unable to modify scsi host link_power_management_policy

[trendzetter]

Might be a regression from Hardy: https://bugs.launchpad.net/bugs/231342

root@ubuntu:/# echo min_power > /sys/class/scsi_host/host0/link_power_management_policy
root@ubuntu:/# cat /sys/class/scsi_host/host0/link_power_management_policy
max_performance
root@ubuntu:/#

ProblemType: Bug
Architecture: amd64
ArecordDevices:
 **** List of CAPTURE Hardware Devices ****
 card 0: Intel [HDA Intel], device 0: ALC269 Analog [ALC269 Analog]
   Subdevices: 1/1
   Subdevice #0: subdevice #0
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: joris 3089 F.... pulseaudio
CRDA: Error: [Errno 2] No such file or directory
Card0.Amixer.info:
 Card hw:0 'Intel'/'HDA Intel at 0xd4400000 irq 22'
   Mixer name : 'Intel G45 DEVCTG'
   Components : 'HDA:10ec0269,1025022b,00100004 HDA:80862802,80860101,00100000'
   Controls : 23
   Simple ctrls : 12
Date: Fri Sep 11 17:04:21 2009
DistroRelease: Ubuntu 9.10
HibernationDevice: RESUME=UUID=3f4200c1-4289-4ec8-9515-11e12561cd79
MachineType: Acer Aspire 5810T
Package: linux-image-2.6.31-10-generic 2.6.31-10.32
PccardctlIdent:

PccardctlStatus:

ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.31-10-generic root=UUID=d9aeca27-f2de-4b4b-ac98-9ff714300820 ro quiet splash libata.force=noncq usbcore.autosuspend=1
ProcEnviron:
 LANGUAGE=en_US:en
 LANG=en_US.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-10.32-generic
RelatedPackageVersions:
 linux-backports-modules-2.6.31-10-generic 2.6.31-10.12
 linux-firmware 1.16
SourcePackage: linux
Uname: Linux 2.6.31-10-generic x86_64
dmi.bios.date: 06/24/2009
dmi.bios.vendor: INSYDE
dmi.bios.version: V1.23
dmi.board.asset.tag: Base Board Asset Tag
dmi.board.name: Aspire 5810T
dmi.board.vendor: Acer
dmi.board.version: Base Board Version
dmi.chassis.type: 1
dmi.chassis.vendor: Chassis Manufacturer
dmi.chassis.version: Chassis Version
dmi.modalias: dmi:bvnINSYDE:bvrV1.23:bd06/24/2009:svnAcer:pnAspire5810T:pvrV1.23:rvnAcer:rnAspire5810T:rvrBaseBoardVersion:cvnChassisManufacturer:ct1:cvrChassisVersion:
dmi.product.name: Aspire 5810T
dmi.product.version: V1.23
dmi.sys.vendor: Acer

[Leann Ogasawara (leannogasawara)]

Hi trendzetter,

I'm unfortunately unable to reproduce the issue here, but am on a newer 2.6.31-10.34 kernel.

root@emiko:~# cat /proc/version_signature
Ubuntu 2.6.31-10.34-generic
root@emiko:~# cat /sys/class/scsi_host/host0/link_power_management_policy
max_performance
root@emiko:~# echo min_power > /sys/class/scsi_host/host0/link_power_management_policy
root@emiko:~# cat /sys/class/scsi_host/host0/link_power_management_policy
min_power

[trendzetter (trendzetter)]

I noticed that it is only host0 that is fixed on max_performance. I also have host1 to 5 that are fine (configurable). I'm not sure on which host my sata disc is connected.
The fan on my brand new ultra energy efficient laptop keeps spinning all the time which should not be the case I guess. Powertop keeps telling me I should look into this issue with /sys/class/scsi_host/host0/link_power_management_policy

[Leann Ogasawara (leannogasawara)]

I've marked bug 431822 as a duplicate of this bug. A subscriber there noted "It's interesting to note that of the six scsi_hosts, 0, 4, 5 remain in max_performance whereas 1,2,3 can be set to min_power."

[Nixblicker (el-lunatico)]

I can confirm this buug on my Acer timeline 3810T machine running with jaunty 64 bit.

Funny thing is, although i am unable to change the policy setting manually, powertop seems to be able to toggle that feature as it stops noticing me about this possible tweak. And i think but cannot proof, that i has a impact on power consumption.
Or may this probably be related to a fix i had to use when i was installing Jaunty to make my Hitachi 320 GB disk accept the AHCI feature - i just remember i had to switch off NCQ to stop the HD from producing sata lockups and system hangs.

[ThomasMartitz (thomas-martitz)]

Bug #231342 seems like a duplicate.

I have this problem too (Acer 3810TG). Writing to that file hs no effect, and powertop keeps suggesting that.

[Miguel Branco (arlanthir)]

I confirm this bug on my Timeline 4810t, hosts 0 and 5 don't keep the settings while the others do.

[Alexander van Loon (avanloon)]

Confirming, using an Acer TravelMate TimeLine 8371 here (which is nearly identical to the TimeLine 3810T). Like the others said, no matter what I tried with the various power saving scripts or manual commands, only PowerTOP seems to be able to change it.

