[PATCH] perf tools: check perf_event_paranoid and kptr_restrict on 'perf top'

[PATCH] perf tools: check perf_event_paranoid and kptr_restrict on 'perf top'

[Jackie Liu]

Perf top will segfault, we should give prompt information like perf
record instead of crashing directly.

Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
---
 tools/perf/builtin-top.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 3673c04d16b6..b257fadba3bd 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -1239,6 +1239,14 @@ static int __cmd_top(struct perf_top *top)
 			return ret;
 	}
 
+	if (symbol_conf.kptr_restrict && !evlist__exclude_kernel(top->evlist)) {
+		pr_warning(
+"Kernel address maps (/proc/{kallsyms,modules}) are restricted.\n\n"
+"Check /proc/sys/kernel/kptr_restrict and /proc/sys/kernel/perf_event_paranoid.\n\n"
+"Kernel samples will not be resolved.\n");
+		return -1;
+	}
+
 	ret = callchain_param__setup_sample_type(&callchain_param);
 	if (ret)
 		return ret;
-- 
2.25.1

Re: [PATCH] perf tools: check perf_event_paranoid and kptr_restrict on 'perf top'

[Jiri Olsa]

On Tue, Feb 23, 2021 at 02:24:40PM +0800, Jackie Liu wrote:
> Perf top will segfault, we should give prompt information like perf
> record instead of crashing directly.
> 
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
> Cc: Mark Rutland <mark.rutland@arm.com>
> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
> Cc: Jiri Olsa <jolsa@redhat.com>
> Cc: Namhyung Kim <namhyung@kernel.org>
> Signed-off-by: Jackie Liu <liuyun01@kylinos.cn>
> ---
>  tools/perf/builtin-top.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
> index 3673c04d16b6..b257fadba3bd 100644
> --- a/tools/perf/builtin-top.c
> +++ b/tools/perf/builtin-top.c
> @@ -1239,6 +1239,14 @@ static int __cmd_top(struct perf_top *top)
>  			return ret;
>  	}
>  
> +	if (symbol_conf.kptr_restrict && !evlist__exclude_kernel(top->evlist)) {
> +		pr_warning(
> +"Kernel address maps (/proc/{kallsyms,modules}) are restricted.\n\n"
> +"Check /proc/sys/kernel/kptr_restrict and /proc/sys/kernel/perf_event_paranoid.\n\n"
> +"Kernel samples will not be resolved.\n");
> +		return -1;
> +	}
> +

hm, why not fix that crash by checking prog_addrs != NULL before using it

  perf_event__synthesize_one_bpf_prog:
  ...
     __u64 *prog_addrs = (__u64 *)(uintptr_t)(info->jited_ksyms);

it looks like we're already checking on kptr_restrict in
perf_event__process_sample and allow it on some level

jirka

>  	ret = callchain_param__setup_sample_type(&callchain_param);
>  	if (ret)
>  		return ret;
> -- 
> 2.25.1
>