pipewire/src/pipewire/impl-node.c
Wim Taymans 45e3af5cdc impl-port: make passive mode as an enum
Hopefully easier to understand.
2026-03-12 13:46:49 +01:00

2922 lines
84 KiB
C

/* PipeWire */
/* SPDX-FileCopyrightText: Copyright © 2018 Wim Taymans */
/* SPDX-License-Identifier: MIT */
#include "config.h"
#include <string.h>
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <errno.h>
#include <time.h>
#include <malloc.h>
#include <limits.h>
#include <spa/support/system.h>
#include <spa/pod/parser.h>
#include <spa/pod/filter.h>
#include <spa/pod/dynamic.h>
#include <spa/node/utils.h>
#include <spa/debug/types.h>
#include <spa/utils/string.h>
#include <spa/utils/json-pod.h>
#define PW_API_NODE_IMPL SPA_EXPORT
#include "pipewire/impl-node.h"
#include "pipewire/private.h"
PW_LOG_TOPIC_EXTERN(log_node);
#define PW_LOG_TOPIC_DEFAULT log_node
#define DEFAULT_SYNC_TIMEOUT ((uint64_t)(5 * SPA_NSEC_PER_SEC))
/** \cond */
struct impl {
struct pw_impl_node this;
enum pw_node_state pending_state;
uint32_t pending_id;
struct pw_work_queue *work;
int last_error;
struct spa_list param_list;
struct spa_list pending_list;
unsigned int cache_params:1;
unsigned int pending_play:1;
struct spa_command *pending_request_process;
char *group;
char *link_group;
char *sync_group;
};
static const char * const global_keys[] = {
PW_KEY_OBJECT_PATH,
PW_KEY_MODULE_ID,
PW_KEY_FACTORY_ID,
PW_KEY_CLIENT_ID,
PW_KEY_CLIENT_API,
PW_KEY_DEVICE_ID,
PW_KEY_PRIORITY_SESSION,
PW_KEY_PRIORITY_DRIVER,
PW_KEY_APP_NAME,
PW_KEY_NODE_DESCRIPTION,
PW_KEY_NODE_NAME,
PW_KEY_NODE_NICK,
PW_KEY_NODE_SESSION,
PW_KEY_MEDIA_CLASS,
PW_KEY_MEDIA_TYPE,
PW_KEY_MEDIA_CATEGORY,
PW_KEY_MEDIA_ROLE,
NULL
};
#define pw_node_resource(r,m,v,...) pw_resource_call(r,struct pw_node_events,m,v,__VA_ARGS__)
#define pw_node_resource_info(r,...) pw_node_resource(r,info,0,__VA_ARGS__)
#define pw_node_resource_param(r,...) pw_node_resource(r,param,0,__VA_ARGS__)
struct resource_data {
struct pw_impl_node *node;
struct pw_resource *resource;
struct spa_hook resource_listener;
struct spa_hook object_listener;
uint32_t subscribe_ids[MAX_PARAMS];
uint32_t n_subscribe_ids;
/* for async replies */
int seq;
int end;
struct spa_hook listener;
};
SPA_EXPORT
struct pw_node_peer *pw_node_peer_ref(struct pw_impl_node *onode, struct pw_impl_node *inode)
{
struct pw_node_peer *peer;
spa_list_for_each(peer, &onode->peer_list, link) {
if (peer->target.id == inode->info.id) {
pw_log_debug("exiting peer %p from %p to %p", peer, onode, inode);
peer->ref++;
return peer;
}
}
peer = calloc(1, sizeof(*peer));
if (peer == NULL)
return NULL;
peer->ref = 1;
peer->output = onode;
copy_target(&peer->target, &inode->rt.target);
spa_list_append(&onode->peer_list, &peer->link);
pw_log_debug("new peer %p from %p to %p", peer, onode, inode);
pw_impl_node_add_target(onode, &peer->target);
return peer;
}
SPA_EXPORT
void pw_node_peer_unref(struct pw_node_peer *peer)
{
if (peer == NULL || --peer->ref > 0)
return;
spa_list_remove(&peer->link);
pw_log_debug("remove peer %p from %p to %p", peer, peer->output, peer->target.node);
pw_impl_node_remove_target(peer->output, &peer->target);
free(peer);
}
static inline void activate_target(struct pw_impl_node *node, struct pw_node_target *t)
{
struct pw_node_activation_state *state = &t->activation->state[0];
if (!t->active) {
if (!node->async) {
if (!node->exported) {
SPA_ATOMIC_INC(state->required);
SPA_ATOMIC_INC(state->pending);
}
}
t->active = true;
pw_log_debug("%p: target state:%p id:%d pending:%d/%d %d:%d:%d",
node, state, t->id, state->pending, state->required,
node->async, node->driving, node->exported);
}
}
static inline void deactivate_target(struct pw_impl_node *node, struct pw_node_target *t, uint64_t trigger)
{
if (t->active) {
struct pw_node_activation_state *state = &t->activation->state[0];
if (!node->async) {
/* the driver copies the required to the pending state
* so first try to resume the node and then decrement the
* required state. This way we either resume with the old value
* or we don't when the driver has not yet copied */
if (trigger != 0)
t->trigger(t, trigger);
if (!node->exported)
SPA_ATOMIC_DEC(state->required);
}
t->active = false;
pw_log_debug("%p: target state:%p id:%d pending:%d/%d %d:%d:%d trigger:%"PRIu64,
node, state, t->id, state->pending, state->required,
node->async, node->driving, node->exported, trigger);
}
}
/* called from data-loop when all the targets of a node need to be triggered */
static inline void trigger_targets(struct pw_impl_node *node, int status, uint64_t nsec)
{
struct pw_node_target *ta;
pw_log_trace_fp("%p: (%s-%u) trigger targets %"PRIu64,
node, node->name, node->info.id, nsec);
spa_list_for_each(ta, &node->rt.target_list, link)
ta->trigger(ta, nsec);
}
/** \endcond */
/* Called from the node data loop when a node needs to be scheduled by
* the given driver.
*
* - the node adds the source to the data loop
* - from all the targets of the node, we can now be scheduled ourselves
* so activate our target.
* - When we get scheduled, we will activate our peer targets
*/
static int
do_node_prepare(struct spa_loop *loop, bool async, uint32_t seq,
const void *data, size_t size, void *user_data)
{
struct pw_impl_node *this = user_data;
struct pw_node_target *t;
uint64_t dummy;
int res;
pw_log_trace("%p: prepare %d remote:%d exported:%d", this, this->rt.prepared,
this->remote, this->exported);
if (this->rt.prepared)
return 0;
if (!this->remote) {
/* clear the eventfd in case it was written to while the node was stopped */
res = spa_system_eventfd_read(this->rt.target.system, this->source.fd, &dummy);
if (SPA_UNLIKELY(res != -EAGAIN && res != 0))
pw_log_warn("%p: read failed %m", this);
spa_loop_add_source(loop, &this->source);
}
if (!this->remote || this->rt.target.activation->client_version < 1)
SPA_ATOMIC_STORE(this->rt.target.activation->status, PW_NODE_ACTIVATION_FINISHED);
spa_list_for_each(t, &this->rt.target_list, link)
activate_target(this, t);
this->rt.prepared = true;
return 0;
}
static void add_node_to_graph(struct pw_impl_node *node)
{
pw_loop_locked(node->data_loop, do_node_prepare, 1, NULL, 0, node);
}
/* called from the node data loop and undoes the changes done in do_node_prepare. */
static int
do_node_unprepare(struct spa_loop *loop, bool async, uint32_t seq,
const void *data, size_t size, void *user_data)
{
struct pw_impl_node *this = user_data;
struct pw_node_target *t;
int old_state;
uint64_t trigger = 0;
pw_log_trace("%p: unprepare %d remote:%d exported:%d", this, this->rt.prepared,
this->remote, this->exported);
if (!this->rt.prepared)
return 0;
/* The remote client will INACTIVE itself and remove itself from the loop to avoid
* being scheduled.
* The server will mark remote nodes as FINISHED and trigger the peers. This will
* make sure the remote node will not trigger the peers anymore when it will
* stop (it only triggers peers when it has PENDING_TRIGGER (<= AWAKE)). We have
* to trigger the peers on the server because the client might simply be dead and
* not able to trigger anything.
*/
old_state = SPA_ATOMIC_XCHG(this->rt.target.activation->status, PW_NODE_ACTIVATION_INACTIVE);
if (PW_NODE_ACTIVATION_PENDING_TRIGGER(old_state))
trigger = get_time_ns(this->rt.target.system);
if (!this->remote)
spa_loop_remove_source(loop, &this->source);
spa_list_for_each(t, &this->rt.target_list, link)
deactivate_target(this, t, trigger);
this->rt.prepared = false;
return 0;
}
static void remove_node_from_graph(struct pw_impl_node *node)
{
pw_loop_locked(node->data_loop, do_node_unprepare, 1, NULL, 0, node);
}
static void node_deactivate(struct pw_impl_node *this)
{
struct pw_impl_port *port;
struct pw_impl_link *link;
pw_log_debug("%p: deactivate", this);
/* make sure the node doesn't get woken up while not active */
remove_node_from_graph(this);
spa_list_for_each(port, &this->input_ports, link) {
spa_list_for_each(link, &port->links, input_link)
pw_impl_link_deactivate(link);
}
spa_list_for_each(port, &this->output_ports, link) {
spa_list_for_each(link, &port->links, output_link)
pw_impl_link_deactivate(link);
}
}
static int idle_node(struct pw_impl_node *this)
{
struct impl *impl = SPA_CONTAINER_OF(this, struct impl, this);
int res = 0;
pw_log_debug("%p: idle node state:%s pending:%s pause-on-idle:%d", this,
pw_node_state_as_string(this->info.state),
pw_node_state_as_string(impl->pending_state),
this->pause_on_idle);
if (impl->pending_state <= PW_NODE_STATE_IDLE)
return 0;
if (!this->pause_on_idle)
return 0;
node_deactivate(this);
res = spa_node_send_command(this->node,
&SPA_NODE_COMMAND_INIT(SPA_NODE_COMMAND_Pause));
if (res < 0)
pw_log_debug("%p: pause node error %s", this, spa_strerror(res));
return res;
}
static void node_activate(struct pw_impl_node *this)
{
struct pw_impl_port *port;
pw_log_debug("%p: activate", this);
spa_list_for_each(port, &this->output_ports, link) {
struct pw_impl_link *link;
spa_list_for_each(link, &port->links, output_link)
pw_impl_link_activate(link);
}
spa_list_for_each(port, &this->input_ports, link) {
struct pw_impl_link *link;
spa_list_for_each(link, &port->links, input_link)
pw_impl_link_activate(link);
}
}
static int start_node(struct pw_impl_node *this)
{
struct impl *impl = SPA_CONTAINER_OF(this, struct impl, this);
int res = 0;
node_activate(this);
if (impl->pending_state >= PW_NODE_STATE_RUNNING)
return 0;
pw_log_debug("%p: start node driving:%d driver:%d prepared:%d", this,
this->driving, this->driver, this->rt.prepared);
this->lazy = this->rt.position && SPA_FLAG_IS_SET(this->rt.position->clock.flags,
SPA_IO_CLOCK_FLAG_LAZY);
if (!(this->driving && this->driver)) {
impl->pending_play = true;
res = spa_node_send_command(this->node,
&SPA_NODE_COMMAND_INIT(SPA_NODE_COMMAND_Start));
} else {
/* driver nodes will wait until all other nodes are started before
* they are started */
spa_clear_ptr(impl->pending_request_process, free);
res = EBUSY;
}
if (res < 0)
pw_log_error("(%s-%u) start node error %d: %s", this->name, this->info.id,
res, spa_strerror(res));
return res;
}
static void emit_info_changed(struct pw_impl_node *node, bool flags_changed)
{
if (node->info.change_mask == 0 && !flags_changed)
return;
pw_impl_node_emit_info_changed(node, &node->info);
if (node->global && node->info.change_mask != 0) {
struct pw_resource *resource;
if (node->info.change_mask & PW_NODE_CHANGE_MASK_PROPS)
pw_global_update_keys(node->global, node->info.props, global_keys);
spa_list_for_each(resource, &node->global->resource_list, link)
pw_node_resource_info(resource, &node->info);
}
node->info.change_mask = 0;
}
static int resource_is_subscribed(struct pw_resource *resource, uint32_t id)
{
struct resource_data *data = pw_resource_get_user_data(resource);
uint32_t i;
for (i = 0; i < data->n_subscribe_ids; i++) {
if (data->subscribe_ids[i] == id)
return 1;
}
return 0;
}
static int notify_param(void *data, int seq, uint32_t id,
uint32_t index, uint32_t next, struct spa_pod *param)
{
struct pw_impl_node *node = data;
struct pw_resource *resource;
spa_list_for_each(resource, &node->global->resource_list, link) {
if (!resource_is_subscribed(resource, id))
continue;
pw_log_debug("%p: resource %p notify param %d", node, resource, id);
pw_node_resource_param(resource, seq, id, index, next, param);
}
return 0;
}
static void emit_params(struct pw_impl_node *node, uint32_t *changed_ids, uint32_t n_changed_ids)
