link error: undefined reference to `<T as core::convert::From<T>>::from'

[tmandry]

From the 1.55 crater run #87749, while building the crate issue-macros-0.1.4:

[INFO] [stdout]   = note: /usr/bin/ld: /opt/rustwide/target/debug/deps/issue_macros-597542d719c97701.issue_macros.be0037e3-cgu.4.rcgu.o: in function `<T as core::convert::Into<U>>::into':
[INFO] [stdout]           /rustc/53fd98ca776cb875bc9e5514f56b52eb74f9e7a9/library/core/src/convert/mod.rs:540: undefined reference to `<T as core::convert::From<T>>::from'
[INFO] [stdout]           collect2: error: ld returned 1 exit status

Full log

This doesn't happen in 1.54. I'm not sure what's going on yet, next step is probably to reproduce locally.

Possibly related to #87002 (but that's for 1.54 and is fixed in 1.55)

[apiraino]

@tmandry I was curious to bisect this, but I can't find a code repository for the crate issue-macro. Any idea? thanks :)

[Friz64]

It's macros in https://github.com/KaiserKarel/issue-rs.

[apiraino]

ah thanks. I've bisected with cargo test (because cargo build does not trigger the crash) and the bisection seems to point to nightly-2021-08-11 but the tarball is not yet available (will update this comment later).

EDIT: bisection results seems to point to commit 4e90017

searched nightlies: from nightly-2021-08-01 to nightly-2021-08-16
regressed nightly: nightly-2021-08-13
searched commits: from ccffcaf to 0fa3190
regressed commit: 4e90017

bisected with cargo-bisect-rustc v0.6.0

Host triple: x86_64-unknown-linux-gnu
Reproduce with:

cargo bisect-rustc --start=2021-08-01 --script=./test.sh 

[tmiasko]

With -Clink-args=-Wl,--no-demangle and extra logging one can determine that undefined symbol is supposed to be provided by the hyper crate. The hyper crate is linked as dylib, at the same time in it was build only as crate-type = ["lib", "staticlib", "cdylib"]. The problem seems to be that cdylib is being misidentified by crate locator as dylib.

The issue appears to be pre-existing and merely exposed by changes in the standard library. Seems like a duplicate of #82972.

[apiraino]

Assigning priority as discussed in the Zulip thread of the Prioritization Working Group.

@rustbot label -I-prioritize +P-medium

[Mark-Simulacrum]

Closing as essentially a duplicate of #82972