Radeon module loaded before agpgart

I have the same problem.

Not I added a rc script to do something like this.

sudo rmmod radeon
sudo rmmod ttm
sudo rmmod drm
sudo modprobe agpgart radeon
sudo /etc/init.d/gdm restart

I can confirm this too. A better workaround is to put the following in /etc/modprobe.d/radeon.conf e.g.:

install radeon /sbin/modprobe agpgart; /sbin/modprobe --ignore-install radeon

Looking at dmesg it seems like agpgart and radeon are initialized in parallel on my system.

[ 1.872289] Linux agpgart interface v0.103
[ 1.924626] usb 4-1: new full speed USB device using ohci_hcd and address 2
[ 2.047508] [drm] Initialized drm 1.1.0 20060810
[ 2.094656] [drm] radeon default to kernel modesetting DISABLED.
[ 2.095009] pci 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[ 2.095302] [drm] Initialized radeon 1.31.0 20080528 for 0000:01:00.0 on minor 0
[ 2.099769] agpgart-sis 0000:00:00.0: SiS chipset [1039/0648]
[ 2.104715] agpgart-sis 0000:00:00.0: AGP aperture is 64M @ 0xe8000000

My previous fix didn't work every time so now I have changed it to:

install radeon /sbin/modprobe agpgart; sleep 0.5; /sbin/modprobe --ignore-install radeon

I also noticed that problem. Most times when I start my mythbuntu machine I have no 3D accelaration:

(WW) RADEON(0): [agp] AGP not available
(EE) RADEON(0): [agp] AGP failed to initialize. Disabling the DRI.
(WW) RADEON(0): Direct rendering disabled

And yes, it is caused by drm and radeon drivers loaded before agpgart module.

Sometimes it is possible to stop gdm service, then unload radeon, drm and ttm kernel modules, load them again and get 3D acceleration working. But sometimes I get only black screen and top shows that Xorg is eating almost 100% of my cpu.

Problem seems to be similar to describe in bug https://bugs.launchpad.net/ubuntu/+source/linux/+bug/430694

My machine is an AMD Sempron 2400+ / VIA KT600 based desktop pc with Radeon 8500DV (R200) graphics.

Confirmed form me (kubuntu 9.10 up to date)

The workaround with /etc/modprobe.d/radeon.conf don't work for me

This workaround work for me :
edit /etc/init/kdm.conf and add two ligne before exec kdm

...
    initctl emit starting-dm DM=kdm

        # fix pour imposer un chargement de radeon en dernier
        modprobe -r radeon
        modprobe radeon
    exec kdm

end script

May by this workaround can help

Thanks Cyril, your workaround works :) (with a small difference: I use gdm, so I've added instructions for reload radeon driver to my /etc/init/gdm.conf).

I've also found that extremely high cpu usage was caused by reloading not only radeon, but also ttm and drm modules. When I reload only radeon then Xorg starts normally with 3D acceleration enabled.

I have the same problem.
I've tried Cyril's workaround, and all I get after rebooting is a no signal blank screen and no command available.

My problem still exists. Reloading radeon module worked fine untill I powered down machine. Then I had my X server using 100% of CPU and black screen.

I have tried with EXA and XAA acceleration modes, both with similar results. It seems to me that I can get accelerated X only when my desktop first starts in non-accelerated mode and then I reload radeon module and start X again.
But even then when I try to play some 3D games (e.g chromium), then I see artifacts on the screen (eg. red and white vertical stripes) during reboot (restoring text mode), so accelerated modes still have problems.

When I use Cyril's method during startup X hangs with 100% CPU usage. Maybe graphic card is initialized in some way when it probes different screen modes in non-agp mode (reloading modules before gdm start prevents that probing in non-accelerated mode).

Unfortunately I can't do more tests at ths time. I have had to return to my previous, non-accelerated mode because my wife use that computer for TV watching ;) Maybe I find some time during weekend to test it again in more digested way. If you have any suggestions how to test, please post it here or send me by email.

Today I did a few tests more, and finally got best results with KMS enabled. Now I have working X server with DRI2, 3D acceleration and really fast switching between VTs and X desktop. 3D performance is worse than it was before with DRI, but I have no issues when powering down (previously was: red&white artifacts, system hangs) and didn't experience either black screen or high CPU usage after startup :)

You can easy test KVM without restarting, just stop gdm and reload radeon driver with option modeset=1, then start gdm again.

   sudo stop gdm
   sudo modprobe -r radeon
   sudo modprobe radeon modeset=1
   sudo start gdm

To enable KVM automatically I use modified Cyril's method (just added modeset=1):
    ...
    initctl emit starting-dm DM=gdm

# 3D acceleration FIX: re-init of radeon driver (should be loaded AFTER agpgart kernel module); enable KVM
    modprobe -r radeon ; modprobe radeon modeset=1

    exec gdm-binary $CONFIG_FILE

(I mean KMS, not KVM)

Same problem here.
Work with

sudo stop gdm
sudo modprobe -r radeon drm
sudo modprobe radeon modeset=0
sudo start gdm

I confirm the same problem with 10.04 LTS Lucid Lynx Alpha1

i have this bug too with the stock karmic kernel, but it's non-issue with the 2.6.32-996.200912071003 drm-next mainline kernel

I install "linux-image-2.6.32-996-generic_2.6.32-996.200912071003_i386"
and works fine!

