[<prev] [next>] [day] [month] [year] [list]
Message-ID: <17bdc644-329a-42fb-aba2-d7f80f2a1037@arm.com>
Date: Mon, 11 Aug 2025 12:33:58 +0530
From: Anshuman Khandual <anshuman.khandual@....com>
To: Yujun Dong <yujundong@...cal-lab.net>, linux-perf-users@...r.kernel.org
Cc: Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...hat.com>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Namhyung Kim <namhyung@...nel.org>, Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...nel.org>, Ian Rogers <irogers@...gle.com>,
Adrian Hunter <adrian.hunter@...el.com>,
"Liang, Kan" <kan.liang@...ux.intel.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] perf branch: Fix heap out-of-bounds write in
branch_type_count()
On 09/08/25 3:08 PM, Yujun Dong wrote:
> The branch_type_count() function writes to st->new_counts[flags->new_type]
> when flags->type is PERF_BR_EXTEND_ABI. However, while the array
> st->new_counts is sized for PERF_BR_NEW_MAX (8) entries, the field
> flags->new_type is a 4-bit unsigned value and may hold values up to 15.
>
> This mismatch allows crafted perf data to trigger a heap out-of-bounds
> write when flags->new_type >= 8, leading to memory corruption.
Crafted ? How could flags->new_type >= 8 when PERF_BR_NEW_MAX is capped at 8.
Is this a real scenario that happened on a system ?
>
> Add a bounds check to ensure flags->new_type is less than
> PERF_BR_NEW_MAX before accessing the new_counts array.
But it might make sense to add this check just to be on the safer side.
>
> Fixes: 0ddea8e2a0c2 ("perf branch: Extend branch type classification")
> Signed-off-by: Yujun Dong <yujundong@...cal-lab.net>
> ---
> tools/perf/util/branch.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/tools/perf/util/branch.c b/tools/perf/util/branch.c
> index 3712be067464..8ea6628c7735 100644
> --- a/tools/perf/util/branch.c
> +++ b/tools/perf/util/branch.c
> @@ -21,10 +21,12 @@ void branch_type_count(struct branch_type_stat *st, struct branch_flags *flags,
> if (flags->type == PERF_BR_UNKNOWN || from == 0)
> return;
>
> - if (flags->type == PERF_BR_EXTEND_ABI)
> - st->new_counts[flags->new_type]++;
> - else
> + if (flags->type == PERF_BR_EXTEND_ABI) {
> + if (flags->new_type < PERF_BR_NEW_MAX)
> + st->new_counts[flags->new_type]++;
> + } else {
> st->counts[flags->type]++;
> + }
>
> if (flags->type == PERF_BR_COND) {
> if (to > from)
Powered by blists - more mailing lists