Re: [Linuxwacom-discuss] wireless sleep timeout

[Chris B. <ch...@cn...>]

On Mon, Mar 25, 2013 at 11:28 AM, Jason Gerecke <kil...@gm...>wrote:

> On Sun, Mar 24, 2013 at 7:59 PM, Chris Bagwell <ch...@cn...>wrote:
>
>> On Sun, Mar 17, 2013 at 11:56 AM, Tamas Papp <tk...@gm...> wrote:
>>
>>> Hi,
>>>
>>> I have a Wacom Bamboo (CTH-470) which I am using with a wireless module,
>>> with Debian.  If I am not using the tablet for a while, it goes to
>>> sleep, which does not play well with xournal.  Is there a way I can
>>> control the sleep timeout, eg with xsetwacom?
>>>
>>
>> Short answer is no; there is such option.
>>
>> Longer answer is it should be possible to enhance the Linux driver to
>> adjust the sleep timer if someone was motivated enough.   Its been almost a
>> year since I looked but I recall the Windows driver exposing a setting for
>> this.  Same issue will exist for Intuos5 which shares same wireless module.
>>
>> I'd have to research again but it seemed iffy feature to implement at the
>> time.  Maybe something like you could adjust between 1 and 5 minutes and
>> that didn't seem wide enough range to me.  Maybe an on/off setting would be
>> useful though.
>>
>> Anyone else on the list find the sleep timeout to short or to long?
>>
>> Chris
>>
>>
> It looks like power-saving for the wireless module has a few
> implementation details that we'd want to be careful about exposing. For
> instance, the sleep timeout can be reliably set to any value from 5-60
> minutes. Above that, and sleep mode is disabled; below that, and sleep mode
> is reset to the hardware default of 15 minutes.
>

Ah, both 5 or 15 minutes are an extremely long time and I guess I was
thinking most people would want an on/off contoller before tweaking those
values to much.



> If a userspace program is unaware of this, it could inadvertently run into
> either of these cases while setting a timeout. When reading the current
> timeout, if the tablet happily reports numbers outside the 5-60 range then
> the kernel would need to adjust the value to match reality.
>
>
For the 5-60 range validation part, I most often see a sysfs item that is
read-only and exposes either a list of all valid possible values or 2 read
only files that list min and max range.  Something like:

/sys/*/wacom/sleep_timer   <--- read/write to both control and see current
value
/sys/*/wacom/sleep_timer_min  <--- read only - 5
/sys/*/wacom/sleep_timer_max <--- read only - 60

So that way we can support hardware with different ranges if they come to
exist and also gracefully ignore writes to sleep_timer with values below 5.

For the sleep timer OFF feature (values above 60), I'd expose that to user
different then hardware does it and ignore writes to sleep_timer above max
level as well.  I'd instead create a /sys/*/wacom/sleep file with valid
values of 0/1.

Some sysfs do use magic values such as saying sleep_max value means OFF but
I think a seperate controler would be more self documenting.

Chris