From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org,
David Airlie <airlied@linux.ie>,
"open list:DRM DRIVERS FOR RENESAS"
<dri-devel@lists.freedesktop.org>,
open list <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 05/17] drm: rcar-du: Split CRTC handling to support hardware indexing
Date: Thu, 26 Apr 2018 23:30:52 +0300 [thread overview]
Message-ID: <8207827.UIER1Ri0ED@avalon> (raw)
In-Reply-To: <20180426165346.494-6-kieran.bingham+renesas@ideasonboard.com>
Hi Kieran,
Thank you for the patch.
On Thursday, 26 April 2018 19:53:34 EEST Kieran Bingham wrote:
> The DU CRTC driver does not support distinguishing between a hardware
> index, and a software (CRTC) index in the event that a DU channel might
> not be populated by the hardware.
>
> Support this by adapting the rcar_du_device_info structure to store a
> bitmask of available channels rather than a count of CRTCs. The count
> can then be obtained by determining the hamming weight of the bitmask.
>
> This allows the rcar_du_crtc_create() function to distinguish between
> both index types, and non-populated DU channels will be skipped without
> leaving a gap in the software CRTC indexes.
>
> Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
> ---
> drivers/gpu/drm/rcar-du/rcar_du_crtc.c | 26 ++++++++++++++------------
> drivers/gpu/drm/rcar-du/rcar_du_crtc.h | 3 ++-
> drivers/gpu/drm/rcar-du/rcar_du_drv.c | 20 ++++++++++----------
> drivers/gpu/drm/rcar-du/rcar_du_drv.h | 4 ++--
> drivers/gpu/drm/rcar-du/rcar_du_kms.c | 17 ++++++++++++-----
> 5 files changed, 40 insertions(+), 30 deletions(-)
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c index 5a15dfd66343..36ce194c13b5
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> @@ -902,7 +902,8 @@ static irqreturn_t rcar_du_crtc_irq(int irq, void *arg)
> * Initialization
> */
>
> -int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int index)
> +int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int swindex,
> + unsigned int hwindex)
> {
> static const unsigned int mmio_offsets[] = {
> DU0_REG_OFFSET, DU1_REG_OFFSET, DU2_REG_OFFSET, DU3_REG_OFFSET
> @@ -910,7 +911,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> struct rcar_du_device *rcdu = rgrp->dev;
> struct platform_device *pdev = to_platform_device(rcdu->dev);
> - struct rcar_du_crtc *rcrtc = &rcdu->crtcs[index];
> + struct rcar_du_crtc *rcrtc = &rcdu->crtcs[swindex];
> struct drm_crtc *crtc = &rcrtc->crtc;
> struct drm_plane *primary;
> unsigned int irqflags;
> @@ -922,7 +923,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> /* Get the CRTC clock and the optional external clock. */
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_CRTC_IRQ_CLOCK)) {
> - sprintf(clk_name, "du.%u", index);
> + sprintf(clk_name, "du.%u", hwindex);
> name = clk_name;
> } else {
> name = NULL;
> @@ -930,16 +931,16 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> rcrtc->clock = devm_clk_get(rcdu->dev, name);
> if (IS_ERR(rcrtc->clock)) {
> - dev_err(rcdu->dev, "no clock for CRTC %u\n", index);
> + dev_err(rcdu->dev, "no clock for CRTC %u\n", swindex);
How about
dev_err(rcdu->dev, "no clock for DU channel %u\n", hwindex);
I think that would be clearer, because at this stage we're dealing with
hardware resources, so matching the datasheet numbers seems better to me.
