Re: [PATCH v1 1/2] scsi: ufs: add required delay after gating reference clock

From: Can Guo <cang@codeaurora.org>
To: Stanley Chu <stanley.chu@mediatek.com>
Cc: linux-scsi@vger.kernel.org, martin.petersen@oracle.com,
	avri.altman@wdc.com, alim.akhtar@samsung.com, jejb@linux.ibm.com,
	beanhuo@micron.com, asutoshd@codeaurora.org,
	matthias.bgg@gmail.com, bvanassche@acm.org,
	linux-mediatek@lists.infradead.org,
	linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org, kuohong.wang@mediatek.com,
	peter.wang@mediatek.com, chun-hung.wu@mediatek.com,
	andy.teng@mediatek.com
Subject: Re: [PATCH v1 1/2] scsi: ufs: add required delay after gating reference clock
Date: Wed, 19 Feb 2020 10:35:27 +0800	[thread overview]
Message-ID: <a8cd5beee0a1e12a40da752c6cd9b5de@codeaurora.org> (raw)
In-Reply-To: <56c1fc80919491d058d904fcc7301835@codeaurora.org>

Hi Stanely,

On 2020-02-17 21:42, Can Guo wrote:
> On 2020-02-17 21:34, Stanley Chu wrote:
>> Hi Can,
>> 
>> On Mon, 2020-02-17 at 21:22 +0800, Can Guo wrote:
>>> On 2020-02-17 21:12, Stanley Chu wrote:
>>> > Hi Can,
>>> >
>>> >
>>> >> >  			} else if (!on && clki->enabled) {
>>> >> >  				clk_disable_unprepare(clki->clk);
>>> >> > +				wait_us = hba->dev_info.clk_gating_wait_us;
>>> >> > +				if (ref_clk && wait_us)
>>> >> > +					usleep_range(wait_us, wait_us + 10);
>>> >>
>>> >> Hi St,anley,
>>> >>
>>> >> If wait_us is 1us, it would be inappropriate to use usleep_range()
>>> >> here.
>>> >> You have checks of the delay in patch #2, but why it is not needed
>>> >> here?
>>> >>
>>> >> Thanks,
>>> >> Can Guo.
>>> >
>>> > You are right. I could make that delay checking as common function so
>>> > it
>>> > can be used here as well to cover all possible values.
>>> >
>>> > Thanks for suggestion.
>>> > Stanley
>>> 
>>> Hi Stanley,
>>> 
>>> One more thing, as in patch #2, you have already added delays in your
>>> ufshcd_vops_setup_clocks(OFF, PRE_CHANGE) path, plus this delay here,
>>> don't you delay for 2*bRefClkGatingWaitTime in ufshcd_setup_clocks()?
>>> As the delay added in your vops also delays the actions of turning
>>> off all the other clocks in ufshcd_setup_clocks(), you don't need the
>>> delay here again, do you agree?
>> 
>> MediaTek driver is not using reference clocks named as "ref_clk" 
>> defined
>> in device tree, thus the delay specific for "ref_clk" in
>> ufshcd_setup_clocks() will not be applied in MediaTek platform.
>> 
>> This patch is aimed to add delay for this kind of "ref_clk" used by 
>> any
>> future vendors.
>> 
>> Anyway thanks for the reminding : )
>> 
>>> 
>>> Thanks,
>>> Can Guo.
>> 
>> 
>> Thanks,
>> Stanley
> 
> Hi Stanley,
> 
> Then we are unluckily hit by this change. We have ref_clk in DT, thus
> this change would add unwanted delays to our platforms. but still we
> disable device's ref_clk in vops. :)
> 
> Could you please hold on patch #1 first? I need sometime to have a
> dicussion with my colleagues on this.
> 
> Thanks.
> Can Guo.

Since we all need this delay here, how about put the delay in the
entrence of ufshcd_setup_clocks(), before vops_setup_clocks()?
If so, we can remove all the delays we added in our vops since the
delay anyways delays everything inside ufshcd_setup_clocks().

Meanwhile, if you want to modify the delay
(hba->dev_info.clk_gating_wait_us) for some reasons, say for specific
UFS devices, you still can do it in vops_apply_dev_quirks().

What do you say?

Thanks,
Can Guo.