Re: [RFC v6 01/23] clocksource: sh_cmt: compute rate before registration again

[Nicolai Stange]

Thomas Gleixner <tglx@xxxxxxxxxxxxx> writes:

> On Fri, 9 Sep 2016, Nicolai Stange wrote:
>> @@ -339,17 +339,14 @@ static int sh_cmt_enable(struct sh_cmt_channel *ch, unsigned long *rate)
>>  	sh_cmt_start_stop_ch(ch, 0);
>>  
>>  	/* configure channel, periodic mode and maximum timeout */
>> -	if (ch->cmt->info->width == 16) {
>> -		*rate = clk_get_rate(ch->cmt->clk) / 512;
>> +	if (ch->cmt->info->width == 16)
>>  		sh_cmt_write_cmcsr(ch, SH_CMT16_CMCSR_CMIE |
>>  				   SH_CMT16_CMCSR_CKS512);
>> -	} else {
>> -		*rate = clk_get_rate(ch->cmt->clk) / 8;
>> +	else
>>  		sh_cmt_write_cmcsr(ch, SH_CMT32_CMCSR_CMM |
>>  				   SH_CMT32_CMCSR_CMTOUT_IE |
>>  				   SH_CMT32_CMCSR_CMR_IRQ |
>>  				   SH_CMT32_CMCSR_CKS_RCLK8);
>> -	}
>
> Removing the braces here is just wrong.

Completely convinced, will redo.



>> @@ -824,6 +810,12 @@ static int sh_cmt_register_clockevent(struct sh_cmt_channel *ch,
>>  	ced->suspend = sh_cmt_clock_event_suspend;
>>  	ced->resume = sh_cmt_clock_event_resume;
>>  
>> +	/* TODO: calculate good shift from rate and counter bit width */
>> +	ced->shift = 32;
>> +	ced->mult = div_sc(ch->cmt->rate, NSEC_PER_SEC, ced->shift);
>
>
> Errm. clockevents_calc_mult_shift()

Or even clockevents_config_and_register()?

The reason why I left this mult/shift initialization as is (it has only
been moved from sh_cmt_clock_event_start() by this patch) is that I was
unsure about why this conversion hadn't been done before. That TODO
comment seems to imply that a shift value as calculated by
clockevents_config() isn't always the perfect choice...

However, do you want me to
a.) simply change this patch to use clockevents_config_and_register(),
b.) add a separate patch to this series doing this cleanup,
c.) send such a patch on its own
d.) or leave this initalization as is?

Thanks,

Nicolai Stange