Date: Mon, 18 Aug 1997 15:46:31 -0700 From: John Polstra <jdp@polstra.com> To: hackers@freebsd.org Subject: Re: [Fwd: Re: Please Help Me Understand dlopen()] Message-ID: <199708182246.PAA04812@austin.polstra.com> In-Reply-To: <33F869F3.446B9B3D@FreeBSD.org> References: <33F869F3.446B9B3D@FreeBSD.org>
index | next in thread | previous in thread | raw e-mail
> Hmmm - any comments, ld hackers?
I didn't see it in the newsgroup until Joerg brought it to my
attention today. (I don't follow the newsgroups, since I already
have all the FREE XXX SEX SITE PASSWORDS and FANTASTIC MEGA-INCOME
OPPORTUNITIES that I need at the moment.)
I agree that it's a bug, and I followed up there saying I planned to
fix it. One fix would be switching to ELF, of course ... ;-)
I think the fix should assume that the application most likely wants
to look up a normal C symbol. So it should do something like this:
/*
* First try: look for a normal C symbol like (in assembly
* language) "_foo".
*/
if (sym starts with "_") /* Assume caller already added "_" */
trysym = sym;
else /* Add "_" for caller */
trysym = concat("_", sym); /* Not C, but you get the idea */
if (lookup(trysym))
return success;
/*
* Second try: look for a strange C symbol like (in assembly
* language) "__foo", or an assembly language symbol "foo" with no
* underscore at all.
*/
if (sym starts with "_") /* Assume C symbol really starts with "_" */
trysym = concat("_", sym);
else /* Assume caller wants assembly symbol */
trysym = sym;
if (lookup(trysym))
return success;
/*
* Give up.
*/
return failure;
No matter what heuristic is used, it could do the wrong thing in
some cases.
John
--
John Polstra jdp@polstra.com
John D. Polstra & Co., Inc. Seattle, Washington USA
"Self-knowledge is always bad news." -- John Barth
help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199708182246.PAA04812>