Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uwm.edu!wuarchive!cs.utexas.edu!bcm!rice!rice!sun-spots-request
From: auspex!guy@uunet.uu.net (Guy Harris)
Newsgroups: comp.sys.sun
Subject: Re: dlopen broken?
Keywords: Miscellaneous
Message-ID: <1990Aug9.022250.7092@rice.edu>
Date: 7 Aug 90 18:00:18 GMT
Sender: sun-spots-request@rice.edu
Organization: Sun-Spots
Lines: 28
Approved: Sun-Spots@rice.edu
Originator: spots@titan.rice.edu
X-Sun-Spots-Digest: Volume 9, Issue 296, message 10
X-Refs:  Original: v9n296

>Is the program interface to the dynamic linker (dlopen) broken?

If "dlopen()" drops core, it's pretty much by definition broken.
Unfortunately, I don't know what the fix is.  However, you are making what
may be another error in your sample code:

>     symbol = "_p2";
>     symAdr = dlsym(handle,symbol);
>     printf("returned symbol address of %s is %x\n",symbol,symAdr);
>     (*symAdr)();

Presumably, the function "_p2()" is defined as

	_p2()
	{
		...
	}

in the ".so" file you've loaded.  If it's "p2()", not "_p2()", you should
hand "p2", not "_p2", to "dlsym".  The loader may happen to let you get
away with sticking the underscore in front of it, but you're not supposed
to put it there, the fact that in Sun's C implementation on non-386-based
machines the compiler sticks an underscore there nonwithstanding.  Not all
C implementations put the underscore there, and if the second argument to
"dlsym()" were supposed to have the underscore there iff the C
implementation puts it there, portability would be blown out of the water
(remember, S5R4 has "dlopen()", too, and most S5R4 implementations
probably won't put the underscore there).