On Tue, 2022-06-28 at 10:38 +0800, CK Hu wrote:
> On Tue, 2022-06-28 at 10:28 +0800, Rex-BC Chen wrote:
> > On Tue, 2022-06-28 at 10:15 +0800, CK Hu wrote:
> > > Hi, Bo-Chen:
> > >
> > > On Fri, 2022-06-24 at 11:09 +0800, Bo-Chen Chen wrote:
> > > > Dp_intf supports YUV422 as output format. In MT8195 Chrome
> > > > project,
> > > > YUV422 output format is used for 4K resolution.
> > > >
> > > > To support this, it is also needed to support color format
> > > > transfer.
> > > > Color format transfer is a new feature for both dpi and dpintf
> > > > of
> > > > MT8195.
> > > >
> > > > The input format could be RGB888 and output format for dp_intf
> > > > should
> > > > be
> > > > YUV422. Therefore, we add a mtk_dpi_matrix_sel() helper to
> > > > update
> > > > the
> > > > DPI_MATRIX_SET register depending on the color format.
> > > >
> > > > Signed-off-by: Guillaume Ranquet <granquet@xxxxxxxxxxxx>
> > > > Signed-off-by: Bo-Chen Chen <rex-bc.chen@xxxxxxxxxxxx>
> > > > Reviewed-by: AngeloGioacchino Del Regno <
> > > > angelogioacchino.delregno@xxxxxxxxxxxxx>
> > > > ---
> > > > drivers/gpu/drm/mediatek/mtk_dpi.c | 34
> > > > ++++++++++++++++++++++++-
> > > > drivers/gpu/drm/mediatek/mtk_dpi_regs.h | 3 +++
> > > > 2 files changed, 36 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/gpu/drm/mediatek/mtk_dpi.c
> > > > b/drivers/gpu/drm/mediatek/mtk_dpi.c
> > > > index 9e4250356342..438bf3bc5e4a 100644
> > > > --- a/drivers/gpu/drm/mediatek/mtk_dpi.c
> > > > +++ b/drivers/gpu/drm/mediatek/mtk_dpi.c
> > > > @@ -128,6 +128,7 @@ struct mtk_dpi_yc_limit {
> > > > * @num_output_fmts: Quantity of supported output formats.
> > > > * @is_ck_de_pol: Support CK/DE polarity.
> > > > * @swap_input_support: Support input swap function.
> > > > + * @color_fmt_trans_support: Enable color format transfer.
> > > > * @dimension_mask: Mask used for HWIDTH, HPORCH, VSYNC_WIDTH
> > > > and
> > > > VSYNC_PORCH
> > > > * (no shift).
> > > > * @hvsize_mask: Mask of HSIZE and VSIZE mask (no shift).
> > > > @@ -144,6 +145,7 @@ struct mtk_dpi_conf {
> > > > u32 num_output_fmts;
> > > > bool is_ck_de_pol;
> > > > bool swap_input_support;
> > > > + bool color_fmt_trans_support;
> > > > u32 dimension_mask;
> > > > u32 hvsize_mask;
> > > > u32 channel_swap_shift;
> > > > @@ -412,6 +414,31 @@ static void
> > > > mtk_dpi_config_disable_edge(struct
> > > > mtk_dpi *dpi)
> > > > mtk_dpi_mask(dpi, dpi->conf->reg_h_fre_con, 0,
> > > > EDGE_SEL_EN);
> > > > }
> > > >
> > > > +static void mtk_dpi_matrix_sel(struct mtk_dpi *dpi,
> > > > + enum mtk_dpi_out_color_format
> > > > format)
> > > > +{
> > > > + u32 matrix_sel = 0;
> > > > +
> > > > + if (!dpi->conf->color_fmt_trans_support) {
> > > > + dev_info(dpi->dev, "matrix_sel is not
> > > > supported.\n");
> > > > + return;
> > > > + }
> > > > +
> > > > + switch (format) {
> > > > + case MTK_DPI_COLOR_FORMAT_YCBCR_422:
> > > > + case MTK_DPI_COLOR_FORMAT_YCBCR_422_FULL:
> > > > + case MTK_DPI_COLOR_FORMAT_YCBCR_444:
> > > > + case MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL:
> > >
> > > I think the transform formula are different for full range and
> > > non-
> > > full
> > > range. Please make sure '0x2' is for full range or non-full
> > > range.
> > > If
> > > you are not sure, you could provide the transform matrix of '0x2'
> > > so
> > > we
> > > could find out it's full or non-full.
> > >
> > > > + case MTK_DPI_COLOR_FORMAT_XV_YCC:
> > > > + if (dpi->mode.hdisplay <= 720)
> > > > + matrix_sel = 0x2;
> > >
> > > Symbolize '0x2'.
> > >
> > > > + break;
> > > > + default:
> > > > + break;
> > > > + }
> > > > + mtk_dpi_mask(dpi, DPI_MATRIX_SET, matrix_sel,
> > > > INT_MATRIX_SEL_MASK);
> > > > +}
> > > > +
> > > > static void mtk_dpi_config_color_format(struct mtk_dpi *dpi,
> > > > enum
> > > > mtk_dpi_out_color_format
> > > > format)
> > > > {
> > > > @@ -419,6 +446,7 @@ static void
> > > > mtk_dpi_config_color_format(struct
> > > > mtk_dpi *dpi,
> > > > (format == MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL)) {
> > > > mtk_dpi_config_yuv422_enable(dpi, false);
> > > > mtk_dpi_config_csc_enable(dpi, true);
> > > > + mtk_dpi_matrix_sel(dpi, format);
> > >
> > > Why mt8173 support MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL but it
> > > does
> > > not
> > > call mtk_dpi_matrix_sel()? It seems that mt8173 also need to call
> > > mtk_dpi_matrix_sel() but lost and this patch looks like a bug fix
> > > for
> > > all SoC DPI driver.
