Bugzilla

Let's tentatively reclassify this as networking, since network requests seem to be the dominant source of trouble here, and I imagine the fix would be some sort of mitigation in networking code.

(In Florian's profile from comment 23, he only had the tab open for 1 second, but that gave the testcase an opportunity to generate over 12,000 HTTP requests for https://hskupin.info/tmp/hang-on-load/count.js , which then locked up the parent process for a while, as he noted.  Presumably the site's JS is explicitly making those requests; but hopefully there's something we can do to limit the impact on other tabs & on the browser frontend, by making the parent process service those requests less-eagerly, possibly after some threshold is reached if we want to keep things eager/snappy at first?)

Let's tentatively reclassify this as networking, since network requests seem to be the dominant source of trouble here, and I imagine the fix would be some sort of mitigation in networking code.

(In Florian's profile from comment 23, he only had the tab open for 1 second, but that gave the testcase an opportunity to generate over 12,000 HTTP requests for https://hskupin.info/tmp/hang-on-load/count.js , which then locked up the parent process for a while, as he noted.  Presumably the site's JS is explicitly making those requests; but hopefully there's something we can do to limit the impact on other tabs & on the browser frontend, by making the parent process service those requests less-eagerly.  Maybe we'd only apply a mitigation after some threshold is reached, too, so that there'd only be a noticeable change-in-behavior for pathological cases like the attached testcase?)