fgrlx causes hanging on logout/shutdown/reboot

starthis · 2006-01-05 18:30:18

Since (I think) the update of udev where hotplug was removed, I can't logout/shutdown/reboot anymore, and many other issues arose.

Though fglrx modules loads fine at startup, and X loads normally, there seems to be a kernel bug when doing shutdown related things. (a simple gdm-stop is enough to hang the system)

The error message is like this :

kernel BUG at mm/swap.c:49
Invalid operand : 0000 [#3]
...

(I dunno how to dump the error to disk (If u know how I'd like to know), but here is a simular one : http://www.mail-archive.com/desktop-bug … 01189.html )
The above link suggests its gdm related, but it isn't.

So this seems like a distro independant problem. Maybe related to x86_64 procs? I didn't find much info googling.

Anyway removing the fglrx.ko module and falling back to the standard module eliminates the problem. (I thought that adding fgrlx to the blacklist in rc.conf would be enough, but then it still loads.)

Anybody else experiencing this problem?

Jules · 2006-01-05 23:27:58

I'm experiencing this problem as well, and haven't found a proper solution yet.
Unfortunately I upgraded the ATI driver and the udev update on the same day and can't recall which one actualy caused the problem. Right now I'm using the non-accelerated radeon driver.

Jacob · 2006-01-06 01:19:40

I too have had similar problems with my X.org freezing when I exit. I also get it when I close glxgears or fgl_gears.

I did notice that when I recompiled ATI drivers 8.20.8 against kernel 2.6.15 as I loaded the fglrx module into memory as the system started up I was plauged with a few errors (in my dmesg). This makes me believe that the problem isn't with udev, but with the ATI drivers in conjunction with 2.6.15 kernel.

I've since downgraded the kernel, udev, hal, reinstall hotplug and the old 2.6.14.5 kernel, and flagged all of them in pacman.conf

I'll be waiting until ATI releases another driver before I try again.

