Hello ng,

I hope this is not off topic, if so please do tell me where this thread
should go.
Or where to report this 'bug'

I just tried upgrading to kernel 2.6.15 with XOrg 6.8.2 and ATI 8.20.8
driver.
Kernel upgrade went OK as did the install of the ATI driver.
But after starting up X, I got some error messages on the console.
I saved a dmesg on disk - see below where I have included the relevant
lines - and then my system just hung.
So I reverted to my old configuration which is just a 2.6.14 kernel
instead.

Best regards

Ib


fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies,
Starnberg, GERMANY' taints kernel.
[fglrx] Maximum main memory to use for locked dma buffers: 929 MBytes.
ACPI: PCI Interrupt 0000:03:00.0[A] -> Link [APC4] -> GSI 19 (level, high)
-> IRQ 16
[fglrx] module loaded - fglrx 8.20.8 [Dec 6 2005] on minor 0
[fglrx] ACPI power management is initialized.
[fglrx] module unloaded - fglrx 8.20.8 [Dec 6 2005] on minor 0
[fglrx] Maximum main memory to use for locked dma buffers: 929 MBytes.
ACPI: PCI Interrupt 0000:03:00.0[A] -> Link [APC4] -> GSI 19 (level, high)
-> IRQ 16
[fglrx] module loaded - fglrx 8.20.8 [Dec 6 2005] on minor 0
[fglrx] ACPI power management is initialized.
[fglrx] Maximum main memory to use for locked dma buffers: 929 MBytes.
[fglrx] AGP detected, AgpState = 0x1f000217 (hardware caps of chipset)
agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0.
agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:03:00.0 into 4x mode
[fglrx] AGP enabled, AgpCommand = 0x1f000314 (selected caps)
Bad page state at free_hot_cold_page (in process 'X', page c1555140)
flags:0x80000414 mapping:00000000 mapcount:0 count:0
Backtrace:
[<c0141b80>] bad_page+0x80/0xc0
[<c01423a1>] free_hot_cold_page+0x61/0x120
[<f9013e45>] drm_ioremap+0x85/0x110 [fglrx]
[<c014cea5>] zap_pte_range+0x195/0x250
[<c014d023>] unmap_page_range+0xc3/0x120
[<c014d175>] unmap_vmas+0xf5/0x1f0
[<c01517d5>] unmap_region+0x95/0x120
[<c0151b1c>] do_munmap+0x10c/0x150
[<c0151ba4>] sys_munmap+0x44/0x70
[<c0103201>] syscall_call+0x7/0xb
Trying to fix it up, but a reboot is needed
[fglrx] free AGP = 54800384
[fglrx] max AGP = 54800384
[fglrx] free LFB = 47165440
[fglrx] max LFB = 47165440
[fglrx] free Inv = 0
[fglrx] max Inv = 0
[fglrx] total Inv = 0
[fglrx] total TIM = 0
[fglrx] total FB = 0
[fglrx] total AGP = 16384
------------[ cut here ]------------
kernel BUG at <bad filename>:58347!
invalid operand: 0000 [#1]
PREEMPT
Modules linked in: fglrx nfs lockd sunrpc autofs snd_pcm_oss snd_mixer_oss
ipv6 uhci_hcd nvidia_agp i2c_nforce2 yealink ohci_hcd ehci_hcd sata_nv
libata ohci1394 ieee1394 shpchp tuner saa7134 video_buf v4l2_common
v4l1_compat ir_kbd_i2c i2c_core ir_common videodev snd_ens1371 gameport
snd_rawmidi snd_seq_device snd_ac97_codec snd_pcm snd_timer snd soundcore
snd_page_alloc snd_ac97_bus 3c59x mii sg agpgart psmouse
CPU: 0
EIP: 0060:[<c0147ff9>] Tainted: P B VLI
EFLAGS: 00213256 (2.6.15)
EIP is at put_page+0x39/0x70
eax: 00000000 ebx: e58a28dc ecx: c1555140 edx: c1555140
esi: b7a37000 edi: c1555140 ebp: e43abf2c esp: e43abe90
ds: 007b es: 007b ss: 0068
Process X (pid: 8137, threadinfo=e43aa000 task=f37580b0)
Stack: e58a28dc c014cea5 c1555140 b7a37000 2aa8a067 2aa8a067 00000000
ffffffff
f7cf2d40 b7a38000 dedffb78 b7a38000 e43abf2c c014d023 c04374a0
f7f23bcc
dedffb78 b7a37000 b7a38000 e43abf2c 00000000 b7a37fff dedffb78
dedffb78
Call Trace:
[<c014cea5>] zap_pte_range+0x195/0x250
[<c014d023>] unmap_page_range+0xc3/0x120
[<c014d175>] unmap_vmas+0xf5/0x1f0
[<c01517d5>] unmap_region+0x95/0x120
[<c0151b1c>] do_munmap+0x10c/0x150
[<c0151ba4>] sys_munmap+0x44/0x70
[<c0103201>] syscall_call+0x7/0xb
Code: 8b 02 89 d1 f6 c4 40 74 03 8b 4a 0c 8b 41 04 40 74 19 83 42 04 ff 0f
98 c0 84 c0 75 04 83 c4 04 c3 89 d0 83 c4 04 e9 47 03 00 00 <0f> 0b eb e3
8b 52 0c 8b 02 89 d1 f6 c4 40 74 03 8b 4a 0c 8b 41
<6>note: X[8137] exited with preempt_count 3
scheduling while atomic: X/0x00000003/8137
[<c0339ad2>] schedule+0x612/0x620
[<c011eae3>] vprintk+0x1a3/0x290
[<c033a7cd>] rwsem_down_read_failed+0x8d/0x170
[<c0121d72>] .text.lock.exit+0x27/0xb5
[<c0120a33>] do_exit+0xf3/0x430
[<c01047d0>] do_invalid_op+0x0/0xd0
[<c010440d>] die+0x17d/0x180
[<c010487e>] do_invalid_op+0xae/0xd0
[<c0147ff9>] put_page+0x39/0x70
[<c0334f9c>] unix_stream_recvmsg+0x10c/0x460
[<c0141ae4>] bad_range+0x34/0x50
[<c0142005>] prep_new_page+0x75/0x80
[<c0142590>] buffered_rmqueue+0x110/0x210
[<f9013817>] drm_free+0xb7/0x150 [fglrx]
[<c0103c97>] error_code+0x4f/0x54
[<c0147ff9>] put_page+0x39/0x70
[<c014cea5>] zap_pte_range+0x195/0x250
[<c014d023>] unmap_page_range+0xc3/0x120
[<c014d175>] unmap_vmas+0xf5/0x1f0
[<c01517d5>] unmap_region+0x95/0x120
[<c0151b1c>] do_munmap+0x10c/0x150
[<c0151ba4>] sys_munmap+0x44/0x70
[<c0103201>] syscall_call+0x7/0xb

