From: Ulf Hansson <email@example.com>
To: Dmitry Osipenko <firstname.lastname@example.org>
Cc: "Viresh Kumar" <email@example.com>,
"Thierry Reding" <firstname.lastname@example.org>,
"Jonathan Hunter" <email@example.com>,
"Viresh Kumar" <firstname.lastname@example.org>,
"Stephen Boyd" <email@example.com>,
"Peter De Schrijver" <firstname.lastname@example.org>,
"Mikko Perttunen" <email@example.com>,
"Peter Chen" <firstname.lastname@example.org>,
"Mark Brown" <email@example.com>,
"Lee Jones" <firstname.lastname@example.org>,
"Uwe Kleine-König" <email@example.com>,
"Nishanth Menon" <firstname.lastname@example.org>,
"Vignesh Raghavendra" <email@example.com>,
"Richard Weinberger" <firstname.lastname@example.org>,
"Miquel Raynal" <email@example.com>,
"Lucas Stach" <firstname.lastname@example.org>, "Stefan Agner" <email@example.com>,
"Adrian Hunter" <firstname.lastname@example.org>,
"Mauro Carvalho Chehab" <email@example.com>,
"Rob Herring" <firstname.lastname@example.org>,
"Michael Turquette" <email@example.com>,
"Linux Kernel Mailing List" <firstname.lastname@example.org>,
"Linux PM" <email@example.com>,
"Linux USB List" <firstname.lastname@example.org>,
"Linux Media Mailing List" <email@example.com>,
Subject: Re: [PATCH v8 01/34] opp: Add dev_pm_opp_sync() helper
Date: Fri, 20 Aug 2021 14:42:08 +0200 [thread overview]
Message-ID: <CAPDyKFqSsAk8a5CTNpRT2z4Wvf8BehJKDbVhUKfHc2Jzj7aTNA@mail.gmail.com> (raw)
On Thu, 19 Aug 2021 at 21:35, Dmitry Osipenko <firstname.lastname@example.org> wrote:
> 19.08.2021 16:07, Ulf Hansson пишет:
> > On Wed, 18 Aug 2021 at 17:43, Dmitry Osipenko <email@example.com> wrote:
> >> 18.08.2021 13:08, Ulf Hansson пишет:
> >>> On Wed, 18 Aug 2021 at 11:50, Viresh Kumar <firstname.lastname@example.org> wrote:
> >>>> On 18-08-21, 11:41, Ulf Hansson wrote:
> >>>>> On Wed, 18 Aug 2021 at 11:14, Viresh Kumar <email@example.com> wrote:
> >>>>>> What we need here is just configure. So something like this then:
> >>>>>> - genpd->get_performance_state()
> >>>>>> -> dev_pm_opp_get_current_opp() //New API
> >>>>>> -> dev_pm_genpd_set_performance_state(dev, current_opp->pstate);
> >>>>>> This can be done just once from probe() then.
> >>>>> How would dev_pm_opp_get_current_opp() work? Do you have a suggestion?
> >>>> The opp core already has a way of finding current OPP, that's what
> >>>> Dmitry is trying to use here. It finds it using clk_get_rate(), if
> >>>> that is zero, it picks the lowest freq possible.
> >>>>> I am sure I understand the problem. When a device is getting probed,
> >>>>> it needs to consume power, how else can the corresponding driver
> >>>>> successfully probe it?
> >>>> Dmitry can answer that better, but a device doesn't necessarily need
> >>>> to consume energy in probe. It can consume bus clock, like APB we
> >>>> have, but the more energy consuming stuff can be left disabled until
> >>>> the time a user comes up. Probe will just end up registering the
> >>>> driver and initializing it.
> >>> That's perfectly fine, as then it's likely that it won't vote for an
> >>> OPP, but can postpone that as well.
> >>> Perhaps the problem is rather that the HW may already carry a non-zero
> >>> vote made from a bootloader. If the consumer driver tries to clear
> >>> that vote (calling dev_pm_opp_set_rate(dev, 0), for example), it would
> >>> still not lead to any updates of the performance state in genpd,
> >>> because genpd internally has initialized the performance-state to
> >>> zero.
> >> We don't need to discover internal SoC devices because we use
> >> device-tree on ARM. For most devices power isn't required at a probe
> >> time because probe function doesn't touch h/w at all, thus devices are
> >> left in suspended state after probe.
> >> We have three components comprising PM on Tegra:
> >> 1. Power gate
> >> 2. Clock state
> >> 3. Voltage state
> >> GENPD on/off represents the 'power gate'.
