Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Linus Walleij]

On Thu, Sep 15, 2016 at 5:52 PM, Mika Westerberg
 wrote:

> When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> driver, it is not possible to select which GPIOs to add to the IRQ domain.
> Instead it just adds all GPIOs which is not always desired. For example
> there might be GPIOs that for some reason just cannot be used as interrupts
> at all.
>
> To make this possible we add valid_mask to each gpio_chip and by default
> assume all GPIOs can be used as interrupts. Drivers can then tune this
> using clear_bit() or similar before they call gpiochip_irqchip_add().
>
> Suggested-by: Linus Walleij 
> Signed-off-by: Mika Westerberg 

Awesome, nice infrastructure.
Will apply once tglx and Marc's comments are fixed.

Yours,
Linus Walleij

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Linus Walleij]

On Thu, Sep 15, 2016 at 5:52 PM, Mika Westerberg
 wrote:

> When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> driver, it is not possible to select which GPIOs to add to the IRQ domain.
> Instead it just adds all GPIOs which is not always desired. For example
> there might be GPIOs that for some reason just cannot be used as interrupts
> at all.
>
> To make this possible we add valid_mask to each gpio_chip and by default
> assume all GPIOs can be used as interrupts. Drivers can then tune this
> using clear_bit() or similar before they call gpiochip_irqchip_add().
>
> Suggested-by: Linus Walleij 
> Signed-off-by: Mika Westerberg 

Awesome, nice infrastructure.
Will apply once tglx and Marc's comments are fixed.

Yours,
Linus Walleij

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Thomas Gleixner]

On Thu, 15 Sep 2016, Mika Westerberg wrote:
> On Thu, Sep 15, 2016 at 05:07:58PM +0100, Marc Zyngier wrote:
> > Mika,
> > 
> > On 15/09/16 16:52, Mika Westerberg wrote:
> > > When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> > > driver, it is not possible to select which GPIOs to add to the IRQ domain.
> > > Instead it just adds all GPIOs which is not always desired. For example
> > > there might be GPIOs that for some reason just cannot be used as 
> > > interrupts
> > > at all.
> > > 
> > > To make this possible we add valid_mask to each gpio_chip and by default
> > > assume all GPIOs can be used as interrupts. Drivers can then tune this
> > > using clear_bit() or similar before they call gpiochip_irqchip_add().
> > > 
> > > Suggested-by: Linus Walleij 
> > > Signed-off-by: Mika Westerberg 
> > > ---
> > >  drivers/gpio/gpiolib.c  | 32 +---
> > >  include/linux/gpio/driver.h |  1 +
> > >  2 files changed, 30 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > > index 53ff25ac66d8..d84c23b47f44 100644
> > > --- a/drivers/gpio/gpiolib.c
> > > +++ b/drivers/gpio/gpiolib.c
> > > @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> > > *data)
> > >   if (status)
> > >   goto err_remove_chip;
> > >   }
> > > +
> > > +#ifdef CONFIG_GPIOLIB_IRQCHIP
> > > + chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> > > +GFP_KERNEL);
> > 
> > Do we really want to make this a mandatory thing? In most cases, I'd
> > expect this valid_mask to have all bits set, so you might as well not
> > allocate it at all in that case (and only allocate it if you actually
> > need it).
> 
> We can make drivers to allocate it and only if set use it in core.

Make it a flag in gpio_chip which allocates it here when requested so we
don't end up with driver writers getting the alloc/free/error handling
wrong over and over.

Thanks,

tglx

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Thomas Gleixner]

