• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //! An ergonomic and easy-to-integrate implementation of the
2 //! [GDB Remote Serial Protocol](https://sourceware.org/gdb/onlinedocs/gdb/Remote-Protocol.html#Remote-Protocol)
3 //! in Rust, with full `#![no_std]` support.
4 //!
5 //! ## Feature flags
6 //!
7 //! By default, both the `std` and `alloc` features are enabled.
8 //!
9 //! When using `gdbstub` in `#![no_std]` contexts, make sure to set
10 //! `default-features = false`.
11 //!
12 //! - `alloc`
13 //!     - Implement `Connection` for `Box<dyn Connection>`.
14 //!     - Log outgoing packets via `log::trace!` (uses a heap-allocated output
15 //!       buffer).
16 //!     - Provide built-in implementations for certain protocol features:
17 //!         - Use a heap-allocated packet buffer in `GdbStub` (if none is
18 //!           provided via `GdbStubBuilder::with_packet_buffer`).
19 //!         - (Monitor Command) Use a heap-allocated output buffer in
20 //!           `ConsoleOutput`.
21 //! - `std` (implies `alloc`)
22 //!     - Implement `Connection` for [`TcpStream`](std::net::TcpStream) and
23 //!       [`UnixStream`](std::os::unix::net::UnixStream).
24 //!     - Implement [`std::error::Error`] for `gdbstub::Error`.
25 //!     - Add a `TargetError::Io` variant to simplify `std::io::Error` handling
26 //!       from Target methods.
27 //! - `paranoid_unsafe`
28 //!     - Please refer to the [`unsafe` in `gdbstub`](https://github.com/daniel5151/gdbstub#unsafe-in-gdbstub)
29 //!       section of the README.md for more details.
30 //!
31 //! ## Getting Started
32 //!
33 //! This section provides a brief overview of the key traits and types used in
34 //! `gdbstub`, and walks though the basic steps required to integrate `gdbstub`
35 //! into a project.
36 //!
37 //! At a high level, there are only three things that are required to get up and
38 //! running with `gdbstub`: a [`Connection`](#the-connection-trait), a
39 //! [`Target`](#the-target-trait), and a [event loop](#the-event-loop).
40 //!
41 //! > _Note:_ I _highly recommended_ referencing some of the
42 //! [examples](https://github.com/daniel5151/gdbstub#examples) listed in the
43 //! project README when integrating `gdbstub` into a project for the first time.
44 //!
45 //! > In particular, the in-tree
46 //! [`armv4t`](https://github.com/daniel5151/gdbstub/tree/master/examples/armv4t)
47 //! example contains basic implementations off almost all protocol extensions,
48 //! making it an incredibly valuable reference when implementing protocol
49 //! extensions.
50 //!
51 //! ### The `Connection` Trait
52 //!
53 //! First things first: `gdbstub` needs some way to communicate with a GDB
54 //! client. To facilitate this communication, `gdbstub` uses a custom
55 //! [`Connection`](conn::Connection) trait.
56 //!
57 //! `Connection` is automatically implemented for common `std` types such as
58 //! [`TcpStream`](std::net::TcpStream) and
59 //! [`UnixStream`](std::os::unix::net::UnixStream).
60 //!
61 //! If you're using `gdbstub` in a `#![no_std]` environment, `Connection` will
62 //! most likely need to be manually implemented on top of whatever in-order,
63 //! serial, byte-wise I/O your particular platform has available (e.g:
64 //! putchar/getchar over UART, using an embedded TCP stack, etc.).
65 //!
66 //! One common way to start a remote debugging session is to simply wait for a
67 //! GDB client to connect via TCP:
68 //!
69 //! ```rust
70 //! use std::io;
71 //! use std::net::{TcpListener, TcpStream};
72 //!
73 //! fn wait_for_gdb_connection(port: u16) -> io::Result<TcpStream> {
74 //!     let sockaddr = format!("localhost:{}", port);
75 //!     eprintln!("Waiting for a GDB connection on {:?}...", sockaddr);
76 //!     let sock = TcpListener::bind(sockaddr)?;
77 //!     let (stream, addr) = sock.accept()?;
78 //!
