Re: [PATCH v3 08/14] drm: rcar-du: Add support for CMM

[Laurent Pinchart]

Hi Jacopo,

On Thu, Sep 05, 2019 at 11:57:57AM +0200, Jacopo Mondi wrote:
> On Tue, Aug 27, 2019 at 07:34:23PM +0300, Laurent Pinchart wrote:
> > On Tue, Aug 27, 2019 at 04:56:19PM +0200, Jacopo Mondi wrote:
> >> On Tue, Aug 27, 2019 at 03:24:22AM +0300, Laurent Pinchart wrote:
> >>> On Sun, Aug 25, 2019 at 03:51:48PM +0200, Jacopo Mondi wrote:
> >>>> Add a driver for the R-Car Display Unit Color Correction Module.
> >>>>
> >>>> In most of Gen3 SoCs, each DU output channel is provided with a CMM unit
> >>>> to perform image enhancement and color correction.
> >>>>
> >>>> Add support for CMM through a driver that supports configuration of
> >>>> the 1-dimensional LUT table. More advanced CMM feature will be
> >>>> implemented on top of this basic one.
> >>>>
> >>>> Signed-off-by: Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
> >>>> ---
> >>>>  drivers/gpu/drm/rcar-du/Kconfig    |   7 +
> >>>>  drivers/gpu/drm/rcar-du/Makefile   |   1 +
> >>>>  drivers/gpu/drm/rcar-du/rcar_cmm.c | 262 +++++++++++++++++++++++++++++
> >>>>  drivers/gpu/drm/rcar-du/rcar_cmm.h |  38 +++++
> >>>>  4 files changed, 308 insertions(+)
> >>>>  create mode 100644 drivers/gpu/drm/rcar-du/rcar_cmm.c
> >>>>  create mode 100644 drivers/gpu/drm/rcar-du/rcar_cmm.h
> >>>>
> >>>> diff --git a/drivers/gpu/drm/rcar-du/Kconfig b/drivers/gpu/drm/rcar-du/Kconfig
> >>>> index 1529849e217e..539d232790d1 100644
> >>>> --- a/drivers/gpu/drm/rcar-du/Kconfig
> >>>> +++ b/drivers/gpu/drm/rcar-du/Kconfig
> >>>> @@ -13,6 +13,13 @@ config DRM_RCAR_DU
> >>>>  	  Choose this option if you have an R-Car chipset.
> >>>>  	  If M is selected the module will be called rcar-du-drm.
> >>>>
> >>>> +config DRM_RCAR_CMM
> >>>> +	bool "R-Car DU Color Management Module (CMM) Support"
> >>>> +	depends on DRM && OF
> >>>> +	depends on DRM_RCAR_DU
> >>>> +	help
> >>>> +	  Enable support for R-Car Color Management Module (CMM).
