[DO NOT MERGE] bootstrap with -Znext-solver=globally

[lcnr]

A revival of #124812.

Current status:

./x.py b --stage 2 passes 🎉

try builds succeed 🎉 🎉 🎉

commits

• support revealing defined opaque post borrowck #133501
• 9456bfe and b21b116 reimplement candidate preference based on rework winnowing to sensibly handle global where-bounds #132325, not yet a separate PR
• ce66d92 is a rebased version of add a normalizes-to fast path #125334, unsure whether I actually want to land this PR for now
• Deeply normalize when computing implied outlives bounds #133517

• Cache in DeepRejectCtxt #133524
• Structurally resolve before adjust_for_branches #133559
• Structurally resolve in probe_adt #133558

r? @ghost

[compiler-errors]

@bors try

[bors]

⌛ Trying commit f631ca8 with merge ca04381...

[compiler-errors]

I also pushed:

• Deeply normalize when computing implied outlives bounds #133517
• Structurally resolve before checking ! in HIR typeck #133518
• Check xform_ret_ty for WF in the new solver to improve method winnowing #133519
• Structurally resolve before applying projection in borrowck #133520
• Structurally resolve before matching on type of projection #133521
• Cache in DeepRejectCtxt #133524

I think this is sufficient to make the compiler build, or at least ./x.py build --stage=2 completes successfully on my local branch.

[bors]

💔 Test failed - checks-actions

[bors]

☔ The latest upstream changes (presumably #133527) made this pull request unmergeable. Please resolve the merge conflicts.

[lqd]

Next failure is on color-print-proc-macro-0.3.6 while building cargo.

pub fn repro() -> impl FnMut() {
    if true {
        || ()
    } else {
        || ()
    }
}

godbolt

[compiler-errors]

@bors try

[bors]

⌛ Trying commit 6210ed3 with merge ac5ccad...

[bors]

💔 Test failed - checks-actions

[compiler-errors]

@bors try

[bors]

⌛ Trying commit 8fa8f40 with merge 550ae0d...

[lqd]

I was reducing the new failure but you've already fixed it ^^

[bors]

💔 Test failed - checks-actions

[compiler-errors]

@bors try

[bors]

⌛ Trying commit 000f7bd with merge bccf3b7b458a76f118143edf3dfd56b763e76c39...

[bors]

☀️ Try build successful - checks-actions
Build commit: bccf3b7 (bccf3b7b458a76f118143edf3dfd56b763e76c39)

[compiler-errors]

@rust-timer build bccf3b7

[rust-timer]

Queued bccf3b7 with parent 66adeaf, future comparison URL.
There is currently 1 preceding artifact in the queue.
It will probably take at least ~1.6 hours until the benchmark run finishes.

[lcnr]

perf results https://perf.rust-lang.org/compare.html?start=66adeaf46b7a646daeed29518ef76235fcdb9726&end=bccf3b7b458a76f118143edf3dfd56b763e76c39

[lqd]

benchmarking bootstrap somehow didn't succeed on the perf collector 🤔 -- could be a timeout maybe.

[lqd]

results are actually incomplete, a benchmark fails in a way that is unexpected by the collector (measureme crashing that issue-46449's -Zself-profile data is invalid) and it's trying to re-run it non stop.

[rust-log-analyzer]

The job mingw-check-tidy failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

COPY host-x86_64/mingw-check/validate-toolstate.sh /scripts/
COPY host-x86_64/mingw-check/validate-error-codes.sh /scripts/

# NOTE: intentionally uses python2 for x.py so we can test it still works.
# validate-toolstate only runs in our CI, so it's ok for it to only support python3.
ENV SCRIPT TIDY_PRINT_DIFF=1 python2.7 ../x.py test \
           --stage 0 src/tools/tidy tidyselftest --extra-checks=py:lint,cpp:fmt
# This file is autogenerated by pip-compile with Python 3.10
# by the following command:
#
#    pip-compile --allow-unsafe --generate-hashes reuse-requirements.in
---
#13 2.862 Building wheels for collected packages: reuse
#13 2.863   Building wheel for reuse (pyproject.toml): started
#13 3.114   Building wheel for reuse (pyproject.toml): finished with status 'done'
#13 3.115   Created wheel for reuse: filename=reuse-4.0.3-cp310-cp310-manylinux_2_35_x86_64.whl size=132720 sha256=026f3bb0f1aa8090b861fd0a0939cb1a782396d84c8aab7875096557d637a0f6
#13 3.115   Stored in directory: /tmp/pip-ephem-wheel-cache-75g9g79s/wheels/3d/8d/0a/e0fc6aba4494b28a967ab5eaf951c121d9c677958714e34532
#13 3.118 Installing collected packages: boolean-py, binaryornot, tomlkit, reuse, python-debian, markupsafe, license-expression, jinja2, chardet, attrs
#13 3.513 Successfully installed attrs-23.2.0 binaryornot-0.4.4 boolean-py-4.0 chardet-5.2.0 jinja2-3.1.4 license-expression-30.3.0 markupsafe-2.1.5 python-debian-0.1.49 reuse-4.0.3 tomlkit-0.13.0
#13 3.513 WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
#13 4.064 Collecting virtualenv
#13 4.064 Collecting virtualenv
#13 4.139   Downloading virtualenv-20.28.0-py3-none-any.whl (4.3 MB)
#13 4.237      ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.3/4.3 MB 44.8 MB/s eta 0:00:00
#13 4.296 Collecting platformdirs<5,>=3.9.1
#13 4.308   Downloading platformdirs-4.3.6-py3-none-any.whl (18 kB)
#13 4.335 Collecting distlib<1,>=0.3.7
#13 4.355      ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 469.0/469.0 KB 82.3 MB/s eta 0:00:00
#13 4.395 Collecting filelock<4,>=3.12.2
#13 4.407   Downloading filelock-3.16.1-py3-none-any.whl (16 kB)
#13 4.407   Downloading filelock-3.16.1-py3-none-any.whl (16 kB)
#13 4.488 Installing collected packages: distlib, platformdirs, filelock, virtualenv
#13 4.671 Successfully installed distlib-0.3.9 filelock-3.16.1 platformdirs-4.3.6 virtualenv-20.28.0
#13 DONE 4.8s

