[PATCH v2 18/22] drm/tegra: dc: Support memory bandwidth management

From: Dmitry Osipenko
Date: Sun Mar 29 2020 - 21:10:09 EST

Display controller (DC) performs isochronous memory transfers, and thus,
has a requirement for a minimum memory bandwidth that shall be fulfilled,
otherwise framebuffer data can't be fetched fast enough and this results
in a DC's data-FIFO underflow that follows by a visual corruption.

The Memory Controller drivers provide facility for memory bandwidth
management via interconnect API. This patch wires up the interconnect
API support to the DC driver.

Signed-off-by: Dmitry Osipenko <digetx@xxxxxxxxx>
---
drivers/gpu/drm/tegra/dc.c | 271 +++++++++++++++++++++++++++++++++-
drivers/gpu/drm/tegra/dc.h | 8 +
drivers/gpu/drm/tegra/drm.c | 19 +++
drivers/gpu/drm/tegra/plane.c | 1 +
drivers/gpu/drm/tegra/plane.h | 4 +-
5 files changed, 299 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/tegra/dc.c b/drivers/gpu/drm/tegra/dc.c
index 1a7b08f35776..b540ac6ffdc4 100644
--- a/drivers/gpu/drm/tegra/dc.c
+++ b/drivers/gpu/drm/tegra/dc.c
@@ -519,6 +519,136 @@ static void tegra_dc_setup_window(struct tegra_plane *plane,
tegra_plane_setup_blending(plane, window);
}

+static unsigned long
+tegra_plane_memory_bandwidth(struct drm_plane_state *state,
+ struct tegra_dc_window *window,
+ unsigned int num,
+ unsigned int denum)
+{
+ struct tegra_plane_state *tegra_state;
+ struct drm_crtc_state *crtc_state;
+ const struct drm_format_info *fmt;
+ struct tegra_dc_window win;
+ unsigned long long bandwidth;
+ unsigned int bpp_plane;
+ unsigned int bpp;
+ unsigned int mul;
+ unsigned int i;
+
+ if (!state->fb || !state->visible)
+ return 0;
+
+ crtc_state = drm_atomic_get_new_crtc_state(state->state, state->crtc);
+ tegra_state = to_tegra_plane_state(state);
+
+ if (!window)
+ window = &win;
+
+ window->src.w = drm_rect_width(&state->src) >> 16;
+ window->src.h = drm_rect_height(&state->src) >> 16;
+ window->dst.w = drm_rect_width(&state->dst);
+ window->dst.h = drm_rect_height(&state->dst);
+ window->tiling = tegra_state->tiling;
+
+ fmt = state->fb->format;
+
+ /*
+ * Note that real memory bandwidth vary depending on format and
+ * memory layout, we are not taking that into account because small
+ * estimation error isn't important since bandwidth is rounded up
+ * anyway.
+ */
+ for (i = 0, bpp = 0; i < fmt->num_planes; i++) {
+ bpp_plane = fmt->cpp[i] * 8;
+
+ /*
+ * Sub-sampling is relevant for chroma planes only and vertical
+ * readouts are not cached, hence only horizontal sub-sampling
+ * matters.
+ */
+ if (i > 0)
+ bpp_plane /= fmt->hsub;
+
+ bpp += bpp_plane;
+ }
+
+ /*
+ * Horizontal downscale takes extra bandwidth which roughly depends
+ * on the scaled width.
+ */
+ if (window->src.w > window->dst.w)
+ mul = (window->src.w - window->dst.w) * bpp / 2048 + 1;
+ else
+ mul = 1;
+
+ /*
+ * Ignore cursor window if its width is small enough such that
+ * data-prefetch FIFO will easily help to overcome temporal memory
+ * pressure.
+ *
+ * Window A has a 128bit x 128 deep read FIFO, while windows B/C
+ * have a 128bit x 64 deep read FIFO.
+ *
+ * This allows us to not overestimate memory frequency requirement.
+ * Even if it will happen that cursor gets a temporal underflow, this
+ * won't be fatal.
+ */
+ if (state->plane->type == DRM_PLANE_TYPE_CURSOR &&
+ mul == 1 && window->src.w * bpp <= 128 * 16)
+ return 0;
+
+ /* mode.clock in kHz, bandwidth in kbit/s */
+ bandwidth = kbps_to_icc(crtc_state->mode.clock * bpp * mul);
+
+ /* the requested bandwidth should be higher than required */
+ bandwidth *= num;
+ do_div(bandwidth, denum);
+
+ return min_t(u64, bandwidth, ULONG_MAX);
+}
+
+static unsigned long
+tegra20_plane_memory_bandwidth(struct drm_plane_state *state)
+{
+ return tegra_plane_memory_bandwidth(state, NULL, 29, 10);
+}
+
+static unsigned long
+tegra30_plane_memory_bandwidth(struct drm_plane_state *state)
+{
+ struct tegra_dc_window window;
+ unsigned long bandwidth;
+
+ bandwidth = tegra_plane_memory_bandwidth(state, &window, 29, 10);
+
+ /* x2: memory overfetch for tiled framebuffer and DDR3 */
+ if (window.tiling.mode == TEGRA_BO_TILING_MODE_TILED)
+ bandwidth *= 2;
+
+ return bandwidth;
+}
+
+static unsigned long
+tegra114_plane_memory_bandwidth(struct drm_plane_state *state)
+{
+ struct tegra_dc_window window;
+ unsigned long bandwidth;
+
+ bandwidth = tegra_plane_memory_bandwidth(state, &window, 12, 10);
+
+ /* x2: memory overfetch for tiled framebuffer and DDR3 */
+ if (window.tiling.mode == TEGRA_BO_TILING_MODE_TILED)
+ bandwidth *= 2;
+
+ return bandwidth;
+}
+
+static unsigned long
+tegra124_plane_memory_bandwidth(struct drm_plane_state *state)
+{
+ return tegra_plane_memory_bandwidth(state, NULL, 12, 10);
+}
+
static const u32 tegra20_primary_formats[] = {
DRM_FORMAT_ARGB4444,
DRM_FORMAT_ARGB1555,
@@ -608,8 +738,10 @@ static int tegra_plane_atomic_check(struct drm_plane *plane,
int err;

