Bug 707

Summary: Radeon 7000 freezes with dri enabled
Product: xorg Reporter: Alan Olsen <alan>
Component: Driver/RadeonAssignee: Xorg Project Team <xorg-team>
Status: RESOLVED FIXED QA Contact: Joel Franco <joel.franco>
Severity: normal    
Priority: high CC: ajax, joel.franco, lukejr+freedesktopbugs
Version: 6.9.0   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
Radeon 7000 Xorg.0.log with failure error messages.
none
Listing on the permissions for /dev/dri/card0
none
The dmesg log from the system with the problem
none
gdb attach & bt of X
none
kdm log
none
Xorg logfile
none
dmesg
none
xorg.conf
none
Xorg.0.log none

Description Alan Olsen 2004-06-01 16:52:33 UTC
Using Fedora Core 2 xorg-x11-6.7.0-2 with an ATI Radeon 7000 card.  

When the dri module is loaded, xorg will freeze and the screen will get a very
odd look to it.  (Not certain if it is "blooming" or what.  It as if it loses
all sync.) The system becomes very unresponsive after that. (Not certain if it
is totally hung or just xorg fubared.  I was not in a place where I could test it.)

If I comment out the dri module line, xorg will work fine, but without hardware
acceleration.
Comment 1 nastos 2004-06-03 15:31:21 UTC
I am experiencing similar things.  I have three systems 
with ATI Radeon 7000. On two systems, that both use an 
ASUS P4S800D-E (SIS655TX) the system freezes when X 
tries to load.  By freezes, I mean: (i) ctrl-alt-backspace 
does not kill X, (ii) ctrl-alt-del does not reboot the 
machine, (iii) can no longer ping the card, etc... 
Commenting out the 'Load dri' line solves the problem. 
The third system though, is an AMD Athlon 64 on an 
ASUS K8V board, and everything loads fine with dri. 
Comment 2 Alan Olsen 2004-06-06 04:45:59 UTC
I have some additional information.

What is happening is that X is trying to open /dev/dri/card0.  For some weird
reason the open fails.  (Giving an unknown error code 999 according to the
logs.)  Instead of failing gracfully and not loading dri, it trys to go on from
there and nothing comes up.

I am attaching the log file, a list of the permissions on the device and a copy
of dmesg from that system.

I have tested the 8k stacks issue and it is not the cause of this problem.
Comment 3 Alan Olsen 2004-06-06 04:48:07 UTC
Created attachment 346 [details]
Radeon 7000 Xorg.0.log with failure error messages.

Here is the log file for the Radeon 7000 dri problem.  Look for /dev/dri/card0
to find where it is choking.
Comment 4 Alan Olsen 2004-06-06 04:49:57 UTC
Created attachment 347 [details]
Listing on the permissions for /dev/dri/card0

Here are the permissions listed for /dev/dri/card0.  They appear to be correct.
Comment 5 Alan Olsen 2004-06-06 04:51:36 UTC
Created attachment 348 [details]
The dmesg log from the system with the problem

This is the dmesg log from the system with the problem.  I suspect it is a bug
with the SiS AGP chip on the motherboard, but I am not certain.
Comment 6 Alan Olsen 2004-06-06 05:40:20 UTC
I think I have a cause now.

The SiS 741 AGP chipset is not supported.  Instead of figuring that out and
giving an "unsupported message", it just goes out to lunch.

As Bender would say "We're boned!".
Comment 7 Eric Anholt 2004-06-16 03:51:58 UTC
Could you remove the agp from your kernel and try running your Radeon in PCI
mode?  You may need to set Option "BusType" "PCI" in the Driver section of
xorg.conf, but probably not at this point.  If that resolves the issue, I'll
close the bug assuming that it's AGP at fault like you say.
Comment 8 Luke-Jr 2004-07-10 03:55:00 UTC
I am having this issue on my system with a Radeon 9200SE, but only with Load 
"glx"; "dri" by itself is fine, but that doesn't enable the GLX extention. 
The display freezes and I suspect the system itself continues to run 
(network/HD activity; I should test w/ audio playing, but I don't want to 
reboot ATM). 
Interestingly, my PC is similar to the one nastos said he had working... 
Motherboard: Asus K8V 
CPU: Athlon64 3200+ 
AGP Bridge: 0000:00:00.0 Host bridge: VIA Technologies, Inc. VT8385 [K8T800 
AGP] Host Bridge (rev 01) 
 
I think this makes it unlikely that the problem is that particular AGP 
bridge... 
Comment 9 Luke-Jr 2004-07-13 15:27:27 UTC
dmesg output: 
+mtrr: 0xe8000000,0x8000000 overlaps existing 0xe8000000,0x200000 
+[drm] Initialized radeon 1.11.0 20020828 on minor 0: ATI Technologies Inc 
RV280 [Radeon 9200 SE] 
+mtrr: 0xe8000000,0x8000000 overlaps existing 0xe8000000,0x200000 
+agpgart: Found an AGP 3.5 compliant device at 0000:00:00.0. 
+agpgart: Putting AGP V3 device at 0000:00:00.0 into 0x mode 
+agpgart: Putting AGP V3 device at 0000:01:00.0 into 0x mode 
+[drm] Loading R200 Microcode 
 
Other relevant debug info will be attached... but I'm suspecting the 0x 
stuff... 
Comment 10 Luke-Jr 2004-07-13 15:32:42 UTC
Created attachment 475 [details]
gdb attach & bt of X
Comment 11 Luke-Jr 2004-07-13 15:32:59 UTC
Created attachment 476 [details]
kdm log
Comment 12 Luke-Jr 2004-07-13 15:33:14 UTC
Created attachment 477 [details]
Xorg logfile
Comment 13 svetljo 2004-08-02 12:22:15 UTC
(In reply to comment #2)
> I have some additional information.
> 
> What is happening is that X is trying to open /dev/dri/card0.  For some weird
> reason the open fails.  (Giving an unknown error code 999 according to the
> logs.)  Instead of failing gracfully and not loading dri, it trys to go on from
> there and nothing comes up.
> 
> I am attaching the log file, a list of the permissions on the device and a copy
> of dmesg from that system.
> 
> I have tested the 8k stacks issue and it is not the cause of this problem.



does the error message go away in case you don't specify the BusID ?
/* here it does */
Comment 14 Eric Anholt 2004-08-14 18:32:29 UTC
All of these reports look like AGP driver bugs to me.  The first two are SiS
chipsets resulting in hangs.  The third is a newer VIA AGPv3 chipset with a v3
card, with visible agp issues in the dmesg ("Putting AGP V3 device at
0000:00:00.0 into 0x mode").  I would suggest that these issues get reported to
the linux kernel bugzilla.

Comment 15 Joel Franco 2006-02-16 12:06:16 UTC
(In reply to comment #14)
> All of these reports look like AGP driver bugs to me.  The first two are SiS
> chipsets resulting in hangs.  The third is a newer VIA AGPv3 chipset with a v3
> card, with visible agp issues in the dmesg ("Putting AGP V3 device at
> 0000:00:00.0 into 0x mode").  I would suggest that these issues get reported
to the linux kernel bugzilla.

i have a onboard radeon 7000M (IBM xSeries 226 - appears to be PCI in
documentation). Following your comments i have compiled the 2.6.15.4 with
agpgart modules all disabled. With DRI and GLX extensions loaded, the Xorg 6.9
(Debian testing) the screen get's black (no image). Without the glx extensions,
it runs but no DRI.

I have experienced a lot lockups with this video onboard in this machine. I have
tried a lot configurations and i just get it to run secure (but a lot slow in
windows management, especially with a complex wallpaper) without DRI and GLX.
Comment 16 Joel Franco 2006-02-16 12:13:22 UTC
Created attachment 4619 [details]
dmesg

the dmesg after the Xorg loaded with dri and glx loaded
Comment 17 Joel Franco 2006-02-16 12:13:59 UTC
Created attachment 4620 [details]
xorg.conf

My xorg.conf
Comment 18 Joel Franco 2006-02-16 12:16:11 UTC
Created attachment 4621 [details]
Xorg.0.log
Comment 19 Joel Franco 2006-02-16 12:18:28 UTC
My Xorg process get full CPU usage with DRI.

joel@venus:~/tmpCLTXyI$ w
 23:17:21 up 16 min,  1 user,  load average: 1,01, 0,79, 0,38
USER     TTY      FROM              LOGIN@   IDLE   JCPU   PCPU WHAT
joel     pts/0    192.168.0.11     23:05    0.00s  5.07s  0.12s w
joel@venus:~/tmpCLTXyI$ 

This is the system iddle.

top - 23:18:07 up 17 min,  1 user,  load average: 1.00, 0.82, 0.40
Tasks:  78 total,   2 running,  76 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.4% us, 23.9% sy,  0.0% ni, 75.2% id,  0.5% wa,  0.0% hi,  0.0% si
Mem:   2075672k total,   150028k used,  1925644k free,     2788k buffers
Swap:  1004052k total,        0k used,  1004052k free,    74868k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND           
                                                                               
                     
 4653 root      25   0 37632 5204 3196 R 93.8  0.3   7:40.19 Xorg              
                                                                               
                      
 4769 joel      15   0  2196 1020  768 R  8.5  0.0   0:00.32 top               
                                                                               
                      
    1 root      16   0  1924  648  556 S  0.0  0.0   0:00.48 init              
                                                                               
                      
    2 root      RT   0     0    0    0 S  0.0  0.0   0:00.00 migration/0       
                                                                               
                      
    3 root      34  19     0    0    0 S  0.0  0.0   0:00.00 ksoftirqd/0       
                                                                               
                      
    4 root      RT   0     0    0    0 S  0.0  0.0   0:00.00 migration/1       
                                                                               
                      
    5 root      34  19     0    0    0 S  0.0  0.0   0:00.00 ksoftirqd/1       
                                                                               
                      
Comment 20 Joel Franco 2006-02-17 22:53:13 UTC
Based with some bugs related to multi processor, i compiled a new kernel without
SMP: still locks with dri and glx.

