View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0005254 | Spring engine | General | public | 2016-05-26 16:09 | 2016-05-27 14:59 |
| Reporter | DeinFreund | Assigned To | abma | ||
| Priority | normal | Severity | crash | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Product Version | 101.0+git | ||||
| Summary | 0005254: Sudden crashes | ||||
| Description | The latest spring dev crashes in the first few minutes of each game, without any apparent reason. The games are AI only. I'll attach a infologs of different crashes, demos are not being written. http://hastebin.com/ekucugahog.xml http://hastebin.com/ogagotetun.xml http://hastebin.com/ubesirisix.xml http://hastebin.com/ojagaxesuw.xml | ||||
| Tags | No tags attached. | ||||
| Checked infolog.txt for Errors | |||||
|
|
Is there any way I can get those stack traces to show files/line numbers? Or some other way to get more verbose logging regarding the cause of crash? |
|
|
https://springrts.com/wiki/Development:Getting_Started#Buildbot_and_Stacktrace_translator |
|
|
Ok, here are the translated stacktraces: http://hastebin.com/zarebojilo.sm rts/lib/lua/include/LuaUser.cpp:191 is on two of them. |
|
|
Sometimes there are units teleporting around before the game crashes, maybe some memory glitch caused by too much pointer juggling? Trace + Infolog: http://paste.springfiles.com/view/281523d3 |
|
|
are you sure you're not just out of memory? |
|
|
https://i.imgur.com/HcL0mQG.png The drop is Spring crashing. I don't think memory is limiting here. |
|
|
weird, maybe your AI has memory issues? |
|
|
Ok, the issue is related to the java interface. It doesn't happen when my AI isn't playing. |
|
|
AFAIK spring 32 bit is limited to <4GB Memory. -> memory usage of spring is relevant, not the available memory. |
|
|
the bump in the image seems <4 GB |
|
|
from infolog.txt: <SkirmishAI: CSI (team 2)>: Initialized thread 1 are you sure the spring api isn't accessed outside of the spring-mainthread? the ai interface isn't threadsafe! (!!!) |
|
|
also: can i/we reproduce this issue somehow? where is the sourcecode? |
|
|
an other paste from your provided infolog.txt: [f=0001410] Error: [ErrorMessageBox][1] msg="not enough memory" IsMainThread()=1 fromMain=0 |
|
|
I've been replacing/turning off all the multithreaded parts now but the crashes still happen. I'll check memory consumption on next run to see if maybe the 4 GB limit causes problems. |
|
|
Another crash, but Spring never went above 1.2 GB. I couldn't see any big spikes in other runs either. Is there maybe a way to log all JavaInterface calls? |
|
|
If you want to try to reproduce, I'm playing CSI vs CAI on Comet Catcher Redux, latest Zero-K Stable. AI build: http://www.mediafire.com/download/44fbnguc2t1648n/zkcbai.zip AI source: http://www.mediafire.com/download/cgouaoj3qkv0x4c/src.zip The pathfinder precalculation is still being done multithreaded but that doesn't cause the crashing. The crash can take up to 10 minutes to occur. Originally multithreaded classes are PathPrecalculaterThread.java, BackgroundPathfinder.java, BuilderHandler.java. |
|
|
AI crashes first: http://paste.springfiles.com/view/raw/af56dea8 then spring hangs and then spring crashes in frame -1 with "failed to allocate memory". (64 Bit Ubuntu / 16GB RAM) |
|
|
First exception is expected, it has to create that file on first run. The 2nd exception is a bit odd, but neither of those should actually crash the AI. I.e. return non-zero. I assume the AI GUI didn't load for you. I can check if it's the GUI that's causing some issues, because AWT runs in it's own thread by default. |
|
|
Just noticed one of my newer optimizations actually causes the GUI to access the Interface :/ Will check that now. |
|
|
Ok, I can confirm that the issue was caused by the GUI thread accessing the interface. Thanks for your help in finding this. If you think it's acceptable for Spring to crash here then the issue is resolved, otherwise I'd like to see Spring throw an exception when accessed from outside the main thread. |
|
|
Checking the current thread is much more demanding than one would think. Even assertions can make the debug version barely playable. |
|
|
A simple mutex to check whether the call happens outside update/other callins would have already helped here. (Because my own debug output showed that there were interface calls during those times) This shouldn't cost more than a handful of cycles per callin. |
|
|
It will. Maybe not on linux, but definitely on windows. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2016-05-26 16:09 | DeinFreund | New Issue | |
| 2016-05-26 16:10 | DeinFreund | Note Added: 0016351 | |
| 2016-05-26 16:13 |
|
Note Added: 0016352 | |
| 2016-05-26 16:15 | DeinFreund | Note Edited: 0016351 | |
| 2016-05-26 16:19 | DeinFreund | Note Added: 0016353 | |
| 2016-05-26 16:20 | DeinFreund | Note Edited: 0016353 | |
| 2016-05-26 17:12 | DeinFreund | Note Added: 0016355 | |
| 2016-05-26 17:39 | hokomoko | Note Added: 0016356 | |
| 2016-05-26 17:39 | hokomoko | Note View State: 0016353: private | |
| 2016-05-26 17:39 | hokomoko | Note View State: 0016353: public | |
| 2016-05-26 18:07 | DeinFreund | Note Added: 0016357 | |
| 2016-05-26 18:15 | hokomoko | Note Added: 0016358 | |
| 2016-05-26 18:42 | DeinFreund | Note Added: 0016359 | |
| 2016-05-26 19:03 | abma | Note Added: 0016360 | |
| 2016-05-26 19:04 | abma | Note Edited: 0016360 | |
| 2016-05-26 19:22 | hokomoko | Note Added: 0016361 | |
| 2016-05-26 19:54 | abma | Note Added: 0016362 | |
| 2016-05-26 19:55 | abma | Note Added: 0016363 | |
| 2016-05-26 19:57 | abma | Note Added: 0016364 | |
| 2016-05-26 21:10 | DeinFreund | Note Added: 0016365 | |
| 2016-05-26 21:17 | DeinFreund | Note Added: 0016366 | |
| 2016-05-26 21:43 | DeinFreund | Note Added: 0016367 | |
| 2016-05-26 21:47 | DeinFreund | Note Edited: 0016367 | |
| 2016-05-27 07:58 | abma | Note Added: 0016368 | |
| 2016-05-27 08:05 | abma | Note Edited: 0016368 | |
| 2016-05-27 08:09 | abma | Note Edited: 0016368 | |
| 2016-05-27 08:37 | DeinFreund | Note Added: 0016369 | |
| 2016-05-27 08:40 | DeinFreund | Note Added: 0016370 | |
| 2016-05-27 09:28 | DeinFreund | Note Added: 0016371 | |
| 2016-05-27 09:28 | DeinFreund | Note Edited: 0016371 | |
| 2016-05-27 09:33 | hokomoko | Note Added: 0016372 | |
| 2016-05-27 09:50 | DeinFreund | Note Added: 0016373 | |
| 2016-05-27 09:54 | hokomoko | Note Added: 0016374 | |
| 2016-05-27 14:59 | abma | Status | new => closed |
| 2016-05-27 14:59 | abma | Assigned To | => abma |
| 2016-05-27 14:59 | abma | Resolution | open => no change required |
