[Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

kas111 · 2012-04-16 21:20:03

I have gotten a segfault listed in the Xorg log for the last 3 (3.2.14, 3.3.1 and 3.3.2) kernels and have to downgrade to 3.2.13.1-x86-64 in order to get any video output. This is a stock kernel and the current stock xf-86-video-ati drivers, and an otherwise stock setup. Nothing else needs to be downgraded to get working video. In all kernels newer than 3.2.13-1-x86.64, I end up with no video and the monitor led shows no signal. Any settings of "radeon.modeset=" and nomodeset don't help. Appending 3 to the stock grub command line does exactly the same thing and no video comes upon any of the VTs or terminals in video mode 3. Chrooting with an install CD and downgrading to 3.2.13-1-x86_64 seems the only way to get video back and that does with no other changes or downgrades from a fully upgraded otherwise system. The fall-back kernels do the same thing. I have had a similar problem occasionally in the past but it always went away on the next kernel update ... and radeon.modeset=0 worked in those cases. My intel video x86 laptop similarly updated boots and runs 3.3.2-1-ARCH fine.

this is from Xorg.0.log ... nothing else shows up in the logs ... but simply downgrading the kernel does get rid of the problem.

Video card is an ATI RV730 (HD4650).

(EE) AIGLX error: r600 does not export required DRI extension
[ 20.309] (EE) AIGLX: reverting to software rendering
[ 20.311] (II) AIGLX: Loaded and initialized swrast
[ 20.311] (II) GLX: Initialized DRISWRAST GL provider for screen 0
[ 20.311] (II) RADEON(0): Setting screen physical size to 423 x 317
[ 20.396]
[ 20.396] Backtrace:
[ 20.397] 0: /usr/bin/Xorg (xorg_backtrace+0x36) [0x55d2c6]
[ 20.397] 1: /usr/bin/Xorg (0x400000+0x160cf9) [0x560cf9]
[ 20.397] 2: /lib/libpthread.so.0 (0x7fbaa1eaa000+0xf140) [0x7fbaa1eb9140]
[ 20.397] 3: /usr/lib/xorg/modules/drivers/radeon_drv.so (0x7fba9ee9b000+0xb675e) [0x7fba9ef5175e]
[ 20.397] 4: /usr/lib/xorg/modules/drivers/radeon_drv.so (0x7fba9ee9b000+0xb502d) [0x7fba9ef5002d]
[ 20.397] 5: /usr/lib/xorg/modules/libexa.so (0x7fba9e42e000+0x9044) [0x7fba9e437044]
[ 20.397] 6: /usr/lib/xorg/modules/libexa.so (0x7fba9e42e000+0xaae9) [0x7fba9e438ae9]
[ 20.397] 7: /usr/bin/Xorg (0x400000+0xf1ccd) [0x4f1ccd]
[ 20.397] 8: /usr/bin/Xorg (miPaintWindow+0x220) [0x53f340]
[ 20.397] 9: /usr/bin/Xorg (miWindowExposures+0xc8) [0x53f488]
[ 20.397] 10: /usr/lib/xorg/modules/extensions/libdri.so (DRIWindowExposures+0x99) [0x7fba9fe51909]
[ 20.397] 11: /usr/bin/Xorg (0x400000+0x8310c) [0x48310c]
[ 20.397] 12: /usr/bin/Xorg (MapWindow+0x309) [0x45d659]
[ 20.397] 13: /usr/bin/Xorg (0x400000+0x23107) [0x423107]
[ 20.397] 14: /lib/libc.so.6 (__libc_start_main+0xf5) [0x7fbaa0d5f455]
[ 20.397] 15: /usr/bin/Xorg (0x400000+0x2343d) [0x42343d]
[ 20.397]
[ 20.397] Segmentation fault at address 0x7c00000103
[ 20.397]
Fatal server error:
[ 20.397] Caught signal 11 (Segmentation fault). Server aborting
[ 20.397]

Last edited by kas111 (2012-04-18 13:47:04)

Lone_Wolf · 2012-04-17 12:19:21

Might be a problem with Kernel Mode Setting, try adding vga=<some mode> to the grub bootline for testing purposes.

other things to check :
is there a newer bios version for your Motherboard ?
What version of linux-firmware are you using ?

kas111 · 2012-04-17 13:42:41

Lone_Wolf wrote:

Might be a problem with Kernel Mode Setting, try adding vga=<some mode> to the grub bootline for testing purposes.
other things to check :
is there a newer bios version for your Motherboard ?
What version of linux-firmware are you using ?

