Invalidate all dereferences when encountering non-local assignments #132527

DianQK · 2024-11-02T15:11:59Z

This PR removes the computation value by traversing SSA locals through for_each_assignment_mut.

Because the for_each_assignment_mut traversal skips statements which have side effects, such as dereference assignments, the computation may be unsound. Instead of for_each_assignment_mut, we compute values by traversing in reverse postorder.

Because we compute and use the symbolic representation of values on the fly, I invalidate all old values when encountering a dereference assignment. The current approach does not prevent the optimization of a clone to a copy.

In the future, we may add an alias model, or dominance information for dereference assignments, or SSA form to help GVN.

r? cjgillot

cc @jieyouxu #132356
cc @RalfJung #133474

DianQK · 2024-11-02T15:13:37Z

@bors try @rust-timer queue

bors · 2024-11-02T15:14:47Z

⌛ Trying commit 64d0f48 with merge 0011660...

[WIP] Invalidate all dereferences when encountering non-local assignments Fixes rust-lang#132353. r? ghost

bors · 2024-11-02T17:07:59Z

☀️ Try build successful - checks-actions
Build commit: 0011660 (0011660807de999202661fd942d6b7f77fdc9c8e)

rust-timer · 2024-11-02T19:18:14Z

Finished benchmarking commit (0011660): comparison URL.

Overall result: ❌✅ regressions and improvements - please read the text below

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

This is the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.3%, 0.7%]	3
Regressions ❌ (secondary)	0.5%	[0.2%, 0.7%]	10
Improvements ✅ (primary)	-1.2%	[-6.4%, -0.2%]	14
Improvements ✅ (secondary)	-0.5%	[-0.6%, -0.5%]	10
All ❌✅ (primary)	-0.9%	[-6.4%, 0.7%]	17

Max RSS (memory usage)

Results (primary -0.1%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.3%	[2.1%, 5.8%]	4
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-6.9%	[-10.5%, -3.3%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.1%	[-10.5%, 5.8%]	6

Cycles

Results (primary -1.6%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.8%	[0.8%, 0.8%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.1%	[-5.8%, -0.8%]	5
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.6%	[-5.8%, 0.8%]	6

Binary size

Results (primary -0.1%, secondary -0.3%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	0.1%	[0.0%, 0.4%]	33
Regressions ❌ (secondary)	0.0%	[0.0%, 0.1%]	14
Improvements ✅ (primary)	-0.3%	[-2.4%, -0.0%]	23
Improvements ✅ (secondary)	-0.7%	[-1.2%, -0.0%]	10
All ❌✅ (primary)	-0.1%	[-2.4%, 0.4%]	56

Bootstrap: 780.173s -> 782.091s (0.25%)
Artifact size: 335.37 MiB -> 335.25 MiB (-0.03%)

DianQK · 2024-11-16T01:15:10Z

I will update this after #132461.

DianQK · 2024-12-04T10:32:33Z

@bors try @rust-timer queue

bors · 2024-12-04T10:34:44Z

🔒 Merge conflict

This pull request and the master branch diverged in a way that cannot be automatically merged. Please rebase on top of the latest master branch, and let the reviewer approve again.

How do I rebase?

Assuming self is your fork and upstream is this repository, you can resolve the conflict following these steps:

git checkout gvn-stmt-iter (switch to your branch)
git fetch upstream master (retrieve the latest master)
git rebase upstream/master -p (rebase on top of it)
Follow the on-screen instruction to resolve conflicts (check git status if you got lost).
git push self gvn-stmt-iter --force-with-lease (update this PR)

You may also read Git Rebasing to Resolve Conflicts by Drew Blessing for a short tutorial.

Please avoid the "Resolve conflicts" button on GitHub. It uses git merge instead of git rebase which makes the PR commit history more difficult to read.

Sometimes step 4 will complete without asking for resolution. This is usually due to difference between how Cargo.lock conflict is handled during merge and rebase. This is normal, and you should still perform step 5 to update this PR.

Error message

Auto-merging tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-unwind.mir
CONFLICT (content): Merge conflict in tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-unwind.mir
Auto-merging tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-abort.mir
CONFLICT (content): Merge conflict in tests/mir-opt/pre-codegen/vec_deref.vec_deref_to_slice.PreCodegen.after.panic-abort.mir
Automatic merge failed; fix conflicts and then commit the result.

bors · 2024-12-04T10:43:25Z

☔ The latest upstream changes (presumably #133841) made this pull request unmergeable. Please resolve the merge conflicts.

DianQK · 2024-12-04T10:54:11Z

@bors try @rust-timer queue

[WIP] Invalidate all dereferences when encountering non-local assignments Fixes rust-lang#132353. r? ghost

bors · 2024-12-21T03:18:15Z

⌛ Trying commit 199b7f9 with merge 0754bc9...

bors · 2024-12-21T04:57:53Z

💔 Test failed - checks-actions

bors · 2024-12-22T22:00:36Z

☔ The latest upstream changes (presumably #134330) made this pull request unmergeable. Please resolve the merge conflicts.

DianQK · 2024-12-22T22:11:48Z

The job x86_64-mingw-1 failed! Check out the build log: (web) (plain)
Click to see the possible cause of the failure (guessed by this bot)

Should fix by #134608.

DianQK · 2024-12-22T22:30:07Z

☔ The latest upstream changes (presumably #134330) made this pull request unmergeable. Please resolve the merge conflicts.

I rebased.

cjgillot · 2024-12-23T00:59:32Z

compiler/rustc_mir_transform/src/gvn.rs

+            }
+            // Function calls maybe invalidate nested deref, and non-local assignments maybe invalidate deref.
+            // Currently, no distinction is made between these two cases.
+            self.invalidate_derefs();


I fear we are missing many cases of indirect writes. For example, inline asm.

Instead of addressing to which terminator may write something, I adopted a more conservative approach, which terminator must not write anything.

DianQK · 2024-12-23T01:11:37Z

@bors r-

DianQK · 2025-01-07T12:01:24Z

@rustbot ready

bors · 2025-01-17T00:49:18Z

☔ The latest upstream changes (presumably #135592) made this pull request unmergeable. Please resolve the merge conflicts.

…ments `for_each_assignment_mut` can skip assignment statements with side effects, which can result in some assignment statements retrieving outdated value. For example, it may skip a dereference assignment statement.

This reverts commit 906f66f.

This reverts commit 917dd82.

DianQK · 2025-01-24T15:13:05Z

I have rebased.

@oli-obk Since @cjgillot may be busy, would you like to review it again?
btw, I think you might have forgotten to update your GitHub profile status? :p

bors · 2025-01-26T12:40:44Z

☔ The latest upstream changes (presumably #135753) made this pull request unmergeable. Please resolve the merge conflicts.

oli-obk · 2025-01-28T12:03:46Z

would you like to review it again?

I think I'll leave it to the expert who keeps finding things I missed ;)

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Nov 2, 2024

DianQK changed the title ~~[WIP] Invalidate all dereferences when encountering non-local assignments.~~ [WIP] Invalidate all dereferences when encountering non-local assignments Nov 2, 2024