{
uint32_t i;
int res;
if (node->global == NULL)
return;
pw_log_debug("%p: emit %d params", node, n_changed_ids);
for (i = 0; i < n_changed_ids; i++) {
struct pw_resource *resource;
int subscribed = 0;
/* first check if anyone is subscribed */
spa_list_for_each(resource, &node->global->resource_list, link) {
if ((subscribed = resource_is_subscribed(resource, changed_ids[i])))
break;
}
if (!subscribed)
continue;
if ((res = pw_impl_node_for_each_param(node, 1, changed_ids[i], 0, UINT32_MAX,
NULL, notify_param, node)) < 0) {
pw_log_error("%p: error %d (%s)", node, res, spa_strerror(res));
}
}
}
static void node_update_state(struct pw_impl_node *node, enum pw_node_state state, int res, char *error)
{
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
enum pw_node_state old = node->info.state;
bool emit_pending_request_process = false;
switch (state) {
case PW_NODE_STATE_RUNNING:
pw_log_debug("%p: start node driving:%d driver:%d prepared:%d", node,
node->driving, node->driver, node->rt.prepared);
if (res >= 0) {
add_node_to_graph(node);
}
if (node->driving && node->driver) {
res = spa_node_send_command(node->node,
&SPA_NODE_COMMAND_INIT(SPA_NODE_COMMAND_Start));
if (res < 0) {
state = PW_NODE_STATE_ERROR;
error = spa_aprintf("Start error: %s", spa_strerror(res));
remove_node_from_graph(node);
} else if (impl->pending_request_process != NULL) {
emit_pending_request_process = true;
}
}
break;
case PW_NODE_STATE_IDLE:
case PW_NODE_STATE_SUSPENDED:
case PW_NODE_STATE_ERROR:
if (state != PW_NODE_STATE_IDLE || node->pause_on_idle)
if (old != PW_NODE_STATE_CREATING)
remove_node_from_graph(node);
break;
default:
break;
}
free((char*)node->info.error);
node->info.error = error;
node->info.state = state;
impl->pending_state = state;
pw_log_debug("%p: (%s) %s -> %s (%s)", node, node->name,
pw_node_state_as_string(old), pw_node_state_as_string(state), error);
if (old == state)
return;
if (state == PW_NODE_STATE_ERROR) {
pw_log_error("(%s-%u) %s -> error (%s)", node->name, node->info.id,
pw_node_state_as_string(old), error);
} else {
pw_log_info("(%s-%u) %s -> %s", node->name, node->info.id,
pw_node_state_as_string(old), pw_node_state_as_string(state));
}
pw_impl_node_emit_state_changed(node, old, state, error);
if (emit_pending_request_process) {
pw_log_debug("%p: request process:%p", node, impl->pending_request_process);
spa_node_send_command(node->node, impl->pending_request_process);
spa_clear_ptr(impl->pending_request_process, free);
}
node->info.change_mask |= PW_NODE_CHANGE_MASK_STATE;
emit_info_changed(node, false);
if (state == PW_NODE_STATE_ERROR && node->global) {
struct pw_resource *resource;
spa_list_for_each(resource, &node->global->resource_list, link)
pw_resource_error(resource, res, error);
}
if (old == PW_NODE_STATE_RUNNING &&
state == PW_NODE_STATE_IDLE &&
node->suspend_on_idle) {
pw_impl_node_set_state(node, PW_NODE_STATE_SUSPENDED);
}
}
static int suspend_node(struct pw_impl_node *this)
{
int res = 0;
struct pw_impl_port *p;
pw_log_debug("%p: suspend node state:%s", this,
pw_node_state_as_string(this->info.state));
if (this->info.state > 0 && this->info.state <= PW_NODE_STATE_SUSPENDED)
return 0;
spa_list_for_each(p, &this->input_ports, link) {
if (p->busy_count > 0) {
pw_log_debug("%p: can't suspend, input port %d busy:%d",
this, p->port_id, p->busy_count);
return -EBUSY;
}
}
spa_list_for_each(p, &this->output_ports, link) {
if (p->busy_count > 0) {
pw_log_debug("%p: can't suspend, output port %d busy:%d",
this, p->port_id, p->busy_count);
return -EBUSY;
}
}
node_deactivate(this);
pw_log_debug("%p: suspend node driving:%d driver:%d prepared:%d", this,
this->driving, this->driver, this->rt.prepared);
res = spa_node_send_command(this->node,
&SPA_NODE_COMMAND_INIT(SPA_NODE_COMMAND_Suspend));
if (res == -ENOTSUP)
res = spa_node_send_command(this->node,
&SPA_NODE_COMMAND_INIT(SPA_NODE_COMMAND_Pause));
if (res < 0 && res != -EIO)
pw_log_warn("%p: suspend node error %s", this, spa_strerror(res));
spa_list_for_each(p, &this->input_ports, link) {
if ((res = pw_impl_port_set_param(p, SPA_PARAM_Format, 0, NULL)) < 0)
pw_log_warn("%p: error unset format input: %s",
this, spa_strerror(res));
/* force CONFIGURE in case of async */
p->state = PW_IMPL_PORT_STATE_CONFIGURE;
}
spa_list_for_each(p, &this->output_ports, link) {
if ((res = pw_impl_port_set_param(p, SPA_PARAM_Format, 0, NULL)) < 0)
pw_log_warn("%p: error unset format output: %s",
this, spa_strerror(res));
/* force CONFIGURE in case of async */
p->state = PW_IMPL_PORT_STATE_CONFIGURE;
}
node_update_state(this, PW_NODE_STATE_SUSPENDED, 0, NULL);
return res;
}
static void
clear_info(struct pw_impl_node *this)
{
pw_free_strv(this->groups);
pw_free_strv(this->link_groups);
pw_free_strv(this->sync_groups);
free(this->name);
free((char*)this->info.error);
}
static int reply_param(void *data, int seq, uint32_t id,
uint32_t index, uint32_t next, struct spa_pod *param)
{
struct resource_data *d = data;
pw_log_debug("%p: resource %p reply param %d", d->node, d->resource, seq);
pw_node_resource_param(d->resource, seq, id, index, next, param);
return 0;
}
static int node_enum_params(void *object, int seq, uint32_t id,
uint32_t index, uint32_t num, const struct spa_pod *filter)
{
struct resource_data *data = object;
struct pw_resource *resource = data->resource;
struct pw_impl_node *node = data->node;
int res;
pw_log_debug("%p: resource %p enum params seq:%d id:%d (%s) index:%u num:%u",
node, resource, seq, id,
spa_debug_type_find_name(spa_type_param, id), index, num);
if ((res = pw_impl_node_for_each_param(node, seq, id, index, num,
filter, reply_param, data)) < 0) {
pw_resource_errorf(resource, res,
"enum params id:%d (%s) failed", id,
spa_debug_type_find_name(spa_type_param, id));
}
return 0;
}
static int node_subscribe_params(void *object, uint32_t *ids, uint32_t n_ids)
{
struct resource_data *data = object;
struct pw_resource *resource = data->resource;
uint32_t i;
n_ids = SPA_MIN(n_ids, SPA_N_ELEMENTS(data->subscribe_ids));
data->n_subscribe_ids = n_ids;
for (i = 0; i < n_ids; i++) {
data->subscribe_ids[i] = ids[i];
pw_log_debug("%p: resource %p subscribe param id:%d (%s)",
data->node, resource, ids[i],
spa_debug_type_find_name(spa_type_param, ids[i]));
node_enum_params(data, 1, ids[i], 0, UINT32_MAX, NULL);
}
return 0;
}
static void remove_busy_resource(struct resource_data *d)
{
if (d->end != -1) {
spa_hook_remove(&d->listener);
d->end = -1;
pw_impl_client_set_busy(d->resource->client, false);
}
}
static void result_node_sync(void *data, int seq, int res, uint32_t type, const void *result)
{
struct resource_data *d = data;
pw_log_debug("%p: sync result %d %d (%d/%d)", d->node, res, seq, d->seq, d->end);
if (seq == d->end)
remove_busy_resource(d);
}
static int node_set_param(void *object, uint32_t id, uint32_t flags,
const struct spa_pod *param)
{
struct resource_data *data = object;
struct pw_resource *resource = data->resource;
struct pw_impl_node *node = data->node;
struct pw_impl_client *client = resource->client;
int res;
static const struct spa_node_events node_events = {
SPA_VERSION_NODE_EVENTS,
.result = result_node_sync,
};
pw_log_debug("%p: resource %p set param id:%d (%s) %08x", node, resource,
id, spa_debug_type_find_name(spa_type_param, id), flags);
res = spa_node_set_param(node->node, id, flags, param);
if (res < 0) {
pw_resource_errorf(resource, res,
"set param id:%d (%s) flags:%08x failed", id,
spa_debug_type_find_name(spa_type_param, id), flags);
} else if (SPA_RESULT_IS_ASYNC(res)) {
pw_impl_client_set_busy(client, true);
if (data->end == -1)
spa_node_add_listener(node->node, &data->listener,
&node_events, data);
data->seq = res;
data->end = spa_node_sync(node->node, res);
}
return 0;
}
static int node_send_command(void *object, const struct spa_command *command)
{
struct resource_data *data = object;
struct pw_impl_node *node = data->node;
uint32_t id = SPA_NODE_COMMAND_ID(command);
int res;
pw_log_debug("%p: got command %d (%s)", node, id,
spa_debug_type_find_name(spa_type_node_command_id, id));
switch (id) {
case SPA_NODE_COMMAND_Suspend:
res = suspend_node(node);
break;
default:
res = spa_node_send_command(node->node, command);
break;
}
return res;
}
static const struct pw_node_methods node_methods = {
PW_VERSION_NODE_METHODS,
.subscribe_params = node_subscribe_params,
.enum_params = node_enum_params,
.set_param = node_set_param,
.send_command = node_send_command
};
static void resource_destroy(void *data)
{
struct resource_data *d = data;
remove_busy_resource(d);
spa_hook_remove(&d->resource_listener);
spa_hook_remove(&d->object_listener);
}
static void resource_pong(void *data, int seq)
{
struct resource_data *d = data;
struct pw_resource *resource = d->resource;
pw_log_debug("%p: resource %p: got pong %d", d->node,
resource, seq);
}
static const struct pw_resource_events resource_events = {
PW_VERSION_RESOURCE_EVENTS,
.destroy = resource_destroy,
.pong = resource_pong,
};
static int
global_bind(void *object, struct pw_impl_client *client, uint32_t permissions,
uint32_t version, uint32_t id)
{
struct pw_impl_node *this = object;
struct pw_global *global = this->global;
struct pw_resource *resource;
struct resource_data *data;
resource = pw_resource_new(client, id, permissions, global->type, version, sizeof(*data));
if (resource == NULL)
goto error_resource;
data = pw_resource_get_user_data(resource);
data->node = this;
data->resource = resource;
data->end = -1;
pw_resource_add_listener(resource,
&data->resource_listener,
&resource_events, data);
pw_resource_add_object_listener(resource,
&data->object_listener,
&node_methods, data);
pw_log_debug("%p: bound to %d", this, resource->id);
pw_global_add_resource(global, resource);
this->info.change_mask = PW_NODE_CHANGE_MASK_ALL;
pw_node_resource_info(resource, &this->info);
this->info.change_mask = 0;
return 0;
error_resource:
pw_log_error("%p: can't create node resource: %m", this);
return -errno;
}
static void global_free(void *data)
{
struct pw_impl_node *this = data;
spa_hook_remove(&this->global_listener);
this->global = NULL;
pw_impl_node_destroy(this);
}
static const struct pw_global_events global_events = {
PW_VERSION_GLOBAL_EVENTS,
.free = global_free,
};
static inline void insert_driver(struct pw_context *context, struct pw_impl_node *node)
{
struct pw_impl_node *n, *t;
spa_list_for_each_safe(n, t, &context->driver_list, driver_link) {
if (n->priority_driver < node->priority_driver)
break;
if (n->priority_driver == 0 && node->priority_driver == 0) {
/* no priority is set, we prefer the driver that does
* lazy scheduling. */
if (n->supports_request > 0 && node->supports_lazy > 0) {
if (n->supports_request <= node->supports_lazy)
break;
}
}
}
spa_list_append(&n->driver_link, &node->driver_link);
pw_context_emit_driver_added(context, node);
}
static inline void remove_driver(struct pw_context *context, struct pw_impl_node *node)
{
spa_list_remove(&node->driver_link);
pw_context_emit_driver_removed(context, node);
}
static int
do_update_position(struct spa_loop *loop,
bool async, uint32_t seq, const void *data, size_t size, void *user_data)
{
struct pw_impl_node *node = user_data;
void *position = *(void**)data;
pw_log_trace("%p: set position %p", node, position);
node->rt.position = position;
if (position) {
node->target_rate = node->rt.position->clock.target_rate;
