From: "Andreas Färber" <afaerber@suse.de>
To: Daniel Lezcano <daniel.lezcano@linaro.org>
Cc: arm@kernel.org, linux-arm-kernel@lists.infradead.org,
mp-cs@actions-semi.com, 96boards@ucrobotics.com,
support@lemaker.org, linux-kernel@vger.kernel.org,
Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH v3 04/25] clocksource: Add Owl timer
Date: Tue, 28 Feb 2017 18:08:06 +0100 [thread overview]
Message-ID: <1994faec-7936-aa96-87b4-df48f75f179b@suse.de> (raw)
In-Reply-To: <20170228164745.GC30601@mai>
Am 28.02.2017 um 17:47 schrieb Daniel Lezcano:
> On Tue, Feb 28, 2017 at 07:35:14AM +0100, Andreas Färber wrote:
>> The Actions Semi S500 SoC provides four timers, 2Hz0/1 and 32-bit TIMER0/1.
>>
>> Use TIMER0 as clocksource and TIMER1 as clockevents.
>>
>> Based on LeMaker linux-actions tree.
>>
>> An S500 datasheet can be found on the LeMaker Guitar pages:
>> http://www.lemaker.org/product-guitar-download-29.html
>>
>> Signed-off-by: Andreas Färber <afaerber@suse.de>
>> ---
>> v2 -> v3:
>> * Cleared interrupt pending flag for Timer1
>> * Adopted named interrupts for Timer1
>> * Extended commit message (Daniel)
>> * Adopted BIT() macros (Daniel)
>> * Adopted PTR_ERR() (Daniel)
>> * Adopted request_irq() (Daniel)
>> * Factored timer reset out (Daniel)
>> * Adopted CLOCK_EVT_FEAT_DYNIRQ (Daniel)
>> * Adopted clk input for rate (Daniel)
>> * Prepared for S900, adopting S500 DT compatible
>>
>> v2: new
>>
>> drivers/clocksource/Kconfig | 7 ++
>> drivers/clocksource/Makefile | 1 +
>> drivers/clocksource/owl-timer.c | 193 ++++++++++++++++++++++++++++++++++++++++
>> 3 files changed, 201 insertions(+)
>> create mode 100644 drivers/clocksource/owl-timer.c
>>
>> diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig
>> index 3356ab8..2551365 100644
>> --- a/drivers/clocksource/Kconfig
>> +++ b/drivers/clocksource/Kconfig
>> @@ -107,6 +107,13 @@ config ORION_TIMER
>> help
>> Enables the support for the Orion timer driver
>>
>> +config OWL_TIMER
>> + bool "Owl timer driver" if COMPILE_TEST
>> + depends on GENERIC_CLOCKEVENTS
>> + select CLKSRC_MMIO
>> + help
>> + Enables the support for the Actions Semi Owl timer driver.
>> +
>> config SUN4I_TIMER
>> bool "Sun4i timer driver" if COMPILE_TEST
>> depends on GENERIC_CLOCKEVENTS
>> diff --git a/drivers/clocksource/Makefile b/drivers/clocksource/Makefile
>> index d227d13..801b65a 100644
>> --- a/drivers/clocksource/Makefile
>> +++ b/drivers/clocksource/Makefile
>> @@ -53,6 +53,7 @@ obj-$(CONFIG_CLKSRC_PISTACHIO) += time-pistachio.o
>> obj-$(CONFIG_CLKSRC_TI_32K) += timer-ti-32k.o
>> obj-$(CONFIG_CLKSRC_NPS) += timer-nps.o
>> obj-$(CONFIG_OXNAS_RPS_TIMER) += timer-oxnas-rps.o
>> +obj-$(CONFIG_OWL_TIMER) += owl-timer.o
>>
>> obj-$(CONFIG_ARC_TIMERS) += arc_timer.o
>> obj-$(CONFIG_ARM_ARCH_TIMER) += arm_arch_timer.o
>> diff --git a/drivers/clocksource/owl-timer.c b/drivers/clocksource/owl-timer.c
>> new file mode 100644
>> index 0000000..1b1e26d
>> --- /dev/null
>> +++ b/drivers/clocksource/owl-timer.c
>> @@ -0,0 +1,193 @@
>> +/*
>> + * Actions Semi Owl timer
>> + *
>> + * Copyright 2012 Actions Semi Inc.
