SATA ALPM min_power causes > 10s delay on resume

SATA ALPM min_power causes > 10s delay on resume

[Jeffrey W. Baker]

I started using the min_power ALPM setting on my ThinkPad to save power
(which works great, thanks!) but it causes a small problem.  When the
machine resumes from suspend-to-ram it hangs for about 10s and then
prints this on the console:

ata1: COMRESET failed (errno=-16)

It then proceeds normally.  The full messages are:

ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
ata1.00: configured for UDMA/100
ata1: failed to recover some devices, retrying in 5 secs
ata1: soft resetting link
ata1: SATA link down (SStatus 611 SControl 300)
ata1: failed to recover some devices, retrying in 5 secs
ata1: hard resetting link
ata1: port is slow to respond, please be patient (Status 0x80)
ata1: COMRESET failed (errno=-16)
ata1: hard resetting link
ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)

After which point the machine works normally.  The SATA controller is an
Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) SATA AHCI Controller (rev
03) and the SATA target is a SAMSUNG MCBQE32G Rev: PS10.  The kernel is
2.6.24.  I haven't tried anything newer.

If the ALPM setting is left at its default of max_performance then
resume from suspend is instant.  Is there any way to work around this?
ALPM is worth a solid 30 minutes of extra battery life, so obviously I'd
like to use it.  If nothing else, it might be helpful to have a module
parameter to shorten the timeout on known-working hardware.

-jwb

Re: SATA ALPM min_power causes > 10s delay on resume

[Grant Grundler]

On Sun, Aug 24, 2008 at 10:27 AM, Jeffrey W. Baker <jwbaker@gmail.com> wrote:
> I started using the min_power ALPM setting on my ThinkPad to save power
> (which works great, thanks!) but it causes a small problem.  When the
> machine resumes from suspend-to-ram it hangs for about 10s and then
> prints this on the console:
>
> ata1: COMRESET failed (errno=-16)
>
> It then proceeds normally.  The full messages are:
>
> ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> ata1.00: configured for UDMA/100
> ata1: failed to recover some devices, retrying in 5 secs
> ata1: soft resetting link
> ata1: SATA link down (SStatus 611 SControl 300)
> ata1: failed to recover some devices, retrying in 5 secs
> ata1: hard resetting link
> ata1: port is slow to respond, please be patient (Status 0x80)
> ata1: COMRESET failed (errno=-16)
> ata1: hard resetting link
> ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)

Can you enable CONFIG_PRINTK_TIME in your kernel?

It sounds like the ahci driver is having problems spinning up
the disk after putting it into a "low power state" (which I
have no clue might be). Getting more accurate time
stamps might be a useful additional clue.

hth,
grant

>
> After which point the machine works normally.  The SATA controller is an
> Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) SATA AHCI Controller (rev
> 03) and the SATA target is a SAMSUNG MCBQE32G Rev: PS10.  The kernel is
> 2.6.24.  I haven't tried anything newer.
>
> If the ALPM setting is left at its default of max_performance then
> resume from suspend is instant.  Is there any way to work around this?
> ALPM is worth a solid 30 minutes of extra battery life, so obviously I'd
> like to use it.  If nothing else, it might be helpful to have a module
> parameter to shorten the timeout on known-working hardware.
>
> -jwb
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ide" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: SATA ALPM min_power causes > 10s delay on resume

[Jeffrey W. Baker]

On Mon, 2008-08-25 at 17:14 -0700, Grant Grundler wrote:
> On Sun, Aug 24, 2008 at 10:27 AM, Jeffrey W. Baker <jwbaker@gmail.com> wrote:
> > I started using the min_power ALPM setting on my ThinkPad to save power
> > (which works great, thanks!) but it causes a small problem.  When the
> > machine resumes from suspend-to-ram it hangs for about 10s and then
> > prints this on the console:
> >
> > ata1: COMRESET failed (errno=-16)
> >
> > It then proceeds normally.  The full messages are:
> >
> > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > ata1.00: configured for UDMA/100
> > ata1: failed to recover some devices, retrying in 5 secs
> > ata1: soft resetting link
> > ata1: SATA link down (SStatus 611 SControl 300)
> > ata1: failed to recover some devices, retrying in 5 secs
> > ata1: hard resetting link
> > ata1: port is slow to respond, please be patient (Status 0x80)
> > ata1: COMRESET failed (errno=-16)
> > ata1: hard resetting link
> > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> 
> Can you enable CONFIG_PRINTK_TIME in your kernel?

Output follows ...