The MB is about 3 or 4 yr old but a friend with exactly the same one had a world of trouble updating bios on his (it bricked and Gigabyte
replaced it) so I have left mine alone since it had been working fine for several years,

I will try VGA= directly after this reply and report back if it is successful. It sure would beat chrooting after every try.

I forgot to mention I am running xfce4 if that makes any difference. I've been running it since the gnome 3 switch with
no problems except as mentioned. KMS has been an ongoing/occasional problem with the video card.

Thanks for the reply
kas111

kas111 · 2012-04-17 14:27:59

Vga=791 does get me a message saying that X is respawning too fast but will allow me to login on a text console. nomodeset vga=791 or radeon.modeset vga=791 do the same thing.

K

Lone_Wolf · 2012-04-18 10:42:30

Then we have atleast found a workaround, and it definitely points to a problem with KMS and the kernel radeon module.

changelog for kernel 3.2.14
If i read that correctly there were 3 commits related to drm/radeon/kms , possibly 1 of those caused your problems.

You could switch to the LTS kernel for the time being, and i also think you should file a kernel bug on the kernel mailing list.

Edit :
I looked a bit more closely, and there are 2 commits for drm/radeon/kms, both by Alex Deucher.
This one seems the most likely culprit :

commit 16d4db3ec3dfdd10bbafef9a4f62fc9374592d2e
Author: Alex Deucher <alexander.deucher@amd.com>
Date:   Fri Mar 16 12:22:09 2012 -0400

    drm/radeon/kms: fix analog load detection on DVI-I connectors
    
    commit e00e8b5e760cbbe9067daeae5454d67c44c8d035 upstream.
    
    We digital encoders have a detect function as well (for
    DP to VGA bridges), so we make sure we choose the analog
    one here.
    
    Fixes:
    https://bugs.freedesktop.org/show_bug.cgi?id=47007
    
    Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
    Signed-off-by: Dave Airlie <airlied@redhat.com>
    Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Last edited by Lone_Wolf (2012-04-18 10:45:53)

kas111 · 2012-04-18 13:30:37

I guess there is no excuse for me being lazy ... switching cables from DVI to VGA at the video
card end fixed the problem ... 3.3.2-1-x86-64 is now booting into gfx mode (5) with no
problems and a grub kernel line with nothing but vmlinuz26, root=the UUID of the root
drive and ro. Now if I can figure out how to change the heading to [Solved] we'll be good.

Thanks, Lone_Wolf, for your help.

kas111

ericab · 2012-05-07 21:49:21

any news on this ? i too have the same problem, and switching from dvi to vga is not acceptable to me.
nomodeset vga=791 does not work.

video card is: Radeon HD 4200
kernel is: 3.3.4-2-ARCH

Lone_Wolf · 2012-05-08 13:30:53

ericab,

there are 5 types of DVI conectors :

DVI-A analog only
DVI-I single link, analog + digital
DVI-I dual link, analog + digital
DVI-D single link, Digital only
DVI-D dual link, digital only
(this wikipedia page has pics of the connectors)

the problem appears to be specifically with the DVI - I connector.
This connector can both supply a digital and analog connection, and it looks like since kernel 3.2.14 the radeon kernelmodule automatically selects the ANALOG connection for this connector. This means a digital monitor connected to a DVI-I connector on the card doesn't get any signal.

Check what connectors your videocard and monitor have, there may be another option then switching back to vga.

ericab · 2012-05-08 22:50:18

Lone_Wolf;
hi, im sorry i probably should have elaborated on my setup a bit more.
the machine in question is actually a headless machine, and im faking a connected and powered on monitor in the xorg.conf, by providing an EDID binary;
does this make fixing this issue any easier ? harder ?
my first post; i say "switching from dvi to vga is not acceptable to me", as if i was actually using a cable but im not; but assumed that the edid is defaulting to using DVI. can i modify the edid somehow; or even xorg to tell it i want to use the VGA post on the monitor ? the edid binary i use is actually the monitor i have in my office to access my headless machines via VNC, and does have DVI & VGA inputs; so.....
there's got to be a way to tell xorg to use VGA instead of dvi;
any ideas ?

