| Repost: Verifier Related HotSwap Bug: Why is nobody intressted in this bug report? [message #68543] |
Fri, 27 June 2003 03:27  |
Eclipse User |
|
|
|
package de.neze.experiments;
/**
* class VerifierRelatedHotSwapBug
* <blockquote>
* Reason:
* Hot code replace failed - verifier detected internal inconsistency
* or security problem
* </blockquote>
*
* Reproduce the meassage with the above reason:
* <ol><li>
* Set an break point to the indecated line below
* </li><li>
* Start this class in the debugger of eclipse (2.1.1 perhap earlier) against
* Suns 1.4.2 beta.
* </li><li>
* edit the line above the break point (without compilation errors)
* </li><li>
* resume the suspended thread: this gives the message noted above.
* </li></ol>
*
* For IBM's JDK 1.4.1 the reason for problems is:
* <blockquote>
* Reason: The target VM does not support hot code replace
* </blockquote>
* Why has IBM's JDK 1,4,1 no hotswap (I believe it has hotswap)?
*
* I detected another hotswap related bug (with a workaround),
* let me know if someody is intressted.
*
* Kind regard
*
* mle
*/
public class VerifierRelatedHotSwapBug {
/**
* class InnerClass
*/
public static class InnerClass {
private InnerClass m_oInnerSubClass;
private final void setInnerSubClass(final InnerSubClass o) {
m_oInnerSubClass = o;
}
}
/**
* class InnerSubClass
*/
public static class InnerSubClass extends InnerClass {
}
/**
* some init
*/
static {
new InnerClass();
// Class oClass = InnerClass.class;
}
/**
* main()
*/
public static void main(String[] arguments) {
System.out.println("Edit this line and hotswap: 1 23456789123456789");
System.out.println("This is the place for your breakboint");
}
}
|
|
|
|
|
|
|
| Another HCR Problem - with Eclipse 2.1.1 RC2 linux gtk [message #69587 is a reply to message #69390] |
Sun, 29 June 2003 06:38  |
| Eclipse User |
|
|
|
Hello again,
Because I had success with my brutal bug-report offer ;-)
and because I'm too tiered for real work :-(
I make the announcement from my initial posting true;
of course I know there is still no intereset :-(
* I detected another hotswap related bug (with a workaround),
* let me know if someody is intressted.
If I start my application sometimes a thread stops in a blocking
native method (read() from multicast socket) and the thread view
shows this thread as suspended but without any stack trace. If I
now HCR in a class (not nessessary related to the suspended thread)
then I get the following HCR-problem message box:
Hot code replace failed
....
Reason:
com.sun.jdi.IncompatibleThreadStateException: \
Thread was not suspended occured retrieving stack frames.
This is perhaps a JVM bug (SUN J2SE 1.4.2 final), but I get this error
with some other JVMs too, and there exists a workaround:
Resume the Thread shown as suspended.
This workaround makes HCR work (even if it failed before in the same
debug session).
Another observation is that after the resume a "real" suspend shows the
correct call stack for the thread.
I can't offer a source code example because I believe there exist a
race condition, hard to debug ... :-(
Kind regards
mle
|
|
|
Powered by
FUDForum. Page generated in 0.07575 seconds
Home