This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
glibc-2.2.5 bug triggered by Java interpreter
- From: "Ulrich Weigand" <Ulrich dot Weigand at de dot ibm dot com>
- To: java at gcc dot gnu dot org
- Cc: gcc at gcc dot gnu dot org
- Date: Wed, 2 Oct 2002 17:03:55 +0200
- Subject: glibc-2.2.5 bug triggered by Java interpreter
- Sensitivity:
Hello,
while trying to get the Java interpreter running, I was side-tracked by
a bug in the vanilla glibc 2.2.5, which happens to be triggered by the
gij code.
What gij does is to use dlopen on file names that contain '$' signs,
like dlopen ("lib-PR242$Inner", ...). This causes a bug in a subroutine
of dlopen to trigger; the routine _dl_dst_count goes into an endless
loop and dlopen never returns.
While this bug is already fixed in the current glibc CVS head, the
question remains why gij is doing that; seeing as '$' in the library
name has some special semantics for dlopen even with that bug fixed
(it can trigger some keyword replacement AFAIK), couldn't this have
unexpected side effect if a class has just the right name to trigger
one of those keyword checks?
Mit freundlichen Gruessen / Best Regards
Ulrich Weigand
--
Dr. Ulrich Weigand
Linux for S/390 Design & Development
IBM Deutschland Entwicklung GmbH, Schoenaicher Str. 220, 71032 Boeblingen
Phone: +49-7031/16-3727 --- Email: Ulrich.Weigand@de.ibm.com