Skip to content

armgit / PS5LM

Public Imported from GitHub
0

llama.cpp on a jailbroken PS5: run any GGUF model natively on the console, starting with Qwen 3.8

v0.8.1: the chat server survives network drops

Abhiramfb15fb722f62
diff --git a/CHANGELOG.md b/CHANGELOG.md
index b0e8cb8..8108e85 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,27 @@ v0.1.x are the upstream payload releases by cobanov.
## [Unreleased]
+## [0.8.1] - 2026-10-13
+
+### Fixed
+- The chat server stopped answering after the network dropped: when the
+ console's link went down, accept() failed with `SCE_NET_EINACTIVEDISABLED`
+ (errno 163, the same `0x804101a3` a dropped download reported), cpp-httplib
+ took that as fatal and closed the listening socket, and the model stayed
+ loaded with nothing to reach it. llama-server now binds again once a second
+ until it works (`patches/0005-listener-survives-network-drops.patch`).
+ Checked on the console: the listener came back within a second after each
+ of dozens of drops. A reply streaming at the moment of a drop is still cut.
+
+### Added
+- README: the web UI with a sample chat.
+
+### Found
+- The repeated "Ethernet disconnected" notifications are the console's link
+ going down and up (`eth0: link state changed to DOWN` in the kernel log,
+ up to eight times in a few seconds), with or without PS5LM running: a cable,
+ port or switch problem, not the app.
+
## [0.8.0] - 2026-10-13
### Added
diff --git a/README.md b/README.md
index 0e40c09..c5186b0 100644
--- a/README.md
+++ b/README.md
@@ -54,6 +54,11 @@ For comparison, the 27B on the CPU as a payload, streamed from disk, made about
**On your network**
+<p align="center">
+ <img src="docs/media/app/webui-chat.jpg" width="820" alt="llama.cpp's web UI served by the PS5, answering with Llama 3.2 3B">
+ <br><sub>The chat in a PC's browser, served by the PS5: Llama 3.2 3B on the console's GPU at 105 tok/s, explaining that a games console cannot run a chatbot on its own.</sub>
+</p>
+
- **Chat:** llama.cpp's web UI at `http://<console IP>:8081`, with file tools and, for vision models such as Gemma 4, images (put the model's `mmproj` file beside it).
- **API:** OpenAI Chat Completions and Responses, and Anthropic Messages, on the same port. opencode, codex, Claude Code and aider work against it: see [docs/CLI.md](docs/CLI.md).
diff --git a/docs/media/app/webui-chat.jpg b/docs/media/app/webui-chat.jpg
new file mode 100644
index 0000000..683db8f
Binary files /dev/null and b/docs/media/app/webui-chat.jpg differ
diff --git a/patches/0005-listener-survives-network-drops.patch b/patches/0005-listener-survives-network-drops.patch
new file mode 100644
index 0000000..e5decc4
--- /dev/null
+++ b/patches/0005-listener-survives-network-drops.patch
@@ -0,0 +1,55 @@
+diff --git a/tools/server/server-http.cpp b/tools/server/server-http.cpp
+index 33cbb156c..7ac78f9ad 100644
+--- a/tools/server/server-http.cpp
++++ b/tools/server/server-http.cpp
+@@ -10,6 +10,7 @@
+ #include <future>
+ #include <memory>
+ #include <string>
++#include <atomic>
+ #include <thread>
+
+ //
+@@ -23,6 +24,7 @@ public:
+ std::vector<std::thread> threads; // one thread per listener
+ std::unique_ptr<httplib::ThreadPool> pool; // single pool shared among all listeners
+ int n_threads_http = 0;
++ std::atomic<bool> stopping{false}; // stop() was asked: a listener that ends stays ended
+ };
+
+ class server_http_task_queue : public httplib::TaskQueue {
+@@ -528,9 +530,23 @@ bool server_http_context::start() {
+ pimpl->pool = std::make_unique<httplib::ThreadPool>(pimpl->n_threads_http, pimpl->n_threads_http + 1024);
+ for (size_t i = 0; i < pimpl->servers.size(); ++i) {
+ const auto & srv = pimpl->servers[i];
+- pimpl->threads.emplace_back([srv = srv.get(), addr = listening_addresses[i]] {
+- if (!srv->listen_after_bind()) {
+- SRV_ERR("listener on %s stopped unexpectedly\n", addr.c_str());
++ // PS5LM: when the network drops (link down, a new DHCP address),
++ // accept() fails with an error httplib takes as fatal and closes the
++ // listening socket: the model stays loaded with nothing to reach it.
++ // Bind again, once a second, until it works or stop() is asked.
++ const std::string host = pimpl->hosts[i];
++ const int bind_port = port;
++ pimpl->threads.emplace_back([srv = srv.get(), addr = listening_addresses[i], host, bind_port,
++ stopping = &pimpl->stopping] {
++ while (!srv->listen_after_bind() && !*stopping) {
++ SRV_ERR("listener on %s stopped unexpectedly (errno %d); listening again\n", addr.c_str(), errno);
++ do {
++ std::this_thread::sleep_for(std::chrono::seconds(1));
++ srv->stop(); // clears the failed state; the socket is closed already
++ } while (!*stopping && !srv->bind_to_port(host, bind_port));
++ if (!*stopping) {
++ SRV_INF("listener on %s is back\n", addr.c_str());
++ }
+ }
+ });
+ srv->wait_until_ready();
+@@ -546,6 +562,7 @@ bool server_http_context::start() {
+ }
+
+ void server_http_context::stop() const {
++ pimpl->stopping = true;
+ for (const auto & srv : pimpl->servers) {
+ if (srv) {
+ srv->stop();
diff --git a/ps5/app/version.hpp b/ps5/app/version.hpp
index c60081c..fce3fef 100644
--- a/ps5/app/version.hpp
+++ b/ps5/app/version.hpp
@@ -4,4 +4,4 @@
// The app's version, semver. The one place it is set: scripts read it from
// here (scripts/release-app.sh), and git tags are v<this>. Bump it with an
// entry in CHANGELOG.md.
-#define PS5LM_VERSION "0.8.0"
+#define PS5LM_VERSION "0.8.1"