/20100101 Firefox/130.0

Actual results:

The request is aborted with the console message:
Cross-Origin Request Blocked: The Same Origin Policy disallows reading the remote resource at https://..... (Reason: CORS request did not succeed). Status code: (null)

The server CORS configuration is correct, and the preflight request passed. It is working fine on Chrome and Safari.

Expected results:

The request should have completed correctly without cancelling after 30s.

Comment 1

[Simon Brami]

I just tried with Firefox 126.0 and it works, so there's definitely a bug that has been introduced in more recent versions.
The user agent working:
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:126.0) Gecko/20100101 Firefox/126.0

Comment 2

[Francesco Lodolo [:flod]]

Would you be able to try use mozregression to identify the change that cause this problem?
https://mozilla.github.io/mozregression/quickstart.html

Comment 3

[Kershaw Chang [:kershaw]]

Please see comment #2. Thanks.

Comment 4

[Simon Brami]

Hey, found this revision using mozregression : https://phabricator.services.mozilla.com/D212540

Comment 5

[Kershaw Chang [:kershaw]]

(In reply to Simon Brami from comment #4)

Hey, found this revision using mozregression : https://phabricator.services.mozilla.com/D212540

Could you please capture an HTTP log? Note that in about:logging, please set Logging preset to HTTP/3 and select logging to file, and send the file to [e-mail].
Thanks.

Comment 6

[Simon Brami]

Done !

Comment 7

[Valentin Gosu [:valentin] (he/him)]

Kershaw, could you have a look at the logs? Thanks!

Comment 8

[Valentin Gosu [:valentin] (he/him)]

Assuming that the bug does indeed happen only with HTTP/3, I do see we made some changes to the keep_alive code in neqo in https://phabricator.services.mozilla.com/D212097

Comment 9

[Kershaw Chang [:kershaw]]

This is the same as bug 1914309.
Unfortunately, this is regressed by bug 1899706.