Testing of recent commits

Ben Lindstrom mouring at etoh.eviladmin.org
Fri Nov 21 06:34:17 EST 2003


I could have swore I was told it was not required, but I don't have
HP/UX.  And my initial check in comments actually elude to HP/UX
maybe problematic.

However, with your change we should do:

                pw_password = spw->sp_pwdp;
-# endif
-# if defined(HAVE_GETPWANAM) && !defined(DISABLE_SHADOW)
+# elif defined(HAVE_GETPWANAM) && !defined(DISABLE_SHADOW)
        struct passwd_adjunct *spw;

Because they are all exclusive cases.  None of them should ever overlap.

If your diff plus this one passes then submit a patch and I'll take
care of it.

- Ben


On Thu, 20 Nov 2003, Darren Cole wrote:

> I get a very similar problem on HP-UX 10.26 as described by Lutz
> Jaenicke <Lutz.Jaenicke at aet.TU-Cottbus.DE> once I made this change
> (against current anonymous cvs):
> -----begin diff-----
> RCS file: /cvs/openssh/openbsd-compat/xcrypt.c,v
> retrieving revision 1.5
> diff -u -r1.5 xcrypt.c
> --- openbsd-compat/xcrypt.c     25 Sep 2003 10:18:34 -0000      1.5
> +++ openbsd-compat/xcrypt.c     19 Nov 2003 23:07:51 -0000
> @@ -89,7 +89,7 @@
>   {
>          char *pw_password = pw->pw_passwd;
>
> -# if defined(HAVE_SHADOW_H) && !defined(DISABLE_SHADOW)
> +# if defined(HAVE_SHADOW_H) && !defined(DISABLE_SHADOW) &&
> !defined(HAVE_SECUREWARE)
>          struct spwd *spw = getspnam(pw->pw_name);
>
>          if (spw != NULL)
> -----end diff-----
> The above is needed otherwise you can end up trying to define variable
> spw twice when you use shadow passwords on a Secureware enabled system
> (i.e. Trusted CMW HP-UX 10.26).  I know there are other problems, but
> it got me to the below errors:
>
> gmake[1]: Entering directory
> `/home/oeddev/projects/openssh/openssh/openbsd-compat'
> gcc -g -O2 -Wall -Wpointer-arith -Wno-uninitialized -I. -I.. -I. -I./..
> -I/opt/openssl/include
> -I/opt/zlib/include  -D_HPUX_SOURCE -D_XOPEN_SOURCE
> -D_XOPEN_SOURCE_EXTENDED=1 -DHAVE_CONFIG_H
> -c getrrsetbyname.c
> getrrsetbyname.c: In function `getrrsetbyname':
> getrrsetbyname.c:191: warning: implicit declaration of function
> `res_init'
> getrrsetbyname.c:207: warning: implicit declaration of function
> `res_query'
> getrrsetbyname.c:265: error: `T_SIG' undeclared (first use in this
> function)
> getrrsetbyname.c:265: error: (Each undeclared identifier is reported
> only once
> getrrsetbyname.c:265: error: for each function it appears in.)
> getrrsetbyname.c: In function `parse_dns_response':
> getrrsetbyname.c:366: error: `HFIXEDSZ' undeclared (first use in this
> function)
> getrrsetbyname.c: In function `parse_dns_qsection':
> getrrsetbyname.c:437: warning: implicit declaration of function
> `dn_expand'
> gmake[1]: *** [getrrsetbyname.o] Error 1
> gmake[1]: Leaving directory
> `/home/oeddev/projects/openssh/openssh/openbsd-compat'
> gmake: *** [openbsd-compat/libopenbsd-compat.a] Error 2
>
> > ----------------------------------------------------------------------
> > HP-UX 10.20 is shipped with Bind4.
>
> I am not quite sure what version of Bind our systems have, or even if
> bind is installed at all on fielded systems.  Cygwin's solutions sounds
> like it was to include minires.  Installing a newer bind (or whatever)
> really isn't the most desirable option here.  I would rather see
> openssh continue to compile without the need for updated bind,
> additional minires, or some other extra piece of code.
>
> Just my two cents.
>
> --
>    Darren Cole
>    Software Engineer
>    email: dcole at keysoftsys.com
>
> _______________________________________________
> openssh-unix-dev mailing list
> openssh-unix-dev at mindrot.org
> http://www.mindrot.org/mailman/listinfo/openssh-unix-dev
>




More information about the openssh-unix-dev mailing list