Re: "/sys/conf/kmod.mk", line 111: Malformed conditional (${MK_CLANG_IS_CC}

On 2012-04-17 17:59, Arno J. Klaassen wrote:
...
> I verified and re-verfied : kernel and world perfectly in sync; /usr/src
> is a symlink but /sys points to the right place and (before patch)
> /usr/share/mk/bsd.own.mk and /usr/src/share/mk/bsd.own.mk are
> indentical.
>
> Still I have to apply something like the following patch to get
> VirtualBox kernel module compiled (no idea who/where MK_CLANG_IS_CC
> gets set to "no", AFAIK not by me (I don't use src.conf and nothing
> about CLANG in make.conf).

I was going to say that all the MK_XXX defines get set in bsd.own.mk,
but apparently there is an exception specifically for ports!  It has the
following fragment:

   [...]
   .if !defined(_WITHOUT_SRCCONF)
   #
   # Define MK_* variables (which are either "yes" or "no") for users
   # to set via WITH_*/WITHOUT_* in /etc/src.conf and override in the
   # make(1) environment.

while bsd.port.mk has:

   [...]
   # Needed to keep bsd.own.mk from reading in /etc/src.conf
   # and setting MK_* variables when building ports.
   _WITHOUT_SRCCONF=
   [...]
   .include <bsd.own.mk>
   .include "${BSDPORTMK}"

So those variables intentionally don't get set.  This is bit of a pity,
since particularly the information that '/usr/bin/cc' can be gcc or
clang (indicated by MK_CLANG_IS_CC) is rather relevant for ports.

Of course, that setting is normally configured in src.conf(5), which
emphatically does *not* apply to ports, but since it influences what
/usr/bin/cc will be, it is still of interest.  E.g. you can't detect
anymore whether someone is using clang by checking ${CC}...

I'm not sure what the best solution is in this case, I would really like
some feedback from ports people. :)