From: Paul Cercueil <paul@crapouillou.net>
To: Zhou Yanjie <zhouyanjie@wanyeetech.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
linux-kernel@vger.kernel.org, devicetree@vger.kernel.org,
robh+dt@kernel.org, tglx@linutronix.de,
dongsheng.qiu@ingenic.com, aric.pzqi@ingenic.com,
rick.tyliu@ingenic.com, yanfei.li@ingenic.com,
sernia.zhou@foxmail.com, zhenwenjin@gmail.com
Subject: Re: [PATCH v6 2/2] clocksource: Ingenic: Add support for the Ingenic X1000 OST.
Date: Sat, 18 Jul 2020 18:11:10 +0200 [thread overview]
Message-ID: <MAAODQ.8WWI41VIHCH92@crapouillou.net> (raw)
In-Reply-To: <36aec769-da25-20dd-2ca4-351f7c5369fb@wanyeetech.com>
Le sam. 18 juil. 2020 à 23:55, Zhou Yanjie <zhouyanjie@wanyeetech.com>
a écrit :
> Hi Paul,
>
> 在 2020/7/18 下午11:44, Paul Cercueil 写道:
>> Hi Zhou,
>>
>> Le sam. 18 juil. 2020 à 21:42, Zhou Yanjie
>> <zhouyanjie@wanyeetech.com> \x7fa écrit :
>>> Hello Paul and Daniel,
>>>
>>> 在 2020/7/18 下午9:12, Paul Cercueil 写道:
>>>> Hi Daniel,
>>>>
>>>> Le ven. 17 juil. 2020 à 10:02, Daniel Lezcano
>>>> \x7f\x7f\x7f\x7f<daniel.lezcano@linaro.org> a écrit :
>>>>> On 17/07/2020 08:13, Zhou Yanjie wrote:
>>>>>> Hi Daniel,
>>>>>>
>>>>>> 在 2020/7/17 下午12:20, Daniel Lezcano 写道:
>>>>>>> On 10/07/2020 19:02, 周琰杰 (Zhou Yanjie) wrote:
>>>>>>>> X1000 and SoCs after X1000 (such as X1500 and X1830) had a
>>>>>>>> \x7f\x7f\x7f\x7f\x7f\x7f\x7fseparate
>>>>>>>> OST, it no longer belongs to TCU. This driver will register
>>>>>>>> both a
>>>>>>>> clocksource and a sched_clock to the system.
>>>>>>>>
>>>>>>>> Tested-by: 周正 (Zhou Zheng) <sernia.zhou@foxmail.com>
>>>>>>>> Co-developed-by: 漆鹏振 (Qi Pengzhen)
>>>>>>>> <aric.pzqi@ingenic.com>
>>>>>>>> Signed-off-by: 漆鹏振 (Qi Pengzhen) <aric.pzqi@ingenic.com>
>>>>>>>> Signed-off-by: 周琰杰 (Zhou Yanjie)
>>>>>>>> <zhouyanjie@wanyeetech.com>
>>>>>>>> Reviewed-by: Paul Cercueil <paul@crapouillou.net>
>>>>>>>> ---
>>>>>>>>
>>>>>>>> Notes:
>>>>>>>> v1->v2:
>>>>>>>> Fix compile warnings.
>>>>>>>> Reported-by: kernel test robot <lkp@intel.com>
>>>>>>>> v2->v3:
>>>>>>>> No change.
>>>>>>>> v3->v4:
>>>>>>>> 1.Rename "ost" to "sysost"
>>>>>>>> 1.Remove unrelated changes.
>>>>>>>> 2.Remove ost_clock_parent enum.
>>>>>>>> 3.Remove
>>>>>>>> ost->percpu_timer_channel/ost->global_timer_channel.
>>>>>>>> 4.Set up independent .recalc_rate/.set_rate for
>>>>>>>> percpu/global
>>>>>>>> timer.
>>>>>>>> 5.No longer call functions in variable declarations.
>>>>>>>> v4->v5:
>>>>>>>> Use "of_io_request_and_map()" instead "of_iomap()".
>>>>>>>> Suggested-by: Paul Cercueil <paul@crapouillou.net>
>>>>>>>> v5->v6:
>>>>>>>> No change.
>>>>>>>>
>>>>>>>> drivers/clocksource/Kconfig | 11 +
>>>>>>>> drivers/clocksource/Makefile | 1 +
>>>>>>>> drivers/clocksource/ingenic-sysost.c | 539
>>>>>>>> +++++++++++++++++++++++++++++++++++
>>>>>>>> 3 files changed, 551 insertions(+)
>>>>>>>> create mode 100644 drivers/clocksource/ingenic-sysost.c
>>>>>>>>
>>>>>>>> diff --git a/drivers/clocksource/Kconfig
>>>>>>>> \x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7fb/drivers/clocksource/Kconfig
>>>>>>>> index 91418381fcd4..1bca8b8fb30f 100644
>>>>>>>> --- a/drivers/clocksource/Kconfig
>>>>>>>> +++ b/drivers/clocksource/Kconfig
>>>>>>>> @@ -696,6 +696,17 @@ config INGENIC_TIMER
>>>>>>>> help
>>>>>>>> Support for the timer/counter unit of the Ingenic JZ
>>>>>>>> SoCs.
