Re: [PATCH 2/2] perf report: Fix memory leaks in the hierarchy mode

[Namhyung Kim <namhyung@kernel.org>]

On Thu, Mar 06, 2025 at 09:19:09PM -0800, Namhyung Kim wrote:
> On Thu, Mar 06, 2025 at 08:50:46PM -0800, Ian Rogers wrote:
> > On Thu, Mar 6, 2025 at 6:01 PM Namhyung Kim <namhyung@kernel.org> wrote:
> > >
> > > On Thu, Mar 06, 2025 at 09:07:00AM -0800, Namhyung Kim wrote:
> > > > Hi Ian,
> > > >
> > > > On Thu, Mar 06, 2025 at 08:55:05AM -0800, Ian Rogers wrote:
> > > > > On Wed, Mar 5, 2025 at 11:51 PM Namhyung Kim <namhyung@kernel.org> wrote:
> > > > > >
> > > > > > Ian told me that there are many memory leaks in the hierarchy mode.  I
> > > > > > can easily reproduce it with the follwing command.
> > > > > >
> > > > > >   $ make DEBUG=1 EXTRA_CFLAGS=-fsanitize=leak
> > > > > >
> > > > > >   $ perf record --latency -g -- ./perf test -w thloop
> > > > > >
> > > > > >   $ perf report -H --stdio
> > > > > >   ...
> > > > > >   Indirect leak of 168 byte(s) in 21 object(s) allocated from:
> > > > > >       #0 0x7f3414c16c65 in malloc ../../../../src/libsanitizer/lsan/lsan_interceptors.cpp:75
> > > > > >       #1 0x55ed3602346e in map__get util/map.h:189
> > > > > >       #2 0x55ed36024cc4 in hist_entry__init util/hist.c:476
> > > > > >       #3 0x55ed36025208 in hist_entry__new util/hist.c:588
> > > > > >       #4 0x55ed36027c05 in hierarchy_insert_entry util/hist.c:1587
> > > > > >       #5 0x55ed36027e2e in hists__hierarchy_insert_entry util/hist.c:1638
> > > > > >       #6 0x55ed36027fa4 in hists__collapse_insert_entry util/hist.c:1685
> > > > > >       #7 0x55ed360283e8 in hists__collapse_resort util/hist.c:1776
> > > > > >       #8 0x55ed35de0323 in report__collapse_hists /home/namhyung/project/linux/tools/perf/builtin-report.c:735
> > > > > >       #9 0x55ed35de15b4 in __cmd_report /home/namhyung/project/linux/tools/perf/builtin-report.c:1119
> > > > > >       #10 0x55ed35de43dc in cmd_report /home/namhyung/project/linux/tools/perf/builtin-report.c:1867
> > > > > >       #11 0x55ed35e66767 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:351
> > > > > >       #12 0x55ed35e66a0e in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:404
> > > > > >       #13 0x55ed35e66b67 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:448
> > > > > >       #14 0x55ed35e66eb0 in main /home/namhyung/project/linux/tools/perf/perf.c:556
> > > > > >       #15 0x7f340ac33d67 in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58
> > > > > >   ...
> > > > > >
> > > > > >   $ perf report -H --stdio 2>&1 | grep -c '^Indirect leak'
> > > > > >   93
> > > > > >
> > > > > > I found that hist_entry__delete() missed to release child entries in the
> > > > > > hierarchy tree (hroot_{in,out}).  It needs to iterate the child entries
> > > > > > and call hist_entry__delete() recursively.
> > > > > >
> > > > > > After this change:
> > > > > >
> > > > > >   $ perf report -H --stdio 2>&1 | grep -c '^Indirect leak'
> > > > > >   0
> > > > > >
> > > > > > Reported-by: Ian Rogers <irogers@google.com>
> > > > > > Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> > > > > > ---
> > > > > >  tools/perf/util/hist.c | 9 +++++++++
> > > > > >  1 file changed, 9 insertions(+)
> > > > > >
> > > > > > diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
> > > > > > index fbf131aeae7ffe9b..bbc6a299b5106c3b 100644
> > > > > > --- a/tools/perf/util/hist.c
> > > > > > +++ b/tools/perf/util/hist.c
> > > > > > @@ -1385,6 +1385,15 @@ void hist_entry__delete(struct hist_entry *he)
> > > > > >  {
> > > > > >         struct hist_entry_ops *ops = he->ops;
> > > > > >
> > > > > > +       while (!RB_EMPTY_ROOT(&he->hroot_out.rb_root)) {
> > > > > > +               struct rb_node *node = rb_first(&he->hroot_out.rb_root);
> > > > > > +               struct hist_entry *child = rb_entry(node, struct hist_entry, rb_node);
> > > > > > +
> > > > > > +               rb_erase_init(node, &he->hroot_out.rb_root);
> > > > > > +
> > > > > > +               hist_entry__delete(child);
> > > > > > +       }
> > > > >
> > > > > Thanks for the fix! A nit, iterating the rbtree of N nodes and calling
> > > > > erase on the first entry, an O(log N) operation, means this is a O(N *
> > > > > log N). rbtree.h has rbtree_postorder_for_each_entry_safe:
> > > > > https://web.git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/include/linux/rbtree.h?h=perf-tools-next#n81
> > > > > ```
> > > > >  * rbtree_postorder_for_each_entry_safe - iterate in post-order over rb_root of
> > > > >  * given type allowing the backing memory of @pos to be invalidated
> > > > > ```
> > > > > which is O(N). I think this code would be better something like:
> > > > > ```
> > > > > struct hist_entry *pos, *tmp;
> > > > > rbtree_postorder_for_each_entry_safe(pos, tmp, he->hroot_out.rb_root, rb_node)
> > > > >         hist_entry__delete(pos);
> > > > > ```
> > > >
> > > > Thanks for your review!  I was wondering if there's something like this.
> > > > Will update with that.
> > >
> > > I found that the following comments:
> > >
> > >  * Note, however, that it cannot handle other modifications that re-order the
> > >  * rbtree it is iterating over. This includes calling rb_erase() on @pos, as
> > >  * rb_erase() may rebalance the tree, causing us to miss some nodes.
> > >
> > > So I cannot use this here.
> > 
> > I thought the tree is no longer in use after the hist_entry__delete,
> > so you don't need to maintain the child/parent pointers as everything
> > is going to get freed. Here is what I think is a similar use:
> > https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/trace/trace_stat.c#n51
> 
> Oh, I see.  So it should not call rb_erase().  Will check again.

It works!  Will send v2 soon.

Thanks,
Namhyung