openssh unix dev - Sep 2003 - 3.6p2 build errors on buffer_get with latest portable/SNAP

Tried the user discussion list to no avail, can't get 3.6.p2 
portable running due to buffer_get errors.  Does the
latest portable SNAP incorporate the latest patches?

Running Red Hat 8.0

AS SHIPPED
/usr/sbin/sshd
...
cool, listens on 22 with 3.4p1

WITH 3.6.1p2
./configure
make
make install
...
/usr/local/sbin/sshd -t -f /usr/local/etc/sshd_config
buffer_get: trying to get more bytes 1 than in buffer 0
make: [check-config] Error 255 (ignored)

WITH 3.5p1
./configure
make
make install
...
cool, listens on 22 with 3.5p1

WITH SNAP-20030830 (had to go to 8/30 since strlcat.h was missing!?
./configure
make
make install
...
/usr/local/sbin/sshd -t -f /usr/local/etc/sshd_config
buffer_get: trying to get more bytes 1 than in buffer 0
make: [check-config] Error 255 (ignored)

What's up?

cs

"STEWARD, Curtis (Jamestown)" wrote:
> Tried the user discussion list to no avail, can't get 3.6.p2
> portable running due to buffer_get errors.  Does the
> latest portable SNAP incorporate the latest patches?
The snapshots contain everything committed to the CVS tree at the point
that they were generated.
> Running Red Hat 8.0
[snip]
> /usr/local/sbin/sshd -t -f /usr/local/etc/sshd_config
> buffer_get: trying to get more bytes 1 than in buffer 0
> make: [check-config] Error 255 (ignored)
I build and test on Redhat 8 and I've never seen those errors.  What
versions of OpenSSL and zlib do you have?  Which version of gcc?

-- 
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.

I've been running gcc 3.2, OpenSSL 0.9.7b and zlib 1.1.4.  After
receiving your reply I tried both OpenSSL 0.9.6.b and pulled
/usr/local/bin and /usr/local/sbin out of the PATH only
to get the same results.  I don't have any OS patches on 8.0,
so I can only guess I have something flakey on my box since
your up and running with the same release?  Let me know if 
you have any other ideas.

Thanks,

cs

-----Original Message-----
From: Darren Tucker [mailto:dtucker at zip.com.au]
Sent: Tuesday, September 09, 2003 8:32 PM
To: STEWARD, Curtis (Jamestown)
Cc: 'openssh-unix-dev at mindrot.org'
Subject: Re: 3.6p2 build errors on buffer_get with latest portable/SNAP


"STEWARD, Curtis (Jamestown)" wrote:
> Tried the user discussion list to no avail, can't get 3.6.p2
> portable running due to buffer_get errors.  Does the
> latest portable SNAP incorporate the latest patches?
The snapshots contain everything committed to the CVS tree at the point
that they were generated.
> Running Red Hat 8.0
[snip]
> /usr/local/sbin/sshd -t -f /usr/local/etc/sshd_config
> buffer_get: trying to get more bytes 1 than in buffer 0
> make: [check-config] Error 255 (ignored)
I build and test on Redhat 8 and I've never seen those errors.  What
versions of OpenSSL and zlib do you have?  Which version of gcc?

-- 
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.

I took out /usr/local path's just to eliminate something
that might be interfering.  I've tested both sshd_config and
ssh_host_rsa_key as provided by the "make install" with 
the same errors.  Here's the debug:

# /usr/local/sbin/sshd -ddd -t
debug2: read_server_config: filename /usr/local/etc/sshd_config
debug1: sshd version OpenSSH_3.7p1
buffer_get: trying to get more bytes 1 than in buffer 0

But look at my 8.0 rpm's... ?!
# rpm -q openssl openssl-devel zlib zlib-devel gcc glibc-devel
openssl-0.9.6b-29
openssl-devel-0.9.6b-29
zlib-1.1.4-4
zlib-devel-1.1.4-4
gcc-3.2-7
glibc-devel-2.2.93-5