>> + * Author: Actions Semi, Inc.
>> + *
>> + * Copyright (c) 2017 SUSE Linux GmbH
>> + * Author: Andreas Färber
>> + *
>> + * This program is free software; you can redistribute it and/or modify it
>> + * under the terms of the GNU General Public License as published by the
>> + * Free Software Foundation; either version 2 of the License, or (at your
>> + * option) any later version.
>> + */
>> +
>> +#include <linux/clk.h>
>> +#include <linux/clockchips.h>
>> +#include <linux/interrupt.h>
>> +#include <linux/irq.h>
>> +#include <linux/irqreturn.h>
>> +#include <linux/sched_clock.h>
>> +#include <linux/of.h>
>> +#include <linux/of_address.h>
>> +#include <linux/of_irq.h>
>> +
>> +#define OWL_Tx_CTL 0x0
>> +#define OWL_Tx_CMP 0x4
>> +#define OWL_Tx_VAL 0x8
>> +
>> +#define OWL_Tx_CTL_PD BIT(0)
>> +#define OWL_Tx_CTL_INTEN BIT(1)
>> +#define OWL_Tx_CTL_EN BIT(2)
>> +
>> +#define OWL_MAX_Tx 2
>> +
>> +struct owl_timer_info {
>> + int timer_offset[OWL_MAX_Tx];
>> +};
>> +
>> +static const struct owl_timer_info *owl_timer_info;
>> +
>> +static void __iomem *owl_timer_base;
>> +
>> +static inline void __iomem *owl_timer_get_base(unsigned timer_nr)
>> +{
>> + if (timer_nr >= OWL_MAX_Tx)
>> + return NULL;
>
> The driver is supposed to know what is doing. owl_timer_get_base won't be
> called with an invalid index. This test will have a cost as it is called from
> owl_timer_sched_read.
OK
>> +
>> + return owl_timer_base + owl_timer_info->timer_offset[timer_nr];
>
> Actually, you just need a clkevt base @ and a clocksource base @.
>
> Instead of computing again and again the base, why not just precompute:
>
> owl_clksrc_base = owl_timer_base + owl_timer_info->timer_offset[OWL_TIMER0]
> owl_clkevt_base = owl_timer_base + owl_timer_info->timer_offset[OWL_TIMER1]
>
> at init time.
>
> And use these variables directly in the functions.
Either that, or revert to previous simpler behavior...
>> +}
>> +
>> +static inline void owl_timer_reset(unsigned index)
>> +{
>> + void __iomem *base;
>> +
>> + base = owl_timer_get_base(index);
>> + if (!base)
>> + return;
>
> Same here, this test is pointless.
Seems like you didn't look at the following patch yet. It sets two S500
offsets as -1, i.e. non-existant, which then results in NULL here.
>> + writel(0, base + OWL_Tx_CTL);
>> + writel(0, base + OWL_Tx_VAL);
>> + writel(0, base + OWL_Tx_CMP);
>> +}
>
> I suggest:
>
> static inline void owl_timer_reset(void __iomem *addr)
> {
> writel(0, addr + OWL_Tx_CTL);
> writel(0, addr + OWL_Tx_VAL);
> writel(0, addr + OWL_Tx_CMP);
> }
OK
>> +
>> +static u64 notrace owl_timer_sched_read(void)
>> +{
>> + return (u64)readl(owl_timer_get_base(0) + OWL_Tx_VAL);
>
> return (u64)readl(owl_clksrc_base + OWL_Tx_VAL);
>
>> +}
>> +
>
> static inline int owl_timer_set_state_disable(struct clock_event_device *evt)
> {
> return writel(0, owl_clkevt_base + OWL_Tx_CTL);
> }
That I don't like. Disabling is just setting a bit. We save a readl by
just writing where we know it's safe. An API like this is not safe.