thanks,
eric

***edit
ive noticed my xorg error is a bit different then OP:

[    37.039] (EE) AIGLX error: r600 does not export required DRI extension
[    37.039] (EE) AIGLX: reverting to software rendering
[    37.045] (II) AIGLX: Loaded and initialized swrast
[    37.045] (II) GLX: Initialized DRISWRAST GL provider for screen 0
[    37.046] (II) RADEON(0): Setting screen physical size to 338 x 254
[    37.057]
[    37.057] Backtrace:
[    37.057] 0: /usr/bin/X (xorg_backtrace+0x36) [0x55f646]
[    37.057] 1: /usr/bin/X (0x400000+0x163379) [0x563379]
[    37.057] 2: /lib/libpthread.so.0 (0x7f8404776000+0xf140) [0x7f8404785140]
[    37.057] 3: /usr/lib/xorg/modules/drivers/radeon_drv.so (0x7f840175f000+0xb675e) [0x7f840181575e]
[    37.057] 4: /usr/lib/xorg/modules/drivers/radeon_drv.so (0x7f840175f000+0xb502d) [0x7f840181402d]
[    37.057] 5: /usr/lib/xorg/modules/libexa.so (0x7f8400cef000+0x90d4) [0x7f8400cf80d4]
[    37.057] 6: /usr/lib/xorg/modules/libexa.so (0x7f8400cef000+0xab79) [0x7f8400cf9b79]
[    37.057] 7: /usr/bin/X (0x400000+0xf316d) [0x4f316d]
[    37.057] 8: /usr/bin/X (miPaintWindow+0x220) [0x5415f0]
[    37.057] 9: /usr/bin/X (miWindowExposures+0xc8) [0x541738]
[    37.057] 10: /usr/lib/xorg/modules/extensions/libdri.so (DRIWindowExposures+0x99) [0x7f8401e8ab19]
[    37.057] 11: /usr/bin/X (0x400000+0x8407c) [0x48407c]
[    37.057] 12: /usr/bin/X (MapWindow+0x32e) [0x45e01e]
[    37.057] 13: /usr/bin/X (0x400000+0x234d7) [0x4234d7]
[    37.057] 14: /lib/libc.so.6 (__libc_start_main+0xf5) [0x7f840362b455]
[    37.058] 15: /usr/bin/X (0x400000+0x2380d) [0x42380d]
[    37.058]
[    37.058] Segmentation fault at address 0x7c00000103

Last edited by ericab (2012-05-08 23:11:32)

ericab · 2012-05-14 01:15:34

BuMP

Lone_Wolf · 2012-05-14 16:11:29

Ericab, the following MIGHT work

First, you need to identify the outputnames for your connection.
run xrandr to get them.

xrandr
Screen 0: minimum 320 x 200, current 1920 x 1200, maximum 8192 x 8192
DVI-0 connected 1920x1200+0+0 (normal left inverted right x axis y axis) 519mm x 324mm
   1920x1200      60.0*+
   1600x1200      60.0  
   1680x1050      60.0  
   1280x1024      75.0  
   1440x900       75.0     59.9  
   1280x960       60.0  
   1152x864       75.0     60.0  
   1280x720       60.0  
   1024x768       75.1     70.1     60.0  
   832x624        74.6  
   800x600        72.2     75.0     60.3     56.2  
   640x480        72.8     75.0     66.7     60.0  
   720x400        70.1  
VGA-0 disconnected (normal left inverted right x axis y axis)
HDMI-0 disconnected (normal left inverted right x axis y axis)
DisplayPort-0 disconnected (normal left inverted right x axis y axis)

DVI-0 , VGA-0 , HDMI-0 , DisplayPort-0 are the connections present on my system.

Let's assume you want to have X always use the VGA-0 output .