> It sounds like the ahci driver is having problems spinning up
> the disk after putting it into a "low power state" (which I
> have no clue might be). Getting more accurate time
> stamps might be a useful additional clue.

The disk is an SSD so there's no need for delay to allow it to spin up.
Here are the messages from a normal resume, pipe grep ata:

[    0.699797] ata5: port disabled. ignoring.
[    2.028774] ata3: SATA link down (SStatus 0 SControl 300)
[    2.028792] ata2: SATA link down (SStatus 0 SControl 0)
[    2.192670] ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
[    2.193535] ata1.00: ACPI cmd ef/02:00:00:00:00:a0 succeeded
[    2.193537] ata1.00: ACPI cmd f5/00:00:00:00:00:a0 filtered out
[    2.194998] ata1.00: ACPI cmd ef/10:03:00:00:00:a0 succeeded
[    2.195381] ata1.00: configured for UDMA/100
[    2.209049] ata1.00: configured for UDMA/100
[    2.209051] ata1: EH complete

And here are the messages on a slow resume:

[    0.700191] ata5: port disabled. ignoring.
[    2.029795] ata2: SATA link down (SStatus 0 SControl 0)
[    2.029814] ata3: SATA link down (SStatus 0 SControl 300)
[    2.193770] ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
[    2.194631] ata1.00: ACPI cmd ef/02:00:00:00:00:a0 succeeded
[    2.194634] ata1.00: ACPI cmd f5/00:00:00:00:00:a0 filtered out
[    2.196161] ata1.00: ACPI cmd ef/10:03:00:00:00:a0 succeeded
[    2.196545] ata1.00: configured for UDMA/100
[    2.197289] ata1: failed to recover some devices, retrying in 5 secs
[    3.644507] ata1: soft resetting link
[    3.644518] ata1: SATA link down (SStatus 611 SControl 300)
[    3.644530] ata1: failed to recover some devices, retrying in 5 secs
[    4.430864] ata1: hard resetting link
[    5.298009] ata1: port is slow to respond, please be patient (Status 0x80)
[    6.010843] ata1: COMRESET failed (errno=-16)
[    6.010847] ata1: hard resetting link
[    6.085653] ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
[    6.086267] ata1.00: ACPI cmd ef/02:00:00:00:00:a0 succeeded
[    6.086269] ata1.00: ACPI cmd f5/00:00:00:00:00:a0 filtered out
[    6.086380] ata1.00: ACPI cmd ef/10:03:00:00:00:a0 succeeded
[    6.087124] ata1.00: ACPI cmd ef/02:00:00:00:00:a0 succeeded
[    6.087128] ata1.00: ACPI cmd f5/00:00:00:00:00:a0 filtered out
[    6.087350] ata1.00: ACPI cmd ef/10:03:00:00:00:a0 succeeded
[    6.087550] ata1.00: configured for UDMA/100
[    6.088119] ata1.00: configured for UDMA/100
[    6.088122] ata1: EH complete

I don't know how accurate these times are supposed to be.  On my
wallclock it takes substantially more than 6 seconds.

-jwb

Re: SATA ALPM min_power causes > 10s delay on resume

[Jeffrey W. Baker]

On Mon, 2008-08-25 at 20:14 -0700, Jeffrey W. Baker wrote:
> On Mon, 2008-08-25 at 17:14 -0700, Grant Grundler wrote:
> > On Sun, Aug 24, 2008 at 10:27 AM, Jeffrey W. Baker <jwbaker@gmail.com> wrote:
> > > I started using the min_power ALPM setting on my ThinkPad to save power
> > > (which works great, thanks!) but it causes a small problem.  When the
> > > machine resumes from suspend-to-ram it hangs for about 10s and then
> > > prints this on the console:
> > >
> > > ata1: COMRESET failed (errno=-16)
> > >
> > > It then proceeds normally.  The full messages are:
> > >
> > > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > > ata1.00: configured for UDMA/100
> > > ata1: failed to recover some devices, retrying in 5 secs
> > > ata1: soft resetting link
> > > ata1: SATA link down (SStatus 611 SControl 300)
> > > ata1: failed to recover some devices, retrying in 5 secs
> > > ata1: hard resetting link
> > > ata1: port is slow to respond, please be patient (Status 0x80)
> > > ata1: COMRESET failed (errno=-16)
> > > ata1: hard resetting link
> > > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > 
> > Can you enable CONFIG_PRINTK_TIME in your kernel?
> 
> Output follows ...
> 
> > It sounds like the ahci driver is having problems spinning up
> > the disk after putting it into a "low power state" (which I
> > have no clue might be). Getting more accurate time
> > stamps might be a useful additional clue.
> 
> The disk is an SSD so there's no need for delay to allow it to spin
> up.

