daemon: rpc/ — Unix-socket transport + generated dispatch wiring (build step 1)

First real code in daemon/. veloxd now listens on
$XDG_RUNTIME_DIR/velox/velox.sock (0600, SO_PEERCRED same-UID check),
frames NDJSON, and routes every method through the generated
velox::proto::dispatch(). The CLI and GUI have a server to talk to.

Modules:
- rpc/ndjson.hpp    — newline-delimited framing, 8 MiB frame cap, CRLF-
                       tolerant, partial-tail buffering. Header-only, tested.
- rpc/event_loop    — single-threaded poll(2) reactor; never blocks the
                       loop. stop()/wake() are async-signal-safe (eventfd).
- rpc/runtime_dir   — $XDG_RUNTIME_DIR/velox resolution, 0700, owner-checked;
                       refuses an insecure fallback rather than using /tmp.
- rpc/uds_server    — listener + non-blocking per-conn read/write with
                       backpressure; handles session.hello (protocol-major
                       check -> -32001, sessionId, transport=uds) and
                       session.subscribe in the server layer; routes the
                       rest through dispatch().
- rpc/dispatcher    — VeloxDispatcher : proto::Dispatcher, all 39 methods.
                       download.list answers an empty table; the rest return
                       "not implemented" (-> -32603) until the store lands.
- main.cpp          — abstract-namespace single-instance lock, signal ->
                       clean shutdown, socket unlinked on exit.

Tests (ASan+UBSan and TSan clean):
- veloxd.ndjson         — framing edge cases
- veloxd.uds_roundtrip  — real socket: hello ok / version mismatch / empty
                          list / -32601 / -32700 / pipelined requests, and a
                          guard on the -32603 collapse documented in P1.

Known gap, filed not worked around: daemon/docs/proto-requests-m1.md P1 —
the generated Dispatcher has no error channel below -32603, so handlers
cannot yet return -32010/-32011/-32013 with their data payloads. The
server layer handles -32001/-32002/-32003 around dispatch(); genuine
in-handler errors collapse to -32603 until PROTO gives handlers a real
error return. Three error fixtures are non-conformant until then.

Not in this drop: rpc/ws_server (next; needs the store for hashed pairing
tokens), store/, sched/, cli/. WS reuses this event loop.

Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01Upd9WhG9oppieig5nRDLig
This commit is contained in:
2026-09-10 15:27:20 +04:00
co-authored by Claude Sonnet 5
parent 170bcfdb3e
commit e60d6669d8
18 changed files with 1563 additions and 0 deletions
+69
View File
@@ -0,0 +1,69 @@
#pragma once
// A single-threaded poll(2) reactor. Every RPC listener and connection registers its fd
// here; the loop never blocks on disk or DNS (AGENT-DAEMON.md build step 1 — "Never block
// the RPC loop"). Long work is handed to CORE's pools later; this class only multiplexes
// readiness.
//
// Thread model: run() executes on one thread. add_fd/mod_fd/del_fd are called from
// callbacks on that same thread. stop() and wake() are async-signal-safe and safe to call
// from any thread or a signal handler — they only write() a byte to an internal eventfd.
#include <atomic>
#include <cstdint>
#include <functional>
#include <unordered_map>
namespace velox::daemon::rpc {
enum Interest : unsigned {
kNone = 0,
kRead = 1u << 0,
kWrite = 1u << 1,
};
class EventLoop {
public:
// Called when the fd is readable and/or writable. `events` is the subset of the fd's
// registered Interest that fired. A callback may add/modify/remove any fd, including
// its own, and may call stop().
using Callback = std::function<void(int fd, unsigned events)>;
EventLoop();
~EventLoop();
EventLoop(const EventLoop&) = delete;
EventLoop& operator=(const EventLoop&) = delete;
// Register `fd` (must be non-blocking) for `interest`. Replaces any prior registration.
void add_fd(int fd, unsigned interest, Callback cb);
// Change the interest mask for an already-registered fd.
void mod_fd(int fd, unsigned interest);
// Stop watching `fd`. Does not close it — ownership stays with the caller.
void del_fd(int fd);
// Run until stop() is called. Re-entrant calls are not supported.
void run();
// Ask run() to return after the current poll wakeup. Async-signal-safe.
void stop() noexcept;
// Force one poll() wakeup without stopping — used when interest changed from outside a
// callback. Async-signal-safe.
void wake() noexcept;
private:
struct Entry {
unsigned interest;
Callback cb;
};
void drain_wakeup() noexcept;
int wake_fd_; // eventfd, always registered
bool running_ = false;
std::atomic<bool> stop_requested_ = false; // set from stop(), read by run()
std::unordered_map<int, Entry> fds_;
};
} // namespace velox::daemon::rpc