This is the mail archive of the libc-hacker@sources.redhat.com mailing list for the glibc project.

Note that libc-hacker is a closed list. You may look at the archives of this list, but subscription and posting are not open.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

ld.so changes


I've checked in some changes which make ld.so pass the additional
tests I checked in yesterday.  The changes are minimal in extend, this
was a major design aspect.  There are other ways to solve the problem
but now seem to be cheaper.

If you look at the code you'll see that yet another list is now
allocated dynamically, if necessary.  This is unfortunately
unavoidable.  The effect should still be minimal: the dynamic
allocation should be necessary only in rare cases since it's done only
for dynamically loaded objects.  This might not exactly be what
Solaris does but at this point I don't care.  This is so very much
esoteric that it's not worth caring.  For reference: what doesn't work
is if a DSO, which is a dependency of the application, is incomplete
and will need a definition from an object which is a dependency of a
dynamically loaded object which has the first (incomplete) DSO as a
dependency.  See, it's not even possible to explain this easy enough
to understand.


Anyhow, the test suite is passed and I don't expect major problems.
But I'll do more testing tomorrow.  There are a few more boundary
cases to be tested and I also have to check for memory leaks.  It
would be good to get some testing with programs which heavily depend
on dynamic loading.

-- 
---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]