Solaris 2.6 getting a number of packet_read fatal errors

[Mark D. Baushke]

I have upgraded to OpenSSH 3.0.1p1 on a solaris 2.6 box and have
started to see a larger number of fatal() error messages showing up in
my syslog files than was happening in previous releases.

The message is

    fatal: Read from socket failed: Resource temporarily unavailable

which is coming from packet.c in the packet_read function where a read
is return a len of -1 and the errno is EAGAIN. I would have suspected
that if EAGAIN is found it should try to reread the socket... The
select just before the read seems to handle the EAGAIN errno in a
reasonable manner.

Is anyone else seeing anything similar?

I am not personally seeing the fatal error in my connections, so I am
not sure what is triggering them.

	Thanks,
	-- Mark