Thanks a.tom!!

I'm also experiencing the same issue with the following messages in xorg.0.log:
"RADEON(0): [agp] AGP not available"
"RADEON(0): [agp] AGP failed to initialize. Disabling the DRI."
"RADEON(0): [agp] You may want to make sure the agpgart kernel module is loaded before the radeon kernel module."

I found that enabling the Control + Alt + Backspace option to restart the X server followed by pressing Control + Alt + Backspace allowed me to log back into Ubuntu with 3D acceleration working second time round. As this requires logging in twice I will probably try the other work arounds mentioned above but a proper fix would be appreciated.

Hej there
I am a Danish Linux newbie.
I have the trouble of using 3d rendering on ati 9600 pro and mobo asrock k7Upgrade-600.
It's a VIA KT400/KT400A/KT600 chipset...

Following those steps opposed by Cyril and Marcin I got it to work
However I noticed that whenever 3D worked I got a message inside KDE (kubuntu) (kernel 2.6.31-17 and -14) asking if I wanted to disable alsa drivers permanently for the missing device. ???
That was late last night...

This morning when starting up I just got blank screens and lock-ups.
What the f....... I thought over things a bit and then hard reboot several times. Same result.

Now something got me thinking of thislsa warning and I as an windows oldie thought perhaps there is some conflict between my agp and my AC97 device so into bios, disable snd device, and reboot.
”wupti.com” it worked.

Now 3D rendering, menus, tty shift, everything is working like a charm.
But no sound.....

:-|
/erling

I've had it too and Lars Ljung's solution works for me.

I try to install Alpha3 without success!
Then in boot options write radeon.modeset=0 and after that i can install it!
The result is X freeze every time when i do something!
Upgrade to Beta1 without difference!
I try to boot from cd Lucid 10.04 Beta1 without success!
Work with radeon.modeset=0 in boot options, and then i must to run

sudo stop gdm
sudo modprobe -r radeon drm
sudo modprobe radeon modeset=0
sudo start gdm

like 9.10 and after that, is everything ok but only from cd boot!

I guess nothing has changed in comparison with 9.10!

For me it is ok, the bug has disappear without workaround. I try to make some tests and I come back.

Unfortunately it has came back after a reboot, and i can't explain it.

I can confirm that Marcin's workaround works for me.

$ lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc Radeon Mobility X700 (PCIE)

$ uname -a
Linux abel 2.6.31-20-generic #58-Ubuntu SMP Fri Mar 12 05:23:09 UTC 2010 i686 GNU/Linux

I confirm the same problem with 10.04 LTS Lucid Lynx Beta2.

I try to boot from cd and i can' t.
Work with radeon.modeset=0 in boot options and then i must run

sudo stop gdm
sudo modprobe -r radeon drm
sudo start gdm

and after that, all is fine!

I confirm the same problem with 10.04 LTS Lucid Lynx RC
I try to boot from cd and i can' t.
i write radeon.modeset=0 in boot options but doesn' t work!

I can confirm this bug too. Using 10.04 LTS Lucid Lynx RC (LiveCD only atm). Can't boot with KMS enabled. At some point, boot process hangs repeatedly showing message "BUG: soft lockup - CPU #0 stuck for 61s! [plymouthd:1183]" every minute or so. Somehow I were able to boot into command prompt and after looking trough logs, I saw error messages involving AGP GART and agpgart-amd64 loading after DRM initialization just as described in bug description here. I can boot successfully if I enable "nomodeset" boot option but splash (plymouth) is shown in low res and low color mode and DRI (including 3D acceleration) is not working.

Fedora 13 Beta works correctly but judging by lsmod an dmesg, Fedora seems to have agpgart-amd64 module built into the kernel. It loads before DRM initialization as expected.

Computer:
ASUS K8N (nforce3) motherboard
Athlon64 3200+
Radeon 9250 (RV280), 256MB

I try to install 10.04 LTS Lucid Lynx, but the problem persists.

I try to boot from cd and i can' t.
Work with radeon.modeset=0 in boot options and then i must run

sudo stop gdm
sudo modprobe -r radeon drm
sudo start gdm

and after that, all is fine!

