Bug 750

Summary:	"buffer_append_space: alloc 10506240 not supported" on AIX5 when attempting to use compression
Product:	Portable OpenSSH	Reporter:	andrew_wrobel
Component:	ssh	Assignee:	OpenSSH Bugzilla mailing list <openssh-bugs>
Status:	CLOSED INVALID
Severity:	major
Priority:	P2
Version:	-current
Hardware:	All
OS:	AIX

Description andrew_wrobel 2003-10-29 05:22:10 AEDT

Using openssh-3.7.1p2 on AIX 4.3.3, 5.1 and 5.2.

  When attempting to connect, with compression, to sshd running on an AIX 5 
machine, sshd kills the connection after accepting the users password.

  syslog, on the AIX 5 machine, reports the following error message:

fatal: buffer_append_space: alloc 10506240 not supported

  Attempt to connect, without compression, to sshd running on AIX 5.x works 
fine.

  AIX 4.3.3 doesn't have the issue at all.  I've found references to this 
problem going back to openssh 3.4 without any resolution.  Any help/suggestion 
would be greatly appreciated.

Comment 1 Darren Tucker 2003-10-29 09:02:11 AEDT

I've seen this a few times before and it's usually an issue with zlib.  Please 
try recompiling zlib, then recompiling openssh.  Make sure you're using zlib 1.
1.4, and that there isn't an old zlib somewhere in your link path.

Comment 2 andrew_wrobel 2003-10-29 09:31:28 AEDT

I had seen references to that as well.

I have recompiled zlib 1.1.4, verified that an older zlib isn't located on the 
system.

I've recompiled openssh a couple of different times, using different configure 
options, still the same issue.

Would compiling openssh WITHOUT zlib fix the problem?

Comment 3 Darren Tucker 2003-10-29 09:51:22 AEDT

zlib is required, you can't compile without it.

You're not using the gcc-2.9x that came on the AIX toolbox CD are you?  I've 
personally seen that miscompile openssl, and I suspect it of causing a number of 
other problems.  If you're using it, I strongly suggest you upgrade to a newer 
gcc.  If that's not possible, compile zlib and openssh (and openssl if you 
compiled it with that compiler) without optimization and see if the problem 
persists.

Comment 4 Darren Tucker 2003-12-22 20:44:47 AEDT

2 months no reply = closed bug.

Comment 5 Damien Miller 2004-04-14 12:24:19 AEST

Mass change of RESOLVED bugs to CLOSED