node->target_quantum = node->rt.position->clock.target_duration;
}
return 0;
}
SPA_EXPORT
int pw_impl_node_set_io(struct pw_impl_node *this, uint32_t id, void *data, size_t size)
{
int res;
struct pw_impl_port *port;
switch (id) {
case SPA_IO_Position:
if (data != NULL && size < sizeof(struct spa_io_position))
return -EINVAL;
pw_log_debug("%p: set position %p", this, data);
pw_loop_locked(this->data_loop,
do_update_position, SPA_ID_INVALID, &data, sizeof(void*), this);
break;
case SPA_IO_Clock:
if (data != NULL && size < sizeof(struct spa_io_clock))
return -EINVAL;
pw_log_debug("%p: set clock %p", this, data);
this->rt.clock = data;
if (this->rt.clock) {
this->info.id = this->rt.clock->id;
this->rt.target.id = this->info.id;
}
break;
}
this->driving = this->driver && this->rt.clock && this->rt.position &&
this->rt.position->clock.id == this->rt.clock->id;
pw_log_debug("%p: driving:%d clock-id:%d driver-id:%d", this, this->driving,
this->rt.clock ? this->rt.clock->id : SPA_ID_INVALID,
this->rt.position ? this->rt.position->clock.id : SPA_ID_INVALID);
spa_list_for_each(port, &this->input_ports, link)
spa_node_set_io(port->mix, id, data, size);
spa_list_for_each(port, &this->output_ports, link)
spa_node_set_io(port->mix, id, data, size);
res = spa_node_set_io(this->node, id, data, size);
if (this->rt.position &&
((res >= 0 && !SPA_RESULT_IS_ASYNC(res)) ||
this->rt.target.activation->client_version < 1))
this->rt.target.activation->active_driver_id = this->rt.position->clock.id;
pw_log_debug("%p: set io: %s", this, spa_strerror(res));
return res;
}
static int
do_add_target(struct spa_loop *loop,
bool async, uint32_t seq, const void *data, size_t size, void *user_data)
{
struct pw_node_target *t = user_data;
struct pw_impl_node *node = *(struct pw_impl_node**)data;
pw_log_debug("%p: target:%p id:%d added:%d prepared:%d", node, t, t->id, t->added, node->rt.prepared);
if (!t->added) {
spa_list_append(&node->rt.target_list, &t->link);
t->added = true;
if (node->rt.prepared)
activate_target(node, t);
}
return 0;
}
SPA_EXPORT
int pw_impl_node_add_target(struct pw_impl_node *node, struct pw_node_target *t)
{
pw_loop_locked(node->data_loop,
do_add_target, SPA_ID_INVALID, &node, sizeof(void *), t);
if (t->node)
pw_impl_node_emit_peer_added(node, t->node);
return 0;
}
static int
do_remove_target(struct spa_loop *loop,
bool async, uint32_t seq, const void *data, size_t size, void *user_data)
{
struct pw_node_target *t = user_data;
struct pw_impl_node *node = *(struct pw_impl_node**)data;
pw_log_debug("%p: target:%p id:%d added:%d prepared:%d", node, t, t->id, t->added, node->rt.prepared);
if (t->added) {
spa_list_remove(&t->link);
t->added = false;
if (node->rt.prepared) {
int old_state = SPA_ATOMIC_LOAD(node->rt.target.activation->status);
uint64_t trigger = 0;
if (PW_NODE_ACTIVATION_PENDING_TRIGGER(old_state))
trigger = get_time_ns(node->rt.target.system);
deactivate_target(node, t, trigger);
}
}
return 0;
}
SPA_EXPORT
int pw_impl_node_remove_target(struct pw_impl_node *node, struct pw_node_target *t)
{
/* we also update the target list for remote nodes so that the profiler
* can inspect the nodes as well */
pw_loop_locked(node->data_loop,
do_remove_target, SPA_ID_INVALID, &node, sizeof(void *), t);
if (t->node)
pw_impl_node_emit_peer_removed(node, t->node);
return 0;
}
static void update_io(struct pw_impl_node *node)
{
struct pw_node_target *t = &node->rt.target;
pw_log_debug("%p: id:%d", node, node->info.id);
pw_impl_node_set_io(node, SPA_IO_Clock, &t->activation->position.clock,
sizeof(struct spa_io_clock));
pw_impl_node_set_io(node, SPA_IO_Position, &t->activation->position,
sizeof(struct spa_io_position));
}
SPA_EXPORT
int pw_impl_node_register(struct pw_impl_node *this,
struct pw_properties *properties)
{
struct pw_context *context = this->context;
struct pw_impl_port *port;
pw_log_debug("%p: register remote:%d exported:%d", this, this->remote, this->exported);
if (this->registered)
goto error_existed;
this->global = pw_global_new(context,
PW_TYPE_INTERFACE_Node,
PW_VERSION_NODE,
PW_NODE_PERM_MASK,
properties,
global_bind,
this);
if (this->global == NULL)
return -errno;
spa_list_append(&context->node_list, &this->link);
if (this->driver)
insert_driver(context, this);
this->registered = true;
this->info.id = this->global->id;
this->rt.target.id = this->info.id;
this->rt.target.activation->position.clock.id = this->global->id;
this->from_driver_peer = pw_node_peer_ref(this, this);
this->to_driver_peer = pw_node_peer_ref(this, this);
pw_properties_setf(this->properties, PW_KEY_OBJECT_ID, "%d", this->global->id);
pw_properties_setf(this->properties, PW_KEY_OBJECT_SERIAL, "%"PRIu64,
pw_global_get_serial(this->global));
pw_global_update_keys(this->global, &this->properties->dict, global_keys);
pw_impl_node_initialized(this);
pw_global_add_listener(this->global, &this->global_listener, &global_events, this);
pw_global_register(this->global);
if (this->node)
update_io(this);
spa_list_for_each(port, &this->input_ports, link)
pw_impl_port_register(port, NULL);
spa_list_for_each(port, &this->output_ports, link)
pw_impl_port_register(port, NULL);
if (this->active)
pw_context_recalc_graph(context, "register active node");
return 0;
error_existed:
pw_properties_free(properties);
return -EEXIST;
}
SPA_EXPORT
int pw_impl_node_initialized(struct pw_impl_node *this)
{
pw_log_debug("%p initialized", this);
pw_impl_node_emit_initialized(this);
node_update_state(this, PW_NODE_STATE_SUSPENDED, 0, NULL);
return 0;
}
static void remove_segment_owner(struct pw_impl_node *driver, uint32_t node_id)
{
struct pw_node_activation *a = driver->rt.target.activation;
SPA_ATOMIC_CAS(a->segment_owner[0], node_id, 0);
SPA_ATOMIC_CAS(a->segment_owner[1], node_id, 0);
}
SPA_EXPORT
int pw_impl_node_set_driver(struct pw_impl_node *node, struct pw_impl_node *driver)
{
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
struct pw_impl_node *old = node->driver_node;
bool was_driving, no_driver = (driver == NULL);
if (no_driver)
driver = node;
spa_list_remove(&node->follower_link);
spa_list_append(&driver->follower_list, &node->follower_link);
if (old == driver)
return 0;
remove_segment_owner(old, node->info.id);
pw_log_debug("%p: driver %p driving:%u", node, driver, node->driving);
pw_log_info("(%s-%u) -> change driver (%s-%d -> %s-%d)",
node->name, node->info.id,
old->name, old->info.id, driver->name, driver->info.id);
/* make sure the old driver doesn't trigger the node anymore */
pw_node_peer_unref(spa_steal_ptr(node->from_driver_peer));
/* make sure the node doesn't trigger the old driver anymore */
pw_node_peer_unref(spa_steal_ptr(node->to_driver_peer));
node->driver_node = driver;
node->moved = true;
/* first send new driver target to node, the node is not yet being
* scheduled so it won't trigger yet */
node->to_driver_peer = pw_node_peer_ref(node, driver);
was_driving = node->driving;
/* then set the new driver node activation */
pw_impl_node_set_io(node, SPA_IO_Position,
&driver->rt.target.activation->position,
sizeof(struct spa_io_position));
/* When a node was driver (and is waiting for all nodes to complete
* the Start command) cancel the pending state and let the new driver
* calculate a new state so that the Start command is sent to the
* node */
if (was_driving && !node->driving)
impl->pending_state = node->info.state;
/* and then make the driver trigger the node */
node->from_driver_peer = pw_node_peer_ref(driver, node);
pw_impl_node_emit_driver_changed(node, old, driver);
if (no_driver) {
/* We don't have a driver, so remove the property */
pw_properties_set(node->properties, PW_KEY_NODE_DRIVER_ID, NULL);
} else if (node->driver_node->global) {
/* Expose the driver ID if it is available as a global */
pw_properties_setf(node->properties, PW_KEY_NODE_DRIVER_ID, "%u",
pw_global_get_id(node->driver_node->global));
}
node->info.change_mask |= PW_NODE_CHANGE_MASK_PROPS;
pw_impl_node_emit_info_changed(driver, &node->info);
return 0;
}
struct match {
struct pw_impl_node *node;
int count;
};
#define MATCH_INIT(n) ((struct match){ .node = (n) })
static int execute_match(void *data, const char *location, const char *action,
const char *val, size_t len)
{
struct match *match = data;
struct pw_impl_node *this = match->node;
if (spa_streq(action, "update-props")) {
match->count += pw_properties_update_string(this->properties, val, len);
}
return 1;
}
static void check_properties(struct pw_impl_node *node)
{
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
struct pw_context *context = node->context;
const char *str, *recalc_reason = NULL;
struct spa_fraction frac;
uint32_t value;
bool driver, trigger, sync, async;
struct match match;
match = MATCH_INIT(node);
pw_context_conf_section_match_rules(context, "node.rules",
&node->properties->dict, execute_match, &match);
value = pw_properties_get_uint32(node->properties, PW_KEY_PRIORITY_DRIVER, 0);
if (value != node->priority_driver) {
pw_log_debug("%p: priority driver %d -> %d", node, node->priority_driver, value);
node->priority_driver = value;
if (node->registered && node->driver) {
remove_driver(context, node);
insert_driver(context, node);
recalc_reason = "driver priority changed";
}
}
node->supports_lazy = pw_properties_get_uint32(node->properties, PW_KEY_NODE_SUPPORTS_LAZY, 0);
node->supports_request = pw_properties_get_uint32(node->properties, PW_KEY_NODE_SUPPORTS_REQUEST, 0);
if ((str = pw_properties_get(node->properties, PW_KEY_NODE_NAME)) &&
(node->name == NULL || !spa_streq(node->name, str))) {
free(node->name);
node->name = strdup(str);
snprintf(node->rt.target.name, sizeof(node->rt.target.name), "%s", node->name);
pw_log_debug("%p: name '%s'", node, node->name);
}
node->pause_on_idle = pw_properties_get_bool(node->properties, PW_KEY_NODE_PAUSE_ON_IDLE, true);
node->suspend_on_idle = pw_properties_get_bool(node->properties, PW_KEY_NODE_SUSPEND_ON_IDLE, false);
node->transport_sync = pw_properties_get_bool(node->properties, PW_KEY_NODE_TRANSPORT_SYNC, false);
impl->cache_params = pw_properties_get_bool(node->properties, PW_KEY_NODE_CACHE_PARAMS, true);
driver = pw_properties_get_bool(node->properties, PW_KEY_NODE_DRIVER, false);
node->exclusive = pw_properties_get_bool(node->properties, PW_KEY_NODE_EXCLUSIVE, false);
node->reliable = pw_properties_get_bool(node->properties, PW_KEY_NODE_RELIABLE, false);
if (node->driver != driver) {
pw_log_debug("%p: driver %d -> %d", node, node->driver, driver);
node->driver = driver;
if (node->registered) {
if (driver)
insert_driver(context, node);
else
remove_driver(context, node);
}
if (driver && node->driver_node == node)
node->driving = true;
recalc_reason = "driver changed";
}
/* not scheduled automatically so we add an additional required trigger */
trigger = pw_properties_get_bool(node->properties, PW_KEY_NODE_TRIGGER, false);
if (trigger != node->trigger) {
node->trigger = trigger;
if (trigger)
SPA_ATOMIC_INC(node->rt.target.activation->state[0].required);
else
SPA_ATOMIC_DEC(node->rt.target.activation->state[0].required);
}
/* group defines what nodes are scheduled together */
str = pw_properties_get(node->properties, PW_KEY_NODE_GROUP);
if (!spa_streq(str, impl->group)) {
pw_log_info("%p: group '%s'->'%s'", node, impl->group, str);
free(impl->group);
impl->group = str ? strdup(str) : NULL;
pw_free_strv(node->groups);
node->groups = impl->group ?
