[Xcb] mem leak with ecore_xcb

Vincent Torri vtorri at univ-evry.fr
Wed Apr 11 02:45:35 PDT 2007


After having found a nasty bug, ecore_xcb behaves well. Nevertheless, when 
running a test through valgrind, I have those 2 messages when the prog 
shuts down:

==31763== 24 bytes in 1 blocks are still reachable in loss record 2 of 3
==31763==    at 0x401A6F0: malloc (vg_replace_malloc.c:149)
==31763==    by 0x4494E6C: XauFileName (AuFileName.c:67)
==31763== 936 bytes in 26 blocks are definitely lost in loss record 3 of 3
==31763==    at 0x401A6F0: malloc (vg_replace_malloc.c:149)
==31763==    by 0x42864D5: read_packet (xcb_in.c:144)

does someone see where the problem is ?

Also, when running over a ssh connection, the quantity of allocated memory 
and the number of malloc/free increase. Is it normal ?

thank you


Remainder: note that there is always that valgrind output:

==31763== 3 errors in context 1 of 1:
==31763== Syscall param writev(vector[...]) points to uninitialised 
==31763==    at 0x41F738B: writev (in /lib/tls/libc-2.3.3.so)
==31763==    by 0x4284C0A: _xcb_conn_wait (xcb_conn.c:167)
==31763==  Address 0x44E8AF9 is 4,353 bytes inside a block of size 8,552 
==31763==    at 0x4019A2F: calloc (vg_replace_malloc.c:279)
==31763==    by 0x42846CE: xcb_connect_to_fd (xcb_conn.c:221)

More information about the Xcb mailing list