> return PTR_ERR(rcrtc->clock);
> }
>
> - sprintf(clk_name, "dclkin.%u", index);
> + sprintf(clk_name, "dclkin.%u", hwindex);
> clk = devm_clk_get(rcdu->dev, clk_name);
> if (!IS_ERR(clk)) {
> rcrtc->extclock = clk;
> } else if (PTR_ERR(rcrtc->clock) == -EPROBE_DEFER) {
> - dev_info(rcdu->dev, "can't get external clock %u\n", index);
> + dev_info(rcdu->dev, "can't get external clock %u\n", hwindex);
> return -EPROBE_DEFER;
> }
>
> @@ -948,13 +949,13 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) spin_lock_init(&rcrtc->vblank_lock);
>
> rcrtc->group = rgrp;
> - rcrtc->mmio_offset = mmio_offsets[index];
> - rcrtc->index = index;
> + rcrtc->mmio_offset = mmio_offsets[hwindex];
> + rcrtc->index = hwindex;
>
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_VSP1_SOURCE))
> primary = &rcrtc->vsp->planes[rcrtc->vsp_pipe].plane;
> else
> - primary = &rgrp->planes[index % 2].plane;
> + primary = &rgrp->planes[hwindex % 2].plane;
This shouldn't make a difference because when RCAR_DU_FEATURE_VSP1_SOURCE
isn't set we're running on Gen2, and don't need to deal with indices, but from
a conceptual point of view, wouldn't the software index be better here ?
Missing hardware channels won't be visible from userspace, so taking the first
plane of the group as the primary plane would seem better to me.
> ret = drm_crtc_init_with_planes(rcdu->ddev, crtc, primary, NULL,
> rcdu->info->gen <= 2 ?
> @@ -970,7 +971,8 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> /* Register the interrupt handler. */
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_CRTC_IRQ_CLOCK)) {
> - irq = platform_get_irq(pdev, index);
> + /* The IRQ's are associated with the CRTC (sw)index */
s/index/index./
> + irq = platform_get_irq(pdev, swindex);
> irqflags = 0;
> } else {
> irq = platform_get_irq(pdev, 0);
> @@ -978,7 +980,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) }
>
> if (irq < 0) {
> - dev_err(rcdu->dev, "no IRQ for CRTC %u\n", index);
> + dev_err(rcdu->dev, "no IRQ for CRTC %u\n", swindex);
> return irq;
> }
>
> @@ -986,7 +988,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) dev_name(rcdu->dev), rcrtc);
> if (ret < 0) {
> dev_err(rcdu->dev,
> - "failed to register IRQ for CRTC %u\n", index);
> + "failed to register IRQ for CRTC %u\n", swindex);
> return ret;
> }
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h index 518ee2c60eb8..5f003a16abc5
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> @@ -99,7 +99,8 @@ enum rcar_du_output {
> RCAR_DU_OUTPUT_MAX,
> };
>
> -int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int index);
> +int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int swindex,
> + unsigned int hwindex);
> void rcar_du_crtc_suspend(struct rcar_du_crtc *rcrtc);
> void rcar_du_crtc_resume(struct rcar_du_crtc *rcrtc);
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> b/drivers/gpu/drm/rcar-du/rcar_du_drv.c index 05745e86d73e..d6ebc628fc22
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> @@ -40,7 +40,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7743_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
I'd write it BIT(1) | BIT(0) to match the usual little-endian order. Same
comment for the other info structure instances.
> .routes = {
> /*
> * R8A7743 has one RGB output and one LVDS output
> @@ -61,7 +61,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7745_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7745 has two RGB outputs
> @@ -80,7 +80,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7745_info = { static const struct rcar_du_device_info
> rcar_du_r8a7779_info = {
> .gen = 2,
> .features = 0,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7779 has two RGB outputs and one (currently unsupported)
> @@ -102,7 +102,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7790_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> .quirks = RCAR_DU_QUIRK_ALIGN_128B,
> - .num_crtcs = 3,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2),
> .routes = {
> /*
> * R8A7790 has one RGB output, two LVDS outputs and one
> @@ -129,7 +129,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7791_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A779[13] has one RGB output, one LVDS output and one
> @@ -151,7 +151,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7792_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /* R8A7792 has two RGB outputs. */
> [RCAR_DU_OUTPUT_DPAD0] = {
> @@ -169,7 +169,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7794_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7794 has two RGB outputs and one (currently unsupported)
> @@ -191,7 +191,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7795_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 4,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2) | BIT(3),
> .routes = {
> /*
> * R8A7795 has one RGB output, two HDMI outputs and one
> @@ -223,7 +223,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7796_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 3,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2),
> .routes = {
> /*
> * R8A7796 has one RGB output, one LVDS output and one HDMI
> @@ -251,7 +251,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a77970_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 1,
> + .channel_mask = BIT(0),
> .routes = {
> /* R8A77970 has one RGB output and one LVDS output. */
> [RCAR_DU_OUTPUT_DPAD0] = {
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> b/drivers/gpu/drm/rcar-du/rcar_du_drv.h index 5c7ec15818c7..7a5de66deec2
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> @@ -52,7 +52,7 @@ struct rcar_du_output_routing {
> * @gen: device generation (2 or 3)
> * @features: device features (RCAR_DU_FEATURE_*)
> * @quirks: device quirks (RCAR_DU_QUIRK_*)
> - * @num_crtcs: total number of CRTCs
> + * @channel_mask: bit mask of supported DU channels
Nitpicking, how about channels_mask ?
