Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Vineet Gupta <vineet.gupta1@synopsys.com>
To:     Alexey Brodkin <alexey.brodkin@synopsys.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
CC:     "linux-snps-arc@lists.infradead.org" 
        <linux-snps-arc@lists.infradead.org>,
        Daniel Lezcano <daniel.lezcano@linaro.org>,
        Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH] clocksource/drivers/arc_timer: Utilize generic
 sched_clock
Thread-Topic: [PATCH] clocksource/drivers/arc_timer: Utilize generic
 sched_clock
Thread-Index: AQHUZSQrx0Qmu49VI06mKwaw7Xomww==
Date:   Tue, 16 Oct 2018 16:03:27 +0000
Message-ID: <C2D7FE5348E1B147BCA15975FBA23075012B0AE356@US01WEMBX2.internal.synopsys.com>
References: <20181016074504.2042-1-abrodkin@synopsys.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 10/16/2018 12:45 AM, Alexey Brodkin wrote:=0A=
> It turned out we used to use default implementation of sched_clock()=0A=
> from kernel/sched/clock.c which was as precise as 1/HZ, i.e.=0A=
> by default we had 10 msec granularity of time measurement.=0A=
>=0A=
> Now given ARC built-in timers are clocked with the same frequency as=0A=
> CPU cores we may get much higher precision of time tracking.=0A=
=0A=
Can you do LMBench runs with and w/o and see if there's any other changes. =
I'm=0A=
hoping lat_ctx will be more consistent.=0A=
=0A=
> diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig=0A=
> index dec0dd88ec15..3268dad4effe 100644=0A=
> --- a/drivers/clocksource/Kconfig=0A=
> +++ b/drivers/clocksource/Kconfig=0A=
> @@ -290,6 +290,7 @@ config CLKSRC_MPS2=0A=
>  =0A=
>  config ARC_TIMERS=0A=
>  	bool "Support for 32-bit TIMERn counters in ARC Cores" if COMPILE_TEST=
=0A=
> +	depends on GENERIC_SCHED_CLOCK=0A=
=0A=
It needs to select, not depends on=0A=
=0A=
> @@ -88,6 +89,11 @@ static u64 arc_read_gfrc(struct clocksource *cs)=0A=
>  	return (((u64)h) << 32) | l;=0A=
>  }=0A=
>  =0A=
> +static u64 arc_gfrc_clock_read(void)=0A=
=0A=
Needs to be notrace like other such routines.=0A=
=0A=
>=0A=
> +=0A=
>  static struct clocksource arc_counter_timer1 =3D {=0A=
>  	.name   =3D "ARC Timer1",=0A=
>  	.rating =3D 300,=0A=
> @@ -209,6 +229,8 @@ static int __init arc_cs_setup_timer1(struct device_n=
ode *node)=0A=
>  	write_aux_reg(ARC_REG_TIMER1_CNT, 0);=0A=
>  	write_aux_reg(ARC_REG_TIMER1_CTRL, TIMER_CTRL_NH);=0A=
>  =0A=
> +	sched_clock_register(arc_timer1_clock_read, 64, arc_timer_freq);=0A=
=0A=
TIMER1 is 32 bits wide.=0A=
=0A=
-Vineet=0A=