Re: [PATCH 01/35] perf tools: Fix dso_id inode generation comparison

[Ian Rogers <irogers@google.com>]

On Tue, Jul 19, 2022 at 3:18 AM Adrian Hunter <adrian.hunter@intel.com> wrote:
>
> On 18/07/22 17:57, Arnaldo Carvalho de Melo wrote:
> > Em Mon, Jul 11, 2022 at 12:31:44PM +0300, Adrian Hunter escreveu:
> >> Synthesized MMAP events have zero ino_generation, so do not compare zero
> >> values.
> >>
> >> Fixes: 0e3149f86b99 ("perf dso: Move dso_id from 'struct map' to 'struct dso'")
> >> Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
> >> ---
> >>  tools/perf/util/dsos.c | 10 ++++++++--
> >>  1 file changed, 8 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/tools/perf/util/dsos.c b/tools/perf/util/dsos.c
> >> index b97366f77bbf..839a1f384733 100644
> >> --- a/tools/perf/util/dsos.c
> >> +++ b/tools/perf/util/dsos.c
> >> @@ -23,8 +23,14 @@ static int __dso_id__cmp(struct dso_id *a, struct dso_id *b)
> >>      if (a->ino > b->ino) return -1;
> >>      if (a->ino < b->ino) return 1;
> >>
> >> -    if (a->ino_generation > b->ino_generation) return -1;
> >> -    if (a->ino_generation < b->ino_generation) return 1;
> >> +    /*
> >> +     * Synthesized MMAP events have zero ino_generation, so do not compare
> >> +     * zero values.
> >> +     */
> >> +    if (a->ino_generation && b->ino_generation) {
> >> +            if (a->ino_generation > b->ino_generation) return -1;
> >> +            if (a->ino_generation < b->ino_generation) return 1;
> >> +    }
> >
> > But comparing didn't harm right? when its !0 now we may have three
> > comparisions instead of 2 :-\
> >
> > The comment has some value tho, so I'm merging this :-)
>
> Thanks. I found it harmful because the mismatch resulted in a new
> dso that did not have a build ID whereas the original dso did have
> a build ID.  The build ID was essential because the object was not
> found otherwise.

That's good to know, could we add that also to the comment? Perhaps:

Synthesized MMAP events have zero ino_generation, avoid comparing them
with MMAP events with actual ino_generation.

Thanks,
Ian