You will need a DEVICE and MONITOR section in an xorg.conf file, say /etc/X11/xorg.conf.d/10-MyFakeMonitor.conf
That file will look something like this :

Section "Device"
  Identifier "videocard"
  Driver "radeon"
  Option "Monitor-VGA-0" "fakemonitor"
EndSection

Section "Monitor"
 Identifier "fakemonitor"
EndSection

ericab · 2012-05-14 19:09:00

thanks for the reply LoneWolf,

xrandr returns: Cant open Display

my /etc/X11/xorg.conf.d/20-radeon.conf
contains this:

Section "Device"
Identifier "Radeon"
Driver "radeon"
#Option "ConnectedMonitor" "HDMI-0"
Option "Monitor-VGA-0" "Monitor0"
#Option "CustomEDID" "HDMI-0:/etc/X11/viewsonic.edid"
#Option "UseEDID" "true"
EndSection

Section "InputDevice"
Identifier "Mouse0"
Driver "mouse"
Option "Protocol" "auto"
Option "Device" "/dev/psaux"
Option "Emulate3Buttons" "no"
Option "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"
Identifier "Keyboard0"
Driver "kbd"
EndSection

Section "Monitor"
Identifier "Monitor0"
VendorName "ViewSonic"
ModelName "ViewSonic VX2025wm"
HorizSync 30.0 - 82.0
VertRefresh 50.0 - 75.0
Option "DPMS"
EndSection

Section "Screen"
Identifier "Screen0"
Device "Radeon"
Monitor "Monitor0"
DefaultDepth 24
SubSection "Display"
Depth 24
Modes "1280x960"
EndSubSection
EndSection

previously the hashed out items in the DEVICE section worked, but
i've removed them, and tried your suggestion.
xorg fails with the same error.

Lone_Wolf · 2012-05-15 20:50:26

Ok, seems we have to look deeper into this.

please post the full Xorg0.log .

Tsynique · 2012-05-15 23:10:09

Lone_Wolf wrote:

Ok, seems we have to look deeper into this.
please post the full Xorg0.log .

I have a very similar problem with Samsung R540 JS01EE laptop equipped with ATI Radeon HD 545v (32-bit fresh, stock Arch Linux installation, everything updated).

Here is my Xorg.log

Lone_Wolf · 2012-05-16 10:54:19

@Tsynique

[   140.375] (II) [KMS] drm report modesetting isn't supported.

This is weird, as KMS is supported by the radeon drivers since a long time.

[   141.242] (EE) AIGLX error: r600 does not export required DRI extension
[   141.242] (EE) AIGLX: reverting to software rendering
[   141.247] (II) AIGLX: Loaded and initialized swrast
[   141.247] (II) GLX: Initialized DRISWRAST GL provider for screen 0

The lack of KMS causes X to fall back to SW rendering.
While this may not be the cause of the problems, it definitely needs to be corrected.

Check the Arch radeon wiki on KMS on how to enable KMS.

Tsynique · 2012-05-16 15:13:05

Thanks a lot, the article solved my problem.

Here's what I did:

1. Removed nomodeset from /boot/grub/menu.lst (long ago I remember this would just give me a black screen right after modesetting but now it's ok)
2. In the MODULES variable inside /etc/mkinitcpio.conf I added intel_agp radeon. As the article states the order of these two is important.
3. Ran # mkinitcpio -p linux
4. Added a kernel option radeon.agpmode=8 in /boot/grub/menu.lst.
5. Reboot then startx and all went fine! I even tested glx-gears.

Here's GRUB config file /boot/grub/menu.lst:

title Arch Linux
root (hd0,6)
kernel /vmlinuz-linux root=/dev/sda6 ro radeon.agpmode=8
initrd /initramfs-linux.img
boot

Also, the MODULES line from /etc/mkinitcpio.conf:

MODULES="intel_agp radeon"

Thanks for the help again!

Last edited by Tsynique (2012-05-16 15:13:21)

ericab · 2012-05-16 17:31:17

Tsynique;
you post solved my problem !
thank you, and thank you Lone_Wolf !

Last edited by ericab (2012-05-16 17:31:34)

Arch Linux

#1 2012-04-16 21:20:03

[Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#2 2012-04-17 12:19:21

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#3 2012-04-17 13:42:41

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#4 2012-04-17 14:27:59

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#5 2012-04-18 10:42:30

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#6 2012-04-18 13:30:37

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#7 2012-05-07 21:49:21

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#8 2012-05-08 13:30:53

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#9 2012-05-08 22:50:18

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#10 2012-05-14 01:15:34

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#11 2012-05-14 16:11:29

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#12 2012-05-14 19:09:00

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#13 2012-05-15 20:50:26

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#14 2012-05-15 23:10:09

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#15 2012-05-16 10:54:19

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#16 2012-05-16 15:13:05

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

#17 2012-05-16 17:31:17

Re: [Solved] kernel - no video with 3.2.14, thru 3.3.2-1-x86_64 with xf86-

Board footer