Linux/AM5728: PLL clock setting for 3 video outputs

I tried to swap PLL1_3 and PLL2_3 in dpi_get_clk_src_dra7xx().

However, video2 PLL still inactive.

How to enable video2 PLL in a general way?

Can you please share the diff of the changes you made and the linux kernel version you are using?

I just swap DSS_CLK_SRC_PLL1_3 and DSS_CLK_SRC_PLL2_3 in OMAP_DSS_CHANNEL_LCD2 case in function dpi_get_clk_src_dra7xx.

My linux kernel is 4.4.32 base on psdk 3.02.05.

static enum dss_clk_source dpi_get_clk_src_dra7xx(enum omap_channel channel)
{
	/*
	 * Possible clock sources:
	 * LCD1: FCK/PLL1_1/HDMI_PLL
	 * LCD2: FCK/PLL1_3/HDMI_PLL (DRA74x: PLL2_3)
	 * LCD3: FCK/PLL1_3/HDMI_PLL (DRA74x: PLL2_1)
	 */
	switch (channel) {
	case OMAP_DSS_CHANNEL_LCD:
	{
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL1_1))
			return DSS_CLK_SRC_PLL1_1;
	}
	case OMAP_DSS_CHANNEL_LCD2:
	{
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL1_3))
			return DSS_CLK_SRC_PLL2_3;
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL2_3))
			return DSS_CLK_SRC_PLL1_3;
	}
	case OMAP_DSS_CHANNEL_LCD3:
	{
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL2_1))
			return DSS_CLK_SRC_PLL2_1;
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL1_3))
			return DSS_CLK_SRC_PLL1_3;
	}
	default:
		break;
	}

	return DSS_CLK_SRC_FCK;
}

Hi Manisha,

it's my mistake
It can work with following change. 

Thank you again.


	case OMAP_DSS_CHANNEL_LCD2:
	{
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL2_1))
			return DSS_CLK_SRC_PLL2_1;
		if (dss_pll_find_by_src(DSS_CLK_SRC_PLL1_3))
			return DSS_CLK_SRC_PLL1_3;
	}

Hi manisha,

Still got error after swap PLL2_3 & PLL1_3.

In function dss_pll_write_config_type_a(),

omapdss PLL error: cannot lock DSS DPLL.

It cause dpi_set_mode fail.

Do you have any comments or suggestions for this?

Suggested changes were tested at our end with kernel v4.9, and verified that both video plls getting enabled, and vsyncs at 60fps for both LCD outputs works fine.

Did you also take care of the changes required in the dts configuration?

Definitions for a second LCD in AM5 EVM dts file is needed , and set it to use the second port and then the dpi.c need to be changed as mentioned above. 

Hi Manisha,

We will try it with kernel v4.9.

I think our dts setting is ok.

&dss {
	status = "okay";
	ports {
		#address-cells = <1>;
		#size-cells = <0>;

		port@0 {
			reg = <0>;
			dpi_out_1: endpoint {
				remote-endpoint = <&hdmi_encoder_1_in>;
				data-lines = <24>;
			};
		};
		port@1 {
			reg = <1>;
			dpi_out_2: endpoint {
				remote-endpoint = <&hdmi_encoder_2_in>;
				data-lines = <24>;
			};
		};
	};
};

Hi Manisha,

We switched our resource to do other tasks.

Currently, we use video1 PLL to drive dpi1 and dpi2 with different DIV M4 & M6.

Regards,

Hugo

Hi Hugo,

Does that mean both LCD1 and LCD2 out are working for you now? If not, I noticed that dpi_get_clk_src_dra7xx() wasn't modified in your code properly. You should try below - 

case OMAP_DSS_CHANNEL_LCD2:
{

if (dss_pll_find_by_src(DSS_CLK_SRC_PLL2_3))
return DSS_CLK_SRC_PLL2_3;
if (dss_pll_find_by_src(DSS_CLK_SRC_PLL1_3))
return DSS_CLK_SRC_PLL1_3;
}

Regards,

Manisha

Hi Manisha,

I have tried this modification, it's still not work.

video2 PLL :  Disabled

Regards,

Hugo

Hi Hugo,

Please find attached the patch that we created to test LCD1 and LCD2 along with HDMI.

0001-hack-for-am5-evm-vout1-2.patch

Since we do not have hardware with three displays, we tested the patch using kmstest application (kmstest --flip). The application source code can be found here - 

It is based on the latest 4.4 kernel. You need to change the kernel config, though, to get 3 displays: change CONFIG_DRM_OMAP_NUM_CRTCS to 3. 

When the kernel is compiled  with the patch and then ran the kmstest exmaple application, it did page flipping on all the displays -

Moving bar on the LCD and the HDMI is seen, which shows that the real displays are working. Additionally we see prints from kmstest that shows 3 displays getting fps ~60. This tells that also the third display is getting clocked correctly and getting vsyncs every 16.6ms.

Please compare the patch against your implementation and let us know.