Page 1 of 3 123 LastLast
Results 1 to 10 of 26

Thread: FGLRX failing at DKMS

  1. #1
    Join Date
    Nov 2009
    Beans
    Hidden!
    Distro
    Xubuntu 10.04 Lucid Lynx

    FGLRX failing at DKMS

    Hi, this morning I wiped the computer a clean sheet and have been having some issues with getting FGLRX back up and running. I run Xubuntu 10.04.1 (on 2.6.32-25 kernel but, the same problem happens with lesser versions) and everything else seems to work fine. I've managed to compile the latest version of Bluefish hiccup free, etc, etc.

    More than anything, the fan noise of my graphics card when running like this is a nightmare! I'm at the end of my Linux knowledge leash when things like DKMS spit errors at me, please help.

    To start, here is the lspci output in relation to my card:
    VGA compatible controller: ATI Technologies Inc Radeon HD 3870
    Subsystem: ATI Technologies Inc Device 2542
    Flags: bus master, fast devsel, latency 0, IRQ 25
    Memory at b0000000 (64-bit, prefetchable) [size=256M]
    Memory at cfee0000 (64-bit, non-prefetchable) [size=64K]
    I/O ports at 8c00 [size=256]
    [virtual] Expansion ROM at cfe00000 [disabled] [size=128K]
    Capabilities: <access denied>
    Kernel driver in use: radeon
    Kernel modules: radeon
    The problem whilst trying to install the ATI driver throws up this message from the install script:
    Building for architecture x86_64
    Building initial module for 2.6.32-25-generic

    Error! Bad return status for module build on kernel: 2.6.32-25-generic (x86_64)
    Consult the make.log in the build directory
    /var/lib/dkms/fglrx/8.771/build/ for more information.
    dpkg: error processing fglrx (--configure):
    Following the trail to the DKMS make log
    DKMS make.log for fglrx-8.771 for kernel 2.6.32-25-generic (x86_64)
    Sun Sep 19 18:07:02 BST 2010
    AMD kernel module generator version 2.1
    doing Makefile based build for kernel 2.6.x and higher
    rm -rf *.c *.h *.o *.ko *.GCC* .??* *.symvers
    make -C /lib/modules/2.6.32-25-generic/build SUBDIRS=/var/lib/dkms/fglrx/8.771/build/2.6.x modules
    make[1]: Entering directory `/usr/src/linux-headers-2.6.32-25-generic'
    CC [M] /var/lib/dkms/fglrx/8.771/build/2.6.x/firegl_public.o
    CC [M] /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_acpi.o
    CC [M] /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_agp.o
    CC [M] /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_debug.o
    CC [M] /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_ioctl.o
    /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_ioctl.c: In function ‘KCL_IOCTL_AllocUserSpace32’:
    /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_ioctl.c:196: error: implicit declaration of function ‘compat_alloc_user_space’
    /var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_ioctl.c:196: warning: return makes pointer from integer without a cast
    make[2]: *** [/var/lib/dkms/fglrx/8.771/build/2.6.x/kcl_ioctl.o] Error 1
    make[1]: *** [_module_/var/lib/dkms/fglrx/8.771/build/2.6.x] Error 2
    make[1]: Leaving directory `/usr/src/linux-headers-2.6.32-25-generic'
    make: *** [kmod_build] Error 2
    build failed with return value 2
    I've tried the open source, x-swat and proprietary driver, installing by Synaptic/Apt-get, the shell script, making the .deb files, purging and then retrying...all with the same result

    I have been scooting around the forums and web for a good few hours now to no avail.
    . . . with sweetness through mine ear, dissolve me into ecstasies
    . . . from that height the cessation of night, A supernova of sensations

  2. #2
    Join Date
    Apr 2008
    Beans
    8

    Re: FGLRX failing at DKMS

    Exact same problem here. I traced the problem down to the build log and it appears the module isn't compiling successfully.

    This mailing list post seems to give a reason why, but doesn't provide fix.

    http://www.mail-archive.com/ubuntu-b...sg2516506.html

    Also, the ppa-purge package doesn't work (same problem), so I can't revert back without trying to uninstall the packages myself somehow.