I just did an experiment, and a workaround for this problem is to enable
max_performance immediately before sleep and return to min_power during
resume.

-jwb

Re: SATA ALPM min_power causes > 10s delay on resume

[Tejun Heo]

Jeffrey W. Baker wrote:
> I started using the min_power ALPM setting on my ThinkPad to save power
> (which works great, thanks!) but it causes a small problem.  When the
> machine resumes from suspend-to-ram it hangs for about 10s and then
> prints this on the console:
> 
> ata1: COMRESET failed (errno=-16)
> 
> It then proceeds normally.  The full messages are:
> 
> ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> ata1.00: configured for UDMA/100
> ata1: failed to recover some devices, retrying in 5 secs
> ata1: soft resetting link
> ata1: SATA link down (SStatus 611 SControl 300)
> ata1: failed to recover some devices, retrying in 5 secs
> ata1: hard resetting link
> ata1: port is slow to respond, please be patient (Status 0x80)
> ata1: COMRESET failed (errno=-16)
> ata1: hard resetting link
> ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> 
> After which point the machine works normally.  The SATA controller is an
> Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) SATA AHCI Controller (rev
> 03) and the SATA target is a SAMSUNG MCBQE32G Rev: PS10.  The kernel is
> 2.6.24.  I haven't tried anything newer.
> 
> If the ALPM setting is left at its default of max_performance then
> resume from suspend is instant.  Is there any way to work around this?
> ALPM is worth a solid 30 minutes of extra battery life, so obviously I'd
> like to use it.  If nothing else, it might be helpful to have a module
> parameter to shorten the timeout on known-working hardware.

Which kernel version are you using?

-- 
tejun

Re: SATA ALPM min_power causes > 10s delay on resume

[Jeffrey W. Baker]

On Sat, 2008-08-30 at 13:02 +0200, Tejun Heo wrote:
> Jeffrey W. Baker wrote:
> > I started using the min_power ALPM setting on my ThinkPad to save power
> > (which works great, thanks!) but it causes a small problem.  When the
> > machine resumes from suspend-to-ram it hangs for about 10s and then
> > prints this on the console:
> > 
> > ata1: COMRESET failed (errno=-16)
> > 
> > It then proceeds normally.  The full messages are:
> > 
> > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > ata1.00: configured for UDMA/100
> > ata1: failed to recover some devices, retrying in 5 secs
> > ata1: soft resetting link
> > ata1: SATA link down (SStatus 611 SControl 300)
> > ata1: failed to recover some devices, retrying in 5 secs
> > ata1: hard resetting link
> > ata1: port is slow to respond, please be patient (Status 0x80)
> > ata1: COMRESET failed (errno=-16)
> > ata1: hard resetting link
> > ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
> > 
> > After which point the machine works normally.  The SATA controller is an
> > Intel Corporation 82801HBM/HEM (ICH8M/ICH8M-E) SATA AHCI Controller (rev
> > 03) and the SATA target is a SAMSUNG MCBQE32G Rev: PS10.  The kernel is
> > 2.6.24.  I haven't tried anything newer.
> > 
> > If the ALPM setting is left at its default of max_performance then
> > resume from suspend is instant.  Is there any way to work around this?
> > ALPM is worth a solid 30 minutes of extra battery life, so obviously I'd
> > like to use it.  If nothing else, it might be helpful to have a module
> > parameter to shorten the timeout on known-working hardware.
> 
> Which kernel version are you using?

2.6.24 (2.6.24-19-generic package in Ubuntu 8.04)

Re: SATA ALPM min_power causes > 10s delay on resume

[Tejun Heo]

Jeffrey Baker wrote:
> 2.6.24 (2.6.24-19-generic package in Ubuntu 8.04).

Can you please give more recent kernel a try?  2.6.26.3 preferably?

Thanks.

-- 
tejun

Re: SATA ALPM min_power causes > 10s delay on resume

[Jeffrey W. Baker]

On Sun, 2008-08-31 at 11:35 +0200, Tejun Heo wrote: 
> Jeffrey Baker wrote:
> > 2.6.24 (2.6.24-19-generic package in Ubuntu 8.04).
> 
> Can you please give more recent kernel a try?  2.6.26.3 preferably?

I wasn't able to try 2.6.26.3 but in 2.6.27 this seems fixed.

-jwb