[Intel-gfx] [PATCH i-g-t 1/3] skl_ddb_allocation: Add a standalone version of the DDB allocator

Damien Lespiau damien.lespiau at intel.com
Tue Oct 14 18:33:58 CEST 2014


This is useful to debug the allocation algorithm.

Signed-off-by: Damien Lespiau <damien.lespiau at intel.com>
---
 tests/.gitignore           |   1 +
 tests/Makefile.sources     |   1 +
 tests/skl_ddb_allocation.c | 378 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 380 insertions(+)
 create mode 100644 tests/skl_ddb_allocation.c

diff --git a/tests/.gitignore b/tests/.gitignore
index bb90685..3b1fb8b 100644
--- a/tests/.gitignore
+++ b/tests/.gitignore
@@ -149,5 +149,6 @@ prime_nv_test
 prime_self_import
 prime_udl
 single-tests.txt
+skl_ddb_allocation
 template
 testdisplay
diff --git a/tests/Makefile.sources b/tests/Makefile.sources
index 5202ab2..89596e6 100644
--- a/tests/Makefile.sources
+++ b/tests/Makefile.sources
@@ -2,6 +2,7 @@ noinst_PROGRAMS = \
 	gem_alive \
 	gem_stress \
 	ddi_compute_wrpll \
+	skl_ddb_allocation \
 	$(TESTS_progs) \
 	$(TESTS_progs_M) \
 	$(HANG) \
diff --git a/tests/skl_ddb_allocation.c b/tests/skl_ddb_allocation.c
new file mode 100644
index 0000000..7faed0e
--- /dev/null
+++ b/tests/skl_ddb_allocation.c
@@ -0,0 +1,378 @@
+#include <stdint.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <string.h>
+
+/**
+ * container_of - cast a member of a structure out to the containing structure
+ * @ptr:	the pointer to the member.
+ * @type:	the type of the container struct this is embedded in.
+ * @member:	the name of the member within the struct.
+ *
+ */
+#define container_of(ptr, type, member) ({			\
+	typeof( ((type *)0)->member ) *__mptr = (ptr);	\
+	(type *)( (char *)__mptr - offsetof(type,member) );})
+
+#define div_u64(a, b)	((a) / (b))
+
+/*
+ * Stub a few defines/structures
+ */
+
+#define I915_MAX_PIPES	3
+#define I915_MAX_PLANES	3
+#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
+
+#define for_each_pipe(p) for ((p) = 0; (p) < 3; (p)++)
+#define for_each_plane(pipe, p) for ((p) = 0; (p) < 3; (p)++)
+
+#define for_each_crtc(dev, crtc) \
+	for (int i = 0; i < 3 && (crtc = &crtcs[i].base); i++)
+
+#define for_each_intel_crtc(dev, intel_crtc) \
+	for (int i = 0; i < 3, intel_crtc = &crtcs[i]; i++)
+
+enum pipe {
+	PIPE_A,
+	PIPE_B,
+	PIPE_C,
+};
+
+enum plane {
+	PLANE_1,
+	PLANE_2,
+	PLANE_3,
+};
+
+#define pipe_name(p) ((p) + 'A')
+
+struct drm_device {
+	void *dev_private;
+};
+
+struct drm_i915_private {
+	struct drm_device *dev;
+};
+
+struct drm_crtc {
+	struct drm_device *dev;
+	bool active;
+};
+
+static bool intel_crtc_active(struct drm_crtc *crtc)
+{
+	return crtc->active;
+}
+
+struct intel_crtc {
+	struct drm_crtc base;
+	enum pipe pipe;
+};
+
+static int intel_num_planes(struct intel_crtc *crtc)
+{
+	return 3;
+}
+
+struct intel_crtc crtcs[I915_MAX_PIPES];
+
+#define to_intel_crtc(x) container_of(x, struct intel_crtc, base)
+
+/*
+ * DDB code
+ */
+
+struct intel_wm_config {
+	unsigned int num_pipes_active;
+};
+
+struct intel_plane_wm_parameters {
+	uint32_t horiz_pixels;
+	uint32_t vert_pixels;
+	uint8_t bytes_per_pixel;
+	bool enabled;
+	bool scaled;
+};
+
+struct skl_pipe_wm_parameters {
+	bool active;
+	uint32_t pipe_htotal;
+	uint32_t pixel_rate; /* in KHz */
+	struct intel_plane_wm_parameters plane[I915_MAX_PLANES];
+	struct intel_plane_wm_parameters cursor;
+};
+
+struct skl_ddb_entry {
