On 08/07/2012 02:41 PM, Mauro Carvalho Chehab wrote:
Em 06-08-2012 09:21, Antti Palosaari escreveu:
On 08/05/2012 08:44 PM, Mauro Carvalho Chehab wrote:
The dvb-usb-v2 core doesn't know anything about CI. So, the
driver needs to handle it by hand. This patch stops CI just
before stopping URB's/RC, and restarts it before URB/RC start.
It should be noticed that suspend/resume is not yet working properly,
as the PM model requires the implementation of reset_resume:
dvb_usb_az6007 1-6:1.0: no reset_resume for driver dvb_usb_az6007?
But this is not implemented there at dvb-usb-v2 yet.
That is true, but it is coming:
http://blog.palosaari.fi/2012/07/dvb-power-management-on-suspend.html
http://git.linuxtv.org/anttip/media_tree.git/shortlog/refs/heads/dvb_core3
At the time I added initial suspend/resume support for dvb-usb-v2 I left those out purposely as I saw some study and changes are needed for DVB-core/frontend.
Normally suspend keeps USB-device powered and calls .resume() on resume. But on certain conditions USB device could lose power
during suspend and on that case reset_resume() is called, and if there is no reset_resume() is calls disconnect() (and probe() after that).
This should depend on BIOS settings, and what of the following type of suspend[1]
was done:
S1: All processor caches are flushed, and the CPU(s) stops executing instructions.
Power to the CPU(s) and RAM is maintained; devices that do not indicate they
must remain on may be powered down.
S2: CPU powered off. Dirty cache is flushed to RAM.
S3: Commonly referred to as Standby, Sleep, or Suspend to RAM. RAM remains powered
S4: Hibernation or Suspend to Disk. All content of main memory is saved to non-volatile
memory such as a hard drive, and is powered down.
[1] http://en.wikipedia.org/wiki/Advanced_Configuration_and_Power_Interface
That was something I was already aware. There is even S5 and S4b
mentioned by Kernel documentation. But in real life you have to care only:
S3, Suspend, suspend to ram
S4, Hibernation, suspend to disk
And from the USB-driver point of view those are covered by there three
callbacks:
.suspend()
.resume()
.reset_resume()
* if reset_resume() does not exits .disconnect() + .probe() is called
instead
What is my current understanding S3 level leaves USB/PCI powered
normally, but device driver should drop device to low power state. In
case of DVB -device this means all sub-drivers should put sleep.
S4 naturally powers everything off. Also worth to mention laptops will
switch from S3 to S4 if battery drains empty during S3.
There are also some per-device sysfs nodes that control how PM will work for them.
See:
$ tree /sys/devices/pci0000:00/0000:00:1d.7/usb1/1-8/dvb/dvb0.frontend0
/sys/devices/pci0000:00/0000:00:1d.7/usb1/1-8/dvb/dvb0.frontend0
├── dev
├── device -> ../../../1-8
├── power
│ ├── async
│ ├── autosuspend_delay_ms
│ ├── control
│ ├── runtime_active_kids
│ ├── runtime_active_time
│ ├── runtime_enabled
│ ├── runtime_status
│ ├── runtime_suspended_time
│ └── runtime_usage
├── subsystem -> ../../../../../../../class/dvb
└── uevent
There are a number of pm functions that can change the power management behavior
as well.
Not sure how to control it, but, IMHO, for a media device, it only makes sense
to keep it powered on suspend if the device has IR and if the power button of
the IR could be used to wake up the hardware. Otherwise, the better is to just
power it off, to save battery (for notebooks).
yeah, and it was already done.
Maybe it makes sense to talk with Raphael Wysocki to be sure that it will cover
all possible cases: auto-suspend, S1/S2/S3/S4 and "wake on IR").
That IR was something I wasn't noticed at all. Currently it stops IR
polling too. If that kind of functionality is needed it is surely some
more work as you cannot stop IR-polling. Maybe I will skip it that time
as I don't have time for it currently :) If someone wish to learn how
USB polling remote could be used for wake-up computer then feel free to
do that.
regards
Antti
--
http://palosaari.fi/
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
