From: Leo Yan <leo.yan@linaro.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>, Marc Zyngier <maz@kernel.org>,
Mark Rutland <mark.rutland@arm.com>,
Catalin Marinas <catalin.marinas@arm.com>,
linux-kernel@vger.kernel.org,
linux-arm-kernel@lists.infradead.org, mingo@redhat.com,
acme@kernel.org, alexander.shishkin@linux.intel.com,
jolsa@redhat.com, daniel.lezcano@linaro.org, tglx@linutronix.de,
sboyd@codeaurora.org, john.stultz@linaro.org
Subject: Re: [PATCH 2/5] arm64: perf: Implement correct cap_user_time
Date: Tue, 12 May 2020 22:03:01 +0800 [thread overview]
Message-ID: <20200512140301.GC20352@leoy-ThinkPad-X240s> (raw)
In-Reply-To: <20200512124450.824507755@infradead.org>
Hi Peter,
On Tue, May 12, 2020 at 02:41:00PM +0200, Peter Zijlstra wrote:
> As reported by Leo; the existing implementation is broken when the
> clock and counter don't intersect at 0.
>
> Use the sched_clock's struct clock_read_data information to correctly
> implement cap_user_time and cap_user_time_zero.
>
> Note that the ARM64 counter is architecturally only guaranteed to be
> 56bit wide (implementations are allowed to be wider) and the existing
> perf ABI cannot deal with wrap-around.
>
> This implementation should also be faster than the old; seeing how we
> don't need to recompute mult and shift all the time.
>
> Reported-by: Leo Yan <leo.yan@linaro.org>
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> ---
> arch/arm64/kernel/perf_event.c | 36 +++++++++++++++++++++++++++---------
> 1 file changed, 27 insertions(+), 9 deletions(-)
>
> --- a/arch/arm64/kernel/perf_event.c
> +++ b/arch/arm64/kernel/perf_event.c
> @@ -19,6 +19,7 @@
> #include <linux/of.h>
> #include <linux/perf/arm_pmu.h>
> #include <linux/platform_device.h>
> +#include <linux/sched_clock.h>
> #include <linux/smp.h>
>
> /* ARMv8 Cortex-A53 specific event types. */
> @@ -1165,28 +1166,45 @@ device_initcall(armv8_pmu_driver_init)
> void arch_perf_update_userpage(struct perf_event *event,
> struct perf_event_mmap_page *userpg, u64 now)
> {
> - u32 freq;
> - u32 shift;
> + struct clock_read_data *rd;
> + unsigned int seq;
>
> /*
> * Internal timekeeping for enabled/running/stopped times
> * is always computed with the sched_clock.
> */
> - freq = arch_timer_get_rate();
> userpg->cap_user_time = 1;
> + userpg->cap_user_time_zero = 1;
> +
> + do {
> + rd = sched_clock_read_begin(&seq);
> +
> + userpg->time_mult = rd->mult;
> + userpg->time_shift = rd->shift;
> + userpg->time_zero = rd->epoch_ns;
> +
> + /*
> + * This isn't strictly correct, the ARM64 counter can be
> + * 'short' and then we get funnies when it wraps. The correct
> + * thing would be to extend the perf ABI with a cycle and mask
> + * value, but because wrapping on ARM64 is very rare in
> + * practise this 'works'.
> + */
> + userpg->time_zero -= (rd->epoch_cyc * rd->mult) >> rd->shift;
> +
> + } while (sched_clock_read_retry(seq));
> +
> + userpg->time_offset = userpg->time_zero - now;
>
> - clocks_calc_mult_shift(&userpg->time_mult, &shift, freq,
> - NSEC_PER_SEC, 0);
> /*
> * time_shift is not expected to be greater than 31 due to
> * the original published conversion algorithm shifting a
> * 32-bit value (now specifies a 64-bit value) - refer
> * perf_event_mmap_page documentation in perf_event.h.
> */
> - if (shift == 32) {
> - shift = 31;
> + if (userpg->shift == 32) {
Thanks a lot for the patch set, some typos:
s/shift/time_shift
> + userpg->shift = 31;
s/shift/time_shift
Thanks,
Leo
> userpg->time_mult >>= 1;
> }
> - userpg->time_shift = (u16)shift;
> - userpg->time_offset = -now;
> +
> }
>
>
next prev parent reply other threads:[~2020-05-12 14:03 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-12 12:40 [PATCH 0/5] arm64: perf: Proper cap_user_time* support Peter Zijlstra
2020-05-12 12:40 ` [PATCH 1/5] sched_clock: Expose struct clock_read_data Peter Zijlstra
2020-05-12 12:41 ` [PATCH 2/5] arm64: perf: Implement correct cap_user_time Peter Zijlstra
2020-05-12 14:03 ` Leo Yan [this message]
2020-05-12 14:08 ` Peter Zijlstra
2020-05-12 16:05 ` kbuild test robot
2020-05-12 12:41 ` [PATCH 3/5] arm64: perf: Only advertise cap_user_time for arch_timer Peter Zijlstra
2020-05-12 12:41 ` [PATCH 4/5] perf: Add perf_event_mmap_page::cap_user_time_short ABI Peter Zijlstra
2020-05-12 12:41 ` [PATCH 5/5] arm64: perf: Add cap_user_time_short Peter Zijlstra
2020-05-12 14:11 ` Leo Yan
2020-05-12 16:59 ` kbuild test robot
2020-07-13 6:08 ` [PATCH 0/5] arm64: perf: Proper cap_user_time* support Leo Yan
2020-07-13 10:11 ` Will Deacon
2020-07-13 12:58 ` Leo Yan
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200512140301.GC20352@leoy-ThinkPad-X240s \
--to=leo.yan@linaro.org \
--cc=acme@kernel.org \
--cc=alexander.shishkin@linux.intel.com \
--cc=catalin.marinas@arm.com \
--cc=daniel.lezcano@linaro.org \
--cc=john.stultz@linaro.org \
--cc=jolsa@redhat.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mark.rutland@arm.com \
--cc=maz@kernel.org \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=sboyd@codeaurora.org \
--cc=tglx@linutronix.de \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).