1 files changed, 464 insertions, 0 deletions
diff --git a/src/examples/video-src-alloc.c b/src/examples/video-src-alloc.c
new file mode 100644
index 0000000..ef364fd
--- /dev/null
+++ b/src/examples/video-src-alloc.c
@@ -0,0 +1,464 @@
+/* PipeWire
+ *
+ * Copyright © 2018 Wim Taymans
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ [title]
+ Allocating buffer memory and sending fds to the server.
+ [title]
+ */
+
+#include "config.h"
+
+#include <stdio.h>
+#include <errno.h>
+#include <signal.h>
+#include <math.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <sys/mman.h>
+
+#include <spa/param/video/format-utils.h>
+
+#include <pipewire/pipewire.h>
+
+#define BPP		3
+#define CURSOR_WIDTH	64
+#define CURSOR_HEIGHT	64
+#define CURSOR_BPP	4
+
+#define MAX_BUFFERS	64
+
+#define M_PI_M2 ( M_PI + M_PI )
+
+struct data {
+	struct pw_thread_loop *loop;
+	struct spa_source *timer;
+
+	struct pw_stream *stream;
+	struct spa_hook stream_listener;
+
+	struct spa_video_info_raw format;
+	int32_t stride;
+
+	int counter;
+	uint32_t seq;
+
+	double crop;
+	double accumulator;
+};
+
+static void draw_elipse(uint32_t *dst, int width, int height, uint32_t color)
+{
+	int i, j, r1, r2, r12, r22, r122;
+
+	r1 = width/2;
+	r12 = r1 * r1;
+	r2 = height/2;
+	r22 = r2 * r2;
+	r122 = r12 * r22;
+
+	for (i = -r2; i < r2; i++) {
+		for (j = -r1; j < r1; j++) {
+			dst[(i + r2)*width+(j+r1)] =
+				(i * i * r12 + j * j * r22 <= r122) ? color : 0x00000000;
+		}
+	}
+}
+
+/* called when we should push a new buffer in the queue */
+static void on_process(void *userdata)
+{
+	struct data *data = userdata;
+	struct pw_buffer *b;
+	struct spa_buffer *buf;
+	uint32_t i, j;
+	uint8_t *p;
+	struct spa_meta *m;
+	struct spa_meta_header *h;
+	struct spa_meta_region *mc;
+	struct spa_meta_cursor *mcs;
+
+	if ((b = pw_stream_dequeue_buffer(data->stream)) == NULL) {
+		pw_log_warn("out of buffers: %m");
+		return;
+	}
+
+	buf = b->buffer;
+	if ((p = buf->datas[0].data) == NULL)
+		return;
+
+	if ((h = spa_buffer_find_meta_data(buf, SPA_META_Header, sizeof(*h)))) {
+#if 0
+		struct timespec now;
+		clock_gettime(CLOCK_MONOTONIC, &now);
+		h->pts = SPA_TIMESPEC_TO_NSEC(&now);
+#else
+		h->pts = -1;
+#endif
+		h->flags = 0;
+		h->seq = data->seq++;
+		h->dts_offset = 0;
+	}
+	if ((m = spa_buffer_find_meta(buf, SPA_META_VideoDamage))) {
+		struct spa_meta_region *r = spa_meta_first(m);
+
+		if (spa_meta_check(r, m)) {
+			r->region.position = SPA_POINT(0,0);
+			r->region.size = data->format.size;
+			r++;
+		}
+		if (spa_meta_check(r, m))
+			r->region = SPA_REGION(0,0,0,0);
+	}
+	if ((mc = spa_buffer_find_meta_data(buf, SPA_META_VideoCrop, sizeof(*mc)))) {
+		data->crop = (sin(data->accumulator) + 1.0) * 32.0;
+		mc->region.position.x = data->crop;
+		mc->region.position.y = data->crop;
+		mc->region.size.width = data->format.size.width - data->crop*2;
+		mc->region.size.height = data->format.size.height - data->crop*2;
+	}
+	if ((mcs = spa_buffer_find_meta_data(buf, SPA_META_Cursor, sizeof(*mcs)))) {
+		struct spa_meta_bitmap *mb;
+		uint32_t *bitmap, color;
+
+		mcs->id = 1;
+		mcs->position.x = (sin(data->accumulator) + 1.0) * 160.0 + 80;
+		mcs->position.y = (cos(data->accumulator) + 1.0) * 100.0 + 50;
+		mcs->hotspot.x = 0;
+		mcs->hotspot.y = 0;
+		mcs->bitmap_offset = sizeof(struct spa_meta_cursor);
+
+		mb = SPA_PTROFF(mcs, mcs->bitmap_offset, struct spa_meta_bitmap);
+		mb->format = SPA_VIDEO_FORMAT_ARGB;
+		mb->size.width = CURSOR_WIDTH;
+		mb->size.height = CURSOR_HEIGHT;
+		mb->stride = CURSOR_WIDTH * CURSOR_BPP;
+		mb->offset = sizeof(struct spa_meta_bitmap);
+
+		bitmap = SPA_PTROFF(mb, mb->offset, uint32_t);
+		color = (cos(data->accumulator) + 1.0) * (1 << 23);
+		color |= 0xff000000;
+
+		draw_elipse(bitmap, mb->size.width, mb->size.height, color);