Jacob

Jules · 2006-01-06 08:06:40

Here is the error from my kernel.log

Jan 5 23:52:31 inspiron fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
Jan 5 23:52:31 inspiron [fglrx] Maximum main memory to use for locked dma buffers: 927 MBytes.
Jan 5 23:52:31 inspiron ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 16
Jan 5 23:52:31 inspiron [fglrx] module loaded - fglrx 8.20.8 [Dec 6 2005] on minor 0
Jan 5 23:52:32 inspiron [fglrx] Maximum main memory to use for locked dma buffers: 927 MBytes.
Jan 5 23:52:32 inspiron Bad page state at free_hot_cold_page (in process 'X', page c17b3828)
Jan 5 23:52:32 inspiron flags:0x80000414 mapping:00000000 mapcount:0 count:0
Jan 5 23:52:32 inspiron Backtrace:
Jan 5 23:52:32 inspiron [<c01492a7>] bad_page+0x87/0xc0
Jan 5 23:52:32 inspiron [<c0149c3c>] free_hot_cold_page+0x4c/0x140
Jan 5 23:52:32 inspiron [<c01557c8>] zap_pte_range+0x238/0x2f0
Jan 5 23:52:32 inspiron [<c0155976>] unmap_page_range+0xf6/0x130
Jan 5 23:52:32 inspiron [<c0155a80>] unmap_vmas+0xd0/0x250
Jan 5 23:52:32 inspiron [<c015a592>] unmap_region+0xb2/0x160
Jan 5 23:52:32 inspiron [<c015a939>] do_munmap+0x109/0x150
Jan 5 23:52:32 inspiron [<c015a9d1>] sys_munmap+0x51/0x80
Jan 5 23:52:32 inspiron [<c01031f7>] sysenter_past_esp+0x54/0x75
Jan 5 23:52:32 inspiron Trying to fix it up, but a reboot is needed
Jan 5 23:52:32 inspiron [fglrx] free PCIe = 54804480
Jan 5 23:52:32 inspiron [fglrx] max PCIe = 54804480
Jan 5 23:52:32 inspiron [fglrx] free LFB = 116322304
Jan 5 23:52:32 inspiron [fglrx] max LFB = 116322304
Jan 5 23:52:32 inspiron [fglrx] free Inv = 0
Jan 5 23:52:32 inspiron [fglrx] max Inv = 0
Jan 5 23:52:32 inspiron [fglrx] total Inv = 0
Jan 5 23:52:32 inspiron [fglrx] total TIM = 0
Jan 5 23:52:32 inspiron [fglrx] total FB = 0
Jan 5 23:52:32 inspiron [fglrx] total PCIe = 16384
Jan 5 23:54:22 inspiron ------------[ cut here ]------------
Jan 5 23:54:22 inspiron kernel BUG at mm/swap.c:49!
Jan 5 23:54:22 inspiron invalid operand: 0000 [#1]
Jan 5 23:54:22 inspiron PREEMPT SMP
Jan 5 23:54:22 inspiron Modules linked in: fglrx hci_usb bluetooth sbp2 ohci1394 eth1394 ieee1394 uhci_hcd ehci_hcd usbhid usblp usb_storage yenta_socket rsrc_nonstatic pcmcia_core ipw2200 ieee80211 ieee80211_crypt b44 mii sg snd_intel8x0m snd_intel8x0 snd_ac97_codec snd_ac97_bus snd_pcm_oss snd_pcm snd_timer snd_page_alloc snd_mixer_oss snd soundcore shpchp pci_hotplug serio_raw pcspkr evdev rtc intel_agp agpgart usbcore ntfs xfs reiserfs jfs ext3 jbd ata_piix sr_mod ahci libata sd_mod ide_floppy ide_cd cdrom ide_disk generic ide_core
Jan 5 23:54:22 inspiron CPU: 0
Jan 5 23:54:22 inspiron EIP: 0060:[<c01504f3>] Tainted: P B VLI
Jan 5 23:54:22 inspiron EFLAGS: 00010256 (2.6.15-ARCH)
Jan 5 23:54:22 inspiron EIP is at put_page+0x63/0x80
Jan 5 23:54:22 inspiron eax: 00000000 ebx: f6e4485c ecx: c17b3828 edx: c17b3828
Jan 5 23:54:22 inspiron esi: b7a17000 edi: c17b3828 ebp: 00000020 esp: f6e27e94
Jan 5 23:54:22 inspiron ds: 007b es: 007b ss: 0068
Jan 5 23:54:22 inspiron Process X (pid: 2545, threadinfo=f6e26000 task=f64f8070)
Jan 5 23:54:22 inspiron Stack: c01557c8 c17b3828 b7a17000 36c3a067 f645f900 ffffffff 00000000 c17b819c
Jan 5 23:54:22 inspiron f64feb78 b7a18000 f6e27f28 f64feb78 c0155976 c1907900 f64fbddc f64feb78
Jan 5 23:54:22 inspiron b7a17000 b7a18000 f6e27f28 00000000 f64feb78 b7a17fff 00000001 b7a17000
Jan 5 23:54:22 inspiron Call Trace:
Jan 5 23:54:22 inspiron [<c01557c8>] zap_pte_range+0x238/0x2f0
Jan 5 23:54:22 inspiron [<c0155976>] unmap_page_range+0xf6/0x130
Jan 5 23:54:22 inspiron [<c0155a80>] unmap_vmas+0xd0/0x250
Jan 5 23:54:22 inspiron [<c015a592>] unmap_region+0xb2/0x160
Jan 5 23:54:22 inspiron [<c015a939>] do_munmap+0x109/0x150
Jan 5 23:54:22 inspiron [<c015a9d1>] sys_munmap+0x51/0x80
Jan 5 23:54:22 inspiron [<c01031f7>] sysenter_past_esp+0x54/0x75
Jan 5 23:54:22 inspiron Code: 00 8b 52 0c 8b 02 89 d1 f6 c4 40 74 03 8b 4a 0c 8b 41 04 40 74 1f f0 83 42 04 ff 0f 98 c0 84 c0 74 cc 8b 4a 38 89 54 24 04 ff e1 <0f> 0b 31 00 6c 0b 3b c0 eb ad 0f 0b 29 00 6c 0b 3b c0 eb d7 89
Jan 5 23:54:22 inspiron <6>note: X[2545] exited with preempt_count 2
Jan 5 23:54:22 inspiron scheduling while atomic: X/0x00000002/2545
Jan 5 23:54:22 inspiron [<c039488d>] schedule+0x9cd/0xd60
Jan 5 23:54:22 inspiron [<c0122118>] call_console_drivers+0x158/0x180
Jan 5 23:54:22 inspiron [<c012268b>] release_console_sem+0x7b/0xc0
Jan 5 23:54:22 inspiron [<c012250e>] vprintk+0x28e/0x310
Jan 5 23:54:22 inspiron [<c0395935>] rwsem_down_read_failed+0xa5/0x1b0
Jan 5 23:54:22 inspiron [<c0126220>] .text.lock.exit+0x27/0x87
Jan 5 23:54:22 inspiron [<c0124e08>] do_exit+0xf8/0x420
Jan 5 23:54:22 inspiron [<c010457b>] die+0x18b/0x190
Jan 5 23:54:22 inspiron [<c0104910>] do_invalid_op+0x0/0xd0
Jan 5 23:54:22 inspiron [<c01049c8>] do_invalid_op+0xb8/0xd0
Jan 5 23:54:22 inspiron [<c01504f3>] put_page+0x63/0x80
Jan 5 23:54:22 inspiron [<c0149202>] bad_range+0x42/0x60
Jan 5 23:54:22 inspiron [<c0149202>] bad_range+0x42/0x60
Jan 5 23:54:22 inspiron [<f9a9bc87>] drm_free+0xb7/0x150 [fglrx]
Jan 5 23:54:22 inspiron [<c0103d77>] error_code+0x4f/0x54
Jan 5 23:54:22 inspiron [<f9a9007b>] intel_840_configure+0xb/0x140 [fglrx]
Jan 5 23:54:22 inspiron [<c01504f3>] put_page+0x63/0x80
Jan 5 23:54:22 inspiron [<c01557c8>] zap_pte_range+0x238/0x2f0
Jan 5 23:54:22 inspiron [<c0155976>] unmap_page_range+0xf6/0x130
Jan 5 23:54:22 inspiron [<c0155a80>] unmap_vmas+0xd0/0x250
Jan 5 23:54:22 inspiron [<c015a592>] unmap_region+0xb2/0x160
Jan 5 23:54:22 inspiron [<c015a939>] do_munmap+0x109/0x150
Jan 5 23:54:22 inspiron [<c015a9d1>] sys_munmap+0x51/0x80
Jan 5 23:54:22 inspiron [<c01031f7>] sysenter_past_esp+0x54/0x75

