RFR: 8296089: Remove debug agent code for special handling of Thread.resume()

Chris Plummer cjplummer at openjdk.org
Mon Oct 31 19:21:06 UTC 2022


The debug agent sets a breakpoint in Thread.resume() so it can prevent the debugger from suspending threads while in the resume call:

             /*
              * Track the resuming thread by marking it as being within
              * a resume and by setting up for notification on
              * a frame pop or exception. We won't allow the debugger
              * to suspend threads while any thread is within a
              * call to resume. This (along with the block below)
              * ensures that when the debugger
              * suspends a thread it will remain suspended.
              */
             trackAppResume(resumer);

Now that Thread.resume() is unsupported and just throws UnsupportedOperationException, all debug agent code related to this support can be removed. It's at least a couple of hundred lines of code, and with a fair amount of confusing synchronization. It will be nice to see it go. 

Also, there is a little bit of code to deal with the app calling Thread.suspend(). I'm a bit unsure of it's removal, because I'm not certain if we also need to consider another JVMTI agent doing a suspend. However, we don't seem to defend against another JVMTI agent resuming a thread, so maybe the debug agent is just not expected to work if another JVMTI agent interferes.  I would appreciate some insight on this possibility. I've called out this code in the inlined review comments.

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

Commit messages:
 - Get rid of Thread.resume() and Thread.suspend() support

Changes: https://git.openjdk.org/jdk/pull/10922/files
 Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=10922&range=00
  Issue: https://bugs.openjdk.org/browse/JDK-8296089
  Stats: 352 lines in 3 files changed: 1 ins; 346 del; 5 mod
  Patch: https://git.openjdk.org/jdk/pull/10922.diff
  Fetch: git fetch https://git.openjdk.org/jdk pull/10922/head:pull/10922

PR: https://git.openjdk.org/jdk/pull/10922


More information about the serviceability-dev mailing list