RFR: 8332488: Add JVMTI DataDumpRequest to the debug agent. [v3]

[Chris Plummer]

On Thu, 8 Aug 2024 21:05:00 GMT, Leonid Mesnik <lmesnik at openjdk.org> wrote:

> > FYI, this is what the (entire) debuggee output looks like for this test:
> > ```
> > Deuggee output:
> > ```
> 
> Should we see the same report for CTRL+\ or after sending SIGQUIT signal? I see it should cause this request.

Yes, I believe so. In fact during testing I think more than once I saw the dump just by issuing a random jcmd. I believe what happened is the debug agent had suspended all threads so that the attach would time out (due to Signal Handler thread being suspended). The failure_handler would timeout a number of jcmds in sequence. I think this causes a number of SIGQUIT signals to queue up. Once the debugger eventually resumed all threads, the first SIGQUIT would cause the attach listener to start up, and subsequent ones caused a thread dump and data dump. Kind of strange, but doesn't really have anything to with my changes, and I've seen thread dumps like this before due to attach issues.

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

PR Comment: https://git.openjdk.org/jdk/pull/20367#issuecomment-2276668328