Ib Højme <ib@hojme.dk> wrote:
> I hope this is not off topic, if so please do tell me where this thread
> should go.

You are not off-topic as you have built this upon a Slackware
installation. However, this forum might not have the best answers.

> fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies,
> Starnberg, GERMANY' taints kernel.

A tainted kernel together with a problem is usually hard to solve.

> [fglrx] AGP enabled, AgpCommand = 0x1f000314 (selected caps)
> Bad page state at free_hot_cold_page (in process 'X', page c1555140)
> flags:0x80000414 mapping:00000000 mapcount:0 count:0
> Backtrace:
> [<c0141b80>] bad_page+0x80/0xc0
> [<c01423a1>] free_hot_cold_page+0x61/0x120
> [<f9013e45>] drm_ioremap+0x85/0x110 [fglrx]
> [<c014cea5>] zap_pte_range+0x195/0x250
> [<c014d023>] unmap_page_range+0xc3/0x120
> [<c014d175>] unmap_vmas+0xf5/0x1f0
> [<c01517d5>] unmap_region+0x95/0x120
> [<c0151b1c>] do_munmap+0x10c/0x150
> [<c0151ba4>] sys_munmap+0x44/0x70
> [<c0103201>] syscall_call+0x7/0xb
> Trying to fix it up, but a reboot is needed
> [fglrx] free AGP = 54800384
> [fglrx] max AGP = 54800384
> [fglrx] free LFB = 47165440
> [fglrx] max LFB = 47165440
> [fglrx] free Inv = 0
> [fglrx] max Inv = 0
> [fglrx] total Inv = 0
> [fglrx] total TIM = 0
> [fglrx] total FB = 0
> [fglrx] total AGP = 16384
> ------------[ cut here ]------------
> kernel BUG at <bad filename>:58347!

