[ltp] ! (Re: New ATI Driver) - Mixed luck (words of caution)

[D. Sen]

Having used the new ATI driver for the last three days, I would urge 
caution on its use. Its failings has forced me to go back to using the 
open source radeon driver.

The fglrx driver did seem to work in that it provided DRI as well as 
ACPI suspend/resume to disk. However, every second suspend/resume cycle 
caused the following types of error to be posted on the syslog (posted 
below).

The problem happens everytime and while it doesnt lock the machine up, 
it does cause it behave abnormally requiring an eventual restart of the 
machine. The last time this happened, it caused data loss (I did have 
backups - so it wasnt too catastrophic) and I have since gone back to 
using the radeon driver which is amazingly stable. My machine stays up 
indefinitely through multiple suspend/resume cycles every day.

So, for now I am going to live without DRI support. Stability counts for 
a lot more.

DS

Nov 20 22:52:03 localhost kernel: ------------[ cut here ]------------
Nov 20 22:52:03 localhost kernel: kernel BUG at 
drivers/block/ll_rw_blk.c:3034!
Nov 20 22:52:03 localhost kernel: invalid operand: 0000 [#1]
Nov 20 22:52:03 localhost kernel: PREEMPT
Nov 20 22:52:03 localhost kernel: Modules linked in: sg vmnet vmmon 
wlan_wep ath_pci ath_rate_sample wlan ath_hal fglrx ipv6 sd_mod rfcomm 
l2cap bluetooth snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device 
snd_pcm_oss snd_mixer_oss snd_intel8x0 snd_ac97_codec snd_ac97_bus 
snd_pcm snd_timer snd_page_alloc snd soundcore parport_pc lp parport 
ipt_state ipt_multiport ipt_conntrack iptable_mangle ip_nat_irc 
ip_nat_tftp ip_nat_ftp iptable_nat ip_nat ip_conntrack_irc 
ip_conntrack_tftp ip_conntrack_ftp ip_conntrack nfnetlink iptable_filter 
ip_tables af_packet pcmcia firmware_class crc32 yenta_socket 
rsrc_nonstatic pcmcia_core e1000 ntfs nls_iso8859_1 intel_agp agpgart 
nvram evdev usbmouse usbhid ehci_hcd uhci_hcd usbcore video thermal 
processor ibm_acpi fan button battery ac joydev
Nov 20 22:52:03 localhost kernel: CPU:    0
Nov 20 22:52:03 localhost kernel: EIP:    0060:[<c029f223>]    Tainted: 
P   M  VLI
Nov 20 22:52:03 localhost kernel: EFLAGS: 00010246   (2.6.14.2)
Nov 20 22:52:03 localhost kernel: EIP is at submit_bio+0x103/0x110
Nov 20 22:52:03 localhost kernel: eax: 00000003   ebx: 00000200   ecx: 
00000202   edx: 00000001
Nov 20 22:52:03 localhost kernel: esi: d7762bc0   edi: 00000001   ebp: 
d7762bc0   esp: d7b67dc0
Nov 20 22:52:03 localhost kernel: ds: 007b   es: 007b   ss: 0068
Nov 20 22:52:03 localhost kernel: Process klaptop_acpi_he (pid: 13378, 
threadinfo=d7b66000 task=d7667a70)
Nov 20 22:52:03 localhost kernel: Stack: 0000001c 00000001 00000000 
00000000 c21158c0 c0174990 c2115900 00000010
Nov 20 22:52:03 localhost kernel:        0000000c 00000200 00000200 
00000001 c0174f74 f7c4d830 d7762bc0 00000200
Nov 20 22:52:03 localhost kernel:        00000000 00000001 c022537f 
00000001 d7762bc0 00000200 00000000 00000000
Nov 20 22:52:03 localhost kernel: Call Trace:
Nov 20 22:52:03 localhost kernel:  [<c0174990>] bio_alloc_bioset+0x130/0x1e0
Nov 20 22:52:03 localhost kernel:  [<c0174f74>] bio_add_page+0x34/0x40
Nov 20 22:52:03 localhost kernel:  [<c022537f>] _pagebuf_ioapply+0x19f/0x2d0
Nov 20 22:52:03 localhost kernel:  [<c01169e0>] 
default_wake_function+0x0/0x20
Nov 20 22:52:03 localhost kernel:  [<c01169e0>] 
default_wake_function+0x0/0x20
Nov 20 22:52:03 localhost kernel:  [<c02254f8>] pagebuf_iorequest+0x48/0x1b0
Nov 20 22:52:03 localhost kernel:  [<c020789a>] 
xlog_state_sync_all+0x15a/0x280
Nov 20 22:52:03 localhost kernel:  [<c01169e0>] 
default_wake_function+0x0/0x20
Nov 20 22:52:03 localhost last message repeated 2 times
Nov 20 22:52:03 localhost kernel:  [<c022bd38>] xfs_bdstrat_cb+0x38/0x50
Nov 20 22:52:03 localhost kernel:  [<c0221935>] xfs_bwrite+0x65/0xf9
Nov 20 22:52:03 localhost kernel:  [<c02042e7>] xfs_log_force+0x47/0x80
Nov 20 22:52:03 localhost kernel:  [<c0218bbc>] xfs_syncsub+0x16c/0x380
Nov 20 22:52:03 localhost kernel:  [<c0358e19>] schedule+0x3b9/0x6c0
Nov 20 22:52:03 localhost kernel:  [<c02180ed>] xfs_sync+0x2d/0x40
Nov 20 22:52:03 localhost kernel:  [<c022cbbe>] linvfs_sync_super+0x4e/0x120
Nov 20 22:52:03 localhost kernel:  [<c035a2ca>] __down+0xea/0x130
Nov 20 22:52:03 localhost kernel:  [<c01169e0>] 
default_wake_function+0x0/0x20
Nov 20 22:52:03 localhost kernel:  [<c017691e>] sync_filesystems+0x12e/0x170
Nov 20 22:52:03 localhost kernel:  [<c01708bc>] do_sync+0x6c/0xf0
Nov 20 22:52:03 localhost kernel:  [<c017094f>] sys_sync+0xf/0x20
Nov 20 22:52:03 localhost kernel:  [<c010327b>] sysenter_past_esp+0x54/0x75
Nov 20 22:52:03 localhost kernel: Code: 24 0c 8b 81 9c 00 00 00 89 5c 24 
04 c7 04 24 2c 29 38 c0 89 44 24 08 e8 9c ba e7 ff e9 55 ff ff ff f6 46 
0d 01 0f 85 58 ff ff ff <0f> 0b da 0b f0 c6 37 c0 e9 4b ff ff ff 55 57 
31 ff 56 53 83 ec
Nov 20 22:52:03 localhost kernel:  <7>Back to C!

Florian Reitmeir wrote:
> On Son, 20 Nov 2005, D. Sen wrote:
> 
> 
>>I have had success using external agp and changing my ACPI suspend 
>>script to just have that one line.
>>Thanks very much for your help.
> 
> 
> nice to hear :-)
> 
> 
>>Florian Reitmeir wrote:
>>
>>>>Here is my suspend script for ACPI. Can anyone poke a hole through it?
>>>
>>>i can suspend/wake up simply by only use
>>>echo -n mem /sys/power/state
>>>i don't remove any modules or stop any daemons..
> 
>