/* no need for further checks if the plane is being disabled */
- if (!state->crtc)
+ if (!state->crtc) {
+ plane_state->memory_bandwidth = 0;
return 0;
+ }

err = tegra_plane_format(state->fb->format->format,
&plane_state->format,
@@ -662,6 +794,8 @@ static int tegra_plane_atomic_check(struct drm_plane *plane,
if (err < 0)
return err;

+ plane_state->memory_bandwidth = dc->soc->plane_memory_bandwidth(state);
+
return 0;
}

@@ -1186,6 +1320,7 @@ tegra_crtc_atomic_duplicate_state(struct drm_crtc *crtc)
copy->pclk = state->pclk;
copy->div = state->div;
copy->planes = state->planes;
+ copy->memory_bandwidth = state->memory_bandwidth;

return &copy->base;
}
@@ -1777,6 +1912,8 @@ static void tegra_crtc_atomic_disable(struct drm_crtc *crtc,
err = host1x_client_suspend(&dc->client);
if (err < 0)
dev_err(dc->dev, "failed to suspend: %d\n", err);
+
+ icc_set_bw(dc->icc_bandwidth, 0, 0);
}

static void tegra_crtc_atomic_enable(struct drm_crtc *crtc,
@@ -1788,6 +1925,9 @@ static void tegra_crtc_atomic_enable(struct drm_crtc *crtc,
u32 value;
int err;

+ icc_set_bw(dc->icc_bandwidth, state->memory_bandwidth,
+ state->memory_bandwidth);
+
err = host1x_client_resume(&dc->client);
if (err < 0) {
dev_err(dc->dev, "failed to resume: %d\n", err);
@@ -1901,6 +2041,9 @@ static void tegra_crtc_atomic_enable(struct drm_crtc *crtc,
static void tegra_crtc_atomic_begin(struct drm_crtc *crtc,
struct drm_crtc_state *old_crtc_state)
{
+ struct tegra_dc_state *dc_old_state = to_dc_state(old_crtc_state);
+ struct tegra_dc_state *dc_state = to_dc_state(crtc->state);
+ struct tegra_dc *dc = to_tegra_dc(crtc);
unsigned long flags;

if (crtc->state->event) {
@@ -1915,6 +2058,25 @@ static void tegra_crtc_atomic_begin(struct drm_crtc *crtc,

