Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Message-ID: <1525082351.12322.230.camel@mtkswgap22>
Subject: Re: [PATCH v2 2/2] soc: mediatek: add a fixed wait for SRAM stable
From:   Sean Wang <sean.wang@mediatek.com>
To:     Matthias Brugger <matthias.bgg@gmail.com>
CC:     <rjw@rjwysocki.net>, <khilman@baylibre.com>,
        <ulf.hansson@linaro.org>, <linux-mediatek@lists.infradead.org>,
        <linux-pm@vger.kernel.org>, <linux-arm-kernel@lists.infradead.org>,
        <linux-kernel@vger.kernel.org>, "Weiyi Lu" <weiyi.lu@mediatek.com>
Date:   Mon, 30 Apr 2018 17:59:11 +0800
In-Reply-To: <b45efd5d-755c-6d5c-b226-763801e811ea@gmail.com>
References: <2e16481e1477dfae0cfb24568d8111da81d92628.1524472331.git.sean.wang@mediatek.com>
         <0ef8e87ba7156e626d1a1a48388f222ce917099b.1524472331.git.sean.wang@mediatek.com>
         <51840588-96e4-2520-f3d7-a61e74da6814@gmail.com>
         <1524476340.12322.14.camel@mtkswgap22>
         <0ec15c8a-ca90-26c3-1ea6-00bf0d48b62a@gmail.com>
         <1525072080.12322.212.camel@mtkswgap22>
         <b45efd5d-755c-6d5c-b226-763801e811ea@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, 2018-04-30 at 11:10 +0200, Matthias Brugger wrote:
> 
> On 04/30/2018 09:08 AM, Sean Wang wrote:
> > On Fri, 2018-04-27 at 11:46 +0200, Matthias Brugger wrote:
> >> Hi Sean,
> >>
> >> On 04/23/2018 11:39 AM, Sean Wang wrote:
> >>> On Mon, 2018-04-23 at 11:31 +0200, Matthias Brugger wrote:
> >>>>
> >>>> On 04/23/2018 10:36 AM, sean.wang@mediatek.com wrote:
> >>>>> From: Sean Wang <sean.wang@mediatek.com>
> >>>>>
> >>>>> MT7622_POWER_DOMAIN_WB doesn't send an ACK when its managed SRAM becomes
> >>>>> stable, which is not like the behavior the other power domains should
> >>>>> have. Therefore, it's necessary for such a power domain to have a fixed
> >>>>> and well-predefined duration to wait until its managed SRAM can be allowed
> >>>>> to access by all functions running on the top.
> >>>>>
> >>>>> v1 -> v2:
> >>>>>  - use MTK_SCPD_FWAIT_SRAM flag as an indication requiring force waiting.
> >>>>>
> >>>>> Signed-off-by: Sean Wang <sean.wang@mediatek.com>
> >>>>> Cc: Matthias Brugger <matthias.bgg@gmail.com>
> >>>>> Cc: Ulf Hansson <ulf.hansson@linaro.org>
> >>>>> Cc: Weiyi Lu <weiyi.lu@mediatek.com>
> >>>>> ---
> >>>>>  drivers/soc/mediatek/mtk-scpsys.c | 24 ++++++++++++++++++------
> >>>>>  1 file changed, 18 insertions(+), 6 deletions(-)
> >>>>>
> >>>>> diff --git a/drivers/soc/mediatek/mtk-scpsys.c b/drivers/soc/mediatek/mtk-scpsys.c
> >>>>> index b1b45e4..d4f1a63 100644
> >>>>> --- a/drivers/soc/mediatek/mtk-scpsys.c
> >>>>> +++ b/drivers/soc/mediatek/mtk-scpsys.c
> >>>>> @@ -32,6 +32,7 @@
> >>>>>  #define MTK_POLL_TIMEOUT    (jiffies_to_usecs(HZ))
> >>>>>  
> >>>>>  #define MTK_SCPD_ACTIVE_WAKEUP		BIT(0)
> >>>>> +#define MTK_SCPD_FWAIT_SRAM		BIT(1)
> >>>>>  #define MTK_SCPD_CAPS(_scpd, _x)	((_scpd)->data->caps & (_x))
> >>>>>  
> >>>>>  #define SPM_VDE_PWR_CON			0x0210
> >>>>> @@ -237,11 +238,22 @@ static int scpsys_power_on(struct generic_pm_domain *genpd)
> >>>>>  	val &= ~scpd->data->sram_pdn_bits;
> >>>>>  	writel(val, ctl_addr);
> >>>>>  
> >>>>> -	/* wait until SRAM_PDN_ACK all 0 */
> >>>>> -	ret = readl_poll_timeout(ctl_addr, tmp, (tmp & pdn_ack) == 0,
> >>>>> -				 MTK_POLL_DELAY_US, MTK_POLL_TIMEOUT);
> >>>>> -	if (ret < 0)
> >>>>> -		goto err_pwr_ack;
> >>>>> +	/* Either wait until SRAM_PDN_ACK all 0 or have a force wait */
> >>>>> +	if (!MTK_SCPD_CAPS(scpd, MTK_SCPD_FWAIT_SRAM)) {
> >>
> >> After having another look on the patch, could you change the order of the if:
> >> So that we check for the existence of the MTK_SCPD_FWAIT_SRAM and sleep and in
> >> the else branch we to the readl_poll_timeout.
> >>
> >> I think in the future this will make the code easier to understand as you can
> >> easily oversee the '!' negation in the if.
> >>
> >> Regards,
> >> Matthias
> >>
> > 
> > Initial thought on the patch is that I would like to save a branch
> > instruction for a most possibly executed block. Or would it be better to
> > add a compiler to branch prediction information? something like that  
> > 
> >         /* Either wait until SRAM_PDN_ACK all 0 or have a force wait */
> >         if (unlikely(MTK_SCPD_CAPS(scpd, MTK_SCPD_FWAIT_SRAM))) {
> >                 /*
> >                  * Currently, MTK_SCPD_FWAIT_SRAM is necessary only for
> >                  * MT7622_POWER_DOMAIN_WB and thus just a trivial setup
> > is
> >                  * applied here.
> >                  */
> >                 usleep_range(12000, 12100);
> > ...
> >  
> 
> Is this a performance critical path? I thought if you turn on the power domain
> for some peripherals, it does not matter if you need a few CPU cycles more or less.

thanks for your advice

it's not a critical path.

i'll send a new patch according to the result.

> Regards,
> Matthias