[might69 (pmoyanoperez)]

I confirm this bug on my Packar Bell Butterfly FM001-SP (Acer Timeline 3810TG clone)

[Andy Whitcroft (apw)]

Link power control is complex. Firstly not all host controllers support it, and you must determine if that support is there from the host controller flags. Secondly power saving has little meaning where there is not a device present, and indeed if you enable powersaving there hotplug no longer works. Where link control is not appropriate any attempt to set power modes is silently ignored.

The reporter indicates that host0 is the only one which is not taking a value, that the other 5 do indeed take the setting. If we look at his boot dmesg we see the following:

Firstly note the 'pm' in the controller flags, this indicates the there is power management support:

    [ 3.899062] ahci 0000:00:1f.2: flags: 64bit ncq sntf ilck stag pm led clo pmp pio slum part ems

These are the sata host channels and their capabilities, note that ata3/4 are fake entries so should be ignored behaviour wise:

    [ 3.921463] ata1: SATA max UDMA/133 abar m2048@0xd4405000 port 0xd4405100 irq 26
    [ 3.921468] ata2: SATA max UDMA/133 abar m2048@0xd4405000 port 0xd4405180 irq 26
    [ 3.921471] ata3: DUMMY
    [ 3.921473] ata4: DUMMY
    [ 3.921477] ata5: SATA max UDMA/133 abar m2048@0xd4405000 port 0xd4405300 irq 26
    [ 3.921482] ata6: SATA max UDMA/133 abar m2048@0xd4405000 port 0xd4405380 irq 26

Note that the first link is down and second is up, likely the disk is on ata2, ie. on host1. Host 0 we would therefore likely expect not to be able to control as there is no link:

    [ 4.288992] ata1: SATA link down (SStatus 0 SControl 300)
    [ 4.984177] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
    [ 5.892161] ata5: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
    [ 6.292115] ata6: SATA link down (SStatus 0 SControl 300)

As long as you are able to control host1 and host 4 you are controlling the links which exist as far as I can tell.

[Andy Whitcroft (apw)]

Could the other reporters please attach a full boot dmesg.

[Attilio (nexor-tele2)]

Added dmesg log, in my case the affected links (0 and 5) are actually empty:

[root@acer attilio]# cat /sys/class/scsi_host/host0/link_power_management_policy
max_performance
[root@acer attilio]# cat /sys/class/scsi_host/host1/link_power_management_policy
min_power
[root@acer attilio]# cat /sys/class/scsi_host/host2/link_power_management_policy
min_power
[root@acer attilio]# cat /sys/class/scsi_host/host3/link_power_management_policy
min_power
[root@acer attilio]# cat /sys/class/scsi_host/host4/link_power_management_policy
min_power
[root@acer attilio]# cat /sys/class/scsi_host/host5/link_power_management_policy
max_performance

[Alexander van Loon (avanloon)]

Andy, thanks a lot for taking the time to take a look at this bug, I appreciate the explanation very much. I never used dmesg before, but after searching a bit I assume that posting the resulting file of the command ‘dmesg | tee <filename>’ should be okay.

So if I understand the explanation correctly, you are implying that this is a bug in PowerTOP, because it is incorrectly suggesting that it would be useful to enable power saving for host0? The latest version of PowerTOP, 1.13, still does this on my completely up to date Maverick installation. Should I notify the PowerTOP developers of this bug?

[Alexander van Loon (avanloon)]

Oh, and if it’s any use, inputting the same cat commands as Attilio gives me max_performance for hosts 0, 4 and 5 and min_power for the rest.

[Jeremy Foshee (jeremyfoshee)]

This bug report was marked as Incomplete and has not had any updated comments for quite some time. As a result this bug is being closed. Please reopen if this is still an issue in the current Ubuntu development release http://cdimage.ubuntu.com/daily-live/current/ . Also, please be sure to provide any requested information that may have been missing. To reopen the bug, click on the current status under the Status column and change the status back to "New". Thanks.

[This is an automated message. Apologies if it has reached you inappropriately; please just reply to this message indicating so.]

[Alexander van Loon (avanloon)]

I just started PowerTOP on my notebook with Kubuntu 10.10 and I still noticed the problem. I don’t know if it still occurs with the latest daily Live CD of what is going to become Kubuntu 11.04, but please reopen the bug. I’ll definitely test with a daily Live CD when I have more time on my hands.

[Alexander van Loon (avanloon)]

Hmm, somehow my comment was eaten by Launchpad?

[Alexander van Loon (avanloon)]

Well Launchpad didn’t eat my comment, I thought so when I didn’t see comment #16 appearing immediately. Anyway, I’ve installed the latest daily Live CD of Kubuntu 11.04 on my notebook today and have been monitoring PowerTOP for half an hour approximately. I didn’t see a message of PowerTOP complaining that SCSI link power management wasn’t working properly.

To verify I tried the cat /sys/class/scsi_host/host0/link_power_management_policy (also for hosts 1 to 5) command again, and this time ‘min_power’ was reported for all six hosts. So I guess it fixed for my machine in combination with the latest 11.04 development branch then. Then it seems the bug can remain closed, but hopefully anyone else could verify that it is also fixed on their machines?