Hello Changhuang,
On Sat, Sep 14, 2024 at 11:53:31AM +0000, Changhuang Liang wrote:
> > On Wed, Jul 24, 2024 at 02:20:17AM +0000, Changhuang Liang wrote:
> > > > On Wed, Jul 10, 2024 at 11:11:57AM +0200, Jacopo Mondi wrote:
> > > > > On Tue, Jul 09, 2024 at 01:38:11AM GMT, Changhuang Liang wrote:
> > > > > > Add JH7110 ISP module definitions for user space.
> > > > > >
> > > > > > Signed-off-by: Changhuang Liang <changhuang.liang@xxxxxxxxxxxxxxxx>
> > > > > > Signed-off-by: Zejian Su <zejian.su@xxxxxxxxxxxxxxxx>
> > > > > > ---
> > > > > > MAINTAINERS | 1 +
> > > > > > include/uapi/linux/jh7110-isp.h | 739 ++++++++++++++++++++++++++++++++
> > > > >
> > > > > With the recently merged support for the RaspberryPi PiSP BE we
> > > > > have introduced include/uapi/linux/media/raspberry.
> > > > >
> > > > > Would you consider placing this in
> > > > > include/uapi/linux/media/startfive/ ?
> > > >
> > > > That sounds like a good idea.
> > > >
> > > > > > 2 files changed, 740 insertions(+) create mode 100644
> > > > > > include/uapi/linux/jh7110-isp.h
> > > > > >
> > > > > > diff --git a/MAINTAINERS b/MAINTAINERS index
> > > > > > 507f04a80499..890604eb0d64 100644
> > > > > > --- a/MAINTAINERS
> > > > > > +++ b/MAINTAINERS
> > > > > > @@ -21305,6 +21305,7 @@ S: Maintained
> > > > > > F: Documentation/admin-guide/media/starfive_camss.rst
> > > > > > F: Documentation/devicetree/bindings/media/starfive,jh7110-camss.yaml
> > > > > > F: drivers/staging/media/starfive/camss
> > > > > > +F: include/uapi/linux/jh7110-isp.h
> > > > > >
> > > > > > STARFIVE CRYPTO DRIVER
> > > > > > M: Jia Jie Ho <jiajie.ho@xxxxxxxxxxxxxxxx>
> > > > > > diff --git a/include/uapi/linux/jh7110-isp.h b/include/uapi/linux/jh7110-isp.h
> > > > > > new file mode 100644 index 000000000000..4939cd63e771
> > > > > > --- /dev/null
> > > > > > +++ b/include/uapi/linux/jh7110-isp.h
> > > > > > @@ -0,0 +1,739 @@
> > > > > > +/* SPDX-License-Identifier: ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) */
> > > > > > +/*
> > > > > > + * jh7110-isp.h
> > > > > > + *
> > > > > > + * JH7110 ISP driver - user space header file.
> > > > > > + *
> > > > > > + * Copyright © 2023 StarFive Technology Co., Ltd.
> > > > > > + *
> > > > > > + * Author: Zejian Su <zejian.su@xxxxxxxxxxxxxxxx>
> > > > > > + *
> > > > > > + */
> > > > > > +
> > > > > > +#ifndef __JH7110_ISP_H_
> > > > > > +#define __JH7110_ISP_H_
> > > > > > +
> > > > >
> > > > > Do you need to include
> > > > >
> > > > > #include <linux/types.h>
> > > > >
> > > > > > +/**
> > > > >
> > > > > Is this kernel-doc or a single * would do ?
> > > > >
> > > > > > + * ISP Module Diagram
> > > > > > + * ------------------
> > > > > > + *
> > > > > > + * Raw +-----+ +------+ +------+ +----+
> > > > > > + * ---->| OBC |--->| OECF |--->| LCCF |--->| WB |-----+
> > > > > > + * +-----+ +------+ +------+ +----+ |
> > > > > > + * |
> > > > > > + * +--------------------------------------------------+
> > > > > > + * |
> > > > > > + * | +-----+ +-----+ +-----+ +-----+
> > > > > > + * +--->| DBC |--->| CTC |--->| CFA |--->| CAR |------+
> > > > > > + * +-----+ +-----+ +-----+ +-----+ |
> > > > > > + * |
> > > > > > + * +--------------------------------------------------+
> > > > > > + * |
> > > > > > + * | +-----+ +--------+ +-----+ +------+
> > > > > > + * +--->| CCM |--->| GMARGB |--->| R2Y |--->| YCRV |--+
> > > > > > + * +-----+ +--------+ +-----+ +------+ |
> > > > > > + * |
> > > > > > + * +--------------------------------------------------+
> > > > > > + * |
> > > > > > + * | +-------+ +-------+ +-----+ +----+
> > > > > > + * +--->| SHARP |--->| DNYUV |--->| SAT |--->| SC |
> > > > > > + * +-------+ +-------+ +-----+ +----+
> > > > > > + *
> > > >
> > > > The diagram is useful, thank you. A glossary would also be nice,
> > > > maybe as
> > > >
> > > > * - OBC: Optical Black Correction
> > > > * - OECF: Opto-Electric Conversion Function
> > > > * ...