+	uint16_t start, end;	/* in number of blocks */
+};
+
+static inline uint16_t skl_ddb_entry_size(const struct skl_ddb_entry *entry)
+{
+	/* end not set, clearly no allocation here. start can be 0 though */
+	if (entry->end == 0)
+		return 0;
+
+	return entry->end - entry->start + 1;
+}
+
+static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1,
+				       const struct skl_ddb_entry *e2)
+{
+	if (e1->start == e2->start && e1->end == e2->end)
+		return true;
+
+	return false;
+}
+
+struct skl_ddb_allocation {
+	struct skl_ddb_entry plane[I915_MAX_PIPES][I915_MAX_PLANES];
+	struct skl_ddb_entry cursor[I915_MAX_PIPES];
+};
+
+/*
+ * On gen9, we need to allocate Display Data Buffer (DDB) portions to the
+ * different active planes.
+ */
+
+#define SKL_DDB_SIZE		896	/* in blocks */
+
+static void
+skl_ddb_get_pipe_allocation_limits(struct drm_device *dev,
+				   struct drm_crtc *for_crtc,
+				   const struct intel_wm_config *config,
+				   const struct skl_pipe_wm_parameters *params,
+				   struct skl_ddb_entry *alloc /* out */)
+{
+	struct drm_crtc *crtc;
+	unsigned int pipe_size, ddb_size;
+	int nth_active_pipe;
+
+	if (!params->active) {
+		alloc->start = 0;
+		alloc->end = 0;
+		return;
+	}
+
+	ddb_size = SKL_DDB_SIZE;
+	ddb_size -= 4; /* 4 blocks for bypass path allocation */
+
+	nth_active_pipe = 0;
+	for_each_crtc(dev, crtc) {
+		if (!intel_crtc_active(crtc))
+			continue;
+
+		if (crtc == for_crtc)
+			break;
+
+		nth_active_pipe++;
+	}
+
+	pipe_size = ddb_size / config->num_pipes_active;
+	alloc->start = nth_active_pipe * ddb_size / config->num_pipes_active;
+	alloc->end = alloc->start + pipe_size - 1;
+}
+
+static unsigned int skl_cursor_allocation(const struct intel_wm_config *config)
+{
+	if (config->num_pipes_active == 1)
+		return 32;
+
+	return 8;
+}
+
+static unsigned int
+skl_plane_relative_data_rate(const struct intel_plane_wm_parameters *p)
+{
+	return p->horiz_pixels * p->vert_pixels * p->bytes_per_pixel;
+}
+
+/*
+ * We don't overflow 32 bits. Worst case is 3 planes enabled, each fetching
+ * a 8192x4096 at 32bpp framebuffer:
+ *   3 * 4096 * 8192  * 4 < 2^32
+ */
+static unsigned int
+skl_get_total_relative_data_rate(struct intel_crtc *intel_crtc,
+				 const struct skl_pipe_wm_parameters *params)
+{
+	unsigned int total_data_rate = 0;
+	int plane;
+
+	for (plane = 0; plane < intel_num_planes(intel_crtc); plane++) {
+		const struct intel_plane_wm_parameters *p;
+
+		p = &params->plane[plane];
+		if (!p->enabled)
+			continue;
+
+		total_data_rate += skl_plane_relative_data_rate(p);
+	}
+
+	return total_data_rate;
+}
+
+static void
+skl_allocate_pipe_ddb(struct drm_crtc *crtc,
+		      const struct intel_wm_config *config,
+		      const struct skl_pipe_wm_parameters *params,
+		      struct skl_ddb_allocation *ddb /* out */)
+{
+	struct drm_device *dev = crtc->dev;
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	enum pipe pipe = intel_crtc->pipe;
+	struct skl_ddb_entry alloc;
+	uint16_t alloc_size, start, cursor_blocks;
+	unsigned int total_data_rate;
+	int plane;
+
+	skl_ddb_get_pipe_allocation_limits(dev, crtc, config, params, &alloc);
+	alloc_size = skl_ddb_entry_size(&alloc);
+	if (alloc_size == 0) {
+		memset(ddb->plane[pipe], 0, sizeof(ddb->plane[pipe]));
+		memset(&ddb->cursor[pipe], 0, sizeof(ddb->cursor[pipe]));
+		return;
+	}
+
+	cursor_blocks = skl_cursor_allocation(config);
+	ddb->cursor[pipe].start = alloc.end - cursor_blocks + 1;