So, you have a kernel bug and no kernel developer will want to look at
your problem as your kernel is tainted. Even if they wanted to solve your
problem they would not be able to as it seems as if the bug also occurs in
the binary driver for which only ATI has the source.

You now have the following options:

1) Ask ATI to help you. They have the source of the buggy driver, they are
the only ones that can fix the bug.

2) Reconfigure your X configutation and hope that the bug doesn't appear
or at least doesn't appear as often. I once played with the fglrx
driver for my Radeon 9200 and different settings in XF86Config could
at least make it start and run for a while.

3) Switch to an opensource driver. If you have a low-end card like Radeon
9250, 9200, 9100, 9000 or 8500 I would really recomend this. X.org
6.8.2 has out-of-the box support for those cards including hardware 3D
acceleration. If you have a newer/better ATI card there is experimental
support in X.org 6.9.0. However, I've heard that driver is slower than
fglrx and it is also rather buggy.

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc7(at)uthyres.com Examples of addresses which go to spammers:
root@variousus.net root@localhost

Thanks for your reply, I was afraid that the bug might be problematic :-)

The ATI driver are delevered as is as I understand it, eg. ATI will not
provide
any support. That might have changed - I have not look for this in a
while. So
now might be a good time. However I have been quite pleased with the ATI
driver
as it has been quite stable ans fast for my ATI 9100 based video card. So
I will
look into using the Xorg 6.8 driver.

Thanks again.


Best regards

Ib


On Thu, 05 Jan 2006 20:48:18 +0100, Henrik Carlqvist
<Henrik.Carlqvist@deadspam.com> wrote:

> Ib Højme <ib@hojme.dk> wrote:
>> I hope this is not off topic, if so please do tell me where this thread
>> should go.
>
> You are not off-topic as you have built this upon a Slackware
> installation. However, this forum might not have the best answers.
>
>> fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies,
>> Starnberg, GERMANY' taints kernel.
>
> A tainted kernel together with a problem is usually hard to solve.
>
>> [fglrx] AGP enabled, AgpCommand = 0x1f000314 (selected caps)
>> Bad page state at free_hot_cold_page (in process 'X', page c1555140)
>> flags:0x80000414 mapping:00000000 mapcount:0 count:0
>> Backtrace:
>> [<c0141b80>] bad_page+0x80/0xc0
>> [<c01423a1>] free_hot_cold_page+0x61/0x120
>> [<f9013e45>] drm_ioremap+0x85/0x110 [fglrx]
>> [<c014cea5>] zap_pte_range+0x195/0x250
>> [<c014d023>] unmap_page_range+0xc3/0x120
>> [<c014d175>] unmap_vmas+0xf5/0x1f0
>> [<c01517d5>] unmap_region+0x95/0x120
>> [<c0151b1c>] do_munmap+0x10c/0x150
>> [<c0151ba4>] sys_munmap+0x44/0x70
>> [<c0103201>] syscall_call+0x7/0xb
>> Trying to fix it up, but a reboot is needed
>> [fglrx] free AGP = 54800384
>> [fglrx] max AGP = 54800384
>> [fglrx] free LFB = 47165440
>> [fglrx] max LFB = 47165440
>> [fglrx] free Inv = 0
>> [fglrx] max Inv = 0
>> [fglrx] total Inv = 0
>> [fglrx] total TIM = 0
>> [fglrx] total FB = 0
>> [fglrx] total AGP = 16384
>> ------------[ cut here ]------------
>> kernel BUG at <bad filename>:58347!
>
> So, you have a kernel bug and no kernel developer will want to look at
> your problem as your kernel is tainted. Even if they wanted to solve your
> problem they would not be able to as it seems as if the bug also occurs
> in
> the binary driver for which only ATI has the source.
>
> You now have the following options:
>
> 1) Ask ATI to help you. They have the source of the buggy driver, they
> are
> the only ones that can fix the bug.
>
> 2) Reconfigure your X configutation and hope that the bug doesn't appear
> or at least doesn't appear as often. I once played with the fglrx
> driver for my Radeon 9200 and different settings in XF86Config could
> at least make it start and run for a while.
>
> 3) Switch to an opensource driver. If you have a low-end card like Radeon
> 9250, 9200, 9100, 9000 or 8500 I would really recomend this. X.org
> 6.8.2 has out-of-the box support for those cards including hardware 3D
> acceleration. If you have a newer/better ATI card there is
> experimental
> support in X.org 6.9.0. However, I've heard that driver is slower than
> fglrx and it is also rather buggy.
>
> regards Henrik

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

> I hope this is not off topic, if so please do tell me where this thread
> should go.
> Or where to report this 'bug'

There is an unofficial bugzilla at http://ati.cchtml.com - some of the
ATi developers do read it, but it's no guarantee that they will fix the
problems that appear.

