FS#12332 - xf86-video-siliconmotion-1.6 crashes Xorg 7.4

Attached to Project: Arch Linux
Opened by mentalo (mentalo) - Wednesday, 03 December 2008, 13:35 GMT
Last edited by Aaron Griffin (phrakture) - Wednesday, 24 December 2008, 17:47 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Jan de Groot (JGC)
Architecture x86_64
Severity High
Priority Normal
Reported Version None
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 1
Private No

Details

Description:
After the update to Xorg 7.4 and xf86-video-siliconmotion-1.6 it's impossible to start a new xserver with siliconmotion module.

Acer Travelmate C100 with Lynx3DM

Additional info:
* package version(s)
xf86-video-siliconmotion-1.6.0-1
xorg-server 1.5.3-2

* config and/or log files etc.
Xorg.0.log:
(II) Silicon Motion: driver (version 1.6.0) for Silicon Motion Lynx chipsets:
Lynx, LynxE, Lynx3D, LynxEM, LynxEM+, Lynx3DM, Cougar3DR
(II) Primary Device is: PCI 00@00:02:0
(WW) Falling back to old probe method for siliconmotion
...
(EE) Silicon MotionUnknown EDID version 128
...
(WW) Silicon Motionxf86UnMapVidMem: cannot find region for [0xb7703000,0x200000]
...
Fatal server error:
AddScreen/ScreenInit failed for driver 0


Steps to reproduce:
Try to start X with Xorg 7.4 and xf86-video-siliconmotion-1.6
   Xorg.0.log (23.1 KiB)
This task depends upon

Closed by  Aaron Griffin (phrakture)
Wednesday, 24 December 2008, 17:47 GMT
Reason for closing:  Upstream
Additional comments about closing:  Fixed upstream
Comment by Glenn Matthys (RedShift) - Friday, 05 December 2008, 20:53 GMT
Can you post your xorg.conf please?
Comment by Tim Besard (maleadt) - Wednesday, 10 December 2008, 21:53 GMT
As a Debian user, I have to acknowledge this bug-report. The resulting crash seems to differ though: using enlightenment (e16), I get served with a heavily corrupted screen. Starting XDM results in an incredibly slow rendering screen (1 line per 5 seconds), again heavily corrupted. Xorg.conf was default autogenerated (i.e. only device naming), with no special settings.
Comment by mentalo (mentalo) - Monday, 15 December 2008, 07:20 GMT
sorry, didnt recognize your answer, cause my mailaccount was full :/

Comment by mentalo (mentalo) - Monday, 15 December 2008, 07:22 GMT
oops, damn late. here is xorg.conf

only difference, is that atm module is "vesa" instead of "siliconmotion"
   xorg.conf (2.3 KiB)
Comment by Tim Besard (maleadt) - Monday, 15 December 2008, 09:58 GMT
Try all patches mentioned at http://bugs.freedesktop.org/show_bug.cgi?id=18816 above the GIT version of the driver, it fixed all primary issues (X doesn't start), and many more (VT blank after X closure, slight XAA corruption, heavy EXA corruption).
Comment by mentalo (mentalo) - Sunday, 21 December 2008, 00:47 GMT
yes, worked for me. thanks for your comment. will the patches be included in package?
Comment by Tim Besard (maleadt) - Sunday, 21 December 2008, 08:24 GMT
Has been merged into GIT master: http://cgit.freedesktop.org/xorg/driver/xf86-video-siliconmotion/log/

Remember though you should manually force UseBIOS to false with certain chipsets, as mine (a Lynx 3D). It seems that that specific patch hasn't been merged into master.
Comment by Jan de Groot (JGC) - Sunday, 21 December 2008, 11:42 GMT
Can I solve this bug by packaging git master of this driver? Any recommended patches for the UseBIOS thing mentioned in the last post?
Comment by mentalo (mentalo) - Sunday, 21 December 2008, 12:15 GMT
maleadt, thank you for your good testing with the module. have you solved the crashing when switching to tty? same happens here.
i use the up-to-date git-version, where patches were already merged.
Comment by Tim Besard (maleadt) - Sunday, 21 December 2008, 14:20 GMT
@Jan De Groot:
Ignore what I said about one patch not being merged into master, it actually has :)
The patches contains a "detection method" to see whether the BIOS can be relied upon. Patch [1] also contains a fragment to force not to use this detection method (and force NOT to use the BIOS) on a single chipset which is known to have problems with it. Some other chipsets might have similar problems, but have not yet been blacklisted in the driver. So if you experience any problems, be sure to try the "UseBIOS" option (Option "UseBIOS" "off").

@mentalo
All problems have been fixed, except for one situation where I switch the VT while the X server is still loading. Francisco Jerez (thank him for fixing all these bugs, I merely provided information) commented that this it not a driver bug [2], and I haven't found the time to find who's causing it. Regular VT switches did permanently blank the screen earlier on, but this has been fixed for me [3]. Mind though I focused on EXA acceleration, which now works perfectly AFAIK, so XAA acceleration might still contain some bugs as I didn't test it as extensively.


References:
1: http://cgit.freedesktop.org/xorg/driver/xf86-video-siliconmotion/commit/?id=47bcfb505886d16ff864f26a9cfa5f43b5da710e
2: http://bugs.freedesktop.org/show_bug.cgi?id=18816#c39
3: http://cgit.freedesktop.org/xorg/driver/xf86-video-siliconmotion/commit/?id=5a07709ae41600bc02205753c64d764f11838240
Comment by mentalo (mentalo) - Monday, 22 December 2008, 11:18 GMT
so, my problems are gone. packing the git-version would be nice. thank you for support.

Loading...