>>>>>>>> +config INGENIC_SYSOST
>>>>>>>> + bool "Clocksource/timer using the SYSOST in Ingenic X
>>>>>>>> SoCs"
>>>>>>> We usually use silent options and let the platform's Kconfig
>>>>>>> \x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7fenable it.
>>>>>>> We show up the option only when COMPILE_TEST is enabled.
>>>>>>>
>>>>>>> Is there a reason to do it differently?
>>>>>>
>>>>>>
>>>>>> Do you mean
>>>>>>
>>>>>> bool "Clocksource/timer using the SYSOST in Ingenic X SoCs"
>>>>>>
>>>>>> or
>>>>>>
>>>>>> default MACH_INGENIC ?
>>>>>
>>>>> Both, no default here.
>>>>>
>>>>> eg.
>>>>>
>>>>> bool "Clocksource/timer using the SYSOST in Ingenic X SoCs" if
>>>>> \x7f\x7f\x7f\x7f\x7f\x7fCOMPILE_TEST
>>>>>
>>>>> and
>>>>>
>>>>> in arch/mips/Kconfig in the config MACH_INGENIC section :
>>>>>
>>>>> ...
>>>>> select INGENIC_SYSOST
>>>>> ...
>>>>
>>>> Disagreed. That's not how we do things on MIPS. Selecting
>>>> \x7f\x7f\x7fMACH_INGENIC \x7fmeans "this kernel will support Ingenic SoCs",
>>>> but not \x7f\x7f\x7fthat it will \x7fonly support these. Hence the depends on
>>>> MIPS / \x7f\x7f\x7fdefault MACH_INGENIC.
>>>>
>>>> As for the select INGENIC_SYSOST, this driver only applies to a
>>>> few \x7f\x7f\x7f\x7fSoCs, I certainly don't want it to be force-enabled. I
>>>> don't even \x7f\x7f\x7fwait \x7fit to be force-enabled on X1000, since it is
>>>> optional there too.
>>>>
>>>> Cheers,
>>>> -Paul
>>>
>>>
>>> If we still need to keep the "default MACH_INGENIC", then Daniel
>>> can \x7f\x7fdirectly apply the v6 version.
>>>
>>> If we need to use the silent options, maybe we can enable them
>>> \x7f\x7fseparately according to
>>> \x7f\x7fMACH_JZ4740/MACH_JZ4770/MACH_JZ4780/MACH_X1000/MACH_X1830.
>>>
>>> In fact, I think X1000 and X1830 need to enable this driver in most
>>> \x7f\x7fcases, because the current test has found that use TCU to provide
>>> \x7f\x7fclocksource and clockevent will cause data loss/error when
>>> \x7f\x7ftransmitting data through spi or ethernet. And these errors no
>>> longer \x7f\x7fappear after using OST.
>>
>> This is likely because the clock is too fast, try to reduce it by a
>> \x7ffactor 2 or 4, it should behave better.
>>
>
> Yes, it is indeed because the clock is too fast, especially the TCU
> only has 16-bit, which makes this problem more serious. I even
> reduced the timing clock to the lowest possible 23kHz (divided by
> 1024). The problem has been alleviated, but it cannot completely
> solved, and finally there is no problem after opening the OST.
>
>
>> If it turns out OST is really needed, then it should be selected
>> from \x7fMACH_X1000/X1830, not MACH_INGENIC.
>>
>
> Sure, I will do it in v8.
>
> Thanks and best regards!
Please do it in a separate patch then (but in the same patchset).
Cheers,
-Paul
>
>> Cheers,
>> -Paul
>>
>>> Thanks and best regards!
>>>
>>>
>>>>
>>>>>
>>>>>> This driver has some origins from "INGENIC_TIMER" driver and
>>>>>> "INGENIC_OST" driver.
>>>>>> Early Ingenic processors used TCU (timer/counter unit, has 6 or
>>>>>> 8
>>>>>> generic timer channels) to provide clocksource and clockevent
>>>>>> \x7f\x7f\x7f\x7f\x7f(both \x7f\x7f\x7fwith
>>>>>> only 16bit precision). This part of the processor can only use
>>>>>> "INGENIC_TIMER" driver.