> >>>> +
> >>>>  config DRM_RCAR_DW_HDMI
> >>>>  	tristate "R-Car DU Gen3 HDMI Encoder Support"
> >>>>  	depends on DRM && OF
> >>>> diff --git a/drivers/gpu/drm/rcar-du/Makefile b/drivers/gpu/drm/rcar-du/Makefile
> >>>> index 6c2ed9c46467..4d1187ccc3e5 100644
> >>>> --- a/drivers/gpu/drm/rcar-du/Makefile
> >>>> +++ b/drivers/gpu/drm/rcar-du/Makefile
> >>>> @@ -15,6 +15,7 @@ rcar-du-drm-$(CONFIG_DRM_RCAR_LVDS)	+= rcar_du_of.o \
> >>>>  rcar-du-drm-$(CONFIG_DRM_RCAR_VSP)	+= rcar_du_vsp.o
> >>>>  rcar-du-drm-$(CONFIG_DRM_RCAR_WRITEBACK) += rcar_du_writeback.o
> >>>>
> >>>> +obj-$(CONFIG_DRM_RCAR_CMM)		+= rcar_cmm.o
> >>>>  obj-$(CONFIG_DRM_RCAR_DU)		+= rcar-du-drm.o
> >>>>  obj-$(CONFIG_DRM_RCAR_DW_HDMI)		+= rcar_dw_hdmi.o
> >>>>  obj-$(CONFIG_DRM_RCAR_LVDS)		+= rcar_lvds.o
> >>>> diff --git a/drivers/gpu/drm/rcar-du/rcar_cmm.c b/drivers/gpu/drm/rcar-du/rcar_cmm.c
> >>>> new file mode 100644
> >>>> index 000000000000..55361f5701e8
> >>>> --- /dev/null
> >>>> +++ b/drivers/gpu/drm/rcar-du/rcar_cmm.c
> >>>> @@ -0,0 +1,262 @@
> >>>> +// SPDX-License-Identifier: GPL-2.0+
> >>>> +/*
> >>>> + * rcar_cmm.c -- R-Car Display Unit Color Management Module
> >>>> + *
> >>>> + * Copyright (C) 2019 Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
> >>>> + */
> >>>> +
> >>>> +#include <linux/io.h>
> >>>> +#include <linux/module.h>
> >>>> +#include <linux/of.h>
> >>>> +#include <linux/platform_device.h>
> >>>> +#include <linux/pm_runtime.h>
> >>>> +
> >>>> +#include <drm/drm_color_mgmt.h>
> >>>> +
> >>>> +#include "rcar_cmm.h"
> >>>> +
> >>>> +#define CM2_LUT_CTRL		0x0000
> >>>> +#define CM2_LUT_CTRL_LUT_EN	BIT(0)
> >>>> +#define CM2_LUT_TBL_BASE	0x0600
> >>>> +#define CM2_LUT_TBL(__i)	(CM2_LUT_TBL_BASE + (__i) * 4)
> >>>> +
> >>>> +struct rcar_cmm {
> >>>> +	void __iomem *base;
> >>>> +	bool enabled;
> >>>> +
> >>>> +	/*
> >>>> +	 * @lut:		1D-LUT status
> >>>> +	 * @lut.enabled:	1D-LUT enabled flag
> >>>> +	 * @lut.size:		Number of entries in the LUT table
> >>>
> >>> Please see my review of patch 13/14, I wonder if we could drop this
> >>> field.
> >>>
> >>>> +	 * @lut.table:		Table of 1D-LUT entries scaled to HW support
> >>>> +	 *			precision (8-bits per color component)
> >>>> +	 */
> >>>> +	struct {
> >>>> +		bool enabled;
> >>>> +		unsigned int size;
> >>>> +		u32 table[CMM_GAMMA_LUT_SIZE];
> >>>> +	} lut;
> >>>> +};
> >>>> +
> >>>> +static inline int rcar_cmm_read(struct rcar_cmm *rcmm, u32 reg)
> >>>> +{
> >>>> +	return ioread32(rcmm->base + reg);
> >>>> +}
> >>>> +
> >>>> +static inline void rcar_cmm_write(struct rcar_cmm *rcmm, u32 reg, u32 data)
> >>>> +{
> >>>> +	iowrite32(data, rcmm->base + reg);
> >>>> +}
> >>>> +
> >>>> +/*
> >>>> + * rcar_cmm_lut_extract() - Scale down to hw precision the DRM LUT table
> >>>
> >>> s/hw/hardware/ (and below too)
> >>>
> >>>> + *			    entries and store them.
> >>>> + * @rcmm: Pointer to the CMM device
> >>>> + * @size: Number of entries in the table
> >>>> + * @drm_lut: DRM LUT table
> >>>> + */
> >>>> +static void rcar_cmm_lut_extract(struct rcar_cmm *rcmm, size_t size,
> >>>> +				 const struct drm_color_lut *drm_lut)
> >>>> +{
> >>>> +	unsigned int i;
> >>>> +
> >>>> +	for (i = 0; i < size; ++i) {
> >>>> +		const struct drm_color_lut *lut = &drm_lut[i];
> >>>> +
> >>>> +		rcmm->lut.table[i] = drm_color_lut_extract(lut->red, 8) << 16
> >>>> +				   | drm_color_lut_extract(lut->green, 8) << 8
> >>>> +				   | drm_color_lut_extract(lut->blue, 8);
> >>>> +	}
> >>>> +
> >>>> +	rcmm->lut.size = size;
> >>>> +}
> >>>> +
> >>>> +/*
> >>>> + * rcar_cmm_lut_load() - Write to hw the LUT table entries from the local table.
