COMRESET failure when sata link power management policy is set to min_power on thinkpad T61

[VS]

After changing the S-ATA link power management policy to min_powed (echo min_power > /sys/class/scsi_host/host0/link_power_management_policy), my laptop (a Thinkpad T61, model 7664-18G) is relatively slow to resume from suspend to ram. A look in the syslog (relevant section attached) reveals a few error messages (as you can see, it pauses for at least ten seconds, hence the slow resume).

[ 3.912093] ata1: soft resetting link
[ 3.912105] ata1: SATA link down (SStatus 611 SControl 300)
[ 3.912114] ata1: failed to recover some devices, retrying in 5 secs
[ 4.576808] ata1: hard resetting link
[ 5.309541] ata1: port is slow to respond, please be patient (Status 0x80)
[ 5.919953] ata1: COMRESET failed (errno=-16)
[ 5.919959] ata1: hard resetting link
[ 5.986791] ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 300)

Changing the link power management policy back to max_performance makes the problem go away.

I've tried upgrading the bios to the current version, 2.17, but no change.

[Chris Jones (cmsj)]

I can reliably reproduce this on a Thinkpad X300, and it is mentioned at http://www.thinkwiki.org/wiki/How_to_reduce_power_consumption#SATA_Link_Power_Management
My assumption is that the kernel should be putting the link back into full power before it suspends - or be able to detect that the link is in low power mode on resume, and handle it appropriately.

[kiev1 (sys-sys-admin)]

This kernel bug arise up at copying of large file

all more users run into this misfortune
developers can not do nothing
development of linux reached a deaf impasse

kernel 2.6.24-17-generic is similarly subject to this vulnerability allowing to lose all information on SATA driver and physically to spoil hd driver!!!

http://lkml.org/lkml/2007/6/14/154
http://kerneltrap.org/mailarchive/li...07/6/14/103765
http://kerneltrap.org/node/16175

"System hang from time to time" http://bugzilla.kernel.org/show_bug.cgi?id=8300

"sata hotplug removal of drive freezes all 2.6.21 kernels"
http://bugzilla.kernel.org/show_bug.cgi?id=8421

"(sata_via) system freeze in random time"
http://bugzilla.kernel.org/show_bug.cgi?id=9115

"kernel freezes with on clockevent warning"
http://bugzilla.kernel.org/show_bug.cgi?id=9834

"[pata_ali] Unspecified hang on Acer laptop"
http://bugzilla.kernel.org/show_bug.cgi?id=9898

"System freezes after I/O on pata_jmicron device"
http://bugzilla.kernel.org/show_bug.cgi?id=10296

Computer hangs up (freeze) and reports SATA errors
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/217920

sata controller/drive reports exception Emask 0x10 SAct 0x0 SErr 0x40d0002 action 0x2 frozen
https://bugs.launchpad.net/ubuntu/+bug/164183

port is slow to respond" error freezes OS for seconds while reading HDs
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/229747

Port slow to respond on SiI3512 with sata_sil
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/159521

sata controller/drive reports exception Emask 0x10 SAct 0x0 SErr 0x40d0002 action 0x2 frozen
https://bugs.launchpad.net/ubuntu/+bug/164183

sata disk speed negotiation with heavy disk activity
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/187146

very bad I/O performance with linux kernel in Hardy
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/221437

Hardy 'hangs' - 'ata1: port is slow to respond' in dmesg
https://bugs.launchpad.net/ubuntu/+bug/226600

"System hang from time to time" http://bugzilla.kernel.org/show_bug.cgi?id=8300

"sata hotplug removal of drive freezes all 2.6.21 kernels" http://bugzilla.kernel.org/show_bug.cgi?id=8421

"(sata_via) system freeze in random time" http://bugzilla.kernel.org/show_bug.cgi?id=9115

"System freezes after I/O on pata_jmicron device" http://bugzilla.kernel.org/show_bug.cgi?id=10296

"weird message in syslog" http://ubuntuforums.org/showthread.php?t=760046

Development of linux reached an impasse.

Nobody can not do nothing!

!!! :(((((((((((((((

[Chris Jones (cmsj)]

kiev1: those bugs are unrelated to this one - this bug is purely about a harmless timewasting link reset when resuming from ram suspend. Not all COMRESET messages have the same cause.

[kiev1 (sys-sys-admin)]

from ram suspend takes place reading of large file -> and here the given shows up this kernels bug

[Chris Jones (cmsj)]

kiev1: I am not sure exactly what you mean, but this bug is purely about the amount of time it takes to resume when the SATA link is being aggressively power managed. there is *no* dataloss, corruption or hardware damage associated with this bug. The bugs you mention are different and not related to this one.

[kiev1 (sys-sys-admin)]

I disable all power manager (hald stop, add kernel options noapic acpi=off irgpoll) and problem was present
information of destroy as a result of hanging up of server - sometimes a server was simply disconnected from hdd and did not see him - even reboot did not work is a
report of "reboot command not found"
see http://bugzilla.kernel.org/show_bug.cgi?id=9115

[Chris Jones (cmsj)]

kiev1: please stop spamming this bug report. your bug(s) are different from this one. This page is not relevant to you. Please either report a new bug or subscribe to one which matches your bug(s).

[Leann Ogasawara (leannogasawara)]

Hi Vegar and Chris,

Care to quickly confirm this is still an issue with the upcoming Intrepid Ibex 8.10 release which has a 2.6.26 based kernel? http://www.ubuntu.com/testing . You should be able to test Suspend via a LiveCD. Thanks.

[VS (storvann)]

Appears to be be fixed, at least with the live CD. I'll try to install the .26 kernel on my hardy install later this week, to confirm 100%.

[Leann Ogasawara (leannogasawara)]

Hi Veger,

Thanks for testing and the feedback. Based on your last comment I'm going to tentatively mark this "Fix Released" for Intrepid. If you notice any sort of regressions regarding this issue, please feel free to flip the status back to "New". Unfortunately against Hardy this does not qualify for a Stable Release Update. Thanks.

[Chris Jones (cmsj)]

I can confirm that this is fixed for my running current Intrepid (so 2.6.27)