linux-kernel - Re: [PATCH v1 2/2] perf evsel: for_each

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Tue, 23 May 2023 10:04:15 -0700
From:   Ian Rogers <irogers@...gle.com>
To:     Adrian Hunter <adrian.hunter@...el.com>
Cc:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Mark Rutland <mark.rutland@....com>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        Jiri Olsa <jolsa@...nel.org>,
        Namhyung Kim <namhyung@...nel.org>,
        Kan Liang <kan.liang@...ux.intel.com>,
        Sandipan Das <sandipan.das@....com>,
        James Clark <james.clark@....com>,
        Dmitrii Dolgov <9erthalion6@...il.com>,
        Changbin Du <changbin.du@...wei.com>,
        Rob Herring <robh@...nel.org>,
        Xing Zhengjun <zhengjun.xing@...ux.intel.com>,
        linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 2/2] perf evsel: for_each_group fixes

On Tue, May 23, 2023 at 7:33 AM Adrian Hunter <adrian.hunter@...el.com> wrote:
>
> On 23/05/23 07:44, Ian Rogers wrote:
> > Address/memory sanitizer was reporting issues in evsel__group_pmu_name
> > because the for_each_group_evsel loop didn't terminate when the head
> > was reached, the head would then be cast and accessed as an evsel
> > leading to invalid memory accesses. Fix for_each_group_member and
> > for_each_group_evsel to terminate at the list head. Note,
> > evsel__group_pmu_name no longer iterates the group, but the problem is
> > present regardless.
> >
> > Fixes: 717e263fc354 ("perf report: Show group description when event group is enabled")
> > Signed-off-by: Ian Rogers <irogers@...gle.com>
> > ---
> >  tools/perf/util/evsel.h         | 24 ++++++++++++++++--------
> >  tools/perf/util/evsel_fprintf.c |  1 +
> >  2 files changed, 17 insertions(+), 8 deletions(-)
> >
> > diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
> > index 820771a649b2..6a64543c7612 100644
> > --- a/tools/perf/util/evsel.h
> > +++ b/tools/perf/util/evsel.h
> > @@ -462,16 +462,24 @@ static inline int evsel__group_idx(struct evsel *evsel)
> >  }
> >
> >  /* Iterates group WITHOUT the leader. */
> > -#define for_each_group_member(_evsel, _leader)                                       \
> > -for ((_evsel) = list_entry((_leader)->core.node.next, struct evsel, core.node); \
> > -     (_evsel) && (_evsel)->core.leader == (&_leader->core);                                  \
> > -     (_evsel) = list_entry((_evsel)->core.node.next, struct evsel, core.node))
> > +#define for_each_group_member_head(_evsel, _leader, _head)                           \
> > +for ((_evsel) = list_entry((_leader)->core.node.next, struct evsel, core.node);              \
> > +     (_evsel) && (&(_evsel)->core.node != (_head)) &&                                \
>
> Extra parentheses perhaps not needed e.g. just
>
>         &(_evsel)->core.node != (_head)
>
> > +     (_evsel)->core.leader == (&_leader->core);                                      \
>
> Parentheses look odd, maybe should be:
>
>         &(_leader)->core
>
> > +     (_evsel) = list_entry((_evsel)->core.node.next, struct evsel, core.node))
> > +
> > +#define for_each_group_member(_evsel, _leader)                               \
> > +     for_each_group_member_head(_evsel, _leader, &(_leader)->evlist->core.entries)
>
> Did you consider using (_leader)->core.nr_members so that it is not
> necessary to assume the evlist back pointer is populated.

Thanks! I'll address the other comments in v2. Wrt
nr_members/evsel->evlist, I think we can use it to avoid the whole
list scan in evsel__compute_group_pmu_name. We could use it here but
in all the non-parse_events__sort_events_and_fix_groups cases we have
the evsel->evlist and testing for the list head seems more consistent
with other list_for_each cases.

Thanks,
Ian

> >
> >  /* Iterates group WITH the leader. */
> > -#define for_each_group_evsel(_evsel, _leader)                                        \
> > -for ((_evsel) = _leader;                                                     \
> > -     (_evsel) && (_evsel)->core.leader == (&_leader->core);                                  \
> > -     (_evsel) = list_entry((_evsel)->core.node.next, struct evsel, core.node))
> > +#define for_each_group_evsel_head(_evsel, _leader, _head)                            \
> > +for ((_evsel) = _leader;                                                             \
> > +     (_evsel) && (&(_evsel)->core.node != (_head)) &&                                \
> > +     (_evsel)->core.leader == (&_leader->core);                                      \
> > +     (_evsel) = list_entry((_evsel)->core.node.next, struct evsel, core.node))
> > +
> > +#define for_each_group_evsel(_evsel, _leader)                                \
> > +     for_each_group_evsel_head(_evsel, _leader, &(_leader)->evlist->core.entries)
> >
> >  static inline bool evsel__has_branch_callstack(const struct evsel *evsel)
> >  {
> > diff --git a/tools/perf/util/evsel_fprintf.c b/tools/perf/util/evsel_fprintf.c
> > index cc80ec554c0a..036a2171dc1c 100644
> > --- a/tools/perf/util/evsel_fprintf.c
> > +++ b/tools/perf/util/evsel_fprintf.c
> > @@ -2,6 +2,7 @@
> >  #include <inttypes.h>
> >  #include <stdio.h>
> >  #include <stdbool.h>
> > +#include "util/evlist.h"
> >  #include "evsel.h"
> >  #include "util/evsel_fprintf.h"
> >  #include "util/event.h"
>