pw_strv_parse(impl->group, strlen(impl->group), INT_MAX, NULL) : NULL;
node->freewheel = pw_strv_find(node->groups, "pipewire.freewheel") >= 0;
recalc_reason = "group changed";
}
/* link group defines what nodes are logically linked together */
str = pw_properties_get(node->properties, PW_KEY_NODE_LINK_GROUP);
if (!spa_streq(str, impl->link_group)) {
pw_log_info("%p: link group '%s'->'%s'", node, impl->link_group, str);
free(impl->link_group);
impl->link_group = str ? strdup(str) : NULL;
pw_free_strv(node->link_groups);
node->link_groups = impl->link_group ?
pw_strv_parse(impl->link_group, strlen(impl->link_group), INT_MAX, NULL) : NULL;
recalc_reason = "link group changed";
}
/* sync group defines what nodes are part of the same sync */
str = pw_properties_get(node->properties, PW_KEY_NODE_SYNC_GROUP);
if (str == NULL)
str = "group.sync.0";
if (!spa_streq(str, impl->sync_group)) {
pw_log_info("%p: sync group '%s'->'%s'", node, impl->sync_group, str);
free(impl->sync_group);
impl->sync_group = str ? strdup(str) : NULL;
pw_free_strv(node->sync_groups);
node->sync_groups = impl->sync_group ?
pw_strv_parse(impl->sync_group, strlen(impl->sync_group), INT_MAX, NULL) : NULL;
recalc_reason = "sync group changed";
}
sync = pw_properties_get_bool(node->properties, PW_KEY_NODE_SYNC, false);
if (sync != node->sync) {
pw_log_info("%p: sync %d -> %d", node, node->sync, sync);
node->sync = sync;
recalc_reason = "sync changed";
}
str = pw_properties_get(node->properties, PW_KEY_NODE_TRANSPORT);
if (str != NULL) {
node->transport = spa_atob(str) ?
PW_NODE_ACTIVATION_COMMAND_START :
PW_NODE_ACTIVATION_COMMAND_STOP;
pw_log_info("%p: transport %d", node, node->transport);
pw_properties_set(node->properties, PW_KEY_NODE_TRANSPORT, NULL);
recalc_reason = "transport changed";
}
async = pw_properties_get_bool(node->properties, PW_KEY_NODE_ASYNC, false);
async &= !node->driver;
if (async != node->async) {
pw_log_info("%p: async %d -> %d", node, node->async, async);
node->async = async;
SPA_FLAG_UPDATE(node->rt.target.activation->flags, PW_NODE_ACTIVATION_FLAG_ASYNC, async);
}
if ((str = pw_properties_get(node->properties, PW_KEY_NODE_PASSIVE)) == NULL) {
if ((str = pw_properties_get(node->properties, PW_KEY_MEDIA_CLASS)) != NULL &&
(strstr(str, "/Sink") != NULL || strstr(str, "/Source") != NULL))
str = "follow";
else
str = "false";
}
if (spa_streq(str, "out")) {
node->passive_mode[SPA_DIRECTION_OUTPUT] = PASSIVE_MODE_TRUE;
}
if (spa_streq(str, "out-follow")) {
node->passive_mode[SPA_DIRECTION_OUTPUT] = PASSIVE_MODE_FOLLOW;
}
else if (spa_streq(str, "in")) {
node->passive_mode[SPA_DIRECTION_INPUT] = PASSIVE_MODE_TRUE;
}
else if (spa_streq(str, "in-follow")) {
node->passive_mode[SPA_DIRECTION_INPUT] = PASSIVE_MODE_FOLLOW;
}
else if (spa_streq(str, "follow")) {
node->passive_mode[SPA_DIRECTION_INPUT] = PASSIVE_MODE_FOLLOW;
node->passive_mode[SPA_DIRECTION_OUTPUT] = PASSIVE_MODE_FOLLOW;
}
else {
node->passive_mode[SPA_DIRECTION_OUTPUT] =
node->passive_mode[SPA_DIRECTION_INPUT] =
spa_atob(str) ? PASSIVE_MODE_TRUE : PASSIVE_MODE_FALSE;
}
node->want_driver = pw_properties_get_bool(node->properties, PW_KEY_NODE_WANT_DRIVER, false);
node->always_process = pw_properties_get_bool(node->properties, PW_KEY_NODE_ALWAYS_PROCESS, false);
if (node->always_process)
node->want_driver = true;
if ((str = pw_properties_get(node->properties, PW_KEY_NODE_LATENCY))) {
if (sscanf(str, "%u/%u", &frac.num, &frac.denom) == 2 && frac.denom != 0) {
if (node->latency.num != frac.num || node->latency.denom != frac.denom) {
pw_log_info("(%s-%u) latency:%u/%u -> %u/%u", node->name,
node->info.id, node->latency.num,
node->latency.denom, frac.num, frac.denom);
node->latency = frac;
recalc_reason = "quantum changed";
}
}
}
if ((str = pw_properties_get(node->properties, PW_KEY_NODE_MAX_LATENCY))) {
if (sscanf(str, "%u/%u", &frac.num, &frac.denom) == 2 && frac.denom != 0) {
if (node->max_latency.num != frac.num || node->max_latency.denom != frac.denom) {
pw_log_info("(%s-%u) max-latency:%u/%u -> %u/%u", node->name,
node->info.id, node->max_latency.num,
node->max_latency.denom, frac.num, frac.denom);
node->max_latency = frac;
recalc_reason = "max quantum changed";
}
}
}
node->lock_quantum = pw_properties_get_bool(node->properties, PW_KEY_NODE_LOCK_QUANTUM, false);
value = pw_properties_get_uint32(node->properties, PW_KEY_NODE_FORCE_QUANTUM, 0);
if (node->force_quantum != value) {
node->force_quantum = value;
node->stamp = ++context->stamp;
recalc_reason = "force quantum changed";
}
if ((str = pw_properties_get(node->properties, PW_KEY_NODE_RATE))) {
if (sscanf(str, "%u/%u", &frac.num, &frac.denom) == 2 && frac.denom != 0) {
if (node->rate.num != frac.num || node->rate.denom != frac.denom) {
pw_log_info("(%s-%u) rate:%u/%u -> %u/%u", node->name,
node->info.id, node->rate.num,
node->rate.denom, frac.num, frac.denom);
node->rate = frac;
recalc_reason = "node rate changed";
}
}
}
node->lock_rate = pw_properties_get_bool(node->properties, PW_KEY_NODE_LOCK_RATE, false);
value = pw_properties_get_uint32(node->properties, PW_KEY_NODE_FORCE_RATE, SPA_ID_INVALID);
if (value == 0)
value = node->rate.denom;
if (value == SPA_ID_INVALID)
value = 0;
if (node->force_rate != value) {
pw_log_info("(%s-%u) force-rate:%u -> %u", node->name,
node->info.id, node->force_rate, value);
node->force_rate = value;
node->stamp = ++context->stamp;
recalc_reason = "force rate changed";
}
pw_log_debug("%p: driver:%d recalc:%s active:%d", node, node->driver,
recalc_reason, node->active);
if (recalc_reason != NULL && node->active)
pw_context_recalc_graph(context, recalc_reason);
}
static const char *str_status(uint32_t status)
{
switch (status) {
case PW_NODE_ACTIVATION_NOT_TRIGGERED:
return "not-triggered";
case PW_NODE_ACTIVATION_TRIGGERED:
return "triggered";
case PW_NODE_ACTIVATION_AWAKE:
return "awake";
case PW_NODE_ACTIVATION_FINISHED:
return "finished";
case PW_NODE_ACTIVATION_INACTIVE:
return "inactive";
}
return "unknown";
}
static inline void update_xrun_stats(struct pw_node_activation *a, uint32_t count, uint64_t trigger, uint64_t delay)
{
a->xrun_count += count;
a->xrun_time = trigger;
a->xrun_delay = delay;
a->max_delay = SPA_MAX(a->max_delay, delay);
}
static inline void debug_xrun_target(struct pw_impl_node *driver,
struct pw_node_target *t, int status, uint64_t nsec)
{
struct pw_node_activation *a = t->activation;
struct pw_node_activation_state *state = &a->state[0];
int suppressed;
enum spa_log_level level = SPA_LOG_LEVEL_DEBUG;
if ((suppressed = spa_ratelimit_test(&driver->rt.rate_limit, nsec)) >= 0)
level = SPA_LOG_LEVEL_INFO;
pw_log(level, "(%s-%u) xrun state:%p pending:%d/%d s:%"PRIu64" a:%"PRIu64" f:%"PRIu64
" waiting:%"PRIu64" process:%"PRIu64" status:%s (%d suppressed)",
t->name, t->id, state,
state->pending, state->required,
a->signal_time,
a->awake_time,
a->finish_time,
a->awake_time - a->signal_time,
a->finish_time - a->awake_time,
str_status(status), suppressed);
}
static inline void debug_xrun_graph(struct pw_impl_node *driver, uint64_t nsec, uint32_t old_status)
{
int suppressed;
enum spa_log_level level = SPA_LOG_LEVEL_DEBUG;
struct pw_node_target *t;
if ((suppressed = spa_ratelimit_test(&driver->rt.rate_limit, nsec)) >= 0)
level = SPA_LOG_LEVEL_INFO;
pw_log(level, "(%s-%u) graph xrun %s (%d suppressed)",
driver->name, driver->info.id, str_status(old_status), suppressed);
spa_list_for_each(t, &driver->rt.target_list, link) {
struct pw_node_activation *a = t->activation;
struct pw_node_activation_state *state = &a->state[0];
uint32_t status = SPA_ATOMIC_LOAD(a->status);
if (status == PW_NODE_ACTIVATION_TRIGGERED ||
status == PW_NODE_ACTIVATION_AWAKE) {
pw_log(level, "(%s-%u) xrun state:%p pending:%d/%d s:%"PRIu64" a:%"PRIu64" f:%"PRIu64
" waiting:%"PRIu64" process:%"PRIu64" status:%s",
t->name, t->id, state,
state->pending, state->required,
a->signal_time,
a->awake_time,
a->finish_time,
a->awake_time - a->signal_time,
a->finish_time - a->awake_time,
str_status(status));
}
}
}
static void debug_sync_timeout(struct pw_impl_node *driver, uint64_t nsec)
{
struct pw_node_target *t;
enum spa_log_level level = SPA_LOG_LEVEL_DEBUG;
int suppressed;
if ((suppressed = spa_ratelimit_test(&driver->rt.rate_limit, nsec)) >= 0)
level = SPA_LOG_LEVEL_INFO;
pw_log(level, "(%s-%u) sync timeout, going to RUNNING (%d suppressed)",
driver->name, driver->info.id, suppressed);
spa_list_for_each(t, &driver->rt.target_list, link) {
struct pw_node_activation *a = t->activation;
struct pw_node_activation_state *state = &a->state[0];
uint32_t status = SPA_ATOMIC_LOAD(a->status);
if (!a->pending_sync)
continue;
pw_log(level, "(%s-%u) sync state:%p pending:%d/%d s:%"PRIu64" a:%"PRIu64" f:%"PRIu64
" waiting:%"PRIu64" process:%"PRIu64" status:%s",
t->name, t->id, state,
state->pending, state->required,
a->signal_time,
a->awake_time,
a->finish_time,
a->awake_time - a->signal_time,
a->finish_time - a->awake_time,
str_status(status));
}
}
static inline void calculate_stats(struct pw_impl_node *this, struct pw_node_activation *a)
{
uint64_t signal_time = a->signal_time;
uint64_t prev_signal_time = a->prev_signal_time;
uint64_t process_time = a->finish_time - a->signal_time;
uint64_t period_time = signal_time - prev_signal_time;
if (SPA_LIKELY(signal_time > prev_signal_time)) {
float load = (float) process_time / (float) period_time;
a->cpu_load[0] = (a->cpu_load[0] + load) / 2.0f;
a->cpu_load[1] = (a->cpu_load[1] * 7.0f + load) / 8.0f;
a->cpu_load[2] = (a->cpu_load[2] * 31.0f + load) / 32.0f;
}
pw_log_trace_fp("%p: graph completed wait:%"PRIu64" run:%"PRIu64
" busy:%"PRIu64" period:%"PRIu64" cpu:%f:%f:%f", this,
a->awake_time - signal_time,
a->finish_time - a->awake_time,
process_time, period_time,
a->cpu_load[0], a->cpu_load[1], a->cpu_load[2]);
}
/* The main processing entry point of a node. This is called from the data-loop and usually
* as a result of signaling the eventfd of the node.
