Bug 896170 - Xorg display errors radeon - Failed to find memory space for buffer 0xffff8802034b2c48 eviction
Summary: Xorg display errors radeon - Failed to find memory space for buffer 0xffff880...
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Fedora
Classification: Fedora
Component: xorg-x11
Version: 18
Hardware: x86_64
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: X/OpenGL Maintenance List
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2013-01-16 19:10 UTC by Wolfgang Rupprecht
Modified: 2014-02-05 23:19 UTC (History)
3 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2014-02-05 23:19:02 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
screenshot of failed display. (15.55 MB, image/png)
2013-01-16 19:10 UTC, Wolfgang Rupprecht
no flags Details
Xorg.0.log file (74.58 KB, text/plain)
2013-01-16 19:15 UTC, Wolfgang Rupprecht
no flags Details
drm and TTM errors from /var/log/messages (12.95 KB, text/plain)
2013-01-16 19:21 UTC, Wolfgang Rupprecht
no flags Details

Description Wolfgang Rupprecht 2013-01-16 19:10:49 UTC
Created attachment 679752 [details]
screenshot of failed display.

Description of problem:
Xorg startup and user login takes forever and the log shows failed buffer allocation errors.  Most icons and jpegs (like the background) are distorted past recognition.

2013-01-16T04:46:38.463444-08:00 arbol kernel: [    4.483876] [drm] Initialized radeon 2.24.0 20080528 for 0000:01:00.0 on minor 0
2013-01-16T04:46:50.883021-08:00 arbol kernel: [   19.909046] [TTM] Failed to find memory space for buffer 0xffff8802034b2c48 eviction
2013-01-16T04:46:50.883041-08:00 arbol kernel: [   19.909053] [TTM] No space for ffff8802034b2c48 (9660 pages, 38640K, 37M)
2013-01-16T04:46:50.883043-08:00 arbol kernel: [   19.909056] [TTM]   placement[0]=0x00070002 (1)
2013-01-16T04:46:50.883044-08:00 arbol kernel: [   19.909058] [TTM]     has_type: 1
2013-01-16T04:46:50.883046-08:00 arbol kernel: [   19.909060] [TTM]     use_type: 1
2013-01-16T04:46:50.883047-08:00 arbol kernel: [   19.909061] [TTM]     flags: 0x0000000A
2013-01-16T04:46:50.883049-08:00 arbol kernel: [   19.909062] [TTM]     gpu_offset: 0x20000000
2013-01-16T04:46:50.883050-08:00 arbol kernel: [   19.909064] [TTM]     size: 131072
2013-01-16T04:46:50.883051-08:00 arbol kernel: [   19.909065] [TTM]     available_caching: 0x00070000
2013-01-16T04:46:50.883080-08:00 arbol kernel: [   19.909066] [TTM]     default_caching: 0x00010000
2013-01-16T04:46:50.883164-08:00 arbol kernel: [   19.909187] [TTM] Failed to find memory space for buffer 0xffff8802034b2c48 eviction


Version-Release number of selected component (if applicable):
xorg-x11-server-Xephyr.x86_64           1.13.1-4.fc18          installed        
xorg-x11-server-Xorg.x86_64             1.13.1-4.fc18          installed        
xorg-x11-server-common.x86_64           1.13.1-4.fc18          installed        
xorg-x11-server-devel.x86_64            1.13.1-4.fc18          installed        
xorg-x11-drv-ati.x86_64                 7.0.0-0.8.20121015gitbd9e2c064.fc18


How reproducible:
effects are variable, but errors in /var/log/messages are consistent.

Steps to Reproduce:
1. boot system
2. observe 30-sec  delay during Xorg startup
3. log in
4. observe 30-sec delay during login
5. observed broken icons and background image
6. observer errors tagged with kernel and drm or kernel and TTM in /var/log/messages.
  
Actual results:
bronken icons and background

Expected results:
no broken icons and background

Additional info:
this is a 3 monitor setup, but I've seen similar on a radeon with one monitor also.

Comment 1 Wolfgang Rupprecht 2013-01-16 19:15:50 UTC
Created attachment 679753 [details]
Xorg.0.log file

Xorg log file from above failure.

Comment 2 Wolfgang Rupprecht 2013-01-16 19:19:28 UTC
lspci shows this for the frame ATI Radeon buffer:

01:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI Cedar [FirePro 2460] (prog-if 00 [VGA controller])
	Subsystem: Advanced Micro Devices [AMD] nee ATI Device 2002
	Flags: bus master, fast devsel, latency 0, IRQ 44
	Memory at d0000000 (64-bit, prefetchable) [size=256M]
	Memory at fe8e0000 (64-bit, non-prefetchable) [size=128K]
	I/O ports at a000 [size=256]
	Expansion ROM at fe8c0000 [disabled] [size=128K]
	Capabilities: [50] Power Management version 3
	Capabilities: [58] Express Legacy Endpoint, MSI 00
	Capabilities: [a0] MSI: Enable+ Count=1/1 Maskable- 64bit+
	Capabilities: [100] Vendor Specific Information: ID=0001 Rev=1 Len=010 <?>
	Capabilities: [150] Advanced Error Reporting
	Kernel driver in use: radeon

Comment 3 Wolfgang Rupprecht 2013-01-16 19:21:21 UTC
Created attachment 679755 [details]
drm and TTM errors from /var/log/messages

Comment 4 Wolfgang Rupprecht 2013-01-16 19:25:02 UTC
I guess I should add that this card and monitor set worked fine under f16 and f17 for the past year.  The problem is a new lossage that started immediately after the f18 installation (via fedup --network 18).

Comment 5 Wolfgang Rupprecht 2013-01-27 18:06:59 UTC
I just noticed that if I only power up the 1200 x 1600 monitor of the 3-monitor set and reboot, then the above lossage doesn't happen.  

Any hints on how I can work around this problem?  It has been almost 2 weeks with not a peep from any developer.  Surely the above error messages indicate what is failing and suggest a work around to someone familiar with this code.

Comment 6 Wolfgang Rupprecht 2013-01-29 11:06:58 UTC
I tried a quick test with a 3.8.0 kernel and I'm still seeing failures.  While I don't see any broken icons, the desktop background image and some application windows are trashed beyond recognition.

kernel.x86_64                     3.8.0-0.rc5.git0.1.fc19 @/kernel-3.8.0-0.rc5.git0.1.fc19.x86_64

2013-01-29T02:50:10.063796-08:00 arbol kernel: [ 1964.922643] [TTM] Failed to find memory space for buffer 0xffff88018a5b5a10 eviction
2013-01-29T02:50:10.063821-08:00 arbol kernel: [ 1964.922654] [TTM] No space for ffff88018a5b5a10 (9660 pages, 38640K, 37M)
2013-01-29T02:50:10.063823-08:00 arbol kernel: [ 1964.922660] [TTM]   placement[0]=0x00010002 (1)
2013-01-29T02:50:10.063824-08:00 arbol kernel: [ 1964.922664] [TTM]     has_type: 1
2013-01-29T02:50:10.063825-08:00 arbol kernel: [ 1964.922667] [TTM]     use_type: 1
2013-01-29T02:50:10.063827-08:00 arbol kernel: [ 1964.922671] [TTM]     flags: 0x0000000A
2013-01-29T02:50:10.063828-08:00 arbol kernel: [ 1964.922674] [TTM]     gpu_offset: 0x20000000
2013-01-29T02:50:10.063829-08:00 arbol kernel: [ 1964.922678] [TTM]     size: 131072
2013-01-29T02:50:10.063830-08:00 arbol kernel: [ 1964.922682] [TTM]     available_caching: 0x00070000
2013-01-29T02:50:10.063832-08:00 arbol kernel: [ 1964.922685] [TTM]     default_caching: 0x00010000
2013-01-29T02:50:10.065008-08:00 arbol kernel: [ 1964.922882] [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!

Comment 7 Jérôme Glisse 2013-01-30 20:20:31 UTC
Please try if installing mesa package from this scratch build helps :

http://koji.fedoraproject.org/koji/taskinfo?taskID=4915728

You need to at least install mesa-libGL mesa-dri-drivers (from this build)

Comment 8 Wolfgang Rupprecht 2013-01-30 22:12:38 UTC
That seems to have done the trick.  No more broken jpegs and no TTM error messages in /var/log/message.

I installed all the rpms from http://kojipkgs.fedoraproject.org//work/tasks/5729/4915729

One nit left though.  There is still a ~30 second delay both 1) after the machine boots and before the login screen finishes drawing and 2) after one logs in and one's desktop screen gets drawn.   I had thought that this was a result of the failed TTM, but it appears that is not the case.  I get no TTM error messages but the delay persists.  This didn't happen under F16 and F17, but may well be an unrelated lossage.

Comment 9 Jérôme Glisse 2013-01-31 15:16:07 UTC
I am not seeing the long delay on login (thought i was seeing some rendering issue like your when in dual screen), i will retest again today.

To check if this delay is GPU/graphic related please boot with only a single screen connected and login with a single screen connected.

Comment 10 Fedora Update System 2013-01-31 20:05:51 UTC
mesa-9.0.1-4.fc18 has been submitted as an update for Fedora 18.
https://admin.fedoraproject.org/updates/mesa-9.0.1-4.fc18

Comment 11 Wolfgang Rupprecht 2013-01-31 21:07:51 UTC
I would give it karma on bodi if it would let me.  Captcha is inscrutable and bodi throws an internal 500 error when I try to log in.  (Login problems are probably IPv6 related.  I've seen this issue before.)

Comment 12 Wolfgang Rupprecht 2013-02-01 12:47:00 UTC
There is a much longer delay with 3 monitors than with 1 monitor (1200 x 1600).

login from password after carriage return to cinnamon bottom bar drawn:
1 monitor:   11 sec
3 monitors:  28 sec

logout from clicking the logout confirmation button to gdm redraw of login selector screen:
1 monitor:    7 sec
3 monitors:  58 sec

The logout is most dramatic with over 8x the delay.

Comment 13 Jérôme Glisse 2013-02-05 15:52:24 UTC
Can you try with no monitor in portrait mode ie all monitor non rotated.

Comment 14 Wolfgang Rupprecht 2013-02-05 18:44:47 UTC
Sorry for any confusion if I presented the pixel count in a backwards fashion.  All monitors are in landscape modem with the long-axis horizontal.

Comment 15 Fedora Update System 2013-02-08 02:08:47 UTC
mesa-9.0.1-4.fc18 has been pushed to the Fedora 18 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 16 Wolfgang Rupprecht 2013-02-08 23:31:55 UTC
The delay still persists with mesa-9.0.1-4.fc18 from the stable directory.     Rendering issues appear fixed.

Comment 17 Wolfgang Rupprecht 2013-03-20 14:19:19 UTC
The render problems just came back with the latest mesa.  (from this morning's yum update).
mesa-dri-drivers.x86_64                      9.1-1.fc18                @updates 
mesa-dri-filesystem.x86_64                   9.1-1.fc18                @updates 
mesa-libEGL.i686                             9.1-1.fc18                @updates 
mesa-libEGL.x86_64                           9.1-1.fc18                @updates 
mesa-libEGL-devel.x86_64                     9.1-1.fc18                @updates 
mesa-libGL.i686                              9.1-1.fc18                @updates 
mesa-libGL.x86_64                            9.1-1.fc18                @updates 
mesa-libGL-devel.x86_64                      9.1-1.fc18                @updates 
mesa-libGLES.x86_64                          9.1-1.fc18                @updates 
mesa-libGLES-devel.x86_64                    9.1-1.fc18                @updates 
mesa-libGLw.x86_64                           6.5.1-13.fc18             installed
mesa-libGLw-devel.x86_64                     6.5.1-13.fc18             installed
mesa-libOSMesa.x86_64                        9.1-1.fc18                @updates 
mesa-libOSMesa-devel.x86_64                  9.1-1.fc18                @updates 
mesa-libgbm.i686                             9.1-1.fc18                @updates 
mesa-libgbm.x86_64                           9.1-1.fc18                @updates 
mesa-libgbm-devel.x86_64                     9.1-1.fc18                @updates 
mesa-libglapi.i686                           9.1-1.fc18                @updates 
mesa-libglapi.x86_64                         9.1-1.fc18                @updates 
mesa-libwayland-egl.x86_64                   9.1-1.fc18                @updates 
mesa-libwayland-egl-devel.x86_64             9.1-1.fc18                @updates 
mesa-libxatracker.x86_64                     9.1-1.fc18                @updates 
mesa-libxatracker-devel.x86_64               9.1-1.fc18                @updates 

