* Re: [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self()
2022-02-01 21:40 [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self() Rob Herring
@ 2022-02-01 23:59 ` Arnaldo Carvalho de Melo
2022-02-02 10:33 ` John Garry
2022-02-02 16:48 ` Jiri Olsa
2 siblings, 0 replies; 5+ messages in thread
From: Arnaldo Carvalho de Melo @ 2022-02-01 23:59 UTC (permalink / raw)
To: Rob Herring
Cc: Will Deacon, Peter Zijlstra, Ingo Molnar, Mark Rutland,
Alexander Shishkin, Jiri Olsa, Namhyung Kim, Masayoshi Mizuma,
linux-perf-users, linux-kernel
Em Tue, Feb 01, 2022 at 03:40:56PM -0600, Rob Herring escreveu:
> Add the arm64 variants for read_perf_counter() and read_timestamp().
> Unfortunately the counter number is encoded into the instruction, so the
> code is a bit verbose to enumerate all possible counters.
>
> Tested-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> Signed-off-by: Rob Herring <robh@kernel.org>
Looks sane, will give some time to Jiri to react.
- Arnaldo
> ---
> Arm64 kernel support landed in 5.17, but the corresponding libperf
> support didn't get picked up.
>
> v9:
> - Rebase on v5.17-rc
> - Add Tested-by
> v8:
> - Set attr.config1 to request user access on arm64
> v7:
> - Move enabling of libperf user read test for arm64 to this patch
> ---
> tools/lib/perf/mmap.c | 98 +++++++++++++++++++++++++++++++
> tools/lib/perf/tests/test-evsel.c | 5 +-
> 2 files changed, 102 insertions(+), 1 deletion(-)
>
> diff --git a/tools/lib/perf/mmap.c b/tools/lib/perf/mmap.c
> index f7ee07cb5818..0d1634cedf44 100644
> --- a/tools/lib/perf/mmap.c
> +++ b/tools/lib/perf/mmap.c
> @@ -13,6 +13,7 @@
> #include <internal/lib.h>
> #include <linux/kernel.h>
> #include <linux/math64.h>
> +#include <linux/stringify.h>
> #include "internal.h"
>
> void perf_mmap__init(struct perf_mmap *map, struct perf_mmap *prev,
> @@ -294,6 +295,103 @@ static u64 read_timestamp(void)
>
> return low | ((u64)high) << 32;
> }
> +#elif defined(__aarch64__)
> +#define read_sysreg(r) ({ \
> + u64 __val; \
> + asm volatile("mrs %0, " __stringify(r) : "=r" (__val)); \
> + __val; \
> +})
> +
> +static u64 read_pmccntr(void)
> +{
> + return read_sysreg(pmccntr_el0);
> +}
> +
> +#define PMEVCNTR_READ(idx) \
> + static u64 read_pmevcntr_##idx(void) { \
> + return read_sysreg(pmevcntr##idx##_el0); \
> + }
> +
> +PMEVCNTR_READ(0);
> +PMEVCNTR_READ(1);
> +PMEVCNTR_READ(2);
> +PMEVCNTR_READ(3);
> +PMEVCNTR_READ(4);
> +PMEVCNTR_READ(5);
> +PMEVCNTR_READ(6);
> +PMEVCNTR_READ(7);
> +PMEVCNTR_READ(8);
> +PMEVCNTR_READ(9);
> +PMEVCNTR_READ(10);
> +PMEVCNTR_READ(11);
> +PMEVCNTR_READ(12);
> +PMEVCNTR_READ(13);
> +PMEVCNTR_READ(14);
> +PMEVCNTR_READ(15);
> +PMEVCNTR_READ(16);
> +PMEVCNTR_READ(17);
> +PMEVCNTR_READ(18);
> +PMEVCNTR_READ(19);
> +PMEVCNTR_READ(20);
> +PMEVCNTR_READ(21);
> +PMEVCNTR_READ(22);
> +PMEVCNTR_READ(23);
> +PMEVCNTR_READ(24);
> +PMEVCNTR_READ(25);
