Re: null ptr deref in try_dlopen

[Gary V. Vaughan]

Ralf Wildenhues wrote:
> Actually, while looking at try_dlopen, a couple of questions have come
> up (don't worry about these for the next alpha or so, I'm sorry for the
> unfortunate timing).

No problem.  If I can get my fortran compiler to work, I'm planning on 
branching for the next release later today, so we can limit ourselves to 
fixes only on the release branch.

> - The module name -- does it (should it) contain version information?
>   If yes, how much, and how encoded?

At the moment only the path to the module is used to identify it.  And
that's all thats needed I think.  It would be perfectly possible to 
build a version checking layer above lt_dlopenext using 
lt_dlsym("version") or the like, but I think libltdl is low enough level
that it needn't enforce a particular scheme.  Maybe we could implement 
an optional recommended scheme in a future release?

> - If yes, shoud it be possible for a program (a library) to be able
>   to load two different versions of the same module simultaneously?

Yes, they will have different path names, and libltdl doesn't care about
anything else.

> - Would you eventually accept a rewrite of try_dlopen?

Absolutely, the current implementation is horrid.  I've been picking 
away at the edges of libltdl for a year or so now in the hope of getting
to the point where I can actually follow what the hell is going on in
try_dlopen to tidy it up a bit (a lot!).

> BTW, I have not forgotten about the Autotest stuff, I'll send something
> eventually.

Okay, thanks :-)

Cheers,
        Gary.
--
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature