Re: [2.6.17-rc6-mm1] Oops during sata_promise init

Re: [2.6.17-rc6-mm1] Oops during sata_promise init

[Philipp Baumann]

Hi,

testing the latest mm kernel, i still get an Oops as follows:


BUG: unable to handle kernel NULL pointer dereference at virtual address 
00000008
printing eip:
c0256773
*pde  =  00000000
Oops: 0002 xxxx
4K_STACKS PREEMPT
last sysfs file:
Modules linked in:
CPU:    0
EIP:    0060:[<c0256773>]    Not tainted VLI
EFLAGS:    00010202 (2.6.17-rc6-mm1 #1)
EIP is at pdc_sata_scr_write+0x10/0x14
eax: 00000008  ebx: dfeb429c  ecx: 00000300  edx: 0000002
esi: dfeb429c  edi: 00000300  ebp: 00000002  esp: dffc3e44
ds: 007b  es: 007b  ss: 0068
Process idle (pid: 1, threadinfo=dffc3000 task=dffc2ab0)
Stack:  c02cb980 c024c84b dfeb429c dfeb0c78 fffb82ca dfeb429c c024e4b5 
c0244541
    c02e60a6 c3000000 c01ea753 c0256281 dfeb429c dfeb0c78 00000002 c024e5bc
    dfeb0834 c0250cee dfeac408 e1493800 00000053 c02e6ef3 e0802300 c0802338
Call Trace:
[<c024c84b>] sata_scr_write_flush+0x24/0x37
[<c024e4b5>] __sata_phy_reset+0x2e/0x12d
[<c0244541>] scsi_add_host+0xc0/0x192
[<c01ea753>] __delay+0x6/0x7
[<c0256281>] pdc_reset_port+0x1f/0x34
[<c024e5bc>] sata_phy_reset+0x8/0x18
[<c0250cee>] ata_device_add+0x580/0x6b1
[<c0256c41>] pdc_ata_init_one+0x2a9/0x3c8
[<c01f5732>] pci_device_probe+0x40/0x5b
[<c023f70b>] driver_probe_device+0x3b/0xaa
[<c023f830>] __driver_attach+0x5a/0x5c
[<c023f1c4>] bus_for_each_dev+0x3a/0x58
[<c023f67a>] driver_attach+0x16/0x1a
[<c023f7d6>] --driver_attach+0x0/0x5c
[<c023ee9c>] bus_add_driver+0x6f/0x11f
[<c01f5889>] __pci_register_driver+0x34/0x4f
[<c010028c>] init+0x6c/0x225
[<c02bd34e>] __kprobes_text_start+0x6/0x14
[<c0100220>] init+0x0/0x225
[<c0100220>] init+0x0/0x225
[<c0101005>] kernel_thread_helper+0x5/0xb
Code: 02 77 0a 8d 04 12 01 c0 03 41 60 8b 00 c3 8b 80 78 1f 00 00 8b 40 
08 8b 40 40 c3 53 89 c3 83 fa 02 77 0a 8d
04 12 01 c0 03 43 60 <89> 08 5b c3 55 57 56 53 83 ec 10 89 c6 9c 5d fa 
b8 00 f0 ff ff
EIP: [<c0256773>] pdc_sata_scr_write+0x10/0x14 SS:ESP 0068:dffc3e44
<0>Kernel panic - not syncing: Attempted to kill init!

any advice?

kind regards

prx

Re: [2.6.17-rc6-mm1] Oops during sata_promise init

[Andrew Morton]

