[0/3] drm/sun4i: dsi: misc timing fixes

Message

Icenowy Zheng Oct. 1, 2019, 8:02 a.m. UTC

This patchset fixes some portion of timing calculation in sun6i_mipi_dsi
driver according to the BSP driver.

Two of the patches are reverting, one is fixing some misread of the BSP
source code, another is fixing a wrong refactor that actually breaks the
formula.

The other non-reverting patch is fixing a porch error which is usually
seen in the original driver commit. Most of porch errors are then fixed,
but this one gets ignored.

By applying these patches, several DSI panels are tested to be driven
properly by the timing provided by the vendor, including the LCD panel
of PinePhone "Don't Be Evil" DevKit, the final PinePhone panel and the
panel on PineTab. Without these patches they need dirty timing hacks to
work.

Icenowy Zheng (3):
  Revert "drm/sun4i: dsi: Change the start delay calculation"
  drm/sun4i: dsi: fix DRQ calculation
  Revert "drm/sun4i: dsi: Rework a bit the hblk calculation"

 drivers/gpu/drm/sun4i/sun6i_mipi_dsi.c | 16 +++++-----------
 1 file changed, 5 insertions(+), 11 deletions(-)

Comments

Maxime Ripard Oct. 2, 2019, 10:36 a.m. UTC | #1

Hi,

On Tue, Oct 01, 2019 at 04:02:50PM +0800, Icenowy Zheng wrote:
> This patchset fixes some portion of timing calculation in sun6i_mipi_dsi
> driver according to the BSP driver.
>
> Two of the patches are reverting, one is fixing some misread of the BSP
> source code, another is fixing a wrong refactor that actually breaks the
> formula.
>
> The other non-reverting patch is fixing a porch error which is usually
> seen in the original driver commit. Most of porch errors are then fixed,
> but this one gets ignored.
>
> By applying these patches, several DSI panels are tested to be driven
> properly by the timing provided by the vendor, including the LCD panel
> of PinePhone "Don't Be Evil" DevKit, the final PinePhone panel and the
> panel on PineTab. Without these patches they need dirty timing hacks to
> work.

Thanks for going after that issue. Can you provide references to the
BSP on the various patches?

Ideally, having the panel drivers, and the panel datasheet would help.

Thanks!
Maxime

PS: where can we get one of those devices?

Icenowy Zheng Oct. 3, 2019, 3:58 a.m. UTC | #2

在 2019-10-02三的 12:36 +0200，Maxime Ripard写道：
> Hi,
> 
> On Tue, Oct 01, 2019 at 04:02:50PM +0800, Icenowy Zheng wrote:
> > This patchset fixes some portion of timing calculation in
> > sun6i_mipi_dsi
> > driver according to the BSP driver.
> > 
> > Two of the patches are reverting, one is fixing some misread of the
> > BSP
> > source code, another is fixing a wrong refactor that actually
> > breaks the
> > formula.
> > 
> > The other non-reverting patch is fixing a porch error which is
> > usually
> > seen in the original driver commit. Most of porch errors are then
> > fixed,
> > but this one gets ignored.
> > 
> > By applying these patches, several DSI panels are tested to be
> > driven
> > properly by the timing provided by the vendor, including the LCD
> > panel
> > of PinePhone "Don't Be Evil" DevKit, the final PinePhone panel and
> > the
> > panel on PineTab. Without these patches they need dirty timing
> > hacks to
> > work.
> 
> Thanks for going after that issue. Can you provide references to the
> BSP on the various patches?

For patch 1: [1] for setting delay 1 in DSI controller, [2] for setting
real delay in TCON controller.

For patch 2: [3]

Patch 3 is reverting a breaking change, so I didn't check it in the
BSP. It can be verified by mathmatical calculation.

[1] 
https://github.com/ayufan-pine64/linux-pine64/blob/my-hacks-1.2-with-drm/drivers/video/sunxi/disp2/disp/de/lowlevel_sun50iw1/de_dsi.c#L730

[2] 
https://github.com/ayufan-pine64/linux-pine64/blob/my-hacks-1.2-with-drm/drivers/video/sunxi/disp2/disp/de/lowlevel_sun50iw1/de_lcd.c#L369

[3] 
https://github.com/ayufan-pine64/linux-pine64/blob/my-hacks-1.2-with-drm/drivers/video/sunxi/disp2/disp/de/lowlevel_sun50iw1/de_dsi.c#L780

> 
> Ideally, having the panel drivers, and the panel datasheet would
> help.
> 
> Thanks!
> Maxime
> 
> PS: where can we get one of those devices?