linux-kernel - Re: [PATCH 1/1] dt-bindings: clock: Add new OST support for the upcoming new driver.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <a16d9fc7-07f7-6a56-c0e4-5d39b34577a9@wanyeetech.com>
Date:   Thu, 15 Oct 2020 00:25:59 +0800
From:   Zhou Yanjie <zhouyanjie@...yeetech.com>
To:     Rob Herring <robh@...nel.org>
Cc:     devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
        sernia.zhou@...mail.com
Subject: Re: [PATCH 1/1] dt-bindings: clock: Add new OST support for the
 upcoming new driver.

Hi Rob,

在 2020/10/13 下午9:29, Rob Herring 写道:
> On Thu, Oct 08, 2020 at 02:14:07AM +0800, 周琰杰 (Zhou Yanjie) wrote:
>> The new OST has one global timer and two or four percpu timers, so there will be three
>> combinations in the upcoming new OST driver: the original GLOBAL_TIMER + PERCPU_TIMER,
>> the new GLOBAL_TIMER + PERCPU_TIMER0/1 and GLOBAL_TIMER + PERCPU_TIMER0/1/2/3, For this,
>> add the macro definition about OST_CLK_PERCPU_TIMER0/1/2/3. And in order to ensure that
>> all the combinations work normally, the original ABI values of OST_CLK_PERCPU_TIMER and
>> OST_CLK_GLOBAL_TIMER need to be exchanged to ensure that in any combinations, the clock
>> can be registered (by calling clk_hw_register()) from index 0.
> Wrap lines at <80 characters.


Sorry, I will fix it in the next version.


>
>> I'm sure that exchanging the ABI values of OST_CLK_PERCPU_TIMER and OST_CLK_GLOBAL_TIMER
>> will not affect the existing related drivers and the SoCs whitch using these drivers, so
>> we should be able to exchange them safely.
> Why exactly is this okay?


These definitions are currently only used in two places, one is when 
using "assigned-clocks" to configure the clock in the DTS file; the 
other is when registering the clock in the sysost driver. When 
exchanging the values of "OST_CLK_PERCPU_TIMER" and 
"OST_CLK_GLOBAL_TIMER", the corresponding values will be changed 
synchronously when the clock is driven to register and when the clock is 
configured through "assigned-clocks". Therefore, there is no situation 
that causes the wrong clock to the configured. And it has been tested on 
X1000 and X1000 and X1830 SoCs, all working fine.

I'm sorry because I didn't think carefully when writing this header 
file, and caused the current trouble.


Thanks and best regards!


>> Tested-by: 周正 (Zhou Zheng) <sernia.zhou@...mail.com>
>> Signed-off-by: 周琰杰 (Zhou Yanjie) <zhouyanjie@...yeetech.com>
>> ---
>>   include/dt-bindings/clock/ingenic,sysost.h | 10 +++++++---
>>   1 file changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/include/dt-bindings/clock/ingenic,sysost.h b/include/dt-bindings/clock/ingenic,sysost.h
>> index 9ac88e90babf..063791b01ab3 100644
>> --- a/include/dt-bindings/clock/ingenic,sysost.h
>> +++ b/include/dt-bindings/clock/ingenic,sysost.h
>> @@ -1,12 +1,16 @@
>>   /* SPDX-License-Identifier: GPL-2.0 */
>>   /*
>> - * This header provides clock numbers for the ingenic,tcu DT binding.
>> + * This header provides clock numbers for the Ingenic OST DT binding.
>>    */
>>   
>>   #ifndef __DT_BINDINGS_CLOCK_INGENIC_OST_H__
>>   #define __DT_BINDINGS_CLOCK_INGENIC_OST_H__
>>   
>> -#define OST_CLK_PERCPU_TIMER	0
>> -#define OST_CLK_GLOBAL_TIMER	1
>> +#define OST_CLK_PERCPU_TIMER	1
>> +#define OST_CLK_GLOBAL_TIMER	0
>> +#define OST_CLK_PERCPU_TIMER0	1
>> +#define OST_CLK_PERCPU_TIMER1	2
>> +#define OST_CLK_PERCPU_TIMER2	3
>> +#define OST_CLK_PERCPU_TIMER3	4
>>   
>>   #endif /* __DT_BINDINGS_CLOCK_INGENIC_OST_H__ */
>> -- 
>> 2.11.0
>>