sata_sil problems with recent kernels

sata_sil problems with recent kernels

[Dale Blount]

Hi,

Excuse me if this has been covered or fixed, I couldn't find anything in
the archives.

I upgraded from 2.6.11.7 to 2.6.20.1 today and found all the drives
connected to 2 brands of sata_sil sata controllers not working.  The
drives are also (now) of various brands, Maxtor 300GB and 500GB
Seagates.

On boot (or after modprobing sata_sil if not configured to load in my
initrd), I get the below messages in dmesg and no drives showing up
in /dev.

There's 4 drives total, and 6 BUGs so I'm a little confused.  One of the
drive (old sdb, which happened to be the first drive on sata_sil
controllers) is known to have bad sectors, but I didn't think it could
cause this... I'm away from the machine currently, so I haven't tested a
single drive to see if it fails also.

I rebooted back into 2.6.11.7 and everything was peachy (minus the known
bad disk).

Any suggestions or hints to more data I could provide to be diagnosed?

Thanks,

Dale

***************

ata3: SATA max UDMA/100 cmd 0xF8848080 ctl 0xF884808A bmdma 0xF8848000
irq 18
ata4: SATA max UDMA/100 cmd 0xF88480C0 ctl 0xF88480CA bmdma 0xF8848008
irq 18
scsi2 : sata_sil
ata3: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata3.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata3.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata3: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c011e519>] __wake_up_common+0x39/0x70
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata3.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata3.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata3: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c011e519>] __wake_up_common+0x39/0x70
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata3.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata3.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata3: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
scsi3 : sata_sil
ata4: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c011e519>] __wake_up_common+0x39/0x70
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata4.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata4.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata4: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c011e519>] __wake_up_common+0x39/0x70
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata4.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata4.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata4: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
BUG: at drivers/ata/libata-core.c:4365 ata_hsm_move()
 [<f882c9ae>] ata_hsm_move+0x1be/0x7d0 [libata]
 [<c033eb42>] schedule_timeout+0x52/0xd0
 [<c011e519>] __wake_up_common+0x39/0x70
 [<c0130170>] process_timeout+0x0/0x10
 [<f882fe97>] ata_pio_task+0x27/0x100 [libata]
 [<c01378d3>] run_workqueue+0x93/0x140
 [<f882fe70>] ata_pio_task+0x0/0x100 [libata]
 [<c0138447>] worker_thread+0x147/0x170
 [<c0121360>] default_wake_function+0x0/0x10
 [<c0138300>] worker_thread+0x0/0x170
 [<c013b2eb>] kthread+0xbb/0xf0
 [<c013b230>] kthread+0x0/0xf0
 [<c0103e4f>] kernel_thread_helper+0x7/0x18
 =======================
ata4.00: zero err_mask for failed internal command, assuming
AC_ERR_OTHER
ata4.00: failed to IDENTIFY (I/O error, err_mask=0x100)
ata4: SATA link up 1.5 Gbps (SStatus 113 SControl 310)

Re: sata_sil problems with recent kernels

[Dale Blount]

On Fri, 2007-02-23 at 12:00 -0500, Dale Blount wrote:
> Hi,
> 
> Excuse me if this has been covered or fixed, I couldn't find anything in
> the archives.
> 
> I upgraded from 2.6.11.7 to 2.6.20.1 today and found all the drives
> connected to 2 brands of sata_sil sata controllers not working.  The
> drives are also (now) of various brands, Maxtor 300GB and 500GB
> Seagates.


I have some additional information coming from more testing and I also
CC'd linux-ide and Jeff Garzik.
Original message (with BUG) can be found here: 
http://permalink.gmane.org/gmane.linux.kernel/497140


I'm using my distro's modular kernel (ArchLinux).  The 2.6.19.x kernels
from the distro also have the problem.  However, Knoppix 5.1.1 which
ships with 2.6.19.1 detects all the drives on sata_sil controllers
without errors, and PLD linux live rescue CD (2.6.19.2) detects 2 of the
3 currently connected, the third is just absent, no BUG like when using
Arch.

I noticed Knoppix had sata_sil=y, so I tried building it in as well with
the same BUG results.

My .config (distro standard) can be found here:
http://tinyurl.com/yul8eu

I will be happy to test/provide more information on request.

Thanks again,

Dale

Re: sata_sil problems with recent kernels

[Dale Blount]

On Tue, 2007-02-27 at 13:54 -0500, Dale Blount wrote:
> On Fri, 2007-02-23 at 12:00 -0500, Dale Blount wrote:
> > Hi,
> > 
> > Excuse me if this has been covered or fixed, I couldn't find anything in
> > the archives.
> > 
> > I upgraded from 2.6.11.7 to 2.6.20.1 today and found all the drives
> > connected to 2 brands of sata_sil sata controllers not working.  The
> > drives are also (now) of various brands, Maxtor 300GB and 500GB
> > Seagates.

For the archives, the fix is documented here:

http://article.gmane.org/gmane.linux.ide/16304

Dale

Re: sata_sil problems with recent kernels

[Tejun Heo]

Dale Blount wrote:
> On Tue, 2007-02-27 at 13:54 -0500, Dale Blount wrote:
>> On Fri, 2007-02-23 at 12:00 -0500, Dale Blount wrote:
>>> Hi,
>>>
>>> Excuse me if this has been covered or fixed, I couldn't find anything in
>>> the archives.
>>>
>>> I upgraded from 2.6.11.7 to 2.6.20.1 today and found all the drives
>>> connected to 2 brands of sata_sil sata controllers not working.  The
>>> drives are also (now) of various brands, Maxtor 300GB and 500GB
>>> Seagates.
> 
> For the archives, the fix is documented here:
> 
> http://article.gmane.org/gmane.linux.ide/16304

Yeap, that's the correct fix and the problem is introduced in 2.6.20 by
using polling IDENTIFY by default.  Well, the bug has always been there
but no one was issuing polling command to sata_sil upto 2.6.19.

This will probably included in the next -stable iteration.  Thanks.

-- 
tejun

Re: sata_sil problems with recent kernels

[Patrick Ale]

On 3/5/07, Tejun Heo <htejun@gmail.com> wrote:
> Dale Blount wrote:

>
> This will probably included in the next -stable iteration.  Thanks.
>
> --
> tejun


I can confirm that sata_sil is working great under 2.6.21-rc1.
I have an onboard Silicon Image SATA RAID controller and a PCI Add-on
card, also with silicon image chipset.

2x500GB Samsung
1x400GB Samsung

I am getting maximum troughput and no errors of any kind.


Patrick