happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

add stdin_file

This commit is contained in:
Xuan Son Nguyen 2025-11-24 14:21:21 +01:00
parent 2c6b58f785
commit 539cbf003e
2 changed files with 9 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
tools/server/server-models.cpp
View File

@ -407,6 +407,8 @@ void server_models::load(const std::string & name, const std::vector<std::string
if (result != 0) { if (result != 0) {
throw std::runtime_error("failed to spawn server instance"); throw std::runtime_error("failed to spawn server instance");
} }
inst.stdin_file = subprocess_stdin(inst.subproc.get());
} }
// start a thread to manage the child process // start a thread to manage the child process
@ -457,13 +459,12 @@ void server_models::load(const std::string & name, const std::vector<std::string
cv.notify_all(); cv.notify_all();
} }
static void interrupt_subprocess(subprocess_s * proc) { static void interrupt_subprocess(FILE * stdin_file) {
// because subprocess.h does not provide a way to send SIGINT, // because subprocess.h does not provide a way to send SIGINT,
// we will send a command to the child process to exit gracefully // we will send a command to the child process to exit gracefully
FILE * p_stdin = subprocess_stdin(proc); if (stdin_file) {
if (p_stdin) { fprintf(stdin_file, "%s\n", CMD_EXIT);
fprintf(p_stdin, "%s\n", CMD_EXIT); fflush(stdin_file);
fflush(p_stdin);
} }
} }
@ -473,7 +474,7 @@ void server_models::unload(const std::string & name) {
if (it != mapping.end()) { if (it != mapping.end()) {
if (it->second.meta.is_active()) { if (it->second.meta.is_active()) {
SRV_INF("unloading model instance name=%s\n", name.c_str()); SRV_INF("unloading model instance name=%s\n", name.c_str());
interrupt_subprocess(it->second.subproc.get()); interrupt_subprocess(it->second.stdin_file);
// status change will be handled by the managing thread // status change will be handled by the managing thread
} else { } else {
SRV_WRN("model instance name=%s is not loaded\n", name.c_str()); SRV_WRN("model instance name=%s is not loaded\n", name.c_str());
@ -488,7 +489,7 @@ void server_models::unload_all() {
for (auto & [name, inst] : mapping) { for (auto & [name, inst] : mapping) {
if (inst.meta.is_active()) { if (inst.meta.is_active()) {
SRV_INF("unloading model instance name=%s\n", name.c_str()); SRV_INF("unloading model instance name=%s\n", name.c_str());
interrupt_subprocess(inst.subproc.get()); interrupt_subprocess(inst.stdin_file);
// status change will be handled by the managing thread // status change will be handled by the managing thread
} }
// moving the thread to join list to avoid deadlock // moving the thread to join list to avoid deadlock

1
tools/server/server-models.h
View File

@ -74,6 +74,7 @@ private:
std::shared_ptr<subprocess_s> subproc; // shared between main thread and monitoring thread std::shared_ptr<subprocess_s> subproc; // shared between main thread and monitoring thread
std::thread th; std::thread th;
server_model_meta meta; server_model_meta meta;
FILE * stdin_file = nullptr;
}; };
std::mutex mutex; std::mutex mutex;