-
Notifications
You must be signed in to change notification settings - Fork 13.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ICE: Found unstable fingerprints for mir_built #107665
Comments
I, too, have hit this bug with a similar import error, code change, recompile cycle. Good news: by coincidence, I made a ZFS snapshot of the filesystem minutes after it happened. I can restore that and repro at will. Bad news: it seems like if I change anything it invalidates the cache, recompiles properly, and goes away.
I was planning to trim files in Any tips for making this into a docker container? What environment does it check? EDIT: I'm on stable, 1.66.1 |
I have figured out how to make a docker container which reproduces the problem. I did it by nailing down a couple env vars, and copying way too much of my home dir into it. I will prune it down procedurally to something much smaller. Hopefully I'll have a perfect reproduction within the next week or two that is not immensely bloated. |
I have successfully created a Docker container which reproduces the problem, based on some code of mine that is not ready for the public. But I'd be willing to send it to a compiler dev for examination. After significant pruning, the container is ~0.5GB compressed (~1.6 GB uncompressed). But keep in mind it is completely self-contained, so that includes the entire toolchain which reproduces the problem (the then-newest stable). I'll offer a file sharking link privately via e-mail to any rust dev who would take this issue up. Pinging as first ideas:
|
@jhwgh1968 Sure, try this handle at gmail? |
E-mail sent, @workingjubilee! |
Code
A little backstory: this issue ended up fixing itself while I was attempting minimization, so I'm not sure if this bug report is still useful, as I don't know how to replicate the error anymore. I saved a commit when I found the error, but even reverting to it doesn't cause this bug to pop up again. The command used to test was either
cargo test
orcargo test --bin server my_fn
.At the same time, this was an ICE, so I believe it may still be worth reporting.
What ended up fixing things was turning
use super::*
touse super::{only_the_imports_I_need}
so I believe there was something thatsuper::*
was bringing in that was breaking things. Hereafter, when I say "accidentally fixed the issue" I'm referring to this fixing of imports.In practice, this is what I was testing (with identifiying information obfuscated):
Where org was a module in my project. While minimizing, I brought all the non-external code into that one function my_fn, and still received the error. I'm not including it all since I also found that only this test also caused the error:
I'm including the former still it gives some context but because I believe this was caused by something in
super::*
I would ideally submit literally the entire program ---- but since "literally the entire program" is no longer causing the bug to happen for me, I'm not sure what to do. (Literally the entire program in the sense that if I runcargo test
now, with the git-reverted version of my program, I don't get the same error again.)I was still able to test the three main channels before accidentally fixing this though:
Meta
rustc --version --verbose
:Since this section also asked about the nightly and beta versions, I found the following behavior for both of them:
cargo test
passes both tests I have, butcargo test -- --my_fn
returns:So when I then use
cargo test --bin server --my_fn
the test passes fine.Nightly version:
Beta version:
For the record,
cargo test --bin server my_fn
gives what looks like the same error (ascargo test
) on stable. After I accidentally fixed the issue, all three versions are not showing the error. (This was before I accidentally fixed the error)Error output
Backtrace Full
This was the first time I'm submitting a bug report for anything, so any feedback is appreciated. Thank you!
And I apologize once more that I'm unable to find a code snippet which replicates the error. Unreplicable errors are the worst.
The text was updated successfully, but these errors were encountered: