Re: [PATCH v2 1/3] media: V3s: Add support for Allwinner CSI.

[Yong <yong.deng@magewell.com>]

Hi,

On Sun, 30 Jul 2017 09:08:01 +0300
Baruch Siach <baruch@tkos.co.il> wrote:

> Hi Maxime, Yong,
> 
> On Fri, Jul 28, 2017 at 06:02:33PM +0200, Maxime Ripard wrote:
> > Hi, 
> > 
> > Thanks for the second iteration!
> > 
> > On Thu, Jul 27, 2017 at 01:01:35PM +0800, Yong Deng wrote:
> > > Allwinner V3s SoC have two CSI module. CSI0 is used for MIPI interface
> > > and CSI1 is used for parallel interface. This is not documented in
> > > datasheet but by testing and guess.
> > > 
> > > This patch implement a v4l2 framework driver for it.
> > > 
> > > Currently, the driver only support the parallel interface. MIPI-CSI2,
> > > ISP's support are not included in this patch.
> > > 
> > > Signed-off-by: Yong Deng <yong.deng@magewell.com>
> 
> [...]
> 
> > > +#ifdef DEBUG
> > > +static void sun6i_csi_dump_regs(struct sun6i_csi_dev *sdev)
> > > +{
> > > +	struct regmap *regmap = sdev->regmap;
> > > +	u32 val;
> > > +
> > > +	regmap_read(regmap, CSI_EN_REG, &val);
> > > +	printk("CSI_EN_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_IF_CFG_REG, &val);
> > > +	printk("CSI_IF_CFG_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_CAP_REG, &val);
> > > +	printk("CSI_CAP_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_SYNC_CNT_REG, &val);
> > > +	printk("CSI_SYNC_CNT_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_FIFO_THRS_REG, &val);
> > > +	printk("CSI_FIFO_THRS_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_PTN_LEN_REG, &val);
> > > +	printk("CSI_PTN_LEN_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_PTN_ADDR_REG, &val);
> > > +	printk("CSI_PTN_ADDR_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_VER_REG, &val);
> > > +	printk("CSI_VER_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_CH_CFG_REG, &val);
> > > +	printk("CSI_CH_CFG_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_CH_SCALE_REG, &val);
> > > +	printk("CSI_CH_SCALE_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_F0_BUFA_REG, &val);
> > > +	printk("CSI_CH_F0_BUFA_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_F1_BUFA_REG, &val);
> > > +	printk("CSI_CH_F1_BUFA_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_F2_BUFA_REG, &val);
> > > +	printk("CSI_CH_F2_BUFA_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_STA_REG, &val);
> > > +	printk("CSI_CH_STA_REG=0x%x\n",		val);
> > > +	regmap_read(regmap, CSI_CH_INT_EN_REG, &val);
> > > +	printk("CSI_CH_INT_EN_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_INT_STA_REG, &val);
> > > +	printk("CSI_CH_INT_STA_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_FLD1_VSIZE_REG, &val);
> > > +	printk("CSI_CH_FLD1_VSIZE_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_HSIZE_REG, &val);
> > > +	printk("CSI_CH_HSIZE_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_VSIZE_REG, &val);
> > > +	printk("CSI_CH_VSIZE_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_BUF_LEN_REG, &val);
> > > +	printk("CSI_CH_BUF_LEN_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_FLIP_SIZE_REG, &val);
> > > +	printk("CSI_CH_FLIP_SIZE_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_FRM_CLK_CNT_REG, &val);
> > > +	printk("CSI_CH_FRM_CLK_CNT_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_ACC_ITNL_CLK_CNT_REG, &val);
> > > +	printk("CSI_CH_ACC_ITNL_CLK_CNT_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_FIFO_STAT_REG, &val);
> > > +	printk("CSI_CH_FIFO_STAT_REG=0x%x\n",	val);
> > > +	regmap_read(regmap, CSI_CH_PCLK_STAT_REG, &val);
> > > +	printk("CSI_CH_PCLK_STAT_REG=0x%x\n",	val);
> > > +}
> > > +#endif
> > 
> > You can already dump a regmap through debugfs, that's redundant.
> 
> The advantage of in-code registers dump routine is the ability to synchronize 
> the snapshot with the driver code execution. This is particularly important 
> for the capture statistics registers. I have found it useful here.

