[prog] ncurses memory leak?

ed orphan millward at Ms.UManitoba.CA
Sun May 30 12:39:40 EST 2004


I've been using valgrind to track down possible
memory leaks in one of my programs.  It's been
suggested by some on this site that the reported
leaks ( "still reachable 35,912 bytes in 158 blocks")
are caused by global pointers, but I think not.
My program is a simple little linked-list and I've
freed and NULLed every pointer I can find.
After compiling my program, px4.c :
gcc -I/usr/include  -Wall -W -g -o px4 px4.c -lncurses -lform
    and it compiles clean, 
I run it using valgind to find memory leaks:
  valgrind --verbose --leak-check=yes --show-reachable=yes px4
The messages scroll accross my screen real quick, and
I only see one screen of them on the command line, including the summary.
I should have checked sooner, but not until yesterday did I
run valgrind in Gnome terminal and saved all the diagnostics to a text file.
Almost all of them point to /usr/lib/libncurses.so.5.2
For example, this message from valgrind :
==2011== 2857 bytes in 3 blocks are still reachable in loss record
               21 of 22
==2011==    at 0x40026AB0: malloc (vg_replace_malloc.c:153)
==2011==    by 0x40267E08: _nc_doalloc (in /usr/lib/libncurses.so.5.2)
==2011==    by 0x4026A462: tparm (in /usr/lib/libncurses.so.5.2)
==2011==    by 0x40255B66: _nc_mvcur_init (in /usr/lib/libncurses.so.5.2)
   I could be very wrong, but could the memory leaks be in the
ncurses library?  If so, do you think this is serious enough to
contact them?  


   




More information about the Programming mailing list