Bug 382 - xorg is crashing on a fresh install on i586 arch with ATI card when KMS is disabled
Summary: xorg is crashing on a fresh install on i586 arch with ATI card when KMS is di...
Status: RESOLVED DUPLICATE of bug 138
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: Cauldron
Hardware: i586 Linux
Priority: Low normal
Target Milestone: ---
Assignee: Mageia Bug Squad
QA Contact:
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2011-03-15 19:18 CET by John Balcaen
Modified: 2011-03-16 21:48 CET (History)
5 users (show)

See Also:
Source RPM: x11-driver-video-ati
CVE:
Status comment:


Attachments
dmesg of the box (135.24 KB, text/plain)
2011-03-15 19:18 CET, John Balcaen
Details
xorg.log (42.13 KB, text/plain)
2011-03-15 19:20 CET, John Balcaen
Details
xorg.conf (1.52 KB, text/plain)
2011-03-15 19:20 CET, John Balcaen
Details
dmidecode (9.46 KB, text/plain)
2011-03-15 19:23 CET, John Balcaen
Details
result of lsmod (8.55 KB, text/plain)
2011-03-15 19:24 CET, John Balcaen
Details
service dm start && sleep 1 && gdb --pid $(pidof /etc/X11/X) (10.47 KB, text/plain)
2011-03-16 11:46 CET, John Balcaen
Details
:0-log with kernel linus 2.6.38 (17.86 KB, text/plain)
2011-03-16 12:05 CET, John Balcaen
Details
script in order to attach to X from a text terminal (vt1 to vt6) (72 bytes, text/plain)
2011-03-16 12:24 CET, Thierry Vignaud
Details
GDB commands needed by previous script (79 bytes, text/plain)
2011-03-16 12:25 CET, Thierry Vignaud
Details
X-gdb with 2.6.37.4-desktop-1.mga (3.88 KB, text/plain)
2011-03-16 13:10 CET, John Balcaen
Details

Description John Balcaen 2011-03-15 19:18:02 CET
Description of problem:
Xorg is crashing on a fresh mageia alpha 2 install

The install is done with the DVD alpha2 for i586 on a Acer 1690WMLi

[mikala@isis ~]$ uname -a                     
Linux isis.lan.littleboboy.net 2.6.37.3-desktop-1.mga #1 SMP Tue Mar 8 23:42:10 UTC 2011 i686 i686 i386 GNU/Linux 


Version-Release number of selected component (if applicable):
x11-server-1.9.4-2

How reproducible:
always



Reproducible: 

Steps to Reproduce:
Comment 1 John Balcaen 2011-03-15 19:18:34 CET
Created attachment 111 [details]
dmesg of the box
Comment 2 John Balcaen 2011-03-15 19:20:16 CET
Created attachment 112 [details]
xorg.log

The card is a Radeon ATI X700 Mobility
Comment 3 John Balcaen 2011-03-15 19:20:48 CET
Created attachment 113 [details]
xorg.conf
Comment 4 John Balcaen 2011-03-15 19:23:05 CET
Created attachment 114 [details]
dmidecode
Comment 5 John Balcaen 2011-03-15 19:24:30 CET
Created attachment 115 [details]
result of lsmod
Comment 6 D Morgan 2011-03-15 19:39:26 CET
From comment 2: 
Backtrace:
[    67.029] 0: /etc/X11/X (xorg_backtrace+0x37) [0x80e9bf7]
[    67.029] 1: /etc/X11/X (0x8048000+0x5e6aa) [0x80a66aa]
[    67.029] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xffffe40c]
[    67.029] Segmentation fault at address (nil)
[    67.029] 
Fatal server error:
[    67.029] Caught signal 11 (Segmentation fault). Server aborting
[    67.029] 
[    67.029]

CC: (none) => dmorganec

Comment 7 man draker 2011-03-15 22:32:32 CET
The same problem occurs on my Samsung R50 with X700 Mobility (Driver Radeon X 1950 and earlier)

Same Hardware with ML 2010.2 no problem.

CC: (none) => man-draker

Comment 8 Magnus Rasche 2011-03-15 22:40:59 CET
No problems with my installation.

