Re: [PATCH v4 2/5] nvmem: add support for the write-protect pin

[Bartosz Golaszewski]

czw., 30 sty 2020 o 09:06 Geert Uytterhoeven <geert@xxxxxxxxxxxxxx> napisaÅ(a):
>
> Hi Khouloud,
>
> On Tue, Jan 7, 2020 at 10:30 AM Khouloud Touil <ktouil@xxxxxxxxxxxx> wrote:
> > The write-protect pin handling looks like a standard property that
> > could benefit other users if available in the core nvmem framework.
> >
> > Instead of modifying all the memory drivers to check this pin, make
> > the NVMEM subsystem check if the write-protect GPIO being passed
> > through the nvmem_config or defined in the device tree and pull it
> > low whenever writing to the memory.
> >
> > There was a suggestion for introducing the gpiodesc from pdata, but
> > as pdata is already removed it could be replaced by adding it to
> > nvmem_config.
> >
> > Reference: https://lists.96boards.org/pipermail/dev/2018-August/001056.html
> >
> > Signed-off-by: Khouloud Touil <ktouil@xxxxxxxxxxxx>
> > Reviewed-by: Linus Walleij <linus.walleij@xxxxxxxxxx>
> > Acked-by: Srinivas Kandagatla <srinivas.kandagatla@xxxxxxxxxx>
>
> Thanks for your patch!
>
> > --- a/drivers/nvmem/core.c
> > +++ b/drivers/nvmem/core.c
> > @@ -15,6 +15,7 @@
> >  #include <linux/module.h>
> >  #include <linux/nvmem-consumer.h>
> >  #include <linux/nvmem-provider.h>
> > +#include <linux/gpio/consumer.h>
> >  #include <linux/of.h>
> >  #include <linux/slab.h>
> >  #include "nvmem.h"
> > @@ -54,8 +55,14 @@ static int nvmem_reg_read(struct nvmem_device *nvmem, unsigned int offset,
> >  static int nvmem_reg_write(struct nvmem_device *nvmem, unsigned int offset,
> >                            void *val, size_t bytes)
> >  {
> > -       if (nvmem->reg_write)
> > -               return nvmem->reg_write(nvmem->priv, offset, val, bytes);
> > +       int ret;
> > +
> > +       if (nvmem->reg_write) {
> > +               gpiod_set_value_cansleep(nvmem->wp_gpio, 0);
> > +               ret = nvmem->reg_write(nvmem->priv, offset, val, bytes);
> > +               gpiod_set_value_cansleep(nvmem->wp_gpio, 1);
> > +               return ret;
> > +       }
> >
> >         return -EINVAL;
> >  }
> > @@ -338,6 +345,14 @@ struct nvmem_device *nvmem_register(const struct nvmem_config *config)
> >                 kfree(nvmem);
> >                 return ERR_PTR(rval);
> >         }
> > +       if (config->wp_gpio)
> > +               nvmem->wp_gpio = config->wp_gpio;
> > +       else
> > +               nvmem->wp_gpio = gpiod_get_optional(config->dev, "wp",
> > +                                                   GPIOD_OUT_HIGH);
>
> Shouldn't this GPIO be released in nvmem_release(), by calling gpiod_put()?
>

Hi Geert,

Khouloud already sent out a patch but I think it still doesn't fix all
the problems.

While we should call gpiod_put() for the descs we request - we must
not do it for the desc we get over the config structure. Unless... we
make descs reference counted with kref and add gpiod_ref() helper.
That way we could increase the reference counter in the upper branch
of the if and not do it in the lower. Calling gpiod_put() would
internally call kref_put(). Does it make sense? I think that a
function that's called gpiod_put() but doesn't really use reference
counting is misleading anyway.

> Once that's implemented, I assume it will be auto-released on registration
> failure by the call to put_device()?

No, I think this is another leak - why would put_device() lead to
freeing any resources? Am I missing something?

Bart