*
* This code runs on the client and the server, depending on where the node is.
*/
static inline int process_node(void *data, uint64_t nsec)
{
struct pw_impl_node *this = data;
struct pw_impl_port *p;
struct pw_node_activation *a = this->rt.target.activation;
struct spa_system *data_system = this->rt.target.system;
int status;
bool was_awake;
if (!SPA_ATOMIC_CAS(a->status,
PW_NODE_ACTIVATION_TRIGGERED,
PW_NODE_ACTIVATION_AWAKE))
return 0;
a->awake_time = nsec;
pw_log_trace_fp("%p: %s-%d process remote:%u exported:%u %"PRIu64" %"PRIu64,
this, this->name, this->info.id, this->remote, this->exported,
a->signal_time, nsec);
/* when transport sync is not supported, just clear the flag */
if (SPA_UNLIKELY(!this->transport_sync))
a->pending_sync = false;
if (SPA_LIKELY(this->rt.prepared)) {
/* process input mixers */
spa_list_for_each(p, &this->rt.input_mix, rt.node_link)
spa_node_process_fast(p->mix);
/* process the actual node */
status = spa_node_process_fast(this->node);
/* process output tee */
if (status & SPA_STATUS_HAVE_DATA) {
spa_list_for_each(p, &this->rt.output_mix, rt.node_link)
spa_node_process_fast(p->mix);
}
} else {
/* This can happen when we deactivated the node but some links are
* still not shut down. We simply don't schedule the node and make
* sure we trigger the peers in trigger_targets below. */
pw_log_debug("%p: scheduling non-active node %s", this, this->name);
status = SPA_STATUS_HAVE_DATA;
}
a->state[0].status = status;
nsec = get_time_ns(data_system);
was_awake = SPA_ATOMIC_CAS(a->status,
PW_NODE_ACTIVATION_AWAKE,
PW_NODE_ACTIVATION_FINISHED);
a->finish_time = nsec;
pw_log_trace_fp("%p: finished status:%d %"PRIu64" was_awake:%d",
this, status, nsec, was_awake);
/* we don't need to trigger targets when the node was driving the
* graph because that means we finished the graph. */
if (SPA_LIKELY(!this->driving)) {
if ((!this->async || a->server_version < 1) && was_awake)
trigger_targets(this, status, nsec);
} else {
/* calculate CPU time when finished */
a->signal_time = this->driver_start;
calculate_stats(this, a);
pw_impl_node_rt_emit_complete(this);
}
if (SPA_UNLIKELY(status & SPA_STATUS_DRAINED))
pw_impl_node_rt_emit_drained(this);
return status;
}
int pw_impl_node_trigger(struct pw_impl_node *node)
{
uint64_t nsec = get_time_ns(node->rt.target.system);
struct pw_node_target *t = &node->rt.target;
return t->trigger(t, nsec);
}
static void node_on_fd_events(struct spa_source *source)
{
struct pw_impl_node *this = source->data;
if (SPA_UNLIKELY(source->rmask & (SPA_IO_ERR | SPA_IO_HUP))) {
pw_log_warn("%p: got socket error %08x", this, source->rmask);
return;
}
if (SPA_LIKELY(source->rmask & SPA_IO_IN)) {
uint64_t cmd, nsec;
struct spa_system *data_system = this->rt.target.system;
nsec = get_time_ns(data_system);
if (SPA_UNLIKELY(spa_system_eventfd_read(data_system, this->source.fd, &cmd) < 0))
pw_log_warn("%p: read failed %m", this);
else if (SPA_UNLIKELY(cmd > 1)) {
pw_log_info("(%s-%u) client missed %"PRIu64" wakeups",
this->name, this->info.id, cmd - 1);
update_xrun_stats(this->rt.target.activation, cmd - 1,
nsec / 1000, 0);
}
pw_log_trace_fp("%p: remote:%u exported:%u %s-%d got process %"PRIu64,
this, this->remote, this->exported, this->name, this->info.id,
nsec);
process_node(this, nsec);
}
}
static void reset_segment(struct spa_io_segment *seg)
{
spa_zero(*seg);
seg->rate = 1.0;
}
static void reset_position(struct pw_impl_node *this, struct spa_io_position *pos)
{
uint32_t i;
struct settings *s = &this->context->settings;
uint32_t quantum = s->clock_force_quantum == 0 ? s->clock_quantum : s->clock_force_quantum;
uint32_t rate = s->clock_force_rate == 0 ? s->clock_rate : s->clock_force_rate;
this->target_rate = SPA_FRACTION(1, rate);
this->target_quantum = quantum;
this->elapsed = 0;
pos->clock.rate = pos->clock.target_rate = this->target_rate;
pos->clock.duration = pos->clock.target_duration = this->target_quantum;
pos->video.flags = SPA_IO_VIDEO_SIZE_VALID;
pos->video.size = s->video_size;
pos->video.stride = pos->video.size.width * 16;
pos->video.framerate = s->video_rate;
pos->offset = INT64_MIN;
pos->n_segments = 1;
for (i = 0; i < SPA_IO_POSITION_MAX_SEGMENTS; i++)
reset_segment(&pos->segments[i]);
}
SPA_EXPORT
struct pw_impl_node *pw_context_create_node(struct pw_context *context,
struct pw_properties *properties,
size_t user_data_size)
{
struct impl *impl;
struct pw_impl_node *this;
size_t size;
int res;
impl = calloc(1, sizeof(struct impl) + user_data_size);
if (impl == NULL) {
res = -errno;
goto error_exit;
}
spa_list_init(&impl->param_list);
spa_list_init(&impl->pending_list);
this = &impl->this;
this->context = context;
this->name = strdup("node");
this->source.fd = -1;
if (properties == NULL)
properties = pw_properties_new(NULL, NULL);
if (properties == NULL) {
res = -errno;
goto error_clean;
}
this->data_loop = pw_context_acquire_loop(context, &properties->dict);
if (this->data_loop == NULL) {
pw_log_error("can't find data-loop");
res = -ENOENT;
goto error_clean;
}
if (user_data_size > 0)
this->user_data = SPA_PTROFF(impl, sizeof(struct impl), void);
this->properties = properties;
/* the eventfd used to signal the node */
if ((res = spa_system_eventfd_create(this->data_loop->system,
SPA_FD_CLOEXEC | SPA_FD_NONBLOCK)) < 0)
goto error_clean;
pw_log_debug("%p: new fd:%d loop:%s", this, res, this->data_loop->name);
this->source.fd = res;
this->source.func = node_on_fd_events;
this->source.data = this;
this->source.mask = SPA_IO_IN | SPA_IO_ERR | SPA_IO_HUP;
this->source.rmask = 0;
size = sizeof(struct pw_node_activation);
this->activation = pw_mempool_alloc(this->context->pool,
PW_MEMBLOCK_FLAG_READWRITE |
PW_MEMBLOCK_FLAG_SEAL |
PW_MEMBLOCK_FLAG_MAP,
SPA_DATA_MemFd, size);
if (this->activation == NULL) {
res = -errno;
goto error_clean;
}
impl->work = pw_context_get_work_queue(this->context);
impl->pending_id = SPA_ID_INVALID;
spa_list_init(&this->follower_list);
spa_list_init(&this->peer_list);
spa_hook_list_init(&this->listener_list);
spa_hook_list_init(&this->rt_listener_list);
this->info.state = PW_NODE_STATE_CREATING;
this->info.props = &this->properties->dict;
this->info.params = this->params;
spa_list_init(&this->input_ports);
pw_map_init(&this->input_port_map, 64, 64);
spa_list_init(&this->output_ports);
pw_map_init(&this->output_port_map, 64, 64);
spa_list_init(&this->rt.input_mix);
spa_list_init(&this->rt.output_mix);
spa_list_init(&this->rt.target_list);
this->rt.target.activation = this->activation->map->ptr;
this->rt.target.node = this;
this->rt.target.system = this->data_loop->system;
this->rt.target.fd = this->source.fd;
this->rt.target.trigger = trigger_target_v1;
reset_position(this, &this->rt.target.activation->position);
this->rt.target.activation->sync_timeout = DEFAULT_SYNC_TIMEOUT;
this->rt.target.activation->sync_left = 0;
this->rt.target.activation->status = PW_NODE_ACTIVATION_INACTIVE;
this->rt.target.activation->server_version = PW_VERSION_NODE_ACTIVATION;
this->rt.target.activation->client_version = PW_VERSION_NODE_ACTIVATION;
this->rt.rate_limit.interval = 2 * SPA_NSEC_PER_SEC;
this->rt.rate_limit.burst = 1;
this->driver_node = this;
spa_list_append(&this->follower_list, &this->follower_link);
check_properties(this);
return this;
error_clean:
if (this->activation)
pw_memblock_unref(this->activation);
if (this->source.fd != -1)
spa_system_close(this->data_loop->system, this->source.fd);
if (this->data_loop)
pw_context_release_loop(context, this->data_loop);
free(this->name);
free(impl);
error_exit:
pw_properties_free(properties);
errno = -res;
return NULL;
}
SPA_EXPORT
const struct pw_node_info *pw_impl_node_get_info(struct pw_impl_node *node)
{
return &node->info;
}
SPA_EXPORT
void * pw_impl_node_get_user_data(struct pw_impl_node *node)
{
return node->user_data;
}
SPA_EXPORT
struct pw_context * pw_impl_node_get_context(struct pw_impl_node *node)
{
return node->context;
}
SPA_EXPORT
struct pw_global *pw_impl_node_get_global(struct pw_impl_node *node)
{
return node->global;
}
SPA_EXPORT
const struct pw_properties *pw_impl_node_get_properties(struct pw_impl_node *node)
{
return node->properties;
}
static int update_properties(struct pw_impl_node *node, const struct spa_dict *dict, bool filter)
{
static const char * const ignored[] = {
PW_KEY_OBJECT_ID,
PW_KEY_MODULE_ID,
PW_KEY_FACTORY_ID,
