RFR: JDK-8318736: com/sun/jdi/JdwpOnThrowTest.java failed with "transport error 202: bind failed: Address already in use"

[Johannes Bechberger]

On Wed, 25 Oct 2023 16:39:31 GMT, Chris Plummer <cjplummer at openjdk.org> wrote:

>> Fix race condition in debugger port selection, introduced with [JDK-8317920](https://bugs.openjdk.org/browse/JDK-8317920).
>> 
>> Tested on my Mac M1, but it doesn't contain platform-dependent code.
>
> test/lib/jdk/test/lib/JDWP.java line 60:
> 
>> 58:         if (parts.length != 2) {
>> 59:             return null;
>> 60:         }
> 
> This is a bit presumptuous in that it assumes that the first 2 token line of output is the line we are looking for. The reason `parseListenAddress()` tries to match a specific pattern is because sometimes tests are run in a way that causes the debuggee to produce some extra output (like logging output) and this output needs to be skipped.  At the very least you should be checking for a pattern that includes "dt_socket".
> 
> BTW, I tried using `printf` instead of `echo` so the debug agent could be made to generate a "Listening for transport..." line, in which case this new `parseLaunchEchoListenAddress()` would not be needed, but I could not get the argument processing right.  Something about exec() and printf just doesn't work right.

Making too many not enough assumptions. The other method didn't make assumptions on the format of address and connection type, so I thought that I shouldn't too. I could create a temporary shell file that mimicks the other output though.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/16358#discussion_r1372061956