On 13/06/2012 12:25 PM, Poonam Bajaj wrote:
Hi David,
On 6/13/2012 7:35 AM, David Holmes wrote:
Hi Poonam,
It seems to me that rather than passing the ThreadProxy through to
f.sender the frame, which has to be a frame of some thread, should
already know what that thread is and so be able to access it directly.
In other words: shouldn't each CFrame maintain a reference to the
thread it corresponds to?
Yes, it would have made more sense to have a reference of ThreadProxy in
the Frame classes. But that requires restructuring of lot more code. All
the Debugger classes (e.g.
agent/src/share/classes/sun/jvm/hotspot/debugger/linux/LinuxCDebugger.java)
also need to be touched that have the topFrameForThread() where we
create the first frame for the thread.
Oh I see. I was hoping it would actually lead to fewer code changes. It
seems risky to pass in a thread reference when there is only ever one
valid thread reference that can be passed - but no way to verify that
you actually pass the correct one.
This fix was mainly for 6uxx and it didn't seem reasonable to me to
restructure and touch too many classes for this simple fix.
Ok.
David
-----
Thanks,
Poonam
David
On 12/06/2012 11:05 PM, Poonam Bajaj wrote:
Hi,
Please review this fix for bug 6310967
<http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6310967>.
6310967 <http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6310967>:
SA: jstack -m produce failures in output
http://cr.openjdk.java.net/~poonam/6310967/webrev.00/
Problem: jstack -m fails with UnalignedAddressException.The problem is
that while finding the caller frame of a frame in sender() method, we
don't check the validity of the frame pointer (rbp / ebp).
These changes add a simple check that the frame pointer(rbp) should be a
valid pointer on the stack by making sure that it is not less than the
stack
pointer(rsp).
Thanks,
Poonam
