rustdoc query cycle regression

[BoxyUwU]

          I believe this caused a regression on nightly. See https://github.com/aya-rs/aya/actions/runs/15223453504/job/42822502727.

 Documenting aya-log-ebpf v0.1.1 (/home/runner/work/aya/aya/ebpf/aya-log-ebpf)
error[E0391]: cycle detected when computing revealed normalized predicates of `aya_ebpf::EbpfContext::command::{constant#0}`
  --> /home/runner/work/aya/aya/ebpf/aya-ebpf/src/lib.rs:42:38
   |
42 |     fn command(&self) -> Result<[u8; TASK_COMM_LEN], c_long> {
   |                                      ^^^^^^^^^^^^^
   |
note: ...which requires computing normalized predicates of `aya_ebpf::EbpfContext::command::{constant#0}`...
  --> /home/runner/work/aya/aya/ebpf/aya-ebpf/src/lib.rs:42:38
   |
42 |     fn command(&self) -> Result<[u8; TASK_COMM_LEN], c_long> {
   |                                      ^^^^^^^^^^^^^
   = note: ...which again requires computing revealed normalized predicates of `aya_ebpf::EbpfContext::command::{constant#0}`, completing the cycle
   = note: cycle used when normalizing `aya_ebpf::::EbpfContext::command::{constant#0}`
   = note: see https://rustc-dev-guide.rust-lang.org/overview.html#queries and https://rustc-dev-guide.rust-lang.org/query.html for more information

For more information about this error, try `rustc --explain E0391`.
error: could not document `aya-log-ebpf`

Oddly, this seems specific to rustdoc.

Originally posted by @tamird in #139635 (comment)

[BoxyUwU]

I'm not sure how #139635 would have caused this, it doesn't touch anything to do with what queries get called so shouldn't be introducing new query cycles. @tamird what makes you think that's the PR that regressed your crate?

[tamird]

The tests updated in that PR contain the same diagnostic text e.g. b1774b8#diff-d26852ce626e7632e48fd9e08630bf948f008ef8f7de7b00336c6b9cb0eda19f

[tamird]

Ah, I posted this on the wrong PR, it was probably #140553.

[BoxyUwU]

Ah I see what you mean. I'll take a look and see what happened there and verify if it was that PR that caused it.

[BoxyUwU]

@tamird I can't seem to reproduce the failure locally, though I did notice that the crate sometimes uses generic_const_exprs. Is the CI failure happening in a place where the unstable cfg is set and so generic_const_exprs is enabled?

[BoxyUwU]

Ah running cargo xtask public-api --bless reproduces it and stops when removing the enabling of generic_const_exprs. So I think this is a gce only regression, and also seems reasonable to expect that #140553 was the cause