Re: [PATCH] serial: port: Don't suspend if the port is still busy

[Tony Lindgren]

* Yicong Yang <yangyicong@xxxxxxxxxx> [240205 08:55]:
> On 2024/2/5 14:51, Tony Lindgren wrote:
> > Can you please confirm if this still happens also with commit 6f699743aebf
> > ("serial: core: Fix runtime PM handling for pending tx")? It adds a check
> > for -EINPROGRESS.
> 
> Tested nagetive on latest v6.8-rc3. Paste the current code snippet below in __uart_start():

OK thanks for confirming it.

> In our issue case, the dev->power.runtime_status == RPM_RESUMING as analyzed in
> commit. So we cannot pass the pm_runtime_active() check and the chars will still
> be pending.

OK

> Do you mean something like below?
> 
> static int serial_port_runtime_suspend(struct device *dev)
> {
> 	struct serial_port_device *port_dev = to_serial_base_port_device(dev);
> 	struct uart_port *port;
> 	unsigned long flags;
> 	int ret = 0;
> 
> 	port = port_dev->port;
> 
> 	if (port->flags & UPF_DEAD)
> 		return ret;
> 
> 	spin_lock_irqsave(&port->lock, flags);
> 	if (__serial_port_busy(port)) {
> 		port->ops->start_tx(port);
> 		ret = -EBUSY;
> 	}
> 	spin_unlock_irqrestore(&port->lock, flags);
> 
> 	return ret;
> }

Yes the above should work.

> If so will the port fail to suspend after flushing the pending chars? Considering
> underlay driver like amba-pl011 doesn't implement runtime power management, does
> anyone will get the port into suspend routine later? I'm not quite sure about it.

Hmm yeah you may need to also call pm_runtime_mark_last_busy() to
ensure the port gets idled later on. Not sure if PM runtime core does that for
you on returning -EBUSY, worth checking it :)

The PM runtime hierarchy will block the serial port controller driver from
suspending, so the port drivers won't runtime suspend.

> In the patch's implementation the pending chars will be flushed in runtime_resume()
> callback and rpm_resume() will try to call rpm_idle() later.

On serial_port_runtime_suspend() the serial port controller will be active, so
you can call start_tx() directly.

Regards,

Tony