Trying to build OpenSSH-2.1.0 on HP-UX 10.20

[Lutz Jaenicke]

And on we go :-(

I can now start the ssh2 client, but when connecting to the ssh server
in Protocol 2 mode:

serv01 174: /usr/local/openssh/bin/ssh -2 -p 24 serv01
jaenicke at serv01's password: 
Last login: Thu May 11 12:42:30 2000 from serv01.aet.tu-co

Pid 8011 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space,
or stack size exceeded maxssiz.
Connection to serv01 closed by remote host.
Connection to serv01 closed.

I receive a core dump, but it does not call me too much:
(gdb) backtrace
#0  0xc013ede8 in _sigvector () from /usr/lib/libc.1
#1  0xc0148dd4 in signalvector () from /usr/lib/libc.1
#2  0xc0148c74 in signal () from /usr/lib/libc.1
#3  0xe51c in sigchld_handler2 ()
#4  <signal handler called>
#5  debug (fmt=0x0) at log.c:60
#6  0xc013ede8 in _sigvector () from /usr/lib/libc.1

When running in debug mode, sshd claims to receive SIGCHLD in an endless
amount. A trace of the normal sshd (no debug mode) system calls gives:
7967: fork() = 8011
7967: close(7 [/dev/pts/4]) = 0
7967: sigprocmask(2, 0x7b038a80, 0x7b03cb20) = 0
7967: sigprocmask(2, 0x7b03cb20, 0) = 0
7967: sigprocmask(2, 0x7b038a80, 0x7b03cb20) = 0
7967: sigprocmask(2, 0x7b03cb20, 0) = 0
7967: dup(5 [/dev/ptmx]) = 7
7967: dup(5 [/dev/ptmx]) = 11
7967: fcntl(7 [/dev/ptmx], F_GETFL, 0) = 0x20002
7967: fcntl(7 [/dev/ptmx], F_SETFL, 196610) = 0
7967: fcntl(5 [/dev/ptmx], F_GETFL, 0) = 0x30002
7967: select(9, 0x7b03b408, 0x7b03b508, 0, 0) = 2
7967: read(7 [/dev/ptmx], 0x7b03b710, 16384) = 61
7967: write(6, 0x40018bc8, 44) = 44
7967: select(9, 0x7b03b408, 0x7b03b508, 0, 0) = 1
7967: write(6, 0x40018bc8, 100) = 100
7967: select(9, 0x7b03b408, 0x7b03b508, 0, 0) = 1
7967: read(7 [/dev/ptmx], 0x7b03b710, 16384) = 147
7967: select(9, 0x7b03b408, 0x7b03b508, 0, 0) = 1
7967: write(6, 0x40018bc8, 188) = 188
7967: select(9, 0x7b03b408, 0x7b03b508, 0, 0) = 0
7967: sigvector(18, 0x7b03bd28, 0x7b03bd38) = 0
7967: sigvector(18, 0x7b03c3d8, 0x7b03c3e8) = 0
The last message being repeated with increasing addresses >39000 times
(my stack size limit is 64M)

(Yes, I know that 8011 should have received the SIGSEGV. I have another
trace including the forked ones: the child corresponding to 8011 in this
example also receives sigvector(18,...) without end, then dies and the
calling process receives these sigvectors...
The complete trace is availble, be prepared for 4MB of trace :-)

Any ideas on how to track this down?

Best regards,
	Lutz
-- 
Lutz Jaenicke                             Lutz.Jaenicke at aet.TU-Cottbus.DE
BTU Cottbus               http://www.aet.TU-Cottbus.DE/personen/jaenicke/
Lehrstuhl Allgemeine Elektrotechnik                  Tel. +49 355 69-4129
Universitaetsplatz 3-4, D-03044 Cottbus              Fax. +49 355 69-4153