Re: [PATCH v7 4/5] media: platform: mediatek: isp: add mediatek ISP3.0 camsv

["CK Hu (胡俊光)" <ck.hu@mediatek.com>]

On Fri, 2024-11-22 at 10:25 +0100, Julien Stephan wrote:
> External email : Please do not click links or open attachments until you have verified the sender or the content.
> 
> 
> Le ven. 22 nov. 2024 à 09:41, CK Hu (胡俊光) <ck.hu@mediatek.com> a écrit :
> > 
> > Hi, Julien:
> > 
> > On Thu, 2024-11-21 at 09:53 +0100, Julien Stephan wrote:
> > > External email : Please do not click links or open attachments until you have verified the sender or the content.
> > > 
> > > 
> > > From: Phi-bang Nguyen <pnguyen@baylibre.com>
> > > 
> > > This driver provides a path to bypass the SoC ISP so that image data
> > > coming from the SENINF can go directly into memory without any image
> > > processing. This allows the use of an external ISP.
> > > 
> > > Signed-off-by: Phi-bang Nguyen <pnguyen@baylibre.com>
> > > Signed-off-by: Florian Sylvestre <fsylvestre@baylibre.com>
> > > [Paul Elder fix irq locking]
> > > Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
> > > Co-developed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > > Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > > Co-developed-by: Julien Stephan <jstephan@baylibre.com>
> > > Signed-off-by: Julien Stephan <jstephan@baylibre.com>
> > > ---
> > 
> > [snip]
> > 
> > > +static irqreturn_t isp_irq_camsv30(int irq, void *data)
> > > +{
> > > +       struct mtk_cam_dev *cam_dev = (struct mtk_cam_dev *)data;
> > > +       struct mtk_cam_dev_buffer *buf;
> > > +       unsigned int irq_status;
> > > +
> > > +       spin_lock(&cam_dev->buf_list_lock);
> > > +
> > > +       irq_status = mtk_camsv30_read(cam_dev, CAMSV_INT_STATUS);
> > > +
> > > +       if (irq_status & INT_ST_MASK_CAMSV_ERR)
> > > +               dev_err(cam_dev->dev, "irq error 0x%lx\n",
> > > +                       irq_status & INT_ST_MASK_CAMSV_ERR);
> > > +
> > > +       /* De-queue frame */
> > > +       if (irq_status & CAMSV_IRQ_PASS1_DON) {
> > > +               cam_dev->sequence++;
> > > +
> > > +               buf = list_first_entry_or_null(&cam_dev->buf_list,
> > > +                                              struct mtk_cam_dev_buffer,
> > > +                                              list);
> > > +               if (buf) {
> > > +                       buf->v4l2_buf.sequence = cam_dev->sequence;
> > > +                       buf->v4l2_buf.vb2_buf.timestamp =
> > > +                               ktime_get_ns();
> > > +                       vb2_buffer_done(&buf->v4l2_buf.vb2_buf,
> > > +                                       VB2_BUF_STATE_DONE);
> > > +                       list_del(&buf->list);
> > > +               }
> > > +
> > > +               buf = list_first_entry_or_null(&cam_dev->buf_list,
> > > +                                              struct mtk_cam_dev_buffer,
> > > +                                              list);
> > > +               if (buf)
> > > +                       mtk_camsv30_update_buffers_add(cam_dev, buf);
> > 
> > If buf == NULL, so hardware would automatically stop DMA?
> > I don't know how this hardware work.
> > Below is my imagine about this hardware.
> > 
> > 1. Software use CAMSV_IMGO_FBC_RCNT_INC to increase software buffer index.
> > 2. Hardware has a hardware buffer index. After hardware finish one frame, hardware buffer index increase.
> > 3. After software buffer index increase, hardware start DMA.
> > 4. When hardware buffer index is equal to software buffer index, hardware automatically stop DMA.
> > 
> > Does the hardware work as my imagine?
> > If hardware could automatically stop DMA, add comment to describe.
> > If hardware could not automatically stop DMA, software should do something to stop DMA when buf == NULL.
> > 
> 
> You are right except that dma is not stopped but frames are
> automatically dropped by hardware until a new buffer is enqueued and
> software uses CAMSV_IMGO_FBC_RCNT_INC to increase the software buffer
> index.
> 
> What about adding the following comment:
> 
> /*
> * If there is no user buffer available, hardware will drop automatically
> * frames until buf_queue is called
> */

You say DMA is not stopped. Do you mean hardware still write data into latest buffer which would be dequeued to user space?
I think hardware should not write data into the buffer which has been take away by user space.
I think software should do something to stop DMA. Maybe use mtk_camsv30_cmos_vf_hw_disable() to stop DMA.

Regards,
CK

> 
> Let me know if that works for you
> 
> Cheers
> Julien
> 
> > Regards,
> > CK
> > 
> > > +       }
> > > +
> > > +       spin_unlock(&cam_dev->buf_list_lock);
> > > +
> > > +       return IRQ_HANDLED;
> > > +}
> > > +
> > 
> > ************* MEDIATEK Confidentiality Notice ********************
> > The information contained in this e-mail message (including any
> > attachments) may be confidential, proprietary, privileged, or otherwise
> > exempt from disclosure under applicable laws. It is intended to be
> > conveyed only to the designated recipient(s). Any use, dissemination,
> > distribution, printing, retaining or copying of this e-mail (including its
> > attachments) by unintended recipient(s) is strictly prohibited and may
> > be unlawful. If you are not an intended recipient of this e-mail, or believe
> > that you have received this e-mail in error, please notify the sender
> > immediately (by replying to this e-mail), delete any and all copies of
> > this e-mail (including any attachments) from your system, and do not
> > disclose the content of this e-mail to any other person. Thank you!