[android] Planescape crashes on new area

[KLarsen]

Hi,

I recently installed PST on my Asus tf201, and after solving a few issues I got it working. I was able to play through the mortuary and first area of the Hive without issues, however on travelling to any new area in the hive, the game crashes to desktop. I am able to see the new area for about one second before this happen.

GemRB version 0.7.0.1 installed from Google Play

The game is installed in /app-data/net.sourceforge.gemrb/Torment/

 

I've included the output from alogcat below as well as the config file (renamed .txt so it can be uploaded).

 

Any help would be much appreciated, thanks.

alogcat.2012-05-20-16-38-45+0200.txt

GemRB.txt

[lynx]

Please upload the savegame before entering the new area. Just teleporting there from the starting area, I can't reproduce a crash. I've been talking with various citizens and nothing problematic happened so far.

Did you see the exit movie? I also got some ghost animation cutscene, but that may also be due to my shortcircuiting.

[KLarsen]

Sure, here's the link to the save: https://docs.google.com/open?id=0B-nKbAPfmDx2RVJET3dtZ2tQbTQ (it should work, though it's the first time I use Google Drive).

I saw the exit movie when exiting the mortuary, and was able to move around the first area of the Hive, talking with people etc, as well as entering buildings. The crash doesn't happen until leaving the first area for another Hive area.

[lynx]

I'll need the whole directory.

 

Both subareas and the second hive area work for me, so maybe you should just try the latest weekly build.

[KLarsen]

I've tried with the latest build, but got same result. I've now reverted back to the build on google play and am linking all the files from the autosave below.

https://docs.google.com/open?id=0B-nKbAPfmDx2bDJ2OVJhT3M2WXc

https://docs.google.com/open?id=0B-nKbAPfmDx2Vm4tSnA0WjAxZVk

https://docs.google.com/open?id=0B-nKbAPfmDx2a3pwVldEOGhZUlE

https://docs.google.com/open?id=0B-nKbAPfmDx2SE10VDFycXVkbFk

https://docs.google.com/open?id=0B-nKbAPfmDx2SjFncW9sWEJ6RDQ

https://docs.google.com/open?id=0B-nKbAPfmDx2RVJET3dtZ2tQbTQ

[lynx]

next time please just zip it up, this was tedious to download and recreate.

 

I went south through the exit, walked around a bit, talked with the damsel in distress, levelled up and nothing. Do you do anything special to get the crash? Now it looks like an android specific bug, probably due to limited memory or somesuch.

[KLarsen]

Sorry, I didn't have zip on my tablet. The game crashes as soon as I get to the new area, I don't have time to do anything at all. Is there anything I can provide to help you hunt down the bug?

[lynx]

we do have some invalid reads there, so maybe my 64bitness is saving me.

[KLarsen]

I tried installing IWD today to see if that would run, and so far it runs (almost) flawlessly, I can move between areas and in general play the game normally. I don't know if this helps any at all, I just wanted to see if the problems were specific to PST.

[lynx]

good choice, iwd can be played to the end. Don't forget to set the GameType to "how" if you have heart of winter installed, as it's a big update.

[KLarsen]

Yes, I have both HoW and TotLS installed, and the .cfg file configured to how.

 

I just experimented some more with pst: this time I WAS able to move to the South Hive. I saved the game there, reloaded the autosave and travelled to the South Hive again, though this time the game crashed. Reloading the saved game in the South Hive I tried transiting back, no problem, I could also go back south again.

I then tried going to a new area from both the Central Hive and the South Hive, the game crashed.

Could it be some problem with the cache? Perhaps it allowed me to transit to the South Hive the first time as I had recently been running IWD? I will now try running IWD again, go back to PST and go to a completely new Hive area. I'll post back later.

[KLarsen]

No, it still crashes when going back to PST, nor does clearing out the Cache folder first help.

[lynx]

the cache is cleared on each run anyway, so that was a red herring. The bug is likely related to the invalid read I saw (not fixed yet), so for now further debugging isn't needed.