Re: [PATCH v5 2/4] iio: core: Introduce IIO configfs support

[Daniel Baluta]

On 05/04/2015 10:59 PM, Lars-Peter Clausen wrote:
> On 05/04/2015 12:50 PM, Daniel Baluta wrote:
> > This creates an IIO configfs subystem named "iio", with a default
> > "triggers"
> > group.
> >
> > Triggers group is used for handling software triggers. To create a new
> > software
> > trigger one must create a directory inside the trigger directory.
> >
> > Software trigger name MUST follow the following convention:
> >     * <trigger-type>-<trigger-name>
> > Where:
> >     * <trigger_type>, specifies the interrupt source (e.g: hrtimer)
> >     * <trigger-name>, specifies the IIO device trigger name
> >
> > Failing to follow this convention will result in an directory creation
> > error.
> >
> > E.g, assuming that hrtimer trigger type is registered with IIO software
> > trigger core:
> >
> > $ mkdir /config/iio/triggers/hrtimer-instance1
>
> Nice, short and clean. Looks pretty good. It's a bit of a shame that we
> can't have a per type directory, but if that's how configfs works I
> guess there is not much choice.
>
> [...]
> > +static struct config_group *trigger_make_group(struct config_group
> > *group,
> > +                           const char *name)
> > +{
> > +    char *type_name;
> > +    char *trigger_name;
> > +    char buf[MAX_NAME_LEN];
> > +    struct iio_sw_trigger *t;
> > +
> > +    snprintf(buf, MAX_NAME_LEN, "%s", name);
> > +
> > +    /* group name should have the form <trigger-type>-<trigger-name> */
> > +    type_name = buf;
> > +    trigger_name = strchr(buf, '-');
> > +    if (!trigger_name) {
> > +        pr_err("Unable to locate '-' in %s. Use <type>-<name>.\n", buf);
>
> Do we want to print this side channel message? Makes it pretty easy to
> spam the kernel log with a rouge application.

I think this is useful, people can get crazy for random -EINVAL errors 
:). Also the same is done in drivers/usb/gadget/configfs.c.


> > +        return ERR_PTR(-EINVAL);
> > +    }
> > +
> > +    /* replace - with \0, this nicely separates the two strings */
> > +    *trigger_name = '\0';
> > +    trigger_name++;
> > +
> > +    t = iio_sw_trigger_create(type_name, trigger_name);
> > +    if (IS_ERR(t))
> > +        return ERR_CAST(t);
> > +
> > +    config_item_set_name(&t->group.cg_item, name);
> > +
> > +    return &t->group;
> > +}
> > +
> > +static void trigger_drop_group(struct config_group *group,
> > +                   struct config_item *item)
> > +{
> > +    struct iio_sw_trigger *t = to_iio_sw_trigger(item);
> > +
> > +    if (t)
>
> t will never be NULL.

Agree, will fix in v6.
> > +        iio_sw_trigger_destroy(t);
> > +    config_item_put(item);
> > +}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/