Move LocalAnalyzer to rustc_mir::ssa_analyze by bjorn3 · Pull Request #52208 · rust-lang/rust

bjorn3 · 2018-07-10T11:37:59Z

This is useful for other codegen backends as well.

rust-highfive · 2018-07-10T11:38:03Z

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

rust-highfive · 2018-07-10T11:52:12Z

The job x86_64-gnu-llvm-5.0 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.


[00:03:47] travis_fold:start:tidy
travis_time:start:tidy
tidy check
[00:03:47] tidy error: /checkout/src/librustc_codegen_llvm/mir/analyze.rs: incorrect license
[00:03:47] tidy error: /checkout/src/librustc_codegen_llvm/mir/analyze.rs: missing trailing newline
[00:03:49] some tidy checks failed
[00:03:49] 
[00:03:49] 
[00:03:49] command did not execute successfully: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0-tools-bin/tidy" "/checkout/src" "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0/bin/cargo" "--no-vendor" "--quiet"
[00:03:49] 
[00:03:49] 
[00:03:49] failed to run: /checkout/obj/build/bootstrap/debug/bootstrap test src/tools/tidy
[00:03:49] Build completed unsuccessfully in 0:00:52
[00:03:49] Build completed unsuccessfully in 0:00:52
[00:03:49] Makefile:79: recipe for target 'tidy' failed
[00:03:49] make: *** [tidy] Error 1

The command "stamp sh -x -c "$RUN_SCRIPT"" exited with 2.
travis_time:start:002933f8
$ date && (curl -fs --head https://google.com | grep ^Date: | sed 's/Date: //g' || true)
---
travis_time:end:09b585b4:start=1531223009751336422,finish=1531223009761388773,duration=10052351
travis_fold:end:after_failure.3
travis_fold:start:after_failure.4
travis_time:start:0855aaf7
$ head -30 ./obj/build/x86_64-unknown-linux-gnu/native/asan/build/lib/asan/clang_rt.asan-dynamic-i386.vers || true
head: cannot open ‘./obj/build/x86_64-unknown-linux-gnu/native/asan/build/lib/asan/clang_rt.asan-dynamic-i386.vers’ for reading: No such file or directory
travis_fold:end:after_failure.4
travis_fold:start:after_failure.5
travis_time:start:01e68dcb
$ 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)

estebank · 2018-07-12T20:11:16Z

It's gonna take me a while to get to this. @arielb1 can you step in?

bors · 2018-07-13T15:53:05Z

☔ The latest upstream changes (presumably #51987) made this pull request unmergeable. Please resolve the merge conflicts.

eddyb · 2018-07-13T16:32:44Z

r? @eddyb (you can always r? me on codegen-related things unless @nagisa, @arielb1 or @alexcrichton want to take them)

eddyb · 2018-07-13T16:36:03Z

src/librustc_mir/ssa_analyze.rs

+use rustc::ty::layout::{TyLayout, LayoutError};
+
+pub trait LocalAnalyzerCallbacks<'tcx> {
+    fn ty_ssa_allowed(&self, ty: Ty<'tcx>) -> bool;


It would be more useful for this to take a TyLayout.

eddyb · 2018-07-13T16:36:36Z

src/librustc_mir/ssa_analyze.rs

+        debug!("local {} has type {:?}", index, ty);
+        if !analyzer.callbacks.ty_ssa_allowed(ty) {
+            analyzer.not_ssa(mir::Local::new(index));
+        }


This part could arguably stay in rustc_codegen_llvm.

The ty_ssa_allowed function is defined by the user of non_ssa_locals.

eddyb · 2018-07-13T16:37:03Z

src/librustc_mir/ssa_analyze.rs

+                }
+
+                if let mir::ProjectionElem::Field(..) = proj.elem {
+                    if self.callbacks.ty_ssa_allowed(base_ty.to_ty(self.tcx)) {


Can you check that you need this if?

eddyb · 2018-07-13T16:37:53Z

src/librustc_mir/ssa_analyze.rs

+        self.tcx.layout_of(ty::ParamEnv::reveal_all().and(ty))
+            .unwrap_or_else(|e| match e {
+                LayoutError::SizeOverflow(_) => self.tcx.sess.fatal(&e.to_string()),
+                _ => bug!("failed to get layout for `{}`: {}", ty, e)


This is not in rustc_codegen_llvm so ideally we would treat a failure to get the error just like some arbitrary layout that we know nothing about, instead of a fatal error.

eddyb · 2018-07-13T17:01:11Z

IMO this shouldn't live in rustc_mir unless it's somewhat more general (identifying variables that are SSA-like without any concerns for layout). If layout is involved I'd rather it be in the rustc_codegen_ssa crate proposed in #45274 (@irinagpopa will work on that, FWIW).

bjorn3 · 2018-07-14T09:02:13Z

Got it

eddyb · 2018-07-14T16:04:53Z

Note that layout may not need to be directly involved (i.e. this pass could just check for multiple writes, borrows, non-SSA reads, those sorts of things), and rustc_codegen_llvm would just mark each non-immediate/pair local as non-SSA to handle the layout side of things.
But I haven't checked for sure - you could try #52208 (comment) and see if anything breaks.

Move LocalAnalyzer to rustc_mir::ssa_analyze

af01df7

rust-highfive assigned estebank Jul 10, 2018

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jul 10, 2018

Fix tidy errors

96d1515

rust-highfive assigned eddyb and unassigned estebank Jul 13, 2018

eddyb reviewed Jul 13, 2018

View reviewed changes

bjorn3 closed this Jul 14, 2018

bjorn3 deleted the extract_ssa_analyzer branch July 10, 2019 18:23

Uh oh!

Conversation

bjorn3 commented Jul 10, 2018

Uh oh!

rust-highfive commented Jul 10, 2018

Uh oh!

rust-highfive commented Jul 10, 2018

Uh oh!

estebank commented Jul 12, 2018

Uh oh!

bors commented Jul 13, 2018

Uh oh!

eddyb commented Jul 13, 2018

Uh oh!

eddyb Jul 13, 2018

Choose a reason for hiding this comment

Uh oh!

eddyb Jul 13, 2018

Choose a reason for hiding this comment

Uh oh!

bjorn3 Jul 14, 2018

Choose a reason for hiding this comment

Uh oh!

eddyb Jul 13, 2018

Choose a reason for hiding this comment

Uh oh!

eddyb Jul 13, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

eddyb commented Jul 13, 2018

Uh oh!

bjorn3 commented Jul 14, 2018

Uh oh!

eddyb commented Jul 14, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

eddyb Jul 13, 2018 •

edited

Loading