PW_KEY_CLIENT_ID,
PW_KEY_DEVICE_ID,
NULL
};
int changed;
changed = pw_properties_update_ignore(node->properties, dict, filter ? ignored : NULL);
pw_log_debug("%p: updated %d properties", node, changed);
if (changed) {
check_properties(node);
node->info.change_mask |= PW_NODE_CHANGE_MASK_PROPS;
}
return changed;
}
SPA_EXPORT
int pw_impl_node_update_properties(struct pw_impl_node *node, const struct spa_dict *dict)
{
int changed = update_properties(node, dict, false);
emit_info_changed(node, false);
return changed;
}
static void node_info(void *data, const struct spa_node_info *info)
{
struct pw_impl_node *node = data;
uint32_t changed_ids[MAX_PARAMS], n_changed_ids = 0;
bool flags_changed = false;
node->info.max_input_ports = info->max_input_ports;
node->info.max_output_ports = info->max_output_ports;
pw_log_debug("%p: flags:%08"PRIx64" change_mask:%08"PRIx64" max_in:%u max_out:%u",
node, info->flags, info->change_mask, info->max_input_ports,
info->max_output_ports);
if (info->change_mask & SPA_NODE_CHANGE_MASK_FLAGS) {
if (node->spa_flags != info->flags) {
flags_changed = node->spa_flags != 0;
pw_log_debug("%p: flags %"PRIu64"->%"PRIu64, node, node->spa_flags, info->flags);
node->spa_flags = info->flags;
}
}
if (info->change_mask & SPA_NODE_CHANGE_MASK_PROPS) {
update_properties(node, info->props, true);
}
if (info->change_mask & SPA_NODE_CHANGE_MASK_PARAMS) {
uint32_t i;
node->info.change_mask |= PW_NODE_CHANGE_MASK_PARAMS;
node->info.n_params = SPA_MIN(info->n_params, SPA_N_ELEMENTS(node->params));
for (i = 0; i < node->info.n_params; i++) {
uint32_t id = info->params[i].id;
pw_log_debug("%p: param %d id:%d (%s) %08x:%08x", node, i,
id, spa_debug_type_find_name(spa_type_param, id),
node->info.params[i].flags, info->params[i].flags);
node->info.params[i].id = info->params[i].id;
if (node->info.params[i].flags == info->params[i].flags)
continue;
pw_log_debug("%p: update param %d", node, id);
node->info.params[i] = info->params[i];
node->info.params[i].user = 0;
if (info->params[i].flags & SPA_PARAM_INFO_READ)
changed_ids[n_changed_ids++] = id;
}
}
emit_info_changed(node, flags_changed);
if (n_changed_ids > 0)
emit_params(node, changed_ids, n_changed_ids);
if (flags_changed)
pw_context_recalc_graph(node->context, "node flags changed");
}
static void node_port_info(void *data, enum spa_direction direction, uint32_t port_id,
const struct spa_port_info *info)
{
struct pw_impl_node *node = data;
struct pw_impl_port *port = pw_impl_node_find_port(node, direction, port_id);
if (info == NULL) {
if (port) {
pw_log_debug("%p: %s port %d removed", node,
pw_direction_as_string(direction), port_id);
pw_impl_port_destroy(port);
} else {
pw_log_warn("%p: %s port %d unknown", node,
pw_direction_as_string(direction), port_id);
}
} else if (port) {
pw_log_debug("%p: %s port %d changed", node,
pw_direction_as_string(direction), port_id);
pw_impl_port_update_info(port, info);
} else {
int res;
pw_log_debug("%p: %s port %d added", node,
pw_direction_as_string(direction), port_id);
if ((port = pw_context_create_port(node->context, direction, port_id, info,
node->port_user_data_size))) {
if ((res = pw_impl_port_add(port, node)) < 0) {
pw_log_error("%p: can't add port %p: %d, %s",
node, port, res, spa_strerror(res));
pw_impl_port_destroy(port);
}
}
}
}
static void node_result(void *data, int seq, int res, uint32_t type, const void *result)
{
struct pw_impl_node *node = data;
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
pw_log_trace("%p: result seq:%d res:%d type:%u", node, seq, res, type);
if (res < 0)
impl->last_error = res;
if (SPA_RESULT_IS_ASYNC(seq))
pw_work_queue_complete(impl->work, &impl->this, SPA_RESULT_ASYNC_SEQ(seq), res);
pw_impl_node_emit_result(node, seq, res, type, result);
}
static void handle_request_process_command(struct pw_impl_node *node, const struct spa_command *command)
{
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
if (node->driving) {
pw_log_debug("request process %d %d", node->info.state, impl->pending_state);
if (node->info.state == PW_NODE_STATE_RUNNING) {
spa_node_send_command(node->driver_node->node, command);
} else if (impl->pending_state == PW_NODE_STATE_RUNNING) {
spa_clear_ptr(impl->pending_request_process, free);
impl->pending_request_process = (struct spa_command*)spa_pod_copy(&command->pod);
}
}
}
static void node_event(void *data, const struct spa_event *event)
{
struct pw_impl_node *node = data;
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
uint32_t id = SPA_NODE_EVENT_ID(event);
pw_log_debug("%p: event %d (%s)", node, id,
spa_debug_type_find_name(spa_type_node_event_id, id));
switch (id) {
case SPA_NODE_EVENT_Error:
impl->last_error = -EFAULT;
node_update_state(node, PW_NODE_STATE_ERROR,
-EFAULT, strdup("Received error event"));
break;
case SPA_NODE_EVENT_RequestProcess:
if (!node->driving && !node->exported) {
struct spa_command *command;
size_t size = SPA_POD_SIZE(&event->pod);
/* turn the event and all the arguments into a command */
command = alloca(size);
memcpy(command, event, size);
command->body.body.type = SPA_TYPE_COMMAND_Node;
command->body.body.id = SPA_NODE_COMMAND_RequestProcess;
/* send the request process to the driver but only on the
* server size */
handle_request_process_command(node->driver_node, command);
}
break;
default:
pw_log_debug("unhandled event %d", SPA_NODE_EVENT_ID(event));
break;
}
pw_impl_node_emit_event(node, event);
}
static const struct spa_node_events node_events = {
SPA_VERSION_NODE_EVENTS,
.info = node_info,
.port_info = node_port_info,
.result = node_result,
.event = node_event,
};
#define SYNC_CHECK 0
#define SYNC_START 1
#define SYNC_STOP 2
static inline int check_updates(struct pw_impl_node *node, uint32_t *reposition_owner)
{
int res = SYNC_CHECK;
struct pw_node_activation *a = node->rt.target.activation;
uint32_t command;
if (SPA_UNLIKELY(a->position.offset == INT64_MIN))
a->position.offset = a->position.clock.position;
command = SPA_ATOMIC_XCHG(a->command, PW_NODE_ACTIVATION_COMMAND_NONE);
*reposition_owner = SPA_ATOMIC_XCHG(a->reposition_owner, 0);
if (SPA_UNLIKELY(command != PW_NODE_ACTIVATION_COMMAND_NONE)) {
pw_log_debug("%p: update command:%u", node, command);
switch (command) {
case PW_NODE_ACTIVATION_COMMAND_STOP:
a->position.state = SPA_IO_POSITION_STATE_STOPPED;
res = SYNC_STOP;
break;
case PW_NODE_ACTIVATION_COMMAND_START:
a->position.state = SPA_IO_POSITION_STATE_STARTING;
a->sync_left = a->sync_timeout /
((a->position.clock.duration * SPA_NSEC_PER_SEC) /
a->position.clock.rate.denom);
res = SYNC_START;
break;
}
}
return res;
}
static void do_reposition(struct pw_impl_node *driver, struct pw_node_target *target)
{
struct pw_node_activation *a = driver->rt.target.activation;
struct spa_io_segment *dst, *src;
src = &target->activation->reposition;
dst = &a->position.segments[0];
pw_log_info("%p: %u update position:%"PRIu64, driver, target->id, src->position);
dst->version = src->version;
dst->flags = src->flags;
dst->start = src->start;
dst->duration = src->duration;
dst->rate = src->rate;
dst->position = src->position;
if (src->bar.flags & SPA_IO_SEGMENT_BAR_FLAG_VALID)
dst->bar = src->bar;
if (src->video.flags & SPA_IO_SEGMENT_VIDEO_FLAG_VALID)
dst->video = src->video;
if (dst->start == 0)
dst->start = a->position.clock.position - a->position.offset;
switch (a->position.state) {
case SPA_IO_POSITION_STATE_RUNNING:
a->position.state = SPA_IO_POSITION_STATE_STARTING;
a->sync_left = a->sync_timeout /
((a->position.clock.duration * SPA_NSEC_PER_SEC) /
a->position.clock.rate.denom);
break;
}
}
static inline void update_position(struct pw_impl_node *node, int all_ready, uint64_t nsec)
{
struct pw_node_activation *a = node->rt.target.activation;
if (SPA_UNLIKELY(a->position.state == SPA_IO_POSITION_STATE_STARTING)) {
if (!all_ready && --a->sync_left == 0) {
pw_impl_node_rt_emit_timeout(node);
debug_sync_timeout(node, nsec);
all_ready = true;
}
if (all_ready)
a->position.state = SPA_IO_POSITION_STATE_RUNNING;
}
if (SPA_LIKELY(a->position.state == SPA_IO_POSITION_STATE_RUNNING))
node->elapsed += a->position.clock.duration;
a->position.offset = a->position.clock.position - node->elapsed;
}
/* Called from the data-loop and it is the starting point for driver nodes.
* Most of the logic here is to check for reposition updates and transport changes.
