Hi Ulf,
On Thu, Dec 21, 2017 at 01:21:51PM +0100, Ulf Hansson wrote:
> On 20 December 2017 at 11:50, Maxime Ripard
> <maxime.ripard@xxxxxxxxxxxxxxxxxx> wrote:
> > So far, even if our card was not in use, we didn't shut down our main
> > clock, which meant that it was still output on the MMC bus.
> >
> > While this obviously means that we could save some power there, it also
> > created issues when it comes with EMC control since we'll have a perfect
> > peak at the card clock rate.
> >
> > Let's implement runtime_pm with autosuspend so that we will shut down the
> > clock when it's not been in use for quite some time.
> >
> > Signed-off-by: Maxime Ripard <maxime.ripard@xxxxxxxxxxxxxxxxxx>
> > ---
> > drivers/mmc/host/sunxi-mmc.c | 90 ++++++++++++++++++++++++-------------
> > 1 file changed, 60 insertions(+), 30 deletions(-)
> >
> > diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c
> > index 3ce46ebd3488..c6a0bd0e0476 100644
> > --- a/drivers/mmc/host/sunxi-mmc.c
> > +++ b/drivers/mmc/host/sunxi-mmc.c
> > @@ -35,6 +35,7 @@
> > #include <linux/of_gpio.h>
> > #include <linux/of_platform.h>
> > #include <linux/platform_device.h>
> > +#include <linux/pm_runtime.h>
> > #include <linux/regulator/consumer.h>
> > #include <linux/reset.h>
> > #include <linux/scatterlist.h>
> > @@ -1217,29 +1218,11 @@ static int sunxi_mmc_resource_request(struct sunxi_mmc_host *host,
> > return ret;
> > }
> >
> > - ret = clk_prepare_enable(host->clk_mmc);
> > - if (ret) {
> > - dev_err(&pdev->dev, "Enable mmc clk err %d\n", ret);
> > - goto error_disable_clk_ahb;
> > - }
> > -
> > - ret = clk_prepare_enable(host->clk_output);
> > - if (ret) {
> > - dev_err(&pdev->dev, "Enable output clk err %d\n", ret);
> > - goto error_disable_clk_mmc;
> > - }
> > -
> > - ret = clk_prepare_enable(host->clk_sample);
> > - if (ret) {
> > - dev_err(&pdev->dev, "Enable sample clk err %d\n", ret);
> > - goto error_disable_clk_output;
> > - }
> > -
>
> Actually, I think you should keep all the above. Perhaps you may want
> to move it to a separate helper function though, which the
> ->runtime_resume() callbacks can invoke as well.
>
> More reasons to why, see the comment in the ->probe() function.
>
> > if (!IS_ERR(host->reset)) {
> > ret = reset_control_reset(host->reset);
> > if (ret) {
> > dev_err(&pdev->dev, "reset err %d\n", ret);
> > - goto error_disable_clk_sample;
> > + goto error_disable_clk_ahb;
> > }
> > }
> >
> > @@ -1258,12 +1241,6 @@ static int sunxi_mmc_resource_request(struct sunxi_mmc_host *host,
> > error_assert_reset:
> > if (!IS_ERR(host->reset))
> > reset_control_assert(host->reset);
> > -error_disable_clk_sample:
> > - clk_disable_unprepare(host->clk_sample);
> > -error_disable_clk_output:
> > - clk_disable_unprepare(host->clk_output);
> > -error_disable_clk_mmc:
> > - clk_disable_unprepare(host->clk_mmc);
> > error_disable_clk_ahb:
> > clk_disable_unprepare(host->clk_ahb);
> > return ret;
> > @@ -1280,6 +1257,7 @@ static int sunxi_mmc_probe(struct platform_device *pdev)
> > dev_err(&pdev->dev, "mmc alloc host failed\n");
> > return -ENOMEM;
> > }
> > + platform_set_drvdata(pdev, mmc);
> >
> > host = mmc_priv(mmc);
> > host->mmc = mmc;
> > @@ -1340,12 +1318,16 @@ static int sunxi_mmc_probe(struct platform_device *pdev)
> > if (ret)
> > goto error_free_dma;
> >
> > + pm_runtime_set_autosuspend_delay(&pdev->dev, 50);
> > + pm_runtime_use_autosuspend(&pdev->dev);
> > + pm_runtime_enable(&pdev->dev);
> > +
>
> This means in case you don't have CONFIG_PM set when compiling this
> driver, the clocks will never be enabled using runtime PM.