> * @routes: array of CRTC to output routes, indexed by output
> (RCAR_DU_OUTPUT_*) * @num_lvds: number of internal LVDS encoders
> */
> @@ -60,7 +60,7 @@ struct rcar_du_device_info {
> unsigned int gen;
> unsigned int features;
> unsigned int quirks;
> - unsigned int num_crtcs;
> + unsigned int channel_mask;
> struct rcar_du_output_routing routes[RCAR_DU_OUTPUT_MAX];
> unsigned int num_lvds;
> unsigned int dpll_ch;
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> b/drivers/gpu/drm/rcar-du/rcar_du_kms.c index cf5b422fc753..19a445fbc879
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> @@ -559,6 +559,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> struct drm_fbdev_cma *fbdev;
> unsigned int num_encoders;
> unsigned int num_groups;
> + unsigned int swi, hwi;
One variable per line please. I would also call them swindex and hwindex, that
would be clearer in my opinion.
> unsigned int i;
> int ret;
>
> @@ -571,7 +572,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> dev->mode_config.funcs = &rcar_du_mode_config_funcs;
> dev->mode_config.helper_private = &rcar_du_mode_config_helper;
>
> - rcdu->num_crtcs = rcdu->info->num_crtcs;
> + rcdu->num_crtcs = hweight8(rcdu->info->channel_mask);
>
> ret = rcar_du_properties_init(rcdu);
> if (ret < 0)
> @@ -581,7 +582,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> * Initialize vertical blanking interrupts handling. Start with vblank
> * disabled for all CRTCs.
> */
> - ret = drm_vblank_init(dev, (1 << rcdu->info->num_crtcs) - 1);
> + ret = drm_vblank_init(dev, (1 << rcdu->num_crtcs) - 1);
> if (ret < 0)
> return ret;
>
> @@ -623,10 +624,16 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> }
>
> /* Create the CRTCs. */
> - for (i = 0; i < rcdu->num_crtcs; ++i) {
> - struct rcar_du_group *rgrp = &rcdu->groups[i / 2];
> + for (swi = 0, hwi = 0; swi < rcdu->num_crtcs; ++hwi) {
> + struct rcar_du_group *rgrp;
> +
> + /* Skip unpopulated DU channels */
s/channels/channels./
> + if (!(rcdu->info->channel_mask & BIT(hwi)))
> + continue;
> +
> + rgrp = &rcdu->groups[hwi / 2];
>
> - ret = rcar_du_crtc_create(rgrp, i);
> + ret = rcar_du_crtc_create(rgrp, swi++, hwi);
> if (ret < 0)
> return ret;
> }
This is going to turn into an infinite loop if we ever get the num_crtcs
calculation wrong, but I don't see why that should be the case, so I'm OK with
the implementation.
With all those small issues fixed,
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
--
Regards,
Laurent Pinchart
WARNING: multiple messages have this Message-ID (diff)
From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Cc: linux-renesas-soc@vger.kernel.org,
David Airlie <airlied@linux.ie>,
open list <linux-kernel@vger.kernel.org>,
"open list:DRM DRIVERS FOR RENESAS"
<dri-devel@lists.freedesktop.org>
Subject: Re: [PATCH 05/17] drm: rcar-du: Split CRTC handling to support hardware indexing
Date: Thu, 26 Apr 2018 23:30:52 +0300 [thread overview]
Message-ID: <8207827.UIER1Ri0ED@avalon> (raw)
In-Reply-To: <20180426165346.494-6-kieran.bingham+renesas@ideasonboard.com>
Hi Kieran,
Thank you for the patch.
