Use #[expect(…)] instead of #[allow(…)], remove unnecessary allows #2784
Conversation
|
Feel free to fix this for all build targets - but it might be a world of pain :) |
|
I did some more, but I'll leave the rest to someone else. |
|
Also: I cannot guarantee that I haven't broken anything. I've manually checked as much as possible and the CI found stuff as well, but I may have still missed some cfg-dependent stuff (there's a lot in there lol). On the upside, I've only touched linting-related things. So if anyone's computer starts complaining, it's easy to just re-add the warning or change it back from an |
|
|
||
| # start manager in background, then start one client | ||
| timeout 32s ./${FUZZER_NAME} --cores 0 --input ./corpus 2>/dev/null & | ||
| timeout 31s ./${FUZZER_NAME} --cores 1 --input ./corpus | tee fuzz_stdout.log 2>/dev/null || true |
There was a problem hiding this comment.
I added this, otherwise the test doesn't pass. Doesn't either on main btw.
And a few other very small tidbits here and there. Just tidying things with as little change to the logic as possible.
There was a problem hiding this comment.
Weird, how was this not caught before?
There was a problem hiding this comment.
Does it still not work now that you fixed the Opt parser btw?
There was a problem hiding this comment.
I've reverted the change. Most likely something to do with the unwanted code changes. No idea what happened there lol.
|
Not quite sure why I did this. But it's done. |
| @@ -83,6 +83,7 @@ pub fn main() { | |||
| }; | |||
|
|
|||
| // Create an observation channel using the signals map | |||
| #[allow(static_mut_refs)] // only a problem in nightly | |||
There was a problem hiding this comment.
We should not use static mut refs here.
See this discussion as reference
#2770 (comment)
|
This is cool! I'm a bit afraid we'll run into a bunch of new errors on odd platforms though |
|
|
|
The pendantic comment you can ignore |
|
Both are already in the repo. This makes them just a bit more visible. I need to do something else at some point. Separate PR please? |
|
Ah you already didn't :D |
|
As in: the maximum amount of enabled lint possible at all times; without being ridicoulous |
| @@ -24,14 +24,15 @@ use libafl_bolts::{current_nanos, nonzero, rands::StdRand, tuples::tuple_list, A | |||
|
|
|||
| /// Coverage map with explicit assignments due to the lack of instrumentation | |||
| static mut SIGNALS: [u8; 16] = [0; 16]; | |||
| #[allow(static_mut_refs)] // only a problem in nightly | |||
There was a problem hiding this comment.
In any case these allows have to go
There was a problem hiding this comment.
Or otherwise leave a biiiiiig TODO comment
There was a problem hiding this comment.
I've created an issue (#2786) and added a comment to the code with a link to it.
Who needs to graduate anyways… |
|
@devnexen if you ever find the time, it'd be interesting if this breaks something on obscure *Nixes |
|
Ok, will do ! |
|
|
Thank you @devnexen ! :) |
#[expect(…)]returns a lint warning (configured to be an error) if the lint warning is not present without the attribute.This also means I was able to remove a lot of
allowstatements.I suggest in the future,
expectis used wherever possible. The only exceptions I've stumbled across are when they are used in a macro and only appear for some macro calls (thinktrivial_numeric_castsof au64to a type passed to the struct, and it is called foru64among others) and if the warning only appears in certain cfgs.