[OpenAFS-devel] Jaguar pam compilation
David Botsch
dwb7@ccmr.cornell.edu
Thu, 19 Sep 2002 12:19:22 -0400
Out of curiosity, what makes you think it should be a MH_BUNDLE ? Are
the stock pam modules compiled as MH_BUNDLE?
At the moment, with -bundle and either -fno_common, -flat_namespace or
neither of these latter two, I get the following error:
ld: /Users/rootdumb/Desktop/openafs-1.2.7-rc4/lib/liblwp.a(process.o)
has local relocation entries in non-writable section (__TEXT,__text)
make: *** [pam_afs.so.1] Error 1
now, if I leave it with -flat_namespace and also throw a
-read_only_relocs warning, I do get a generated .so file. I'll have to
test it now to see if it works.
On 2002.09.19 11:15 Chaskiel M Grundman wrote:
> --On Wednesday, September 18, 2002 14:22:00 -0400 David Botsch
> <dwb7@ccmr.cornell.edu> wrote:
>
>> (CC) -Xlinker -dylib $(LDFLAGS) -lpam -lpam_misc -o $@ afs_setcred.o
>> afs_auth. o afs_util.o $(SHOBJS) $(LIBS) ;;
>
> I would think that a pam module would need to be an MH_BUNDLE object
> (-Xlinker -bundle), not an MH_DYLIB module like you are doing.
>
> It may also be the case that because you are not telling gcc what you
> are doing (by using -Xlinker), it's doing something that is proper
> for executables but not for dylibs or bundles. Note that while they
> are not listed in the manpage, 10.2's gcc (and I think even older
> ones) accept -dynamiclib and -bundle switches directly.
>
> I suspect that you are *still* going to lose after making this
> change, since the afs code is not compiled with -fno-common (and
> enabling that will probably require lots of changes throughout the
> afs source), or -flat_namespace (I don't remember if -flat_namespace
> is really required or not. that's the hard part about this)
> _______________________________________________
> OpenAFS-devel mailing list
> OpenAFS-devel@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-devel
--
********************************
David William Botsch
Consultant/Advisor II
CCMR Computing Facility
dwb7@ccmr.cornell.edu
********************************