Xorg crashed with " Failed to allocate CMM buffer (size=16384, alignment=4096)!" in AddScreen() as a result of a call to atiddxLoadLogo() in fglrx

[W. Prins]

Ubuntu release: 8.10 64bit
xserver-xorg-core: 2:1.5.2-2ubuntu3
xserver-xorg: 1:7.4~5ubuntu3
xorg-driver-fglrx: 2:8.543-0ubuntu4
Video card: Asus ATI Radeon 9800XT 256MB AGP
Motherboard: Asus K8V SE Deluxe

X Windows will fail to start properly from a cold boot (e.g from power off) using my Radeon 9800XT, reporting the error mesages:
(EE) fglrx(0): Failed to allocate CMM buffer (size=16384, alignment=4096)!
(==) fglrx(0): Using hardware cursor
(EE) fglrx(0): Failed to allocate CMM buffer (size=4096, alignment=4096)!
(EE) fglrx(0): Failed to allocate CMM buffer (size=4096, alignment=4096)!

Booting first to Windows, and then hot-booting (shutdown/restart or reset button) yields a successful boot of X Windows.

Procedure to reproduce:
1.) Install Ubuntu, Install fglrx (8.543), with above hardware
2.) Shut down machine completely (power off)
3.) Attempt to boot Ubuntu. X startup will fail with the above message.
4.) Boot to Windows. Then restart again (don't turn the machine off.)
5.) Attempt to boot to Ubuntu. It will now succeed using fglrx.

Please see my latest comment for updated logs and configuration.

[Eduardo Siemann (eduardosiemann)]

I have same error =/

Ubuntu release: 8.10 final
Video card: Asus ATI Radeon 9550 256MB AGP
Motherboard: Asus K8V-X SE

[W. Prins (wprins)]

I don't think this bug is correctly marked as a duplicate. The error message is different. For the record, I'm still having the problem but have today noticed that the problem only happens from a cold boot -- if you e.g. boot Windows first, then hot-boot Linux (shutdown-reboot Windows or press reset button) and then boot Ubuntu then the fglrx driver works fine with the exact same configuration. If you cold-boot (from total off condition) the PC straight into Ubuntu, then this error results. I've today reinstalled this test box with 8.10 Intrepid, and include for reference the log from the previous failed boot. (I'm currently writing this from a successful boot, made possible by booting Windows first then soft-booting.) Please note I've also posted this same information on the other bug report this is currently marked a duplicate of.

[W. Prins (wprins)]

I've unmarked this bug as duplicate as I did not correctly mark it as a duplicate in the first place.

[W. Prins (wprins)]

For what it's worth, I've reported this problem to ATI directly, via their website.

[Bryce Harrington (bryce)]

Hi walter, can you give the link to the report you made on the ATI website?

[W. Prins (wprins)]

Unfortunately ATI's site does not give back a link, it is a simple submission page that simply thanks you for the submission and doesn't give you any link back to the report. So they have the report but we have no way of knowing what if anything they're doing with it. I'll go and post a link to this page to them so they will (hopefully) get back to us as and when anything is done about this problem.

[W. Prins (wprins)]

Can I mark this report as "Confirmed" seeing as (at least) one other person is having the same issue? Or does it not work like that?

[Emertyl (emertyl)]

Sorry for my english, and hopping you understand me well (I'm French!!).
I'm trying to deal with this problem for a few weeks, but have no solution yet.
I have the same bug: X doesn't start and a window tell me ths following message:

(EE) fglrx(0): Failed to allocate CMM buffer (size=16384, alignment=4096)!
(EE) fglrx(0): Failed to allocate CMM buffer (size=4096, alignment=4096)!
(EE) fglrx(0): Failed to allocate CMM buffer (size=4096, alignment=4096)!

I notice two things more: the dmesg message show me that

[fglrx] GART Table is not in FRAME_BUFFER range

this message appears three times too. Don't know if a link exists between the two sorts of message???

As I disable DRI in my /etc/X11/xorg.conf, I could connect correctly without having the above X-message, and I have a good resolution, but have no 3D...

[W. Prins (wprins)]

Emertyl, do you have Windows on your machine as well? I've found on my machine that first booting into Windows seems to initialize something on my card, and if you then hot-boot (restart Windows or press reset button) to Ubuntu then this problem does not occur. What exact video card do you use?

Google Traduction:
Emertyl, avez-vous Windows sur votre machine ainsi? J'ai trouvé sur ma machine que le premier démarrage de Windows apparaît pour initialiser quelque chose sur ma carte, et si vous avez chaud-boot (redémarrez Windows ou appuyez sur le bouton reset) pour Ubuntu alors ce problème ne se pose pas. Qu'est-ce exact de carte vidéo utilisez-vous?

