STD support for the ESP-IDF framework

[ivmarkov]

Dear all,

This PR is implementing libStd support for the ESP-IDF newlib-based framework, which is the open source SDK provided by Espressif for their MCU family (esp32, esp32s2, esp32c3 and all other forthcoming ones).

Note that this PR has a sibling PR against the libc crate, which implements proper declarations for all ESP-IDF APIs which are necessary for libStd support.

Implementation approach

The ESP-IDF framework - despite being bare metal - offers a relatively complete POSIX API based on newlib. pthread, BSD sockets, file descriptors, and even a small file-system VFS layer. Perhaps the only significant exception is the lack of support for processes, which is to be expected of course on bare metal.

Therefore, the libStd support is implemented as a set of (hopefully small) changes to the sys/unix family of modules, in the form of conditional-compilation branches based either on target_os = "espidf" or in a couple of cases - based on target_env = "newlib" (the latter was already there actually and is not part of this patch).

The PR also contains two new targets:

• riscv32imc-esp-espidf
• riscv32imac-esp-espidf

... which are essentially copies of riscv32imc-unknown-none-elf and riscv32imac-unknown-none-elf, but enriched with proper linker, linker_flavor, families, os, env etc. specifications so that (a) the proper conditional compilation branches in libStd are selected when compiling with these targets and (b) the correct linker is used.

Since support for atomics is a precondition for libStd, the riscv32imc-esp-espidf target additionally is configured in such a way, so as to emit libcalls to the __sync* & __atomic* GCC functions, which are already implemented in the ESP-IDF framework. If this modification is not acceptable, we can also live with only the riscv32imac-esp-espidf target as well. While the RiscV chips of Espressif lack native atomics support, the relevant instructions are transparently emulated in the ESP-IDF framework using invalid instruction trap. This modification was implemented specifically with Rust support in mind.

Target maintainers

In case this PR eventually gets merged, you can list myself as a Target Maintainer.

More importantly, Espressif (the chip vendor) is now actively involved and embracing all Rust-related efforts which were originally a community effort. In light of that, I suppose @MabezDev - who initiated the Rust-on-Espressif efforts back in time and who now works for Espressif won't object to being listed as a maintainer as well.

EDIT: I was hinted (thanks, @Urgau) that answering the Tier 3 policy explicitly might be helpful. Answers below.

Tier 3 Target Policy - answers

A proposed target or target-specific patch that substantially changes code shared with other targets (not just target-specific code) must be reviewed and approved by the appropriate team for that shared code before acceptance.

Hopefully, the changes introduced by the ESP-IDF libStd support are rather on the small side. They are completely contained within the sys/unix set of modules (that is, aside from the obviously necessary one-liners in the unwind crate and in build.rs).

A tier 3 target must have a designated developer or developers (the "target maintainers") on record to be CCed when issues arise regarding the target. (The mechanism to track and CC such developers may evolve over time.)

@ivmarkov
@MabezDev

Targets must use naming consistent with any existing targets; for instance, a target for the same CPU or OS as an existing Rust target should use the same name for that CPU or OS. Targets should normally use the same names and naming conventions as used elsewhere in the broader ecosystem beyond Rust (such as in other toolchains), unless they have a very good reason to diverge. Changing the name of a target can be highly disruptive, especially once the target reaches a higher tier, so getting the name right is important even for a tier 3 target.

The two introduced targets follow as much as possible the naming conventions of the other targets. I.e. taking the bare-metal riscv32imac_unknown_none_elf as a base:

• The name of the new target was derived by replacing none with espidf to designate the target_os.
• _elf was removed, as the non-bare metal targets seem not to have it
• -newlib was deliberately NOT added at the end, as I believe the chance of having two simultaneously active separate targets for the ESP-IDF framework with different C libraries (say, newlib vs musl) is way too small
• Finally, we replaced the middle unknown with esp which is kind of the name of the whole chipset MCU family (and abbreviation from Espressif which is too long). It will stay esp for all RiscV32-based MCUs of the company, as they all use the riscv32imc instruction set. By necessity however (disambiguation), it will be esp32 or esp32s2 or esp32s3 for the Xtensa-based MCUs as all of these have their own variation of the Xtensa architecture. (The Xtensa targets are not part of this PR, even though they would use 1:1 the same LibStd implementation provided here, as they depend on the upstreaming of the Xtensa architecture support in LLVM; this upstreaming this is currently in progress.)

There was also a preceding discussion on the topic here.

Target names should not introduce undue confusion or ambiguity unless absolutely necessary to maintain ecosystem compatibility. For example, if the name of the target makes people extremely likely to form incorrect beliefs about what it targets, the name should be changed or augmented to disambiguate it.

We are explicitly putting an -espidf suffix to designate that the target is specifically for Rust + ESP-IDF

Tier 3 targets may have unusual requirements to build or use, but must not create legal issues or impose onerous legal terms for the Rust project or for Rust developers or users.

Agreed.

The target must not introduce license incompatibilities.

To the best of our knowledge, it doesn't.

Anything added to the Rust repository must be under the standard Rust license (MIT OR Apache-2.0).

MIT + Apache 2.0

The target must not cause the Rust tools or libraries built for any other host (even when supporting cross-compilation to the target) to depend on any new dependency less permissive than the Rust licensing policy. This applies whether the dependency is a Rust crate that would require adding new license exceptions (as specified by the tidy tool in the rust-lang/rust repository), or whether the dependency is a native library or binary. In other words, the introduction of the target must not cause a user installing or running a version of Rust or the Rust tools to be subject to any new license requirements.

Requirements are not changed for any other target.

If the target supports building host tools (such as rustc or cargo), those host tools must not depend on proprietary (non-FOSS) libraries, other than ordinary runtime libraries supplied by the platform and commonly used by other binaries built for the target. For instance, rustc built for the target may depend on a common proprietary C runtime library or console output library, but must not depend on a proprietary code generation library or code optimization library. Rust's license permits such combinations, but the Rust project has no interest in maintaining such combinations within the scope of Rust itself, even at tier 3.

The targets are for bare-metal environment which is not hosting build tools or a compiler.

Targets should not require proprietary (non-FOSS) components to link a functional binary or library.

The linker used by the targets is the GCC linker from the GCC toolchain cross-compiled for riscv. GNU GPL.

"onerous" here is an intentionally subjective term. At a minimum, "onerous" legal/licensing terms include but are not limited to: non-disclosure requirements, non-compete requirements, contributor license agreements (CLAs) or equivalent, "non-commercial"/"research-only"/etc terms, requirements conditional on the employer or employment of any particular Rust developers, revocable terms, any requirements that create liability for the Rust project or its developers or users, or any requirements that adversely affect the livelihood or prospects of the Rust project or its developers or users.
Neither this policy nor any decisions made regarding targets shall create any binding agreement or estoppel by any party. If any member of an approving Rust team serves as one of the maintainers of a target, or has any legal or employment requirement (explicit or implicit) that might affect their decisions regarding a target, they must recuse themselves from any approval decisions regarding the target's tier status, though they may otherwise participate in discussions.
This requirement does not prevent part or all of this policy from being cited in an explicit contract or work agreement (e.g. to implement or maintain support for a target). This requirement exists to ensure that a developer or team responsible for reviewing and approving a target does not face any legal threats or obligations that would prevent them from freely exercising their judgment in such approval, even if such judgment involves subjective matters or goes beyond the letter of these requirements.

Agreed.

Tier 3 targets should attempt to implement as much of the standard libraries as possible and appropriate (core for most targets, alloc for targets that can support dynamic memory allocation, std for targets with an operating system or equivalent layer of system-provided functionality), but may leave some code unimplemented (either unavailable or stubbed out as appropriate), whether because the target makes it impossible to implement or challenging to implement. The authors of pull requests are not obligated to avoid calling any portions of the standard library on the basis of a tier 3 target not implementing those portions.

The targets implement libStd almost in its entirety, except for the missing support for process, as this is a bare metal platform. The process sys\unix module is currently stubbed to return "not implemented" errors.

The target must provide documentation for the Rust community explaining how to build for the target, using cross-compilation if possible. If the target supports running tests (even if they do not pass), the documentation must explain how to run tests for the target, using emulation if possible or dedicated hardware if necessary.

Target does not (yet) support running tests. We would gladly provide all documentation how to build for the target (where?). It is currently hosted in this README.md file, but will likely be moved to the esp-rs organization. Since the build for the target is driven by cargo and all other tooling is downloaded automatically during the build, there is no need for extensive documentation.

Tier 3 targets must not impose burden on the authors of pull requests, or other developers in the community, to maintain the target. In particular, do not post comments (automated or manual) on a PR that derail or suggest a block on the PR based on a tier 3 target. Do not send automated messages or notifications (via any medium, including via @) to a PR author or others involved with a PR regarding a tier 3 target, unless they have opted into such messages.

Agreed.

Backlinks such as those generated by the issue/PR tracker when linking to an issue or PR are not considered a violation of this policy, within reason. However, such messages (even on a separate repository) must not generate notifications to anyone involved with a PR who has not requested such notifications.

Agreed.

Patches adding or updating tier 3 targets must not break any existing tier 2 or tier 1 target, and must not knowingly break another tier 3 target without approval of either the compiler team or the maintainers of the other tier 3 target.

To the best of our knowledge, we believe we are not breaking any other target (be it tier 1, 2 or 3).

In particular, this may come up when working on closely related targets, such as variations of the same architecture with different features. Avoid introducing unconditional uses of features that another variation of the target may not have; use conditional compilation or runtime detection, as appropriate, to let each target run code supported by that target.

To the best of our knowledge, we have not introduced any unconditional use of a feature that affects any other target.

If a tier 3 target stops meeting these requirements, or the target maintainers no longer have interest or time, or the target shows no signs of activity and has not built for some time, or removing the target would improve the quality of the Rust codebase, we may post a PR to remove it; any such PR will be CCed to the target maintainers (and potentially other people who have previously worked on the target), to check potential interest in improving the situation.

Agreed.

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @dtolnay (or someone else) soon.

Please see the contribution instructions for more information.

[rust-highfive]

⚠️ Warning ⚠️

• These commits modify compiler targets. (See the Target Tier Policy.)

[dtolnay]

r? @rust-lang/libs

[the8472]

Note that as part of only supporting fallible allocation in the linux kernel (rust-lang/rfcs#3140) there is ongoing discussion around making std features optional and how to support that on the cargo side. rust-lang/rfcs#3146

I'm not sure what exactly the roadmap on those parts are but it sounds like it would be appropriate here if you only want to support a subset of the standard library in the beginning. But putting whole modules behind feature gates that have yet to be introduced instead of stubbing them could require some work.

[ivmarkov]

Note that as part of only supporting fallible allocation in the linux kernel (rust-lang/rfcs#3140) there is ongoing discussion around making std features optional and how to support that on the cargo side. rust-lang/rfcs#3146

I'm not sure what exactly the roadmap on those parts are but it sounds like it would be appropriate here if you only want to support a subset of the standard library in the beginning. But putting whole modules behind feature gates that have yet to be introduced instead of stubbing them could require some work.

That's very useful info, thanks for sharing!

My problem with it is a little bit that this patch set has been in development since approx last November, and waiting for another Cargo nightly feature to first of all hit nightly, and then suffering through its stabilization pains, change of semantics and so on might be a bit too much. Especially considering that we are already sailing in an ocean of Cargo unstable features, that - when changed or regressed - did hit us in the past:

• build-std
• Extra link arg
• Configurable env
• Artifact dependencies - lower prio; waiting on that, hasn't even hit nightly yet

So I would rather take a conservative approach here and stub what is not supported, if there is no strong, well-grounded objection from the Rust maintainers.

[bors]

☔ The latest upstream changes (presumably #87535) made this pull request unmergeable. Please resolve the merge conflicts.

[Amanieu]

Since support for atomics is a precondition for libStd, the riscv32imc-esp-espidf target additionally is configured in such a way, so as to emit libcalls to the __sync* & __atomic* GCC functions, which are already implemented in the ESP-IDF framework. If this modification is not acceptable, we can also live with only the riscv32imac-esp-espidf target as well. While the RiscV chips of Espressif lack native atomics support, the relevant instructions are transparently emulated in the ESP-IDF framework using invalid instruction trap. This modification was implemented specifically with Rust support in mind.

This is fine as long as the underlying implementation is lock-free (doesn't use a lock internally to emulate atomics). The requirement here is that a signal/interrupt handler should never deadlock if it tries to use an Atomic* which the interrupted thread was in the middle of accessing.

Generally this is implemented by disabling interrupt handlers or by having the interrupt handler restart the atomic sequence in the interrupted thread upon returning. We already have something similar for ARMv4 on Linux (atomic instructions were introduced in ARMv6) where support for atomic operations is provided by the kernel: https://github.com/rust-lang/compiler-builtins/blob/master/src/arm_linux.rs

[ivmarkov]

Now that libc 0.2.99 is released, I've forced pushed the following two updates:

• Modification of ./Cargo.lock and ./library/std/Cargo.toml - only change is that the libc dependency is now raised to 0.2.99
• I've rebased all commits so far into a single one to keep the rust master branch history cleaner. Other than the rebase, no changes to the already reviewed code

[Amanieu]

@bors r+

[bors]

📌 Commit 459eaa6 has been approved by Amanieu

[bors]

⌛ Testing commit 459eaa6 with merge 6bed1f0...

[bors]

☀️ Test successful - checks-actions
Approved by: Amanieu
Pushing 6bed1f0 to master...