On Thursday, 26 April 2018 19:53:34 EEST Kieran Bingham wrote:
> The DU CRTC driver does not support distinguishing between a hardware
> index, and a software (CRTC) index in the event that a DU channel might
> not be populated by the hardware.
>
> Support this by adapting the rcar_du_device_info structure to store a
> bitmask of available channels rather than a count of CRTCs. The count
> can then be obtained by determining the hamming weight of the bitmask.
>
> This allows the rcar_du_crtc_create() function to distinguish between
> both index types, and non-populated DU channels will be skipped without
> leaving a gap in the software CRTC indexes.
>
> Signed-off-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
> ---
> drivers/gpu/drm/rcar-du/rcar_du_crtc.c | 26 ++++++++++++++------------
> drivers/gpu/drm/rcar-du/rcar_du_crtc.h | 3 ++-
> drivers/gpu/drm/rcar-du/rcar_du_drv.c | 20 ++++++++++----------
> drivers/gpu/drm/rcar-du/rcar_du_drv.h | 4 ++--
> drivers/gpu/drm/rcar-du/rcar_du_kms.c | 17 ++++++++++++-----
> 5 files changed, 40 insertions(+), 30 deletions(-)
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c index 5a15dfd66343..36ce194c13b5
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.c
> @@ -902,7 +902,8 @@ static irqreturn_t rcar_du_crtc_irq(int irq, void *arg)
> * Initialization
> */
>
> -int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int index)
> +int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int swindex,
> + unsigned int hwindex)
> {
> static const unsigned int mmio_offsets[] = {
> DU0_REG_OFFSET, DU1_REG_OFFSET, DU2_REG_OFFSET, DU3_REG_OFFSET
> @@ -910,7 +911,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> struct rcar_du_device *rcdu = rgrp->dev;
> struct platform_device *pdev = to_platform_device(rcdu->dev);
> - struct rcar_du_crtc *rcrtc = &rcdu->crtcs[index];
> + struct rcar_du_crtc *rcrtc = &rcdu->crtcs[swindex];
> struct drm_crtc *crtc = &rcrtc->crtc;
> struct drm_plane *primary;
> unsigned int irqflags;
> @@ -922,7 +923,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> /* Get the CRTC clock and the optional external clock. */
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_CRTC_IRQ_CLOCK)) {
> - sprintf(clk_name, "du.%u", index);
> + sprintf(clk_name, "du.%u", hwindex);
> name = clk_name;
> } else {
> name = NULL;
> @@ -930,16 +931,16 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> rcrtc->clock = devm_clk_get(rcdu->dev, name);
> if (IS_ERR(rcrtc->clock)) {
> - dev_err(rcdu->dev, "no clock for CRTC %u\n", index);
> + dev_err(rcdu->dev, "no clock for CRTC %u\n", swindex);
How about
dev_err(rcdu->dev, "no clock for DU channel %u\n", hwindex);
I think that would be clearer, because at this stage we're dealing with
hardware resources, so matching the datasheet numbers seems better to me.
> return PTR_ERR(rcrtc->clock);
> }
>
> - sprintf(clk_name, "dclkin.%u", index);
> + sprintf(clk_name, "dclkin.%u", hwindex);
> clk = devm_clk_get(rcdu->dev, clk_name);
> if (!IS_ERR(clk)) {
> rcrtc->extclock = clk;
> } else if (PTR_ERR(rcrtc->clock) == -EPROBE_DEFER) {
> - dev_info(rcdu->dev, "can't get external clock %u\n", index);
> + dev_info(rcdu->dev, "can't get external clock %u\n", hwindex);
> return -EPROBE_DEFER;
> }
>
> @@ -948,13 +949,13 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) spin_lock_init(&rcrtc->vblank_lock);
>
> rcrtc->group = rgrp;
> - rcrtc->mmio_offset = mmio_offsets[index];
> - rcrtc->index = index;
> + rcrtc->mmio_offset = mmio_offsets[hwindex];
> + rcrtc->index = hwindex;
>
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_VSP1_SOURCE))
> primary = &rcrtc->vsp->planes[rcrtc->vsp_pipe].plane;
> else
> - primary = &rgrp->planes[index % 2].plane;
> + primary = &rgrp->planes[hwindex % 2].plane;
This shouldn't make a difference because when RCAR_DU_FEATURE_VSP1_SOURCE
isn't set we're running on Gen2, and don't need to deal with indices, but from
a conceptual point of view, wouldn't the software index be better here ?