Maybe the same fix is needed as for intel in bug 542251 (https://lists.ubuntu.com/archives/kernel-team/2010-April/009858.html).

I've got the same issue that Vagrale, I can't boot from a live-cd whithout radeonmodeset=0 in kernel options.

[This is an automatic notification.]

Hi Tobe,

This bug was reported against an earlier version of Ubuntu, can you
test if it still occurs on Lucid?

Please note we also provide technical support for older versions of
Ubuntu, but not in the bug tracker. Instead, to raise the issue through
normal support channels, please see:

    http://www.ubuntu.com/support

If you are the original reporter and can still reproduce the issue on
Lucid, please run the following command to refresh the report:

  apport-collect 468413

If you are not the original reporter, please file a new bug report, so
we can work with you as the original reporter instead (you can reference
bug 468413 in your report if you think it may be related):

  ubuntu-bug xorg

If by chance you can no longer reproduce the issue on Lucid or if you
feel it is no longer relevant, please mark the bug report 'Fix Released'
or 'Invalid' as appropriate, at the following URL:

  https://bugs.launchpad.net/ubuntu/+bug/468413

I've tried the latest Lucid release and it works. No more black screen anymore. AGP is recognized correctly.

@vagrale - thanks for providing that info, but we would like to see it from the original reporter. We may want to send this bug report upstream, but upstream has a hard and fast rule that logs from someone other than the original reporter are not acceptable. The reason for that is that quite often with X.org bug reports, two people may *think* they have the same issue based on commonality of symptoms and/or hardware, but in truth they have separate issues and mixing the report and the files can cause the developers some confusion.

If this bug report seems not to be getting attention, I would encourage you to file a new bug report with this data, and simply reference this bug report in your bug description.

For me the problem seems to be solved. The lucid live CD start normal and 3d acceleration works.

Architecture: i386
CurrentDmesg:
 [ 61.540293] lp: driver loaded but no devices found
 [ 62.235869] ppdev: user-space parallel port driver
 [ 66.404009] eth0: no IPv6 routers present
DistroRelease: Ubuntu 10.04
DkmsStatus: Error: [Errno 2] No such file or directory
GdmLog1: Error: command ['gksu', '-D', 'Apport', '--', 'cat', '/var/log/gdm/:0.log.1'] failed with exit code 1: cat: /var/log/gdm/:0.log.1: No such file or directory
GdmLog2: Error: command ['gksu', '-D', 'Apport', '--', 'cat', '/var/log/gdm/:0.log.2'] failed with exit code 1: cat: /var/log/gdm/:0.log.2: No such file or directory
LiveMediaBuild: Ubuntu 10.04 LTS "Lucid Lynx" - Release i386 (20100429)
Lsusb:
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
MachineType: ASUSTek Computer Inc. K8N
Package: xserver-xorg-video-ati 1:6.13.0-1ubuntu5
PackageArchitecture: i386
ProcCmdLine: BOOT_IMAGE=/casper/vmlinuz file=/cdrom/preseed/hostname.seed boot=casper initrd=/casper/initrd.lz quiet splash --
ProcEnviron:
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.32-21.32-generic 2.6.32.11+drm33.2
Tags: lucid lucid
Uname: Linux 2.6.32-21-generic i686
UserGroups: adm admin cdrom dialout lpadmin plugdev sambashare
dmi.bios.date: 02/16/2006
dmi.bios.vendor: American Megatrends Inc.
dmi.bios.version: 1011.005
dmi.board.name: 'K8N'
dmi.board.vendor: ASUSTeK Computer INC.
dmi.board.version: Rev 1.xx
dmi.chassis.asset.tag: Asset-1234567890
dmi.chassis.type: 3
dmi.chassis.vendor: Chassis Manufacture
dmi.chassis.version: Chassis Version
dmi.modalias: dmi:bvnAmericanMegatrendsInc.:bvr1011.005:bd02/16/2006:svnASUSTekComputerInc.:pnK8N:pvrToBeFilledByO.E.M.:rvnASUSTeKComputerINC.:rn'K8N':rvrRev1.xx:cvnChassisManufacture:ct3:cvrChassisVersion:
dmi.product.name: K8N
dmi.product.version: To Be Filled By O.E.M.
dmi.sys.vendor: ASUSTek Computer Inc.
glxinfo: Error: [Errno 2] No such file or directory
system:
 distro: Ubuntu
 codename: lucid
 architecture: i686
 kernel: 2.6.32-21-generic

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

apport information

Sorry for this. apport-collect didn't really work as I expected.

First of all I have to say that I'm *sure* my problem is same as the problem original reporter has. We have almost the same hardware and logs show the same problem.

