2019-09-18 17:52 CEST

View Issue Details Jump to Notes ]
IDProjectCategoryView StatusLast Update
0005504Spring engineGeneralpublic2017-05-30 17:14
ReporterGoogle_Frog 
Assigned Toabma 
PrioritynormalSeverityminorReproducibilityhave not tried
StatusresolvedResolutionunable to reproduce 
Product Version103.0 +git 
Target VersionFixed in Version 
Summary0005504: 103.0.1-712 Explain unusual exit codes
DescriptionI receive infologs that have detected a crash based on an unusual exit code. The infologs do not otherwise look like the game has crashed. What do these exit codes, should Spring be sending them and do they require a fix somewhere?

These infologs have exit code -805306369 and often have fragment shader spam:
 * https://github.com/ZeroK-RTS/CrashReports/issues/532
 * https://github.com/ZeroK-RTS/CrashReports/issues/543
 * https://github.com/ZeroK-RTS/CrashReports/issues/567
 * https://github.com/ZeroK-RTS/CrashReports/issues/568
 * https://github.com/ZeroK-RTS/CrashReports/issues/569
 * https://github.com/ZeroK-RTS/CrashReports/issues/570
 * https://github.com/ZeroK-RTS/CrashReports/issues/571
 * https://github.com/ZeroK-RTS/CrashReports/issues/573

Exit code 1:
 * https://github.com/ZeroK-RTS/CrashReports/issues/535
 * https://github.com/ZeroK-RTS/CrashReports/issues/553

Exit code -1:
 * https://github.com/ZeroK-RTS/CrashReports/issues/546
TagsNo tags attached.
Checked infolog.txt for lua Errors
Attached Files

-Relationships
related to 0005589closedKloot 103.0.1-980 Freeze with truncated infolog and exit code -1073740791 
+Relationships

-Notes

~0017393

Kloot (developer)

Last edited: 2017-03-28 13:23

View 4 revisions

Spring currently only ever sets the exit-code to 0, 1 (server timeout, which explains the "Unable to read data from the transport connection: An existing connection was forcibly closed by the remote host" line in 553), or -1 (potential desync; harmless unless there are actual sync errors). This may as well be removed since it doesn't add useful information.

-805306369 looks like a bug in whatever part of ZK's infrastructure parses it from the log, but should probably be 1 considering the context (532/543/569/570/571/573: "ZkData.TcpTransport socket disconnected: Von der Übertragungsverbindung können keine Daten gelesen werden", 567: "Impossible de lire les données de la connexion de transport") says the same thing in German and French.

"was successfully compiled to run on hardware" does not occur in the engine's source but is returned by certain drivers, which confuses logic in ZK that assumes non-empty shader logs indicate errors.

~0017394

Google_Frog (reporter)

The exit code seems to be read directly from the process: https://github.com/ZeroK-RTS/Zero-K-Infrastructure/commit/49207e1904276b4a4781a9492edba26a99ba1818

I'll ask Licho about it.

~0017396

abma (administrator)

grepped trough the web:

-805306369 -> 0xCFFFFFFF ->

https://msdn.microsoft.com/en-us/library/windows/hardware/ff558784%28v=vs.85%29.aspx

STATUS_APPLICATION_HANG

very likely windows kills the process because it "hangs". so spring itself can't return an exit code.

~0017397

abma (administrator)

the reports with -805306369 seems to have in common old gfx / low gfx memory.

~0017399

Kloot (developer)

Last edited: 2017-03-28 13:38

View 2 revisions

I think the primary common factor there is crappy (wifi?) connections.

~0017400

abma (administrator)

because of "ZkData.TcpTransport socket disconnected:" ?

bad wifi shouldn't cause hangs?!

~0017402

abma (administrator)

Last edited: 2017-03-28 14:32

View 2 revisions

hmm, maybe we should distinguish between infolog.txt which end with

[f=0099959] End widgetHandler:Shutdown (which is from one individual it seems)

and
https://github.com/ZeroK-RTS/CrashReports/issues/567 +
https://github.com/ZeroK-RTS/CrashReports/issues/568

~0017403

Kloot (developer)

not sure, but many of these are apparent hangs during shutdown/reloading which should trigger Spring's hang-detector long before the OS steps in.

~0017404

abma (administrator)

i see

  HangTimeout = 0

->
  [WatchDog::Install] disabled

in infolog.txt :-)

without a stacktrace i don't see what we can do at engine side :)
+Notes

-Issue History
Date Modified Username Field Change
2017-03-28 07:34 Google_Frog New Issue
2017-03-28 12:49 Kloot Note Added: 0017393
2017-03-28 12:51 Kloot Note Edited: 0017393 View Revisions
2017-03-28 12:58 Google_Frog Note Added: 0017394
2017-03-28 13:01 Kloot Note Edited: 0017393 View Revisions
2017-03-28 13:08 abma Note Added: 0017396
2017-03-28 13:14 abma Note Added: 0017397
2017-03-28 13:23 Kloot Note Edited: 0017393 View Revisions
2017-03-28 13:37 Kloot Note Added: 0017399
2017-03-28 13:38 Kloot Note Edited: 0017399 View Revisions
2017-03-28 13:43 abma Note Added: 0017400
2017-03-28 14:30 abma Note Added: 0017402
2017-03-28 14:32 abma Note Edited: 0017402 View Revisions
2017-03-28 14:36 Kloot Note Added: 0017403
2017-03-28 15:26 abma Assigned To => abma
2017-03-28 15:26 abma Status new => resolved
2017-03-28 15:26 abma Resolution open => unable to reproduce
2017-03-28 15:26 abma Note Added: 0017404
2017-05-30 17:14 abma Relationship added related to 0005589
+Issue History