#14 [7/8] COPY host-x86_64/mingw-check/validate-toolstate.sh /scripts/
#14 DONE 0.0s
---
DirectMap4k:      323520 kB
DirectMap2M:     7016448 kB
DirectMap1G:    11534336 kB
##[endgroup]
Executing TIDY_PRINT_DIFF=1 python2.7 ../x.py test            --stage 0 src/tools/tidy tidyselftest --extra-checks=py:lint,cpp:fmt
+ TIDY_PRINT_DIFF=1 python2.7 ../x.py test --stage 0 src/tools/tidy tidyselftest --extra-checks=py:lint,cpp:fmt
    Finished `dev` profile [unoptimized] target(s) in 0.04s
##[endgroup]
WARN: currently no CI rustc builds have rustc debug assertions enabled. Please either set `rust.debug-assertions` to `false` if you want to use download CI rustc or set `rust.download-rustc` to `false`.
downloading https://static.rust-lang.org/dist/2024-10-16/rustfmt-nightly-x86_64-unknown-linux-gnu.tar.xz
---
fmt check
fmt: checked 5685 files
tidy check
tidy: Skipping binary file check, read-only filesystem
##[error]tidy error: /checkout/compiler/rustc_next_trait_solver/src/solve/assembly/mod.rs:728: TODO is used for tasks that should be done before merging a PR; If you want to leave a message in the codebase use FIXME
removing old virtual environment
creating virtual environment at '/checkout/obj/build/venv' using 'python3.10'
Requirement already satisfied: pip in ./build/venv/lib/python3.10/site-packages (24.3.1)
All checks passed!
checking C++ file formatting
some tidy checks failed
some tidy checks failed
Command has failed. Rerun with -v to see more details.
  local time: Thu Nov 28 09:45:24 UTC 2024
  network time: Thu, 28 Nov 2024 09:45:24 GMT
##[error]Process completed with exit code 1.
Post job cleanup.

[Kobzol]

I had to remove the commit from the perf.RLO database, as it was causing issues in the collector, it was crashing.

[lcnr]

results are actually incomplete, a benchmark fails in a way that is unexpected by the collector (measureme crashing that issue-46449's -Zself-profile data is invalid) and it's trying to re-run it non stop.

How can we figure out why that happened so that we can fix it?

[lqd]

I wanted to make a proper repro for y'all but I'm still working on fixing decodeme and the collector so this doesn't brick it in the future anymore.

If it can't wait, and you want to reproduce in rustc-perf: ./target/debug/collector bench_local +bccf3b7b458a76f118143edf3dfd56b763e76c39 --include issue-46449 --self-profile --scenarios IncrPatched --profiles Check will generate the invalid .mm_profdata that makes analyzeme crash.

As-is, the file will be hidden in a tmp folder, so the setup the above command does basically goes like this: cargo check on the benchmark with self-profiling turned on, then applying one (or all) of the patches in the same folder and cargo checking again (always with self-profiling) to reproduce the issue (hopefully).

[lqd]

I'll see if I can get an actual MCVE tomorrow, but in the meantime, it seems this incremental ICE is the root cause.

From the benchmark:

[package]
name = "issue-46449-byte-filling-slowing-down-sroa"
version = "0.1.0"

[dependencies]
futures = "=0.1.17"

#![allow(unexpected_cfgs)]

extern crate futures;

use futures::{Future, Poll};

const BUFFER_SIZE: usize = 1;

#[cfg(before)]
pub struct Error(::std::io::Error);

#[cfg(after)]
pub struct Error();

struct Dummy<T>(T);
impl<T> Future for Dummy<T> {
    type Item = T;
    type Error = Error;
    fn poll(&mut self) -> Poll<Self::Item, Self::Error> {
        loop {}
    }
}

pub fn run() -> Box<dyn Future<Item = (), Error = Error>> {
    let c2s = Dummy([0u8; BUFFER_SIZE]).then(move |_| Ok(0));
    let s2c = Dummy(()).then(move |_| Ok(0));
    let fut = c2s.select(s2c)
        .and_then(move |_| Ok(()))
        .map_err(|(err, _)| err);
    Box::new(fut)
}

> cargo +bccf3b7b458a76f118143edf3dfd56b763e76c39 rustc -q -- --cfg before && cargo +bccf3b7b458a76f118143edf3dfd56b763e76c39 rustc -q -- --cfg after

thread 'rustc' panicked at compiler/rustc_middle/src/query/on_disk_cache.rs:729:21:
Failed to convert DefPathHash DefPathHash(Fingerprint(10053302304625752304, 298815228590218976))
...