*/
static int node_ready(void *data, int status)
{
struct pw_impl_node *node = data;
struct pw_impl_node *driver = node->driver_node;
struct pw_node_activation *a = node->rt.target.activation;
struct pw_node_activation_state *state = &a->state[0];
struct spa_system *data_system = node->rt.target.system;
struct pw_node_target *t, *reposition_target = NULL;;
struct pw_impl_port *p;
struct spa_io_clock *cl = &node->rt.position->clock;
int sync_type, all_ready, update_sync, target_sync, old_status;
uint32_t owner[2], reposition_owner, pending;
uint64_t min_timeout = UINT64_MAX, nsec;
pw_log_trace_fp("%p: ready driver:%d exported:%d %p status:%d prepared:%d", node,
node->driver, node->exported, driver, status, node->rt.prepared);
if (SPA_UNLIKELY(!node->rt.prepared)) {
/* This can happen when we are stopping a node and removed it from the
* graph but we still have not completed the Pause/Suspend command on
* the node. In that case, the node might still emit ready events,
* which we should simply ignore here. */
pw_log_info("%p: ready non-active node %s in state %d", node, node->name, node->info.state);
return -EIO;
}
if (SPA_UNLIKELY(node != driver)) {
pw_log_warn("%p: ready non-driver node %s", node, node->name);
return -EIO;
}
nsec = get_time_ns(data_system);
while (true) {
old_status = SPA_ATOMIC_LOAD(a->status);
if (SPA_LIKELY(old_status == PW_NODE_ACTIVATION_FINISHED))
/* all good, graph completed */
break;
if (SPA_ATOMIC_CAS(a->status, old_status, PW_NODE_ACTIVATION_TRIGGERED)) {
/* if we got triggered but did not run the processing yet we don't
* really have an error so we can skip the error reporting. We need
* to run recovery anyway because the ready callback is already
* emitted */
if (old_status != PW_NODE_ACTIVATION_TRIGGERED) {
/* otherwise, something was wrong and we debug */
debug_xrun_graph(node, nsec, old_status);
pw_impl_node_rt_emit_incomplete(driver);
}
SPA_FLAG_SET(cl->flags, SPA_IO_CLOCK_FLAG_XRUN_RECOVER);
process_node(node, nsec);
SPA_FLAG_CLEAR(cl->flags, SPA_IO_CLOCK_FLAG_XRUN_RECOVER);
break;
}
}
sync_type = check_updates(node, &reposition_owner);
owner[0] = SPA_ATOMIC_LOAD(a->segment_owner[0]);
owner[1] = SPA_ATOMIC_LOAD(a->segment_owner[1]);
again:
all_ready = sync_type == SYNC_CHECK;
update_sync = !all_ready;
target_sync = sync_type == SYNC_START ? true : false;
pending = 0;
spa_list_for_each(t, &driver->rt.target_list, link) {
struct pw_node_activation *ta = t->activation;
uint32_t id = t->id;
ta->driver_id = driver->info.id;
retry_status:
pw_node_activation_state_reset(&ta->state[0]);
if (ta->active_driver_id != ta->driver_id) {
pw_log_trace_fp("%p: (%s-%u) %d waiting for driver %d<>%d", t->node,
t->name, t->id, ta->status,
ta->active_driver_id, ta->driver_id);
continue;
}
/* we don't change the state of inactive nodes and don't use them
* for reposition. The pending will be at least 1 and they might
* get decremented to 0 but since the status is inactive, we don't
* do the atomic CAS from NOT_TRIGGERED to TRIGGERED and we don't
* write the eventfd. */
old_status = SPA_ATOMIC_LOAD(ta->status);
if (SPA_UNLIKELY(old_status == PW_NODE_ACTIVATION_INACTIVE))
continue;
/* if this fails, the node might just have stopped and we need to retry */
if (SPA_UNLIKELY(!SPA_ATOMIC_CAS(ta->status, old_status, PW_NODE_ACTIVATION_NOT_TRIGGERED)))
goto retry_status;
if (!SPA_FLAG_IS_SET(ta->flags, PW_NODE_ACTIVATION_FLAG_ASYNC))
pending++;
if (old_status == PW_NODE_ACTIVATION_TRIGGERED ||
old_status == PW_NODE_ACTIVATION_AWAKE) {
update_xrun_stats(ta, 1, nsec / 1000, 0);
debug_xrun_target(node, t, old_status, nsec);
}
/* this is the node with reposition info */
if (SPA_UNLIKELY(id == reposition_owner))
reposition_target = t;
/* update extra segment info if it is the owner */
if (SPA_UNLIKELY(id == owner[0]))
a->position.segments[0].bar = ta->segment.bar;
if (SPA_UNLIKELY(id == owner[1]))
a->position.segments[0].video = ta->segment.video;
min_timeout = SPA_MIN(min_timeout, ta->sync_timeout);
if (SPA_UNLIKELY(update_sync)) {
ta->pending_sync = target_sync;
ta->pending_new_pos = target_sync;
} else {
all_ready &= ta->pending_sync == false;
}
ta->prev_signal_time = ta->signal_time;
ta->prev_awake_time = ta->awake_time;
ta->prev_finish_time = ta->finish_time;
}
node->driver_start = nsec;
a->sync_timeout = SPA_MIN(min_timeout, DEFAULT_SYNC_TIMEOUT);
if (SPA_UNLIKELY(reposition_target != NULL)) {
do_reposition(node, reposition_target);
sync_type = SYNC_START;
reposition_owner = 0;
reposition_target = NULL;
goto again;
}
state->pending = pending;
update_position(node, all_ready, nsec);
/* move output with previous cycle, this makes the async nodes
* pick up the new data immediately */
if (status & SPA_STATUS_HAVE_DATA) {
spa_list_for_each(p, &node->rt.output_mix, rt.node_link)
spa_node_process_fast(p->mix);
}
a->position.clock.cycle++;
pw_impl_node_rt_emit_start(node);
/* now signal all the nodes we drive */
trigger_targets(node, status, nsec);
return 0;
}
static int node_reuse_buffer(void *data, uint32_t port_id, uint32_t buffer_id)
{
struct pw_impl_node *node = data;
struct pw_impl_port *p;
spa_list_for_each(p, &node->rt.input_mix, rt.node_link) {
if (p->port_id != port_id)
continue;
spa_node_port_reuse_buffer(p->mix, 0, buffer_id);
break;
}
return 0;
}
static int node_xrun(void *data, uint64_t trigger, uint64_t delay, struct spa_pod *info)
{
struct pw_impl_node *this = data;
struct pw_node_activation *a = this->rt.target.activation;
struct spa_system *data_system = this->rt.target.system;
uint64_t nsec = get_time_ns(data_system);
int suppressed;
update_xrun_stats(a, 1, trigger, delay);
if ((suppressed = spa_ratelimit_test(&this->rt.rate_limit, nsec)) >= 0) {
struct spa_fraction rate;
if (a) {
struct spa_io_clock *cl = &a->position.clock;
rate.num = cl->rate.num * cl->duration;
rate.denom = cl->rate.denom;
} else {
rate = SPA_FRACTION(0,0);
}
pw_log_info("(%s-%d) XRun! rate:%u/%u count:%u time:%"PRIu64
" delay:%"PRIu64" max:%"PRIu64" (%d suppressed)",
this->name, this->info.id,
rate.num, rate.denom, a->xrun_count,
trigger, delay, a->max_delay,
suppressed);
}
pw_impl_node_rt_emit_xrun(this);
return 0;
}
static const struct spa_node_callbacks node_callbacks = {
SPA_VERSION_NODE_CALLBACKS,
.ready = node_ready,
.reuse_buffer = node_reuse_buffer,
.xrun = node_xrun,
};
static int handle_node_param(struct pw_impl_node *node, const char *key, const char *value)
{
const struct spa_type_info *ti;
uint8_t buffer[1024];
struct spa_pod_builder b = SPA_POD_BUILDER_INIT(buffer, sizeof(buffer));
struct spa_pod *pod;
int res;
ti = spa_debug_type_find_short(spa_type_param, key);
if (ti == NULL)
return -ENOENT;
if ((res = spa_json_to_pod(&b, 0, ti, value, strlen(value))) < 0)
return res;
if ((pod = spa_pod_builder_deref(&b, 0)) == NULL)
return -ENOSPC;
if ((res = pw_impl_node_set_param(node, ti->type, 0, pod)) < 0)
return res;
return 0;
}
SPA_EXPORT
int pw_impl_node_set_implementation(struct pw_impl_node *node,
struct spa_node *spa_node)
{
int res;
const struct spa_dict_item *it;
pw_log_debug("%p: implementation %p", node, spa_node);
if (node->node) {
pw_log_error("%p: implementation existed %p", node, node->node);
return -EEXIST;
}
node->node = spa_node;
spa_node_set_callbacks(node->node, &node_callbacks, node);
res = spa_node_add_listener(node->node, &node->listener, &node_events, node);
again:
spa_dict_for_each(it, &node->properties->dict) {
if (spa_strstartswith(it->key, "node.param.")) {
if ((res = handle_node_param(node, &it->key[11], it->value)) < 0)
pw_log_warn("can't set node param: %s", spa_strerror(res));
pw_properties_set(node->properties, it->key, NULL);
goto again;
}
}
if (node->registered)
update_io(node);
return res;
}
SPA_EXPORT
struct spa_node *pw_impl_node_get_implementation(struct pw_impl_node *node)
{
return node->node;
}
SPA_EXPORT
void pw_impl_node_add_listener(struct pw_impl_node *node,
struct spa_hook *listener,
const struct pw_impl_node_events *events,
void *data)
{
spa_hook_list_append(&node->listener_list, listener, events, data);
}
struct listener_data {
struct spa_hook *listener;
const struct pw_impl_node_rt_events *events;
void *data;
};
static int
do_add_rt_listener(struct spa_loop *loop,
bool async, uint32_t seq, const void *data, size_t size, void *user_data)
{
struct pw_impl_node *node = user_data;
const struct listener_data *d = data;
spa_hook_list_append(&node->rt_listener_list,
d->listener, d->events, d->data);
return 0;
}
SPA_EXPORT
void pw_impl_node_add_rt_listener(struct pw_impl_node *node,
struct spa_hook *listener,
const struct pw_impl_node_rt_events *events,
void *data)
{
struct listener_data d = { .listener = listener, .events = events, .data = data };
pw_loop_locked(node->data_loop,
do_add_rt_listener, SPA_ID_INVALID, &d, sizeof(d), node);
}
static int do_remove_listener(struct spa_loop *loop,
bool async, uint32_t seq, const void *data, size_t size, void *user_data)
{
struct spa_hook *listener = user_data;
spa_hook_remove(listener);
return 0;
}
SPA_EXPORT
void pw_impl_node_remove_rt_listener(struct pw_impl_node *node,
struct spa_hook *listener)
{
pw_loop_locked(node->data_loop,
do_remove_listener, SPA_ID_INVALID, NULL, 0, listener);
}
/** Destroy a node
* \param node a node to destroy
*
* Remove \a node. This will stop the transfer on the node and
* free the resources allocated by \a node.
