diff options
| author | Thomas Lively <tlively@google.com> | 2024-12-19 17:45:05 -0800 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-12-19 17:45:05 -0800 |
| commit | 7c42700738fa9f9aff29dd930e9fd7f6cf71fc29 (patch) | |
| tree | 75b8df675e6da753d1209f7b02c9781c8db6cc4e /test/lit/passes | |
| parent | 74b2b064e59beee84e88afaa952a8c51cf9309a4 (diff) | |
| download | binaryen-7c42700738fa9f9aff29dd930e9fd7f6cf71fc29.tar.gz binaryen-7c42700738fa9f9aff29dd930e9fd7f6cf71fc29.tar.bz2 binaryen-7c42700738fa9f9aff29dd930e9fd7f6cf71fc29.zip | |
Do not optimize atomic gets in GUFA (#7161)
Conservatively avoid introducing synchronization bugs by not optimizing
atomic struct.gets at all in GUFA. It is possible that we could be more
precise in the future.
Also remove obsolete logic dealing with the types of null values as a
drive-by. All null values now have bottom types, so the type mismatch
this code checked for is impossible.
Diffstat (limited to 'test/lit/passes')
| -rw-r--r-- | test/lit/passes/gufa-refs.wast | 53 |
1 files changed, 53 insertions, 0 deletions
diff --git a/test/lit/passes/gufa-refs.wast b/test/lit/passes/gufa-refs.wast index 80fd32386..ec67cf40c 100644 --- a/test/lit/passes/gufa-refs.wast +++ b/test/lit/passes/gufa-refs.wast @@ -6137,3 +6137,56 @@ ) ) ) + +;; Atomic accesses require special handling +(module + ;; CHECK: (type $A (shared (struct (field i32)))) + (type $A (shared (struct (field i32)))) + + ;; CHECK: (type $1 (func)) + + ;; CHECK: (func $gets (type $1) + ;; CHECK-NEXT: (local $0 (ref $A)) + ;; CHECK-NEXT: (local.set $0 + ;; CHECK-NEXT: (struct.new_default $A) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: (drop + ;; CHECK-NEXT: (i32.const 0) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: (drop + ;; CHECK-NEXT: (struct.atomic.get acqrel $A 0 + ;; CHECK-NEXT: (local.get $0) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: (drop + ;; CHECK-NEXT: (struct.atomic.get $A 0 + ;; CHECK-NEXT: (local.get $0) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: ) + ;; CHECK-NEXT: ) + (func $gets + (local (ref $A)) + (local.set 0 + (struct.new_default $A) + ) + (drop + ;; This is optimizable. It reads from shared memory, but there is only one + ;; possible value that can be read. + (struct.get $A 0 + (local.get 0) + ) + ) + (drop + ;; We do not (yet) optimize atomic gets. + (struct.atomic.get acqrel $A 0 + (local.get 0) + ) + ) + (drop + ;; We do not (yet) optimize atomic gets. + (struct.atomic.get $A 0 + (local.get 0) + ) + ) + ) +) |