> > > >
> > > > I think that would be easier to read than the comments above each
> > > > macro below. Up to you.
> > > >
> > > > > > + */
> > > > > > +
> > > > > > +/* Auto White Balance */
> > > > > > +#define JH7110_ISP_MODULE_WB_SETTING (1U << 0)
> > > > > > +/* Color Artifact Removal */
> > > > > > +#define JH7110_ISP_MODULE_CAR_SETTING (1U << 1)
> > > > > > +/* Color Correction Matrix */
> > > > > > +#define JH7110_ISP_MODULE_CCM_SETTING (1U << 2)
> > > > > > +/* Color Filter Arrays */
> > > > > > +#define JH7110_ISP_MODULE_CFA_SETTING (1U << 3)
> > > > > > +/* Crosstalk Correction */
> > > > > > +#define JH7110_ISP_MODULE_CTC_SETTING (1U << 4)
> > > > > > +/* Defect Bad Pixel Correction */
> > > > > > +#define JH7110_ISP_MODULE_DBC_SETTING (1U << 5)
> > > > > > +/* Denoise YUV */
> > > > > > +#define JH7110_ISP_MODULE_DNYUV_SETTING (1U << 6)
> > > > > > +/* RGB Gamma */
> > > > > > +#define JH7110_ISP_MODULE_GMARGB_SETTING (1U << 7)
> > > > > > +/* Lens Correction Cosine Fourth */
> > > > > > +#define JH7110_ISP_MODULE_LCCF_SETTING (1U << 8)
> > > > > > +/* Optical Black Correction */
> > > > > > +#define JH7110_ISP_MODULE_OBC_SETTING (1U << 9)
> > > > > > +/* Opto-Electric Conversion Function */
> > > > > > +#define JH7110_ISP_MODULE_OECF_SETTING (1U << 10)
> > > > > > +/* RGB To YUV */
> > > > > > +#define JH7110_ISP_MODULE_R2Y_SETTING (1U << 11)
> > > > > > +/* Saturation */
> > > > > > +#define JH7110_ISP_MODULE_SAT_SETTING (1U << 12)
> > > > > > +/* Sharpen */
> > > > > > +#define JH7110_ISP_MODULE_SHARP_SETTING (1U << 13)
> > > > > > +/* Y Curve */
> > > > > > +#define JH7110_ISP_MODULE_YCRV_SETTING (1U << 14)
> > > > > > +/* Statistics Collection */
> > > > > > +#define JH7110_ISP_MODULE_SC_SETTING (1U << 15)
> > > >
> > > > Unless there's a specific reason to keep the current order, maybe
> > > > you could sort those macros in the same order as in the module diagram ?
> > > >
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_wb_gain - auto white balance gain
> > > > > > + *
> > > > > > + * @gain_r: gain value for red component.
> > > > > > + * @gain_g: gain value for green component.
> > > > > > + * @gain_b: gain value for blue component.
> > > >
> > > > I suppose the gains are expressed as fixed-point integers. This
> > > > needs more details, what are the limits, and how many bits of
> > > > integer and fractional parts are there ?
> > > >
> > > > Same comment for all the other values below.
> > > >
> > > > > > + */
> > > > > > +struct jh7110_isp_wb_gain {
> > > > > > + __u16 gain_r;
> > > > > > + __u16 gain_g;
> > > > > > + __u16 gain_b;
> > > > > > +};
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_wb_setting - Configuration used by auto
> > > > > > +white balance gain
> > > > > > + *
> > > > > > + * @enabled: enabled setting flag.
> > > > > > + * @gains: auto white balance gain setting.