> >>>> + *
> >>>
> >>> No need for a blank line
> >>>
> >>>> + * @rcmm: Pointer to the CMM device
> >>>> + */
> >>>> +static void rcar_cmm_lut_load(struct rcar_cmm *rcmm)
> >>>
> >>> I would name this rcar_cmm_lut_write().
> >>
> >> I won't, as I would like to convey the LUT tables is loaded from the
> >> local cache after it has been scaled down to the hardware supported
> >> precision.
> >
> > "load" hints a read though, and here you write the LUT to the hardware.
> > Without reading the comments I would have thought this function would
> > read the LUT back from the hardware.
> >
> >>>> +{
> >>>> +	unsigned int i;
> >>>> +
> >>>> +	for (i = 0; i < rcmm->lut.size; ++i) {
> >>>> +		u32 entry = rcmm->lut.table[i];
> >>>> +
> >>>> +		rcar_cmm_write(rcmm, CM2_LUT_TBL(i), entry);
> >>>
> >>> You don't need the local entry variable.
> >>
> >> True, but the code is nicer to read and the compiler should be smart
> >> enough to optimize it away
> >
> > I'm not sure about nicer to read, I find the opposite personally, but
> > it's your code :-)
> >
> >>>> +	}
> >>>> +}
> >>>> +
> >>>> +/**
> >>>> + * rcar_cmm_setup() - configure the CMM unit
> >>>
> >>> s/configure/Configure/ and s/$/./, or the other way around for the other
> >>> functions (I don't mine which one, but let's stay consistent).
> >>
> >> Oh right, sorry for the confusion
> >
> > It's just my OCD kicking in :-)
> >
> >>>> + *
> >>>
> >>> No need for a blank line (same for the functions below).
> >>>
> >>>> + * @pdev: The platform device associated with the CMM instance
> >>>> + * @config: The CRTC-provided configuration.
> >>>> + *
> >>>> + * Configure the CMM unit with the CRTC-provided configuration.
> >>>> + * Currently enabling, disabling and programming of the 1-D LUT unit is
> >>>> + * supported.
> >>>> + */
> >>>> +int rcar_cmm_setup(struct platform_device *pdev,
> >>>> +		   const struct rcar_cmm_config *config)
> >>>> +{
> >>>> +	struct rcar_cmm *rcmm = platform_get_drvdata(pdev);
> >>>> +
> >>>> +	if (config->lut.size > CMM_GAMMA_LUT_SIZE)
> >>>> +		return -EINVAL;
> >>>> +
> >>>> +	/*
> >>>> +	 * As rcar_cmm_setup() is called by atomic commit tail helper, it might
> >>>> +	 * be called when the CMM is disabled. As we can't program the hardware
> >>>> +	 * in that case, store the configuration internally and apply it when
> >>>> +	 * the CMM will be enabled by the CRTC through rcar_cmm_enable().
> >>>> +	 */
> >>>> +	if (!rcmm->enabled) {
> >>>> +		if (!config->lut.enable)
> >>>> +			return 0;
> >>>> +
> >>>> +		rcar_cmm_lut_extract(rcmm, config->lut.size, config->lut.table);
> >>>> +		rcmm->lut.enabled = true;
> >>>> +
> >>>> +		return 0;
> >>>> +	}
> >>>> +
> >>>> +	/* Stop LUT operations if requested. */
> >>>> +	if (!config->lut.enable) {
> >>>> +		if (rcmm->lut.enabled) {
> >>>> +			rcar_cmm_write(rcmm, CM2_LUT_CTRL, 0);
> >>>> +			rcmm->lut.enabled = false;
> >>>> +			rcmm->lut.size = 0;
> >>>> +		}
> >>>> +
> >>>> +		return 0;
> >>>> +	}
> >>>> +
> >>>> +	/*
> >>>> +	 * Enable LUT and program the new gamma table values.
