lkml.org 
[lkml]   [2020]   [Jul]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH v18 3/3] Input: new da7280 haptic driver
Date
Hello Dmitry and Uwe,

Wednesday, July 29, 2020 3:37 PM, Dmitry Torokhov wrote:

> On Wed, Jul 29, 2020 at 11:59:40AM +0900, Roy Im wrote:
> > Adds support for the Dialog DA7280 LRA/ERM Haptic Driver with multiple
> > mode and integrated waveform memory and wideband support.
> > It communicates via an I2C bus to the device.
>
> A few questions/suggestions...
>
> >
> > Reviewed-by: Jes Sorensen <Jes.Sorensen@gmail.com>.
> >
> > Signed-off-by: Roy Im <roy.im.opensource@diasemi.com>
> >
> > ---
> > v18:
> > - Corrected comments in Kconfig
> > - Updated to preferred style for multi line comments in c file.
> > v17:
> > - fixed an issue.
> > v16:
> > - Corrected some code and updated description in Kconfig.
> > v15:
> > - Removed some defines and updated some comments.
> > v14:
> > - Updated pwm related code, alignments and comments.
> > v13:
> > - Updated some conditions in pwm function and alignments.
> > v12: No changes.
> > v11:
> > - Updated the pwm related code, comments and typo.
> > v10:
> > - Updated the pwm related function and added some comments.
> > v9:
> > - Removed the header file and put the definitions into the c file.
> > - Updated the pwm code and error logs with %pE
>
> I believe the %pE is to format an escaped buffer, you probably want to %pe (lowercase) to print errors. I am also not quite sure
> if we want to use it in cases when we have non-pointer error, or we should stick with %d as most of the kernel does.

Right, it should be %pe as you and Uwe said, Uwe suggested %pe to understand easier.. do you still prefer to stick with %d?

>
> ...
> > +
> > +/* DA7280_ACTUATOR3 (Address 0x0e) */
> > +#define DA7280_IMAX_MASK (31 << 0)
>
> We have GENMASK(h,l) macro in include/linux/bits.h that could be used here and in other mask definitions.

I will do.

>
> > +
> > + bool legacy;
> > + struct delayed_work work_duration;
> > + struct work_struct work_playback;
> > + struct work_struct work_setgain;
>
> How do we ensure that all these works do not clash with each other?
> As far as I can see we could have the "duration" work executing simultaneously with playback...

You are right, I will use use cancel_delayed_work_sync()/duration and cancel_work_sync()/playback before scheduling playback work. And same as for the work_setgain. For setgain, there is no problem to run this regardless the playback.

>
> > +static int da7280_haptics_playback(struct input_dev *dev,
> > + int effect_id, int val)
> > +{
> > + struct da7280_haptic *haptics = input_get_drvdata(dev);
> > +
> > + if (!haptics->op_mode) {
> > + dev_warn(haptics->dev,
> > + "Any effects are not uploaded yet\n");
>
> "No effects have been uploaded"?

Ok, let me update so.

>
> > + return -EPERM;
>
> I'd say EINVAL.

OK

>
> > +static DEVICE_ATTR_RW(ps_seq_id);
> > +static DEVICE_ATTR_RW(ps_seq_loop);
> > +static DEVICE_ATTR_RW(gpi_seq_id0);
> > +static DEVICE_ATTR_RW(gpi_seq_id1);
> > +static DEVICE_ATTR_RW(gpi_seq_id2);
> > +static DEVICE_ATTR_WO(patterns);
>
> Should this be a binary attribute instead of having string parsing in the kernel?

I have carefully reviewed my driver again with your comments, now the upload effect covers this attributes, so I would like to remove them and add some code more for gpi_seq_idx update.

Thanks for your comments.

Kinds regards,
Roy

\
 
 \ /
  Last update: 2020-07-29 16:10    [W:0.110 / U:0.164 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site