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.
I encountered a deadlock with a dev build of NB 6.9 (100514-539d9a81eb38) after changing some files externally. The jstack output is attached.
Could you please attach the output?
Created attachment 99129 [details] jstack output when NetBeans was deadlocked Hmmm, I thought I had.
platform/fs
Probably duplicate of http://netbeans.org/bugzilla/show_bug.cgi?id=185900
I wish it was duplicate. But this one is more serious, it shows that events are delivered under the "idleIO" mode and as such any code which reacts to them is also running in idleIO mode. Possibly P1, depends how often this problem happens. Another question to ask is why the code does not leave the waitIOLoad method? After a while the load shall drop down, shall it not?
Fixed in core-main#9519a4901110. I am not sure what are the next steps and whether this needs to be fixed for 6.9. However the code shall not be better than it was before. If you decide to merge to 6.9 then Tonda please: hg pull -r 9519a4901110 http://hg.netbeans.org/core-main/ hg merge
> I am not sure ...whether this needs to be fixed for 6.9. > However the code shall not be better than it was before. Not better than before? Not sure if for 6.9? If the report shows that "events are delivered under the idleIO mode and as such any code which reacts to them is also running in idleIO mode" and the patch fixes it, it should prevent the deadlock and we want to have it integrated in 6.9 IMO. Or can the change lead to some other problems? Can someone do a review for you? Let's wait with backport to release69 after this is more clear.
Integrated into 'main-golden', will be available in build *201005250001* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main/rev/9519a4901110 User: Jaroslav Tulach <jtulach@netbeans.org> Log: #186291: Delivering events from refreshSlow outside of idleIO block
"not be better" this should have been: "now be better". The only person that comes to my mind to do review is Tomáš Z.
Tomasi, can you review? Thanks.
I don't have so deep knowledge of master fs. Jarda explained me what the patch does and it seems compatible with Jarda description. The patch is probably OK.
Backported to release69 for RC2.
*** Bug 186732 has been marked as a duplicate of this bug. ***