Jurgen,
As addition: the word AHCI is not in the motherboard manual. I'll do
some gooling for this.
Regards,
Tom.
J?rgen Keil wrote:
>> Any clue how I can trace or debug this? Any dtracing (or similar) that can
>> be done?
>> Currenly the only clue I have is a Watt meter on the other end pf the
>> powersupply...
>
> I experimented with the following dtrace script, but noticed that
> pm_set_power is not called for a HDD controlled by the ata driver:
>
> ==============================
> #!/usr/sbin/dtrace -s
>
> #pragma D option flowindent
>
>
> fbt::pm_set_power:entry
> {
> this->dip = (struct dev_info *)arg0;
> printf(".../%...@%s/%...@%s, comp %d, level %d, dir %d",
> stringof(this->dip->devi_parent->devi_node_name),
> stringof(this->dip->devi_parent->devi_addr_buf),
> stringof(this->dip->devi_node_name),
> stringof(this->dip->devi_addr_buf),
> (int)arg1, (int)arg2, (int)arg3);
> }
>
>
> fbt::pm_set_power:return
> {
> }
>
>
> fbt::pm_scan_dev:entry
> {
> this->dip = (struct dev_info *)arg0;
> printf(".../%...@%s/%...@%s",
> stringof(this->dip->devi_parent->devi_node_name),
> stringof(this->dip->devi_parent->devi_addr_buf),
> stringof(this->dip->devi_node_name),
> stringof(this->dip->devi_addr_buf));
> }
>
>
> fbt::pm_scan_dev:return
> {
> }
> ==============================
>
>> Regards,
>>
>> Tom
>>
>> J?rgen Keil wrote:
>>>> I'm trying to workout how I must configure power
>>>> management so that it will idle sata disks.
>>>>
>>>> I've added
>>>> autopm enable
>>>> device-thresholds /dev/dsk/c0d0s0 60s
>>>> device-thresholds /dev/dsk/c0d1s0 60s
>>>> to /etc/power.conf
>>>> Also did a /usr/sbin/pmconfig
>>> I have (that is, dtpower has) added physical device paths
>>> in /etc/power.conf, for the s-ata disks:
>> This is a system running from a 4GB flash disk.
>> Therefore I had to minimize the package (that is I did a core-up
>> approach). There is no X-windows running.
>> the deamon is running - it does manage the cpu....
>> (kstat -m cpu_info helps here)
>
> And that 4GB flash disk is using a S-ATA interface?
>
>
>>> device-dependency-property removable-media /dev/fb
>>> autoS3 default
>>> S3-support enable
>>> autopm enable
>>> autoshutdown 30 9:00 9:00 noshutdown
>>> device-thresholds /pci at 0,0/pci-ide at d/ide at 1/sd at 1,0 1800s
>>> device-thresholds /pci at 0,0/pci1043,81c0 at e/disk at 0,0 1800s
>>> device-thresholds /pci at 0,0/pci1043,81c0 at f/disk at 0,0 1800s
>>> device-thresholds /pci at 0,0/pci1043,81c0 at f/disk at 1,0 1800s
>>>
>>>
>>>> And added
>>>> standby=60;
>>>> to /kernel/drv/ata.conf
>>> As far as I know, this setting is more or less obsolete by the
>>> putback for 6455736 (in snv_77 and newer):
>>> Synopsis: ata/dadk/cmdk should support DDI_SUSPEND/DDI_RESUME
>> I'm using build 91.
>
> Hmm, after a closer look at the ata driver source I found that
> ATA_USE_AUTOPM define. I think this needs to be defined
> before the ata driver / p-ata hdds work with /etc/power.conf.
> But I can't find the place where ATA_USE_AUTOPM might be
> defined. Apparently p-ata hdds can not yet be controlled by
> powerd / /etc/power.conf (unless you recompile the ata
> driver from source and define ATA_USE_AUTOPM somewhere):
>
> http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/uts/intel/io/dktp/controller/ata/ata_common.c#123
>
>
> That should explain why my dtrace experiments didn't
> show pm_set_power calls for ata disks ...
>
>
>
>>>> Both disks a a mirror in a zfs pool. The OS and swap
>>>> runs from a other disk.
>>>>
>>>> Monitoring the power usage - I most strongly think that the disks aren't
>>>> idled. But I did not find any command to actually get this from the OS.
>>>>
>>>> I also tried with the physical device paths in /etc/power.conf - same
>>>> result.
>>> What exactly did you use as physical device paths?
>> /pci at 0,0/pci-ide at f/ide at 0/cmdk at 0,0
>> /pci at 0,0/pci-ide at f/ide at 1/cmdk at 0,0
>>
>> the paths are as show by format.
>
> Aha, so you're using the p-ata driver ...
>
> But the disks are s-ata? What kind of s-ata controller
> are you using? Maybe you should have enabled AHCI
> mode in the system's BIOS; it seems that the s-ata
> controller emulates a standard legacy p-ata pci-ide
> controller, so Solaris' p-ata driver "ata" is used...
>
>>>> How can I trace/debug/monitor the power.conf config?
>>>> Any clue on what I might have missed?
>>> --
>>> This message posted from opensolaris.org
>>> _______________________________________________
>>> pm-discuss mailing list
>>> pm-discuss at opensolaris.org
>>>
>> http://mail.opensolaris.org/mailman/listinfo/pm-discuss
>> _______________________________________________
>> pm-discuss mailing list
>> pm-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/pm-discuss
> --
> This message posted from opensolaris.org
> _______________________________________________
> pm-discuss mailing list
> pm-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/pm-discuss
