Bug 16336

Summary: i915 crashes when loading X.org server with vesa driver
Product: Drivers Reporter: Artem S. Tashkinov (aros)
Component: Video(DRI - Intel)Assignee: drivers_video-dri-intel (drivers_video-dri-intel)
Status: RESOLVED WILL_NOT_FIX    
Severity: normal    
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.34 Subsystem:
Regression: No Bisected commit-id:
Attachments: Bug appearance
Bug appearance II

Description Artem S. Tashkinov 2010-07-04 11:13:02 UTC 
This bug is a follow-up of a freedesktop bug https://bugs.freedesktop.org/show_bug.cgi?id=28896

Software:

Fedora 13 i686 with all updates installed
Intel driver 2.12 vanilla
Kernel 2.6.34 vanilla

Hardware:

00:02.0 VGA compatible controller: Intel Corporation Core Processor Integrated
Graphics Controller (rev 12)

Steps to reproduce:

Run vesa X.org server next to already running intel X.org server.

Result: PC is unusable, both X servers die, all VTs die, dozens of errors in
dmesg and in the first X.org server log.
Comment 1 Chris Wilson 2010-07-04 12:33:16 UTC 
You have two drivers conflicting over control of the same hardware. Don't do that.
Comment 2 Artem S. Tashkinov 2010-07-04 14:32:39 UTC 
X.org server trying to run using Vesa shouldn't crash my computer and it happens now. I could be happy if kernel just said "Resource or device is busy" without wrecking havoc.

(As a side not I wonder why it's possible (and it perfectly works) to run proprietary NVIDIA X.org server and Vesa X.org server at the same time).
Comment 3 Jesse Barnes 2010-07-23 19:38:26 UTC 
The kernel has no way of doing this right now.  A suitably privileged X server can open /dev/mem or any of the sysfs resource files and start poking regsiters.

The reason NV and VESA work together is that the NV driver isn't a kernel mode driver, when you switch between servers in that config the whole hardware state is saved and restored, whereas in the i915 case, the i915 driver expects to be in sole control of the hw once it loads.
Comment 4 Artem S. Tashkinov 2010-08-09 11:10:42 UTC 
Created attachment 27388 [details]
Bug appearance
Comment 5 Artem S. Tashkinov 2010-08-09 11:11:33 UTC 
Created attachment 27389 [details]
Bug appearance II