Philipp Baumann <pumuckl@i12.com> wrote:
>
> Hi,
> 
> testing the latest mm kernel, i still get an Oops as follows:
> 
> 
> BUG: unable to handle kernel NULL pointer dereference at virtual address 
> 00000008
> printing eip:
> c0256773
> *pde  =  00000000
> Oops: 0002 xxxx
> 4K_STACKS PREEMPT
> last sysfs file:
> Modules linked in:
> CPU:    0
> EIP:    0060:[<c0256773>]    Not tainted VLI
> EFLAGS:    00010202 (2.6.17-rc6-mm1 #1)
> EIP is at pdc_sata_scr_write+0x10/0x14
> eax: 00000008  ebx: dfeb429c  ecx: 00000300  edx: 0000002
> esi: dfeb429c  edi: 00000300  ebp: 00000002  esp: dffc3e44
> ds: 007b  es: 007b  ss: 0068
> Process idle (pid: 1, threadinfo=dffc3000 task=dffc2ab0)
> Stack:  c02cb980 c024c84b dfeb429c dfeb0c78 fffb82ca dfeb429c c024e4b5 
> c0244541
>     c02e60a6 c3000000 c01ea753 c0256281 dfeb429c dfeb0c78 00000002 c024e5bc
>     dfeb0834 c0250cee dfeac408 e1493800 00000053 c02e6ef3 e0802300 c0802338
> Call Trace:
> [<c024c84b>] sata_scr_write_flush+0x24/0x37
> [<c024e4b5>] __sata_phy_reset+0x2e/0x12d
> [<c0244541>] scsi_add_host+0xc0/0x192
> [<c01ea753>] __delay+0x6/0x7
> [<c0256281>] pdc_reset_port+0x1f/0x34
> [<c024e5bc>] sata_phy_reset+0x8/0x18
> [<c0250cee>] ata_device_add+0x580/0x6b1
> [<c0256c41>] pdc_ata_init_one+0x2a9/0x3c8
> [<c01f5732>] pci_device_probe+0x40/0x5b
> [<c023f70b>] driver_probe_device+0x3b/0xaa
> [<c023f830>] __driver_attach+0x5a/0x5c
> [<c023f1c4>] bus_for_each_dev+0x3a/0x58
> [<c023f67a>] driver_attach+0x16/0x1a
> [<c023f7d6>] --driver_attach+0x0/0x5c
> [<c023ee9c>] bus_add_driver+0x6f/0x11f
> [<c01f5889>] __pci_register_driver+0x34/0x4f
> [<c010028c>] init+0x6c/0x225
> [<c02bd34e>] __kprobes_text_start+0x6/0x14
> [<c0100220>] init+0x0/0x225
> [<c0100220>] init+0x0/0x225
> [<c0101005>] kernel_thread_helper+0x5/0xb
> Code: 02 77 0a 8d 04 12 01 c0 03 41 60 8b 00 c3 8b 80 78 1f 00 00 8b 40 
> 08 8b 40 40 c3 53 89 c3 83 fa 02 77 0a 8d
> 04 12 01 c0 03 43 60 <89> 08 5b c3 55 57 56 53 83 ec 10 89 c6 9c 5d fa 
> b8 00 f0 ff ff
> EIP: [<c0256773>] pdc_sata_scr_write+0x10/0x14 SS:ESP 0068:dffc3e44

Thanks.  ap->ioaddr.scr_addr is zero in pdc_sata_scr_write().

Re: [2.6.17-rc6-mm1] Oops during sata_promise init

[Jeff Garzik]

Andrew Morton wrote:
> Thanks.  ap->ioaddr.scr_addr is zero in pdc_sata_scr_write().

Very interesting.  Did I miss the hardware info?

I wonder if its a PATA port, and is going down an incorrect path.  The 
PATA port would naturally not have SATA phy registers (SCRs).

	Jeff

Re: [2.6.17-rc6-mm1] Oops during sata_promise init

[prx]

Hi,

sorry for my late answer.

The hardware which Oopses due to the third (PATA) channel is:
02:0b.0 Mass storage controller: Promise Technology, Inc.
PDC20575 (SATAII150 TX2plus) (rev 02)

I know it's the PATA chan which fails because i commented the
part discovering the third channel out, giving me a working
system :)

Thanks for that suggestion / idea :)

kind regards

prx

Re: [2.6.17-rc6-mm1] Oops during sata_promise init

[Jeff Garzik]

prx wrote:
> Hi,
> 
> sorry for my late answer.
> 
> The hardware which Oopses due to the third (PATA) channel is:
> 02:0b.0 Mass storage controller: Promise Technology, Inc.
> PDC20575 (SATAII150 TX2plus) (rev 02)
> 
> I know it's the PATA chan which fails because i commented the
> part discovering the third channel out, giving me a working
> system :)

OK, thanks.

So it sounds like either #upstream broke #promise-sata-pata, or simply a 
bug got introduced into the only-used-in-mm-trees version of sata_promise.c.

I will look into it, but since Promise PATA-on-SATA support is still 
highly experimental, I can't promise an immediate response.  This isn't 
something that is in 2.6.17-rc, or even something going into 2.6.18.

	Jeff