starthis · 2006-01-06 17:26:19

Unfortunately I upgraded the ATI driver and the udev update on the same day and can't recall which one actualy caused the problem.

I had my ati driver updated some days earlier, so I'm pretty sure the problem only started when I upgraded to 2.6.15 kernel.

It indeed looks like its an ATI driver incompatibility with kernl 2.6.15.
More and more posts on the subject appear on the net.

Just now I was starting to believe the ATI drivers reached a stable point :shock:

Lets wait for a new ati driver or kernel...

Jefg60 · 2006-01-08 23:45:37

I'm having this issue too. Just did an ftp install to a clean system, fglrxconfig works, X starts and runs fine but shutting down X causes the pc to hang. This happens if I choose logout, or switch to another virtual terminal and shutdown from there.

vacant · 2006-01-09 11:08:30

I switched from fglrx to radeon when this started happening (9200SE here). Being lazy, I'll watch this thread and hope someone posts when ATI fix it

Jefg60 · 2006-01-09 19:33:31

vacant wrote:

I switched from fglrx to radeon when this started happening (9200SE here). Being lazy, I'll watch this thread and hope someone posts when ATI fix it

thats exactly what I'm doing! in my case its more that I have no idea how to fix it

theD3viL · 2006-01-19 14:22:11

Something new? :? I wanna install arch (now have gentoo - i removed arch couse freezing at logout) ... so .. now drivers working? :?

Jules · 2006-01-19 14:40:17

Hi,

I've tried the 8.21.7 driver, still the same issue.