On Thu, 15 Sep 2016, Mika Westerberg wrote:
> On Thu, Sep 15, 2016 at 05:07:58PM +0100, Marc Zyngier wrote:
> > Mika,
> > 
> > On 15/09/16 16:52, Mika Westerberg wrote:
> > > When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> > > driver, it is not possible to select which GPIOs to add to the IRQ domain.
> > > Instead it just adds all GPIOs which is not always desired. For example
> > > there might be GPIOs that for some reason just cannot be used as 
> > > interrupts
> > > at all.
> > > 
> > > To make this possible we add valid_mask to each gpio_chip and by default
> > > assume all GPIOs can be used as interrupts. Drivers can then tune this
> > > using clear_bit() or similar before they call gpiochip_irqchip_add().
> > > 
> > > Suggested-by: Linus Walleij 
> > > Signed-off-by: Mika Westerberg 
> > > ---
> > >  drivers/gpio/gpiolib.c  | 32 +---
> > >  include/linux/gpio/driver.h |  1 +
> > >  2 files changed, 30 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > > index 53ff25ac66d8..d84c23b47f44 100644
> > > --- a/drivers/gpio/gpiolib.c
> > > +++ b/drivers/gpio/gpiolib.c
> > > @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> > > *data)
> > >   if (status)
> > >   goto err_remove_chip;
> > >   }
> > > +
> > > +#ifdef CONFIG_GPIOLIB_IRQCHIP
> > > + chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> > > +GFP_KERNEL);
> > 
> > Do we really want to make this a mandatory thing? In most cases, I'd
> > expect this valid_mask to have all bits set, so you might as well not
> > allocate it at all in that case (and only allocate it if you actually
> > need it).
> 
> We can make drivers to allocate it and only if set use it in core.

Make it a flag in gpio_chip which allocates it here when requested so we
don't end up with driver writers getting the alloc/free/error handling
wrong over and over.

Thanks,

tglx

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Mika Westerberg]

On Thu, Sep 15, 2016 at 05:07:58PM +0100, Marc Zyngier wrote:
> Mika,
> 
> On 15/09/16 16:52, Mika Westerberg wrote:
> > When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> > driver, it is not possible to select which GPIOs to add to the IRQ domain.
> > Instead it just adds all GPIOs which is not always desired. For example
> > there might be GPIOs that for some reason just cannot be used as interrupts
> > at all.
> > 
> > To make this possible we add valid_mask to each gpio_chip and by default
> > assume all GPIOs can be used as interrupts. Drivers can then tune this
> > using clear_bit() or similar before they call gpiochip_irqchip_add().
> > 
> > Suggested-by: Linus Walleij 
> > Signed-off-by: Mika Westerberg 
> > ---
> >  drivers/gpio/gpiolib.c  | 32 +---
> >  include/linux/gpio/driver.h |  1 +
> >  2 files changed, 30 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > index 53ff25ac66d8..d84c23b47f44 100644
> > --- a/drivers/gpio/gpiolib.c
> > +++ b/drivers/gpio/gpiolib.c
> > @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> > *data)
> > if (status)
> > goto err_remove_chip;
> > }
> > +
> > +#ifdef CONFIG_GPIOLIB_IRQCHIP
> > +   chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> > +  GFP_KERNEL);
> 
> Do we really want to make this a mandatory thing? In most cases, I'd
> expect this valid_mask to have all bits set, so you might as well not
> allocate it at all in that case (and only allocate it if you actually
> need it).

We can make drivers to allocate it and only if set use it in core.

> 
> > +   if (!chip->valid_mask)
> > +   return -ENOMEM;
> 
> You really want to revise your error handling here.

Oops, right. It should be:

if (!chip->valid_mask) {
status = -ENOMEM;
goto err_remove_chip;
}

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Mika Westerberg]

On Thu, Sep 15, 2016 at 05:07:58PM +0100, Marc Zyngier wrote:
> Mika,
> 
> On 15/09/16 16:52, Mika Westerberg wrote:
> > When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> > driver, it is not possible to select which GPIOs to add to the IRQ domain.
> > Instead it just adds all GPIOs which is not always desired. For example
> > there might be GPIOs that for some reason just cannot be used as interrupts
> > at all.
> > 
> > To make this possible we add valid_mask to each gpio_chip and by default
> > assume all GPIOs can be used as interrupts. Drivers can then tune this
> > using clear_bit() or similar before they call gpiochip_irqchip_add().
> > 
> > Suggested-by: Linus Walleij 
> > Signed-off-by: Mika Westerberg 
> > ---
> >  drivers/gpio/gpiolib.c  | 32 +---
> >  include/linux/gpio/driver.h |  1 +
> >  2 files changed, 30 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> > index 53ff25ac66d8..d84c23b47f44 100644
> > --- a/drivers/gpio/gpiolib.c
> > +++ b/drivers/gpio/gpiolib.c
> > @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> > *data)
> > if (status)
> > goto err_remove_chip;
> > }
> > +
> > +#ifdef CONFIG_GPIOLIB_IRQCHIP
> > +   chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> > +  GFP_KERNEL);
> 
> Do we really want to make this a mandatory thing? In most cases, I'd
> expect this valid_mask to have all bits set, so you might as well not
> allocate it at all in that case (and only allocate it if you actually
> need it).

