linux-kernel - Re: [PATCH v2] perf trace: avoid garbage when not printing a trace event's arguments

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]

Message-ID: <CAH0uvohNO517GAD7XH-VvUvU+dPdeZBKDw0Teij50sAPSe9sCQ@mail.gmail.com>
Date: Fri, 1 Nov 2024 14:00:46 -0700
From: Howard Chu <howardchu95@...il.com>
To: Benjamin Peterson <benjamin@...flow.com>
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>, 
	"open list:PERFORMANCE EVENTS SUBSYSTEM" <linux-perf-users@...r.kernel.org>, 
	"open list:PERFORMANCE EVENTS SUBSYSTEM" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] perf trace: avoid garbage when not printing a trace
 event's arguments

Hello Benjamin,

Before your patch:

perf $ ./perf trace -e net:netif_rx_exit
     0.000 irq/79-brcmf_p/1694977 net:netif_rx_exit(6n<)
    28.153 irq/79-brcmf_p/1694977 net:netif_rx_exit(6n<)
    36.429 irq/79-brcmf_p/1694977 net:netif_rx_exit(6n<)
    36.461 irq/79-brcmf_p/1694977 net:netif_rx_exit(6n<)

After:

perf $ ./perf trace -e net:netif_rx_exit
     0.000 irq/79-brcmf_p/1694977 net:netif_rx_exit()
     7.352 irq/79-brcmf_p/1694977 net:netif_rx_exit()
    30.232 irq/79-brcmf_p/1694977 net:netif_rx_exit()
    37.529 irq/79-brcmf_p/1694977 net:netif_rx_exit()

It works beautifully, but I'm thinking can we simplify it by just doing:

+       char bf[2048] = { 0 };
        size_t size = sizeof(bf);

That being said, trace__fprintf_tp_fields() should return void because
currently its return value is not used by any other functions, plus
this line:

return printed + fprintf(trace->output, "%s", bf);

in trace__fprintf_tp_fields() does not make much sense, since that
will be 2 * printed. So I'm thinking maybe:

diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index 748b061f8678..8a628d1e80f3 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -3021,10 +3021,10 @@ static void bpf_output__fprintf(struct trace *trace,
        ++trace->nr_events_printed;
 }

-static size_t trace__fprintf_tp_fields(struct trace *trace, struct
evsel *evsel, struct perf_sample *sample,
+static void trace__fprintf_tp_fields(struct trace *trace, struct
evsel *evsel, struct perf_sample *sample,
                                       struct thread *thread, void
*augmented_args, int augmented_args_size)
 {
-       char bf[2048];
+       char bf[2048] = { 0 };
        size_t size = sizeof(bf);
        struct tep_format_field *field = evsel->tp_format->format.fields;
        struct syscall_arg_fmt *arg = __evsel__syscall_arg_fmt(evsel);
@@ -3087,7 +3087,7 @@ static size_t trace__fprintf_tp_fields(struct
trace *trace, struct evsel *evsel,
                printed += syscall_arg_fmt__scnprintf_val(arg, bf +
printed, size - printed, &syscall_arg, val);
        }

-       return printed + fprintf(trace->output, "%s", bf);
+       fprintf(trace->output, "%s", bf);
 }

 static int trace__event_handler(struct trace *trace, struct evsel *evsel,


Then this will be one more diff line than yours.

But that's just my opinion, yours works just fine, thank you :).
Arnaldo, what do you think?

Tested-by: Howard Chu <howardchu95@...il.com>

Thanks,
Howard

On Fri, Nov 1, 2024 at 10:27 AM Benjamin Peterson <benjamin@...flow.com> wrote:
>
> trace__fprintf_tp_fields may not print any tracepoint arguments. E.g., if the
> argument values are all zero. Previously, this would result in a totally
> uninitialized buffer being passed to fprintf, which could lead to garbage on the
> console. Fix the problem by passing the number of initialized bytes fprintf.
>
> Remove the return value of trace__fprintf_tp_fields, since it was meaningless
> and ignored.
>
> Fixes: f11b2803bb88 ("perf trace: Allow choosing how to augment the tracepoint arguments")
> Signed-off-by: Benjamin Peterson <benjamin@...flow.com>
> ---
>  tools/perf/builtin-trace.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
> index d3f11b90d025..4e785ea29df6 100644
> --- a/tools/perf/builtin-trace.c
> +++ b/tools/perf/builtin-trace.c
> @@ -3021,7 +3021,7 @@ static void bpf_output__fprintf(struct trace *trace,
>         ++trace->nr_events_printed;
>  }
>
> -static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel, struct perf_sample *sample,
> +static void trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel, struct perf_sample *sample,
>                                        struct thread *thread, void *augmented_args, int augmented_args_size)
>  {
>         char bf[2048];
> @@ -3087,7 +3087,7 @@ static size_t trace__fprintf_tp_fields(struct trace *trace, struct evsel *evsel,
>                 printed += syscall_arg_fmt__scnprintf_val(arg, bf + printed, size - printed, &syscall_arg, val);
>         }
>
> -       return printed + fprintf(trace->output, "%s", bf);
> +       fprintf(trace->output, "%.*s", (int)printed, bf);
>  }
>
>  static int trace__event_handler(struct trace *trace, struct evsel *evsel,
> --
> 2.39.5
>