Anyway, the bug is still present in Lucid Final LiveCD. agpgart, amd64_agp, drm and radeon modules are loaded in parallel so there is race condition. The logs I attached up there are of a system that booted successfully ... BUT... it was only 1 out of 4 of my attempts to boot the system. In other three attempts system failed to boot. Sometimes it boots to command line, sometimes it hangs completely, it all depends on order in which modules finished loading.

I will attempt to collect the logs of a system that fails to boot and attach the logs.

I have attached a boot log (dmesg) when system fails to boot properly. As can be clearly seen, agpgart-amd64 is loaded after the drm and radeon modules, too late infact, and system either hangs or drops to command line. If you compare this boot log to the boot log I attached earlier (with apport-collect), you'll see that the order in which modules in question are loaded is completely different. The order seems to be random every time the system is booted. In a boot log of a successful attempt to boot the system, agpgart-amd64 module is loaded after the agpgart module and just before drm. Everything (KMS, DRI, 3D accel.) worked perfectly after that. In a failed attempt to boot, agpgart-amd64 module loaded much, much later.

Boot log snippets:

*******
Fail:
*******
[ 2.938953] Linux agpgart interface v0.103
...
[ 3.058004] [drm] Initialized drm 1.1.0 20060810
...
[ 3.394070] [drm] radeon defaulting to kernel modesetting.
[ 3.394074] [drm] radeon kernel modesetting enabled.
[ 3.394401] ACPI: PCI Interrupt Link [LNKE] enabled at IRQ 19
[ 3.394406] alloc irq_desc for 19 on node -1
[ 3.394408] alloc kstat_irqs on node -1
[ 3.394418] radeon 0000:01:00.0: PCI INT A -> Link[LNKE] -> GSI 19 (level, low) -> IRQ 19
[ 3.397169] [drm] radeon: Initializing kernel modesetting.
[ 3.397258] [drm] register mmio base: 0xFF4F0000
[ 3.397260] [drm] register mmio size: 65536
[ 3.398712] [drm] GPU reset succeed (RBBM_STATUS=0x00000140)
[ 3.398732] [drm] Generation 2 PCI interface, using max accessible memory
[ 3.398746] BUG: unable to handle kernel NULL pointer dereference at 00000040
[ 3.398860] IP: [<f8270a49>] radeon_agp_init+0x19/0x3d0 [radeon]
[ 3.398996] *pde = 00000000
[ 3.399068] Oops: 0000 [#1] SMP
[ 3.399173] last sysfs file: /sys/devices/pci0000:00/0000:00:08.0/host0/uevent
[ 3.399215] Modules linked in: radeon(+) ttm drm_kms_helper drm amd64_agp(+) floppy i2c_algo_bit forcedeth(+) sata_nv(+) pata_amd agpgart
[ 3.399667]
[ 3.399705] Pid: 215, comm: modprobe Not tainted (2.6.32-21-generic #32-Ubuntu) K8N
[ 3.399748] EIP: 0060:[<f8270a49>] EFLAGS: 00010282 CPU: 0
[ 3.399803] EIP is at radeon_agp_init+0x19/0x3d0 [radeon]
[ 3.399843] EAX: f68da400 EBX: f73df000 ECX: d0000000 EDX: 00000000
[ 3.399884] ESI: f68da400 EDI: 00000008 EBP: f6a1fd48 ESP: f6a1fcf4
[ 3.399924] DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
[ 3.399965] Process modprobe (pid: 215, ti=f6a1e000 task=f6a20000 task.ti=f6a1e000)
[ 3.400006] Stack:
[ 3.400020] f6a10020 c00c0000 f70f1000 c00cd000 f6a1fd10 452b1daf f73df000 10000000
[ 3.400020] <0> 00000008 f6a1fd28 c0588f42 f82d9e2c f6a1fd34 f6a1fd40 f829604b f82d9e2c
[ 3.400020] <0> f68da400 f73df000 f73df000 f68da400 00000008 f6a1fd54 f829386f f73df000
[ 3.400020] Call Trace:
[ 3.400020] [<c0588f42>] ? printk+0x1d/0x23
[ 3.400020] [<f829604b>] ? r100_get_accessible_vram+0x11b/0x150 [radeon]
[ 3.400020] [<f829386f>] ? r100_mc_init+0x9f/0xe0 [radeon]
[ 3.400020] [<f8297f72>] ? r100_init+0xd2/0x270 [radeon]
[ 3.400020] [<c058b5af>] ? _spin_lock_irqsave+0x2f/0x50
[ 3.400020] [<c03cbf4b>] ? vga_client_register+0x6b/0x80
...

The original reporter indicates the issue they reported is solved. If anyone else thinks they have a similar problem, it's best to file a new bug report instead of reopening this one.