-----Original Message-----
From: Darren Tucker [mailto:dtucker at zip.com.au]
Sent: Wednesday, September 10, 2003 8:32 AM
To: STEWARD, Curtis (Jamestown)
Cc: 'openssh-unix-dev at mindrot.org'
Subject: Re: 3.6p2 build errors on buffer_get with latest portable/SNAP


"STEWARD, Curtis (Jamestown)" wrote:
> I've been running gcc 3.2, OpenSSL 0.9.7b and zlib 1.1.4.  After
> receiving your reply I tried both OpenSSL 0.9.6.b and pulled
> /usr/local/bin and /usr/local/sbin out of the PATH only
> to get the same results.  I don't have any OS patches on 8.0,
> so I can only guess I have something flakey on my box since
> your up and running with the same release?  Let me know if
> you have any other ideas.
Do you have openssl or zlib libraries in /usr/local?  Something odd in
your sshd_config triggering the problem?  Corrupt public/private keys?

You can use "-ddd" for a little more debug info:
# ./sshd -ddd -t
debug2: read_server_config: filename /usr/local/etc/sshd_config
debug1: sshd version OpenSSH_3.7p1
debug3: Not a RSA1 key file /usr/local/etc/ssh_host_rsa_key.
debug1: read PEM private key done: type RSA
debug1: private host key: #0 type 1 RSA
debug3: Not a RSA1 key file /usr/local/etc/ssh_host_dsa_key.
debug1: read PEM private key done: type DSA
debug1: private host key: #1 type 2 DSA

Here's the versions I have:
# rpm -q openssl openssl-devel zlib zlib-devel gcc glibc-devel
openssl-0.9.6b-33
openssl-devel-0.9.6b-33
zlib-1.1.4-8.8x
zlib-devel-1.1.4-8.8x
gcc-3.2-7
glibc-devel-2.3.2-4.80.6

-- 
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.

Darren,

FYI, I tried a 2nd machine (this time with VMWare and 8.0)
had the same results.  The 2nd machine had identical gcc,
ssl, zlib, etc.  Here's the debug.  From what I could figure
out I could get the error on both buffer_init() 
and buffer_get().  xmalloc()?

BUFFER_INIT
# gdb -q ./sshd
(gdb) set args -t
(gdb) break buffer.c:30
Breakpoint 1 at 0x80687ce: file buffer.c, line 30.
(gdb) break buffer.c:31
Breakpoint 2 at 0x8068670: file buffer.c, line 31.
(gdb) info break
Num Type           Disp Enb Address    What
1   breakpoint     keep y   0x080687ce in buffer_init at buffer.c:30
2   breakpoint     keep y   0x08068670 in buffer_free at buffer.c:31
(gdb) run
Starting program: /root/gz/openssh/sshd -t

Breakpoint 1, buffer_init (buffer=0xbffff1f0) at buffer.c:30
30      }
(gdb) c
Continuing.
buffer_get: trying to get more bytes 1 than in buffer 0

Program exited with code 0377.
(gdb) bt
No stack.
(gdb)

BUFFER_GET
# gdb -q ./sshd
(gdb) set args -t
(gdb) break buffer.c:124
Breakpoint 1 at 0x8068896: file buffer.c, line 124.
(gdb) break buffer.c:125
Breakpoint 2 at 0x806886f: file buffer.c, line 125.
(gdb) info break
Num Type           Disp Enb Address    What
1   breakpoint     keep y   0x08068896 in buffer_get at buffer.c:124
2   breakpoint     keep y   0x0806886f in buffer_get at buffer.c:125
(gdb) run
Starting program: /root/gz/openssh/sshd -t

Breakpoint 1, buffer_get (buffer=0xbffff1f0, buf=0x0, len=1) at buffer.c:124
124                     fatal("buffer_get: trying to get more bytes %d than
in buffer %d",
(gdb) c
Continuing.
buffer_get: trying to get more bytes 1 than in buffer 0

Program exited with code 0377.
(gdb) bt
No stack.
(gdb)



Regards,

cs

-----Original Message-----
From: Darren Tucker [mailto:dtucker at zip.com.au]
Sent: Wednesday, September 10, 2003 9:14 AM
To: STEWARD, Curtis (Jamestown)
Cc: 'openssh-unix-dev at mindrot.org'
Subject: Re: 3.6p2 build errors on buffer_get with latest portable/SNAP


