On Thu, 2014-04-17 at 15:37 +0530, Ritesh Raj Sarraf wrote:
> Ben,
>
> You are one of the core contributors to the Network subsystem in the
> Linux kernel, hence asking you this question.
>
>
> Ethernet devices have an interface to enable / disable the device in:
> /sys/class/net/eth0/device/enabled
No, that's a PCI device attribute ('device' is a symlink to the parent
device).
> If the device is disabled using this interface, what is the expected
> power consumption ?
Unchanged, as this attribute has nothing to do with power control
(that's in the 'power' subdirectory). You should never change it while
there's a driver bound to the device.
> I ask because even after I disable the device, powertop reports state
> that ethernet is consuming power.
>
> Power est. Usage Events/s Category Description
> 6.47 W 23.6 pkts/s Device Network
> interface: wlan0 (iwlwifi)
> 4.01 W 0.0 pkts/s Device Network
> interface: eth0 (e1000e)
>
>
> 15:34:57 rrs@zan:/sys/class/net/eth0/device$ cat enabled
> 0
I would question how powertop determines the power of an individual PCI
device, and whether the result is at all reliable. I wouldn't expect a
current 1G Ethernet port ever to use as much as 4 watts. powertop is
probably trying to include CPU power for servicing the device as well,
but this should be close to 0 while it isn't passing traffic.
Ben.
--
Ben Hutchings
If God had intended Man to program,
we'd have been born with serial I/O ports.
Attachment:
signature.asc
Description: This is a digitally signed message part