2013-03-20T06:58:56.779530-07:00 arbol kernel: [12403.771774] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:57.056536-07:00 arbol kernel: [12404.048478] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:57.361530-07:00 arbol kernel: [12404.352263] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:57.641536-07:00 arbol kernel: [12404.631654] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:57.923529-07:00 arbol kernel: [12404.913671] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:58.206531-07:00 arbol kernel: [12405.195913] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-03-20T06:58:58.558549-07:00 arbol kernel: [12405.546935] [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!

X11 is very unstable and this is the 5th time I've tried to put in this bug report.  Xorg keeps killing the display manager with "snap something went wrong please log out."   (From memory).

To add frosting on the cake yum won't downgrade mesa.   This is a disaster.  The system is unusable in the current state.

Comment 18 Wolfgang Rupprecht 2013-03-21 12:18:37 UTC
Did this patch get dropped from recent mesa builds?  

http://koji.fedoraproject.org/koji/buildinfo?buildID=381632

* Thu Jan 31 2013 Jerome Glisse <jglisse> 9.0.1-4
- force r600g to stay in gpu memory limit

For example, this build doesn't show it:

http://koji.fedoraproject.org/koji/buildinfo?buildID=402325

It also appears to be the build that made it into the updates repository.

Comment 19 Wolfgang Rupprecht 2013-05-03 22:50:13 UTC
2013-05-03T15:33:19.004999-07:00 arbol kernel: [   41.745231] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.140951-07:00 arbol kernel: [   41.880755] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.297982-07:00 arbol kernel: [   42.038104] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.462971-07:00 arbol kernel: [   42.203029] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.495980-07:00 arbol dbus-daemon[561]: ** Message: No devices in use, exit
2013-05-03T15:33:19.621963-07:00 arbol kernel: [   42.361652] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.623028-07:00 arbol apcupsd[2231]: 000.0,000.0,000.0,27.27,00.00,33.0,00.0,000.0,000.0,117.0,100.0,1
2013-05-03T15:33:19.787021-07:00 arbol kernel: [   42.526509] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:19.958949-07:00 arbol kernel: [   42.698989] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:20.126039-07:00 arbol kernel: [   42.865547] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:20.293043-07:00 arbol kernel: [   43.032486] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:20.458027-07:00 arbol kernel: [   43.197119] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:20.628026-07:00 arbol kernel: [   43.367179] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:20.650040-07:00 arbol kernel: [   43.389365] [TTM] Failed to find memory space for buffer 0xffff8808114b2848 eviction
2013-05-03T15:33:20.650062-07:00 arbol kernel: [   43.389374] [TTM] No space for ffff8808114b2848 (9660 pages, 38640K, 37M)
2013-05-03T15:33:20.650064-07:00 arbol kernel: [   43.389378] [TTM]   placement[0]=0x00010002 (1)
2013-05-03T15:33:20.650065-07:00 arbol kernel: [   43.389381] [TTM]     has_type: 1
2013-05-03T15:33:20.650066-07:00 arbol kernel: [   43.389383] [TTM]     use_type: 1
2013-05-03T15:33:20.650067-07:00 arbol kernel: [   43.389385] [TTM]     flags: 0x0000000A
I'm still seeing these.  Most of the time it knocks me back to the GDM login screen.  This is a serious bug.  Could we at least have a work-around?   Any way to turn off hardware rendering?  Slow is better than broken.