>> +static int owl_timer_set_state_shutdown(struct clock_event_device *evt)
>> +{
>> + writel(0, owl_timer_get_base(0) + OWL_Tx_CTL);
>
> return owl_timer_set_state_disable(evt);
>
>> +
>> + return 0;
>> +}
>> +
>> +static int owl_timer_set_state_oneshot(struct clock_event_device *evt)
>> +{
>> + owl_timer_reset(1);
>
> Do you really need to do reset here ? Why not just owl_timer_set_state_disable(evt) ?
Matches downstream, will consider changing.
>> + return 0;
>> +}
>> +
>> +static int owl_timer_tick_resume(struct clock_event_device *evt)
>> +{
>> + return 0;
>> +}
>> +
>> +static int owl_timer_set_next_event(unsigned long evt,
>> + struct clock_event_device *ev)
>> +{
>> + void __iomem *base = owl_timer_get_base(1);
>> +
>> + writel(0, base + OWL_Tx_CTL);
>> +
>> + writel(0, base + OWL_Tx_VAL);
>
Are you suggesting a while line here? The point was disable first, then
initialize (2x), then activate. Maybe add comments instead?
>> + writel(evt, base + OWL_Tx_CMP);
>> +
>> + writel(OWL_Tx_CTL_EN | OWL_Tx_CTL_INTEN, base + OWL_Tx_CTL);
>> +
>> + return 0;
>> +}
>> +
>> +static struct clock_event_device owl_clockevent = {
>> + .name = "owl_tick",
>> + .rating = 200,
>> + .features = CLOCK_EVT_FEAT_ONESHOT |
>> + CLOCK_EVT_FEAT_DYNIRQ,
>> + .set_state_shutdown = owl_timer_set_state_shutdown,
>> + .set_state_oneshot = owl_timer_set_state_oneshot,
>> + .tick_resume = owl_timer_tick_resume,
>> + .set_next_event = owl_timer_set_next_event,
>> +};
>> +
>> +static irqreturn_t owl_timer1_interrupt(int irq, void *dev_id)
>> +{
>> + struct clock_event_device *evt = (struct clock_event_device *)dev_id;
>> +
>> + writel(OWL_Tx_CTL_PD, owl_timer_get_base(1) + OWL_Tx_CTL);
>> +
>> + evt->event_handler(evt);
Is there any guideline as to whether to clear such flag before or after?
>> +
>> + return IRQ_HANDLED;
>> +}
Regards,
Andreas
--
SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)
next prev parent reply other threads:[~2017-02-28 17:08 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-28 6:35 [PATCH v3 00/25] ARM: Initial Actions Semi S500 and S900 enablement Andreas Färber
2017-02-28 6:35 ` [PATCH v3 01/25] dt-bindings: Add vendor prefix for Actions Semi Andreas Färber
2017-02-28 6:35 ` [PATCH v3 02/25] dt-bindings: arm: Document Actions Semi S500 Andreas Färber
2017-02-28 6:35 ` [PATCH v3 03/25] dt-bindings: timer: Document Owl timer Andreas Färber
2017-02-28 12:39 ` Mark Rutland
2017-03-03 6:20 ` Rob Herring
2017-03-03 21:36 ` Andreas Färber
2017-02-28 6:35 ` [PATCH v3 04/25] clocksource: Add " Andreas Färber
2017-02-28 16:47 ` Daniel Lezcano
2017-02-28 17:08 ` Andreas Färber [this message]
2017-02-28 17:39 ` Daniel Lezcano
2017-02-28 18:01 ` Andreas Färber
2017-02-28 18:56 ` Thomas Gleixner