79 //!     // Blocks until a GDB client connects via TCP.
80 //!     // i.e: Running `target remote localhost:<port>` from the GDB prompt.
81 //!
82 //!     eprintln!("Debugger connected from {}", addr);
83 //!     Ok(stream) // `TcpStream` implements `gdbstub::Connection`
84 //! }
85 //! ```
86 //!
87 //! ### The `Target` Trait
88 //!
89 //! The [`Target`](target::Target) trait describes how to control and modify
90 //! a system's execution state during a GDB debugging session, and serves as the
91 //! primary bridge between `gdbstub`'s generic GDB protocol implementation and a
92 //! specific target's project/platform-specific code.
93 //!
94 //! At a high level, the `Target` trait is a collection of user-defined handler
95 //! methods that the GDB client can invoke via the GDB remote serial protocol.
96 //! For example, the `Target` trait includes methods to read/write
97 //! registers/memory, start/stop execution, etc...
98 //!
99 //! **`Target` is the most important trait in `gdbstub`, and must be implemented
100 //! by anyone integrating `gdbstub` into their project!**
101 //!
102 //! Please refer to the [`target` module documentation](target) for in-depth
103 //! instructions on how to implement [`Target`](target::Target) for a particular
104 //! platform.
105 //!
106 //! ## The Event Loop
107 //!
108 //! Once a [`Connection`](#the-connection-trait) has been established and the
109 //! [`Target`](#the-target-trait) has been initialized, all that's left is to
110 //! wire things up and decide what kind of event loop will be used to drive the
111 //! debugging session!
112 //!
113 //! First things first, let's get an instance of `GdbStub` ready to run:
114 //!
115 //! ```rust,ignore
116 //! // Set-up a valid `Target`
117 //! let mut target = MyTarget::new()?; // implements `Target`
118 //!
119 //! // Establish a `Connection`
120 //! let connection: TcpStream = wait_for_gdb_connection(9001);
121 //!
122 //! // Create a new `gdbstub::GdbStub` using the established `Connection`.
123 //! let mut debugger = gdbstub::GdbStub::new(connection);
124 //! ```
125 //!
126 //! Cool, but how do you actually start the debugging session?
127 // use an explicit doc attribute to avoid automatic rustfmt wrapping
128 #![doc = "### `GdbStub::run_blocking`: The quick and easy way to get up and running with `gdbstub`"]
129 //!
130 //! If you've got an extra thread to spare, the quickest way to get up and
131 //! running with `gdbstub` is by using the
132 //! [`GdbStub::run_blocking`](stub::run_blocking) API alongside the
133 //! [`BlockingEventLoop`] trait.
134 //!
135 //! If you are on a more resource constrained platform, and/or don't wish to
136 //! dedicate an entire thread to `gdbstub`, feel free to skip ahead to the
137 //! [following
138 //! section](#gdbstubstatemachine-driving-gdbstub-in-an-async-event-loop--via-interrupt-handlers).
139 //!
140 //! A basic integration of `gdbstub` into a project using the
141 //! `GdbStub::run_blocking` API might look something like this:
142 //!
143 //! ```rust
144 //! # use gdbstub::target::ext::base::BaseOps;
145 //! #
146 //! # struct MyTarget;
147 //! #
148 //! # impl Target for MyTarget {
149 //! #     type Error = &'static str;
150 //! #     type Arch = gdbstub_arch::arm::Armv4t; // as an example
151 //! #     fn base_ops(&mut self) -> BaseOps<Self::Arch, Self::Error> { todo!() }
152 //! # }
153 //! #
154 //! # impl MyTarget {
155 //! #     fn run_and_check_for_incoming_data(
156 //! #         &mut self,
157 //! #         conn: &mut impl Connection
158 //! #     ) -> MyTargetEvent { todo!() }
159 //! #
160 //! #     fn stop_in_response_to_ctrl_c_interrupt(
161 //! #         &mut self
162 //! #     ) -> Result<(), &'static str> { todo!() }
163 //! # }
164 //! #
165 //! # enum MyTargetEvent {
166 //! #     IncomingData,
167 //! #     StopReason(SingleThreadStopReason<u32>),
168 //! # }
169 //! #
170 //! use gdbstub::common::Signal;
171 //! use gdbstub::conn::{Connection, ConnectionExt}; // note the use of `ConnectionExt`
172 //! use gdbstub::stub::{run_blocking, DisconnectReason, GdbStub, GdbStubError};
173 //! use gdbstub::stub::SingleThreadStopReason;
174 //! use gdbstub::target::Target;
175 //!
