Jun 22 22:47:29 *server* sshd[711]: fatal: mmap(65536): Invalid argument And it closes the connection. Tested on two systems, both kernel 2.2.21, on other two systems with kernel 2.4.18 sshd works perfectly, and I'm able to login.
Disabling PrivSep (add "UsePrivilegeSeparation no" to sshd_config) should get it working on 2.2 kernels. 3.3p1 defaults PrivSep to on. Previous releases (that supported it) defaulted to off.
The issue is mmap() on Linux 2.2 kernel does not support any of the useful options in order to allow PrivSep and Compression at the same time. Either disable privsep (like what dtucker@ said) or change 'Compression yes' to 'Compression no'. There is a patch being worked out to straight some of this out.
*** Bug 286 has been marked as a duplicate of this bug. ***
OpenServer 5.0.x also rejects incoming connections. The user will be prompted for their password, but the connection will close immediately after. The fixes suggested (setting "UsePrivilegeSeparation" and/or "Compression" to "no") resolve the inability to connect, though.
*** Bug 288 has been marked as a duplicate of this bug. ***
*** Bug 289 has been marked as a duplicate of this bug. ***
*** Bug 292 has been marked as a duplicate of this bug. ***
*** Bug 293 has been marked as a duplicate of this bug. ***
bug also found on Kernel 2.2.17-14 installed in Cartman Redhat 6.1 not seen on Mac OS X 10.1.5 "Compression no" solves on Linux
Also found on Kernel 2.2.14-5.0 Red Hat 6.2 (Zoot). I can't make Compression no work. With UsePrivilegeSeparation No, it doesn't work either.
*** Bug 299 has been marked as a duplicate of this bug. ***
If neither "Compression no" nor "UsePrivilegeSeparation no" work, are your host keys in ssh1 format? It seems that the new ssh's doesn't understand ssh1 format host keys. If you use ssh2-format rsa/dsa keys, does it work?
OpenSSH 3.4p1 fixed this problem. Kernel 2.2.14-5.0 - Compression yes and UsePrivilegeSeparation yes work Kernel 2.2.12-32 - Compression musr be disabled for UsePrivilegeSeparation to work. I wanted to pass that along, I didn't know if that was "known" or not.
Mass change of RESOLVED bugs to CLOSED