This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Using Aug 6 dev sources, I type: cd nbbuild ant -Dfixedmodules=core,openide,core/windows,core/ui -Dmodules= all (Linux, JDK 1.4, Ant 1.4.1) and NB gets as far as the splash screen before it deadlocks in the window system. Will attach complete log.
Created attachment 7028 [details] Initial log & thread dump
Doesn't seem to be a deadlock. The main thread is waiting for AWT and AWT is runnable doing something. At most it can be some infinite loop but not deadlock.
Yes it is not deadlock. Is this behaviour reproducible?
I tested it with JDK 1.4.0 and ant 1.4.1 with command given above and it works fine. In addition it seems not to be deadlock from thread dump. So I close this issue.
Yes, dump looks like infinite loop rather than deadlock, but it certainly stops the AWT thread either way - splash screen does not repaint, e.g. Sometimes the AWT thread is calling HashMap.put, but anyway it is always in ModeData$CookiesImpl.createTCRefs. Tried with a fresh source dir - checkout out core, openide, nbbuild, used existing (up-to-date) nbextra, JDK 1.3.1_03. Seems to not be reproducible if I use javac. Thinking it might be a Jikes bug (see issue #13535 for example), since I use Jikes when not doing a clean testing build, I tried it after compiling with jikes 1.16 - but then I get a HotSpot error! ---%<--- HotSpot Virtual Machine Error, Internal Error Please report this error at http://java.sun.com/cgi-bin/bugreport.cgi Error happened during: scavenge Error ID: 47454E45524154452F4F502D41500E4350500842 Problematic Thread: prio=1 tid=0x808e488 nid=0x53b5 runnable ---%<--- Not sure what to do with that. 1.4.1_01 also barfs: ---%<--- HotSpot Virtual Machine Error, Internal Error Please report this error at http://java.sun.com/cgi-bin/bugreport.cgi Java VM: Java HotSpot(TM) Client VM (1.4.0_01-b03 mixed mode) Error happened during: generation collection for allocation Error ID: 47454E45524154452F4F502D41500E4350500848 Problematic Thread: prio=1 tid=0x0x811f400 nid=0x5418 runnable ---%<---
From thread dump it does not seem to me that there is any loop. In such case call stack of AWT thread would be much longer IMO. Call stack from thread dump is natural initialization of workspace, mode, tcref - it is fine. Though initialization of toolbar is not very nice: waitFinished() and then waiting for some AWTTask. Anyway it does not cause problem in this case. Dafe please think about some toolbar initialization redesign if possible.
Eh, what is wrong on the toolbar initialization? It simply parses a folders, and then creates the components (which have to be done in AWT). Main thread waits for it before it will show the main window, it would flicker otherwise.
To Marek: I didn't mean an infinite recursion, I meant a situation where something is processed, it fires changes, the changes are found, it is reprocessed, ... similar bugs have caused startup to halt in the past under some conditions. I remember one case involving folder ordering and weak references that was pretty nasty.
verified