"STEWARD, Curtis (Jamestown)" wrote:
> 
> I took out /usr/local path's just to eliminate something
> that might be interfering.  I've tested both sshd_config and
> ssh_host_rsa_key as provided by the "make install" with
> the same errors.  Here's the debug:
> 
> # /usr/local/sbin/sshd -ddd -t
> debug2: read_server_config: filename /usr/local/etc/sshd_config
> debug1: sshd version OpenSSH_3.7p1
> buffer_get: trying to get more bytes 1 than in buffer 0
Try moving the host keys and generating new ones (particularly the SSH V1
ssh_host_key which was not shown in my debugging) for a test.  That would
be my guess.

If it's not that, you can use gdb to set a breakpoint for that line of
code, then use "bt" to get a stack trace to fund out where in the code
the
failing call is coming from:

# gdb -q ./sshd
(gdb) set args -t
(gdb) break buffer.c:124
Breakpoint 1 at 0x8062bfc: file ../buffer.c, line 124.
(gdb) run
[wait for failure]
(gdb) bt
> But look at my 8.0 rpm's... ?!
> # rpm -q openssl openssl-devel zlib zlib-devel gcc glibc-devel
You could try updating those.

-- 
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.

I'm new to some of this so bear with me, I did post a 
buffer_get() error but while debugging I could make it fail
on buffer_init() 31, weird.  Here's the bt without 
the continue:

gdb -q ./sshd
(gdb) set args -t
(gdb) break buffer.c:124
Breakpoint 1 at 0x8068896: file buffer.c, line 124.
(gdb) run
Starting program: /root/gz/openssh/sshd -t

Breakpoint 1, buffer_get (buffer=0xbffff210, buf=0x0, len=1) at buffer.c:124
124                     fatal("buffer_get: trying to get more bytes %d than
in buffer %d",
(gdb) bt
#0  buffer_get (buffer=0xbffff210, buf=0x0, len=1) at buffer.c:124
#1  0x00000000 in ?? ()
(gdb)

-----Original Message-----
From: Darren Tucker [mailto:dtucker at zip.com.au]
Sent: Wednesday, September 10, 2003 7:41 PM
To: STEWARD, Curtis (Jamestown)
Cc: 'openssh-unix-dev at mindrot.org'
Subject: Re: 3.6p2 build errors on buffer_get with latest portable/SNAP


"STEWARD, Curtis (Jamestown)" wrote:
> FYI, I tried a 2nd machine (this time with VMWare and 8.0)
> had the same results.  The 2nd machine had identical gcc,
> ssl, zlib, etc.  Here's the debug.  From what I could figure
> out I could get the error on both buffer_init()
> and buffer_get().  xmalloc()?
The error you posted earlier shows the error coming from buffer_get.

[snip]
> # gdb -q ./sshd
> (gdb) set args -t
> (gdb) break buffer.c:124
> Breakpoint 1 at 0x8068896: file buffer.c, line 124.
> (gdb) break buffer.c:125
> Breakpoint 2 at 0x806886f: file buffer.c, line 125.
> (gdb) info break
> Num Type           Disp Enb Address    What
> 1   breakpoint     keep y   0x08068896 in buffer_get at buffer.c:124
> 2   breakpoint     keep y   0x0806886f in buffer_get at buffer.c:125
> (gdb) run
> Starting program: /root/gz/openssh/sshd -t
> 
> Breakpoint 1, buffer_get (buffer=0xbffff1f0, buf=0x0, len=1) at
buffer.c:124
> 124                     fatal("buffer_get: trying to get more bytes %d
than
> in buffer %d",
> (gdb) c
The bit I wanted to see is the stack trace at this point.  Could you plese
repeat this test, but do "bt" here instead of continuing?
> Continuing.
> buffer_get: trying to get more bytes 1 than in buffer 0

-- 
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.

