Re: [cp-patches] [RFA/JDWP] ClassTypeCommandSet

[Aaron Luchko]

On Thu, 2005-08-04 at 18:22 -0400, Bryce McKinlay wrote:
> Aaron Luchko wrote:
> 
> >+    try
> >+      {
> >+        if ((invokeOpts & 
> >JdwpConstants.InvokeOptions.INVOKE_SINGLE_THREADED) 
> >+            != 0)
> >+          {
> >+            // We must suspend all running threads first
> >+            suspendSuccess = 
> >vm.suspendAllThreadsExcept(Thread.currentThread().
> >+                                                          getThreadGroup());
> >+          }
> >+        return vm.executeMethod(null, thread, clazz, method, values,
> >+                                           false);
> >+      }
> >+    catch (Exception ex)
> >+      {
> >+        if (suspendSuccess)
> >+          { // We must first resume the suspended threads
> >+            
> >vm.resumeAllThreadsExcept(Thread.currentThread().getThreadGroup());
> >+          }
> >  
> >
> 
> What happens if some threads are already suspended by the debugger? 
> Looks like this would resume them which wouldn't be the expected behaviour.

If a thread must be resumed as many times as it's been suspended in
order to resume fully, this is the behavior of resumeAllThreadsExcept.

However you did bring to my attention an error I made in this patch
where I don't resume the suspended threads if the execution was
successful.

Aaron

jdwpClassTypeMethodResult2.patch
Description: Text Data