Missing hardware channels won't be visible from userspace, so taking the first
plane of the group as the primary plane would seem better to me.
> ret = drm_crtc_init_with_planes(rcdu->ddev, crtc, primary, NULL,
> rcdu->info->gen <= 2 ?
> @@ -970,7 +971,8 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index)
>
> /* Register the interrupt handler. */
> if (rcar_du_has(rcdu, RCAR_DU_FEATURE_CRTC_IRQ_CLOCK)) {
> - irq = platform_get_irq(pdev, index);
> + /* The IRQ's are associated with the CRTC (sw)index */
s/index/index./
> + irq = platform_get_irq(pdev, swindex);
> irqflags = 0;
> } else {
> irq = platform_get_irq(pdev, 0);
> @@ -978,7 +980,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) }
>
> if (irq < 0) {
> - dev_err(rcdu->dev, "no IRQ for CRTC %u\n", index);
> + dev_err(rcdu->dev, "no IRQ for CRTC %u\n", swindex);
> return irq;
> }
>
> @@ -986,7 +988,7 @@ int rcar_du_crtc_create(struct rcar_du_group *rgrp,
> unsigned int index) dev_name(rcdu->dev), rcrtc);
> if (ret < 0) {
> dev_err(rcdu->dev,
> - "failed to register IRQ for CRTC %u\n", index);
> + "failed to register IRQ for CRTC %u\n", swindex);
> return ret;
> }
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h index 518ee2c60eb8..5f003a16abc5
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_crtc.h
> @@ -99,7 +99,8 @@ enum rcar_du_output {
> RCAR_DU_OUTPUT_MAX,
> };
>
> -int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int index);
> +int rcar_du_crtc_create(struct rcar_du_group *rgrp, unsigned int swindex,
> + unsigned int hwindex);
> void rcar_du_crtc_suspend(struct rcar_du_crtc *rcrtc);
> void rcar_du_crtc_resume(struct rcar_du_crtc *rcrtc);
>
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> b/drivers/gpu/drm/rcar-du/rcar_du_drv.c index 05745e86d73e..d6ebc628fc22
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_drv.c
> @@ -40,7 +40,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7743_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
I'd write it BIT(1) | BIT(0) to match the usual little-endian order. Same
comment for the other info structure instances.
> .routes = {
> /*
> * R8A7743 has one RGB output and one LVDS output
> @@ -61,7 +61,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7745_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7745 has two RGB outputs
> @@ -80,7 +80,7 @@ static const struct rcar_du_device_info
> rzg1_du_r8a7745_info = { static const struct rcar_du_device_info
> rcar_du_r8a7779_info = {
> .gen = 2,
> .features = 0,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7779 has two RGB outputs and one (currently unsupported)
> @@ -102,7 +102,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7790_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> .quirks = RCAR_DU_QUIRK_ALIGN_128B,
> - .num_crtcs = 3,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2),
> .routes = {
> /*
> * R8A7790 has one RGB output, two LVDS outputs and one
> @@ -129,7 +129,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7791_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A779[13] has one RGB output, one LVDS output and one
> @@ -151,7 +151,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7792_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /* R8A7792 has two RGB outputs. */
> [RCAR_DU_OUTPUT_DPAD0] = {
> @@ -169,7 +169,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7794_info = { .gen = 2,
> .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS,
> - .num_crtcs = 2,
> + .channel_mask = BIT(0) | BIT(1),
> .routes = {
> /*
> * R8A7794 has two RGB outputs and one (currently unsupported)
> @@ -191,7 +191,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7795_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 4,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2) | BIT(3),
> .routes = {
> /*
> * R8A7795 has one RGB output, two HDMI outputs and one
> @@ -223,7 +223,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a7796_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 3,
> + .channel_mask = BIT(0) | BIT(1) | BIT(2),
> .routes = {
> /*
> * R8A7796 has one RGB output, one LVDS output and one HDMI
> @@ -251,7 +251,7 @@ static const struct rcar_du_device_info
> rcar_du_r8a77970_info = { .features = RCAR_DU_FEATURE_CRTC_IRQ_CLOCK
> | RCAR_DU_FEATURE_EXT_CTRL_REGS
> | RCAR_DU_FEATURE_VSP1_SOURCE,
> - .num_crtcs = 1,
> + .channel_mask = BIT(0),
> .routes = {
> /* R8A77970 has one RGB output and one LVDS output. */
> [RCAR_DU_OUTPUT_DPAD0] = {
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> b/drivers/gpu/drm/rcar-du/rcar_du_drv.h index 5c7ec15818c7..7a5de66deec2
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_drv.h
> @@ -52,7 +52,7 @@ struct rcar_du_output_routing {
> * @gen: device generation (2 or 3)
> * @features: device features (RCAR_DU_FEATURE_*)
> * @quirks: device quirks (RCAR_DU_QUIRK_*)
> - * @num_crtcs: total number of CRTCs
> + * @channel_mask: bit mask of supported DU channels
Nitpicking, how about channels_mask ?