crtc->state->event = NULL;
}
+
+ if (old_crtc_state && old_crtc_state->active) {
+ /*
+ * Raise memory bandwidth before changes take effect if it
+ * goes from low to high.
+ */
+ if (dc_old_state->memory_bandwidth < dc_state->memory_bandwidth)
+ icc_set_bw(dc->icc_bandwidth,
+ dc_state->memory_bandwidth,
+ dc_state->memory_bandwidth);
+ } else {
+ /*
+ * Raise memory bandwidth before changes take effect if
+ * CRTC is turning on.
+ */
+ icc_set_bw(dc->icc_bandwidth,
+ dc_state->memory_bandwidth,
+ dc_state->memory_bandwidth);
+ }
}

static void tegra_crtc_atomic_flush(struct drm_crtc *crtc,
@@ -1933,7 +2095,80 @@ static void tegra_crtc_atomic_flush(struct drm_crtc *crtc,
value = tegra_dc_readl(dc, DC_CMD_STATE_CONTROL);
}

+static bool
+tegra_plane_overlaps_other_plane(struct drm_crtc_state *state,
+ const struct drm_plane_state *plane_state)
+{
+ const struct drm_plane_state *other_state;
+ struct drm_plane *plane;
+ struct drm_rect rect;
+
+ drm_atomic_crtc_state_for_each_plane_state(plane, other_state, state) {
+ rect = plane_state->dst;
+
+ if (other_state == plane_state)
+ continue;
+
+ if (!other_state->visible || !other_state->fb)
+ continue;
+
+ if (drm_rect_intersect(&rect, &other_state->dst))
+ return true;
+ }
+
+ return false;
+}
+
+static int tegra_crtc_atomic_check(struct drm_crtc *crtc,
+ struct drm_crtc_state *state)
+{
+ struct tegra_dc_state *dc_state = to_dc_state(state);
+ const struct drm_plane_state *plane_state;
+ const struct tegra_plane_state *tegra;
+ unsigned long long bandwidth = 0;
+ struct drm_plane *plane;
+
+ /*
+ * For overlapping planes pixel's data is fetched for each plane at
+ * the same time, hence bandwidth is accumulated in this case.
+ */
+ drm_atomic_crtc_state_for_each_plane_state(plane, plane_state, state) {
+ tegra = to_tegra_plane_state(plane_state);
+
+ if (tegra_plane_overlaps_other_plane(state, plane_state))
+ bandwidth += tegra->memory_bandwidth;
+ else
+ bandwidth = max_t(u64, bandwidth,
+ tegra->memory_bandwidth);
+ }
+
+ dc_state->memory_bandwidth = min_t(u64, bandwidth, U32_MAX);
+
+ return 0;
+}
+
+void tegra_crtc_atomic_post_commit(struct drm_crtc *crtc,
+ struct drm_crtc_state *old_crtc_state)
+{
+ struct tegra_dc_state *dc_old_state = to_dc_state(old_crtc_state);
+ struct tegra_dc_state *dc_state = to_dc_state(crtc->state);
+ struct tegra_dc *dc = to_tegra_dc(crtc);
+
+ if (!dc_old_state)
+ return;
+
+ /*
+ * Drop memory bandwidth after changes take effect if it goes from
+ * high to low.
+ */
+ if (dc_old_state->memory_bandwidth > dc_state->memory_bandwidth)
+ icc_set_bw(dc->icc_bandwidth,
+ dc_state->memory_bandwidth,
+ dc_state->memory_bandwidth);
+}
+
static const struct drm_crtc_helper_funcs tegra_crtc_helper_funcs = {
+ .atomic_check = tegra_crtc_atomic_check,
.atomic_begin = tegra_crtc_atomic_begin,
.atomic_flush = tegra_crtc_atomic_flush,
.atomic_enable = tegra_crtc_atomic_enable,
@@ -2225,6 +2460,7 @@ static const struct tegra_dc_soc_info tegra20_dc_soc_info = {
.modifiers = tegra20_modifiers,
.has_win_a_without_filters = true,
.has_win_c_without_vert_filter = true,
+ .plane_memory_bandwidth = tegra20_plane_memory_bandwidth,
};

static const struct tegra_dc_soc_info tegra30_dc_soc_info = {
@@ -2244,6 +2480,7 @@ static const struct tegra_dc_soc_info tegra30_dc_soc_info = {
.modifiers = tegra20_modifiers,
.has_win_a_without_filters = false,
.has_win_c_without_vert_filter = false,
+ .plane_memory_bandwidth = tegra30_plane_memory_bandwidth,
};