> > >
> > > Regards,
> > > CK
> > >
> >
> > Hello CK,
> >
> > MT8173 does not support MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL as
> > output
> > format, the output format is:
> >
> > static const u32 mt8173_output_fmts[] = {
> > MEDIA_BUS_FMT_RGB888_1X24,
> > };
> >
> > or do I misunderstand?
>
> In the first patch [1], it define
>
> +enum mtk_dpi_out_color_format {
> + MTK_DPI_COLOR_FORMAT_RGB,
> + MTK_DPI_COLOR_FORMAT_RGB_FULL,
> + MTK_DPI_COLOR_FORMAT_YCBCR_444,
> + MTK_DPI_COLOR_FORMAT_YCBCR_422,
> + MTK_DPI_COLOR_FORMAT_XV_YCC,
> + MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL,
> + MTK_DPI_COLOR_FORMAT_YCBCR_422_FULL
> +};
>
> and this function also process MTK_DPI_COLOR_FORMAT_YCBCR_444,
> MTK_DPI_COLOR_FORMAT_YCBCR_444_FULL, MTK_DPI_COLOR_FORMAT_YCBCR_422,
> and MTK_DPI_COLOR_FORMAT_YCBCR_422_FULL. So I think it want to
> process
> output YUV but the caller of mtk_dpi_config_color_format() just pass
> RGB into this function. If mt8173 does not support YUV output, I
> think
> you should remove YUV processing in this function first, and then add
> back YUV processing in this function.
>
> [1]
>
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/mediatek/mtk_dpi.c?h=v5.19-rc4&id=9e629c17aa8d7a75b8c1d99ed42892cd8ba7cdc4
>
> Regards,
> CK
>
Hello CK,
I don't think it should be remove. After all, it is accepted and from
[1], we can see it assgin output format always as RGB.
+ dpi->color_format = MTK_DPI_COLOR_FORMAT_RGB;
After that, it also added patch of supporting changing output
format[2].
We have support output as YUV422 now. I think it's ok to just add this.
If we remove them and add them back, I think it is a little bit
redundant. And I also can add a commit message like "output format
YUV422 is not support for previous MediaTek SoCs. MT8195 supports
output format as YUV422.."
What do you think?
[2]:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/gpu/drm/mediatek/mtk_dpi.c?h=v5.19-rc4&id=be63f6e8601ff21139da93623754717e92cbd8db
BRs,
Bo-Chen
> >
> > BRs,
> > Bo-Chen
> >
> > > > if (dpi->conf->swap_input_support)
> > > > mtk_dpi_config_swap_input(dpi, false);
> > > > mtk_dpi_config_channel_swap(dpi,
> > > > MTK_DPI_OUT_CHANNEL_SWAP_BGR);
> > > > @@ -426,6 +454,7 @@ static void
> > > > mtk_dpi_config_color_format(struct
> > > > mtk_dpi *dpi,
> > > > (format ==
> > > > MTK_DPI_COLOR_FORMAT_YCBCR_422_FULL)) {
> > > > mtk_dpi_config_yuv422_enable(dpi, true);
> > > > mtk_dpi_config_csc_enable(dpi, true);
> > > > + mtk_dpi_matrix_sel(dpi, format);
> > > > if (dpi->conf->swap_input_support)
> > > > mtk_dpi_config_swap_input(dpi, true);
> > > > else
> > > > @@ -673,7 +702,10 @@ static int
> > > > mtk_dpi_bridge_atomic_check(struct
> > > > drm_bridge *bridge,
> > > > dpi->bit_num = MTK_DPI_OUT_BIT_NUM_8BITS;
> > > > dpi->channel_swap = MTK_DPI_OUT_CHANNEL_SWAP_RGB;
> > > > dpi->yc_map = MTK_DPI_OUT_YC_MAP_RGB;
> > > > - dpi->color_format = MTK_DPI_COLOR_FORMAT_RGB;
> > > > + if (out_bus_format == MEDIA_BUS_FMT_YUYV8_1X16)
> > > > + dpi->color_format =
> > > > MTK_DPI_COLOR_FORMAT_YCBCR_422_FULL;
> > > > + else
> > > > + dpi->color_format = MTK_DPI_COLOR_FORMAT_RGB;
> > > >
> > > > return 0;
> > > > }
> > > > diff --git a/drivers/gpu/drm/mediatek/mtk_dpi_regs.h
> > > > b/drivers/gpu/drm/mediatek/mtk_dpi_regs.h
> > > > index 3a02fabe1662..cca0dccb84a2 100644
> > > > --- a/drivers/gpu/drm/mediatek/mtk_dpi_regs.h
> > > > +++ b/drivers/gpu/drm/mediatek/mtk_dpi_regs.h
> > > > @@ -217,4 +217,7 @@
> > > >
> > > > #define EDGE_SEL_EN BIT(5)
> > > > #define H_FRE_2N BIT(25)
> > > > +
> > > > +#define DPI_MATRIX_SET 0xB4
> > > > +#define INT_MATRIX_SEL_MASK GENMASK(4, 0)
> > > > #endif /* __MTK_DPI_REGS_H */
> > >
> > >
> >
> >
>
>
