BIND 8.2.3 compile blows up on HP-UX 10.20 - will this fix work?

Loomis, Rip GILBERT.R.LOOMIS at saic.com
Mon Feb 26 20:34:10 UTC 2001


All--
I'm the one who submitted the original fix,
to #undef ORD_32.  At the time of the 8.2.2-P5
release, it was defined in
  src/port/hpux10/include/port_before.h
at about line 6.  Unfortunately, on HP-UX 10.20
with GCC as the compiler, the resulting binaries
went into infinite loops when trying to do any
operations involving DSA crypto--to quote the
original problem report:
"The problem is evident in src/lib/cylink/ctk_prime.c
where the function GenPrime() is unable to ever
find a number that bnGcd_16 (in src/lib/cylink/bn16.c)
believes is prime.  It thus sits in a loop forever
trying new numbers over and over and never finding one
that will work."

If you again define ORD_32, then it might fix the
compilation with some of the HP commercial compilers,
but it breaks GCC.  I'd love to know where the
fault really lies...but for right now *please* don't
change it back to defining ORD_32.  We should
identify what's faulty (GCC, the HP compiler, or
the Cylink DSA code) and fix it properly.

My test box:  C180 Visualize workstation, HP-UX 10.20
with latest patch cluster, GCC 2.95.2.  We have the
commercial compiler available, but only under HP-UX
11.00.

Rip Loomis		Voice Number: (410) 953-6874
--------------------------------------------------------
Senior Security Engineer
Center for Information Security Technology
Science Applications International Corporation
http://www.cist.saic.com


> -----Original Message-----
> From: Murphy Brian PRC USAFE CSS/SCN
> [mailto:brian.murphy at ramstein.af.mil]
> Sent: Monday, February 26, 2001 3:44 AM
> To: 'Mark.Andrews at nominum.com'
> Cc: 'bind-workers at isc.org'
> Subject: RE: BIND 8.2.3 compile blows up on HP-UX 10.20 - 
> will this fix
> wo rk?
> 
> 
> 
> BIND 8.2.2-P7 compiled fine.
> 
> Does the #define listed below break anything, or is it correct for the
> machine (assuming this is setting to 32 bit, big_endian)?  Again, the
> compile completed and named started without errors after 
> making the #define
> change.
> 
> Lemme know - Tks,
> 
> murph
> 
> -----Original Message-----
> From: Mark.Andrews at nominum.com [mailto:Mark.Andrews at nominum.com]
> Sent: Saturday, February 24, 2001 1:29 AM
> To: Murphy Brian PRC USAFE CSS/SCN
> Cc: 'bind-workers at isc.org'
> Subject: Re: BIND 8.2.3 compile blows up on HP-UX 10.20 - 
> will this fix
> work? 
> 
> 
> 
> 	HPUX 10.20 early versions of the HP C/ANSI C compiler do
> 	not work, version B.10.20.15 in known to work.
> 
> > 
> > Sending this to the list per Paul Vixie's request - comments?
> > 
> > When trying to compile BIND 8.2.3 running "make all" on 
> HP-UX 10.20 on an
> HP
> > 9000/715 system with the HP-UX ANSI C compiler, the compile 
> blew up on
> > cylink and the ctk_endian.h definitions - here is the error:
> > 
> >         : ld -x -r bsafe_link.o -o bsafe_link.out && \
> >          : mv bsafe_link.out bsafe_link.o
> >         if test ! -d threaded ; then mkdir threaded ; else true ; fi
> >         cc -Ae -z -I../../port/hpux10/include -I../../include -g
> > -Wl,+vnocompatw
> > arnings  -D_REENTRANT -I../cylink -I../dnssafe -DCYLINK_DSS 
> -DHMAC_MD5
> > -DUSE_MD5
> >  -DDNSSAFE -c cylink_link.c -o threaded/cylink_link.o
> > cpp: "ctk_endian.h", line 135: error 4062: Not defined 
> basic word type
> > ORD_32 or
> >  ORD_16.
> > *** Error exit code 1
> > 
> > Stop.
> > *** Error exit code 1
> > 
> > Stop.
> > *** Error exit code 1
> > 
> > Stop.
> > 
> > 
> > The fix action was to add the following at line 115 in
> > /var/obj/bind/lib/cylink/SRC/ctk_endian.h:
> > 
> > #define ORD_32
> > 
> > which allowed the compile to complete and apparently the 
> binary is working
> > (the compiling box we are using is cfg'd to forward, so not 
> much of a test
> > of the compile).
> > 
> > Could there be any problems with this?  I would like to 
> port the binaries
> to
> > our production systems soon, given the announced problem 
> with pre-8.2.3
> > releases.
> > 
> > Lemme know if more is needed - Tks,
> > 
> > murph
> > 
> --
> Mark Andrews, Nominum Inc.
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE: +61 2 9871 4742                 INTERNET: 
> Mark.Andrews at nominum.com
> 


More information about the bind-workers mailing list