On Tue, Sep 6, 2016 at 1:43 PM, William Roberts <bill.c.roberts@xxxxxxxxx> wrote: > <snip> >>>>> >>>>> Also, there are some memory leaks in there; run it under valgrind, e.g. >>>>> valgrind --leak-check=full matchpathcon /etc >>>> >>>> OK I'll run that test. >> >> I cant reproduce: > bad send... Can you send your valgrind output? Are you sure its not there > prior to my patch? The only heap alloc I add is the strdup, closef should > always be called despite any flow changes in process_file() as the caller, > init() always goes to label finish. Ok I see the condition now, I had to statically link in libselinux to matchpathcon to get it to not use the system one, even though I set LD_LIBRARY_PATH. Once I did that, I got the leaks. The condition on which it occurs, is because process_file() can be called N times and reseats the data->path pointer. This should have been obvious since mmap area is a linked list. > > > -- > Respectfully, > > William C Roberts -- Respectfully, William C Roberts _______________________________________________ Selinux mailing list Selinux@xxxxxxxxxxxxx To unsubscribe, send email to Selinux-leave@xxxxxxxxxxxxx. To get help, send an email containing "help" to Selinux-request@xxxxxxxxxxxxx.