[W. Prins (wprins)]

Confirming as another person has reported having this problem.

[Emertyl (emertyl)]

I don't have Windows any more, only Ubuntu 8.10 Intrepid Ibex.
My video card is an ATI Radeon 9600SE 128MB DDR AGP.

(thanks for the translation :):) but your English is very good, and I understand it well :) )

[Jeff Trull (jetrull)]

I have a K8V-X SE and although I do not use the "fglrx" driver I know there is a kernel issue with these boards that messes up AGP and DRI. One way to tell you have this particular issue is if you can use DRI in 64b mode but not 32b. Have any of you tried installing a 64b (x86_64) kernel? If it works but 32b (i686) does not, you may have the same problem I have... a kernel fix is on the way.

[Thomas Staverman (thomas-staverman)]

I have the same problem also. Ubunto 8.10 and, i think, ati radeon 9550. No windows on this machine.
I can't choose a proper resolution for my monitor, which is pretty bad when watching movies...

[W. Prins (wprins)]

@ Jeff: The machine I have the problem on runs 64bit IIRC, I've just updated the description to reflect that.

@Thomas: Try disabling DRI as per Emertyl's post. (edit /etc/X11/xorg.conf and add a section:)

Section "Module"
   Disable "dri"
EndSection

(Untested/off the top of my head so please backup your xorg.conf file first and don't try this is you don't know how to undo the change if need be. I'll test/verify/confirm this workaround tonight when I've had a chance to try it.)

[W. Prins (wprins)]

To be clear: If you already have a "Module" section you should add the line to it (and if there's already a line "Load DRI" then then new line replaces it.

[Bryce Harrington (bryce)]

I've posted a new version of the -fglrx driver to our xorg-edgers PPA,
would you mind testing it either on Jaunty or Karmic and see if it
resolves this bug?

Get fglrx 8.620 here:

  https://edge.launchpad.net/~xorg-edgers/+archive/ppa

[W. Prins (wprins)]

Hi, I've finally gotten back to this issue, having upgraded/reinstalled the box in question. What I tried:

1.) Installed the latest/daily version of Karmic. Noticed the open source ATI driver worked well, including acceleration. Noticed the "hardware drivers" applet didn't offer any proprietary drivers. Proceeded to add the PPA above to the system, apt-get update, apt-get upgrade, which pulled in a bunch of stuff, still no offer of a driver, so I ended up doing "apt-get install xorg-driver-fglrx". Rebooted, ended up with an unusable system: System attempts to start X but displays graphical corruption, and finally locks hard.

2.) Installed Release version of Jaunty + all updates. Again noticed the open source ATI driver worked well, including acceleration. Again noticed the "hardware drivers" applet didn't offer any proprietary drivers. Decided to skip the PPA and just install the official driver: "apt-get install xorg-driver-fglrx". Rebooted, again ended up with an unusable system: System attempts to start X but displays graphical corruption, and finally locks hard. Had to fix it by rebooting into recovery and manually removing xorg-driver-fglrx.

Notes:
I've noticed that ATI has moved this video card to Legacy support, which means that the latest version of the Catalyst drivers apparently does not contain support for the ATI 9800XT anymore, and the last version they provide will only receive essential fixes etc. Even so, I had a lot of trouble trying to install this last/latest version posted on ATI's site for the card into Windows -- basically it wouldn't install, and I eventually ended up installing an older version of the driver.

Anyway, I mention this because it may be possible that the latest fglrx may be using the latest catalyst driver for Linux, which presumably means that the card is likely not supported anymore (I've not checked the readme's or anything but I'd be surprised if this isn't the case, given the state of affairs in Windows land...)

Also, I notice that the open source driver seems to be working rather well now, so it may be an idea to close this bug as "wont fix".

Nevertheless, thanks for the effort to try and help resolve this anyway, it is appreciated.

[Emertyl (emertyl)]

The problem may be linked to the motherboard. I realized that my chipset was an nvidia one:

~$ lspci | grep AGP
00:0b.0 PCI bridge: nVidia Corporation nForce3 250Gb AGP Host to PCI Bridge (rev a2)

So I tried a gforce graphic card and it works perfectly now!

[Bryce Harrington (bryce)]

> Anyway, I mention this because it may be possible that the latest fglrx may be using the latest catalyst driver for Linux, which presumably means that the card is likely not supported anymore (I've not checked the readme's or anything but I'd be surprised if this isn't the case, given the state of affairs in Windows land...)

> Also, I notice that the open source driver seems to be working rather well now, so it may be an idea to close this bug as "wont fix".

Right, yeah ATI dropped fglrx support for older cards, and we're focusing on the open source -ati driver for them now.