Don't emit `Unevaluated` from `const_eval` #56723

oli-obk · 2018-12-11T19:57:08Z

rust-highfive · 2018-12-11T19:57:11Z

(rust_highfive has picked a reviewer for you, use r? to override)

matthewjasper · 2018-12-11T20:45:24Z

rust-highfive · 2018-12-11T21:25:16Z

The job mingw-check of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

travis_time:end:04b9a33c:start=1544558344989913189,finish=1544558420352388686,duration=75362475497
$ git checkout -qf FETCH_HEAD
travis_fold:end:git.checkout

Encrypted environment variables have been removed for security reasons.
See https://docs.travis-ci.com/user/pull-requests/#Pull-Requests-and-Security-Restrictions
$ export SCCACHE_BUCKET=rust-lang-ci-sccache2
$ export SCCACHE_REGION=us-west-1
Setting environment variables from .travis.yml
$ export IMAGE=mingw-check
---
[00:07:07] configure: build.locked-deps    := True
[00:07:07] configure: llvm.ccache          := sccache
[00:07:07] configure: build.cargo-native-static := True
[00:07:07] configure: dist.missing-tools   := True
[00:07:07] configure: build.configure-args := ['--enable-sccache', '--disable-manage-submodu ...
[00:07:07] configure: writing `config.toml` in current directory
[00:07:07] configure: 
[00:07:07] configure: run `python /checkout/x.py --help`
[00:07:07] configure: 
---
[00:09:44]     Checking syntax_ext v0.0.0 (/checkout/src/libsyntax_ext)
[00:11:09] error[E0080]: could not evaluate static initializer
[00:11:09]     --> src/librustc/ty/sty.rs:1989:1
[00:11:09]      |
[00:11:09] 1989 | static_assert!(MEM_SIZE_OF_LAZY_CONST: ::std::mem::size_of::<LazyConst<'_>>() == 24);
[00:11:09]      | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ index out of bounds: the len is 1 but the index is 1
[00:11:09]      = note: this error originates in a macro outside of the current crate (in Nightly builds, run with -Z external-macro-backtrace for more info)
[00:11:09] 
[00:11:10] error: aborting due to previous error
[00:11:10] 
[00:11:10] 
[00:11:10] For more information about this error, try `rustc --explain E0080`.
[00:11:10] error: Could not compile `rustc`.
[00:11:10] 
[00:11:10] To learn more, run the command again with --verbose.
[00:11:10] command did not execute successfully: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo" "check" "--target" "i686-pc-windows-gnu" "-j" "4" "--release" "--locked" "--color" "always" "--features" "" "--manifest-path" "/checkout/src/rustc/Cargo.toml" "--message-format" "json"
[00:11:10] failed to run: /checkout/obj/build/bootstrap/debug/bootstrap check --target=i686-pc-windows-gnu --host=i686-pc-windows-gnu
[00:11:10] Build completed unsuccessfully in 0:02:54
travis_time:end:0e63e16a:start=1544558428984192020,finish=1544559099453644431,duration=670469452411
The command "stamp sh -x -c "$RUN_SCRIPT"" exited with 1.
---
travis_time:end:02ca9624:start=1544559099902914048,finish=1544559099909545488,duration=6631440
travis_fold:end:after_failure.3
travis_fold:start:after_failure.4
travis_time:start:0ba53794
$ ln -s . checkout && for CORE in obj/cores/core.*; do EXE=$(echo $CORE | sed 's|obj/cores/core\.[0-9]*\.!checkout!\(.*\)|\1|;y|!|/|'); if [ -f "$EXE" ]; then printf travis_fold":start:crashlog\n\033[31;1m%s\033[0m\n" "$CORE"; gdb --batch -q -c "$CORE" "$EXE" -iex 'set auto-load off' -iex 'dir src/' -iex 'set sysroot .' -ex bt -ex q; echo travis_fold":"end:crashlog; fi; done || true
travis_fold:end:after_failure.4
travis_fold:start:after_failure.5
travis_time:start:1441e3ec
travis_time:start:1441e3ec
$ cat ./obj/build/x86_64-unknown-linux-gnu/native/asan/build/lib/asan/clang_rt.asan-dynamic-i386.vers || true
cat: ./obj/build/x86_64-unknown-linux-gnu/native/asan/build/lib/asan/clang_rt.asan-dynamic-i386.vers: No such file or directory
travis_fold:end:after_failure.5
travis_fold:start:after_failure.6
travis_time:start:07329432
$ dmesg | grep -i kill

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

RalfJung · 2018-12-12T16:07:49Z

src/librustc/traits/error_reporting.rs

+                            "_Self".to_owned(),
+                            Some(format!("[{}; _]", self.tcx.type_of(def.did).to_string())),
+                        ));
+                    }


So we don't print anything when it is not evaluated? That is different from the previous code.

