RFR: 8260366: ExtendedSocketOptions <clinit> can deadlock in some circumstances [v2]

Jaikiran Pai jpai at openjdk.java.net
Wed Feb 17 13:42:59 UTC 2021


On Wed, 17 Feb 2021 13:37:19 GMT, Jaikiran Pai <jpai at openjdk.org> wrote:

>> Since the test should be fast, I'd also advise to stick several identical @run lines (maybe e.g. 5 of them) to increase the probability of the deadlock to happen...
>
>> Since the test should be fast, I'd also advise to stick several identical @run lines (maybe e.g. 5 of them) to increase the probability of the deadlock to happen...
> 
> Done. The PR has been updated to run this test multiple times now.

> The --add-exports should not be needed - have you tried simply adding:
> 
> ```
> @modules java.base/sun.net.ext:+open
>                   jdk.net
> ```
> 
> before `@run`? This also has the additional benefit to declare which modules are required to run the test.

I had forgotten about the `@modules` directive. I have now updated the PR use that instead. The only minor difference between what you suggested and my updated PR is that I decided to use `:open` instead of `:+open` for the `sun.net.ext` package, since I don't use the types in that package at compile time, in that test.

-------------

PR: https://git.openjdk.java.net/jdk/pull/2601


More information about the net-dev mailing list