Cheers,
Jules

theD3viL · 2006-01-19 14:52:33

What about the driver from official site? It must work i suppose? But if dont, how to remove it? :oops:

Jules · 2006-01-19 15:54:04

The 8.21.7 is the original ati driver. If it doesn't work use the non-accelerated radeon driver instead, or let hwd write xorg.conf (hwd -x).

theD3viL · 2006-01-19 16:01:51

irc logs:

about the ati freeze on logout, stay with kernel26arckck-2.16.14 and ati-drivers from community

Try it..

mmccaskill · 2006-01-20 03:11:44

Here you go kiddies.

Make sure you abs first.

Put in /var/abs/community/modules/ati-drivers

fglrx-2.6.15.diff

--- firegl_public.c.old    2005-12-09 19:21:33.000000000 +0100
+++ firegl_public.c    2005-12-18 14:44:15.000000000 +0100
@@ -2621,7 +2621,7 @@
 
     pMmPage = virt_to_page(kaddr);
 
-#if 0
+#if LINUX_VERSION_CODE >= 0x02060f
     // WARNING WARNINIG WARNNING WARNNING WARNNING WARNNING WARNNING WARNNING
     // Don't increment page usage count, cause ctx pages are allocated
     // with drm_alloc_pages, which marks all pages as reserved. Reserved

Update the PKGBUILD

PKGBUILD

# Contributor: Link Dupont <link@subpop.net>,  qwerty <qwerty@riot.eu.org>

pkgname=ati-drivers
pkgver=8.20.8
pkgrel=5
pkgdesc="ATI precompiled drivers for r350, r300, r250 and r200 chipsets.  Built for the kernel26archck pkg in [community]"
url="http://www.ati.com"
license=""
depends=('xorg>=6.8.0' 'kernel26archck')
makedepends=('rpmunpack' 'cpio')
conflicts=('ati-drivers-module')
install=ati-drivers.install
source=(fglrx_6_8_0-$pkgver-1.i386.rpm fireglcontrol.desktop)
md5sums=('bec227dc16abed4b84754137808b225c' '5c5aa064e53a64447296ec5b062d22e2')

_kernel=2.6.15-archck
    
build() {
    cd $startdir/src
    rpmunpack < $startdir/src/fglrx_6_8_0-$pkgver-1.i386.rpm | gzip -d | cpio -idmuv
    rm -rf $startdir/src/usr/X11R6/bin/*.bz2
    cd $startdir/src/lib/modules/fglrx/build_mod
    
    # Patch for 2.6.15 to eliminate freeze on exit
    patch -p0 < $stardir/fglrx-2.6.15.diff

    # Build the kernel module
    cp 2.6.x/Makefile .
    make -C /usr/src/linux-${_kernel} SUBDIRS="`pwd`" modules || return 1

    # Install the kernel module
    install -m 644 -D $startdir/src/lib/modules/fglrx/build_mod/fglrx.ko 
        $startdir/pkg/lib/modules/${_kernel}/video/fglrx.ko

    # Remove src and documents - install remaining files
    rm -rf $startdir/src/usr/{src,share/{docs,applnk,gnome}}
    cp -R $startdir/src/usr $startdir/pkg

    # correct dir permissions
    find $startdir/pkg/usr -type d -exec chmod 755 {} ;

    # create DE file
    install -m 755 -D $startdir/src/fireglcontrol.desktop 
      $startdir/pkg/usr/share/applications/fireglcontrol.desktop

    # cheap, dirty fix for file conflict, see ati-drivers.install (co/ mmgm)
    mv $startdir/pkg/usr/X11R6/lib/libGL.so.1.2 $startdir/pkg/usr/X11R6/lib/ati.libGL.so.1.2
}

dimi · 2006-01-20 20:05:21

Didn't work for me... It's still crashing.

By the way, there's a typo in the "patch" line, it should read $startdir, I think:

# Patch for 2.6.15 to eliminate freeze on exit 
patch -p0 < $startdir/fglrx-2.6.15.diff

EDIT: Whoops, tried it again, it works great. Thanks!

