Cause of the leak is not known (at least to me). I was able to find combination of 3 tests that cause minimum of leaks. Running any combination of just two tests from that group doesn't leak. I believe that I could find more then just this one combination... We should fix that leak because just adding new virtual hosts to server-location.txt file rises the final leak and we have to rise the threshold with it.

Given bug 473845 fixes all the other leaks, this would be the last thing to do to get to zero browser-chrome leaks for good, which makes leak detection and debugging significantly easier for everyone. This component is as good as any (that has blocking flags) until we know where the problem lies...

Bug 435743 comment 29: { From Dave Townsend (Mossop) 2009-01-21 05:16:17 PST This is explained in bug 441359 comment 24 } Bug 441359 comment 24: { every server-locations.txt entry marked "privileged" has a principal stored in the nsScriptSecurityManager. The browser chrome tests leak nsScriptSecurityManager, so any new privileged entry shows up as a leaked principal/uri. }

Bug 473845 fixes all the browser-chrome leaks. I'm not entirely sure why the other bugs are dependencies of this.

Confirming that patch from bug 473845 fixes all the leaks. This bug can be closed after that bug has been fixed.

Fixed bug bug 473845.