Re: [PATCH v1] Increase MAX_NR_CPUS and MAX_CACHES

From: Arnaldo Carvalho de Melo
Date: Mon Jun 24 2019 - 14:51:16 EST


Em Thu, Jun 20, 2019 at 02:36:30PM -0500, Kyle Meyer escreveu:
> From: Kyle Meyer <kyle.meyer@xxxxxxx>
>
> Attempting to profile 1024 or more CPUs with perf causes two errors:
>
> perf record -a
> [ perf record: Woken up X times to write data ]
> way too many cpu caches..
> [ perf record: Captured and wrote X MB perf.data (X samples) ]
>
> perf report -C 1024
> Error: failed to set cpu bitmap
> Requested CPU 1024 too large. Consider raising MAX_NR_CPUS
>
> Increasing MAX_NR_CPUS from 1024 to 2048 and redefining MAX_CACHES as
> MAX_NR_CPUS * 4 returns normal functionality to perf:
>
> perf record -a
> [ perf record: Woken up X times to write data ]
> [ perf record: Captured and wrote X MB perf.data (X samples) ]
>
> perf report -C 1024

So, I'm applying the tools/perf/ part, leaving the rest for Daniel do
consider, ok?

- Arnaldo

> ...
>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Signed-off-by: Kyle Meyer <kyle.meyer@xxxxxxx>
> ---
> samples/bpf/map_perf_test_kern.c | 2 +-
> samples/bpf/map_perf_test_user.c | 2 +-
> tools/perf/perf.h | 2 +-
> tools/perf/util/header.c | 2 +-
> 4 files changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/samples/bpf/map_perf_test_kern.c b/samples/bpf/map_perf_test_kern.c
> index 2b2ffb97018b..342738a1e386 100644
> --- a/samples/bpf/map_perf_test_kern.c
> +++ b/samples/bpf/map_perf_test_kern.c
> @@ -11,7 +11,7 @@
> #include "bpf_helpers.h"
>
> #define MAX_ENTRIES 1000
> -#define MAX_NR_CPUS 1024
> +#define MAX_NR_CPUS 2048
>
> struct bpf_map_def SEC("maps") hash_map = {
> .type = BPF_MAP_TYPE_HASH,
> diff --git a/samples/bpf/map_perf_test_user.c b/samples/bpf/map_perf_test_user.c
> index fe5564bff39b..da3c101ca776 100644
> --- a/samples/bpf/map_perf_test_user.c
> +++ b/samples/bpf/map_perf_test_user.c
> @@ -22,7 +22,7 @@
> #include "bpf_load.h"
>
> #define TEST_BIT(t) (1U << (t))
> -#define MAX_NR_CPUS 1024
> +#define MAX_NR_CPUS 2048
>
> static __u64 time_get_ns(void)
> {
> diff --git a/tools/perf/perf.h b/tools/perf/perf.h
> index 711e009381ec..74d0124d38f3 100644
> --- a/tools/perf/perf.h
> +++ b/tools/perf/perf.h
> @@ -26,7 +26,7 @@ static inline unsigned long long rdclock(void)
> }
>
> #ifndef MAX_NR_CPUS
> -#define MAX_NR_CPUS 1024
> +#define MAX_NR_CPUS 2048
> #endif
>
> extern const char *input_name;
> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
> index 06ddb6618ef3..abc9c2145efe 100644
> --- a/tools/perf/util/header.c
> +++ b/tools/perf/util/header.c
> @@ -1121,7 +1121,7 @@ static int build_caches(struct cpu_cache_level caches[], u32 size, u32 *cntp)
> return 0;
> }
>
> -#define MAX_CACHES 2000
> +#define MAX_CACHES (MAX_NR_CPUS * 4)
>
> static int write_cache(struct feat_fd *ff,
> struct perf_evlist *evlist __maybe_unused)
> --
> 2.12.3

--

- Arnaldo