Re: [PATCH] drm/panel: ilitek-ili9805: Use _multi variants

[Doug Anderson]

Hi,

On Fri, Jul 17, 2026 at 3:12 AM Michail Tatas <michail.tatas@xxxxxxxxx> wrote:
>
> On Thu, Jul 16, 2026 at 04:31:53PM -0700, Doug Anderson wrote:
>
> Hello Doug, thanks a lot for the review!
>
> > Hi,
> >
> > On Thu, Jul 16, 2026 at 3:09 PM Michail Tatas <michail.tatas@xxxxxxxxx> wrote:
> > >
> > > @@ -159,36 +159,23 @@ static int ili9805_power_off(struct ili9805 *ctx)
> > >
> > >  static int ili9805_activate(struct ili9805 *ctx)
> > >  {
> > > -       struct mipi_dsi_device *dsi = ctx->dsi;
> > > -       struct device *dev = &dsi->dev;
> > > -       int i, ret;
> > > +       struct mipi_dsi_multi_context dsi_ctx = { .dsi = ctx->dsi };
> > > +       int i;
> > >
> > >         for (i = 0; i < ctx->desc->init_length; i++) {
> > >                 const struct ili9805_instr *instr = &ctx->desc->init[i];
> > >
> > > -               ret = mipi_dsi_dcs_write_buffer(ctx->dsi, instr->data, instr->len);
> > > -               if (ret < 0)
> > > -                       return ret;
> > > +               mipi_dsi_dcs_write_buffer_multi(&dsi_ctx, instr->data, instr->len);
> > >
> > >                 if (instr->delay > 0)
> > > -                       msleep(instr->delay);
> > > -       }
> >
> > What you've done is an improvement, but it's not all the way there.
> > Specifically, we'd really want to get rid of the whole "struct
> > ili9805_instr" type and instead each panel should have an init
> > function. The "struct ili9805_desc" should have a pointer to the init
> > function instead of a pointer to the init data.
>
> I am more than happy to do these changes as well. My only concern is
> that I do not have hardware to test the panel. Are compiling and static
> analysis sufficient tests for these changes ?

That is inevitable with these cleanups. We're doing our best with
static analysis and code review. If someone comes along later and
finds that something broke, we'll fix it.

-Doug