Hi Philipp, Thanks for answering.
2014-11-25 9:16 GMT+01:00 Philipp Zabel <p.za...@pengutronix.de>: > Hi Jean-Michel, > > Am Montag, den 24.11.2014, 16:19 +0100 schrieb Jean-Michel Hautbois: >> Hi, >> >> I am working on using the CSI bus of i.MX6 with a adv7611 chip. >> I started to work with Steve Longerbeam's tree, and here is the >> current tree I am using : >> https://github.com/Vodalys/linux-2.6-imx/tree/mx6-camera-staging-v2-vbx >> >> This is a WiP tree, and not intended to be complete right now. >> But at least, it should be possible to get a picture. >> I will try to be as complete and synthetic as possible... >> >> Right now, I am configuring the ADV7611 in "16-Bit SDR ITU-R BT.656 >> 4:2:2 Mode 0" (Table 73 in Appendix C of the Reference Manual). > > ITU-R BT.656 only specifies 8-bit (or 10-bit) streams, the 16-bit BT.656 > SDR/DDR modes with two values on the bus at the same time are somewhat > nonstandard. As far as I can tell, this mode should correspond to the > CSI's BT.1120 SDR mode (Figure 37-20 in MX6DQ Reference Manual v1), so > I'd expect CSI_SENS_CONF to be configured as DATA_WIDTH=1 (8-bit > components), SENS_DATA_FORMAT=1 (YUV422), SENS_PRCTL=5 (progressive > BT.1120 SDR). OK, so I tested in a brutal way : diff --git a/drivers/gpu/ipu-v3/ipu-csi.c b/drivers/gpu/ipu-v3/ipu-csi.c index 293262d..ff48819 100644 --- a/drivers/gpu/ipu-v3/ipu-csi.c +++ b/drivers/gpu/ipu-v3/ipu-csi.c @@ -342,10 +342,16 @@ static void fill_csi_bus_cfg(struct ipu_csi_bus_config *csicfg, break; case V4L2_MBUS_BT656: csicfg->ext_vsync = 0; - if (V4L2_FIELD_HAS_BOTH(mbus_fmt->field)) - csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR656_INTERLACED; + if (mbus_fmt->code == V4L2_MBUS_FMT_YUYV8_2X8) + if (V4L2_FIELD_HAS_BOTH(mbus_fmt->field)) + csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR1120_PROGRESSIVE_SDR; + else + csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR1120_INTERLACED_SDR; else - csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR656_PROGRESSIVE; + if (V4L2_FIELD_HAS_BOTH(mbus_fmt->field)) + csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR656_INTERLACED; + else + csicfg->clk_mode = IPU_CSI_CLK_MODE_CCIR656_PROGRESSIVE; break; case V4L2_MBUS_CSI2: And before launching capture, I configure manually register 0x3 of ADV7611 in order to have the SDR 4:2:2 mode I want. It works better, but still have a little issue : I am expecting : 0x23 0x72 0x23 0xd4 ... I am getting : 0x23 0xc8 0x23 0x50 If I take binary values : 0x72 => 01110010b 0xc8 => 11001000b => 0x72 << 2 0xd4 => 11010100 0x50 => 01010000 => 0xd4 << 2 In my DT, I have specified : csi0: endpoint@0 { reg = <0>; bus-width = <16>; data-shift = <4>; /* Lines 19:4 used */ }; pinctrl_ipu1_csi0: ipu1_csi0grp { fsl,pins = < MX6QDL_PAD_EIM_D27__IPU1_CSI0_DATA00 0x80000000 MX6QDL_PAD_EIM_D26__IPU1_CSI0_DATA01 0x80000000 MX6QDL_PAD_EIM_D30__IPU1_CSI0_DATA03 0x80000000 MX6QDL_PAD_EIM_D31__IPU1_CSI0_DATA02 0x80000000 MX6QDL_PAD_CSI0_DAT4__IPU1_CSI0_DATA04 0x80000000 MX6QDL_PAD_CSI0_DAT5__IPU1_CSI0_DATA05 0x80000000 MX6QDL_PAD_CSI0_DAT6__IPU1_CSI0_DATA06 0x80000000 MX6QDL_PAD_CSI0_DAT7__IPU1_CSI0_DATA07 0x80000000 MX6QDL_PAD_CSI0_DAT8__IPU1_CSI0_DATA08 0x80000000 MX6QDL_PAD_CSI0_DAT9__IPU1_CSI0_DATA09 0x80000000 MX6QDL_PAD_CSI0_DAT10__IPU1_CSI0_DATA10 0x80000000 MX6QDL_PAD_CSI0_DAT11__IPU1_CSI0_DATA11 0x80000000 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 /* Clock and Data only : BT.656 mode */ MX6QDL_PAD_CSI0_PIXCLK__IPU1_CSI0_PIXCLK 0x80000000 /*MX6QDL_PAD_CSI0_MCLK__IPU1_CSI0_HSYNC 0x80000000 MX6QDL_PAD_CSI0_VSYNC__IPU1_CSI0_VSYNC 0x80000000 MX6QDL_PAD_CSI0_DATA_EN__IPU1_CSI0_DATA_EN 0x80000000*/ >; }; Can it be linked to the data-shift ? Thanks again. Regards, JM -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