+	ddb->cursor[pipe].end = alloc.end;
+
+	alloc_size -= cursor_blocks;
+	alloc.end -= cursor_blocks;
+
+	/*
+	 * Each active plane get a portion of the remaining space, in
+	 * proportion to the amount of data they need to fetch from memory.
+	 *
+	 * FIXME: we may not allocate every single block here.
+	 */
+	total_data_rate = skl_get_total_relative_data_rate(intel_crtc, params);
+
+	start = alloc.start;
+	for (plane = 0; plane < intel_num_planes(intel_crtc); plane++) {
+		const struct intel_plane_wm_parameters *p;
+		unsigned int data_rate;
+		uint16_t plane_blocks;
+
+		p = &params->plane[plane];
+		if (!p->enabled)
+			continue;
+
+		data_rate = skl_plane_relative_data_rate(p);
+
+		/*
+		 * promote the expression to 64 bits to avoid overflowing, the
+		 * result is < available as data_rate / total_data_rate < 1
+		 */
+		plane_blocks = div_u64((uint64_t)alloc_size * data_rate,
+				       total_data_rate);
+
+		ddb->plane[pipe][plane].start = start;
+		ddb->plane[pipe][plane].end = start + plane_blocks - 1;
+
+		start += plane_blocks;
+	}
+
+}
+
+static void skl_ddb_print(struct skl_ddb_allocation *ddb)
+{
+	struct skl_ddb_entry *entry;
+	enum pipe pipe;
+	int plane;
+
+	printf("%-15s%8s%8s%8s\n", "", "Start", "End", "Size");
+
+	for_each_pipe(pipe) {
+		printf("Pipe %c\n", pipe_name(pipe));
+
+		for_each_plane(pipe, plane) {
+			entry = &ddb->plane[pipe][plane];
+			printf("  Plane%-8d%8u%8u%8u\n", plane + 1,
+			       entry->start, entry->end,
+			       skl_ddb_entry_size(entry));
+		}
+
+		entry = &ddb->cursor[pipe];
+		printf("  %-13s%8u%8u%8u\n", "Cursor", entry->start,
+		       entry->end, skl_ddb_entry_size(entry));
+	}
+
+
+}
+
+static struct drm_device drm_device;
+static struct drm_i915_private drm_i915_private;
+
+static void init_stub(void)
+{
+	int i;
+
+	drm_device.dev_private = &drm_i915_private;
+	drm_i915_private.dev = &drm_device;
+
+	for (i = 0; i < I915_MAX_PIPES; i++) {
+		crtcs[i].base.dev = &drm_device;
+		crtcs[i].pipe = i;
+	}
+}
+
+struct wm_input {
+	struct intel_wm_config config;
+	struct skl_pipe_wm_parameters params[I915_MAX_PIPES];
+};
+
+static void wm_input_reset(struct wm_input *in)
+{
+	memset(in, 0, sizeof(*in));
+}
+
+static void wm_enable_plane(struct wm_input *in,
+			    enum pipe pipe, enum plane plane,
+			    uint32_t width, uint32_t height, int bpp)
+{
+	enum pipe i;
+
+	in->params[pipe].active = 1;
+
+	in->config.num_pipes_active = 0;
+	for_each_pipe(i)
+		if (in->params[i].active)
+			in->config.num_pipes_active++;
+
+	in->params[pipe].plane[plane].horiz_pixels = width;
+	in->params[pipe].plane[plane].vert_pixels = height;
+	in->params[pipe].plane[plane].bytes_per_pixel = bpp;
+	in->params[pipe].plane[plane].enabled = true;
+}
+
+static void skl_ddb_allocate(struct wm_input *in,
+			     struct skl_ddb_allocation *out)
+{
+	struct drm_crtc *crtc;
+
+	for_each_crtc(, crtc) {
+		enum pipe pipe = to_intel_crtc(crtc)->pipe;
+
+		skl_allocate_pipe_ddb(crtc,
+				      &in->config, &in->params[pipe], out);
+	}
+}
+
+int main(int argc, char **argv)
+{
+	struct wm_input in;
+	static struct skl_ddb_allocation ddb;
+
+	init_stub();
+
+	wm_input_reset(&in);
+	wm_enable_plane(&in, PIPE_A, PLANE_1, 1280, 1024, 4);
+	wm_enable_plane(&in, PIPE_A, PLANE_2,  100,  100, 4);
+	skl_ddb_allocate(&in, &ddb);
+	skl_ddb_print(&ddb);
+
+	return 0;
+}
-- 
1.8.3.1




More information about the Intel-gfx mailing list