Crash problem in net.dll

Hello everybody, I got one crash issue on OpenJDK7 windows build. I captured the stack trace of such a crash, which happens intermittently on my 64bit Windows 2008 server. But unfortunately I have no simple test case to reproduce this problem. RtlInterlockedFlushSList+0x2ea (0x779F2A7F [ntdl

Re: Crash problem in net.dll

Jonathan, I think the change looks reasonable. But, we should make the equivalent change at the other site where enumAddresses_win() is called in the same file. - Michael. On 29/08/11 09:14, Jonathan Lu wrote: Hello everybody, I got one crash issue on OpenJDK7 windows build. I captured the

Re: Crash problem in net.dll

Hi Michael, Thanks for reminding, I checked the source under src/windows folder and found two more places to change. Here's the updated patch: diff -r 9b8c96f96a0f src/windows/native/java/net/NetworkInterface.c --- a/src/windows/native/java/net/NetworkInterface.cMon Jun 27 13:21:34 2011

hg: jdk8/tl/jdk: 7059542: JNDI name operations should be locale independent

Changeset: 02c2d38f4271 Author:xuelei Date: 2011-08-29 05:55 -0700 URL: http://hg.openjdk.java.net/jdk8/tl/jdk/rev/02c2d38f4271 7059542: JNDI name operations should be locale independent Reviewed-by: weijun ! src/share/classes/com/sun/jndi/ldap/ClientId.java ! src/share/classes/com

Datagram socket leak

There appears to be a socket leak in both DatagramSocket and MulticastSocket constructors. Both classes have constructors that create a socket and then attempt to bind. The bind can fail with a variety of exceptions none of which are caught by the constructor. Thus, the actual system socket t

Re: Crash problem in net.dll

Jonathan, Bug id 7084560 has been created to track this. That fix looks fine to me. Thanks, Michael. On 29/08/11 10:20, Jonathan Lu wrote: Hi Michael, Thanks for reminding, I checked the source under src/windows folder and found two more places to change. Here's the updated patch: diff -r

Re: Crash problem in net.dll

Just to be clear, I will run the tests on the fix and push the change assuming everything is ok. There is a webrev here http://cr.openjdk.java.net/~michaelm/7084560/webrev.1/ Thanks, Michael. On 29/08/11 15:33, Michael McMahon wrote: Jonathan, Bug id 7084560 has been created to track this. Tha

Re: Datagram socket leak

I haven't looked at the code in question in detail, but from the code snippet, won't the finally close the socket ( since it is not isBound )? -Chris. On 08/29/11 03:27 PM, Salter, Thomas A wrote: There appears to be a socket leak in both DatagramSocket and MulticastSocket constructors. Both c

hg: jdk8/tl/jdk: 2 new changesets

Changeset: d1c04dac850d Author:mullan Date: 2011-08-29 12:22 -0400 URL: http://hg.openjdk.java.net/jdk8/tl/jdk/rev/d1c04dac850d 7024771: "\\<>" in attribute value part of X500Principal constructor parameter makes strange effect Reviewed-by: vinnie ! src/share/classes/sun/security/

Re: Datagram socket leak

[take two!] Tom, This specific area of code was changed recently due to CR 7035556 [1], changeset [2], and this issue was discussed during the code review [3]. Essentially, bind() already closes the impl internally before throwing the exception. Does this resolve the issue for you? Or do you

RE: Datagram socket leak

I believe you're referring to the close() in the catch clause following the call to getImpl().bind. The problem I encountered was when the Datagram.bind threw an exception before it got that far. In my case, the checkListen was throwing a SecurityException, but any of the earlier throws would

Re: Datagram socket leak

Ah ok. I finally get it. In which case I think you original changes should be fine. Do you want to make similar changes to MulticastSocket and post the diffs? Also, I think a testcase would be useful here. I know it's not strictly specified that the socket should be closed if the constructor

RE: Datagram socket leak

I can post diffs, but I don't know what the process is for test cases. Not to quibble, but I'd say it's more than desirable that standard classes don't consume limited resources. It's not like the application has any way of closing the sockets once the constructor throws. Tom -Original M

RE: Datagram socket leak

Here's what I changed. I'm working with the fcs source bundle for b147, 27_jun_2011, so I may not have the latest source base. Left base folder: new Right base folder: b147 File: src\share\classes\java\net\DatagramSocket.java 186,189d185 < finally { < if( !isBound() ) <

hg: jdk8/tl/jdk: 7083576: add javax/xml/crypto into jdk_security2

Changeset: 6d6d75421e8a Author:weijun Date: 2011-08-30 10:46 +0800 URL: http://hg.openjdk.java.net/jdk8/tl/jdk/rev/6d6d75421e8a 7083576: add javax/xml/crypto into jdk_security2 Reviewed-by: mullan ! test/Makefile ! test/javax/xml/crypto/dsig/SecurityManager/XMLDSigWithSecMgr.java