Re: Explicit LOAD and dynamic library loading

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Stephen Scheck <singularsyntax(at)gmail(dot)com>
Cc: pgsql-general <pgsql-general(at)postgresql(dot)org>
Subject: Re: Explicit LOAD and dynamic library loading
Date: 2013-06-13 14:38:10
Message-ID: 22065.1371134290@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Stephen Scheck <singularsyntax(at)gmail(dot)com> writes:
> [postgres(at)dev1 lib]$ ldd bar.so
> linux-vdso.so.1 => (0x00007fff1c7ff000)
> libc.so.6 => /lib64/libc.so.6 (0x00007fa4c96ac000)
> /lib64/ld-linux-x86-64.so.2 (0x00007fa4c9c5d000)

So you're missing any reference to foo.so; not surprising it fails.
You need to make sure that "-lfoo" or something similar gets into the
link command for bar.so. You might still have some rpath issues after
that, but right now the runtime linker doesn't even know it should be
loading foo.so.

> After digging into the PGXS docs a bit more, I think if I change the
> Makefile a bit this may work:

> MODULES = foo bar
> MODULE_big = bar
> OBJS = bar.o
> SHLIB_LINK = foo.o

Oh, you're trying to build two separate shlibs in one Makefile? I don't
think that our Makefile infrastructure is smart enough to handle that,
at least not if cross-references are required. You'd probably be well
advised to split them into two separate source-code directories.

Another question worth asking yourself, if you are building like this,
is why you're bothering with two .so's at all. Is there a strong reason
not to just make them into one library?

regards, tom lane

In response to

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Stephen Scheck 2013-06-13 18:26:35 Re: Explicit LOAD and dynamic library loading
Previous Message Tom Lane 2013-06-13 13:59:34 Re: Why hash join cost calculation need reduction