> >>>> +	 *
> >>>> +	 * FIXME: In order to have stable operations it is required to first
> >>>> +	 * enable the 1D-LUT and then program its table entries. This seems to
> >>>> +	 * contradict what the chip manual reports, and will have to be
> >>>> +	 * reconsidered when implementing support for double buffering.
> >>>> +	 */
> >>>> +	if (!rcmm->lut.enabled) {
> >>>> +		rcar_cmm_write(rcmm, CM2_LUT_CTRL, CM2_LUT_CTRL_LUT_EN);
> >>>> +		rcmm->lut.enabled = true;
> >>>> +	}
> >>>> +
> >>>> +	rcar_cmm_lut_extract(rcmm, config->lut.size, config->lut.table);
> >>>> +	rcar_cmm_lut_load(rcmm);
> >>>> +
> >>>> +	return 0;
> >>>> +}
> >>>> +EXPORT_SYMBOL_GPL(rcar_cmm_setup);
> >>>> +
> >>>> +/**
> >>>> + * rcar_cmm_enable() - enable the CMM unit
> >>>> + *
> >>>> + * @pdev: The platform device associated with the CMM instance
> >>>> + *
> >>>> + * Enable the CMM unit by enabling the parent clock and enabling the CMM
> >>>> + * components, such as 1-D LUT, if requested.
> >>>> + */
> >>>> +int rcar_cmm_enable(struct platform_device *pdev)
> >>>> +{
> >>>> +	struct rcar_cmm *rcmm = platform_get_drvdata(pdev);
> >>>> +	int ret;
> >>>> +
> >>>> +	if (!rcmm)
> >>>> +		return -EPROBE_DEFER;
> >>>
> >>> This function is called in rcar_du_crtc_atomic_enable(), so that's not
> >>> the right error code. It seems we need another function for the CMM API
> >>> to defer probing :-/ I would call it rcar_cmm_init(). This check would
> >>> then be removed.
> >>
> >> I agree about the return code, but not the name, as this function
> >> actually enables the CMM.
> >
> > I meant creating a new rcar_cmm_init() function that would just have the
> > !rcmm check.
> >
> >> PROBE_DEFER does not make any sense here, I
> >> wonder where it come from, as the probing of CMM and DU has long
> >> happened once we get here (at least, I assume so, if we receive a
> >> gamma_table, userspace has already been running, and both DU and CMM
> >> should have probed. Otherwise, we can exploit the newly created device
> >> link, and make sure DU probes after the CMM).
> >>
> >> I would just change the return value here, and possibly use the device
> >> link to ensure the correct probing sequence.
> >
> > How does device link help here ?
> 
> Currently it doesn't, as we are creating a stateless link.
> 
> But if we go for a managed device link (which is the default, by the
> way, you have to opt-out from it) we can guarantee the CMM has probed
> before the DU probes, so that we have a guarantee when we get here
> !rcmm cannot happen.
> 
> https://www.kernel.org/doc/html/v5.2-rc7/driver-api/device_link.html
> "The consumer devices are not probed before the supplier is bound to a driver,
>  and theyâre unbound before the supplier is unbound."
> 
> As we create the link, the CMM is the supplier of DU, so we could just
> drop the DL_FLAG_STATELESS flag in device_link_add() in 10/14.
> 
> Does this match your understanding ?

Except there's a bit of a chicken and egg issue, as you call
device_link_add() from rcar_du_cmm_init(), which thus require the DU
driver to probe first :-) For this to work we would probably need an
early initcall in the DU driver.

