Re: [PATCH v1] driver core: Add state_synced sysfs file for devices that support it

[Saravana Kannan]

On Wed, May 13, 2020 at 1:22 AM Greg Kroah-Hartman
<gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
>
> On Tue, May 12, 2020 at 06:34:15PM -0700, Saravana Kannan wrote:
> > This can be used to check if a device supports sync_state() callbacks
> > and therefore keeps resources left on by the bootloader enabled till all
> > its consumers have probed.
> >
> > This can also be used to check if sync_state() has been called for a
> > device or whether it is still trying to keep resources enabled because
> > they were left enabled by the bootloader and all its consumers haven't
> > probed yet.
> >
> > Signed-off-by: Saravana Kannan <saravanak@xxxxxxxxxx>
> > ---
> >  .../ABI/testing/sysfs-devices-state_synced    | 24 +++++++++++++++++++
> >  drivers/base/dd.c                             | 16 +++++++++++++
> >  2 files changed, 40 insertions(+)
> >  create mode 100644 Documentation/ABI/testing/sysfs-devices-state_synced
> >
> > diff --git a/Documentation/ABI/testing/sysfs-devices-state_synced b/Documentation/ABI/testing/sysfs-devices-state_synced
> > new file mode 100644
> > index 000000000000..0c922d7d02fc
> > --- /dev/null
> > +++ b/Documentation/ABI/testing/sysfs-devices-state_synced
> > @@ -0,0 +1,24 @@
> > +What:                /sys/devices/.../state_synced
> > +Date:                May 2020
> > +Contact:     Saravana Kannan <saravanak@xxxxxxxxxx>
> > +Description:
> > +             The /sys/devices/.../state_synced attribute is only present for
> > +             devices whose bus types or driver provides the .sync_state()
> > +             callback. The number read from it (0 or 1) reflects the value
> > +             of the device's 'state_synced' field. A value of 0 means the
> > +             .sync_state() callback hasn't been called yet. A value of 1
> > +             means the .sync_state() callback has been called.
> > +
> > +             Generally, if a device has sync_state() support and has some of
> > +             the resources it provides enabled at the time the kernel starts
> > +             (Eg: enabled by hardware reset or bootloader or anything that
> > +             run before the kernel starts), then it'll keep those resources
> > +             enabled and in a state that's compatible with the state they
> > +             were in at the start of the kernel. The device will stop doing
> > +             this only when the sync_state() callback has been called --
> > +             which happens only when all its consumer devices are registered
> > +             and have probed successfully. Resources that were left disabled
> > +             at the time the kernel starts are not affected or limited in
> > +             any way by sync_state() callbacks.
> > +
> > +
> > diff --git a/drivers/base/dd.c b/drivers/base/dd.c
> > index 48ca81cb8ebc..72599436ae84 100644
> > --- a/drivers/base/dd.c
> > +++ b/drivers/base/dd.c
> > @@ -458,6 +458,13 @@ static void driver_deferred_probe_add_trigger(struct device *dev,
> >               driver_deferred_probe_trigger();
> >  }
> >
> > +static ssize_t state_synced_show(struct device *dev,
> > +                              struct device_attribute *attr, char *buf)
> > +{
> > +     return sprintf(buf, "%u\n", dev->state_synced);
> > +}
> > +static DEVICE_ATTR_RO(state_synced);
> > +
> >  static int really_probe(struct device *dev, struct device_driver *drv)
> >  {
> >       int ret = -EPROBE_DEFER;
> > @@ -531,9 +538,16 @@ static int really_probe(struct device *dev, struct device_driver *drv)
> >               goto dev_groups_failed;
> >       }
> >
> > +     if (dev_has_sync_state(dev) &&
> > +         device_create_file(dev, &dev_attr_state_synced)) {
> > +             dev_err(dev, "state_synced sysfs add failed\n");
> > +             goto dev_sysfs_state_synced_failed;
> > +     }
>
> Why not add this to the groups above this and only enable it if needed
> at runtime?

Those groups above seem to be driver specific groups. Looking at the
code, some drivers seem to be setting them. Also, this attribute can
only be decided after a driver has successfully bound to the device
because dev_has_sync_state() has to check the bus and the driver for
sync_state() support.

> The is_visible() callback should be what you need to use here.

If this is an attribute specific property, it might work. I'll take a look.

>  That
> will save you lots of housekeeping as well as properly remove the
> attribute when the device is removed from the system (which you didn't
> explicitly do in this patch...)

I had a nagging feeling I was missing something. Duh! I'll make sure I
do proper clean up in v2.

Thanks for the review.

-Saravana