Re: [v2 0/2] Add driver for BOE tv110c9m-ll3 panel

[Doug Anderson]

Hi,

On Thu, Aug 26, 2021 at 10:17 AM Doug Anderson <dianders@xxxxxxxxxx> wrote:
>
> Hi,
>
> On Mon, Aug 23, 2021 at 6:41 PM yangcong
> <yangcong5@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> >
> > Hi Sam,
> >   This driver support boe tv1110c9m-ll3 and inx hj110iz-01a panel.The IC chip is used NT36523, which is a new IC.
> >
> > 1: panel-boe-tv101wum-nl6.c driver cannot meet the timing requirements of the current two panel.
> >
> > 2: The screen cannot be work in HS mode, panel-boe-tv101wum-nl6.c will switch to HS mode when boe_panel_enter_sleep_mode.
> >
> > static int boe_panel_enter_sleep_mode(struct boe_panel *boe)
> > {
> >         struct mipi_dsi_device *dsi = boe->dsi;
> >         int ret;
> >
> >         dsi->mode_flags &= ~MIPI_DSI_MODE_LPM;
> >         ...
> >         ...
> > }
>
> It's really up to Sam how he wants to proceed here, but certainly you
> could support things with the existing driver even if there are
> differences. In general you can add more things to the `struct
> panel_desc` in the driver and then make them different for your panel.
> Look, for instance, at `discharge_on_disable`. Not all panels
> supported by this driver do that, so you could support your "cannot
> work in HS mode" in a similar way.
>
> For the timings, you could also add another bit to the `struct
> panel_desc` to select a different delay for your panel or (if it's
> just a small delay) you could just increase it across the board. I
> guess you need a 10 ms delay instead of a 5 ms delay in probe? I'd
> just make it 10 ms across the board and call it done. Similarly looks
> like something needs .5 ms => 1 ms. Again, this is likely fine across
> the board for all panels.
>
> Unless Sam comes back and says "no, wait, keep it two drivers!" then
> I'd suggest that you post a new version that works as Sam suggests. In
> the worst case if having it combined into one driver looks too ugly
> then we can always go back to a split driver.

Breadcrumbs: it looks like yangcong has attempted this in the
confusingly numbered v3 of his other patch series. See:

https://lore.kernel.org/r/20210827082407.101053-4-yangcong5@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

-Doug