Based on my retriggering on try, about 20-30% of XUL Fennec and 50% of native Fennec crashtest-1 runs time out. Despite the timeout hitting at around the same test every time, disabling the last few tests, or even every test in that directory, just pushes the timeout a few tests later. It would pay to sanity check whether it helps in staging first, but that probably means it's just more work than a Tegra can do without taking a nap, and it's time to split it in 3.
I should have bounced this back to the queue earlier
Created attachment 594380 [details] [diff] [review] fix Conveniently, there's really no need for staging. I got fed up with C1, and hid it on m-i and m-c, and I'll hide it anywhere else I have to look as I have to look, so there's zero risk to taking an untested patch, because there's nowhere to go but up now.
Comment on attachment 594380 [details] [diff] [review] fix http://hg.mozilla.org/build/buildbot-configs/rev/48d40e93397d
hidden: m-i, m-c, m-a, s-c, fx-team
I don't see C3 on https://tbpl.mozilla.org/?noignore=1, just c1 and c2. Do we need a reconfig for this?
Two, rather than just one, actually.
Not the most profound of effects, but fixed nonetheless.