OK, I am making a little progress. I got hold of the dmalloc package for
debugging memory (allocation) problems. BTW, has anyone tried using dmalloc
with nessus?
1. When I run nessusd in the i386 system with the dmalloc "dmallocth"
library, i get lots of information (boy is it slow processing) about a large
number of relatively small memory areas which were allocated but never freed
(small is around 25 bytes). However, nessusd does work properly. Now when I
say large I mean around 29,000. I plan to look into this by maodifying
nessusd to more tightly couple with dmalloc so that I can see where stuff is
being allocated.
2. I finally managed to hack dmalloc/nessud so that I could use dmalloc. I
never get far enough to see if there are a large number of small unfreed
chunks of memory. Instead, dmalloc starts complaining that it is unable to
extend the heap. Now on the x86_64 pointers are 8 bytes rather than four and
thus the sizes of some allocated memeory may be much larger and fill the
heap. I am going to try and change some of the malloc parameters to get what
happens to change.
3. BTW, can someone explain why, when I run nessusd under a debugger such as
gdb, it does not run many of the tests unless I select "all but dangerous"
plugins?
Gene
--
debugging memory (allocation) problems. BTW, has anyone tried using dmalloc
with nessus?
1. When I run nessusd in the i386 system with the dmalloc "dmallocth"
library, i get lots of information (boy is it slow processing) about a large
number of relatively small memory areas which were allocated but never freed
(small is around 25 bytes). However, nessusd does work properly. Now when I
say large I mean around 29,000. I plan to look into this by maodifying
nessusd to more tightly couple with dmalloc so that I can see where stuff is
being allocated.
2. I finally managed to hack dmalloc/nessud so that I could use dmalloc. I
never get far enough to see if there are a large number of small unfreed
chunks of memory. Instead, dmalloc starts complaining that it is unable to
extend the heap. Now on the x86_64 pointers are 8 bytes rather than four and
thus the sizes of some allocated memeory may be much larger and fill the
heap. I am going to try and change some of the malloc parameters to get what
happens to change.
3. BTW, can someone explain why, when I run nessusd under a debugger such as
gdb, it does not run many of the tests unless I select "all but dangerous"
plugins?
Gene
--