>>>>>>
>>>>>> Later processors provide an independent 32bit or 64bit timer
>>>>>> channel
>>>>>> (still under TCU, known as ost channel, this channel can not
>>>>>> \x7f\x7f\x7f\x7f\x7fgenerate
>>>>>> interrupt) to provid higher precision clocksource. The
>>>>>> "INGENIC_OST"
>>>>>> driver is for this channel. These processors can use
>>>>>> "INGENIC_TIMER"
>>>>>> driver, but using "INGENIC_OST" driver to provide higher
>>>>>> precision
>>>>>> clocksource would be a better choice (clockevent still needs to
>>>>>> be
>>>>>> provided by generic timer channel of TCU, and still 16bit
>>>>>> \x7f\x7f\x7f\x7f\x7fprecision).
>>>>>>
>>>>>> And the recent processors provide a SYSOST components, it is
>>>>>> \x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7findependent
>>>>>> from TCU, including a 64bit timer channel for clocksource and a
>>>>>> \x7f\x7f\x7f\x7f\x7f32bit
>>>>>> timer channel for clockevent. Although these processors can
>>>>>> also use
>>>>>> "INGENIC_TIMER" driver, but the better choice is completely
>>>>>> \x7f\x7f\x7f\x7f\x7f\x7f\x7f\x7findependent
>>>>>> use of "INGENIC_SYSOST" driver to provide higher precision
>>>>>> \x7f\x7f\x7f\x7f\x7fclocksource
>>>>>> and clockevent.
>>>>>
>>>>> Ok, the rating should do the job then.
>>>>>
>>>>> Thanks for the explanation.
>>>>>
>>>>>> You may have already noticed that this independent SYSOST
>>>>>> \x7f\x7f\x7f\x7f\x7fcomponent is
>>>>>> like an upgraded and streamlined TCU, which only retains one
>>>>>> generic
>>>>>> timer channel that can generate interrupts, upgrade it from
>>>>>> 16bit to
>>>>>> 32bit, and then retain the 64bit ost channel. so the driver
>>>>>> code and
>>>>>> Kconfig code of this patch is largely referenced
>>>>>> "INGENIC_TIMER" driver and "INGENIC_OST" driver.
>>>>>>
>>>>>> Thanks and best regards!
>>>>>>
>>>>>>>> + default MACH_INGENIC
>>>>>>>> + depends on MIPS || COMPILE_TEST
>>>>>>>> + depends on COMMON_CLK
>>>>>>>> + select MFD_SYSCON
>>>>>>>> + select TIMER_OF
>>>>>>>> + select IRQ_DOMAIN
>>>>>>>> + help
>>>>>>>> + Support for the SYSOST of the Ingenic X Series SoCs.
>>>>>>>> +
>>>>>>> [ ... ]
>>>>>>>
>>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> <http://www.linaro.org/> Linaro.org │ Open source software for
>>>>> ARM \x7f\x7f\x7f\x7fSoCs
>>>>>
>>>>> Follow Linaro: <http://www.facebook.com/pages/Linaro> Facebook |
>>>>> <http://twitter.com/#!/linaroorg> Twitter |
>>>>> <http://www.linaro.org/linaro-blog/> Blog
>>>>
>>
next prev parent reply other threads:[~2020-07-18 16:11 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-10 17:02 [PATCH v6 0/2] Add support for the OST in Ingenic X1000 周琰杰 (Zhou Yanjie)
2020-07-10 17:02 ` [PATCH v6 1/2] dt-bindings: timer: Add Ingenic X1000 OST bindings 周琰杰 (Zhou Yanjie)
2020-07-13 15:52 ` Rob Herring
2020-07-10 17:02 ` [PATCH v6 2/2] clocksource: Ingenic: Add support for the Ingenic X1000 OST 周琰杰 (Zhou Yanjie)
2020-07-17 4:20 ` Daniel Lezcano
2020-07-17 6:13 ` Zhou Yanjie
2020-07-17 8:02 ` Daniel Lezcano
2020-07-17 14:40 ` Zhou Yanjie
2020-07-18 13:12 ` Paul Cercueil
2020-07-18 13:42 ` Zhou Yanjie
2020-07-18 15:44 ` Paul Cercueil
2020-07-18 15:55 ` Zhou Yanjie
2020-07-18 16:11 ` Paul Cercueil [this message]
2020-07-18 16:18 ` Zhou Yanjie
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=MAAODQ.8WWI41VIHCH92@crapouillou.net \
--to=paul@crapouillou.net \
--cc=aric.pzqi@ingenic.com \
--cc=daniel.lezcano@linaro.org \
--cc=devicetree@vger.kernel.org \
--cc=dongsheng.qiu@ingenic.com \
--cc=linux-kernel@vger.kernel.org \
--cc=rick.tyliu@ingenic.com \
--cc=robh+dt@kernel.org \
--cc=sernia.zhou@foxmail.com \
--cc=tglx@linutronix.de \
--cc=yanfei.li@ingenic.com \
--cc=zhenwenjin@gmail.com \
--cc=zhouyanjie@wanyeetech.com \
/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).