Ok, this is what I did:

  1)  I reinstalled psyche (8.0) w/3.7p1 - no problems with OpenSSH.
  2)  Put on 3.7p1 on a vm (VMWare) - no problems with OpenSSH.
  3)  Put on my usual config onto vm - no problems with OpenSSH.
  4)  At this point I can only assume the other two examples
fail because of some bogus configuration that I can't duplicate...  
Anyways I proceeded in forcing the core dump and received a
memory access error on the old machine:

# gdb -q ./sshd core.27823
Core was generated by `./sshd -t'.
Program terminated with signal 6, Aborted.
Reading symbols from /lib/libutil.so.1...done.
Loaded symbols for /lib/libutil.so.1
Reading symbols from /usr/lib/libz.so.1...done.
Loaded symbols for /usr/lib/libz.so.1
Reading symbols from /lib/libnsl.so.1...done.
Loaded symbols for /lib/libnsl.so.1
Reading symbols from /lib/libcrypto.so.2...done.
Loaded symbols for /lib/libcrypto.so.2
Reading symbols from /lib/libcrypt.so.1...done.
Loaded symbols for /lib/libcrypt.so.1
Reading symbols from /lib/i686/libc.so.6...done.
Loaded symbols for /lib/i686/libc.so.6
Reading symbols from /lib/libdl.so.2...done.
Loaded symbols for /lib/libdl.so.2
Reading symbols from /lib/ld-linux.so.2...done.
Loaded symbols for /lib/ld-linux.so.2
#0  0x42028cc1 in kill () from /lib/i686/libc.so.6
(gdb) bt
#0  0x42028cc1 in kill () from /lib/i686/libc.so.6
#1  0x42028ac8 in raise () from /lib/i686/libc.so.6
#2  0x4202a019 in abort () from /lib/i686/libc.so.6
#3  0x080688b2 in buffer_get ()
Cannot access memory at address 0x3
(gdb)

It looks like I can run now despite the above, if you want 
to continue to pursue the memory problem let me know, I'll 
hang onto the problematic machine :)

cs

-----Original Message-----
From: Darren Tucker [mailto:dtucker at zip.com.au]
Sent: Thursday, September 11, 2003 9:31 PM
To: STEWARD, Curtis (Jamestown)
Cc: 'openssh-unix-dev at mindrot.org'
Subject: Re: 3.6p2 build errors on buffer_get with latest portable/SNAP


"STEWARD, Curtis (Jamestown)" wrote:
> I'm new to some of this so bear with me, I did post a
> buffer_get() error but while debugging I could make it fail
> on buffer_init() 31, weird.  Here's the bt without
> the continue:
> (gdb) bt
> #0  buffer_get (buffer=0xbffff210, buf=0x0, len=1) at buffer.c:124
> #1  0x00000000 in ?? ()
Hmm, there should be more here, I don't know what there isn't.

Plan B: if you insert an abort(); immediately before the fatal at
buffer.c:124 then run it normally, you should get a core dump which you
can generate the backtrace from.  It should look something like this:

# ./sshd -t
[core dumps]
# gdb -q ./sshd core
Core was generated by `./sshd -t'.
[snip]
#0  0x4020bfd1 in kill () from /lib/libc.so.6
(gdb) bt
#0  0x4020bfd1 in kill () from /lib/libc.so.6
#1  0x4020bc94 in raise () from /lib/libc.so.6
#2  0x4020d04d in abort () from /lib/libc.so.6
#3  0x08062bd7 in buffer_get () at ../buffer.c:123
#4  0x08062a1c in buffer_get_char (buffer=0xbfffd4f0) at ../bufaux.c:262
#5  0x08061ac5 in key_load_public_rsa1 (fd=3,
    filename=0x8079e80 "/usr/local/etc/ssh_host_rsa_key",
commentp=0x0)
    at ../authfile.c:268
#6  0x080622f3 in key_load_private (
    filename=0x8079e80 "/usr/local/etc/ssh_host_rsa_key",
    passphrase=0x8078c1a "", commentp=0x0) at ../authfile.c:573
#7  0x0804d8ae in main (ac=2, av=0x8092f68) at ../sshd.c:978
#8  0x401fa4ed in __libc_start_main () from /lib/libc.so.6

-- 
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.