Re: [PATCH 1/3] perf record: Fix per-thread option.

[Alexey Bayduraev]

On 12.04.2022 09:21, Ian Rogers wrote:

> From: Alexey Bayduraev <alexey.bayduraev@xxxxxxxxx>

Thanks,

The tag
Signed-off-by: Alexey Bayduraev <alexey.bayduraev@xxxxxxxxx>
can be added to this patch.

Regards,
Alexey

> 
> Per-thread mode doesn't have specific CPUs for events, add checks for
> this case.
> 
> Minor fix to a pr_debug by Ian Rogers <irogers@xxxxxxxxxx> to avoid an
> out of bound array access.
> 
> Reported-by: Ian Rogers <irogers@xxxxxxxxxx>
> Fixes: 7954f71689f9 ("perf record: Introduce thread affinity and mmap masks")
> Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
> ---
>  tools/perf/builtin-record.c | 22 +++++++++++++++++-----
>  1 file changed, 17 insertions(+), 5 deletions(-)
> 
> diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
> index ba74fab02e62..069825c48d40 100644
> --- a/tools/perf/builtin-record.c
> +++ b/tools/perf/builtin-record.c
> @@ -989,8 +989,11 @@ static int record__thread_data_init_maps(struct record_thread *thread_data, stru
>  	struct mmap *overwrite_mmap = evlist->overwrite_mmap;
>  	struct perf_cpu_map *cpus = evlist->core.user_requested_cpus;
>  
> -	thread_data->nr_mmaps = bitmap_weight(thread_data->mask->maps.bits,
> -					      thread_data->mask->maps.nbits);
> +	if (cpu_map__is_dummy(cpus))
> +		thread_data->nr_mmaps = nr_mmaps;
> +	else
> +		thread_data->nr_mmaps = bitmap_weight(thread_data->mask->maps.bits,
> +						      thread_data->mask->maps.nbits);
>  	if (mmap) {
>  		thread_data->maps = zalloc(thread_data->nr_mmaps * sizeof(struct mmap *));
>  		if (!thread_data->maps)
> @@ -1007,16 +1010,17 @@ static int record__thread_data_init_maps(struct record_thread *thread_data, stru
>  		 thread_data->nr_mmaps, thread_data->maps, thread_data->overwrite_maps);
>  
>  	for (m = 0, tm = 0; m < nr_mmaps && tm < thread_data->nr_mmaps; m++) {
> -		if (test_bit(cpus->map[m].cpu, thread_data->mask->maps.bits)) {
> +		if (cpu_map__is_dummy(cpus) ||
> +		    test_bit(cpus->map[m].cpu, thread_data->mask->maps.bits)) {
>  			if (thread_data->maps) {
>  				thread_data->maps[tm] = &mmap[m];
>  				pr_debug2("thread_data[%p]: cpu%d: maps[%d] -> mmap[%d]\n",
> -					  thread_data, cpus->map[m].cpu, tm, m);
> +					  thread_data, perf_cpu_map__cpu(cpus, m).cpu, tm, m);
>  			}
>  			if (thread_data->overwrite_maps) {
>  				thread_data->overwrite_maps[tm] = &overwrite_mmap[m];
>  				pr_debug2("thread_data[%p]: cpu%d: ow_maps[%d] -> ow_mmap[%d]\n",
> -					  thread_data, cpus->map[m].cpu, tm, m);
> +					  thread_data, perf_cpu_map__cpu(cpus, m).cpu, tm, m);
>  			}
>  			tm++;
>  		}
> @@ -3329,6 +3333,9 @@ static void record__mmap_cpu_mask_init(struct mmap_cpu_mask *mask, struct perf_c
>  {
>  	int c;
>  
> +	if (cpu_map__is_dummy(cpus))
> +		return;
> +
>  	for (c = 0; c < cpus->nr; c++)
>  		set_bit(cpus->map[c].cpu, mask->bits);
>  }
> @@ -3680,6 +3687,11 @@ static int record__init_thread_masks(struct record *rec)
>  	if (!record__threads_enabled(rec))
>  		return record__init_thread_default_masks(rec, cpus);
>  
> +	if (cpu_map__is_dummy(cpus)) {
> +		pr_err("--per-thread option is mutually exclusive to parallel streaming mode.\n");
> +		return -EINVAL;
> +	}
> +
>  	switch (rec->opts.threads_spec) {
>  	case THREAD_SPEC__CPU:
>  		ret = record__init_thread_cpu_masks(rec, cpus);
>