tight2: Use worker per stream rather than worker per tile

tight-encoder-v2
Andri Yngvason 2020-07-09 11:48:25 +00:00
parent ed054fe00d
commit cd2c296fa9
2 changed files with 60 additions and 86 deletions

View File

@ -7,6 +7,7 @@
struct tight_tile;
struct pixman_region16;
struct aml_work;
struct tight_encoder_v2 {
uint32_t width;
@ -17,9 +18,7 @@ struct tight_encoder_v2 {
struct tight_tile* grid;
z_stream zs[4];
uint8_t zs_mask;
pthread_mutex_t zs_mutex;
pthread_cond_t zs_cond;
struct aml_work* zs_worker[4];
const struct rfb_pixel_format* dfmt;
const struct rfb_pixel_format* sfmt;

View File

@ -39,12 +39,19 @@ enum tight_tile_state {
struct tight_tile {
enum tight_tile_state state;
struct tight_encoder_v2* parent;
size_t size;
uint8_t type;
char buffer[MAX_TILE_SIZE];
};
struct tight_zs_worker_ctx {
struct tight_encoder_v2* encoder;
int index;
};
static void do_tight_zs_work(void*);
static void on_tight_zs_work_done(void*);
static int tight_encoder_v2_init_stream(z_stream* zs)
{
int rc = deflateInit2(zs,
@ -74,6 +81,27 @@ static inline uint32_t tight_tile_height(struct tight_encoder_v2* self,
return y + TSL > self->height ? self->height - y : TSL;
}
static int tight_init_zs_worker(struct tight_encoder_v2* self, int index)
{
struct tight_zs_worker_ctx* ctx = calloc(1, sizeof(*ctx));
if (!ctx)
return -1;
ctx->encoder = self;
ctx->index = index;
self->zs_worker[index] =
aml_work_new(do_tight_zs_work, on_tight_zs_work_done, ctx, free);
if (!self->zs_worker[index])
goto failure;
return 0;
failure:
free(ctx);
return -1;
}
int tight_encoder_v2_init(struct tight_encoder_v2* self, uint32_t width,
uint32_t height)
{
@ -90,17 +118,15 @@ int tight_encoder_v2_init(struct tight_encoder_v2* self, uint32_t width,
if (!self->grid)
return -1;
for (uint32_t y = 0; y < self->grid_height; ++y)
for (uint32_t x = 0; x < self->grid_width; ++x)
tight_tile(self, x, y)->parent = self;
tight_encoder_v2_init_stream(&self->zs[0]);
tight_encoder_v2_init_stream(&self->zs[1]);
tight_encoder_v2_init_stream(&self->zs[2]);
tight_encoder_v2_init_stream(&self->zs[3]);
pthread_mutex_init(&self->zs_mutex, NULL);
pthread_cond_init(&self->zs_cond, NULL);
tight_init_zs_worker(self, 0);
tight_init_zs_worker(self, 1);
tight_init_zs_worker(self, 2);
tight_init_zs_worker(self, 3);
pthread_mutex_init(&self->dst_mutex, NULL);
@ -117,8 +143,10 @@ void tight_encoder_v2_destroy(struct tight_encoder_v2* self)
pthread_mutex_destroy(&self->dst_mutex);
pthread_cond_destroy(&self->zs_cond);
pthread_mutex_destroy(&self->zs_mutex);
aml_unref(self->zs_worker[3]);
aml_unref(self->zs_worker[2]);
aml_unref(self->zs_worker[1]);
aml_unref(self->zs_worker[0]);
deflateEnd(&self->zs[3]);
deflateEnd(&self->zs[2]);
@ -172,41 +200,6 @@ static int calc_bytes_per_cpixel(const struct rfb_pixel_format* fmt)
: fmt->bits_per_pixel / 8;
}
static int tight_acquire_zstream_unlocked(struct tight_encoder_v2* self)
{
int i;
for (i = 0; i < 4; ++i)
if (!(self->zs_mask & (1 << i)))
break;
if (i >= 4)
return -1;
self->zs_mask |= (1 << i);
return i;
}
static int tight_acquire_zstream(struct tight_encoder_v2* self)
{
pthread_mutex_lock(&self->zs_mutex);
int i;
while ((i = tight_acquire_zstream_unlocked(self)) < 0)
pthread_cond_wait(&self->zs_cond, &self->zs_mutex);
pthread_mutex_unlock(&self->zs_mutex);
return i;
}
static void tight_release_zstream(struct tight_encoder_v2* self, int index)
{
pthread_mutex_lock(&self->zs_mutex);
self->zs_mask &= ~(1 << index);
pthread_cond_signal(&self->zs_cond);
pthread_mutex_unlock(&self->zs_mutex);
}
static int tight_deflate(struct tight_tile* tile, void* src,
size_t len, z_stream* zs, bool flush)
{
@ -234,12 +227,9 @@ static int tight_deflate(struct tight_tile* tile, void* src,