2013-05-03T15:33:20.650069-07:00 arbol kernel: [   43.389388] [TTM]     gpu_offset: 0x20000000
2013-05-03T15:33:20.650087-07:00 arbol kernel: [   43.389390] [TTM]     size: 131072
2013-05-03T15:33:20.650089-07:00 arbol kernel: [   43.389392] [TTM]     available_caching: 0x00070000
2013-05-03T15:33:20.650090-07:00 arbol kernel: [   43.389395] [TTM]     default_caching: 0x00010000
2013-05-03T15:33:20.713960-07:00 arbol kernel: [   43.452836] [drm:radeon_cs_ioctl] *ERROR* Failed to parse relocation -12!
2013-05-03T15:33:20.973969-07:00 arbol kernel: [   43.712562] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.086025-07:00 arbol kernel: [   43.825071] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.188945-07:00 arbol kernel: [   43.928281] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.289019-07:00 arbol kernel: [   44.027433] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.394019-07:00 arbol kernel: [   44.132883] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.494974-07:00 arbol kernel: [   44.233393] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit
2013-05-03T15:33:21.595972-07:00 arbol kernel: [   44.335135] radeon_gem_object_create:69 alloc size 339Mb bigger than 256Mb limit

Comment 20 lefred 2013-05-14 08:16:25 UTC
I've exactly the same issue, and it seems to happen more and more often :(

[  170.384151] radeon_gem_object_create:69 alloc size 325Mb bigger than 256Mb limit
[  170.525861] radeon_gem_object_create:69 alloc size 325Mb bigger than 256Mb limit


kernel:  3.8.11-200.fc18.x86_64

mesa packages:

mesa-libEGL-devel-9.1-3.fc18.x86_64
mesa-dri-filesystem-9.1-3.fc18.x86_64
mesa-libxatracker-9.1-3.fc18.x86_64
mesa-libglapi-9.1-3.fc18.i686
mesa-libGLU-9.0.0-1.fc18.x86_64
mesa-libEGL-9.1-3.fc18.x86_64
mesa-libglapi-9.1-3.fc18.x86_64
mesa-libgbm-9.1-3.fc18.i686
mesa-libEGL-9.1-3.fc18.i686
mesa-libgbm-9.1-3.fc18.x86_64
mesa-libGL-9.1-3.fc18.i686
mesa-libGL-devel-9.1-3.fc18.x86_64
mesa-libGL-9.1-3.fc18.x86_64
mesa-dri-drivers-9.1-3.fc18.x86_64

I also have 3 screens.

Comment 21 Wolfgang Rupprecht 2013-05-14 09:27:12 UTC
I find that this is 100% reproducible with the current F18 by full-sizing a window on one of the 2560x1440 monitors.  It certainly seems like a buffer-size problem.  The larger the buffer, the more likely the problem is to occur.

If hardware mesa is so broken, can someone please just fix the code so we can turn HW mesa off again and just use SW?

Comment 22 Fedora End Of Life 2013-12-21 15:43:54 UTC
This message is a reminder that Fedora 18 is nearing its end of life.
Approximately 4 (four) weeks from now Fedora will stop maintaining
and issuing updates for Fedora 18. It is Fedora's policy to close all
bug reports from releases that are no longer maintained. At that time
this bug will be closed as WONTFIX if it remains open with a Fedora 
'version' of '18'.

Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version' 
to a later Fedora version prior to Fedora 18's end of life.

Thank you for reporting this issue and we are sorry that we may not be 
able to fix it before Fedora 18 is end of life. If you would still like 
to see this bug fixed and are able to reproduce it against a later version 
of Fedora, you are encouraged  change the 'version' to a later Fedora 
version prior to Fedora 18's end of life.

Although we aim to fix as many bugs as possible during every release's 
lifetime, sometimes those efforts are overtaken by events. Often a 
more recent Fedora release includes newer upstream software that fixes 
bugs or makes them obsolete.

Comment 23 Fedora End Of Life 2014-02-05 23:19:02 UTC
Fedora 18 changed to end-of-life (EOL) status on 2014-01-14. Fedora 18 is
no longer maintained, which means that it will not receive any further
security or bug fix updates. As a result we are closing this bug.

If you can reproduce this bug against a currently maintained version of
Fedora please feel free to reopen this bug against that version. If you
are unable to reopen this bug, please file a new report against the
current release. If you experience problems, please add a comment to this
bug.

Thank you for reporting this bug and we are sorry it could not be fixed.


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