On 09/13/2012 04:39 PM, Jack Wang wrote:

quoted

quoted

quoted

quoted

For hard disk runtime power off, Is there (hardware?) requirement for

hard

quoted

quoted

drive to support similar feature?

No requirement for hard drive, as it does not need to remote wake up
itself while powered off.

Please note that there is a v7 sent yesterday, you may want to take a
look at that instead of v6.

And the runtime power off of hard drive is dropped from that patch set,
I'll submit it in another patch set(pretty small changes needed).

Thanks,
Aaron

[Jack Wang] Thanks Aaron for quick replay, for disk runtime power off,

will

quoted

quoted

it   a different proach from block runtime power management patch set

from

quoted

quoted

Lin Ming?

I suppose you mean the "block layer runtime pm" from Lin Ming?
No, they are different things.

That patch set introduced a mechanism to decide when a device can enter
runtime suspend based on block layer request, while my patch cares about
when it is runtime suspended, how to remove its power.

Thanks,
Aaron

[Jack Wang] Thanks, but I still not get how to remove hard disk power?
Through acpi? But libata acpi not cover scsi disk right? Sorry if my
question is too stupid.

When scsi disk is runtime suspended, the ata port it attached to will be
runtime suspended, and then the power will be removed through ACPI.

Please check libata-acpi.c, function ata_acpi_set_state.

The parent/child relationship of the devices:
ata_host
  ata_port
    scsi_host
      scsi_target
        scsi_device

If you have more questions on how it works, please send email only
to me, other people might be too busy to read such emails :-)

Thanks,
Aaron