static const struct tegra_dc_soc_info tegra114_dc_soc_info = {
@@ -2263,6 +2500,7 @@ static const struct tegra_dc_soc_info tegra114_dc_soc_info = {
.modifiers = tegra20_modifiers,
.has_win_a_without_filters = false,
.has_win_c_without_vert_filter = false,
+ .plane_memory_bandwidth = tegra114_plane_memory_bandwidth,
};

static const struct tegra_dc_soc_info tegra124_dc_soc_info = {
@@ -2282,6 +2520,7 @@ static const struct tegra_dc_soc_info tegra124_dc_soc_info = {
.modifiers = tegra124_modifiers,
.has_win_a_without_filters = false,
.has_win_c_without_vert_filter = false,
+ .plane_memory_bandwidth = tegra124_plane_memory_bandwidth,
};

static const struct tegra_dc_soc_info tegra210_dc_soc_info = {
@@ -2301,6 +2540,7 @@ static const struct tegra_dc_soc_info tegra210_dc_soc_info = {
.modifiers = tegra124_modifiers,
.has_win_a_without_filters = false,
.has_win_c_without_vert_filter = false,
+ .plane_memory_bandwidth = tegra124_plane_memory_bandwidth,
};

static const struct tegra_windowgroup_soc tegra186_dc_wgrps[] = {
@@ -2349,6 +2589,7 @@ static const struct tegra_dc_soc_info tegra186_dc_soc_info = {
.has_nvdisplay = true,
.wgrps = tegra186_dc_wgrps,
.num_wgrps = ARRAY_SIZE(tegra186_dc_wgrps),
+ .plane_memory_bandwidth = tegra124_plane_memory_bandwidth,
};

static const struct tegra_windowgroup_soc tegra194_dc_wgrps[] = {
@@ -2397,6 +2638,7 @@ static const struct tegra_dc_soc_info tegra194_dc_soc_info = {
.has_nvdisplay = true,
.wgrps = tegra194_dc_wgrps,
.num_wgrps = ARRAY_SIZE(tegra194_dc_wgrps),
+ .plane_memory_bandwidth = tegra124_plane_memory_bandwidth,
};

static const struct of_device_id tegra_dc_of_match[] = {
@@ -2503,6 +2745,7 @@ static int tegra_dc_couple(struct tegra_dc *dc)

static int tegra_dc_probe(struct platform_device *pdev)
{
+ const char *level = KERN_ERR;
struct tegra_dc *dc;
int err;

@@ -2571,8 +2814,6 @@ static int tegra_dc_probe(struct platform_device *pdev)

err = tegra_dc_rgb_probe(dc);
if (err < 0 && err != -ENODEV) {
- const char *level = KERN_ERR;
-
if (err == -EPROBE_DEFER)
level = KERN_DEBUG;

@@ -2581,6 +2822,25 @@ static int tegra_dc_probe(struct platform_device *pdev)
return err;
}

+ /*
+ * The display controller memory bandwidth management isn't trivial
+ * because it requires the knowledge about the DC hardware state
+ * in order to make a proper decisions. It's not easy to convey
+ * that information to the ICC provider, so we will just use the
+ * first interconnect path for the memory bandwidth management and
+ * make all the decisions within the DC driver, for simplicity.
+ */
+ dc->icc_bandwidth = of_icc_get(dc->dev, "display0a");
+ err = PTR_ERR_OR_ZERO(dc->icc_bandwidth);
+ if (err) {
+ if (err == -EPROBE_DEFER)
+ level = KERN_DEBUG;
+
+ dev_printk(level, dc->dev,
+ "failed to get display0a interconnect: %d\n", err);
+ goto remove_rgb;
+ }
+
platform_set_drvdata(pdev, dc);
pm_runtime_enable(&pdev->dev);

@@ -2599,6 +2859,9 @@ static int tegra_dc_probe(struct platform_device *pdev)

disable_pm:
pm_runtime_disable(&pdev->dev);
+ icc_put(dc->icc_bandwidth);
+
+remove_rgb:
tegra_dc_rgb_remove(dc);

return err;
@@ -2616,6 +2879,8 @@ static int tegra_dc_remove(struct platform_device *pdev)
return err;
}

+ icc_put(dc->icc_bandwidth);
+
err = tegra_dc_rgb_remove(dc);
if (err < 0) {
dev_err(&pdev->dev, "failed to remove RGB output: %d\n", err);
diff --git a/drivers/gpu/drm/tegra/dc.h b/drivers/gpu/drm/tegra/dc.h
index 3d8ddccd758f..3a0ff57c5169 100644
--- a/drivers/gpu/drm/tegra/dc.h
+++ b/drivers/gpu/drm/tegra/dc.h
@@ -8,6 +8,7 @@
#define TEGRA_DC_H 1

