rust

This is an incremental compilation cache corruption ICE, not a bug in your code. Fix it by clearing the incremental cache:

1. Run cargo clean -p <crate_name> (e.g. cargo clean -p jmap) to clear only the affected crate, or run full cargo clean to wipe all build artifacts.
2. Recompile with cargo build.

If the error keeps recurring, disable incremental compilation entirely by setting the environment variable CARGO_INCREMENTAL=0 or adding [build] incremental = false to .cargo/config.toml.

Since this is a compiler bug, you should also report it at https://github.com/rust-lang/rust/issues with a minimal reproduction (ideally the code before plus the patch that triggers the ICE on rebuild), including your rustc --version output and the fingerprint hash from the error.

This is a known rustc/LLVM miscompilation bug (rust-lang/rust #164431, affecting versions from ~1.87 through at least nightly 1.101). The optimizer lowers checked_div(...).unwrap_unchecked() into a bare div instruction that can receive a zero divisor after cross-function inlining, producing SIGFPE (exit code 136) at -C opt-level=1 and above, while -C opt-level=0 and Miri run clean. Workarounds until the bug is fixed: 1) Replace checked_div(...).unwrap_unchecked() with wrapping_div(...) — the crash disappears. 2) Add #[inline(never)] to the function being inlined into the caller — this blocks the cross-function inlining that triggers the bug. 3) Compile with -C opt-level=0 as a last resort. Note the bug is also address-layout sensitive: it reproduces with the default PIE build but not reliably with a non-PIE build.

This E0425 error occurs because outdated libc versions (e.g., 0.2.43) lack riscv64 target definitions for core C types like c_char. Fix by upgrading libc to 0.2.190 or later. If libc is pinned by a dependency's Cargo.lock (such as the xsv crate used in cargotest), run cargo update -p libc inside the affected crate's directory, or manually bump the libc version in Cargo.lock/Cargo.toml, then rebuild. Do not follow the compiler's suggestion to import std::ffi::c_char — the missing definitions are a version issue, not an import issue.

This internal compiler error affects rustc 1.98.0-nightly (485ec3fbc 2026-06-10) when compiling recursive inherent associated types (e.g., type Assoc = &'a Foo<'a>::Assoc;) with flags -Znext-solver=globally -Zassumptions-on-binders. The fix: update to nightly 1.99.0 (ba28ff76f 2026-08-13) or newer via rustup update nightly, where the issue no longer reproduces. Note that inherent_associated_types is an incomplete, unstable feature known to cause compiler crashes — avoid it on stable/beta. Also, the triggering code is itself invalid (self-referential associated type produces E0271 type mismatch), so rewriting the associated type to not reference itself resolves it on affected versions.

This is a known type inference regression in Rust 1.99 beta (bisected to PR #158447, duplicate of issue #161913), affecting code where a closure passed to a generic constructor (e.g., Foo::new(&1u8, |input| *input)) could previously infer types. Fix: upgrade to Rust 1.100.0-beta.7 or later, where the regression no longer reproduces — run rustup update or switch channels with rustup default stable. Workaround if you must stay on 1.99 beta: add explicit type annotations to the closure parameter or generic arguments, e.g., |input: &u8| *input, or specify the full type on the binding such as let x: Foo<&u8, u8> = ....