> > > > > > + */
> > > > > > +struct jh7110_isp_wb_setting {
> > > > > > + __u32 enabled;
> > > > > > + struct jh7110_isp_wb_gain gains; };
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_car_setting - Configuration used by color
> > > > > > +artifact removal
> > > > > > + *
> > > > > > + * @enabled: enabled setting flag.
> > > > > > + */
> > > > > > +struct jh7110_isp_car_setting {
> > > > > > + __u32 enabled;
> > > > > > +};
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_ccm_smlow - Color correction matrix
> > > > > > + *
> > > > > > + * @ccm: color transform matrix with size 3 by 3.
> > > > > > + * @offsets: the offsets of R, G, B after the transform by the ccm.
> > > > > > + */
> > > > > > +struct jh7110_isp_ccm_smlow {
> > > > > > + __s32 ccm[3][3];
> > > > > > + __s32 offsets[3];
> > > > > > +};
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_ccm_setting - Configuration used by color
> > > > > > +correction matrix
> > > > > > + *
> > > > > > + * @enabled: enabled setting flag.
> > > > > > + * @ccm_smlow: Color correction matrix.
> > > > > > + */
> > > > > > +struct jh7110_isp_ccm_setting {
> > > > > > + __u32 enabled;
> > > > > > + struct jh7110_isp_ccm_smlow ccm_smlow; };
> > > > > > +
> > > > > > +/**
> > > > > > + * struct jh7110_isp_cfa_params - demosaic parameters
> > > > > > + *
> > > > > > + * @hv_width: detail smooth factor
> > > > > > + * @cross_cov: Cross covariance weighting.
> > > >
> > > > This documentation doesn't tell how to use those paraemeters. This
> > > > comment applies to many other parameters below. There are three main
> > > > options to improve that:
> > > >
> > > > - Expanding the documentation in this header file to clearly explain how
> > > > to compute the parameters values.
> > > >
> > > > - Providing an open userspace implementation of ISP algorithms that
> > > > showcase how to calculate the values.
> > > >
> > > > - Providing detailed hardware documentation for the ISP. This is usually
> > > > not favoured by ISP vendors, and we are not pushing for this, but I
> > > > wanted to mention it for completeness.
> > >
> > > We prefer the first option. It will take a lot of time for us to
> > > supplement the documentation.
> >
> > That is fine. Very broadly speaking, the level of documentation we are aiming
> > for should be enough for a third party developer to implement support for the
> > ISP control algorithms in libcamera. Please let me know if you would like to
> > discuss this in more details, to make sure there's no misunderstanding in the
> > expectations.
>
> After the discussion at our company, our company is not going to open source the ISP algorithm section in the short term.
> That's why I selected the first option at the first time.
>
> As you said, more ISP information is needed, however, due to the suspension of ISP and the dissolution of the team, we
> actually lack ISP information ourselves.
>
> So now we can only get more information about the ISP from the source code.
>
> Therefore, we can only provide the libcamera repository that is not open source ISP algorithm.
> https://github.com/changhuangliang/libcamera/tree/starfive_isp_3a
> (At the beginning, I was responsible for implementing the drivers part, and Zejian was responsible
> for implementing the libcamera, but after zejian left the company, I took over the libcamera part,
> but I am not familiar with this part)
>
> We won't open source in the short term, But we'll probably consider open source later when ISP
> has no commercial value.
>
> So I am confused whether to continue the next version.
A kernel driver must be usable by open and closed userspace alike. With
neither of documentation of the ISP or a reference open-source userspace
implementation being available, this is not possible. Therefore, I'm
afraid we won't be able to accept ISP support in the mainline kernel.
> > > > If you would prefer the second option, any open-source camera
> > > > framework would be acceptable, but in practice the only real option is
> > > > likely libcamera.
> > > >
> > > > This does not mean that you have to open-source all your ISP control
> > > > algorithms. Only the code needed to explain how ISP parameters are
> > > > applied to the image and are computed is needed. Other parts, such
> > > > as for instance AI-based computation of white balance gains, or
> > > > complex AGC calculations, don't need to be open-source.
> > > >
> > > > The explain this requirement different and perhaps more clearly, the
> > > > goal is to make sure that developers who have access to only
> > > > open-source code (ISP kernel driver, this header, any open-source
> > > > userspace code,
> > > > ...) will have enough information to configure and control the ISP.
--
Regards,
Laurent Pinchart
