help with debugging "Close for unknown channel 0" in dropbear-0.53.1

[Bodo Meissner]

On Tue, May 08, 2012 at 07:54:40AM +0800, Matt Johnston wrote:

> If you run Dropbear with "-F -E -v" arguments it will run in the foreground with output to standard output/error.

Hello Matt,

thanks for this info.

I tried this but it did not help because the problem seems to be sensitive to timing.
When I use strace or switch on this trace output, the error disappears.

With debugging on the PC we found out that the library SSH.NET sometimes sends SSH_MSG_CHANNEL_CLOSE twice for the same channel number.
This seems to depend on which end first decides to close the channel.
I filed a bug report for SSH.NET with a description of a possible fix.
http://sshnet.codeplex.com/workitem/1227

I don't know if the behaviour is specified in the SSH protocol or how other SSH servers will handle this.
Would it make sense for dropbear to ignore CHANNEL_CLOSE messages for unknown channels, assuming it has already been closed?


Bodo
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
Url : http://lists.ucc.gu.uwa.edu.au/pipermail/dropbear/attachments/20120508/540ee8c6/attachment.pgp