I have no idea where this code is called, just pointing out a functional change. ;)

oops indeed. This is just for producing more diagnostics, but still. I found a nice way to fix it without more duplication.

src/librustc/ty/relate.rs

src/librustc/ty/wf.rs

src/librustc_codegen_llvm/debuginfo/type_names.rs

RalfJung · 2018-12-12T16:15:36Z

Were Const interned before? Does this mean we now intern both Const and LazyConst?

src/librustc/ty/sty.rs

src/librustc_mir/const_eval.rs

oli-obk · 2018-12-12T16:21:55Z

Were Const interned before? Does this mean we now intern both Const and LazyConst?

Yes they were interned before. Yes we now intern twice. I'll check if we can get away with not interning Const anymore. It's just a size optimization of various enums, but we might not have any Const left, so...

RalfJung · 2018-12-12T16:28:32Z

src/librustc_mir/borrow_check/nll/type_check/mod.rs

@@ -376,6 +376,11 @@ impl<'a, 'b, 'gcx, 'tcx> TypeVerifier<'a, 'b, 'gcx, 'tcx> {
            constant, location
        );

+        let literal = match constant.literal {
+            ty::LazyConst::Evaluated(lit) => lit,
+            ty::LazyConst::Unevaluated(..) => return,


Is it really correct to return early here in case of Unevaluated?

This is #46702

I don't believe this PR makes the situation any more problematic than it was.

Okay.

Cc @eddyb would be good if you could take a quick look at some point, I have no idea what this is doing.^^

RalfJung · 2018-12-12T16:32:06Z

we might not have any Const left

If we don't, that makes me wonder about whether this refactoring is worth it. :/

oli-obk · 2018-12-13T09:23:01Z

If we don't, that makes me wonder about whether this refactoring is worth it. :/

Sorry, I misspoke: We might not have any Const left in the HIR or the MIR, we'd still be using it extensively everywhere else (at least I think so, I wanted to do larger refactorings like that in a separate PR, but I'll do them here now)

oli-obk · 2018-12-14T18:23:31Z

This should theoretically not have a negative perf impact, but I've guessed wrong with this kind of change before.

let's see what perf says

@bors try

bors · 2018-12-14T18:23:44Z

⌛ Trying commit b01765fdfdbec1b70d7d41dccf7db02467721eeb with merge 6ab2e74a91c27b914174a1c9db557e41235bce60...

oli-obk · 2018-12-14T18:24:07Z

@rust-timer build 6ab2e74a91c27b914174a1c9db557e41235bce60

rust-timer · 2018-12-14T18:24:08Z

Success: Queued 6ab2e74a91c27b914174a1c9db557e41235bce60 with parent 7d03617, comparison URL.

rust-timer · 2018-12-14T22:14:00Z

Finished benchmarking try commit 6ab2e74a91c27b914174a1c9db557e41235bce60

bors · 2018-12-14T22:23:44Z

💥 Test timed out

nikomatsakis · 2019-01-02T18:18:05Z

r=me modulo rebase

oli-obk · 2019-01-03T17:59:20Z

@bors r=nikomatsakis

bors · 2019-01-03T17:59:21Z

📌 Commit 03b8928 has been approved by nikomatsakis

@eddyb

Don't emit `Unevaluated` from `const_eval` cc @eddyb @RalfJung

bors · 2019-01-04T17:01:31Z

⌛ Testing commit 03b8928 with merge d6d32ac...

@eddyb

Don't emit `Unevaluated` from `const_eval` cc @eddyb @RalfJung

bors · 2019-01-04T19:39:18Z

☀️ Test successful - status-appveyor, status-travis
Approved by: nikomatsakis
Pushing d6d32ac to master...

bjorn3 · 2019-01-05T12:34:56Z

src/librustc/ty/context.rs

+        impl<'a, 'tcx> Lift<'tcx> for &'a List<$ty> {
+            type Lifted = &'tcx List<$lifted>;
+            fn lift_to_tcx<'b, 'gcx>(&self, tcx: TyCtxt<'b, 'gcx, 'tcx>) -> Option<Self::Lifted> {
+                        if self.is_empty() {


Wrong indentation.

eddyb · 2020-02-07T22:43:52Z

src/librustc/ty/wf.rs

-        self.require_sized(constant.ty, traits::ConstSized);
-        if let ConstValue::Unevaluated(def_id, substs) = constant.val {
+    fn compute_array_len(&mut self, constant: ty::LazyConst<'tcx>) {
+        if let ty::LazyConst::Unevaluated(def_id, substs) = constant {


Why was this change made? If I called something const it's because it's meant to be used with const generics in the future.
This needs to be renamed back, and used for all ty::Consts in all Substs.
The lack of WF const handling probably allows a lot of unintended const expressions.
cc @varkor @yodaldevoid

rust-highfive assigned matthewjasper Dec 11, 2018

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Dec 11, 2018

rust-highfive assigned RalfJung and unassigned matthewjasper Dec 11, 2018