We can make drivers to allocate it and only if set use it in core.

> 
> > +   if (!chip->valid_mask)
> > +   return -ENOMEM;
> 
> You really want to revise your error handling here.

Oops, right. It should be:

if (!chip->valid_mask) {
status = -ENOMEM;
goto err_remove_chip;
}

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Marc Zyngier]

Mika,

On 15/09/16 16:52, Mika Westerberg wrote:
> When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> driver, it is not possible to select which GPIOs to add to the IRQ domain.
> Instead it just adds all GPIOs which is not always desired. For example
> there might be GPIOs that for some reason just cannot be used as interrupts
> at all.
> 
> To make this possible we add valid_mask to each gpio_chip and by default
> assume all GPIOs can be used as interrupts. Drivers can then tune this
> using clear_bit() or similar before they call gpiochip_irqchip_add().
> 
> Suggested-by: Linus Walleij 
> Signed-off-by: Mika Westerberg 
> ---
>  drivers/gpio/gpiolib.c  | 32 +---
>  include/linux/gpio/driver.h |  1 +
>  2 files changed, 30 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> index 53ff25ac66d8..d84c23b47f44 100644
> --- a/drivers/gpio/gpiolib.c
> +++ b/drivers/gpio/gpiolib.c
> @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> *data)
>   if (status)
>   goto err_remove_chip;
>   }
> +
> +#ifdef CONFIG_GPIOLIB_IRQCHIP
> + chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> +GFP_KERNEL);

Do we really want to make this a mandatory thing? In most cases, I'd
expect this valid_mask to have all bits set, so you might as well not
allocate it at all in that case (and only allocate it if you actually
need it).

> + if (!chip->valid_mask)
> + return -ENOMEM;

You really want to revise your error handling here.

Thanks,

M.
-- 
Jazz is not dead. It just smells funny...

Re: [PATCH 1/2] gpiolib: Add possibility to mask which GPIOs are added to IRQ domain

[Marc Zyngier]

Mika,

On 15/09/16 16:52, Mika Westerberg wrote:
> When using GPIO irqchip helpers to setup irqchip for a gpiolib based
> driver, it is not possible to select which GPIOs to add to the IRQ domain.
> Instead it just adds all GPIOs which is not always desired. For example
> there might be GPIOs that for some reason just cannot be used as interrupts
> at all.
> 
> To make this possible we add valid_mask to each gpio_chip and by default
> assume all GPIOs can be used as interrupts. Drivers can then tune this
> using clear_bit() or similar before they call gpiochip_irqchip_add().
> 
> Suggested-by: Linus Walleij 
> Signed-off-by: Mika Westerberg 
> ---
>  drivers/gpio/gpiolib.c  | 32 +---
>  include/linux/gpio/driver.h |  1 +
>  2 files changed, 30 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> index 53ff25ac66d8..d84c23b47f44 100644
> --- a/drivers/gpio/gpiolib.c
> +++ b/drivers/gpio/gpiolib.c
> @@ -1186,6 +1186,18 @@ int gpiochip_add_data(struct gpio_chip *chip, void 
> *data)
>   if (status)
>   goto err_remove_chip;
>   }
> +
> +#ifdef CONFIG_GPIOLIB_IRQCHIP
> + chip->valid_mask = kcalloc(BITS_TO_LONGS(chip->ngpio), sizeof(long),
> +GFP_KERNEL);

Do we really want to make this a mandatory thing? In most cases, I'd
expect this valid_mask to have all bits set, so you might as well not
allocate it at all in that case (and only allocate it if you actually
need it).

> + if (!chip->valid_mask)
> + return -ENOMEM;

You really want to revise your error handling here.

Thanks,

M.
-- 
Jazz is not dead. It just smells funny...