Version: M50
OS: Linux

We received a report of an occasional negative timestamp from Inbox team. Here is the logs (Access restricted to googlers): https://docs.google.com/document/d/14OtxvEZIvI2Dbksu5_MzHRe0bqrmmeUInxd91cYtQnM/edit?ts=57238be3

The errors are reported on Linux and mainly followed what seems to be a mouse click.

Here is a summary of values: 
[160429 15:03:34.73] [16166.845s] -4278799841.17
[160429 15:03:34.85] [16166.963s] -4278799723.17
[160429 15:03:37.03] [16169.148s] -4278797543.1700006
[160429 15:03:38.00] [16170.110s] -4278796575.17

Things to note, the first timestamp's absolute is essentially "Fri Apr 29 2016 12:12:03 GMT-0400 (EDT)". This I suspect is close to page navigation start (performance.timing.navigationStart) which is also the time origin for high-resolution timestamps.

So I suspect we have a small (non-zero?) raw platform timestamp value which becomes negative after (raw_timestamp - time_origin) operation.

This is similar to  https://crbug.com/578243 .