Cerebral · 2006-01-21 21:40:43

dimi wrote:

Didn't work for me... It's still crashing.
By the way, there's a typo in the "patch" line, it should read $startdir, I think

You're right about the typo, however applying this patch worked beautifully for me. I just today upgraded to 2.6.15-archck and was getting the freeze error; now it's all gone thanks to the patch. yay!

giorgosc61 · 2006-01-22 09:38:45

Could an updated version of ati-drivers in community with the patch be made? I have the freezing problems with 2.6.15-archck.

oliv · 2006-01-23 11:09:33

I just see that ati released a 8.21.7 version of the drivers. I try to use it but I have somehow another issue :

Every time I try to load that fglrx module I get:
FATAL: Module fglrx not found.
It seems that there isn't any depmod -A in the driver post-upgrade script. Can someone correct it ?

EDIT: You still have to use the patch with this version. Now I can logout, direct rendering works, and glxgears halt correctly.
EDIT: Yet when I use glxgears, my system load increase to 100%. Is this normal, with a direct rendering ?

starthis · 2006-01-23 17:24:17

The patched 8.21.7 driver also works fine with the regular 2.6.15-ARCH kernel.

EDIT: Yet when I use glxgears, my system load increase to 100%. Is this normal, with a direct rendering ?

I think its normal. Its has the same behaviour here.
Probably it's programmed to do a loop and update as fast as possible :

while (true)
{
DoGLRendering();
Sleep(0);
}

ConnorBehan · 2007-10-05 05:23:46

I am having this same issue with the FOSS r128 drivers with X.org. There seems to be no 100% graceful way to get out of X... poweroff, shutdown now, kill X, xfce's shutdown helper, xfce's logout button and Alt+SysRq+R,Alt+SysRq+E all have a 25% chance of freezing X... I'm used to it now and once I got my friend to shut it down by sshing into my machine but thats about it.

When I do glxinfo I get:

libGL warning: 3D driver claims to not support visual 0x4b... I could try the GATOS ati.2 drivers for my card, does anyone think this will work? If I would need to patch those... where would the file in that be?

Arch Linux

#1 2006-01-05 18:30:18

fgrlx causes hanging on logout/shutdown/reboot

#2 2006-01-05 23:27:58

Re: fgrlx causes hanging on logout/shutdown/reboot

#3 2006-01-06 01:19:40

Re: fgrlx causes hanging on logout/shutdown/reboot

#4 2006-01-06 08:06:40

Re: fgrlx causes hanging on logout/shutdown/reboot

#5 2006-01-06 17:26:19

Re: fgrlx causes hanging on logout/shutdown/reboot

#6 2006-01-08 23:45:37

Re: fgrlx causes hanging on logout/shutdown/reboot

#7 2006-01-09 11:08:30

Re: fgrlx causes hanging on logout/shutdown/reboot

#8 2006-01-09 19:33:31

Re: fgrlx causes hanging on logout/shutdown/reboot

#9 2006-01-19 14:22:11

Re: fgrlx causes hanging on logout/shutdown/reboot

#10 2006-01-19 14:40:17

Re: fgrlx causes hanging on logout/shutdown/reboot

#11 2006-01-19 14:52:33

Re: fgrlx causes hanging on logout/shutdown/reboot

#12 2006-01-19 15:54:04

Re: fgrlx causes hanging on logout/shutdown/reboot

#13 2006-01-19 16:01:51

Re: fgrlx causes hanging on logout/shutdown/reboot

#14 2006-01-20 03:11:44

Re: fgrlx causes hanging on logout/shutdown/reboot

#15 2006-01-20 20:05:21

Re: fgrlx causes hanging on logout/shutdown/reboot

#16 2006-01-21 21:40:43

Re: fgrlx causes hanging on logout/shutdown/reboot

#17 2006-01-22 09:38:45

Re: fgrlx causes hanging on logout/shutdown/reboot

#18 2006-01-23 11:09:33

Re: fgrlx causes hanging on logout/shutdown/reboot

#19 2006-01-23 17:24:17

Re: fgrlx causes hanging on logout/shutdown/reboot

#20 2007-10-05 05:23:46

Re: fgrlx causes hanging on logout/shutdown/reboot

Board footer