Bug 206274 - Segfault with the new 64bit linux_compatibility
Summary: Segfault with the new 64bit linux_compatibility
Status: Closed Overcome By Events
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 10.2-STABLE
Hardware: amd64 Any
: --- Affects Only Me
Assignee: Dmitry Chagin
URL:
Keywords:
Depends on:
Blocks: 247219
  Show dependency treegraph
 
Reported: 2016-01-14 22:01 UTC by Stefan Rumetshofer
Modified: 2020-10-14 15:37 UTC (History)
6 users (show)

See Also:


Attachments
Output of ktrace / kdump (169.57 KB, text/plain)
2016-01-14 22:01 UTC, Stefan Rumetshofer
no flags Details
Output of kdump (rev. 294277) (99.58 KB, text/plain)
2016-01-18 18:03 UTC, Stefan Rumetshofer
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Stefan Rumetshofer 2016-01-14 22:01:42 UTC
Created attachment 165603 [details]
Output of ktrace / kdump

When trying to run the 64bit X-Plane installer it ends with a segfault.

System is a 10.2-STABLE with the linux64.ko module loaded and the nvidia driver installed. 64bit libraries for the driver are in place.

Now i have run ktrace / kdump. For the results see the attachment
Comment 1 Dmitry Chagin freebsd_committer freebsd_triage 2016-01-17 15:40:13 UTC
take
Comment 2 Dmitry Chagin freebsd_committer freebsd_triage 2016-01-18 08:37:01 UTC
(In reply to Stefan Rumetshofer from comment #0)

Stefan, can you, please, update to the latest stable/10 (you need at least r294135) and rebuild and reinstall linux64.ko and usr.bin/kdump, then do
kdump -HAR -m 128 over ktrace.out

your dump looks odd:)
Comment 3 Stefan Rumetshofer 2016-01-18 18:03:28 UTC
Created attachment 165770 [details]
Output of kdump (rev. 294277)
Comment 4 Stefan Rumetshofer 2016-01-18 18:08:36 UTC
(In reply to Dmitry Chagin from comment #2)

After updating to rev. 294277 i rebuilt world and kernel and loaded the linux64.ko. The output is in the attachment.
Comment 5 Dmitry Chagin freebsd_committer freebsd_triage 2016-01-18 18:19:44 UTC
(In reply to Stefan Rumetshofer from comment #4)

Thank you, now I see where the problem is:

   820 100219 X-Plane 10 Installe 0.000006 CALL  L64  linux_arch_prctl(0x1002,0x21cf5a0)
   820 100219 X-Plane 10 Installe 0.000002 RET   L64  linux_arch_prctl  0

seems that SET_FS failed
Comment 6 Stefan Rumetshofer 2016-01-19 17:55:08 UTC
(In reply to Dmitry Chagin from comment #5)

Any chance to fix it?
Comment 7 Stefan Rumetshofer 2016-01-25 06:42:32 UTC
After some testing with the great help from Dmitri it seems that the problem comes from the nvidia driver respectively the 64bit libs. Dmitri and i also tested the installer with the intel HD 3000 and the intel driver, and with this configuration the program doesn't segfault. On my system it quits because of to less VRAM but on Dmitris system the installer is executed.
Now i dont know if this problem affects only this installer or other programs too.
Comment 8 Eitan Adler freebsd_committer freebsd_triage 2018-05-28 19:48:00 UTC
batch change:

For bugs that match the following
-  Status Is In progress 
AND
- Untouched since 2018-01-01.
AND
- Affects Base System OR Documentation

DO:

Reset to open status.


Note:
I did a quick pass but if you are getting this email it might be worthwhile to double check to see if this bug ought to be closed.
Comment 9 Edward Tomasz Napierala freebsd_committer freebsd_triage 2020-06-13 14:51:27 UTC
Stefan, can you still reproduce it?
Comment 10 Stefan Rumetshofer 2020-06-14 11:27:28 UTC
Since i switched from NVIDIA to AMD i can not reproduce this problem. With the Radeon the installer runs fine.
Now i am on 12.1-RELEASE with the drm-kmod package and linux-c7. Also there is a newer version of X-Plane. It's been a while since i tried this version. Afair the installer also runs fine but the simulator itself does'nt run.
Comment 11 Alex S 2020-08-07 18:44:43 UTC
OpenGL with Nvidia used to be broken due to bug #206711. I don't remember whether that issue appeared as a segfault, though. For the game proper see bug #174933, comment #4.

I suggest closing this bug with Overcome by Events.