Re: [core-dev] latest NIO changes trigger OS X crashes

On Thu, Apr 14, 2005 at 16:43 -0400, Sam Berlin wrote:

> Yikes!  You have experienced these with the most recent code?  There 
> was a small period of time in which we experienced kernel panics on OSX 
> while we were using Pipes, but we have since changed the implementation 
> to not use pipes and the bug went away.  It may have appeared again 
> more recently while converting UDPService to using NIO.  We'll look 
> into it some more.  Thanks for the heads up!

I have had that problem with the code from CVS from two days ago, so
this should be fairly recent. I just started a version with today's CVS,
which was markedly visible due to the internal error box popping up as
it is not a sendable version for the bug servlet. I must have set that
to always send in the past. Just for the record, I did see the following
assert repeatedly:

        at com.limegroup.gnutella.Assert.silent(Assert.java:17)
        at
com.limegroup.gnutella.downloader.VerifyingFile.writeBlock(VerifyingF
ile.java:203)
        at
com.limegroup.gnutella.downloader.HTTPDownloader.doDownload(HTTPDownl
oader.java:1640)
        at
com.limegroup.gnutella.downloader.DownloadWorker.doDownload(DownloadW
orker.java:672)
        at
com.limegroup.gnutella.downloader.DownloadWorker.connectAndDownload(D
ownloadWorker.java:362)
        at
com.limegroup.gnutella.downloader.DownloadWorker.run(DownloadWorker.j
ava:202)
        at java.lang.Thread.run(Thread.java:552)
        at
com.limegroup.gnutella.util.ManagedThread.managedRun(ManagedThread.ja
va:60)
        at
com.limegroup.gnutella.util.ManagedThread.run(ManagedThread.java:49)
com.limegroup.gnutella.AssertFailure: trying to write an interval
2565718-256674
1 that wasn't leased verified:[0-524287, 1048576-2097151]
partial:[524288-782399, 800000-1048575, 2097152-2565717,
2621440-2641919, 314572
8-3164159]
discarded:[]
pending:[]
leased:[782400-799999, 2565799-2621439, 2641920-3145727,
3164160-3670015]

-- 
Jens-Uwe Mager	<pgp-mailto:F476EBC2>
_______________________________________________
core-dev mailing list
[email protected]
http://www.limewire.org/mailman/listinfo/core-dev