Make imported functions inexact #7993

kripken · 2025-10-24T20:48:49Z

Defined functions remain exact, but imported ones are inexact.

This is a step along the recent Custom Descriptors spec changes.

New RefFunc::finalize and Literal::makeFunc variants get the module, and look up
the type there.
New Builder::makeRefFunc variant gets a Type and applies it. The HeapType
variant does a lookup on the module (so the Type one is more efficient/applicable
if the IR is not fully built yet).
ReFinalize now updates RefFunc types (following the pattern of a few other places).
C and JS APIs now assume RefFuncs are created after imported functions (so we can
look up the type of the import; see changelog, this seems the least-annoying way to
update here, avoiding new APIs, and less breakage for users - hopefully none, all our
tests here pass as is).
wasm-split adds a cast when a function becomes an inexact import.
Fix GUFA to handle inexact function literals.
Update types in passes and fuzzer as needed.

Update the Literal constructors for funcrefs to take Type instead of HeapType to allow them to be given inexact function references types when the referenced function is an import. Use the new capability to give references to imported functions inexact types in GUFA. Add a test where this change fixes a misoptimization as well as tests where this change simply changes the nature of the misoptimization. Future PRs will fix these tests.

…nc.type

kripken · 2025-11-04T18:30:49Z

Fuzzer noticed that we didn't type imported ref.funcs correctly. Last two (non-merge) commits fix and test that.

tlively · 2025-11-04T21:42:14Z

src/ir/module-splitting.cpp

+        if (wasm.features.hasCustomDescriptors()) {
+          // Add a cast, as the parent may depend on the exactness to validate.
+          replaceCurrent(Builder(wasm).makeRefCast(curr, curr->type));
+        }


We don't have issues about needing this cast even when custom descriptors are disabled?

None that our tests or the fuzzer can find. What concerns you, the exactness? I think you made the binary writer strip that out as needed for all instructions?

For example in the test I commented on here: #7993 (comment). Shouldn't it be an IR validation error that we have an inexact reference being stored in an exact local? The binary writer would strip exactness and make the output valid, but I would have expected our validator to complain about the IR.

I don't follow you. First, the linked comment was resolved. Second, how is an inexact reference stored in an exact local? The cast is ensuring things don't change (keeping the exact type from before, which was presumed valid).

tlively · 2025-11-04T21:44:43Z

src/ir/possible-contents.cpp

+      // This is imported, so it might be anything of the proper type.
+      addRoot(curr);


Don't we still want to track that this is a reference to the imported function? It would just have an inexact type.

But without knowing the identity, I think we can misoptimize? Imagine we have equality for a second, then ref.func a == ref.func a is definitely 1, but ref.func a == ref.func b is not necessarily 0 (can have duplicate imports).

But functions references cannot be compared for equality, so there is nothing to misoptimize, unless I'm missing something.

I did say "imagine" 😆

But, while we don't have ref.eq on functions in wasm userspace, we do have optimizations that compare functions in other ways. E.g. folding an if with ref.eq arms, or GUFA inferences. I admit I don't see an actual bug atm in our optimizer, but a future one is conceivable.

src/passes/ExtractFunction.cpp

src/passes/InstrumentBranchHints.cpp

test/lit/exec/imported-func.wast

tlively · 2025-11-04T23:57:45Z

test/lit/exec/imported-func.wast.second

@@ -0,0 +1,36 @@
+;; Import a function of type $C as type $A, cast to $A, $B, $C. All those casts
+;; should succeed. Exact casts, however TODO


Is this still TODO?

Oops, just some old text. These casts become exact. Cleaned up now.

Let's add some exact casts to the test to show that they work as expected. I don't think the test currently demonstrates that.

Sure, added.

test/lit/passes/gsi-debug.wast

test/lit/passes/gufa.wast

tlively · 2025-11-05T00:00:41Z

test/lit/passes/gufa.wast

+  ;; CHECK-NEXT:  (ref.test (ref (exact $func))
+  ;; CHECK-NEXT:   (ref.func $f)
+  ;; CHECK-NEXT:  )


The comment suggests we should have been able to optimize this.

Good point, I think we need to look at finality in GUFA somehow. I added a TODO.

I suspect this will be fixed by using a non-exact Literal for references to function imports.

test/lit/wasm-split/exact.wast

Co-authored-by: Thomas Lively <[email protected]>

tlively and others added 30 commits October 8, 2025 13:44

marge

7cbdfea

fix

82ef0e0

fix

d857e7d

fix

d1d2ed8

fix

e26f676

work

7fda8cf

work

5ce973b

work

4b167eb

work

7927749

work

160123e

work

b8bb97a

work

89103b9

work

24dea81

work

d4aabd2

work

a4fe585

work

76eac4c

work

020f119

work

21588c7

work

fdd7253

work

7a58395

work

4616380

work

46826b0

work

73024ea

work

4de7694

work

d03376d

work

82061c8

work

1859cf0

work

1791066

format

7278c6a

kripken added 6 commits October 24, 2025 11:17

update

8931d56

format

c2f9d15

fix

44faf20

fix

719bf2c

fix

c24956c

undo

aea5996

kripken requested a review from tlively October 24, 2025 20:48

kripken mentioned this pull request Oct 24, 2025

Allow funcref literals to have inexact types #7959

Open

kripken added 11 commits October 24, 2025 14:22

update.tests

e2d9c70

fix

d360f0b

fix

2d25b8d

fix

bc122d5

fix

5de1db0

Merge remote-tracking branch 'origin/main' into import.func.type

61e0b66

fix signature of branch-hinting function

8341afd

Merge remote-tracking branch 'origin/main' into import.func.type

7d40a94

failing test

457b0e5

fix the types of imported ref.funcs in the interpreter

848151a

Merge remote-tracking branch 'myself/import.func.type' into import.fu…

a6ff39a

…nc.type

kripken added 3 commits November 4, 2025 11:01

fix spec tests

e5081e3

fix another spec test

b99dd75

fmt

87e1094

tlively reviewed Nov 5, 2025

View reviewed changes

kripken and others added 6 commits November 4, 2025 16:59

TODO: ExtractFunction casts

de14c6a

Update test/lit/exec/imported-func.wast

2802d94

Co-authored-by: Thomas Lively <[email protected]>

Update test/lit/exec/imported-func.wast

33189d5

Co-authored-by: Thomas Lively <[email protected]>

Clean up test

aab3067

Simplify gufa test

aa7290d

test exact casts too

89cf0c0

		// This is imported, so it might be anything of the proper type.
		addRoot(curr);

		@@ -0,0 +1,36 @@
		;; Import a function of type $C as type $A, cast to $A, $B, $C. All those casts
		;; should succeed. Exact casts, however TODO

Make imported functions inexact #7993

Are you sure you want to change the base?

Make imported functions inexact #7993

Conversation

kripken commented Oct 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

kripken commented Nov 4, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

kripken commented Oct 24, 2025 •

edited

Loading