+	}
+
+	for (i = 0; i < data->format.size.height; i++) {
+		for (j = 0; j < data->format.size.width * BPP; j++) {
+			p[j] = data->counter + j * i;
+		}
+		p += data->stride;
+		data->counter += 13;
+	}
+
+	data->accumulator += M_PI_M2 / 50.0;
+	if (data->accumulator >= M_PI_M2)
+		data->accumulator -= M_PI_M2;
+
+	buf->datas[0].chunk->offset = 0;
+	buf->datas[0].chunk->size = data->format.size.height * data->stride;
+	buf->datas[0].chunk->stride = data->stride;
+
+	pw_stream_queue_buffer(data->stream, b);
+}
+
+/* trigger the graph when we are a driver */
+static void on_timeout(void *userdata, uint64_t expirations)
+{
+	struct data *data = userdata;
+	pw_log_trace("timeout");
+	pw_stream_trigger_process(data->stream);
+}
+
+/* when the stream is STREAMING, start the timer at 40ms intervals
+ * to produce and push a frame. In other states we PAUSE the timer. */
+static void on_stream_state_changed(void *_data, enum pw_stream_state old, enum pw_stream_state state,
+				    const char *error)
+{
+	struct data *data = _data;
+
+	printf("stream state: \"%s\"\n", pw_stream_state_as_string(state));
+
+	switch (state) {
+	case PW_STREAM_STATE_PAUSED:
+		printf("node id: %d\n", pw_stream_get_node_id(data->stream));
+		pw_loop_update_timer(pw_thread_loop_get_loop(data->loop),
+				data->timer, NULL, NULL, false);
+		break;
+	case PW_STREAM_STATE_STREAMING:
+	{
+		struct timespec timeout, interval;
+
+		timeout.tv_sec = 0;
+		timeout.tv_nsec = 1;
+		interval.tv_sec = 0;
+		interval.tv_nsec = 40 * SPA_NSEC_PER_MSEC;
+
+		if (pw_stream_is_driving(data->stream))
+			pw_loop_update_timer(pw_thread_loop_get_loop(data->loop),
+					data->timer, &timeout, &interval, false);
+		break;
+	}
+	default:
+		break;
+	}
+}
+
+/* we set the PW_STREAM_FLAG_ALLOC_BUFFERS flag when connecting so we need
+ * to provide buffer memory.  */
+static void on_stream_add_buffer(void *_data, struct pw_buffer *buffer)
+{
+	struct data *data = _data;
+	struct spa_buffer *buf = buffer->buffer;
+	struct spa_data *d;
+#ifdef HAVE_MEMFD_CREATE
+	unsigned int seals;
+#endif
+
+	pw_log_info("add buffer %p", buffer);
+	d = buf->datas;
+
+	if ((d[0].type & (1<<SPA_DATA_MemFd)) == 0) {
+		pw_log_error("unsupported data type %08x", d[0].type);
+		return;
+	}
+
+	/* create the memfd on the buffer, set the type and flags */
+	d[0].type = SPA_DATA_MemFd;
+	d[0].flags = SPA_DATA_FLAG_READWRITE;
+#ifdef HAVE_MEMFD_CREATE
+	d[0].fd = memfd_create("video-src-memfd", MFD_CLOEXEC | MFD_ALLOW_SEALING);
+#else
+	d[0].fd = -1;
+#endif
+	if (d[0].fd == -1) {
+		pw_log_error("can't create memfd: %m");
+		return;
+	}
+	d[0].mapoffset = 0;
+	d[0].maxsize = data->stride * data->format.size.height;
+
+	/* truncate to the right size before we set seals */
+	if (ftruncate(d[0].fd, d[0].maxsize) < 0) {
+		pw_log_error("can't truncate to %d: %m", d[0].maxsize);
+		return;
+	}
+#ifdef HAVE_MEMFD_CREATE
+	/* not enforced yet but server might require SEAL_SHRINK later */
+	seals = F_SEAL_GROW | F_SEAL_SHRINK | F_SEAL_SEAL;
+	if (fcntl(d[0].fd, F_ADD_SEALS, seals) == -1) {
+		pw_log_warn("Failed to add seals: %m");
+	}
+#endif
+
+	/* now mmap so we can write to it in the process function above */
+	d[0].data = mmap(NULL, d[0].maxsize, PROT_READ|PROT_WRITE,
+			MAP_SHARED, d[0].fd, d[0].mapoffset);
+	if (d[0].data == MAP_FAILED) {
+		pw_log_error("can't mmap memory: %m");
+		return;
+	}
+}
+
+/* close the memfd we set on the buffers here */
+static void on_stream_remove_buffer(void *_data, struct pw_buffer *buffer)
+{
+	struct spa_buffer *buf = buffer->buffer;
+	struct spa_data *d;
+
+	d = buf->datas;
+	pw_log_info("remove buffer %p", buffer);
+
+	munmap(d[0].data, d[0].maxsize);
+	close(d[0].fd);
+}
+
+/* Be notified when the stream param changes. We're only looking at the
+ * format param.