Agree. It is not used to expose the registers value to user space. If you
think it is redundant, I will delete it.

> 
> [...]
> 
> > > +static int update_buf_addr(struct sun6i_csi *csi, dma_addr_t addr)
> > > +{
> > > +	struct sun6i_csi_dev *sdev = sun6i_csi_to_dev(csi);
> > > +	/* transform physical address to bus address */
> > > +	dma_addr_t bus_addr = addr - 0x40000000;
> > 
> > Like Baruch noticed, you should use PHYS_OFFSET here. The A80 for
> > example has a different RAM base address.
> > 
> > > +
> > > +	regmap_write(sdev->regmap, CSI_CH_F0_BUFA_REG,
> > > +		     (bus_addr + sdev->planar_offset[0]) >> 2);
> 
> Why do you need the bit shift? Does that work for you?
> 
> The User Manuals of both the V3s and the and the A33 (AKA R16) state that the 
> BUFA field size in this register is 31:00, that is 32bit. I have found no 
> indication of this bit shift in the Olimex provided sunxi-vfe[1] driver. On 
> the A33 I have found that only after removing the bit-shift, (some sort of) 
> data started to appear in the buffer.
> 
> [1] https://github.com/hehopmajieh/a33_linux/tree/master/drivers/media/video/sunxi-vfe
> 

The Users Manuals do not document this bit shift. You should see line 10 to
32 in https://github.com/hehopmajieh/a33_linux/blob/master/drivers/media/video/sunxi-vfe/csi/csi_reg.c

> [...]
> 
> > > +static irqreturn_t sun6i_csi_isr(int irq, void *dev_id)
> > > +{
> > > +	struct sun6i_csi_dev *sdev = (struct sun6i_csi_dev *)dev_id;
> > > +	struct regmap *regmap = sdev->regmap;
> > > +	u32 status;
> > > +
> > > +	regmap_read(regmap, CSI_CH_INT_STA_REG, &status);
> > > +
> > > +	if ((status & CSI_CH_INT_STA_FIFO0_OF_PD) ||
> > > +	    (status & CSI_CH_INT_STA_FIFO1_OF_PD) ||
> > > +	    (status & CSI_CH_INT_STA_FIFO2_OF_PD) ||
> > > +	    (status & CSI_CH_INT_STA_HB_OF_PD)) {
> > > +		regmap_write(regmap, CSI_CH_INT_STA_REG, status);
> > > +		regmap_update_bits(regmap, CSI_EN_REG, CSI_EN_CSI_EN, 0);
> > > +		regmap_update_bits(regmap, CSI_EN_REG, CSI_EN_CSI_EN,
> > > +				   CSI_EN_CSI_EN);
> > 
> > You need to enable / disable it at every frame? How do you deal with
> > double buffering? (or did you choose to ignore it for now?)
> 
> These *_OF_PD status bits indicate an overflow error condition.

Right.

> 
> > > +		return IRQ_HANDLED;
> > > +	}
> > > +
> > > +	if (status & CSI_CH_INT_STA_FD_PD) {
> > > +		sun6i_video_frame_done(&sdev->csi.video);
> > > +	}
> > > +
> > > +	regmap_write(regmap, CSI_CH_INT_STA_REG, status);
> > 
> > Isn't it redundant with the one you did in the condition a bit above?
> > 
> > You should also check that your device indeed generated an
> > interrupt. In the occurence of a spourious interrupt, your code will
> > return IRQ_HANDLED, which is the wrong thing to do.
> > 
> > I think you should reverse your logic a bit here to make this
> > easier. You should just check that your status flags are indeed set,
> > and if not just bail out and return IRQ_NONE.
> > 
> > And if they are, go on with treating your interrupt.
> > 
> > > +
> > > +	return IRQ_HANDLED;
> > > +}
> 
> baruch
> 
> -- 
>      http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
> =}------------------------------------------------ooO--U--Ooo------------{=
>    - baruch@tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -


Thanks,
Yong