1 //! Community-created implementations of [`gdbstub::arch::Arch`] for various
2 //! architectures.
3 //!
4 //! _Note:_ If an architecture is missing from this crate, that does _not_ mean
5 //! that it can't be used with `gdbstub`! So-long as there's support for the
6 //! target architecture in GDB, it should be fairly straightforward to implement
7 //! `Arch` manually.
8 //!
9 //! Please consider upstreaming any missing `Arch` implementations you happen to
10 //! implement yourself! Aside from the altruistic motive of improving `gdbstub`,
11 //! upstreaming your `Arch` implementation will ensure that it's kept up-to-date
12 //! with any future breaking API changes.
13 //!
14 //! **Disclaimer:** These implementations are all community contributions, and
15 //! while they are tested (by the PR's author) and code-reviewed, it's not
16 //! particularly feasible to write detailed tests for each architecture! If you
17 //! spot a bug in any of the implementations, please file an issue / open a PR!
18 //!
19 //! # What's with `RegIdImpl`?
20 //!
21 //! Supporting the `Target::read/write_register` API required introducing a new
22 //! [`RegId`] trait + [`Arch::RegId`] associated type. `RegId` is used by
23 //! `gdbstub` to translate raw GDB register ids (a protocol level arch-dependent
24 //! `usize`) into human-readable enum variants.
25 //!
26 //! Unfortunately, this API was added after several contributors had already
27 //! upstreamed their `Arch` implementations, and as a result, there are several
28 //! built-in arch implementations which are missing proper `RegId` enums
29 //! (tracked under [issue #29](https://github.com/daniel5151/gdbstub/issues/29)).
30 //!
31 //! As a stop-gap measure, affected `Arch` implementations have been modified to
32 //! accept a `RegIdImpl` type parameter, which requires users to manually
33 //! specify a `RegId` implementation.
34 //!
35 //! If you're not interested in implementing the `Target::read/write_register`
36 //! methods and just want to get up-and-running with `gdbstub`, it's fine to
37 //! set `RegIdImpl` to `()` and use the built-in stubbed `impl RegId for ()`.
38 //!
39 //! A better approach would be to implement (and hopefully upstream!) a proper
40 //! `RegId` enum. While this will require doing a bit of digging through the GDB
41 //! docs + [architecture XML definitions](https://github.com/bminor/binutils-gdb/tree/master/gdb/features/),
42 //! it's not too tricky to get a working implementation up and running, and
43 //! makes it possible to safely and efficiently implement the
44 //! `Target::read/write_register` API. As an example, check out
45 //! [`ArmCoreRegId`](arm::reg::id::ArmCoreRegId#impl-RegId).
46 //!
47 //! Whenever a `RegId` enum is upstreamed, the associated `Arch`'s `RegIdImpl`
48 //! parameter will be defaulted to the newly added enum. This will simplify the
49 //! API without requiring an explicit breaking API change. Once all `RegIdImpl`
50 //! have a default implementation, only a single breaking API change will be
51 //! required to remove `RegIdImpl` entirely (along with this documentation).
52 
53 #![cfg_attr(not(test), no_std)]
54 #![deny(missing_docs)]
55 
56 pub mod aarch64;
57 pub mod arm;
58 pub mod mips;
59 pub mod msp430;
60 pub mod ppc;
61 pub mod riscv;
62 pub mod x86;
63 
64 // used as part of intra-doc link
65 #[allow(unused_imports)]
66 use gdbstub::arch::*;
67