I see this happening on a nexus4 with debug-gecko and current trunk/master. The phone was idle over the weekend. 11-16 04:14:38.027 F/MOZ_Assert( 1678): Assertion failure: end >= start, at ../../../dom/base/nsJSEnvironment.cpp:1973 11-16 04:14:38.027 F/libc ( 1678): Fatal signal 11 (SIGSEGV) at 0x00000000 (code=1), thread 1678 (Messages) The Messages app wants to log something: 11-16 04:14:38.207 I/DEBUG ( 171): stack: 11-16 04:14:38.207 I/DEBUG ( 171): bef53ba8 b65e7e07 /system/b2g/libxul.so 11-16 04:14:38.207 I/DEBUG ( 171): bef53bac b5a4904f /system/b2g/libxul.so (NS_LogRelease+14) 11-16 04:14:38.207 I/DEBUG ( 171): bef53bb0 b34fe6f0 11-16 04:14:38.207 I/DEBUG ( 171): bef53bb4 00000001 11-16 04:14:38.207 I/DEBUG ( 171): bef53bb8 00000003 11-16 04:14:38.207 I/DEBUG ( 171): bef53bbc ed906943 11-16 04:14:38.207 I/DEBUG ( 171): bef53bc0 4239c4e1 11-16 04:14:38.207 I/DEBUG ( 171): bef53bc4 0004eb4a 11-16 04:14:38.207 I/DEBUG ( 171): bef53bc8 00000000 11-16 04:14:38.207 I/DEBUG ( 171): bef53bcc b52cb2ab /system/b2g/libxul.so 11-16 04:14:38.207 I/DEBUG ( 171): bef53bd0 b63ce79c /system/b2g/libxul.so TimeBetween is only called for GC and CC logging.

I think PRTime isn't guaranteed to be monotonic, so we're just going to hit this sometimes. Maybe it should just be removed.

(In reply to Andrew McCreight [:mccr8] from comment #1) > I think PRTime isn't guaranteed to be monotonic, so we're just going to hit > this sometimes. Maybe it should just be removed. Or change it to a warning?

This should be fixed by patch 2 of bug 948554, which converts this method to use TimeStamp, which I think is guaranteed to be monotonic.

well, hopefully this is okay now.