> >>>> +
> >>>> +	ret = pm_runtime_get_sync(&pdev->dev);
> >>>> +	if (ret < 0)
> >>>> +		return ret;
> >>>> +
> >>>> +	/* Apply the LUT table values saved at rcar_cmm_setup() time. */
> >>>> +	if (rcmm->lut.enabled) {
> >>>> +		rcar_cmm_write(rcmm, CM2_LUT_CTRL, CM2_LUT_CTRL_LUT_EN);
> >>>> +		rcar_cmm_lut_load(rcmm);
> >>>
> >>> You will not like this, but I just realised that we're now reprogramming
> >>> the LUT contents every time the CMM is enabled. Do you think that's
> >>> something we should optimise ? And yes, that would require introducing
> >>
> >> Why so? If we receive an enable after a disable which stops the CMM
> >> clock and we have no guarantees the table entries have been kept, or
> >> what we receive from userspace has changed or not. Why is this an
> >> issue in your opinion?
> >
> > I thought the hardware preserved the LUT ? Skipping the LUT write is an
> > optimisation, so we could do without it in the initial version. I think
> > it would become more important with the CLU though, as we'll have more
> > data entries there. Maybe we should first check how much time the LUT
> > and CLU writes take before deciding to optimise them.
> >
> 
> Yeah, let's post-pone optimizations...
> 
> >>> back an update flag in rcmm->lut :-S Sorry for not realising this when I
> >>> proposed dropping it.
> >>>
> >>>> +	}
> >>>> +
> >>>> +	rcmm->enabled = true;
> >>>> +
> >>>> +	return 0;
> >>>> +}
> >>>> +EXPORT_SYMBOL_GPL(rcar_cmm_enable);
> >>>> +
> >>>> +/**
> >>>> + * rcar_cmm_disable() - disable the CMM unit
> >>>> + *
> >>>> + * @pdev: The platform device associated with the CMM instance
> >>>> + *
> >>>> + * Disable the CMM unit by stopping the parent clock.
> >>>> + */
> >>>> +void rcar_cmm_disable(struct platform_device *pdev)
> >>>> +{
> >>>> +	struct rcar_cmm *rcmm = platform_get_drvdata(pdev);
> >>>> +
> >>>> +	rcar_cmm_write(rcmm, CM2_LUT_CTRL, 0);
> >>>> +
> >>>> +	pm_runtime_put(&pdev->dev);
> >>>> +
> >>>> +	rcmm->lut.enabled = false;
> >>>> +	rcmm->lut.size = 0;
> >>>> +
> >>>> +	rcmm->enabled = false;
> >>>> +}
> >>>> +EXPORT_SYMBOL_GPL(rcar_cmm_disable);
> >>>> +
> >>>> +static int rcar_cmm_probe(struct platform_device *pdev)
> >>>> +{
> >>>> +	struct rcar_cmm *rcmm;
> >>>> +	struct resource *res;
> >>>> +
> >>>> +	rcmm = devm_kzalloc(&pdev->dev, sizeof(*rcmm), GFP_KERNEL);
> >>>> +	if (!rcmm)
> >>>> +		return -ENOMEM;
> >>>> +
> >>>> +	platform_set_drvdata(pdev, rcmm);
> >>>> +
> >>>> +	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> >>>> +	rcmm->base = devm_ioremap_resource(&pdev->dev, res);
> >>>> +	if (IS_ERR(rcmm->base))
> >>>> +		return PTR_ERR(rcmm->base);
> >>>
> >>> You really don't like combining those two calls, do you ? :-)
> >>
> >> devm_of_iomap() ?
> >
> > devm_platform_ioremap_resource()
> 
> Oh stupid, thanks!
