[Hellzed]

Here's the backtrace. I don't currently have the bandwidth to upload a core dump.

Code: [Select]

#0  0x00007fffe9fd18ce in ?? () from /usr/lib/x86_64-linux-gnu/dri/i965_dri.so
#1  0x00007fffe9dfb866 in ?? () from /usr/lib/x86_64-linux-gnu/dri/i965_dri.so
#2  0x00007fffe9dfba5a in ?? () from /usr/lib/x86_64-linux-gnu/dri/i965_dri.so
#3  0x00000000004db71b in gr_opengl_save_screen ()
    at graphics/gropengl.cpp:1028
#4  0x00000000007ce20f in popup_do (pi=0x14a63a0 <Popup_info>, flags=67175424)
    at popup/popup.cpp:821
#5  0x00000000007ce8a9 in popup (flags=67175424, nchoices=3)
    at popup/popup.cpp:1013
#6  0x00000000005de142 in player_tips_popup () at menuui/playermenu.cpp:1347
#7  0x00000000005c4e2d in main_hall_do (frametime=0,25)
    at menuui/mainhallmenu.cpp:840
#8  0x0000000000417aa2 in game_do_state (state=1)
    at freespace2/freespace.cpp:6466
#9  0x00000000004c1acf in gameseq_process_events ()
    at gamesequence/gamesequence.cpp:409
#10 0x0000000000418a58 in game_main (cmdline=0x2261cc0 "")
    at freespace2/freespace.cpp:7062
#11 0x0000000000418c58 in main (argc=1, argv=0x7fffffffdfa8)
    at freespace2/freespace.cpp:7196


[jr2]

Can you 7zip a coredump to a manageable size using the ultra compression settings?  I Googled and apparently, there is possibly also some built in compression options that can be configured for creating core dumps.  Not sure if it will bring the file down to manageable size though.

[Hellzed]

The core dump is not that big in our case :
https://www.dropbox.com/s/fg5znkhc8dzsc6z/core

(does it change anything that it's been compiled with llvm ?)

2nd core, this time with a g++ compiled fs2_open : https://www.dropbox.com/s/3ewifa9t5yzmsph/core_g%2B%2B

[Echelon9]

Although support for Intel integrated graphics has been patchy at best, there has been a recent influx of crash reports on Ubuntu 13.10 (64bit) with this hardware.

Investigating to see if there is a recent change in the underlying platform which can be worked around.

Please follow Mantis #2988 for updates.

[niffiwan]

I've added a command line flag to disable the use of Pixel Buffer Objects.  I suspect that there's a bug in the Intel Linux drivers, or they aren't quite following the OpenGL specification.

Anyway: could you please try compiling & running this code to see if that works around the issue?
https://github.com/niffiwan/fs2open.github.com/commits/mantis2988

You will need to set "-disable_pbo" on the command line, or via the launcher (Advanced -> Troubleshooting -> "Disable OpenGL Pixel Buffer Objects")

[Hellzed]

Still crashes. I'm uploading a core dump right now. Sorry, wrong binary.
Ok, it works around the issue.

(By the way, on a completely different issue, any idea on what this could be ?
http://www.hard-light.net/forums/index.php?topic=83318.msg1725941#msg1725941 )

[niffiwan]

Hurrah & thanks for testing!   I'll see about getting the workaround reviewed & committed to trunk soon.  Or alternatively I'll see if there's some way to reliably auto-detect the use of an Intel driver with the issue & disable PBO automatically...

[Hellzed]

Is this issue somewhere on the intel bugtracker ?
I'm using the xorg-edgers ppa (graphics stack compiled from git for Ubuntu), so if it's a bug in the intel drivers, I'd like to follow this issue and test again once it's fixed in the driver code.

[niffiwan]

I haven't reported the bug upstream yet, I'd like to provide a minimal test case for them but I've got a few other things I need to sort out before moving on to that.

[el mariachi]

This also works for me. Thanks!

[notung]

I've added a command line flag to disable the use of Pixel Buffer Objects.  I suspect that there's a bug in the Intel Linux drivers, or they aren't quite following the OpenGL specification.

Anyway: could you please try compiling & running this code to see if that works around the issue?
https://github.com/niffiwan/fs2open.github.com/commits/mantis2988

You will need to set "-disable_pbo" on the command line, or via the launcher (Advanced -> Troubleshooting -> "Disable OpenGL Pixel Buffer Objects")

I have the same problem (intel integrated graphics card using i965 driver). However I cannot find the commit in git hub. Is it merged already upstream? Otherwise could you please put the patch again or a new branch?

I am using binary version 3.7.0.

Thanks!

[niffiwan]

That link was a branch in my forked repo which I removed once the patch was applied upstream:
https://github.com/scp-fs2open/fs2open.github.com/commit/84728de58716e2ecc0b5098270a5bf5cb29327bd

Also, since FSO is still using SVN as the primary repo (and github is just a clone for now) you can find the same commit here as well:
http://svn.icculus.org/fs2open?limit_changes=0&view=rev&sortby=file&revision=10268

Lastly, the patch is also in the 3.7.2RC1 release
http://www.hard-light.net/forums/index.php?topic=87500.0

[diribative]

I haven't used FS2 Open in many years, so forgive me if I'm missing something obvious, but I think I'm facing the same problem. I tried selecting "fs2_open_3_7_2_RC1.exe" in the launcher, but it still crashes when I try to start with, or delete, a profile. Is there a way I can wipe my profiles clean to confirm there's not a problem along those lines that's interfering? Should I make a new thread? Thanks for any instructions/help you can provide.

Edit: It seems I *was* missing something obvious: this thread is about Linux and I'm on Windows 8.
Edit2: http://www.hard-light.net/forums/index.php?topic=87500.msg1747346#msg1747346 mentions that "Launcher 5.5g" doesn't work on Windows8. I'll have to get wxlauncher instead.

[niffiwan]

That sounds like a wxlauncher issue, so yeah, a new thread in this forum would be good. Providing the wxlauncher debug log would be a good start, see the troubleshooting section of the 1st post in this thread: http://www.hard-light.net/forums/index.php?topic=67950.0