My card is an intel 82865G on-board (intel 810 and later)

If I can help with some output, call me
Comment 9 John Balcaen 2011-03-16 02:49:42 CET
using ssh i've been able to attach a gdb to X11 while it was still running :

(gdb) bt
#0  0xffffe424 in __kernel_vsyscall ()
#1  0xb74a09e0 in __nanosleep_nocancel () from /lib/i686/libc.so.6
#2  0xb74d565c in usleep () from /lib/i686/libc.so.6
#3  0xb71a7c04 in RADEONRestoreMemMapRegisters (pScrn=0x8b00158, restore=0x8b00608) at radeon_driver.c:4200
#4  0xb71a9a65 in RADEONAdjustMemMapRegisters (pScrn=0x8b00158, save=0x8b00608) at radeon_driver.c:4243
#5  0xb71abdc1 in RADEONScreenInit (scrnIndex=0, pScreen=0x8b10be8, argc=8, argv=0xbf8a61f4) at radeon_driver.c:3779
#6  0x0806faed in AddScreen ()
#7  0x080b745e in InitOutput ()
#8  0x08062455 in _start ()
(gdb) bt full
#0  0xffffe424 in __kernel_vsyscall ()
No symbol table info available.
#1  0xb74a09e0 in __nanosleep_nocancel () from /lib/i686/libc.so.6
No symbol table info available.
#2  0xb74d565c in usleep () from /lib/i686/libc.so.6
No symbol table info available.
#3  0xb71a7c04 in RADEONRestoreMemMapRegisters (pScrn=0x8b00158, restore=0x8b00608) at radeon_driver.c:4200
        info = 0x8b04598
        pRADEONEnt = 0x8b005b8
        RADEONMMIO = <value optimized out>
        timeout = <value optimized out>
        mc_fb_loc = 3623866368
        mc_agp_loc = 4294967232
        mc_agp_loc_hi = 6
#4  0xb71a9a65 in RADEONAdjustMemMapRegisters (pScrn=0x8b00158, save=0x8b00608) at radeon_driver.c:4243
        info = 0x8b04598
        fb = 3623866368
        agp = 4294967232
        agp_hi = 1
        changed = 1
#5  0xb71abdc1 in RADEONScreenInit (scrnIndex=0, pScreen=0x8b10be8, argc=8, argv=0xbf8a61f4) at radeon_driver.c:3779
        pScrn = 0x8b00158
        info = 0x8b04598
        hasDRI = 1
#6  0x0806faed in AddScreen ()
No symbol table info available.
#7  0x080b745e in InitOutput ()
No symbol table info available.
#8  0x08062455 in _start ()
No symbol table info available.

Please note that probably not all debug packages are installed but i'll try to do it, gdb does not seem to be able to do it :
warning: Unable to open "librpm-4.6.so" , missing debuginfos notifications will not be displayed

Maybe it should be rebuild against the new rpm.
Regards,
Comment 10 John Balcaen 2011-03-16 02:51:11 CET
just for memory i used  service dm start && sleep 1 && gdb --pid $(pidof /etc/X11/X)  to attach to X before the crash.
Comment 11 John Balcaen 2011-03-16 03:00:47 CET
Ok i managed to install more debug packages & the bt seems to be full now :