static void tight_encode_tile_basic(struct tight_encoder_v2* self,
struct tight_tile* tile, uint32_t x, uint32_t y_start,
uint32_t width, uint32_t height)
uint32_t width, uint32_t height, int zs_index)
{
// TODO Figure out of way to postpone job instead of blocking
int zs_index = tight_acquire_zstream(self);
z_stream* zs = &self->zs[zs_index];
tile->type = TIGHT_BASIC | TIGHT_STREAM(zs_index);
int bytes_per_cpixel = calc_bytes_per_cpixel(self->dfmt);
@ -270,11 +260,9 @@ static void tight_encode_tile_basic(struct tight_encoder_v2* self,
}
static void tight_encode_tile(struct tight_encoder_v2* self,
struct tight_tile* tile)
uint32_t gx, uint32_t gy, int zs_index)
{
intptr_t index = ((intptr_t)tile - (intptr_t)self->grid) / sizeof(*tile);
uint32_t gx = index % self->grid_width;
uint32_t gy = index / self->grid_width;
struct tight_tile* tile = tight_tile(self, gx, gy);
uint32_t x = gx * TSL;
uint32_t y = gy * TSL;
@ -284,7 +272,7 @@ static void tight_encode_tile(struct tight_encoder_v2* self,
tile->size = 0;
tight_encode_tile_basic(self, tile, x, y, width, height);
tight_encode_tile_basic(self, tile, x, y, width, height, zs_index);
//TODO Jpeg
struct rfb_server_fb_rect rect = {
@ -296,15 +284,10 @@ static void tight_encode_tile(struct tight_encoder_v2* self,
};
pthread_mutex_lock(&self->dst_mutex);
// TODO: Check if the tile is basic before releasing stream
tight_release_zstream(self, (tile->type >> 4) & 3);
vec_append(self->dst, &rect, sizeof(rect));
vec_append(self->dst, &tile->type, sizeof(tile->type));
tight_encode_size(self->dst, tile->size);
vec_append(self->dst, tile->buffer, tile->size);
pthread_mutex_unlock(&self->dst_mutex);
tile->state = TIGHT_TILE_READY;
@ -320,17 +303,22 @@ static int tight_encode_rect_count(struct tight_encoder_v2* self)
return vec_append(self->dst, &msg, sizeof(msg));
}
static void do_encode_tile(void* obj)
static void do_tight_zs_work(void* obj)
{
struct tight_tile* tile = aml_get_userdata(obj);
struct tight_encoder_v2* self = tile->parent;
tight_encode_tile(self, tile);
struct tight_zs_worker_ctx* ctx = aml_get_userdata(obj);
struct tight_encoder_v2* self = ctx->encoder;
int index = ctx->index;
for (uint32_t y = 0; y < self->grid_height; ++y)
for (uint32_t x = index; x < self->grid_width; x += 4)
if (tight_tile(self, x, y)->state == TIGHT_TILE_DAMAGED)
tight_encode_tile(self, x, y, index);
}
static void on_encode_tile_done(void* obj)
static void on_tight_zs_work_done(void* obj)
{
struct tight_tile* tile = aml_get_userdata(obj);
struct tight_encoder_v2* self = tile->parent;
struct tight_zs_worker_ctx* ctx = aml_get_userdata(obj);
struct tight_encoder_v2* self = ctx->encoder;
pthread_mutex_lock(&self->wait_mutex);
if (--self->n_jobs == 0)
@ -338,33 +326,20 @@ static void on_encode_tile_done(void* obj)
pthread_mutex_unlock(&self->wait_mutex);
}
static int tight_schedule_encode_tile(struct tight_encoder_v2* self,
uint32_t x, uint32_t y)
static int tight_schedule_zs_work(struct tight_encoder_v2* self, int index)
{
struct tight_tile* tile = tight_tile(self, x, y);
if (tile->state != TIGHT_TILE_DAMAGED)
return 0;
struct aml_work* work = aml_work_new(do_encode_tile,
on_encode_tile_done, tile, NULL);
if (!work)
return -1;
int rc = aml_start(aml_get_default(), work);
int rc = aml_start(aml_get_default(), self->zs_worker[index]);
if (rc >= 0)
++self->n_jobs;
aml_unref(work);
return rc;
}
static int tight_schedule_encoding_jobs(struct tight_encoder_v2* self)
{
for (uint32_t y = 0; y < self->grid_height; ++y)
for (uint32_t x = 0; x < self->grid_width; ++x)
if (tight_schedule_encode_tile(self, x, y) < 0)
return -1;
for (int i = 0; i < 4; ++i)
if (tight_schedule_zs_work(self, i) < 0)
return -1;
return 0;
}