#include <linux/host1x.h>
+#include <linux/interconnect.h>

#include <drm/drm_crtc.h>

@@ -23,6 +24,8 @@ struct tegra_dc_state {
unsigned int div;

u32 planes;
+
+ unsigned long memory_bandwidth;
};

static inline struct tegra_dc_state *to_dc_state(struct drm_crtc_state *state)
@@ -66,6 +69,7 @@ struct tegra_dc_soc_info {
const u64 *modifiers;
bool has_win_a_without_filters;
bool has_win_c_without_vert_filter;
+ unsigned long (*plane_memory_bandwidth)(struct drm_plane_state *state);
};

struct tegra_dc {
@@ -90,6 +94,8 @@ struct tegra_dc {
struct drm_info_list *debugfs_files;

const struct tegra_dc_soc_info *soc;
+
+ struct icc_path *icc_bandwidth;
};

static inline struct tegra_dc *
@@ -150,6 +156,8 @@ int tegra_dc_state_setup_clock(struct tegra_dc *dc,
struct drm_crtc_state *crtc_state,
struct clk *clk, unsigned long pclk,
unsigned int div);
+void tegra_crtc_atomic_post_commit(struct drm_crtc *crtc,
+ struct drm_crtc_state *old_crtc_state);

/* from rgb.c */
int tegra_dc_rgb_probe(struct tegra_dc *dc);
diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
index bd268028fb3d..44f558adddff 100644
--- a/drivers/gpu/drm/tegra/drm.c
+++ b/drivers/gpu/drm/tegra/drm.c
@@ -20,6 +20,7 @@
#include <drm/drm_prime.h>
#include <drm/drm_vblank.h>

+#include "dc.h"
#include "drm.h"
#include "gem.h"

@@ -59,6 +60,22 @@ static const struct drm_mode_config_funcs tegra_drm_mode_config_funcs = {
.atomic_commit = drm_atomic_helper_commit,
};

+static void tegra_atomic_post_commit(struct drm_device *drm,
+ struct drm_atomic_state *old_state)
+{
+ struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+ struct drm_crtc *crtc;
+ unsigned int i;
+
+ for_each_oldnew_crtc_in_state(old_state, crtc, old_crtc_state,
+ new_crtc_state, i) {
+ if (!new_crtc_state->active)
+ continue;
+
+ tegra_crtc_atomic_post_commit(crtc, old_crtc_state);
+ }
+}
+
static void tegra_atomic_commit_tail(struct drm_atomic_state *old_state)
{
struct drm_device *drm = old_state->dev;
@@ -75,6 +92,8 @@ static void tegra_atomic_commit_tail(struct drm_atomic_state *old_state)
} else {
drm_atomic_helper_commit_tail_rpm(old_state);
}
+
+ tegra_atomic_post_commit(drm, old_state);
}

static const struct drm_mode_config_helper_funcs
diff --git a/drivers/gpu/drm/tegra/plane.c b/drivers/gpu/drm/tegra/plane.c
index 9ccfb56e9b01..f94925744105 100644
--- a/drivers/gpu/drm/tegra/plane.c
+++ b/drivers/gpu/drm/tegra/plane.c
@@ -63,6 +63,7 @@ tegra_plane_atomic_duplicate_state(struct drm_plane *plane)
copy->swap = state->swap;
copy->bottom_up = state->bottom_up;
copy->opaque = state->opaque;
+ copy->memory_bandwidth = state->memory_bandwidth;

for (i = 0; i < 2; i++)
copy->blending[i] = state->blending[i];
diff --git a/drivers/gpu/drm/tegra/plane.h b/drivers/gpu/drm/tegra/plane.h
index a158a915109a..5227c7a9ad8b 100644
--- a/drivers/gpu/drm/tegra/plane.h
+++ b/drivers/gpu/drm/tegra/plane.h
@@ -51,10 +51,12 @@ struct tegra_plane_state {
/* used for legacy blending support only */
struct tegra_plane_legacy_blending_state blending[2];
bool opaque;
+
+ unsigned long memory_bandwidth;
};

static inline struct tegra_plane_state *
-to_tegra_plane_state(struct drm_plane_state *state)
+to_tegra_plane_state(const struct drm_plane_state *state)
{
if (state)
return container_of(state, struct tegra_plane_state, base);
--
2.25.1