v4l2 capture: mxc_v4l_read timeout counter 0

[larsweber]

Hello,

im working on a custom board with a mx6dl and a tw9910 video decoder as a tvin source.

It looks like the tw9910 has a lot in common with the adv7180 so i used that driver as a starting point.

I removed the i2c communication from that driver and initialize the tw9910 from userspace.

The tw9910 is decoding the video fine and we verified that it sends the data with a pixelclock of 27Mhz to the CSI0 of the imx.

Hsync and Vsync are provided.

	adv7180: adv7180@21 {
	compatible = "adv,adv7180";
	reg = <0x21>;
	pinctrl-names = "default";
	pinctrl-0 = <&pinctrl_ipu1_4>;
	clocks = <&clks 201>;
	clock-names = "csi_mclk";
	DOVDD-supply = <&reg_3p3v>; /* 3.3v, enabled via 2.8 VGEN6 */
	AVDD-supply = <&reg_3p3v>;  /* 1.8v */
	DVDD-supply = <&reg_3p3v>;  /* 1.8v */
	PVDD-supply = <&reg_3p3v>;  /* 1.8v */
	pwn-gpios = <&gpio1 6 0>;
	csi_id = <0>;
	mclk = <27000000>;
	mclk_source = <0>;
	cvbs = <1>;

    };

	v4l2_cap_1 {
	compatible = "fsl,imx6q-v4l2-capture";
	ipu_id = <0>;
	csi_id = <0>;
	mclk_source = <0>;
	status = "okay";

    };

	pinctrl_ipu1_4: ipu1grp-4 {
		fsl,pins = <
			MX6QDL_PAD_CSI0_DAT12__IPU1_CSI0_DATA12	0x80000000
			MX6QDL_PAD_CSI0_DAT13__IPU1_CSI0_DATA13	0x80000000
			MX6QDL_PAD_CSI0_DAT14__IPU1_CSI0_DATA14	0x80000000
			MX6QDL_PAD_CSI0_DAT15__IPU1_CSI0_DATA15	0x80000000
			MX6QDL_PAD_CSI0_DAT16__IPU1_CSI0_DATA16	0x80000000
			MX6QDL_PAD_CSI0_DAT17__IPU1_CSI0_DATA17	0x80000000
			MX6QDL_PAD_CSI0_DAT18__IPU1_CSI0_DATA18	0x80000000
			MX6QDL_PAD_CSI0_DAT19__IPU1_CSI0_DATA19	0x80000000
			MX6QDL_PAD_CSI0_PIXCLK__IPU1_CSI0_PIXCLK   0x80000000
			MX6QDL_PAD_CSI0_MCLK__IPU1_CSI0_HSYNC	0x80000000
			MX6QDL_PAD_CSI0_VSYNC__IPU1_CSI0_VSYNC	0x80000000

			//Data Enable has to be pulled HIGH!
			MX6QDL_PAD_CSI0_DATA_EN__IPU1_CSI0_DATA_EN 0x80000000
		>;
	};

The CSI0_DATA_EN pin should be pulled high but isn't connected on the board.

So i assume that through the default configuration it is pulled high by the imx.

I also assume that the IPU1 correlates to the ipu_id = <0> and the csi to csi_id = <0>. Is that correct?

The video0 device is showing up but reading from it resolves in an error:

