On Fri, 9 Aug 2002, Uwe Lienig wrote:

> >  You need to add "-ldl" to flags passed to the linker.  The fast hackish
> > way is to add it to the generated Makefile to the variable called EX_LIBS
> > after `./config' is run.  The real solution is to add an entry for
> > "NetBSD-pmax" (or "NetBSD-mipsel") to the perl "Configure" script in the
> > package.  The syntax is shown in the file and you may verify results with
> > `make TABLE'.  The only NetBSD hosts listed now are: "NetBSD-m68",
> > "NetBSD-sparc" and "NetBSD-x86" -- not an impressive list, I'd say.
> 
> Correct me if I'm wrong, but there is no libdl.{a,so} in NetBSD. So adding -ldl
> wouldn't help.

 Hmm, I'm not much familiar with NetBSD (I'm here due to a general
interest in the DECstation) but that's the usual name for the the library
holding dlopen() and friends on various ELF systems.  A few have the
functions in libc, but then you wouldn't get the failure.  My apologies
for providing unverified information. 

> My understanding of NetBSD is, that the kernel of the different platforms is the
> only thing, that is special to the platform (well I know, that this alone isn't
> realy possibile, but from a more general point of view). After that, every
> program is build with the same compile procedure. If this is true, how are the
> openssl-0.9.6e produced on the other supported platforms?

 Well, it's the CPU that is "special" to different platforms, ;-) so the
entries usually differ with the endianness setting, processor-specific
compilation options and processor-specific openssl options (a few
platforms provide hand-coded assembly functions; there are also a few
optimization choices for algorith modifications which work better or worse
depending on a CPU architecture).  Since neither of the existing entries
seem to fit NetBSD/pmax as is, I'd just copy the "NetBSD-m68" entry, since
it's the most generic, and change -DB_ENDIAN to -DL_ENDIAN.  You may want
to twiddle with the openssl optimization options as well.  I'm using
"BN_LLONG MD2_CHAR RC4_INDEX RC4_CHAR RC4_CHUNK DES_UNROLL DES_RISC2
DES_PTR" for my DECstations running Linux -- you might check if they yield
better performance. 

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +