openssh unix dev - Apr 2004 - Solaris 8: RSA_padding_check_PKCS1_type_1:block type is not 01

Hi,

I have a returning problem with one of my sparc Solaris machines. I have a
Ultra2 with two 296MHz processors. All recent combinations of
openssh/openssl have a not permanent problem. If i try to connect to the
machine, i get sometimes these errors:

# ssh root at simba
RSA_public_decrypt failed: error:0407006A:rsa
routines:RSA_padding_check_PKCS1_type_1:block type is not 01
key_verify failed for server_host_key
# ssh root at simba
hash mismatch
key_verify failed for server_host_key
# ssh root at simba
hash mismatch
key_verify failed for server_host_key

And sometimes it works. At the moment i need about 10 tries to get in. If
i manage to get in, i can use the ssh connection for weeks without
problem.

a ssh -vvv puts out this:

...
debug2: kex_parse_kexinit:
debug2: kex_parse_kexinit: first_kex_follows 0
debug2: kex_parse_kexinit: reserved 0
debug2: mac_init: found hmac-md5
debug1: kex: server->client aes128-cbc hmac-md5 none
debug2: mac_init: found hmac-md5
debug1: kex: client->server aes128-cbc hmac-md5 none
debug1: SSH2_MSG_KEX_DH_GEX_REQUEST sent
debug1: expecting SSH2_MSG_KEX_DH_GEX_GROUP
debug2: dh_gen_key: priv key bits set: 122/256
debug2: bits set: 1049/2048
debug1: SSH2_MSG_KEX_DH_GEX_INIT sent
debug1: expecting SSH2_MSG_KEX_DH_GEX_REPLY
debug3: check_host_in_hostfile: filename /home/gleixner/.ssh/known_hosts
debug3: check_host_in_hostfile: match line 76
debug3: check_host_in_hostfile: filename /home/gleixner/.ssh/known_hosts
debug3: check_host_in_hostfile: match line 76
debug1: Host 'simba' is known and matches the RSA host key.
debug1: Found key in /home/gleixner/.ssh/known_hosts:76
debug2: bits set: 1010/2048
hash mismatch
debug1: ssh_rsa_verify: signature incorrect
key_verify failed for server_host_key
debug1: Calling cleanup 0x80627f0(0x0)




O.K. now the fun:


if i disable one processor (psradm -f 1) then i cannot reproduce the bug!
I tried sone other single/multiprocessor sparc-machines and i cannot
reproduce the bug there. I probably have to say, that only tried
sunfreeware.com packages. At the moment i use:

bash-2.03# pkginfo -l SMCossh
   PKGINST:  SMCossh
      NAME:  openssh
  CATEGORY:  application
      ARCH:  sparc
   VERSION:  3.8p1
   BASEDIR:  /usr/local
    VENDOR:  The OpenSSH Group
    PSTAMP:  Steve Christensen
  INSTDATE:  Apr 21 2004 09:31
     EMAIL:  steve at smc.vnet.net
    STATUS:  completely installed
     FILES:       52 installed pathnames
                   5 shared pathnames
                  11 directories
                  10 executables
                   1 setuid/setgid executables
                3207 blocks used (approx)

bash-2.03# pkginfo -l SMCossld
   PKGINST:  SMCossld
      NAME:  openssl
  CATEGORY:  application
      ARCH:  sparc
   VERSION:  0.9.7d
   BASEDIR:  /usr/local
    VENDOR:  The OpenSSL Group
    PSTAMP:  Steve Christensen
  INSTDATE:  Apr 21 2004 09:31
     EMAIL:  steve at smc.vnet.net
    STATUS:  completely installed
     FILES:     1542 installed pathnames
                  41 directories
                  44 executables
               19902 blocks used (approx)



Thanks for any help.

Flo

Flo Gleixner wrote:
> I have a returning problem with one of my sparc Solaris machines. I have a
> Ultra2 with two 296MHz processors. All recent combinations of
> openssh/openssl have a not permanent problem. If i try to connect to the
> machine, i get sometimes these errors:
[snip]
> if i disable one processor (psradm -f 1) then i cannot reproduce the bug!
> I tried sone other single/multiprocessor sparc-machines and i cannot
> reproduce the bug there. I probably have to say, that only tried
> sunfreeware.com packages. At the moment i use:
This is probably faulty hardware.  We have seen problems with a 300MHz 
UltraSPARC-II's w/2MB cache.  This includes the "hash mismatch"
and
"key_verify failed for server_host_key" errors (although they occurred
infrequently), and it took way too long  (many hours) to generate DSA 
host keys (this was consistent).

OpenSSL's "make test" also failed.

The faulty processor had these markings:
date code = 0598
processor wk/yr = 44/97
processor rev 52 made in uk @ d2d

See also:
http://marc.theaimsgroup.com/?l=openbsd-sparc&m=103826497310917

-- 
Darren Tucker (dtucker at zip.com.au)
GPG key 8FF4FA69 / D9A3 86E9 7EEE AF4B B2D4  37C9 C982 80C7 8FF4 FA69
     Good judgement comes with experience. Unfortunately, the experience
usually comes from bad judgement.