> >> Clock and reset are controlled by device drivers using clk and rst APIs.
> >> Voltage state is represented by GENPD's performance level.
> >> GENPD core assumes that at a first rpm-resume of a consumer device, its
> >> genpd_performance=0. Not true for Tegra because h/w of the device is
> >> preconfigured to a non-zero perf level initially, h/w may not support
> >> zero level at all.
> > I think you may be misunderstanding genpd's behaviour around this, but
> > let me elaborate.
> > In genpd_runtime_resume(), we try to restore the performance state for
> > the device that genpd_runtime_suspend() *may* have dropped earlier.
> > That means, if genpd_runtime_resume() is called prior
> > genpd_runtime_suspend() for the first time, it means that
> > genpd_runtime_resume() will *not* restore a performance state, but
> > instead just leave the performance state as is for the device (see
> > genpd_restore_performance_state()).
> > In other words, a consumer driver may use the following sequence to
> > set an initial performance state for the device during ->probe():
> > ...
> > rate = clk_get_rate()
> > dev_pm_opp_set_rate(rate)
> > pm_runtime_enable()
> > pm_runtime_resume_and_get()
> > ...
> > Note that, it's the consumer driver's responsibility to manage device
> > specific resources, in its ->runtime_suspend|resume() callbacks.
> > Typically that means dealing with clock gating/ungating, for example.
> > In the other scenario where a consumer driver prefers to *not* call
> > pm_runtime_resume_and_get() in its ->probe(), because it doesn't need
> > to power on the device to complete probing, then we don't want to vote
> > for an OPP at all - and we also want the performance state for the
> > device in genpd to be set to zero. Correct?
> > Is this the main problem you are trying to solve, because I think this
> > doesn't work out of the box as of today?
> The main problem is that the restored performance state is zero for the
> first genpd_runtime_resume(), while it's not zero from the h/w perspective.
This should not be a problem, but can be handled by the consumer driver.
genpd_runtime_resume() calls genpd_restore_performance_state() to
restore a performance state for the device. However, in the scenario
you describe, "gpd_data->rpm_pstate" is zero, which makes
genpd_restore_performance_state() to just leave the device's
performance state as is - it will *not* restore the performance state
To make the consumer driver deal with this, it would need to call
dev_pm_opp_set_rate() from within its ->runtime_resume() callback.
> > There is another concern though, but perhaps it's not a problem after
> > all. Viresh told us that dev_pm_opp_set_rate() may turn on resources
> > like clock/regulators. That could certainly be problematic, in
> > particular if the device and its genpd have OPP tables associated with
> > it and the consumer driver wants to follow the above sequence in
> > probe.
> dev_pm_opp_set_rate() won't enable clocks and regulators, but it may
> change the clock rate and voltage. This is also platform/driver specific
> because it's up to OPP user how to configure OPP table. On Tegra we only
> assign clock to OPP table, regulators are unused.
> > Viresh, can you please chime in here and elaborate on some of the
> > magic happening behind dev_pm_opp_set_rate() API - is there a problem
> > here or not?
> >> GENPD core assumes that consumer devices can work at any performance
> >> level. Not true for Tegra because voltage needs to be set in accordance
> >> to the clock rate before clock is enabled, otherwise h/w won't work
> >> properly, perhaps clock may be unstable or h/w won't be latching.
> > Correct. Genpd relies on the callers to use the OPP framework if there
> > are constraints like you describe above.
> > That said, it's not forbidden for a consumer driver to call
> > dev_pm_genpd_set_performance_state() directly, but then it better
> > knows exactly what it's doing.
> >> Performance level should be set to 0 while device is suspended.
> > Do you mean system suspend or runtime suspend? Or both?
> Runtime suspend.
Alright. So that's already taken care of for us in genpd_runtime_suspend().
Or perhaps you have discovered some problem with this?
> >> Performance level needs to be bumped on rpm-resume of a device in
> >> accordance to h/w state before hardware is enabled.
> > Assuming there was a performance state set for the device when
> > genpd_runtime_suspend() was called, genpd_runtime_resume() will
> > restore that state according to the sequence you described.
> What do you think about adding API that will allow drivers to explicitly
> set the restored performance state of a power domain?
> Another option could be to change the GENPD core, making it to set the
> rpm_pstate when dev_pm_genpd_set_performance_state(dev) is invoked and
> device is rpm-suspended, instead of calling the
> genpd->set_performance_state callback.
> Then drivers will be able to sync the perf state at a probe time.
> What do you think?
I don't think it's needed, see my reply earlier above. However your
change touches another problem though, see below.
> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
> index a934c679e6ce..cc15ab9eacc9 100644
> --- a/drivers/base/power/domain.c
> +++ b/drivers/base/power/domain.c
> @@ -435,7 +435,7 @@ static void genpd_restore_performance_state(struct
> device *dev,
> int dev_pm_genpd_set_performance_state(struct device *dev, unsigned int
> struct generic_pm_domain *genpd;
> - int ret;
> + int ret = 0;
> genpd = dev_to_genpd_safe(dev);
> if (!genpd)
> @@ -446,7 +446,10 @@ int dev_pm_genpd_set_performance_state(struct
> device *dev, unsigned int state)
> return -EINVAL;
> - ret = genpd_set_performance_state(dev, state);
> + if (pm_runtime_suspended(dev))
> + dev_gpd_data(dev)->rpm_pstate = state;
> + else
> + ret = genpd_set_performance_state(dev, state);
This doesn't work for all cases. For example, when a consumer driver
deploys runtime PM support in its ->probe() according to the below
We need to call genpd_set_performance_state() independently of whether
the device is runtime suspended or not.
Although, it actually seems like good idea to update
dev_gpd_data(dev)->rpm_pstate = state here, as to make sure
genpd_runtime_resume() doesn't restore an old/invalid value that was
saved while dropping the performance state vote for the device in
Let me send a patch for this shortly, to close this window of a possible error.
> return ret;
next prev parent reply other threads:[~2021-08-20 12:42 UTC|newest]
Thread overview: 119+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-17 1:27 [PATCH v8 00/34] NVIDIA Tegra power management patches for 5.16 Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 01/34] opp: Add dev_pm_opp_sync() helper Dmitry Osipenko
2021-08-17 7:55 ` Viresh Kumar
2021-08-17 15:49 ` Dmitry Osipenko
2021-08-18 3:55 ` Viresh Kumar
2021-08-18 4:12 ` Dmitry Osipenko
2021-08-18 4:29 ` Dmitry Osipenko
2021-08-18 4:30 ` Dmitry Osipenko
2021-08-18 4:34 ` Viresh Kumar
2021-08-18 4:31 ` Viresh Kumar
2021-08-18 4:37 ` Dmitry Osipenko
2021-08-18 4:53 ` Viresh Kumar
2021-08-18 5:21 ` Dmitry Osipenko
2021-08-18 5:58 ` Viresh Kumar
2021-08-18 6:00 ` Viresh Kumar
2021-08-18 6:22 ` Dmitry Osipenko
2021-08-18 6:27 ` Viresh Kumar
2021-08-18 8:29 ` Ulf Hansson
2021-08-18 9:14 ` Viresh Kumar
2021-08-18 9:41 ` Ulf Hansson
2021-08-18 9:42 ` Ulf Hansson
2021-08-18 9:50 ` Viresh Kumar
2021-08-18 10:08 ` Ulf Hansson
2021-08-18 15:43 ` Dmitry Osipenko
2021-08-18 15:46 ` Dmitry Osipenko
2021-08-19 13:07 ` Ulf Hansson
2021-08-19 19:35 ` Dmitry Osipenko
2021-08-20 5:07 ` Viresh Kumar
2021-08-20 12:42 ` Ulf Hansson [this message]
2021-08-21 17:34 ` Dmitry Osipenko
2021-08-23 10:46 ` Ulf Hansson
2021-08-23 15:54 ` Dmitry Osipenko
2021-08-18 15:55 ` Dmitry Osipenko
2021-08-19 6:16 ` Viresh Kumar
2021-08-19 14:55 ` Ulf Hansson
2021-08-20 5:18 ` Viresh Kumar
2021-08-20 12:57 ` Ulf Hansson
2021-08-23 20:24 ` Dmitry Osipenko
2021-08-24 3:04 ` Viresh Kumar
2021-08-22 18:35 ` Dmitry Osipenko
2021-08-25 15:41 ` Dmitry Osipenko
2021-08-26 2:54 ` Viresh Kumar
2021-08-26 2:55 ` Viresh Kumar
2021-08-17 1:27 ` [PATCH v8 02/34] soc/tegra: pmc: Disable PMC state syncing Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 03/34] soc/tegra: Don't print error message when OPPs not available Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 04/34] soc/tegra: Add devm_tegra_core_dev_init_opp_table_simple() Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 05/34] soc/tegra: Use dev_pm_opp_sync() Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 06/34] dt-bindings: clock: tegra-car: Document new tegra-clocks sub-node Dmitry Osipenko
2021-08-18 1:15 ` Rob Herring
2021-08-18 1:44 ` Dmitry Osipenko
2021-08-18 13:52 ` Thierry Reding
2021-08-18 15:04 ` Dmitry Osipenko
2021-08-18 13:59 ` Thierry Reding
2021-08-18 15:05 ` Dmitry Osipenko
2021-08-18 16:39 ` Thierry Reding
2021-08-18 16:57 ` Dmitry Osipenko
2021-08-18 17:16 ` Dmitry Osipenko
2021-08-19 16:31 ` Thierry Reding
2021-08-19 22:20 ` Dmitry Osipenko
