Re: [PATCH] drivers: core: clear wake irq in device_unbind_cleanup

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Tue, Nov 12, 2024 at 01:09:19PM +0000, Peng Fan wrote:
> Hi Greg
> 
> > Subject: Re: [PATCH] drivers: core: clear wake irq in
> > device_unbind_cleanup
> > 
> > On Mon, Nov 11, 2024 at 05:21:30PM +0800, Peng Fan (OSS) wrote:
> > > From: Peng Fan <peng.fan@nxp.com>
> > >
> > > With dev_pm_clear_wake_irq in device_unbind_cleanup, there is no
> > need
> > > to invoke dev_pm_clear_wake_irq in driver remove hook explicitly.
> > >
> > > Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
> > > Cc: Rob Herring <robh@kernel.org>
> > > Cc: Ulf Hansson <ulf.hansson@linaro.org>
> > > Signed-off-by: Peng Fan <peng.fan@nxp.com>
> > > ---
> > >  drivers/base/dd.c | 2 ++
> > >  1 file changed, 2 insertions(+)
> > >
> > > diff --git a/drivers/base/dd.c b/drivers/base/dd.c index
> > > f0e4b4aba885..ea3a871bdd11 100644
> > > --- a/drivers/base/dd.c
> > > +++ b/drivers/base/dd.c
> > > @@ -26,6 +26,7 @@
> > >  #include <linux/wait.h>
> > >  #include <linux/async.h>
> > >  #include <linux/pm_runtime.h>
> > > +#include <linux/pm_wakeirq.h>
> > >  #include <linux/pinctrl/devinfo.h>
> > >  #include <linux/slab.h>
> > >
> > > @@ -556,6 +557,7 @@ static void device_unbind_cleanup(struct
> > device *dev)
> > >  		dev->pm_domain->dismiss(dev);
> > >  	pm_runtime_reinit(dev);
> > >  	dev_pm_set_driver_flags(dev, 0);
> > > +	dev_pm_clear_wake_irq(dev);
> > 
> > I don't understand, you say you don't need to invoke it, yet you are
> > calling it here.
> 
> I mean not need to invoke it in driver.remove hook. With this patch, we
> could remove
> https://elixir.bootlin.com/linux/v6.11.7/source/drivers/input/touchscreen/ti_am335x_tsc.c#L498
> and same to other drivers.

But you did not say that, and you would need to make this as part of a
series.

Also, are you sure that ll drivers want to clear this irq flag?  What is
wrong with just doing it explicitly in the drivers that need it?

> > 
> > What commit id does this fix? 
> 
> I am thinking to take this as a improvement, with core code
> has this, the various drivers no need explicitly invoke it
> in their own driver remove hook.
> 
>  And what bug is this resolving?  What
> > drivers are broken without this?
> 
> See here:
> https://lore.kernel.org/all/ZymxvLMkkktRoCXZ@google.com/

Again, this seems to be a per-driver thing.  What do you break if you
attempt to do this for all drivers?  What about drivers that share irqs?

How was this tested?  On what platfoms?

thanks,

greg k-h