> +PMEVCNTR_READ(26);
> +PMEVCNTR_READ(27);
> +PMEVCNTR_READ(28);
> +PMEVCNTR_READ(29);
> +PMEVCNTR_READ(30);
> +
> +/*
> + * Read a value direct from PMEVCNTR<idx>
> + */
> +static u64 read_perf_counter(unsigned int counter)
> +{
> + static u64 (* const read_f[])(void) = {
> + read_pmevcntr_0,
> + read_pmevcntr_1,
> + read_pmevcntr_2,
> + read_pmevcntr_3,
> + read_pmevcntr_4,
> + read_pmevcntr_5,
> + read_pmevcntr_6,
> + read_pmevcntr_7,
> + read_pmevcntr_8,
> + read_pmevcntr_9,
> + read_pmevcntr_10,
> + read_pmevcntr_11,
> + read_pmevcntr_13,
> + read_pmevcntr_12,
> + read_pmevcntr_14,
> + read_pmevcntr_15,
> + read_pmevcntr_16,
> + read_pmevcntr_17,
> + read_pmevcntr_18,
> + read_pmevcntr_19,
> + read_pmevcntr_20,
> + read_pmevcntr_21,
> + read_pmevcntr_22,
> + read_pmevcntr_23,
> + read_pmevcntr_24,
> + read_pmevcntr_25,
> + read_pmevcntr_26,
> + read_pmevcntr_27,
> + read_pmevcntr_28,
> + read_pmevcntr_29,
> + read_pmevcntr_30,
> + read_pmccntr
> + };
> +
> + if (counter < ARRAY_SIZE(read_f))
> + return (read_f[counter])();
> +
> + return 0;
> +}
> +
> +static u64 read_timestamp(void) { return read_sysreg(cntvct_el0); }
> +
> #else
> static u64 read_perf_counter(unsigned int counter __maybe_unused) { return 0; }
> static u64 read_timestamp(void) { return 0; }
> diff --git a/tools/lib/perf/tests/test-evsel.c b/tools/lib/perf/tests/test-evsel.c
> index 33ae9334861a..89be89afb24d 100644
> --- a/tools/lib/perf/tests/test-evsel.c
> +++ b/tools/lib/perf/tests/test-evsel.c
> @@ -130,6 +130,9 @@ static int test_stat_user_read(int event)
> struct perf_event_attr attr = {
> .type = PERF_TYPE_HARDWARE,
> .config = event,
> +#ifdef __aarch64__
> + .config1 = 0x2, /* Request user access */
> +#endif
> };
> int err, i;
>
> @@ -150,7 +153,7 @@ static int test_stat_user_read(int event)
> pc = perf_evsel__mmap_base(evsel, 0, 0);
> __T("failed to get mmapped address", pc);
>
> -#if defined(__i386__) || defined(__x86_64__)
> +#if defined(__i386__) || defined(__x86_64__) || defined(__aarch64__)
> __T("userspace counter access not supported", pc->cap_user_rdpmc);
> __T("userspace counter access not enabled", pc->index);
> __T("userspace counter width not set", pc->pmc_width >= 32);
> --
> 2.32.0
--
- Arnaldo
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self()
2022-02-01 21:40 [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self() Rob Herring
2022-02-01 23:59 ` Arnaldo Carvalho de Melo
@ 2022-02-02 10:33 ` John Garry
2022-02-02 16:48 ` Jiri Olsa
2 siblings, 0 replies; 5+ messages in thread
From: John Garry @ 2022-02-02 10:33 UTC (permalink / raw)
To: Rob Herring, Will Deacon, Peter Zijlstra, Ingo Molnar,
Arnaldo Carvalho de Melo, Mark Rutland, Alexander Shishkin,
Jiri Olsa, Namhyung Kim
Cc: Masayoshi Mizuma, linux-perf-users, linux-kernel
On 01/02/2022 21:40, Rob Herring wrote:
> Add the arm64 variants for read_perf_counter() and read_timestamp().
