On Sat, 9 Apr 2022 15:00:46 GMT, Weijun Wang <wei...@openjdk.org> wrote:
>>> You might try it on other objects:
>>
>> Nice idea to test object collection. I added two test cases.
>>
>>> The `NativeGSSContext` code still needs to be fixed.
>>
>> Could you have more details? I did not catch the comment about
>> NativeGSSContext.
>
>> Could you have more details? I did not catch the comment about
>> NativeGSSContext.
>
> When `NativeGSSContext(GSSNameElement peer, GSSCredElement myCred, int time,
> GSSLibStub stub)` is called, `pContext` is 0 and you haven't registered the
> cleaner. Later, when `initSecContext()` is called, it calls into
> `cStub.initContext()` and this native method will set `pContext` if a context
> is established. Since you haven't registered the cleaner in the ctor, this
> native context will not be released at the end.
>
> So one solution is to add a `setNativeContext(long pContext)` method and when
> this method is called you register the cleaner. Now, inside the native
> method, instead of setting the `pContext` field directly you can call this
> setter method. Or, move `cStub` and `pContext` into a new static inner class
> and let `disposerFor` work on it.
@wangweij In the following constructor, I'm not sure if the assert right.
NativeGSSContext(long pCtxt, GSSLibStub stub) throws GSSException {
assert(pContext != 0);
...
Should it be `assert(pCtxt != 0);`?
-------------
PR: https://git.openjdk.java.net/jdk/pull/8136