    FYI, I'm running Ubuntu on 2.6.32-24-generic with a Radeon Mobility HD 5470.
    Last edited by Cuchaz; September 19th, 2010 at 06:48 PM.

  3. #3
    Join Date
    Nov 2009
    Beans
    Hidden!
    Distro
    Xubuntu 10.04 Lucid Lynx

    Re: FGLRX failing at DKMS

    It's kind of a relief to know other people are in the same boat about this. Still haven't been able to solve the problem here.

    It got me thinking about what will happen if the display drivers of xorg get demodularised quite a lot *shivers and prays* These sorts of problems would straight up make certain distro's unusable or seriously nerf the ability to update
    . . . with sweetness through mine ear, dissolve me into ecstasies
    . . . from that height the cessation of night, A supernova of sensations

  4. #4
    Join Date
    Apr 2008
    Beans
    8

    Re: FGLRX failing at DKMS

    Yeah, it's annoying that an update completely broke my graphics acceleration. I was able to get my compy working again by reverting to the ati open source driver. Sadly, for my chipset, that means no Compiz. I'm having a hard time getting Cairo-Dock to work correctly w/ Metacity. I hope they can fix fglrx soon.

  5. #5
    Join Date
    Sep 2010
    Beans
    1

    Re: FGLRX failing at DKMS

    I finally managed to downgrade from 10.9 to the original 10.4 from ubuntu.


    1. First completely "PURGE" the linux-image...-24 and the header files.
      Make sure, you have one of the previous kernels installed
    2. Reboot (didn't work for me without it)
    3. Uninstall all fglrx related packages (fglrx throws an error, ignore this))
    4. Reboot (again, it didn't work for me without it)
    5. XOrg should know complain about missing driver and allow you to start with failsettings for this session
    6. Install the all fglrx packages again (reinstall fglrx package)
    7. Reboot and see if it works

    I hope I didn't forget a step.

    Maybe it's relative save to install the linux-image and linux-heades generic packages again (which will reinstall the latest kernel update again, if you don't try to alter the fglrx packages again).

  6. #6
    Join Date
    Nov 2005
    Location
    Tampere, Finland
    Beans
    29
    Distro
    Ubuntu Development Release

    Re: FGLRX failing at DKMS

    I got fglrx-8.771 installed by downgrading kernel 2.6.32-25.44_amd64 to previous version 2.6.32-25.43_amd64 by downgrading packages:
    linux-headers-2.6.32-25
    linux-image-2.6.32-25-generic

    I have now :

    Code:
    linux-headers-2.6.32-25                           2.6.32-25.43
    linux-headers-2.6.32-25-generic                   2.6.32-25.44
    linux-image-2.6.32-25-generic                     2.6.32-25.43
    linux-libc-dev                                    2.6.32-25.44
    After rebooting to older kernel fglrx-8.771 installed fine.

  7. #7
    Join Date
    Sep 2008
    Beans
    45

    Re: FGLRX failing at DKMS

    I have this problem. its really annoying.

    how does one perform this purge : "linux-image...-24 and the header files" ?

    and I dont have a laptop.. this is a problem with fglrx.

    # cat /var/lib/dkms/fglrx/8.723.1/build/make.log
    DKMS make.log for fglrx-8.723.1 for kernel 2.6.32-25-generic (x86_64)
    Mon Sep 20 21:53:07 EDT 2010
    AMD kernel module generator version 2.1
    doing Makefile based build for kernel 2.6.x and higher
    rm -rf *.c *.h *.o *.ko *.GCC* .??* *.symvers
    make -C /lib/modules/2.6.32-25-generic/build SUBDIRS=/var/lib/dkms/fglrx/8.723.1/build/2.6.x modules
    make[1]: Entering directory `/usr/src/linux-headers-2.6.32-25-generic'
    CC [M] /var/lib/dkms/fglrx/8.723.1/build/2.6.x/firegl_public.o
    In file included from /var/lib/dkms/fglrx/8.723.1/build/2.6.x/firegl_public.c:443:
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/drm_proc.h: In function ‘FGLDRM__vma_info’:
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/drm_proc.h:497: warning: format ‘%08lx’ expects type ‘long unsigned int’, but argument 5 has type ‘phys_addr_t’
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/firegl_public.c: In function ‘KCL_SetPageCache_Array’:
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/firegl_public.c:1316: warning: passing argument 1 of ‘KCL_ConvertPageToKernelAddress’ makes pointer from integer without a cast
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/firegl_public.h:325: note: expected ‘void *’ but argument is of type ‘long unsigned int’
    CC [M] /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_acpi.o
    CC [M] /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_agp.o
    CC [M] /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_debug.o
    CC [M] /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_ioctl.o
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_ioctl.c: In function ‘KCL_IOCTL_AllocUserSpace32’:
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_ioctl.c:196: error: implicit declaration of function ‘compat_alloc_user_space’
    /var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_ioctl.c:196: warning: return makes pointer from integer without a cast
    make[2]: *** [/var/lib/dkms/fglrx/8.723.1/build/2.6.x/kcl_ioctl.o] Error 1
    make[1]: *** [_module_/var/lib/dkms/fglrx/8.723.1/build/2.6.x] Error 2
    make[1]: Leaving directory `/usr/src/linux-headers-2.6.32-25-generic'
    make: *** [kmod_build] Error 2
    build failed with return value 2


