Re: [PATCHv2 2/5] driver: serial: omap: prevent runtime PM for "no_console_suspend"

[Sourav Poddar <sourav.poddar@ti.com>]

Hi Grygorii,
On Monday 22 April 2013 08:18 PM, Grygorii Strashko wrote:
> On 04/22/2013 04:43 PM, Sourav Poddar wrote:
>> The driver manages "no_console_suspend" by preventing runtime PM
>> during the suspend path, which forces the console UART to stay awake.
>>
>> Signed-off-by: Sourav Poddar <sourav.poddar@ti.com>
>> ---
>>   drivers/tty/serial/omap-serial.c |    5 ++++-
>>   1 files changed, 4 insertions(+), 1 deletions(-)
>>
>> diff --git a/drivers/tty/serial/omap-serial.c 
>> b/drivers/tty/serial/omap-serial.c
>> index 08332f3..640b14e 100644
>> --- a/drivers/tty/serial/omap-serial.c
>> +++ b/drivers/tty/serial/omap-serial.c
>> @@ -1582,7 +1582,7 @@ static int serial_omap_runtime_suspend(struct 
>> device *dev)
>>       struct uart_omap_port *up = dev_get_drvdata(dev);
>>       struct omap_uart_port_info *pdata = dev->platform_data;
>>   -    if (!up)
>> +    if (!up || (!console_suspend_enabled && uart_console(&up->port)))
>>           return -EINVAL;
> Hi Sourav,
> No ) You will block Runtime PM for console UART forever, but instead
> it need to be blocked only during suspend - autosuspend should 
> continue working.
> But this will be not easy, again, -
> because System suspend isn't synchronized with Runtime PM (I mean,
> serial_omap_suspend/resume() may be called from one thread and
> serial_omap_runtime_suspend/resume() from another at same time).
> And now, serial_omap_suspend() callback is the only one place where you
> can detect that system is going to sleep.
>
Yes, got this point.
As Kevin has already suggested a way to get around this, I will fix this 
portion
in the next version.
> Personally, i don't believe in such approach (my experiences from K3.4 
> said me
> that there will be more problems than benefits).
>
> And, I like combination of "no_console_suspend" in bootargs +
> "ti,no_idle_on_suspend" for console UART in DT, because 1) it's debug 
> option and 2) until
> smth. will be decided about OMAP OCP Bus it can be used.
>
> It's just my opinion.
>
> Regards,
> -grygorii
>>         if (!pdata)
>> @@ -1614,6 +1614,9 @@ static int serial_omap_runtime_resume(struct 
>> device *dev)
>>         int loss_cnt = serial_omap_get_context_loss_count(up);
>>   +    if (!console_suspend_enabled && uart_console(&up->port))
>> +        return -EINVAL;
>> +
>>       if (loss_cnt < 0) {
>>           dev_err(dev, "serial_omap_get_context_loss_count failed : 
>> %d\n",
>>               loss_cnt);
>