<!DOCTYPE html><html><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body>
    On 05/06/2024 10:37, 何品(虎鸣) wrote:<br>
    <blockquote type="cite" cite="mid:b42ded56-d92e-4272-9f34-3485e18401b3.hepin.p@alibaba-inc.com">
      
      <div class="__aliyun_email_body_block">
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;">Thanks, </span></div>
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;">1. it hung
            when we are sharing the common usage of the Virtualthread
            (the default scheduler) cross two modules.</span></div>
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;">2. after
            that, I try to hack the virtual thread builder with
            separated ThreadPoolExecutor.</span></div>
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;">3. but
            still , it hung.</span></div>
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;"><br>
          </span></div>
        <div style="clear:both;"><span style="font-family:Tahoma,Arial,STHeiti,SimSun;">when it
            hung, `<span style="color:#000000;font-family:Helvetica,Tahoma,Arial;font-size:14.0px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:.0px;text-transform:none;word-spacing:.0px;white-space:normal;text-decoration-thickness:initial;text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline;">jcmd
              Thread.print` prints nothing. and dump programly print
              nothing too.</span></span></div>
        <div style="clear:both;"><span style="font-family:Helvetica,Tahoma,Arial;color:#000000;font-size:14.0px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:.0px;text-transform:none;word-spacing:.0px;white-space:normal;text-decoration-thickness:initial;text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline;"><br>
          </span></div>
        <div style="clear:both;"><span style="font-family:Helvetica,Tahoma,Arial;color:#000000;font-size:14.0px;font-style:normal;font-variant-ligatures:normal;font-variant-caps:normal;font-weight:400;letter-spacing:normal;text-align:start;text-indent:.0px;text-transform:none;word-spacing:.0px;white-space:normal;text-decoration-thickness:initial;text-decoration-style:initial;text-decoration-color:initial;float:none;display:inline;">After
            switch back to normal thread, it never hang.</span></div>
        <div style="clear:both;"><span style="font-family:Helvetica,Tahoma,Arial;">Some
            information:<br>
            module A is using `Object.notifyAll, Object.wait` and module
            B is using `CompletableFuture.get` (may > 100 times in
            one rune)</span></div>
        <div style="clear:both;"><span style="font-family:Helvetica,Tahoma,Arial;">I was thinking
            if that could be a problem of notification missing, where in
            Module A, the concurrency is 3000, protected by a <span>semaphore,
              but the underling Carrier Thread is only 128, and 3000
              > 128. </span></span><br>
        </div>
      </div>
    </blockquote>
    <br>
    If it's using synchronized/Object.wait then this may be related to
    pinning. When there is both object monitors and j.u.concurrent locks
    in play then it's possible to create deadlock scenarios due to
    starvation, or selecting a successor or thread to wakeup and the
    thread can't continue because there are no carriers available.
    Object.wait will temporarily increase parallelism to smooth and help
    some cases but it may not help you here, and does nothing when the
    scheduler has been changed to be something other than a ForkJoinPool
    instance.<br>
    <br>
    When you say "prints nothing" then you mean this literally or do you
    mean that jcmd is hung too? If so that's a hint that may be the lock
    for standard output is held by a virtual thread that can't continue
    because there are no carriers available.<br>
    <br>
    It would be interesting to try the latest Loom EA builds which has
    changes to the object monitor implementation that doesn't pin. Would
    you have time to try these builds out?<br>
    <br>
    -Alan<br>
    <br>
    [1] <a class="moz-txt-link-freetext" href="https://jdk.java.net/loom/">https://jdk.java.net/loom/</a><br>
  </body>
</html>