> Unfortunately the counter number is encoded into the instruction, so the
> code is a bit verbose to enumerate all possible counters.
>
> Tested-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> Signed-off-by: Rob Herring <robh@kernel.org>
Ha, I was just looking at this same topic yesterday from an old patchset
[0] and wasn't aware of this separate change.
Tested-by: John Garry <john.garry@huawei.com>
[0] https://lore.kernel.org/all/20210311000837.3630499-1-robh@kernel.org/
BTW, Documentation/arm64/perf.rst has a reference to a file which does
not seem to exit:
.. _tools/perf/arch/arm64/tests/user-events.c:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/tools/perf/arch/arm64/tests/user-events.c
Thanks,
john
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self()
2022-02-01 21:40 [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self() Rob Herring
2022-02-01 23:59 ` Arnaldo Carvalho de Melo
2022-02-02 10:33 ` John Garry
@ 2022-02-02 16:48 ` Jiri Olsa
2022-02-06 12:14 ` Arnaldo Carvalho de Melo
2 siblings, 1 reply; 5+ messages in thread
From: Jiri Olsa @ 2022-02-02 16:48 UTC (permalink / raw)
To: Rob Herring
Cc: Will Deacon, Peter Zijlstra, Ingo Molnar,
Arnaldo Carvalho de Melo, Mark Rutland, Alexander Shishkin,
Namhyung Kim, Masayoshi Mizuma, linux-perf-users, linux-kernel
On Tue, Feb 01, 2022 at 03:40:56PM -0600, Rob Herring wrote:
> Add the arm64 variants for read_perf_counter() and read_timestamp().
> Unfortunately the counter number is encoded into the instruction, so the
> code is a bit verbose to enumerate all possible counters.
>
> Tested-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> Signed-off-by: Rob Herring <robh@kernel.org>
> ---
> Arm64 kernel support landed in 5.17, but the corresponding libperf
> support didn't get picked up.
>
> v9:
> - Rebase on v5.17-rc
> - Add Tested-by
> v8:
> - Set attr.config1 to request user access on arm64
> v7:
> - Move enabling of libperf user read test for arm64 to this patch
> ---
> tools/lib/perf/mmap.c | 98 +++++++++++++++++++++++++++++++
> tools/lib/perf/tests/test-evsel.c | 5 +-
> 2 files changed, 102 insertions(+), 1 deletion(-)
>
> diff --git a/tools/lib/perf/mmap.c b/tools/lib/perf/mmap.c
> index f7ee07cb5818..0d1634cedf44 100644
> --- a/tools/lib/perf/mmap.c
> +++ b/tools/lib/perf/mmap.c
> @@ -13,6 +13,7 @@
> #include <internal/lib.h>
> #include <linux/kernel.h>
> #include <linux/math64.h>
> +#include <linux/stringify.h>
> #include "internal.h"
>
> void perf_mmap__init(struct perf_mmap *map, struct perf_mmap *prev,
> @@ -294,6 +295,103 @@ static u64 read_timestamp(void)
>
> return low | ((u64)high) << 32;
> }
> +#elif defined(__aarch64__)
> +#define read_sysreg(r) ({ \
> + u64 __val; \
> + asm volatile("mrs %0, " __stringify(r) : "=r" (__val)); \
> + __val; \
> +})
if this works on arm I don't have problem, but aat some point
we shuld add arch directory and move it there ;-)
Acked-by: Jiri Olsa <jolsa@redhat.com>
thanks,
jirka
> +
> +static u64 read_pmccntr(void)
> +{
> + return read_sysreg(pmccntr_el0);
> +}
> +
> +#define PMEVCNTR_READ(idx) \
> + static u64 read_pmevcntr_##idx(void) { \
> + return read_sysreg(pmevcntr##idx##_el0); \
> + }
> +
> +PMEVCNTR_READ(0);
> +PMEVCNTR_READ(1);
> +PMEVCNTR_READ(2);
> +PMEVCNTR_READ(3);
> +PMEVCNTR_READ(4);
> +PMEVCNTR_READ(5);
> +PMEVCNTR_READ(6);