> 
> >>>> +
> >>>> +	pm_runtime_enable(&pdev->dev);
> >>>> +
> >>>> +	return 0;
> >>>> +}
> >>>> +
> >>>> +static int rcar_cmm_remove(struct platform_device *pdev)
> >>>> +{
> >>>> +	pm_runtime_disable(&pdev->dev);
> >>>> +
> >>>> +	return 0;
> >>>> +}
> >>>> +
> >>>> +static const struct of_device_id rcar_cmm_of_table[] = {
> >>>> +	{ .compatible = "renesas,cmm-r8a7795", },
> >>>> +	{ .compatible = "renesas,cmm-r8a7796", },
> >>>> +	{ .compatible = "renesas,cmm-r8a77965", },
> >>>> +	{ .compatible = "renesas,cmm-r8a77990", },
> >>>> +	{ .compatible = "renesas,cmm-r8a77995", },
> >>>
> >>> As Geert pointed out, I would drop those entries.
> >>
> >> yes
> >>
> >>>> +	{ .compatible = "renesas,rcar-gen3-cmm", },
> >>>> +	{ .compatible = "renesas,rcar-gen2-cmm", },
> >>>> +	{ },
> >>>> +};
> >>>> +MODULE_DEVICE_TABLE(of, rcar_cmm_of_table);
> >>>> +
> >>>> +static struct platform_driver rcar_cmm_platform_driver = {
> >>>> +	.probe		= rcar_cmm_probe,
> >>>> +	.remove		= rcar_cmm_remove,
> >>>> +	.driver		= {
> >>>> +		.name	= "rcar-cmm",
> >>>> +		.of_match_table = rcar_cmm_of_table,
> >>>> +	},
> >>>> +};
> >>>> +
> >>>> +module_platform_driver(rcar_cmm_platform_driver);
> >>>> +
> >>>> +MODULE_AUTHOR("Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>");
> >>>> +MODULE_DESCRIPTION("Renesas R-Car CMM Driver");
> >>>> +MODULE_LICENSE("GPL v2");
> >>>> diff --git a/drivers/gpu/drm/rcar-du/rcar_cmm.h b/drivers/gpu/drm/rcar-du/rcar_cmm.h
> >>>> new file mode 100644
> >>>> index 000000000000..b0bb7349ebaa
> >>>> --- /dev/null
> >>>> +++ b/drivers/gpu/drm/rcar-du/rcar_cmm.h
> >>>> @@ -0,0 +1,38 @@
> >>>> +/* SPDX-License-Identifier: GPL-2.0+ */
> >>>> +/*
> >>>> + * rcar_cmm.h -- R-Car Display Unit Color Management Module
> >>>> + *
> >>>> + * Copyright (C) 2019 Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
> >>>> + */
> >>>> +
> >>>> +#ifndef __RCAR_CMM_H__
> >>>> +#define __RCAR_CMM_H__
> >>>> +
> >>>> +#define CMM_GAMMA_LUT_SIZE		256
> >>>> +
> >>>> +struct drm_color_lut;
> >>>> +struct platform_device;
> >>>> +
> >>>> +/**
> >>>> + * struct rcar_cmm_config - CMM configuration
> >>>> + *
> >>>> + * @lut:	1D-LUT configuration
> >>>> + * @lut.enable:	1D-LUT enable flag
> >>>> + * @lut.table:	1D-LUT table entries
> >>>> + * @lut.size:	Number of 1D-LUT (max 256)
> >>>
> >>> s/1D-LUT/1D-LUT entries/
> >>
> >> ack, I'll change this.
> >>
> >>>> + */
> >>>> +struct rcar_cmm_config {
> >>>> +	struct {
> >>>> +		bool enable;
> >>>> +		struct drm_color_lut *table;
> >>>> +		unsigned int size;
> >>>> +	} lut;
> >>>> +};
> >>>> +
> >>>> +int rcar_cmm_enable(struct platform_device *pdev);
> >>>> +void rcar_cmm_disable(struct platform_device *pdev);
> >>>> +
> >>>> +int rcar_cmm_setup(struct platform_device *pdev,
> >>>> +		   const struct rcar_cmm_config *config);
> >>>> +
> >>>> +#endif /* __RCAR_CMM_H__ */

-- 
Regards,

Laurent Pinchart