On Fri, Feb 19, 2010 at 2:28 AM, Humberto Abdelnur <
[email protected]> wrote:
> if (!cp || dash != '-' || colon != ':')
>
> to
>
> if (dash != '-' || colon != ':')
>
> do you agreed?
>
Agreed. I'll apply this patch.
> I found 3 leaks, the biggest is in the
> src/ptrace/_UPT_find_proc_info.c:get_unwind_info() function as you
> mentioned,
>
I looked at this some more. The unw_cursor_t keeps track of the previous elf
image mapped. Before mapping a new image, it unmaps the previous image, so
there is no leak.
But in poc.c:
unw_init_remote(&orig_cursor, as, upt);
curr_cursor = orig_cursor;
This assignment makes the cursor lose track of what was previously mapped.
I'll let David comment on if a separate API call is needed to cleanup the
cursor or the implementation could unmap the last mmap when the end of call
chain is reached.
> and there are two other minor leaks in the
> src/dwarf/global.c:dwarf_init(), as I said before, these mempools are never
> unmapped.
>
>
These are by design. I'm not sure you can do much about it, other than
unmapping the memory when libunwind is unloaded. This shouldn't be a large
chunk of memory and more importantly, it's a one time allocation needed for
libunwind to function and hence not a leak.
-Arun
_______________________________________________
Libunwind-devel mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/libunwind-devel