In MVC: mxc_v4l_open

   device name is Mxc Camera

End of mxc_v4l_open: v2f pix widthxheight 720 x 576

End of mxc_v4l_open: crop_bounds widthxheight 720 x 625

End of mxc_v4l_open: crop_defrect widthxheight 720 x 625

End of mxc_v4l_open: crop_current widthxheight 720 x 625

On Open: Input to ipu size is 720 x 625

Inside ipu_csi_init_interface with pixel_fmt=1498831189

Inside ipu_csi_init_interface: Selected CSI_SENS_CONF_DATA_FMT_YUV422_UYVY

imx-ipuv3 2400000.ipu: CSI_SENS_CONF = 0x00000A02

imx-ipuv3 2400000.ipu: CSI_ACT_FRM_SIZE = 0x027002CF

In MVC:mxc_v4l_ioctl

In MVC: mxc_v4l_do_ioctl 80685600

   case VIDIOC_QUERYCAP

cap.capabilities= 5000005In MVC:mxc_v4l_ioctl

In MVC: mxc_v4l_do_ioctl c02c563a

   case VIDIOC_CROPCAP

In MVC:mxc_v4l_ioctl

In MVC: mxc_v4l_do_ioctl 4014563c

   case VIDIOC_S_CROP

   Cropping Input to ipu size 720 x 625

In MVC:mxc_v4l_ioctl

In MVC: mxc_v4l_do_ioctl c0cc5605

   case VIDIOC_S_FMT

In MVC: mxc_v4l2_s_fmt

   type=V4L2_BUF_TYPE_VIDEO_CAPTURE

End of mxc_v4l2_s_fmt: v2f pix widthxheight 720 x 576

End of mxc_v4l2_s_fmt: crop_bounds widthxheight 720 x 625

End of mxc_v4l2_s_fmt: crop_defrect widthxheight 720 x 625

End of mxc_v4l2_s_fmt: crop_current widthxheight 720 x 625

In MVC:mxc_poll

frameRead

ERROR: v4l2 capture: mxc_v4l_read timeout counter 0

imx-ipuv3 2400000.ipu: CSI stop timeout - 5 * 10ms

read error 62, Timer expiredIn MVC:mxc_v4l_close

mxc_v4l_close: release resource

MVC: In mxc_free_frame_buf

In MVC:mxc_free_frames

I tried to change different parameters for the csi capture like: csi_param.Vsync_pol, csi_param.Hsync_pol, csi_param.data_en_pol but i still get no csi_irq_handler interrupt.

Do you have any suggestions where i could look next or what i'm doing wrong?

Thanks for you answers!

[jaimeignacioper]

Hi Lars

You should try to use an exisiting driver, you could follow next link:

linux/tw9910.c at master · spotify/linux · GitHub

Best regards.

Jaime

[larsweber]

Hi Jaime,

thank you for your reply.

I'm using parts of that driver to initialize the tw9910 (but from the userspace at the moment).

I don't want to use the whole driver as it is, because it is not compatible to the device-tree.

From the adv7180 driver i removed every i2c communication.

But i think i found one problem:

I need to set the IOMUX_GPR13 regs because my board isn't specified as compatible to fsl,imx6dl-sabresd / fsl,imx6dl-sabreauto:

regmap_update_bits(gpr, IOMUXC_GPR13, 0x07, 0x04);

Now it seems like the IPU is at least receiving something:

imx-ipuv3 2400000.ipu: IPU Warning - IPU_INT_STAT_5 = 0x00000001

New Frame before end-of-frame error indication of Channel interrupt. This bit is the status bit of New Frame before end-of-frame error interrupt of Channel 0.

Now i'm unsure how to fix that problem. The parameters im variating are:

in probe:

sensor_data.pix.priv = 1;  /* 1 is used to indicate TV in */

in ioctl_g_ifparm:

p->u.bt656.bt_sync_correct = 1; // 0

p->u.bt656.clock_curr = 27000000; // 0=IPU_CSI_CLK_MODE_CCIR656_INTERLACED, !=0=IPU_CSI_CLK_MODE_GATED_CLK

And then i'm using the tw9910 config from the tw9910 driver or the config that is specified by the datasheet for ITU-R BT.656 compatibility (which doesn't seem to match!).

But in these 16 combinations non of them gave me a full image capture.

Additionally from the userspace i'm automatically variating stuff like vsync/hsync polarity.

From the "Data Sheet: Technical Data: i.MX 6Dual/6QuadAutomotive and Infotainment Applications Processors" Figure 65 i think the hsync is not really an hsync-pulse but corresponds to the hactive from the tw9910 which i also tried. Is that correct?

Best regards.

Lars