>
> I think it's good practice to deal with this, therefore I think you
> should enable the clocks as before, and instead indicate that the
> device is already runtime resumed.
>
> In other words, before you call pm_runtime_enable(), invoke
> pm_runtime_set_active().
That's a very good point, I'll do that in the next iteration.
> > ret = mmc_add_host(mmc);
> > if (ret)
> > goto error_free_dma;
> >
> > dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq);
> > - platform_set_drvdata(pdev, mmc);
> > +
> > return 0;
> >
> > error_free_dma:
> > @@ -1361,27 +1343,75 @@ static int sunxi_mmc_remove(struct platform_device *pdev)
> > struct sunxi_mmc_host *host = mmc_priv(mmc);
> >
> > mmc_remove_host(mmc);
> > + pm_runtime_force_suspend(&pdev->dev);
>
> Do you need the clocks to be enabled, while calling disable_irq() and
> sunxi_mmc_reset_host()?
>
> In such case you need to call pm_runtime_get_sync() here. Then move
> pm_runtime_force_suspend() a few lines below, and later call
> pm_runtime_put_noidle().
We don't for disable_irq, but we need it for
sunxi_mmc_reset_host. I'll do what you suggest.
> > disable_irq(host->irq);
> > sunxi_mmc_reset_host(host);
> >
> > if (!IS_ERR(host->reset))
> > reset_control_assert(host->reset);
> >
> > - clk_disable_unprepare(host->clk_sample);
> > + dma_free_coherent(&pdev->dev, PAGE_SIZE, host->sg_cpu, host->sg_dma);
> > + mmc_free_host(mmc);
> > +
> > + return 0;
> > +}
> > +
> > +static int sunxi_mmc_runtime_resume(struct device *dev)
> > +{
> > + struct mmc_host *mmc = dev_get_drvdata(dev);
> > + struct sunxi_mmc_host *host = mmc_priv(mmc);
> > + int ret;
> > +
> > + ret = clk_prepare_enable(host->clk_mmc);
> > + if (ret) {
> > + dev_err(dev, "Enable mmc clk err %d\n", ret);
> > + return ret;
> > + }
> > +
> > + ret = clk_prepare_enable(host->clk_output);
> > + if (ret) {
> > + dev_err(dev, "Enable output clk err %d\n", ret);
> > + goto error_disable_clk_mmc;
> > + }
> > +
> > + ret = clk_prepare_enable(host->clk_sample);
> > + if (ret) {
> > + dev_err(dev, "Enable sample clk err %d\n", ret);
> > + goto error_disable_clk_output;
> > + }
> > +
> > + return 0;
> > +
> > +error_disable_clk_output:
> > clk_disable_unprepare(host->clk_output);
> > +error_disable_clk_mmc:
> > clk_disable_unprepare(host->clk_mmc);
> > - clk_disable_unprepare(host->clk_ahb);
> > + return ret;
> > +}
> >
> > - dma_free_coherent(&pdev->dev, PAGE_SIZE, host->sg_cpu, host->sg_dma);
> > - mmc_free_host(mmc);
> > +static int sunxi_mmc_runtime_suspend(struct device *dev)
> > +{
> > + struct mmc_host *mmc = dev_get_drvdata(dev);
> > + struct sunxi_mmc_host *host = mmc_priv(mmc);
> > +
> > + clk_disable_unprepare(host->clk_sample);
> > + clk_disable_unprepare(host->clk_output);
> > + clk_disable_unprepare(host->clk_mmc);
> >
> > return 0;
> > }
> >
> > +static const struct dev_pm_ops sunxi_mmc_pm_ops = {
> > + SET_RUNTIME_PM_OPS(sunxi_mmc_runtime_suspend,
> > + sunxi_mmc_runtime_resume,
> > + NULL)
> > +};
> > +
> > static struct platform_driver sunxi_mmc_driver = {
> > .driver = {
> > .name = "sunxi-mmc",
> > .of_match_table = of_match_ptr(sunxi_mmc_of_match),
> > + .pm = &sunxi_mmc_pm_ops,
> > },
> > .probe = sunxi_mmc_probe,
> > .remove = sunxi_mmc_remove,
> > --
> > git-series 0.9.1
>
> Otherwise this looks good to me!
>
> BTW, isn't system wide suspend/resume() also important to save power
> for? That's rather simple to implement, after you have enabled runtime
> PM support.
We don't have (unfortunately) any kind of suspend support at the
moment, so I guess that'll come as a single step when we'll have it :)
Thanks!
Maxime
--
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com
Attachment:
signature.asc
Description: PGP signature
