[WIP] Add panic implementation docs #521
Conversation
|
Thanks, this looks useful! I too am trying to better understand the panic machinery. I was wondering if you could define more what a "landing pad" is? I assume it is some reserved section of the stack frame, that does…something related to unwinding? It sounds like it has a list of destructors to run, is that all it does? It might also be helpful to mention that there are two different panic strategies (encoded per crate), which are only loosely coupled with the panic runtimes. For example, the abort runtime can use crates built with unwind strategy, but not the other way around. The abort strategy disables landing pads, but I'd like to know what else changing the strategy does (or maybe landing pads are the only difference?). |
There was a problem hiding this comment.
@Aaron1011 Thanks! This is very helpful info, and I have wondered about it before many times.
My biggest request is to please add links for all of the names you reference (the first time you reference them) to the source code (since there don't appear to be rustdocs for most of these definitions). This helps us check for breakage by just doing linkchecking.
Thanks!
| [cfg(not(test))] | ||
| [panic_handler] | ||
| [unwind(allowed)] |
There was a problem hiding this comment.
It looks like you are missing # before these lines?
There was a problem hiding this comment.
I can't seem to figure out how to stop the # from being interpreted as the start of a comment.
There was a problem hiding this comment.
I don't think md has comments? Can you elaborate on what the concern is?
There was a problem hiding this comment.
When I added # marks, the lines didn't render in my web browser.
There was a problem hiding this comment.
Hmm... Ok, we can figure this out in a followup PR. I think we can go ahead and merge without this.
@Aaron1011 This would be a great addition to the glossary appendix if you have time. |
|
Not knowing about this PR, I wrote a blog post about this very topic. Sorry for the duplication of work! We can likely copy some things from that post into this PR. EDIT: You can find the markdown sources here, for easier copying. |
|
The CI failure seems unrelated |
|
@Aaron1011 Could you rebase on top of the latest master please? Hopefully that will get rid of the CI failures. Also, it would be awesome if you could add as many links as possible, as this helps us to detect changes. |
Co-Authored-By: Who? Me?! <mark-i-m@users.noreply.github.com>
|
@mark-i-m: The CI failures didn't go away after rebasing |
|
@Aaron1011 Could you add a line break not to exceed 100 characters of each line? |
|
@JohnTitor Something odd is going on here... these files are not touched by the PR, and they are working just fine on master.. |
|
Uhm, it's odd. The checker here doesn't ignore lines inside a code block... |
There was a problem hiding this comment.
The unnecessary code blocks cause the failure, it should be gone once you apply my suggestions and do rebase. See also my branch: https://github.com/JohnTitor/rustc-guide/tree/panic
|
Ping @Aaron1011 could you please apply the suggestions? It would be great to merge this chapter. |
Co-Authored-By: Yuki Okushi <huyuumi.dev@gmail.com>
|
@mark-i-m: I've applied @JohnTitor's suggestions. However, I still can't figure out how to get lines starting with |
This PR adds a descirption of the implementation of panicking in Rust, starting from the
panic!macros and ending at the invocation of the platforms-specific unwind logic.This is not fully complete - for one thing, it might be useful to have a description of how this works under Miri.
I developed this while working on the implementation of panics in Miri. Having this document available would have made things a lot easier for me 😄
Closes #486