176 //! enum MyGdbBlockingEventLoop {}
177 //!
178 //! // The `run_blocking::BlockingEventLoop` groups together various callbacks
179 //! // the `GdbStub::run_blocking` event loop requires you to implement.
180 //! impl run_blocking::BlockingEventLoop for MyGdbBlockingEventLoop {
181 //!     type Target = MyTarget;
182 //!     type Connection = Box<dyn ConnectionExt<Error = std::io::Error>>;
183 //!
184 //!     // or MultiThreadStopReason on multi threaded targets
185 //!     type StopReason = SingleThreadStopReason<u32>;
186 //!
187 //!     // Invoked immediately after the target's `resume` method has been
188 //!     // called. The implementation should block until either the target
189 //!     // reports a stop reason, or if new data was sent over the connection.
190 //!     fn wait_for_stop_reason(
191 //!         target: &mut MyTarget,
192 //!         conn: &mut Self::Connection,
193 //!     ) -> Result<
194 //!         run_blocking::Event<SingleThreadStopReason<u32>>,
195 //!         run_blocking::WaitForStopReasonError<
196 //!             <Self::Target as Target>::Error,
197 //!             <Self::Connection as Connection>::Error,
198 //!         >,
199 //!     > {
200 //!         // the specific mechanism to "select" between incoming data and target
201 //!         // events will depend on your project's architecture.
202 //!         //
203 //!         // some examples of how you might implement this method include: `epoll`,
204 //!         // `select!` across multiple event channels, periodic polling, etc...
205 //!         //
206 //!         // in this example, lets assume the target has a magic method that handles
207 //!         // this for us.
208 //!         let event = match target.run_and_check_for_incoming_data(conn) {
209 //!             MyTargetEvent::IncomingData => {
210 //!                 let byte = conn
211 //!                     .read() // method provided by the `ConnectionExt` trait
212 //!                     .map_err(run_blocking::WaitForStopReasonError::Connection)?;
213 //!
214 //!                 run_blocking::Event::IncomingData(byte)
215 //!             }
216 //!             MyTargetEvent::StopReason(reason) => {
217 //!                 run_blocking::Event::TargetStopped(reason)
218 //!             }
219 //!         };
220 //!
221 //!         Ok(event)
222 //!     }
223 //!
224 //!     // Invoked when the GDB client sends a Ctrl-C interrupt.
225 //!     fn on_interrupt(
226 //!         target: &mut MyTarget,
227 //!     ) -> Result<Option<SingleThreadStopReason<u32>>, <MyTarget as Target>::Error> {
228 //!         // notify the target that a ctrl-c interrupt has occurred.
229 //!         target.stop_in_response_to_ctrl_c_interrupt()?;
230 //!
231 //!         // a pretty typical stop reason in response to a Ctrl-C interrupt is to
232 //!         // report a "Signal::SIGINT".
233 //!         Ok(Some(SingleThreadStopReason::Signal(Signal::SIGINT).into()))
234 //!     }
235 //! }
236 //!
237 //! fn gdb_event_loop_thread(
238 //!     debugger: GdbStub<MyTarget, Box<dyn ConnectionExt<Error = std::io::Error>>>,
239 //!     mut target: MyTarget
240 //! ) {
241 //!     match debugger.run_blocking::<MyGdbBlockingEventLoop>(&mut target) {
242 //!         Ok(disconnect_reason) => match disconnect_reason {
243 //!             DisconnectReason::Disconnect => {
244 //!                 println!("Client disconnected")
245 //!             }
246 //!             DisconnectReason::TargetExited(code) => {
247 //!                 println!("Target exited with code {}", code)
248 //!             }
249 //!             DisconnectReason::TargetTerminated(sig) => {
250 //!                 println!("Target terminated with signal {}", sig)
251 //!             }
252 //!             DisconnectReason::Kill => println!("GDB sent a kill command"),
253 //!         },
254 //!         Err(GdbStubError::TargetError(e)) => {
255 //!             println!("target encountered a fatal error: {}", e)
256 //!         }
257 //!         Err(e) => {
258 //!             println!("gdbstub encountered a fatal error: {}", e)
259 //!         }
260 //!     }
261 //! }
262 //! ```
263 // use an explicit doc attribute to avoid automatic rustfmt wrapping
264 #![doc = "### `GdbStubStateMachine`: Driving `gdbstub` in an async event loop / via interrupt handlers"]
265 //!