    Last edited by kelargo; September 21st, 2010 at 03:04 AM.

  8. #8
    Join Date
    Sep 2010
    Beans
    29

    Re: FGLRX failing at DKMS

    A fix has been released for Lucid; the version number is 2:8.723.1-0ubuntu5. If you use either of the Ubuntu-X PPA's, their packages have not been fixed and neither has the official installer from ATI, though it will be fixed for the next release.

    The update might not show up straight away depending on which update server you use. I use the UK server and it took a few hours to show up in lucid-proposed. If you can't wait, try changing the download server in Software sources to the main server.

  9. #9
    Join Date
    Aug 2006
    Location
    Novi Sad, Serbia
    Beans
    54
    Distro
    Ubuntu

    Re: FGLRX failing at DKMS

    Got the same problem, and spent an hour trying to fix things. The only solution I found was to downgrade kernel to 2.6.32-21 (by removing all other linux-image packages) and manually install Ati Catalyst. But only Catalyst 10.7 worked, 10.8 and 10.9 could not be installed (the same/similar DKMS error).

    Geez, how I wish I never bought Ati.
    Last edited by kotnik; September 22nd, 2010 at 09:34 PM.
    -- It's all GNU to me --

  10. #10
    Join Date
    Mar 2006
    Beans
    13

    Re: FGLRX failing at DKMS

    I've been having these same problems with the 2.26.32-xx-Generic kernels on Kubuntu 10.04 64-bit and I was just now able to get the fglrx module to build without having to downgrade kernels.

    Here's what I did. Chances are some kernel developer will start yelling at me for doing this, but oh well.

    As root, open up the file /usr/src/linux-headers-2.6.32-24-generic/arch/x86/include/asm/compat.h in a text editor, inserting your version of the kernel in place of mine. That directory should be correct for both 32-bit and 64-bit systems. Go to the very botton of the file; you'll see

    Code:
    #endif /* _ASM_X86_COMPAT_H */
    .

    Between that line and the curly brace ( } ) two lines above it, paste in the following:

    Code:
    static inline void __user *compat_alloc_user_space(unsigned long len)
    {
    	struct pt_regs *regs = task_pt_regs(current);
    	return (void __user *)regs->sp - len;
    }
    .

    (It isn't really necessary that the function go exactly where I said to put it, but this will make sure that you don't accidentally paste it into another function or something like that.)

    Now save the file and try reinstalling the fglrx drivers. My xorg.conf file was messed up and causing problems for me, so check that before you restart X. With any luck, you should now be rolling in fglrx goodness!

    I don't think that will hurt anything, but you can remove your changes to that file after you build the module if you want.



    You may have noticed that the function I had you paste in looks very similar to another function in there called 'arch_compat_alloc_user_space'. As far as I can tell, they are exactly the same except for the name and parameter type (unsigned long). It appears that 'compat_alloc_user_space' was renamed to 'arch_compat_alloc_user_space' for just this version of the kernel, so really we just put the old one back for the ATI module builder to use. This shouldn't hurt anything since the contents of the two functions are identical.

Page 1 of 3 123 LastLast

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •