From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ulf Hansson Subject: Re: [PATCH 1/7] mmc: omap_hsmmc: Fix PM regression with deferred probe for pm_runtime_reinit Date: Thu, 11 Feb 2016 11:18:34 +0100 Message-ID: References: <1455145370-20301-1-git-send-email-tony@atomide.com> <1455145370-20301-2-git-send-email-tony@atomide.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Cc: linux-omap , "linux-arm-kernel@lists.infradead.org" , linux-mmc , Alan Stern , Kevin Hilman , Nishanth Menon , "Rafael J . Wysocki" , Tero Kristo , Greg Kroah-Hartman , "linux-i2c@vger.kernel.org" , linux-serial@vger.kernel.org, "linux-spi@vger.kernel.org" , Mark Brown , Peter Hurley , Wolfram Sang To: Tony Lindgren Return-path: In-Reply-To: <1455145370-20301-2-git-send-email-tony@atomide.com> Sender: linux-i2c-owner@vger.kernel.org List-Id: linux-spi.vger.kernel.org On 11 February 2016 at 00:02, Tony Lindgren wrote: > Commit 5de85b9d57ab ("PM / runtime: Re-init runtime PM states at probe > error and driver unbind") introduced pm_runtime_reinit() that is used > to reinitialize PM runtime after -EPROBE_DEFER. This allows shutting > down the device after a failed probe. > > However, for drivers using pm_runtime_use_autosuspend() this can cause > a state where suspend callback is never called after -EPROBE_DEFER. > On the following device driver probe, hardware state is different from > the PM runtime state causing omap_device to produce the following > error: > > omap_device_enable() called from invalid state 1 > > And with omap_device and omap hardware being picky for PM, this will > block any deeper idle states in hardware. > > The solution is to fix the drivers to follow the PM runtime documentation: > > 1. For sections of code that needs the device disabled, use > pm_runtime_put_sync_suspend() if pm_runtime_set_autosuspend() has > been set. > > 2. For driver exit code, use pm_runtime_dont_use_autosuspend() before > pm_runtime_put_sync() if pm_runtime_use_autosuspend() has been > set. > > Fixes: 5de85b9d57ab ("PM / runtime: Re-init runtime PM states at probe > error and driver unbind") > Cc: linux-mmc@vger.kernel.org > Cc: Alan Stern > Cc: Kevin Hilman > Cc: Nishanth Menon > Cc: Rafael J. Wysocki > Cc: Ulf Hansson > Cc: Tero Kristo > Signed-off-by: Tony Lindgren > --- > > Ulf, I'd like to merge this along with other related fixes via the > ARM SoC tree if no objections, please review and ack if this look OK > to you. I have some other omap_hsmmc patches queued for 4.6, so I prefer to send this via my mmc tree. I guess that's okay as well!? > > --- > drivers/mmc/host/omap_hsmmc.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c > index b6639ea..32bc112 100644 > --- a/drivers/mmc/host/omap_hsmmc.c > +++ b/drivers/mmc/host/omap_hsmmc.c > @@ -1247,7 +1247,7 @@ static int omap_hsmmc_switch_opcond(struct omap_hsmmc_host *host, int vdd) > int ret; > > /* Disable the clocks */ > - pm_runtime_put_sync(host->dev); > + pm_runtime_put_sync_suspend(host->dev); This has no effect. The mmc core holds a runtime PM usage count for the device anyway, so the clock won't be disabled. Please remove this change from the patch. > if (host->dbclk) > clk_disable_unprepare(host->dbclk); > > @@ -2232,6 +2232,7 @@ err_irq: > dma_release_channel(host->tx_chan); > if (host->rx_chan) > dma_release_channel(host->rx_chan); > + pm_runtime_dont_use_autosuspend(host->dev); > pm_runtime_put_sync(host->dev); > pm_runtime_disable(host->dev); > if (host->dbclk) > @@ -2253,6 +2254,7 @@ static int omap_hsmmc_remove(struct platform_device *pdev) > dma_release_channel(host->tx_chan); > dma_release_channel(host->rx_chan); > > + pm_runtime_dont_use_autosuspend(host->dev); > pm_runtime_put_sync(host->dev); > pm_runtime_disable(host->dev); > device_init_wakeup(&pdev->dev, false); > @@ -2285,7 +2287,7 @@ static int omap_hsmmc_suspend(struct device *dev) > if (host->dbclk) > clk_disable_unprepare(host->dbclk); > > - pm_runtime_put_sync(host->dev); > + pm_runtime_put_sync_suspend(host->dev); This has no effect, as the PM core prevents runtime suspend during the system PM phase. It does so, by increasing the runtime PM usage count (pm_runtime_get_noresume()). Please remove this change from patch. > return 0; > } > > -- > 2.7.0 > Kind regards Uffe