Them i disabled the HyperThreading in the BIOS, and with this new kernel (non
SMP) the problem appears to have disappeared. Now it's working.
Comment 21 Michel Dänzer 2006-05-09 03:48:00 UTC
Does this still happen with a current version of xf86-video-ati?
Comment 22 Joel Franco 2006-06-14 07:14:41 UTC
(In reply to comment #21)
> Does this still happen with a current version of xf86-video-ati?

I got the last radeon driver in
http://dri.freedesktop.org/snapshots/radeon-20060403-linux.i386.tar.bz2 and copy
the radeon_drv.so to my machine replacing the distribution file (ii 
xserver-xorg-video-ati           6.5.8.0-1                X.Org X server -- ATI
display driver).
I only replaced the radeon_drv.so.

It crashs too. It crashs with the HyperThreading enabled and disabled. It
happens while i'm logging to the gdm and before it gets to load all the gnome
components. i cannot ping the machine of another machine and i suppose that it
really crahs.

To i get to run it without problems, i put again the distribution file and set
the ColorDepth to 24 bits with 1600x1200 (xorg.conf) and because the onboard vga
have only 16M, the DRI is not activated with low graphics memory. This way, it
runs slow but is stable.
Comment 23 Michel Dänzer 2006-07-02 14:13:29 UTC
Please try again with current xf86-video-ati git, there has been a DRI stability
fix specifically for RV100 and R300 derivatives.
Comment 24 Joel Franco 2006-07-05 15:03:53 UTC
(In reply to comment #23)
> Please try again with current xf86-video-ati git, there has been a DRI stability
> fix specifically for RV100 and R300 derivatives.

sorry. i use debian testing and it still runs the xorg 7.0.
where i can download the current xf86-video-ati?
and
how can i put it in my dist without do a big files replace?

thank you
Comment 25 Michel Dänzer 2006-07-17 05:22:04 UTC
(In reply to comment #24)
> sorry. i use debian testing and it still runs the xorg 7.0.
> where i can download the current xf86-video-ati?

git clone git://git.freedesktop.org/git/xorg/driver/xf86-video-ati
cd xf86-video-ati
git checkout ati-1-0-branch

> and how can i put it in my dist without do a big files replace?

You can build a custom package using the debian/ directory from the Debian
source package, or you can just build it directly and copy over radeon_drv.so.
Comment 26 Luis Fernando Llana Díaz 2006-08-07 08:27:02 UTC
(In reply to comment #25)
> (In reply to comment #24)
> > sorry. i use debian testing and it still runs the xorg 7.0.
> > where i can download the current xf86-video-ati?
> 
> git clone git://git.freedesktop.org/git/xorg/driver/xf86-video-ati
> cd xf86-video-ati
> git checkout ati-1-0-branch
> 
> > and how can i put it in my dist without do a big files replace?
> 
> You can build a custom package using the debian/ directory from the Debian
> source package, or you can just build it directly and copy over radeon_drv.so.

I have been experienced this problem, and this solution seems to work. I have
been working for a while and the X system has not freezed.
Comment 27 Joel Franco 2006-08-07 08:43:41 UTC
(In reply to comment #26)
> (In reply to comment #25)
> > (In reply to comment #24)
> > > sorry. i use debian testing and it still runs the xorg 7.0.
> > > where i can download the current xf86-video-ati?
> > 
> > git clone git://git.freedesktop.org/git/xorg/driver/xf86-video-ati
> > cd xf86-video-ati
> > git checkout ati-1-0-branch
> > 
> > > and how can i put it in my dist without do a big files replace?
> > 
> > You can build a custom package using the debian/ directory from the Debian
> > source package, or you can just build it directly and copy over radeon_drv.so.
> 
> I have been experienced this problem, and this solution seems to work. I have
> been working for a while and the X system has not freezed.

yes. It really have worked and now my machine is stable again. Since i have
installed this, no more problems. And better: the direct rendering is working
fine. Thank You.
Comment 28 Erik Andren 2006-08-07 13:20:10 UTC
Ok to close this bug or are there still unresolved issues?
Comment 29 Michel Dänzer 2006-08-15 03:49:47 UTC
Closing, feel free to reopen if you still see the problem with current git.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.