Alternatively, you should also try searching the Linux drivers forum at
Rage3D, since some of the ATi developers do post there and the posters
tend to be more knowledgeable about the various bugs/ problems, etc in
getting the proprietary fglrx drivers to work.

http://www.rage3d.com/board/forumdisplay.php?f=88

> I just tried upgrading to kernel 2.6.15 with XOrg 6.8.2 and ATI 8.20.8
> driver.

There is a patch on lkml for fglrx on 2.6.15 to cure the bad page state
problem.

http://lkml.org/lkml/2005/12/11/26

- -Carlos
- --
E-Mail: cathectic AT gmail DOT com
GPG Key ID: 0x23EE722D
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.7 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDvddf0Cfk4yPuci0RArlYAJ4h9+ZC9pLUOJvHynmeuL qrVc1owQCg9a1H
fVkN1jyVCfg+vqA3trZUGkk=
=KaHN
-----END PGP SIGNATURE-----

Ib Højme <ib@hojme.dk> wrote:
> However I have been quite pleased with the ATI driver as it has been
> quite stable ans fast for my ATI 9100 based video card. So I will
> look into using the Xorg 6.8 driver.

If you have a 9100 which basically is a renamed 8500 I would really
recomend the opensource driver in X.org 6.8. I have no experience from any
such card myself, but X.org 6.8 is supposed to have support for hardware
3D acceleration for those cards. These are the settings I use in xorg.conf
for my Radeon 9200 (which is a slower card than 9100/8500):

Section "Module"
....
# This loads the GLX module
Load "glx"
# This loads the DRI module
Load "dri"
EndSection
....
Section "Device"
Identifier "Powercolor Radeon 9200"
Driver "radeon"
#VideoRam 262144
# Insert Clocks lines here if appropriate
Option "AGPMode" "4"
Option "EnablePageFlip" "on"
EndSection
....
Section "DRI"
Mode 0666
EndSection

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc7(at)uthyres.com Examples of addresses which go to spammers:
root@variousus.net root@localhost

Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
> These are the settings I use in xorg.conf
> for my Radeon 9200 (which is a slower card than 9100/8500):

Also I use 16 bit color depth with X.org. That might explain some of the
speed advantages compared to fglrx which I wasn't able to get working with
anything bat 24 bit color depths:

Section "Screen"
Identifier "Screen 1"
Device "Powercolor Radeon 9200"
Monitor "Fujitsu"
DefaultDepth 16
....

regards Henrik
--
The address in the header is only to prevent spam. My real address is:
hc7(at)uthyres.com Examples of addresses which go to spammers:
root@variousus.net root@localhost

Thanks for your reply.

I applied the patch from lkml, re-build the module and now all is well.


Best regards

Ib

On Fri, 06 Jan 2006 03:35:11 +0100, Carlos Corbacho <cathectic@see.my-sig>
wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
>> I hope this is not off topic, if so please do tell me where this thread
>> should go.
>> Or where to report this 'bug'
>
> There is an unofficial bugzilla at http://ati.cchtml.com - some of the
> ATi developers do read it, but it's no guarantee that they will fix the
> problems that appear.
>
> Alternatively, you should also try searching the Linux drivers forum at
> Rage3D, since some of the ATi developers do post there and the posters
> tend to be more knowledgeable about the various bugs/ problems, etc in
> getting the proprietary fglrx drivers to work.
>
> http://www.rage3d.com/board/forumdisplay.php?f=88
>
>> I just tried upgrading to kernel 2.6.15 with XOrg 6.8.2 and ATI 8.20.8
>> driver.
>
> There is a patch on lkml for fglrx on 2.6.15 to cure the bad page state
> problem.
>
> http://lkml.org/lkml/2005/12/11/26
>
> - -Carlos
> - --
> E-Mail: cathectic AT gmail DOT com
> GPG Key ID: 0x23EE722D
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.7 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
> iD8DBQFDvddf0Cfk4yPuci0RArlYAJ4h9+ZC9pLUOJvHynmeuL qrVc1owQCg9a1H
> fVkN1jyVCfg+vqA3trZUGkk=
> =KaHN
> -----END PGP SIGNATURE-----