*/
SPA_EXPORT
void pw_impl_node_destroy(struct pw_impl_node *node)
{
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
struct pw_impl_port *port;
struct pw_impl_node *follower;
struct pw_context *context = node->context;
bool active, had_driver;
active = node->active;
node->active = false;
node->runnable = false;
pw_log_debug("%p: destroy", impl);
pw_log_info("(%s-%u) destroy", node->name, node->info.id);
node_deactivate(node);
suspend_node(node);
pw_impl_node_emit_destroy(node);
pw_log_debug("%p: driver node %p", impl, node->driver_node);
had_driver = node != node->driver_node;
/* remove ourself as a follower from the driver node */
spa_list_remove(&node->follower_link);
pw_node_peer_unref(spa_steal_ptr(node->from_driver_peer));
pw_node_peer_unref(spa_steal_ptr(node->to_driver_peer));
remove_segment_owner(node->driver_node, node->info.id);
spa_list_consume(follower, &node->follower_list, follower_link) {
pw_log_debug("%p: reassign follower %p", impl, follower);
pw_impl_node_set_driver(follower, NULL);
}
if (node->registered) {
spa_list_remove(&node->link);
if (node->driver)
remove_driver(context, node);
}
if (node->node) {
spa_hook_remove(&node->listener);
spa_node_set_callbacks(node->node, NULL, NULL);
}
pw_log_debug("%p: destroy ports", node);
spa_list_consume(port, &node->input_ports, link)
pw_impl_port_destroy(port);
spa_list_consume(port, &node->output_ports, link)
pw_impl_port_destroy(port);
if (node->global) {
spa_hook_remove(&node->global_listener);
pw_global_destroy(node->global);
}
if (active || had_driver)
pw_context_recalc_graph(context,
"active node destroy");
pw_log_debug("%p: free", node);
pw_impl_node_emit_free(node);
spa_hook_list_clean(&node->listener_list);
pw_memblock_unref(node->activation);
pw_param_clear(&impl->param_list, SPA_ID_INVALID);
pw_param_clear(&impl->pending_list, SPA_ID_INVALID);
pw_map_clear(&node->input_port_map);
pw_map_clear(&node->output_port_map);
pw_work_queue_cancel(impl->work, node, SPA_ID_INVALID);
pw_properties_free(node->properties);
spa_clear_ptr(impl->pending_request_process, free);
clear_info(node);
spa_system_close(node->rt.target.system, node->source.fd);
if (node->data_loop)
pw_context_release_loop(context, node->data_loop);
free(impl->group);
free(impl->link_group);
free(impl->sync_group);
free(impl);
#ifdef HAVE_MALLOC_TRIM
int res = malloc_trim(0);
pw_log_debug("malloc_trim(): %d", res);
#endif
}
SPA_EXPORT
int pw_impl_node_for_each_port(struct pw_impl_node *node,
enum pw_direction direction,
int (*callback) (void *data, struct pw_impl_port *port),
void *data)
{
struct spa_list *ports;
struct pw_impl_port *p, *t;
int res;
if (direction == PW_DIRECTION_INPUT)
ports = &node->input_ports;
else
ports = &node->output_ports;
spa_list_for_each_safe(p, t, ports, link)
if ((res = callback(data, p)) != 0)
return res;
return 0;
}
struct result_node_params_data {
struct impl *impl;
void *data;
int (*callback) (void *data, int seq,
uint32_t id, uint32_t index, uint32_t next,
struct spa_pod *param);
int seq;
unsigned int cache:1;
};
static void result_node_params(void *data, int seq, int res, uint32_t type, const void *result)
{
struct result_node_params_data *d = data;
struct impl *impl = d->impl;
switch (type) {
case SPA_RESULT_TYPE_NODE_PARAMS:
{
const struct spa_result_node_params *r = result;
if (d->seq == seq) {
d->callback(d->data, seq, r->id, r->index, r->next, r->param);
if (d->cache)
pw_param_add(&impl->pending_list, seq, r->id, r->param);
}
break;
}
default:
break;
}
}
SPA_EXPORT
int pw_impl_node_for_each_param(struct pw_impl_node *node,
int seq, uint32_t param_id,
uint32_t index, uint32_t max,
const struct spa_pod *filter,
int (*callback) (void *data, int seq,
uint32_t id, uint32_t index, uint32_t next,
struct spa_pod *param),
void *data)
{
int res;
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
struct result_node_params_data user_data = { impl, data, callback, seq, false };
struct spa_hook listener;
struct spa_param_info *pi;
static const struct spa_node_events node_events = {
SPA_VERSION_NODE_EVENTS,
.result = result_node_params,
};
pi = pw_param_info_find(node->info.params, node->info.n_params, param_id);
if (pi == NULL)
return -ENOENT;
if (max == 0)
max = UINT32_MAX;
pw_log_debug("%p: params id:%d (%s) index:%u max:%u cached:%d", node, param_id,
spa_debug_type_find_name(spa_type_param, param_id),
index, max, pi->user);
if (pi->user == 1) {
struct pw_param *p;
uint8_t buffer[4096];
struct spa_pod_dynamic_builder b;
struct spa_result_node_params result;
uint32_t count = 0;
result.id = param_id;
result.next = 0;
spa_list_for_each(p, &impl->param_list, link) {
if (p->id != param_id)
continue;
result.index = result.next++;
if (result.index < index)
continue;
spa_pod_dynamic_builder_init(&b, buffer, sizeof(buffer), 4096);
if (spa_pod_filter(&b.b, &result.param, p->param, filter) == 0) {
pw_log_debug("%p: %d param %u", node, seq, result.index);
result_node_params(&user_data, seq, 0, SPA_RESULT_TYPE_NODE_PARAMS, &result);
count++;
}
spa_pod_dynamic_builder_clean(&b);
if (count == max)
break;
}
res = 0;
} else {
user_data.cache = impl->cache_params &&
(filter == NULL && index == 0 && max == UINT32_MAX);
if (user_data.cache)
pw_param_add(&impl->pending_list, seq, param_id, NULL);
spa_zero(listener);
spa_node_add_listener(node->node, &listener, &node_events, &user_data);
res = spa_node_enum_params(node->node, seq,
param_id, index, max,
filter);
spa_hook_remove(&listener);
if (user_data.cache) {
if (SPA_RESULT_IS_OK(res) && !SPA_RESULT_IS_ASYNC(res)) {
pw_param_update(&impl->param_list, &impl->pending_list, 0, NULL);
pi->user = 1;
}
else {
pw_param_clear(&impl->pending_list, SPA_ID_INVALID);
}
}
}
return res;
}
SPA_EXPORT
int pw_impl_node_set_param(struct pw_impl_node *node,
uint32_t id, uint32_t flags, const struct spa_pod *param)
{
pw_log_debug("%p: set_param id:%d (%s) flags:%08x param:%p", node, id,
spa_debug_type_find_name(spa_type_param, id), flags, param);
return spa_node_set_param(node->node, id, flags, param);
}
SPA_EXPORT
struct pw_impl_port *
pw_impl_node_find_port(struct pw_impl_node *node, enum pw_direction direction, uint32_t port_id)
{
struct pw_impl_port *port, *p;
struct pw_map *portmap;
struct spa_list *ports;
if (direction == PW_DIRECTION_INPUT) {
portmap = &node->input_port_map;
ports = &node->input_ports;
} else {
portmap = &node->output_port_map;
ports = &node->output_ports;
}
if (port_id != PW_ID_ANY)
port = pw_map_lookup(portmap, port_id);
else {
port = NULL;
/* try to find an unlinked port */
spa_list_for_each(p, ports, link) {
if (spa_list_is_empty(&p->links)) {
port = p;
break;
}
/* We can use this port if it can multiplex */
if (SPA_FLAG_IS_SET(p->mix_flags, PW_IMPL_PORT_MIX_FLAG_MULTI))
port = p;
}
}
pw_log_debug("%p: return %s port %d: %p", node,
pw_direction_as_string(direction), port_id, port);
return port;
}
SPA_EXPORT
uint32_t pw_impl_node_get_free_port_id(struct pw_impl_node *node, enum pw_direction direction)
{
uint32_t n_ports, max_ports;
struct pw_map *portmap;
uint32_t port_id;
bool dynamic;
int res;
if (direction == PW_DIRECTION_INPUT) {
max_ports = node->info.max_input_ports;
n_ports = node->info.n_input_ports;
portmap = &node->input_port_map;
dynamic = SPA_FLAG_IS_SET(node->spa_flags, SPA_NODE_FLAG_IN_DYNAMIC_PORTS);
} else {
max_ports = node->info.max_output_ports;
n_ports = node->info.n_output_ports;
portmap = &node->output_port_map;
dynamic = SPA_FLAG_IS_SET(node->spa_flags, SPA_NODE_FLAG_OUT_DYNAMIC_PORTS);
}
pw_log_debug("%p: direction %s n_ports:%u max_ports:%u",
node, pw_direction_as_string(direction), n_ports, max_ports);
if (!dynamic || n_ports >= max_ports) {
res = -ENOSPC;
goto error;
}
port_id = pw_map_insert_new(portmap, NULL);
if (port_id == SPA_ID_INVALID) {
res = -errno;
goto error;
}
pw_log_debug("%p: free port %d", node, port_id);
return port_id;
error:
pw_log_warn("%p: no more port available: %s", node, spa_strerror(res));
errno = -res;
return SPA_ID_INVALID;
}
SPA_EXPORT
struct pw_impl_port *pw_impl_node_get_free_port(struct pw_impl_node *node, enum pw_direction direction)
{
uint32_t port_id = pw_impl_node_get_free_port_id(node, direction);
if (port_id == SPA_ID_INVALID)
return NULL;
spa_node_add_port(node->node, direction, port_id, NULL);
return pw_impl_node_find_port(node, direction, port_id);
}
static void on_state_complete(void *obj, void *data, int res, uint32_t seq)
{
struct pw_impl_node *node = obj;
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
enum pw_node_state state = SPA_PTR_TO_INT(data);
char *error = NULL;
/* driver nodes added -EBUSY. This is then not an error */
if (res == -EBUSY)
res = 0;
impl->pending_id = SPA_ID_INVALID;
impl->pending_play = false;
pw_log_debug("%p: state complete res:%d seq:%d", node, res, seq);
if (impl->last_error < 0) {
res = impl->last_error;
impl->last_error = 0;
}
if (SPA_RESULT_IS_ERROR(res)) {
if (node->info.state == PW_NODE_STATE_SUSPENDED) {
state = PW_NODE_STATE_SUSPENDED;
res = 0;
} else {
error = spa_aprintf("error changing node state: %s", spa_strerror(res));
state = PW_NODE_STATE_ERROR;
}
}
node_update_state(node, state, res, error);
}
/** Set the node state
* \param node a \ref pw_impl_node
* \param state a \ref pw_node_state
* \return 0 on success < 0 on error
*
* Set the state of \a node to \a state.
*/
SPA_EXPORT
int pw_impl_node_set_state(struct pw_impl_node *node, enum pw_node_state state)
{
int res = 0;
struct impl *impl = SPA_CONTAINER_OF(node, struct impl, this);
enum pw_node_state old = impl->pending_state;
pw_log_debug("%p: set state (%s) %s -> %s, active %d pause_on_idle:%d", node,
pw_node_state_as_string(node->info.state),
pw_node_state_as_string(old),
pw_node_state_as_string(state),
node->active,
node->pause_on_idle);
if (old != state)
pw_impl_node_emit_state_request(node, state);
switch (state) {
case PW_NODE_STATE_CREATING:
return -EIO;
case PW_NODE_STATE_SUSPENDED:
res = suspend_node(node);
break;
case PW_NODE_STATE_IDLE:
res = idle_node(node);
break;
case PW_NODE_STATE_RUNNING:
if (node->active)
res = start_node(node);
break;
case PW_NODE_STATE_ERROR:
break;
}
if (SPA_RESULT_IS_ERROR(res))
return res;
if (SPA_RESULT_IS_ASYNC(res)) {
res = spa_node_sync(node->node, res);
}
if (old != state) {
if (impl->pending_id != SPA_ID_INVALID) {
pw_log_debug("cancel state from %s to %s to %s",
pw_node_state_as_string(node->info.state),
pw_node_state_as_string(impl->pending_state),
pw_node_state_as_string(state));
if (impl->pending_state == PW_NODE_STATE_RUNNING &&
state < PW_NODE_STATE_RUNNING &&
impl->pending_play) {
impl->pending_play = false;
idle_node(node);
}
pw_work_queue_cancel(impl->work, node, impl->pending_id);
node->info.state = impl->pending_state;
}
/* driver nodes return EBUSY to add a -EBUSY to the work queue. This
* will wait until all previous items in the work queue are
* completed */
impl->pending_state = state;
if (node->exported) {
/* exported nodes must complete immediately. This is important
* because the server sends ping to check completion. The server
* will only send Start to driver nodes when all clients are
* ready for processing. */
on_state_complete(node, SPA_INT_TO_PTR(state), -EBUSY, 0);
} else {
impl->pending_id = pw_work_queue_add(impl->work,
node, res == EBUSY ? -EBUSY : res,
on_state_complete, SPA_INT_TO_PTR(state));
}
}
return res;
}
SPA_EXPORT
int pw_impl_node_set_active(struct pw_impl_node *node, bool active)
{
bool old = node->active;
if (old != active) {
pw_log_debug("%p: %s registered:%d exported:%d", node,
active ? "activate" : "deactivate",
node->registered, node->exported);
node->active = active;
pw_impl_node_emit_active_changed(node, active);
if (node->registered)
pw_context_recalc_graph(node->context,
active ? "node activate" : "node deactivate");
else if (!active && node->exported)
remove_node_from_graph(node);
}
return 0;
}
SPA_EXPORT
bool pw_impl_node_is_active(struct pw_impl_node *node)
{
return node->active;
}
SPA_EXPORT
int pw_impl_node_send_command(struct pw_impl_node *node, const struct spa_command *command)
{
uint32_t id = SPA_NODE_COMMAND_ID(command);
int res = 0;
switch (id) {
case SPA_NODE_COMMAND_RequestProcess:
handle_request_process_command(node, command);
break;
default:
res = spa_node_send_command(node->node, command);
}
return res;
}