Re: [linux-sunxi] [PATCH v2 2/2] media: cedrus: Implement runtime PM

[Maxime Ripard]

On Sat, May 09, 2020 at 03:03:55PM -0500, Samuel Holland wrote:
> On 5/5/20 7:53 AM, Ezequiel Garcia wrote:
> > On Wed, 22 Apr 2020 at 01:00, Samuel Holland <samuel@xxxxxxxxxxxx> wrote:
> >>
> >> This allows the VE clocks and PLL_VE to be disabled most of the time.
> >> A runtime PM reference is held while streaming.
> >>
> >> Originally-by: Jernej Åkrabec <jernej.skrabec@xxxxxxxxx>
> > 
> > Originally-by is not documented, perhaps just go with Signed-off-by,
> > as Jernej suggested.
> > 
> >> Signed-off-by: Samuel Holland <samuel@xxxxxxxxxxxx>
> >> ---
> >>
> >> v2: moved PM reference to cedrus_{start,stop}_streaming, based on an
> >>     earlier patch by Jernej Skrabec. Removes the need for autosuspend.
> >>     I tested this with running 2x v4l2-request-test in parallel.
> >>
> >> ---
> >>  drivers/staging/media/sunxi/cedrus/cedrus.c   |   7 ++
> >>  .../staging/media/sunxi/cedrus/cedrus_hw.c    | 106 ++++++++++++------
> >>  .../staging/media/sunxi/cedrus/cedrus_hw.h    |   3 +
> >>  .../staging/media/sunxi/cedrus/cedrus_video.c |  33 ++++--
> >>  4 files changed, 104 insertions(+), 45 deletions(-)
> >>
> >> diff --git a/drivers/staging/media/sunxi/cedrus/cedrus.c b/drivers/staging/media/sunxi/cedrus/cedrus.c
> >> index 05a85517ff60..bc27f9430eeb 100644
> >> --- a/drivers/staging/media/sunxi/cedrus/cedrus.c
> >> +++ b/drivers/staging/media/sunxi/cedrus/cedrus.c
> >> @@ -16,6 +16,7 @@
> >>  #include <linux/platform_device.h>
> >>  #include <linux/module.h>
> >>  #include <linux/of.h>
> >> +#include <linux/pm.h>
> >>
> >>  #include <media/v4l2-device.h>
> >>  #include <media/v4l2-ioctl.h>
> >> @@ -551,12 +552,18 @@ static const struct of_device_id cedrus_dt_match[] = {
> >>  };
> >>  MODULE_DEVICE_TABLE(of, cedrus_dt_match);
> >>
> >> +static const struct dev_pm_ops cedrus_dev_pm_ops = {
> >> +       SET_RUNTIME_PM_OPS(cedrus_hw_suspend,
> >> +                          cedrus_hw_resume, NULL)
> >> +};
> >> +
> >>  static struct platform_driver cedrus_driver = {
> >>         .probe          = cedrus_probe,
> >>         .remove         = cedrus_remove,
> >>         .driver         = {
> >>                 .name           = CEDRUS_NAME,
> >>                 .of_match_table = of_match_ptr(cedrus_dt_match),
> >> +               .pm             = &cedrus_dev_pm_ops,
> >>         },
> >>  };
> >>  module_platform_driver(cedrus_driver);
> >> diff --git a/drivers/staging/media/sunxi/cedrus/cedrus_hw.c b/drivers/staging/media/sunxi/cedrus/cedrus_hw.c
> >> index daf5f244f93b..1744e6fcc999 100644
> >> --- a/drivers/staging/media/sunxi/cedrus/cedrus_hw.c
> >> +++ b/drivers/staging/media/sunxi/cedrus/cedrus_hw.c
> >> @@ -19,6 +19,7 @@
> >>  #include <linux/dma-mapping.h>
> >>  #include <linux/interrupt.h>
> >>  #include <linux/clk.h>
> >> +#include <linux/pm_runtime.h>
> >>  #include <linux/regmap.h>
> >>  #include <linux/reset.h>
> >>  #include <linux/soc/sunxi/sunxi_sram.h>
> >> @@ -140,6 +141,64 @@ static irqreturn_t cedrus_irq(int irq, void *data)
> >>         return IRQ_HANDLED;
> >>  }
> >>
> >> +int cedrus_hw_suspend(struct device *device)
> >> +{
> >> +       struct cedrus_dev *dev = dev_get_drvdata(device);
> >> +
> >> +       reset_control_assert(dev->rstc);
> >> +
> >> +       clk_disable_unprepare(dev->ram_clk);
> >> +       clk_disable_unprepare(dev->mod_clk);
> >> +       clk_disable_unprepare(dev->ahb_clk);
> >> +
> > 
> > You can use clk_bulk API here.
> 
> Since this change is already tested, I'd prefer to do that as a separate
> patch.

Given that those three clocks are also pretty different from a semantic point of
view, I'm not sure it's wise to switch to the bulk API anyway.

Maxime

Attachment: signature.asc
Description: PGP signature