> +PMEVCNTR_READ(7);
> +PMEVCNTR_READ(8);
> +PMEVCNTR_READ(9);
> +PMEVCNTR_READ(10);
> +PMEVCNTR_READ(11);
> +PMEVCNTR_READ(12);
> +PMEVCNTR_READ(13);
> +PMEVCNTR_READ(14);
> +PMEVCNTR_READ(15);
> +PMEVCNTR_READ(16);
> +PMEVCNTR_READ(17);
> +PMEVCNTR_READ(18);
> +PMEVCNTR_READ(19);
> +PMEVCNTR_READ(20);
> +PMEVCNTR_READ(21);
> +PMEVCNTR_READ(22);
> +PMEVCNTR_READ(23);
> +PMEVCNTR_READ(24);
> +PMEVCNTR_READ(25);
> +PMEVCNTR_READ(26);
> +PMEVCNTR_READ(27);
> +PMEVCNTR_READ(28);
> +PMEVCNTR_READ(29);
> +PMEVCNTR_READ(30);
> +
> +/*
> + * Read a value direct from PMEVCNTR<idx>
> + */
> +static u64 read_perf_counter(unsigned int counter)
> +{
> + static u64 (* const read_f[])(void) = {
> + read_pmevcntr_0,
> + read_pmevcntr_1,
> + read_pmevcntr_2,
> + read_pmevcntr_3,
> + read_pmevcntr_4,
> + read_pmevcntr_5,
> + read_pmevcntr_6,
> + read_pmevcntr_7,
> + read_pmevcntr_8,
> + read_pmevcntr_9,
> + read_pmevcntr_10,
> + read_pmevcntr_11,
> + read_pmevcntr_13,
> + read_pmevcntr_12,
> + read_pmevcntr_14,
> + read_pmevcntr_15,
> + read_pmevcntr_16,
> + read_pmevcntr_17,
> + read_pmevcntr_18,
> + read_pmevcntr_19,
> + read_pmevcntr_20,
> + read_pmevcntr_21,
> + read_pmevcntr_22,
> + read_pmevcntr_23,
> + read_pmevcntr_24,
> + read_pmevcntr_25,
> + read_pmevcntr_26,
> + read_pmevcntr_27,
> + read_pmevcntr_28,
> + read_pmevcntr_29,
> + read_pmevcntr_30,
> + read_pmccntr
> + };
> +
> + if (counter < ARRAY_SIZE(read_f))
> + return (read_f[counter])();
> +
> + return 0;
> +}
> +
> +static u64 read_timestamp(void) { return read_sysreg(cntvct_el0); }
> +
> #else
> static u64 read_perf_counter(unsigned int counter __maybe_unused) { return 0; }
> static u64 read_timestamp(void) { return 0; }
> diff --git a/tools/lib/perf/tests/test-evsel.c b/tools/lib/perf/tests/test-evsel.c
> index 33ae9334861a..89be89afb24d 100644
> --- a/tools/lib/perf/tests/test-evsel.c
> +++ b/tools/lib/perf/tests/test-evsel.c
> @@ -130,6 +130,9 @@ static int test_stat_user_read(int event)
> struct perf_event_attr attr = {
> .type = PERF_TYPE_HARDWARE,
> .config = event,
> +#ifdef __aarch64__
> + .config1 = 0x2, /* Request user access */
> +#endif
> };
> int err, i;
>
> @@ -150,7 +153,7 @@ static int test_stat_user_read(int event)
> pc = perf_evsel__mmap_base(evsel, 0, 0);
> __T("failed to get mmapped address", pc);
>
> -#if defined(__i386__) || defined(__x86_64__)
> +#if defined(__i386__) || defined(__x86_64__) || defined(__aarch64__)
> __T("userspace counter access not supported", pc->cap_user_rdpmc);
> __T("userspace counter access not enabled", pc->index);
> __T("userspace counter width not set", pc->pmc_width >= 32);
> --
> 2.32.0
>
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [PATCH v9] libperf: Add arm64 support to perf_mmap__read_self()
2022-02-02 16:48 ` Jiri Olsa
@ 2022-02-06 12:14 ` Arnaldo Carvalho de Melo
0 siblings, 0 replies; 5+ messages in thread
From: Arnaldo Carvalho de Melo @ 2022-02-06 12:14 UTC (permalink / raw)
To: Jiri Olsa
Cc: Rob Herring, Will Deacon, Peter Zijlstra, Ingo Molnar,
Mark Rutland, Alexander Shishkin, Namhyung Kim, Masayoshi Mizuma,
linux-perf-users, linux-kernel
Em Wed, Feb 02, 2022 at 05:48:54PM +0100, Jiri Olsa escreveu:
> On Tue, Feb 01, 2022 at 03:40:56PM -0600, Rob Herring wrote:
> > Add the arm64 variants for read_perf_counter() and read_timestamp().