+ *
+ * We are now supposed to call pw_stream_update_params() with success or
+ * failure, depending on if we can support the format. Because we gave
+ * a list of supported formats, this should be ok.
+ *
+ * As part of pw_stream_update_params() we can provide parameters that
+ * will control the buffer memory allocation. This includes the metadata
+ * that we would like on our buffer, the size, alignment, etc.
+ */
+static void
+on_stream_param_changed(void *_data, uint32_t id, const struct spa_pod *param)
+{
+	struct data *data = _data;
+	struct pw_stream *stream = data->stream;
+	uint8_t params_buffer[1024];
+	struct spa_pod_builder b = SPA_POD_BUILDER_INIT(params_buffer, sizeof(params_buffer));
+	const struct spa_pod *params[5];
+
+	if (param == NULL || id != SPA_PARAM_Format)
+		return;
+
+	spa_format_video_raw_parse(param, &data->format);
+
+	data->stride = SPA_ROUND_UP_N(data->format.size.width * BPP, 4);
+
+	params[0] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_ParamBuffers, SPA_PARAM_Buffers,
+		SPA_PARAM_BUFFERS_buffers,  SPA_POD_CHOICE_RANGE_Int(8, 2, MAX_BUFFERS),
+		SPA_PARAM_BUFFERS_blocks,   SPA_POD_Int(1),
+		SPA_PARAM_BUFFERS_size,     SPA_POD_Int(data->stride * data->format.size.height),
+		SPA_PARAM_BUFFERS_stride,   SPA_POD_Int(data->stride),
+		SPA_PARAM_BUFFERS_dataType, SPA_POD_CHOICE_FLAGS_Int(1<<SPA_DATA_MemFd));
+
+	params[1] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta,
+		SPA_PARAM_META_type, SPA_POD_Id(SPA_META_Header),
+		SPA_PARAM_META_size, SPA_POD_Int(sizeof(struct spa_meta_header)));
+
+	params[2] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta,
+		SPA_PARAM_META_type, SPA_POD_Id(SPA_META_VideoDamage),
+		SPA_PARAM_META_size, SPA_POD_CHOICE_RANGE_Int(
+					sizeof(struct spa_meta_region) * 16,
+					sizeof(struct spa_meta_region) * 1,
+					sizeof(struct spa_meta_region) * 16));
+	params[3] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta,
+		SPA_PARAM_META_type, SPA_POD_Id(SPA_META_VideoCrop),
+		SPA_PARAM_META_size, SPA_POD_Int(sizeof(struct spa_meta_region)));
+#define CURSOR_META_SIZE(w,h)	(sizeof(struct spa_meta_cursor) + \
+				 sizeof(struct spa_meta_bitmap) + w * h * CURSOR_BPP)
+	params[4] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_ParamMeta, SPA_PARAM_Meta,
+		SPA_PARAM_META_type, SPA_POD_Id(SPA_META_Cursor),
+		SPA_PARAM_META_size, SPA_POD_Int(
+			CURSOR_META_SIZE(CURSOR_WIDTH,CURSOR_HEIGHT)));
+
+	pw_stream_update_params(stream, params, 5);
+}
+
+static const struct pw_stream_events stream_events = {
+	PW_VERSION_STREAM_EVENTS,
+	.process = on_process,
+	.state_changed = on_stream_state_changed,
+	.param_changed = on_stream_param_changed,
+	.add_buffer = on_stream_add_buffer,
+	.remove_buffer = on_stream_remove_buffer,
+};
+
+static void do_quit(void *userdata, int signal_number)
+{
+	struct data *data = userdata;
+	pw_thread_loop_signal(data->loop, false);
+}
+
+int main(int argc, char *argv[])
+{
+	struct data data = { 0, };
+	const struct spa_pod *params[1];
+	uint8_t buffer[1024];
+	struct spa_pod_builder b = SPA_POD_BUILDER_INIT(buffer, sizeof(buffer));
+
