Attention is currently required from: Jakob Kummerow.
Patch set 1:Commit-Queue +1
1 comment:
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
Patch Set #1, Line 5: // Flags: --experimental-wasm-gc
I don't have access to https://bugs.chromium.org/p/chromium/issues/detail?id=1339321.
I deleted the test case as I assumed it was related to `br_on_non_func` which is the main instruction in the `crash` function below.
Looking at https://chromium-review.googlesource.com/c/v8/v8/+/3724787 it doesn't seem like it. Should I try replacing the branch instruction with a different kind?
To view, visit change 3803226. To unsubscribe, or for help writing mail filters, visit settings.
Attention is currently required from: Matthias Liedtke.
Patch set 1:Code-Review +1
2 comments:
Patchset:
LGTM % comment.
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
Patch Set #1, Line 5: // Flags: --experimental-wasm-gc
I don't have access to https://bugs.chromium.org/p/chromium/issues/detail?id=1339321. […]
Yes, please use one of the other `br_on_foo` instructions instead of deleting the test.
To view, visit change 3803226. To unsubscribe, or for help writing mail filters, visit settings.
2 comments:
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
Patch Set #1, Line 5: // Flags: --experimental-wasm-gc
Yes, please use one of the other `br_on_foo` instructions instead of deleting the test.
Done
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
kExprDrop,
kExprDrop,
Why do we have 2 drops here?
One is for the `ref.func` in line 36 but what is the second drop for?
To view, visit change 3803226. To unsubscribe, or for help writing mail filters, visit settings.
Attention is currently required from: Matthias Liedtke.
Patch set 2:Code-Review +1
1 comment:
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
kExprDrop,
kExprDrop,
Why do we have 2 drops here? […]
the `ref.func` in line 34. Since the loop's type indicates that it consumes a funcref, it's allowed to drop that here.
That said, please use another type of branch here, such as i31 or array or data (sorry for not being more specific earlier). `br_on_non_null` is already being tested in line 22; I believe this test is meant to provide coverage for the shared implementation of generic br_on_* (where Liftoff uses `AbstractTypeCheck`).
To view, visit change 3803226. To unsubscribe, or for help writing mail filters, visit settings.
Patch set 3:Commit-Queue +1
1 comment:
File test/mjsunit/regress/wasm/regress-crbug-1339321.js:
kExprDrop,
kExprDrop,
the `ref.func` in line 34. […]
Done
To view, visit change 3803226. To unsubscribe, or for help writing mail filters, visit settings.
Patch set 3:Commit-Queue +2
V8 LUCI CQ submitted this change.
2 is the latest approved patch-set.
The change was submitted with unreviewed changes in the following files:
```
The name of the file: test/mjsunit/regress/wasm/regress-crbug-1339321.js
Insertions: 2, Deletions: 1.
@@ -34,7 +34,8 @@
kExprRefNull, kAnyRefCode,
kExprLoop, loop_type,
kExprRefNull, kAnyRefCode,
- kExprBrOnNonNull, 0,
+ kGCPrefix, kExprBrOnI31, 0,
+ kExprDrop,
kExprDrop,
kExprEnd, // loop
]);
```
[wasm-gc] Remove br_on_func & br_on_non_func op
Preparation step to remove the subtype relationship between funcref and anyref.
Bug: v8:7748
Change-Id: Ie4b29bfc4b874aaca668082018f5359d1b6e3a2e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3803226
Reviewed-by: Jakob Kummerow <jkum...@chromium.org>
Commit-Queue: Matthias Liedtke <mlie...@chromium.org>
Cr-Commit-Position: refs/heads/main@{#82125}
---
M src/compiler/wasm-compiler.cc
M src/compiler/wasm-compiler.h
M src/wasm/baseline/liftoff-compiler.cc
M src/wasm/function-body-decoder-impl.h
M src/wasm/graph-builder-interface.cc
M src/wasm/wasm-opcodes.h
M test/common/wasm/wasm-macro-gen.h
M test/mjsunit/regress/wasm/regress-crbug-1339321.js
M test/mjsunit/wasm/wasm-module-builder.js
M test/unittests/wasm/function-body-decoder-unittest.cc
10 files changed, 31 insertions(+), 84 deletions(-)