Public/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-09-19 17:24:57 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix a race condition

Browse Source
This commit is contained in:
Xuan Son Nguyen
2026-08-19 13:45:12 +02:00
parent 8ca1955127
commit 0591080c28
1 changed files with 13 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server-context.cpp
+13 -6
View File
@@ -4575,14 +4575,18 @@ void server_routes::init_routes() {
res->content_type = "text/plain; version=0.0.4";
res->status = 200;
if (queue_tasks.is_sleeping()) {
// render response using cached_metrics
auto use_cached_metrics = [&]() {
std::unique_lock<std::mutex> lock(mutex_cache);
res->headers["Process-Start-Time-Unix"] = std::to_string(cached_metrics.t_start);
// render response using cached_metrics
server_task_result_metrics tmp;
tmp.metrics = cached_metrics;
tmp.n_idle_slots = params.n_parallel;
res->data = tmp.to_metrics();
};
if (queue_tasks.is_sleeping()) {
use_cached_metrics();
} else {
// request slots data using task queue
@@ -4594,11 +4598,14 @@ void server_routes::init_routes() {
res->rd.post_task(std::move(task), true); // high-priority task
}
// get the result
auto result = res->rd.next(req.should_stop);
// a task posted right before sleeping is never processed, do not wait for it
auto result = res->rd.next([&]{
return req.should_stop() || queue_tasks.is_sleeping();
});
if (!result) {
// connection was closed
GGML_ASSERT(req.should_stop());
if (!req.should_stop()) {
use_cached_metrics();
}
return res;
}