> * @routes: array of CRTC to output routes, indexed by output
> (RCAR_DU_OUTPUT_*) * @num_lvds: number of internal LVDS encoders
> */
> @@ -60,7 +60,7 @@ struct rcar_du_device_info {
> unsigned int gen;
> unsigned int features;
> unsigned int quirks;
> - unsigned int num_crtcs;
> + unsigned int channel_mask;
> struct rcar_du_output_routing routes[RCAR_DU_OUTPUT_MAX];
> unsigned int num_lvds;
> unsigned int dpll_ch;
> diff --git a/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> b/drivers/gpu/drm/rcar-du/rcar_du_kms.c index cf5b422fc753..19a445fbc879
> 100644
> --- a/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> +++ b/drivers/gpu/drm/rcar-du/rcar_du_kms.c
> @@ -559,6 +559,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> struct drm_fbdev_cma *fbdev;
> unsigned int num_encoders;
> unsigned int num_groups;
> + unsigned int swi, hwi;
One variable per line please. I would also call them swindex and hwindex, that
would be clearer in my opinion.
> unsigned int i;
> int ret;
>
> @@ -571,7 +572,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> dev->mode_config.funcs = &rcar_du_mode_config_funcs;
> dev->mode_config.helper_private = &rcar_du_mode_config_helper;
>
> - rcdu->num_crtcs = rcdu->info->num_crtcs;
> + rcdu->num_crtcs = hweight8(rcdu->info->channel_mask);
>
> ret = rcar_du_properties_init(rcdu);
> if (ret < 0)
> @@ -581,7 +582,7 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> * Initialize vertical blanking interrupts handling. Start with vblank
> * disabled for all CRTCs.
> */
> - ret = drm_vblank_init(dev, (1 << rcdu->info->num_crtcs) - 1);
> + ret = drm_vblank_init(dev, (1 << rcdu->num_crtcs) - 1);
> if (ret < 0)
> return ret;
>
> @@ -623,10 +624,16 @@ int rcar_du_modeset_init(struct rcar_du_device *rcdu)
> }
>
> /* Create the CRTCs. */
> - for (i = 0; i < rcdu->num_crtcs; ++i) {
> - struct rcar_du_group *rgrp = &rcdu->groups[i / 2];
> + for (swi = 0, hwi = 0; swi < rcdu->num_crtcs; ++hwi) {
> + struct rcar_du_group *rgrp;
> +
> + /* Skip unpopulated DU channels */
s/channels/channels./
> + if (!(rcdu->info->channel_mask & BIT(hwi)))
> + continue;
> +
> + rgrp = &rcdu->groups[hwi / 2];
>
> - ret = rcar_du_crtc_create(rgrp, i);
> + ret = rcar_du_crtc_create(rgrp, swi++, hwi);
> if (ret < 0)
> return ret;
> }
This is going to turn into an infinite loop if we ever get the num_crtcs
calculation wrong, but I don't see why that should be the case, so I'm OK with
the implementation.
With all those small issues fixed,
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
--
Regards,
Laurent Pinchart
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel
next prev parent reply other threads:[~2018-04-26 20:30 UTC|newest]
Thread overview: 94+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-04-26 16:53 [PATCH 00/17] r8a77965: M3-N DU Enablement Kieran Bingham
2018-04-26 16:53 ` [PATCH 01/17] dt-bindings: display: renesas: du: Increase indent in output table Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:08 ` Laurent Pinchart
2018-04-26 20:08 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 02/17] dt-bindings: display: renesas: du: Document the R8A77965 bindings Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:57 ` Kieran Bingham
2018-04-26 16:57 ` Kieran Bingham
2018-04-26 20:10 ` Laurent Pinchart
2018-04-26 20:10 ` Laurent Pinchart
2018-04-27 8:40 ` Kieran Bingham
2018-04-27 8:40 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 03/17] pinctrl: sh-pfc: r8a77965: Add DU RGB output pins, groups and functions Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:16 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 04/17] drm: rcar-du: Use the correct naming for ODPM fields in DEFR6 Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:18 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 05/17] drm: rcar-du: Split CRTC handling to support hardware indexing Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:30 ` Laurent Pinchart [this message]
2018-04-26 20:30 ` Laurent Pinchart
2018-04-27 10:15 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 06/17] drm: rcar-du: Allow DU groups to work with " Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:36 ` Laurent Pinchart
2018-04-26 20:36 ` Laurent Pinchart
2018-04-27 10:10 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 07/17] drm: rcar-du: Add R8A77965 support Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:43 ` Laurent Pinchart
2018-04-26 20:43 ` Laurent Pinchart
2018-04-27 10:14 ` Kieran Bingham
2018-04-27 10:14 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 08/17] arm64: dts: r8a77965: Provide sysc header definitions Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 20:53 ` Laurent Pinchart
2018-04-26 20:53 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 09/17] arm64: dts: r8a77965: Use the correct CPG header Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:16 ` Laurent Pinchart
2018-04-26 21:16 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 10/17] arm64: dts: r8a77965: Add FCPF and FCPV instances Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:06 ` Laurent Pinchart
2018-04-26 21:06 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 11/17] arm64: dts: r8a77965: Add VSP instances Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:11 ` Laurent Pinchart
2018-04-26 21:11 ` Laurent Pinchart
2018-04-27 16:33 ` Kieran Bingham
2018-04-27 16:33 ` Kieran Bingham
2018-06-08 9:29 ` Geert Uytterhoeven
2018-06-08 9:29 ` Geert Uytterhoeven
2018-06-08 11:00 ` Laurent Pinchart
2018-06-08 11:00 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 12/17] arm64: dts: r8a77965: Populate the DU instance placeholder Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:15 ` Laurent Pinchart
2018-04-26 21:15 ` Laurent Pinchart
2018-04-27 16:34 ` Kieran Bingham
2018-04-27 16:34 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 13/17] arm64: dts: r8a77965: Add HDMI encoder instance Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:17 ` Laurent Pinchart
2018-04-26 21:17 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 14/17] arm64: dts: r8a77965-salvator-x: Add DU external dot clocks Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:18 ` Laurent Pinchart
2018-04-26 21:18 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 15/17] arm64: dts: r8a77965-salvator-x: Enable HDMI output Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:21 ` Laurent Pinchart
2018-04-26 21:21 ` Laurent Pinchart
2018-04-27 16:22 ` Kieran Bingham
2018-04-27 16:22 ` Kieran Bingham
2018-04-26 16:53 ` [PATCH 16/17] arm64: dts: r8a77965-salvator-xs: Add DU external dot clocks Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:22 ` Laurent Pinchart
2018-04-26 21:22 ` Laurent Pinchart
2018-04-26 16:53 ` [PATCH 17/17] arm64: dts: r8a77965-salvator-xs: Enable HDMI output Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 16:53 ` Kieran Bingham
2018-04-26 21:23 ` [PATCH 00/17] r8a77965: M3-N DU Enablement Laurent Pinchart
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=8207827.UIER1Ri0ED@avalon \
--to=laurent.pinchart@ideasonboard.com \
--cc=airlied@linux.ie \
--cc=dri-devel@lists.freedesktop.org \
--cc=kieran.bingham+renesas@ideasonboard.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-renesas-soc@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.