2021-08-20 2:51 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 07/34] clk: tegra: Support runtime PM and power domain Dmitry Osipenko
2021-08-18 14:07 ` Thierry Reding
2021-08-18 15:05 ` Dmitry Osipenko
2021-08-18 16:42 ` Thierry Reding
2021-08-18 17:11 ` Dmitry Osipenko
2021-08-19 16:54 ` Thierry Reding
2021-08-19 22:09 ` Dmitry Osipenko
2021-08-20 11:42 ` Thierry Reding
2021-08-20 13:08 ` Ulf Hansson
2021-08-21 17:45 ` Dmitry Osipenko
2021-08-23 14:33 ` Thierry Reding
2021-08-23 18:54 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 08/34] dt-bindings: host1x: Document OPP and power domain properties Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 09/34] dt-bindings: host1x: Document Memory Client resets of Host1x, GR2D and GR3D Dmitry Osipenko
2021-08-18 1:16 ` Rob Herring
2021-08-18 1:37 ` Dmitry Osipenko
2021-08-18 2:04 ` Dmitry Osipenko
2021-08-18 2:07 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 10/34] gpu: host1x: Add host1x_channel_stop() Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 11/34] gpu: host1x: Add runtime PM and OPP support Dmitry Osipenko
2021-08-17 12:04 ` Ulf Hansson
2021-08-17 14:02 ` Thierry Reding
2021-08-18 8:35 ` Ulf Hansson
2021-08-18 17:24 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 12/34] drm/tegra: dc: Support OPP and SoC core voltage scaling Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 13/34] drm/tegra: hdmi: Add OPP support Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 14/34] drm/tegra: gr2d: Support power management Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 15/34] drm/tegra: gr3d: " Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 16/34] drm/tegra: vic: Support system suspend Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 17/34] usb: chipidea: tegra: Add runtime PM and OPP support Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 18/34] bus: tegra-gmi: " Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 19/34] pwm: tegra: " Dmitry Osipenko
2021-08-19 13:21 ` Thierry Reding
2021-08-19 14:04 ` Ulf Hansson
2021-08-19 16:17 ` Thierry Reding
2021-08-17 1:27 ` [PATCH v8 20/34] mmc: sdhci-tegra: " Dmitry Osipenko
2021-08-19 17:03 ` Thierry Reding
2021-08-19 22:37 ` Dmitry Osipenko
2021-08-20 11:35 ` Thierry Reding
2021-08-25 9:45 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 21/34] mtd: rawnand: tegra: " Dmitry Osipenko
2021-08-17 8:41 ` Miquel Raynal
2021-08-17 1:27 ` [PATCH v8 22/34] spi: tegra20-slink: Add " Dmitry Osipenko
2021-08-17 12:22 ` Mark Brown
2021-08-17 15:53 ` Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 23/34] media: dt: bindings: tegra-vde: Convert to schema Dmitry Osipenko
2021-08-18 1:17 ` Rob Herring
2021-08-17 1:27 ` [PATCH v8 24/34] media: dt: bindings: tegra-vde: Document OPP and power domain Dmitry Osipenko
2021-08-18 1:17 ` Rob Herring
2021-08-17 1:27 ` [PATCH v8 25/34] media: staging: tegra-vde: Support generic power domain and OPP Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 26/34] soc/tegra: fuse: Add OPP support Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 27/34] soc/tegra: fuse: Reset hardware Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 28/34] soc/tegra: regulators: Prepare for suspend Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 29/34] soc/tegra: pmc: Enable core domain support for Tegra20 and Tegra30 Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 30/34] ARM: tegra: Add OPP tables and power domains to Tegra20 device-trees Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 31/34] ARM: tegra: Add OPP tables and power domains to Tegra30 device-trees Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 32/34] ARM: tegra: Add Memory Client resets to Tegra20 GR2D, GR3D and Host1x Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 33/34] ARM: tegra: Add Memory Client resets to Tegra30 " Dmitry Osipenko
2021-08-17 1:27 ` [PATCH v8 34/34] ARM: tegra20/30: Disable unused host1x hardware Dmitry Osipenko
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).