Superminaren/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-07 20:47:30 +02:00
Code Issues Packages Projects Releases Wiki Activity

refactor on_sleeping_state

Browse Source
This commit is contained in:
Xuan Son Nguyen
2026-08-19 12:35:18 +02:00
parent 9f710a6e77
commit 5b776fdfe8
3 changed files with 17 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server-context.cpp
+4 -6
View File
@@ -898,6 +898,10 @@ private:
void handle_sleeping_state(bool new_state) {
GGML_ASSERT(sleeping != new_state);
if (new_state) {
if (callback_state) {
callback_state(SERVER_STATE_SLEEPING, {});
// note: for sleeping == false, event is emitted by load_model()
}
SRV_INF("%s", "server is entering sleeping state\n");
destroy();
} else {
@@ -4142,12 +4146,6 @@ struct server_res_generator : server_res_spipe {
void server_context::set_state_callback(server_state_callback_t callback) {
impl->callback_state = std::move(callback);
impl->queue_tasks.on_sleeping_state([this](bool sleeping) {
if (sleeping) {
impl->callback_state(SERVER_STATE_SLEEPING, {});
}
// for sleeping == false, event is emitted by load_model()
});
}
//
tools/server/server-queue.cpp
+8 -2
View File
@@ -312,7 +312,10 @@ void server_queue::start_loop(int64_t idle_sleep_ms) {
if (should_sleep()) {
QUE_INF("%s", "entering sleeping state\n");
sleeping = true;
callback_sleeping_state(true);
// Call order cb0 -> cb1 -> cb{N}
for (auto & cb : callback_sleeping_state) {
cb(true);
}
req_stop_sleeping = false;
// wait until we are requested to exit sleeping state
condition_tasks.wait(lock, [&]{
@@ -323,7 +326,10 @@ void server_queue::start_loop(int64_t idle_sleep_ms) {
}
QUE_INF("%s", "exiting sleeping state\n");
req_stop_sleeping = false;
callback_sleeping_state(false);
// Call order cb{N} -> cb1 -> cb0
for (size_t i = callback_sleeping_state.size(); i > 0; i--) {
callback_sleeping_state[i - 1](false);
}
sleeping = false;
time_last_task = ggml_time_ms();
condition_tasks.notify_all(); // notify wait_until_no_sleep()
tools/server/server-queue.h
+5 -12
View File
@@ -44,7 +44,7 @@ private:
// callback functions
std::function<bool(server_task &&, bool)> callback_new_task;
std::function<void(void)> callback_update_slots;
std::function<void(bool)> callback_sleeping_state;
std::vector<std::function<void(bool)>> callback_sleeping_state;
public:
~server_queue() { worker_stop(); }
@@ -127,18 +127,11 @@ public:
}
// Register callback for sleeping state change; multiple callbacks are allowed
// note: when entering sleeping state, the callback is called AFTER sleeping is set to true
// when leaving sleeping state, the callback is called BEFORE sleeping is set to false
// for example: register order cb0, cb1, cb2
// entering sleep: queue.sleeping = true --> cb0(true) --> cb1(true) --> cb2(true)
// leaving sleep: cb2(false) --> cb1(false) --> cb0(false) --> queue.sleeping = false
void on_sleeping_state(std::function<void(bool)> callback) {
if (callback_sleeping_state) {
auto prev_callback = std::move(callback_sleeping_state);
callback_sleeping_state = [prev_callback, callback](bool sleeping) {
prev_callback(sleeping);
callback(sleeping);
};
} else {
callback_sleeping_state = std::move(callback);
}
callback_sleeping_state.push_back(std::move(callback));
}
private: