RE: [PATCH] mmc: atmel-mci: add pm runtime support

From: Yang, Wenyou
Date: Wed Oct 29 2014 - 21:14:16 EST

Next message: Jeff Epler: "Re: incompatible pci sysfs change since 3.12 (5136b2da770d)"
Previous message: Mrs Rebecca Garang: "We can do this Investment"
In reply to: Ulf Hansson: "Re: [PATCH] mmc: atmel-mci: add pm runtime support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hello Ulf,

Thank you very much for so many comments.

I will do it.

> -----Original Message-----
> From: Ulf Hansson [mailto:ulf.hansson@xxxxxxxxxx]
> Sent: Wednesday, October 29, 2014 6:06 PM
> To: Yang, Wenyou
> Cc: Chris Ball; Kevin Hilman; Ferre, Nicolas; Desroches, Ludovic; linux-mmc; linux-
> kernel@xxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH] mmc: atmel-mci: add pm runtime support
>
> On 29 October 2014 03:15, Wenyou Yang <wenyou.yang@xxxxxxxxx> wrote:
>
> Hi Wenyou,
>
> Could you please provide some more information in the commit message.
>
> > Signed-off-by: Wenyou Yang <wenyou.yang@xxxxxxxxx>
> > ---
> > drivers/mmc/host/atmel-mci.c | 116
> > ++++++++++++++++++++++++++++++++++--------
> > 1 file changed, 94 insertions(+), 22 deletions(-)
> >
> > diff --git a/drivers/mmc/host/atmel-mci.c
> > b/drivers/mmc/host/atmel-mci.c index 0b9ddf8..e9d32d0 100644
> > --- a/drivers/mmc/host/atmel-mci.c
> > +++ b/drivers/mmc/host/atmel-mci.c
> > @@ -37,6 +37,8 @@
> >
> > #include <linux/atmel-mci.h>
> > #include <linux/atmel_pdc.h>
> > +#include <linux/pm.h>
> > +#include <linux/pm_runtime.h>
> >
> > #include <asm/cacheflush.h>
> > #include <asm/io.h>
> > @@ -44,6 +46,8 @@
> >
> > #include "atmel-mci-regs.h"
> >
> > +#define AUTOSUSPEND_DELAY 50
> > +
> > #define ATMCI_DATA_ERROR_FLAGS (ATMCI_DCRCE | ATMCI_DTOE |
> ATMCI_OVRE | ATMCI_UNRE)
> > #define ATMCI_DMA_THRESHOLD 16
> >
> > @@ -386,20 +390,19 @@ static int atmci_regs_show(struct seq_file *s, void *v)
> > if (!buf)
> > return -ENOMEM;
> >
> > + pm_runtime_get_sync(&host->pdev->dev);
> > +
> > /*
> > * Grab a more or less consistent snapshot. Note that we're
> > * not disabling interrupts, so IMR and SR may not be
> > * consistent.
> > */
> > - ret = clk_prepare_enable(host->mck);
> > - if (ret)
> > - goto out;
> > -
> > spin_lock_bh(&host->lock);
> > memcpy_fromio(buf, host->regs, ATMCI_REGS_SIZE);
> > spin_unlock_bh(&host->lock);
> >
> > - clk_disable_unprepare(host->mck);
> > + pm_runtime_mark_last_busy(&host->pdev->dev);
> > + pm_runtime_put_autosuspend(&host->pdev->dev);
> >
> > seq_printf(s, "MR:\t0x%08x%s%s ",
> > buf[ATMCI_MR / 4], @@ -449,7 +452,6 @@ static
> > int atmci_regs_show(struct seq_file *s, void *v)
> > val & ATMCI_CFG_LSYNC ? " LSYNC" : "");
> > }
> >
> > -out:
> > kfree(buf);
> >
> > return ret;
> > @@ -1252,6 +1254,8 @@ static void atmci_request(struct mmc_host *mmc,
> struct mmc_request *mrq)
> > WARN_ON(slot->mrq);
> > dev_dbg(&host->pdev->dev, "MRQ: cmd %u\n", mrq->cmd->opcode);
> >
> > + pm_runtime_get_sync(&host->pdev->dev);
> > +
> > /*
> > * We may "know" the card is gone even though there's still an
> > * electrical connection. If so, we really need to communicate
> > @@ -1281,7 +1285,8 @@ static void atmci_set_ios(struct mmc_host *mmc,
> struct mmc_ios *ios)
> > struct atmel_mci_slot *slot = mmc_priv(mmc);
> > struct atmel_mci *host = slot->host;
> > unsigned int i;
> > - bool unprepare_clk;
> > +
> > + pm_runtime_get_sync(&host->pdev->dev);
> >
> > slot->sdc_reg &= ~ATMCI_SDCBUS_MASK;
> > switch (ios->bus_width) {
> > @@ -1297,13 +1302,8 @@ static void atmci_set_ios(struct mmc_host *mmc,
> struct mmc_ios *ios)
> > unsigned int clock_min = ~0U;
> > u32 clkdiv;
> >
> > - clk_prepare(host->mck);
> > - unprepare_clk = true;
> > -
> > spin_lock_bh(&host->lock);
> > if (!host->mode_reg) {
> > - clk_enable(host->mck);
> > - unprepare_clk = false;
> > atmci_writel(host, ATMCI_CR, ATMCI_CR_SWRST);
> > atmci_writel(host, ATMCI_CR, ATMCI_CR_MCIEN);
> > if (host->caps.has_cfg_reg) @@ -1371,8 +1371,6
> > @@ static void atmci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
> > } else {
> > bool any_slot_active = false;
> >
> > - unprepare_clk = false;
> > -
> > spin_lock_bh(&host->lock);
> > slot->clock = 0;
> > for (i = 0; i < ATMCI_MAX_NR_SLOTS; i++) { @@ -1385,17
> > +1383,12 @@ static void atmci_set_ios(struct mmc_host *mmc, struct mmc_ios
> *ios)
> > atmci_writel(host, ATMCI_CR, ATMCI_CR_MCIDIS);
> > if (host->mode_reg) {
> > atmci_readl(host, ATMCI_MR);
> > - clk_disable(host->mck);
> > - unprepare_clk = true;
> > }
> > host->mode_reg = 0;
> > }
> > spin_unlock_bh(&host->lock);
> > }
> >
> > - if (unprepare_clk)
> > - clk_unprepare(host->mck);
> > -
> > switch (ios->power_mode) {
> > case MMC_POWER_OFF:
> > if (!IS_ERR(mmc->supply.vmmc)) @@ -1421,6 +1414,9 @@
> > static void atmci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
> > */
> > break;
> > }
> > +
> > + pm_runtime_mark_last_busy(&host->pdev->dev);
> > + pm_runtime_put_autosuspend(&host->pdev->dev);
> > }
> >
> > static int atmci_get_ro(struct mmc_host *mmc) @@ -1512,6 +1508,9 @@
> > static void atmci_request_end(struct atmel_mci *host, struct mmc_request *mrq)
> > spin_unlock(&host->lock);
> > mmc_request_done(prev_mmc, mrq);
> > spin_lock(&host->lock);
> > +
> > + pm_runtime_mark_last_busy(&host->pdev->dev);
> > + pm_runtime_put_autosuspend(&host->pdev->dev);
> > }
> >
> > static void atmci_command_complete(struct atmel_mci *host, @@ -2417,7
> > +2416,6 @@ static int __init atmci_probe(struct platform_device *pdev)
> >
> > atmci_writel(host, ATMCI_CR, ATMCI_CR_SWRST);
> > host->bus_hz = clk_get_rate(host->mck);
> > - clk_disable_unprepare(host->mck);
> >
> > host->mapbase = regs->start;
> >
> > @@ -2449,6 +2447,12 @@ static int __init atmci_probe(struct
> > platform_device *pdev)
> >
> > setup_timer(&host->timer, atmci_timeout_timer, (unsigned
> > long)host);
> >
> > + pm_runtime_enable(&pdev->dev);
> > + pm_runtime_get_sync(&pdev->dev);
> > + pm_runtime_set_autosuspend_delay(&pdev->dev,
> AUTOSUSPEND_DELAY);
> > + pm_runtime_use_autosuspend(&pdev->dev);
>
> The above is wrong. You will get clock unbalance issues, meaning that the clock
> will never be gated at runtime PM suspend.
>
> Fix it by change the hole chunk of code above to:
>
> pm_runtime_get_noresume();
> pm_runtime_set_active();
> pm_runtime_set_autosuspend_delay();
> pm_runtime_use_autosuspend();
> pm_runtime_enable();
>
> > + pm_suspend_ignore_children(&pdev->dev, 1);
>
> Is this needed?
>
> > +
> > /* We need at least one slot to succeed */
> > nr_slots = 0;
> > ret = -ENODEV;
> > @@ -2491,6 +2495,9 @@ static int __init atmci_probe(struct platform_device
> *pdev)
> > "Atmel MCI controller at 0x%08lx irq %d, %u slots\n",
> > host->mapbase, irq, nr_slots);
> >
> > + pm_runtime_mark_last_busy(&host->pdev->dev);
> > + pm_runtime_put_autosuspend(&pdev->dev);
> > +
> > return 0;
> >
> > err_dma_alloc:
> > @@ -2499,6 +2506,9 @@ err_dma_alloc:
> > atmci_cleanup_slot(host->slot[i], i);
> > }
> > err_init_slot:
> > + pm_runtime_put_sync(&pdev->dev);
> > + pm_runtime_disable(&pdev->dev);
>
> According to upper comment, you must switch order, and you should use
> pm_runtime_put_noidle() instead.
>
> Moreover you needs to have a clk_disable_uprepare() in the error handling and
> prior the above.
>
> > +
> > del_timer_sync(&host->timer);
> > if (host->dma.chan)
> > dma_release_channel(host->dma.chan);
> > @@ -2511,6 +2521,8 @@ static int __exit atmci_remove(struct platform_device
> *pdev)
> > struct atmel_mci *host = platform_get_drvdata(pdev);
> > unsigned int i;
> >
> > + pm_runtime_get_sync(&pdev->dev);
> > +
> > if (host->buffer)
> > dma_free_coherent(&pdev->dev, host->buf_size,
> > host->buffer, host->buf_phys_addr);
> > @@ -2520,11 +2532,9 @@ static int __exit atmci_remove(struct platform_device
> *pdev)
> > atmci_cleanup_slot(host->slot[i], i);
> > }
> >
> > - clk_prepare_enable(host->mck);
> > atmci_writel(host, ATMCI_IDR, ~0UL);
> > atmci_writel(host, ATMCI_CR, ATMCI_CR_MCIDIS);
> > atmci_readl(host, ATMCI_SR);
> > - clk_disable_unprepare(host->mck);
> >
> > del_timer_sync(&host->timer);
> > if (host->dma.chan)
> > @@ -2532,14 +2542,76 @@ static int __exit atmci_remove(struct
> > platform_device *pdev)
> >
> > free_irq(platform_get_irq(pdev, 0), host);
> >
> > + pm_runtime_put_sync(&pdev->dev);
> > + pm_runtime_disable(&pdev->dev);
> > +
> > + clk_disable_unprepare(host->mck);
>
> Same comment as for fixing the error handling in ->probe().
>
> > +
> > return 0;
> > }
> >
> > +#ifdef CONFIG_PM
> > +static int atmci_suspend(struct device *dev) {
> > + struct atmel_mci *host = dev_get_drvdata(dev);
> > +
> > + pm_runtime_get_sync(dev);
> > +
> > + clk_disable_unprepare(host->mck);
> > +
> > + pm_runtime_mark_last_busy(dev);
> > + pm_runtime_put_autosuspend(dev);
> > +
> > + return 0;
> > +}
> > +
> > +static int atmci_resume(struct device *dev) {
> > + struct atmel_mci *host = dev_get_drvdata(dev);
> > + int ret;
> > +
> > + pm_runtime_get_sync(dev);
> > +
> > + ret = clk_prepare_enable(host->mck);
> > +
> > + pm_runtime_mark_last_busy(dev);
> > + pm_runtime_put_autosuspend(dev);
> > +
> > + return ret;
> > +}
> > +
> > +static int atmci_runtime_suspend(struct device *dev) {
> > + struct atmel_mci *host = dev_get_drvdata(dev);
> > +
> > + clk_disable_unprepare(host->mck);
> > +
> > + return 0;
> > +}
> > +
> > +static int atmci_runtime_resume(struct device *dev) {
> > + struct atmel_mci *host = dev_get_drvdata(dev);
> > +
> > + return clk_prepare_enable(host->mck); }
> > +
> > +static const struct dev_pm_ops atmci_dev_pm_ops = {
> > + SET_SYSTEM_SLEEP_PM_OPS(atmci_suspend, atmci_resume)
> > + SET_RUNTIME_PM_OPS(atmci_runtime_suspend,
> > +atmci_runtime_resume, NULL) };
>
> I would suggest the following changes to simplify code and to improve runtime
> PM support a bit.
>
> SET_SYSTEM_SLEEP_PM_OPS(pm_runtime_force_suspend,
> pm_runtime_force_resume) SET_PM_RUNTIME_PM_OPS(atmci_runtime_suspend,
> atmci_runtime_resume, NULL)
>
> Do note the change from SET_RUNTIME_PM_OPS ->
> SET_PM_RUNTIME_PM_OPS.
>
> That means you will be able to remove atmci_suspend|resume() functions entirely.
>
> > +
> > +#define ATMCI_PM_OPS (&atmci_dev_pm_ops)
> > +#else
> > +#define ATMCI_PM_OPS NULL
> > +#endif
>
> This is micro optimization, not needed. Just use the atmci_dev_pm_ops directly
> below instead.
>
> > +
> > static struct platform_driver atmci_driver = {
> > .remove = __exit_p(atmci_remove),
> > .driver = {
> > .name = "atmel_mci",
> > .of_match_table = of_match_ptr(atmci_dt_ids),
> > + .pm = ATMCI_PM_OPS,
> > },
> > };
> >
> > --
> > 1.7.9.5
> >
>
> Kind regards
> Ulf Hansson

Best Regards,
Wenyou Yang

Next message: Jeff Epler: "Re: incompatible pci sysfs change since 3.12 (5136b2da770d)"
Previous message: Mrs Rebecca Garang: "We can do this Investment"
In reply to: Ulf Hansson: "Re: [PATCH] mmc: atmel-mci: add pm runtime support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]