upstream-archive/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-07-23 11:10:55 -05:00
Code Issues Packages Projects Releases Wiki Activity

server: refactor child --> router communication (#24821)

Browse Source 
* server: refactor child --> router communication

* fix wakeup case

* add docs

* improve update_status()

* nits
This commit is contained in:
Xuan-Son Nguyen
2026-06-20 01:02:26 +02:00
committed by GitHub
parent 4b48a53b6c
commit 2b686a9120
8 changed files with 173 additions and 91 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
tools/server/server-context.cpp
View File

@@ -63,11 +63,6 @@ enum slot_state {
SLOT_STATE_GENERATING,
};
enum server_state {
SERVER_STATE_LOADING_MODEL, // Server is starting up, model not fully loaded yet
SERVER_STATE_READY, // Server is ready and model is loaded
};
struct server_slot {
int id;
@@ -773,6 +768,8 @@ public:
// note: chat_params must not be refreshed upon existing sleeping state
server_chat_params chat_params;
server_state_callback_t callback_state = [](server_state, json) -> void {};
server_context_impl() {
mtmd_helper_log_set(common_log_default_callback, nullptr);
}
@@ -1244,8 +1241,8 @@ private:
if (!params_base.model_alias.empty()) {
// backward compat: use first alias as model name
model_name = *params_base.model_alias.begin();
} else if (!params_base.model.name.empty()) {
model_name = params_base.model.name;
} else if (!params_base.model.get_name().empty()) {
model_name = params_base.model.get_name();
} else {
// fallback: derive model name from file name
auto model_path = std::filesystem::path(params_base.model.path);
@@ -3734,8 +3731,11 @@ struct server_res_generator : server_http_res {
}
};
void server_context::on_sleeping_changed(std::function<void(bool)> callback) {
impl->queue_tasks.on_sleeping_state(std::move(callback));
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) {
impl->callback_state(sleeping ? SERVER_STATE_SLEEPING : SERVER_STATE_READY, {});
});
}
// compute the number of tokens before the last user message in the prompt