(gdb) bt
#0  0xffffe424 in __kernel_vsyscall ()
#1  0xb73c59e0 in __nanosleep_nocancel () from /lib/i686/libc.so.6
#2  0xb73fa65c in usleep () from /lib/i686/libc.so.6
#3  0xb70e58dd in legacy_output_dpms (output=0x8abbc70, mode=3) at legacy_output.c:1034
#4  0xb7113694 in radeon_dpms (output=0x8abbc70, mode=3) at radeon_output.c:504
#5  0x080d1424 in xf86PrepareOutputs (scrn=0x8aad158) at xf86Crtc.c:2564
#6  xf86SetDesiredModes (scrn=0x8aad158) at xf86Crtc.c:2621
#7  0xb70d1027 in RADEONScreenInit (scrnIndex=0, pScreen=0x8abdbe8, argc=8, argv=0xbf9fe054) at radeon_driver.c:3883
#8  0x0806faed in AddScreen (pfnInit=0xb70cf120 <RADEONScreenInit>, argc=8, argv=0xbf9fe054) at dispatch.c:3919
#9  0x080b745e in InitOutput (pScreenInfo=0x823fa00, argc=8, argv=0xbf9fe054) at xf86Init.c:773
#10 0x08062455 in main (argc=8, argv=0xbf9fe054, envp=0xbf9fe078) at main.c:209
(gdb) bt full
#0  0xffffe424 in __kernel_vsyscall ()
No symbol table info available.
#1  0xb73379e0 in __nanosleep_nocancel () from /lib/i686/libc.so.6
No symbol table info available.
#2  0xb736c65c in usleep () from /lib/i686/libc.so.6
No symbol table info available.
#3  0xb70578dd in legacy_output_dpms (output=0x8fabc70, mode=3) at legacy_output.c:1034
        tmpPixclksCntl = 4194304
        lvds = 0x8fab680
        pScrn = 0x8f9d158
        info = 0x8fa1598
        save = 0x8f9d608
        RADEONMMIO = 0xb6ff7000 ""
        tmp = <value optimized out>
        radeon_output = 0x8fabb48
        radeon_encoder = <value optimized out>
#4  0xb7085694 in radeon_dpms (output=0x8fabc70, mode=3) at radeon_output.c:504
        info = 0x8fa1598
        radeon_output = 0x8fabb48
#5  0x080d1424 in xf86PrepareOutputs (scrn=0x8f9d158) at xf86Crtc.c:2564
        output = 0x8fabc70
        o = <value optimized out>
#6  xf86SetDesiredModes (scrn=0x8f9d158) at xf86Crtc.c:2621
        crtc = <value optimized out>
        c = <value optimized out>
#7  0xb7043027 in RADEONScreenInit (scrnIndex=0, pScreen=0x8fadbe8, argc=8, argv=0xbfc76024) at radeon_driver.c:3883
        pScrn = 0x8f9d158
        info = 0x8fa1598
        hasDRI = 1
        subPixelOrder = <value optimized out>
        s = <value optimized out>
#8  0x0806faed in AddScreen (pfnInit=0xb7041120 <RADEONScreenInit>, argc=8, argv=0xbfc76024) at dispatch.c:3919
        i = <value optimized out>
        scanlinepad = <value optimized out>
        format = <value optimized out>
        depth = <value optimized out>
        bitsPerPixel = <value optimized out>
        j = <value optimized out>
        k = <value optimized out>
        pScreen = 0x8fadbe8
#9  0x080b745e in InitOutput (pScreenInfo=0x823fa00, argc=8, argv=0xbfc76024) at xf86Init.c:773
        i = <value optimized out>
        j = <value optimized out>
        k = <value optimized out>
        scr_index = <value optimized out>
        modulelist = <value optimized out>
        optionlist = 0x8f94248
        screenpix24 = <value optimized out>
        pix24 = <value optimized out>
        pix24From = <value optimized out>
        pix24Fail = 0
        autoconfig = <value optimized out>
        configured_device = <value optimized out>
#10 0x08062455 in main (argc=8, argv=0xbfc76024, envp=0xbfc76048) at main.c:209
        i = <value optimized out>
        alwaysCheckForInput = {0, 1}

debug packages installed :

[root@isis X11]# rpm -qa |grep debug
libxext-debug-1.2.0-2.mga1
libxfixes-debug-4.0.5-1.mga1
udev-debug-166-1.mga1
libxi-debug-1.4.0-3.mga1
libx11-debug-1.4.1-1.mga1
libxv-debug-1.0.6-1.mga1
libxt-debug-1.1.1-1.mga1
libxau-debug-1.0.6-1.mga1
libxdmcp-debug-1.1.0-1.mga1
freetype2-debug-2.4.4-4.mga1
libxcomposite-debug-0.4.3-1.mga1
x11-server-debug-1.9.4-2.mga1
libxcb-debug-1.7-1.mga1
libxdamage-debug-1.1.3-1.mga1
v4l-utils-debug-0.8.1-1.mga1
dbus-debug-1.4.1-3.mga1
libxtst-debug-1.2.0-1.mga1
libxinerama-debug-1.1.1-2.mga1
x11-driver-video-ati-debug-6.14.0-3.mga1
xcb-util-debug-0.3.6-3.mga1
glibc-debug-2.12.1-10.mga1
libsm-debug-1.2.0-2.mga1
pixman-debug-0.20.2-1.mga1
libxrandr-debug-1.3.1-2.mga1
mesa-debug-7.10-1.mga1
Comment 12 Wolfgang Bornath 2011-03-16 09:44:17 CET
I can confirm the bug.
Error messages on tty:
FATAL: Module fbcon not found

Machine: Lenovo Netbook S10e
Graphics: Intel 945GME

I selected KDE for installation, in the summary I found that X was not configured. Configuration went ok but I saw that xdm was installed.

After reboot xdm comes up but it refuses to take my password. Booting into init 3 I was able to login but startx produced the error of the x server.

CC: (none) => molch.b

Comment 13 Thierry Vignaud 2011-03-16 09:48:48 CET
Please _attach_ the _FULL_ trace + GDB output.
Attach and not paste because:
1) it's then garbaged when bugzilla rewrap lines
2) it makes harder to read this bug report.

Else we miss the actuall error before the trace.

Also could you try with latest kernel as well as with kernel-linus (which is newer than main kernel)

CC: (none) => thierry.vignaud
Source RPM: x11-server-1.9.4-2.mga1.src.rpm => x11-driver-video-ati

Comment 14 Thierry Vignaud 2011-03-16 09:50:14 CET
And also attach your /var/log/gdm/:0.log, :0-greeter.log and :0-slave.log
Comment 15 Wolfgang Bornath 2011-03-16 10:19:46 CET
(In reply to comment #13)
> Please _attach_ the _FULL_ trace + GDB output.

I used John's command string (see comment #10 - resulting in a blank screen. 
I'm very busy preparing for this weekend's Linux event, so I don't have the time for any extra work and the learning I'd have to do (don't know anything about all this trace and gdb stuff), sorry.
Comment 16 Thierry Vignaud 2011-03-16 10:38:32 CET
We need the bits printed by GDB before you typed bt.
You need to attach gdb prior to X11 crashing of course.
Comment 17 John Balcaen 2011-03-16 11:46:32 CET
Created attachment 119 [details]
service dm start && sleep 1 && gdb --pid $(pidof /etc/X11/X)

Result of  service dm start && sleep 1 && gdb --pid $(pidof /etc/X11/X) with the linus kernel.
Currently installed -debug packages : 
[root@isis log]# rpm -qa |grep debug
libxext-debug-1.2.0-2.mga1
openssl-debug-1.0.0c-2.mga1
libxfixes-debug-4.0.5-1.mga1
expat-debug-2.0.1-14.mga1
udev-debug-166-1.mga1
libxi-debug-1.4.0-3.mga1
libx11-debug-1.4.1-1.mga1
libxv-debug-1.0.6-1.mga1
libxt-debug-1.1.1-1.mga1
libdrm-debug-2.4.24-1.mga1
libxau-debug-1.0.6-1.mga1
libxdmcp-debug-1.1.0-1.mga1
freetype2-debug-2.4.4-4.mga1
audit-debug-2.0.5-3.mga1
libfontenc-debug-1.1.0-1.mga1
libxcomposite-debug-0.4.3-1.mga1
x11-server-debug-1.9.4-2.mga1
libxcb-debug-1.7-1.mga1
libpciaccess-debug-0.12.0-1.mga1
libxdamage-debug-1.1.3-1.mga1
v4l-utils-debug-0.8.1-1.mga1
dbus-debug-1.4.1-3.mga1
talloc-debug-2.0.1-2.mga1
x11-driver-video-v4l-debug-0.2.0-7.mga1
libxtst-debug-1.2.0-1.mga1
libxinerama-debug-1.1.1-2.mga1
pam-debug-1.1.3-2.mga1
bzip2-debug-1.0.6-1.mga1
x11-driver-video-ati-debug-6.14.0-3.mga1
xcb-util-debug-0.3.6-3.mga1
glibc-debug-2.12.1-10.mga1
libsm-debug-1.2.0-2.mga1
pixman-debug-0.20.2-1.mga1
zlib-debug-1.2.5-3.mga1
libxrandr-debug-1.3.1-2.mga1
libxfont-debug-1.4.3-1.mga1
mesa-debug-7.10-1.mga1
gcc-debug-4.5.2-2.mga1