+	pw_init(&argc, &argv);
+
+	/* create a thread loop and start it */
+	data.loop = pw_thread_loop_new("video-src-alloc", NULL);
+
+	/* take the lock around all PipeWire functions. In callbacks, the lock
+	 * is already taken for you but it's ok to lock again because the lock is
+	 * recursive */
+	pw_thread_loop_lock(data.loop);
+
+	/* install some handlers to exit nicely */
+	pw_loop_add_signal(pw_thread_loop_get_loop(data.loop), SIGINT, do_quit, &data);
+	pw_loop_add_signal(pw_thread_loop_get_loop(data.loop), SIGTERM, do_quit, &data);
+
+	/* start after the signal handlers are set */
+	pw_thread_loop_start(data.loop);
+
+	/* create a simple stream, the simple stream manages the core
+	 * object for you if you don't want to deal with them.
+	 *
+	 * We're making a new video provider. We need to set the media-class
+	 * property.
+	 *
+	 * Pass your events and a user_data pointer as the last arguments. This
+	 * will inform you about the stream state. The most important event
+	 * you need to listen to is the process event where you need to provide
+	 * the data.
+	 */
+	data.stream = pw_stream_new_simple(
+			pw_thread_loop_get_loop(data.loop),
+			"video-src-alloc",
+			pw_properties_new(
+				PW_KEY_MEDIA_CLASS, "Video/Source",
+				NULL),
+			&stream_events,
+			&data);
+
+	/* make a timer to schedule our frames */
+	data.timer = pw_loop_add_timer(pw_thread_loop_get_loop(data.loop), on_timeout, &data);
+
+	/* build the extra parameter for the connection. Here we make an
+	 * EnumFormat parameter which lists the possible formats we can provide.
+	 * The server will select a format that matches and informs us about this
+	 * in the stream param_changed event.
+	 */
+	params[0] = spa_pod_builder_add_object(&b,
+		SPA_TYPE_OBJECT_Format, SPA_PARAM_EnumFormat,
+		SPA_FORMAT_mediaType,       SPA_POD_Id(SPA_MEDIA_TYPE_video),
+		SPA_FORMAT_mediaSubtype,    SPA_POD_Id(SPA_MEDIA_SUBTYPE_raw),
+		SPA_FORMAT_VIDEO_format,    SPA_POD_Id(SPA_VIDEO_FORMAT_RGB),
+		SPA_FORMAT_VIDEO_size,      SPA_POD_CHOICE_RANGE_Rectangle(
+						&SPA_RECTANGLE(320, 240),
+						&SPA_RECTANGLE(1, 1),
+						&SPA_RECTANGLE(4096, 4096)),
+		SPA_FORMAT_VIDEO_framerate, SPA_POD_Fraction(&SPA_FRACTION(25, 1)));
+
+	/* now connect the stream, we need a direction (input/output),
+	 * an optional target node to connect to, some flags and parameters.
+	 *
+	 * Here we pass PW_STREAM_FLAG_ALLOC_BUFFERS. We should in the
+	 * add_buffer callback configure the buffer memory. This should be
+	 * fd backed memory (memfd, dma-buf, ...) that can be shared with
+	 * the server.  */
+	pw_stream_connect(data.stream,
+			  PW_DIRECTION_OUTPUT,
+			  PW_ID_ANY,
+			  PW_STREAM_FLAG_DRIVER |
+			  PW_STREAM_FLAG_ALLOC_BUFFERS,
+			  params, 1);
+
+	/* unlock, run the loop and wait, this will trigger the callbacks */
+	pw_thread_loop_wait(data.loop);
+
+	/* unlock before stop */
+	pw_thread_loop_unlock(data.loop);
+	pw_thread_loop_stop(data.loop);
+
+	pw_stream_destroy(data.stream);
+
+	/* destroy after dependent objects are destroyed */
+	pw_thread_loop_destroy(data.loop);
+	pw_deinit();
+
+	return 0;
+}