> > Unfortunately the counter number is encoded into the instruction, so the
> > code is a bit verbose to enumerate all possible counters.
> >
> > Tested-by: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
> > Signed-off-by: Rob Herring <robh@kernel.org>
> > ---
> > Arm64 kernel support landed in 5.17, but the corresponding libperf
> > support didn't get picked up.
> >
> > v9:
> > - Rebase on v5.17-rc
> > - Add Tested-by
> > v8:
> > - Set attr.config1 to request user access on arm64
> > v7:
> > - Move enabling of libperf user read test for arm64 to this patch
> > ---
> > tools/lib/perf/mmap.c | 98 +++++++++++++++++++++++++++++++
> > tools/lib/perf/tests/test-evsel.c | 5 +-
> > 2 files changed, 102 insertions(+), 1 deletion(-)
> >
> > diff --git a/tools/lib/perf/mmap.c b/tools/lib/perf/mmap.c
> > index f7ee07cb5818..0d1634cedf44 100644
> > --- a/tools/lib/perf/mmap.c
> > +++ b/tools/lib/perf/mmap.c
> > @@ -13,6 +13,7 @@
> > #include <internal/lib.h>
> > #include <linux/kernel.h>
> > #include <linux/math64.h>
> > +#include <linux/stringify.h>
> > #include "internal.h"
> >
> > void perf_mmap__init(struct perf_mmap *map, struct perf_mmap *prev,
> > @@ -294,6 +295,103 @@ static u64 read_timestamp(void)
> >
> > return low | ((u64)high) << 32;
> > }
> > +#elif defined(__aarch64__)
> > +#define read_sysreg(r) ({ \
> > + u64 __val; \
> > + asm volatile("mrs %0, " __stringify(r) : "=r" (__val)); \
> > + __val; \
> > +})
>
> if this works on arm I don't have problem, but aat some point
> we shuld add arch directory and move it there ;-)
>
> Acked-by: Jiri Olsa <jolsa@redhat.com>
Thanks, applied.