2017-02-28 18:53 ` Thomas Gleixner
2017-02-28 6:35 ` [PATCH v3 05/25] clocksource: owl: Add S900 support Andreas Färber
2017-02-28 17:16 ` Andreas Färber
2017-02-28 17:42 ` Daniel Lezcano
2017-02-28 6:35 ` [PATCH v3 06/25] ARM: Prepare Actions Semi S500 Andreas Färber
2017-02-28 6:35 ` [PATCH v3 07/25] ARM64: Prepare Actions Semi S900 Andreas Färber
2017-02-28 6:35 ` [PATCH v3 08/25] dt-bindings: serial: Document Actions Semi Owl UARTs Andreas Färber
2017-02-28 6:35 ` [PATCH v3 09/25] tty: serial: Add Actions Semi Owl UART earlycon Andreas Färber
2017-02-28 6:35 ` [PATCH v3 10/25] Documentation: kernel-parameters: Document owl earlycon Andreas Färber
2017-02-28 6:35 ` [PATCH v3 11/25] ARM: dts: Add Actions Semi S500 and LeMaker Guitar Andreas Färber
2017-02-28 12:32 ` Mark Rutland
2017-02-28 15:13 ` Andreas Färber
2017-03-01 18:43 ` Mark Rutland
2017-02-28 6:35 ` [PATCH v3 12/25] dt-bindings: Add vendor prefix for uCRobotics Andreas Färber
2017-02-28 6:35 ` [PATCH v3 13/25] dt-bindings: arm: Document Actions Semi S900 Andreas Färber
2017-02-28 6:35 ` [PATCH v3 14/25] ARM64: dts: Add Actions Semi S900 and Bubblegum-96 Andreas Färber
2017-02-28 6:35 ` [PATCH v3 15/25] MAINTAINERS: Add Actions Semi Owl section Andreas Färber
2017-02-28 6:35 ` [PATCH v3 16/25] tty: serial: owl: Implement console driver Andreas Färber
2017-02-28 6:35 ` [PATCH v3 17/25] ARM64: dts: actions: s900-bubblegum-96: Add fake uart5 clock Andreas Färber
2017-02-28 6:35 ` [PATCH v3 18/25] ARM: dts: s500-guitar-bb-rev-b: Add fake uart3 clock Andreas Färber
2017-02-28 6:35 ` [PATCH v3 19/25] dt-bindings: arm: cpus: Add S500 enable-method Andreas Färber
2017-03-03 6:21 ` Rob Herring
2017-02-28 6:35 ` [PATCH v3 20/25] ARM: owl: Implement CPU enable-method for S500 Andreas Färber
2017-03-01 7:19 ` kbuild test robot
2017-03-01 10:40 ` Andreas Färber
2017-03-03 23:00 ` Andreas Färber
2017-02-28 6:35 ` [PATCH v3 21/25] ARM: dts: s500: Set CPU enable-method Andreas Färber
2017-02-28 6:35 ` [PATCH v3 22/25] dt-bindings: power: Add Owl SPS power domains Andreas Färber
2017-03-03 6:21 ` Rob Herring
2017-02-28 6:35 ` [PATCH v3 23/25] soc: actions: Add Owl SPS Andreas Färber
2017-02-28 6:35 ` [PATCH v3 24/25] ARM: dts: s500: Add SPS node Andreas Färber
2017-02-28 6:35 ` [PATCH v3 25/25] ARM: owl: smp: Reimplement SPS power-gating for CPU2 and CPU3 Andreas Färber
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=1994faec-7936-aa96-87b4-df48f75f179b@suse.de \
--to=afaerber@suse.de \
--cc=96boards@ucrobotics.com \
--cc=arm@kernel.org \
--cc=daniel.lezcano@linaro.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mp-cs@actions-semi.com \
--cc=support@lemaker.org \
--cc=tglx@linutronix.de \
/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).