1368048 - Crash reports with same uptime, install date, etc [tb]

Status: RESOLVED INVALID

(Socorro :: General, task)

Description

[Wayne Mery (:wsmwk)]

I do not know yet know whether this is a socorro bug, or just an artifact of this particular crash signature from bug 1368045

Funkieness in these ~200 crash reports makes socorro think there is just one user reporter[1], because they ALL have
Uptime	11,528,028,864 seconds (366 years, 28 weeks and 6 days)
Last Crash	-9,496,989,952 seconds before submission
Install Age	-1,356,873,024 seconds since version was first installed
Install Time	2009-04-22 19:24:48 

But from examining email addresses there are clearly multiple users reporting. For example
bp-e2e4b764-3585-4f5a-beb3-32d1b0170524
bp-19bc84ec-77d7-47bc-87ca-068830170521
bp-cf5e835e-2e7c-4018-ab3b-8178e0170524
bp-f6289f25-884d-43d1-89bb-4f2b10170526

[1] 
* https://crash-stats.mozilla.com/topcrashers/?product=Thunderbird&version=52.1.1&_facets_size=100
* https://crash-stats.mozilla.com/signature/?product=Thunderbird&release_channel=%21release&signature=icaltime_from_timet_with_zone&date=%3E%3D2017-02-26T13%3A14%3A22.000Z&date=%3C2017-05-26T13%3A14%3A22.000Z&_columns=date&_columns=version&_columns=user_comments&_columns=uptime&_columns=install_time&_sort=version&_sort=email&_sort=-date&page=1

Comment 1

[Peter Bengtsson [:peterbe]]

Is that what happens when an individual user has a crazy clock on their computer?

Aren't all those numbers based on the raw crash which is based on Breakpad using the host's clock.

Comment 2

[Wayne Mery (:wsmwk)]

(In reply to Peter Bengtsson [:peterbe] from comment #1)
> Is that what happens when an individual user has a crazy clock on their
> computer?

Dunno, but the "uptime" values only occur with version 52, it must be code related, no? memory corruption?
https://crash-stats.mozilla.com/search/?uptime=%3E%3D11528028864&email=~%40&product=Thunderbird&date=%3E%3D2017-02-28T17%3A18%3A36.000Z&date=%3C2017-05-31T17%3A18%3A36.000Z&_sort=-date&_facets=signature&_facets=email&_columns=date&_columns=signature&_columns=version&_columns=email#crash-reports

Comment 3

[Peter Bengtsson [:peterbe]]

I actually don't know how those numbers work in the processor. 
Perhaps Adrian can help. 

Also, that search URL you pasted yields nothing for me.

Comment 4

[Wayne Mery (:wsmwk)]

sorry, posted wrong link 
https://crash-stats.mozilla.com/search/?uptime=%3E%3D11528028864&product=Thunderbird&date=%3E%3D2017-02-28T17%3A18%3A36.000Z&date=%3C2017-05-31T17%3A18%3A36.000Z&_sort=-date&_facets=signature&_columns=date&_columns=signature&_columns=version&_columns=user_comments#crash-reports

Comment 5

[Peter Bengtsson [:peterbe]]

Adrian, 
Can you explain how "Uptime" works?
I saw that in https://crash-stats.mozilla.com/rawdumps/e2e4b764-3585-4f5a-beb3-32d1b0170524.json there is:
"UptimeTS": "12.414412644" and "StartupTime": "-11644473600"

Quickly grepping around in the processor code it seems that uptime is the result of "crash_time - startup_time". So if the StartupTime is a massive negative number then the uptime is a *massive* positive number. 

Is the bug here that Breakpad sends a huge negative number for StartupTime?
If so, we need to redirect the attention of this bug away from SuperSearch and the webapp and investigate why Breakpad is sending those odd numbers.

Comment 6

[[DEACTIVATED] Adrian Gaudebert]

This is very likely due to bogus data sent from the crash reporter.

Comment 7

[Wayne Mery (:wsmwk)]

I'll take that is "Invalid" (at least for socorro)