[Valgrind-users] Dumping the content of a lost memory block

[Alexander F. <ale...@gm...>]

Hi,

I'm using valgrind 2.4.0 on RH9 Linux (thank you!) and it keeps=20
reporting, that I have 1 memory leak in my C-program:

=3D=3D17058=3D=3D 260 bytes in 1 blocks are definitely lost in loss record =
1 of 2
=3D=3D17058=3D=3D    at 0x1B90359D: malloc (vg_replace_malloc.c:130)
=3D=3D17058=3D=3D    by 0x80497FE: new_node (pref.c:514)
=3D=3D17058=3D=3D    by 0x80499A2: copy_node (pref.c:598)
=3D=3D17058=3D=3D    by 0x8049E45: accept_new (pref.c:739)
=3D=3D17058=3D=3D    by 0x8048FE9: main (pref.c:299)
=3D=3D17058=3D=3D=20
=3D=3D17058=3D=3D LEAK SUMMARY:
=3D=3D17058=3D=3D    definitely lost: 260 bytes in 1 blocks.
=3D=3D17058=3D=3D      possibly lost: 0 bytes in 0 blocks.
=3D=3D17058=3D=3D    still reachable: 4420 bytes in 17 blocks.
=3D=3D17058=3D=3D         suppressed: 0 bytes in 0 blocks.

I'm trying already since few days, but still can't find that leak.
In the new_node() function mentioned above I malloc()=20
a structure (about 20 times) and keep the track to it in a=20
singly-linked tail queue.

Since valgrind probably keeps track to all malloc()ed areas,
is it possible to make it print the content of that lost area?
Then I could identify, which particular node of the list gets lost.

I've tried calling my program with "valgrind --leak-check=3Dyes=20
--db-attach=3Dyes ./myprog", but it doesn't ask to attach...

Regards
Alex