I suggest to temporarily go back to NSS 3.15.3 RTM on mozilla-central, and temporarily disable the calls in PSM to new NSS APIs. Will attach a patch.
https://hg.mozilla.org/integration/mozilla-inbound/rev/a288dd7ac0bb https://hg.mozilla.org/integration/mozilla-inbound/rev/5acd6a53ef97 <briansmith> kaie: I agree with you that we should back out the recent update of NSS <kaie> briansmith, ok cool. want me to push it? <briansmith> wait a sec * kaie waits a sec :) <briansmith> question: should we update to NSS 3.15.3 RTM, or just revert the update to NSS 3.15.4 beta 2? <briansmith> I think we should just backout the update to NSS 3.15.4 beta 2, to restore things to how it was previously <briansmith> and if that fixes things then update to NSS 3.15.3 RTM <kaie> briansmith, if we reverted that update, we'd introduce a snapshot that calls itself a 3.15.3-beta - I'm a bit worried about the confusion that could create <kaie> but I understand that it reduces the set of software that might be responsible for the crashes. <kaie> hmm... <briansmith> I don't think we should worry so much about that, given that it is just a few days <briansmith> I think the way I suggest is better because we can bisect changes <kaie> briansmith, ok. feel free to go ahead with your plan. <briansmith> OK. <briansmith> I am going to do it in three changesets: <briansmith> 1. hg backout 8dbd3f432835 (this will back out bug 932176) <briansmith> 2. hg backout the change that updated to NSS 3.15.4 beta 2 <briansmith> actually, I guess just two changes <kaie> ok <briansmith> if that doesn't fix things then it is likely a problem with TLS 1.2 <briansmith> OK. Thanks for your work on all this! <kaie> you can quote me r=kaie for backing out, if necessary
Assignee: nobody → brian
Severity: normal → critical
Status: NEW → ASSIGNED
Priority: -- → P1
Target Milestone: --- → mozilla28
<briansmith> BTW, there is a step three: back out NSS 3.15.4 beta 1 https://hg.mozilla.org/integration/mozilla-inbound/rev/9f5ebea82f7e
This was undone after investigation revealed that NSS 3.15.4 couldn't have caused the regressions since the regressions started before NSS 3.15.4 landed. See bug 898431 comment 28.
Status: ASSIGNED → RESOLVED
Last Resolved: 5 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.