[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1531209943.3163.31.camel@pengutronix.de>
Date: Tue, 10 Jul 2018 10:05:43 +0200
From: Lucas Stach <l.stach@...gutronix.de>
To: Ulf Hansson <ulf.hansson@...aro.org>,
Leonard Crestez <leonard.crestez@....com>
Cc: Shawn Guo <shawnguo@...nel.org>,
"Rafael J. Wysocki" <rjw@...ysocki.net>,
Fabio Estevam <fabio.estevam@....com>,
Andrey Smirnov <andrew.smirnov@...il.com>,
Robin Gong <yibin.gong@....com>,
Dong Aisheng <aisheng.dong@....com>,
Linux PM <linux-pm@...r.kernel.org>,
Linux ARM <linux-arm-kernel@...ts.infradead.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
dl-linux-imx <linux-imx@....com>,
Sascha Hauer <kernel@...gutronix.de>
Subject: Re: [PATCH v3] soc: imx: gpc: Turn PU domain on/off in sleep on 6qp
Am Dienstag, den 10.07.2018, 00:27 +0200 schrieb Ulf Hansson:
> > On 6 July 2018 at 18:46, Leonard Crestez <leonard.crestez@....com> wrote:
> > On imx6qp power gating on the PU domain is disabled because of errata
> > ERR009619. However power gating during suspend/resume can still be
> > performed.
>
> Sounds a bit hand wavy. Exactly why can it be done during system suspend/resume?
Power gating this domain in normal operation may introduce a glitch in
unrelated modules, potentially disturbing a running display scanout.
This is not a concern when the whole system enters suspend state, so
it's safe to do in this case.
Regards,
Lucas
> >
> > Enable this by implementing SLEEP_PM_OPS in imx_pgc_power_domain_driver.
> >
> > In order to ensure correct ordering add device links from devices inside
> > the PU domain to the pgc platform_device.
>
> In genpd, the PM domain gets powered off after all device's
> corresponding ->suspend_noirq() callback has been invoked - and it
> also considers if there are some wakeup settings enabled, possibly
> avoiding to do the power off.
>
> It sound like the imx's power domain driver ->power_off|on() callback
> should deal with this, to make get a consistent behavior. I understand
> that you currently can't tell under what circumstances the
> ->power_off|on() callbacks are being invoked, that may need some
> changes in genpd.
>
> Currently we have a flag, GENPD_FLAG_ALWAYS_ON which makes genpd to
> always prevent power off. Perhaps we should add a new flag
> "GENPD_FLAG_RUNTIME_ON", which makes genpd to call the
> ->power_on|off() callbacks, but only during system wide PM.
>
> Thoughts?
Yes, this might work for this case.
Regards,
Lucas
Powered by blists - more mailing lists