| Issue |
79333
|
| Summary |
clang warns that completion is called twice but that can be ruled out through static code analysis
|
| Labels |
clang
|
| Assignees |
|
| Reporter |
CodingMarkus
|
clang says that completion may be called twice in the following code:
```
- (void)readPacket:
(void (^)( NSData *_Nullable, NSError *_Nullable))completion
{
NSData * packet = nil;
NSError * error = nil;
[self obtainLock];
{
if (_pendingCallback) {
error = [self makePendingError];
} else {
packet = [self getPacket];
if (packet) {
if (_asyncCompletion) {
dispatch_async(_queue, ^{
completion(packet, nil);
});
packet = nil;
}
} else {
_pendingCallback = completion;
}
}
}
[self releaseLock];
if (packet || error) {
completion(packet, error); // <-- Here clang says: "Completion handler is called twice"
}
}
```
I don't see how this would be the case. Completion is only called in the end, if either `packet` or `error` is set. Completion may only be called earlier when it should finish async but in that case `packet` is explicitly set back to `nil`, so it won't be called again in the end. Or it may be called later, when it is stored to `_pendingCallback`, but that only happens if `packet` and `error` are both `nil`, so there is for sure no call in the end either. When either `packet` or `error` are set in the end, completion was for sure never called before, never stored anywhere and never passed to any external function.
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