I'm not using gdm since kdm is installed (kde4 install)  but i'll switch to gdm to provide logs as requested.
Comment 18 John Balcaen 2011-03-16 12:05:13 CET
Created attachment 120 [details]
:0-log with kernel linus 2.6.38

there's no :0-greeter.log & :0-slave.log & gdm is trying to restart X on every crash with no end 
Should i attach the log with 2.6.37.4 (lasted mga kernel) since it's also failing ?
Thierry Vignaud 2011-03-16 12:19:59 CET

Attachment 119 is obsolete: 0 => 1

Comment 19 Thierry Vignaud 2011-03-16 12:21:15 CET
Yes you can attach it.

But beware:

Using GDB, you actually stoped the X server before it crashes.

You need to first type "cont" (you may have to do it several time
as it may be stoped by some spurious signals like SIGUSR1, SIGUSR2).

When it actually segfaults (you should see SIGSEV somewhere),
then you must type "bt full".

Not before, else it's just some random trace of where was X server
when you interupted it with GDB.

Summary: Mageia alpha2 : xorg is crashing on a fresh install on i586 arch => Mageia alpha2 : xorg is crashing on a fresh install on i586 arch with ATI card

Thierry Vignaud 2011-03-16 12:21:20 CET

Keywords: (none) => NEEDINFO

Comment 20 Thierry Vignaud 2011-03-16 12:22:03 CET
I've some script somewhere either on mdv bugzilla or mga bugzilla in order to perform all of this (without SSH connection from another machine).
Comment 21 Thierry Vignaud 2011-03-16 12:24:36 CET
Created attachment 121 [details]
script in order to attach to X from a text terminal (vt1 to vt6)

I actually found it.
Just run "sh ./Xgdb2.sh" on a text terminal, then switch back to X11 until it segfaults.
Comment 22 Thierry Vignaud 2011-03-16 12:25:13 CET
Created attachment 122 [details]
GDB commands needed by previous script
Thierry Vignaud 2011-03-16 12:25:33 CET

Attachment 121 mime type: application/x-sh => text/plain

Comment 23 John Balcaen 2011-03-16 13:10:05 CET
Created attachment 123 [details]
X-gdb with 2.6.37.4-desktop-1.mga

Ok i guess i found the culprit thanks to your script :)
reading the line : 
(II) [KMS] drm report modesetting isn't supported.
i noticed that kms was not enable with my radeon driver.
So i add a /etc/modprobe.d/radeon.conf
with options radeon modeset=1
& right after X is able to start \o/
So the problem is probably to know when to enable KMS on radeon card (especially with thoses still using fglrx).
If i'm not wrong Anssi is testing that on mandriva cooker.

Regards,
Comment 24 Thierry Vignaud 2011-03-16 13:51:27 CET
OK.
Still it would be good if upstream wouldn't segfault when using UMS...

Priority: Normal => Low
Summary: Mageia alpha2 : xorg is crashing on a fresh install on i586 arch with ATI card => xorg is crashing on a fresh install on i586 arch with ATI card when KMS is disabled
Severity: critical => normal

Comment 25 Thierry Vignaud 2011-03-16 13:52:34 CET
Can you still attach the full trace so that we report it upstream?

CC: (none) => tmb

Comment 26 Thierry Vignaud 2011-03-16 15:04:49 CET
Actually it's a dupe of bug #138
Could you try the patch suggested there?

*** This bug has been marked as a duplicate of bug 138 ***

Status: NEW => RESOLVED
Resolution: (none) => DUPLICATE

Comment 27 Wolfgang Bornath 2011-03-16 18:55:40 CET
(In reply to comment #12)
> I can confirm the bug.
> Error messages on tty:
> FATAL: Module fbcon not found
> 
> Machine: Lenovo Netbook S10e
> Graphics: Intel 945GME

This was caused by a faulty installation!
I did a fresh installation with the same installation medium (md5sum matched) and now the summary showed graphic system as "automatic" where it was "not configured" before. X is running fine, this bug does not apply here.
Comment 28 Thierry Vignaud 2011-03-16 21:48:09 CET
I think you commented on the wrong bug.
You wanted to comment the "summary" bug, didn't you?

Note You need to log in before you can comment on or make changes to this bug.