266 //! `GdbStub::run_blocking` requires that the target implement the
267 //! [`BlockingEventLoop`] trait, which as the name implies, uses _blocking_ IO
268 //! when handling certain events. Blocking the thread is a totally reasonable
269 //! approach in most implementations, as one can simply spin up a separate
270 //! thread to run the GDB stub (or in certain emulator implementations, run the
271 //! emulator as part of the `wait_for_stop_reason` method).
272 //!
273 //! Unfortunately, this blocking behavior can be a non-starter when integrating
274 //! `gdbstub` in projects that don't support / wish to avoid the traditional
275 //! thread-based execution model, such as projects using `async/await`, or
276 //! bare-metal `no_std` projects running on embedded hardware.
277 //!
278 //! In these cases, `gdbstub` provides access to the underlying
279 //! [`GdbStubStateMachine`] API, which gives implementations full control over
280 //! the GDB stub's "event loop". This API requires implementations to "push"
281 //! data to the `gdbstub` implementation whenever new data becomes available
282 //! (e.g: when a UART interrupt handler receives a byte, when the target hits a
283 //! breakpoint, etc...), as opposed to the `GdbStub::run_blocking` API, which
284 //! "pulls" these events in a blocking manner.
285 //!
286 //! See the [`GdbStubStateMachine`] docs for more details on how to use this
287 //! API.
288 //!
289 //! <br>
290 //!
291 //! * * *
292 //!
293 //! <br>
294 //!
295 //! And with that lengthy introduction, I wish you the best of luck in your
296 //! debugging adventures!
297 //!
298 //! If you have any suggestions, feature requests, or run into any problems,
299 //! please start a discussion / open an issue over on the
300 //! [`gdbstub` GitHub repo](https://github.com/daniel5151/gdbstub/).
301 //!
302 //! [`GdbStubStateMachine`]: stub::state_machine::GdbStubStateMachine
303 //! [`BlockingEventLoop`]: stub::run_blocking::BlockingEventLoop
304 
305 #![cfg_attr(not(feature = "std"), no_std)]
306 #![cfg_attr(feature = "paranoid_unsafe", forbid(unsafe_code))]
307 
308 #[cfg(feature = "alloc")]
309 extern crate alloc;
310 
311 #[macro_use]
312 extern crate log;
313 
314 mod protocol;
315 mod util;
316 
317 #[doc(hidden)]
318 pub mod internal;
319 
320 pub mod arch;
321 pub mod common;
322 pub mod conn;
323 pub mod stub;
324 pub mod target;
325 
326 // https://users.rust-lang.org/t/compile-time-const-unwrapping/51619/7
327 //
328 // This works from Rust 1.46.0 onwards, which stabilized branching and looping
329 // in const contexts.
330 macro_rules! unwrap {
331     ($e:expr $(,)*) => {
332         match $e {
333             ::core::option::Option::Some(x) => x,
334             ::core::option::Option::None => {
335                 ["tried to unwrap a None"][99];
336                 loop {}
337             }
338         }
339     };
340 }
341 
342 /// (Internal) The fake Tid that's used when running in single-threaded mode.
343 const SINGLE_THREAD_TID: common::Tid = unwrap!(common::Tid::new(1));
344 /// (Internal) The fake Pid reported to GDB when running in multi-threaded mode.
345 const FAKE_PID: common::Pid = unwrap!(common::Pid::new(1));
346 
347 pub(crate) mod is_valid_tid {
348     pub trait IsValidTid {}
349 
350     impl IsValidTid for () {}
351     impl IsValidTid for crate::common::Tid {}
352 }
353