- Arnaldo
> thanks,
> jirka
>
> > +
> > +static u64 read_pmccntr(void)
> > +{
> > + return read_sysreg(pmccntr_el0);
> > +}
> > +
> > +#define PMEVCNTR_READ(idx) \
> > + static u64 read_pmevcntr_##idx(void) { \
> > + return read_sysreg(pmevcntr##idx##_el0); \
> > + }
> > +
> > +PMEVCNTR_READ(0);
> > +PMEVCNTR_READ(1);
> > +PMEVCNTR_READ(2);
> > +PMEVCNTR_READ(3);
> > +PMEVCNTR_READ(4);
> > +PMEVCNTR_READ(5);
> > +PMEVCNTR_READ(6);
> > +PMEVCNTR_READ(7);
> > +PMEVCNTR_READ(8);
> > +PMEVCNTR_READ(9);
> > +PMEVCNTR_READ(10);
> > +PMEVCNTR_READ(11);
> > +PMEVCNTR_READ(12);
> > +PMEVCNTR_READ(13);
> > +PMEVCNTR_READ(14);
> > +PMEVCNTR_READ(15);
> > +PMEVCNTR_READ(16);
> > +PMEVCNTR_READ(17);
> > +PMEVCNTR_READ(18);
> > +PMEVCNTR_READ(19);
> > +PMEVCNTR_READ(20);
> > +PMEVCNTR_READ(21);
> > +PMEVCNTR_READ(22);
> > +PMEVCNTR_READ(23);
> > +PMEVCNTR_READ(24);
> > +PMEVCNTR_READ(25);
> > +PMEVCNTR_READ(26);
> > +PMEVCNTR_READ(27);
> > +PMEVCNTR_READ(28);
> > +PMEVCNTR_READ(29);
> > +PMEVCNTR_READ(30);
> > +
> > +/*
> > + * Read a value direct from PMEVCNTR<idx>
> > + */
> > +static u64 read_perf_counter(unsigned int counter)
> > +{
> > + static u64 (* const read_f[])(void) = {
> > + read_pmevcntr_0,
> > + read_pmevcntr_1,
> > + read_pmevcntr_2,
> > + read_pmevcntr_3,
> > + read_pmevcntr_4,
> > + read_pmevcntr_5,
> > + read_pmevcntr_6,
> > + read_pmevcntr_7,
> > + read_pmevcntr_8,
> > + read_pmevcntr_9,
> > + read_pmevcntr_10,
> > + read_pmevcntr_11,
> > + read_pmevcntr_13,
> > + read_pmevcntr_12,
> > + read_pmevcntr_14,
> > + read_pmevcntr_15,
> > + read_pmevcntr_16,
> > + read_pmevcntr_17,
> > + read_pmevcntr_18,
> > + read_pmevcntr_19,
> > + read_pmevcntr_20,
> > + read_pmevcntr_21,
> > + read_pmevcntr_22,
> > + read_pmevcntr_23,
> > + read_pmevcntr_24,
> > + read_pmevcntr_25,
> > + read_pmevcntr_26,
> > + read_pmevcntr_27,
> > + read_pmevcntr_28,
> > + read_pmevcntr_29,
> > + read_pmevcntr_30,
> > + read_pmccntr
> > + };
> > +
> > + if (counter < ARRAY_SIZE(read_f))
> > + return (read_f[counter])();
> > +
> > + return 0;
> > +}
> > +
> > +static u64 read_timestamp(void) { return read_sysreg(cntvct_el0); }
> > +
> > #else
> > static u64 read_perf_counter(unsigned int counter __maybe_unused) { return 0; }
> > static u64 read_timestamp(void) { return 0; }
> > diff --git a/tools/lib/perf/tests/test-evsel.c b/tools/lib/perf/tests/test-evsel.c
> > index 33ae9334861a..89be89afb24d 100644
> > --- a/tools/lib/perf/tests/test-evsel.c
> > +++ b/tools/lib/perf/tests/test-evsel.c
> > @@ -130,6 +130,9 @@ static int test_stat_user_read(int event)
> > struct perf_event_attr attr = {
> > .type = PERF_TYPE_HARDWARE,
> > .config = event,
> > +#ifdef __aarch64__
> > + .config1 = 0x2, /* Request user access */
> > +#endif
> > };
> > int err, i;
> >
> > @@ -150,7 +153,7 @@ static int test_stat_user_read(int event)
> > pc = perf_evsel__mmap_base(evsel, 0, 0);
> > __T("failed to get mmapped address", pc);
> >
> > -#if defined(__i386__) || defined(__x86_64__)
> > +#if defined(__i386__) || defined(__x86_64__) || defined(__aarch64__)
> > __T("userspace counter access not supported", pc->cap_user_rdpmc);
> > __T("userspace counter access not enabled", pc->index);
> > __T("userspace counter width not set", pc->pmc_width >= 32);
> > --
> > 2.32.0
> >
--
- Arnaldo
^ permalink raw reply [flat|nested] 5+ messages in thread