From mboxrd@z Thu Jan  1 00:00:00 1970
From: b29396@freescale.com (Dong Aisheng)
Date: Tue, 24 Dec 2013 10:34:13 +0800
Subject: [PATCH 2/2] mmc: sdhci-esdhc-imx: fix warning during module
 remove function
In-Reply-To: <20131223140613.GL26070@S2101-09.ap.freescale.net>
References: <1387796544-18209-1-git-send-email-b29396@freescale.com>
 <1387796544-18209-2-git-send-email-b29396@freescale.com>
 <20131223140613.GL26070@S2101-09.ap.freescale.net>
Message-ID: <20131224023412.GC23230@shlinux2.ap.freescale.net>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Mon, Dec 23, 2013 at 10:06:15PM +0800, Shawn Guo wrote:
> On Mon, Dec 23, 2013 at 07:02:24PM +0800, Dong Aisheng wrote:
> > @@ -1160,8 +1160,6 @@ free_sdhci:
> >  static int sdhci_esdhc_imx_remove(struct platform_device *pdev)
> >  {
> >  	struct sdhci_host *host = platform_get_drvdata(pdev);
> > -	struct sdhci_pltfm_host *pltfm_host = sdhci_priv(host);
> > -	struct pltfm_imx_data *imx_data = pltfm_host->priv;
> >  	int dead = (readl(host->ioaddr + SDHCI_INT_STATUS) == 0xffffffff);
> >  
> >  	sdhci_remove_host(host, dead);
> > @@ -1169,10 +1167,6 @@ static int sdhci_esdhc_imx_remove(struct platform_device *pdev)
> >  	pm_runtime_dont_use_autosuspend(&pdev->dev);
> >  	pm_runtime_disable(&pdev->dev);
> >  
> > -	clk_disable_unprepare(imx_data->clk_per);
> > -	clk_disable_unprepare(imx_data->clk_ipg);
> > -	clk_disable_unprepare(imx_data->clk_ahb);
> 
> It's obviously a bad change to me.  We should definitely have these
> clk_disable_unprepare() calls in sdhci_esdhc_imx_remove() to match the
> clk_prepare_enable() calls in sdhci_esdhc_imx_probe().  Otherwise, at
> least for !CONFIG_PM_RUNTIME build, it's broken.
> 

How about add the !CONFIG_RUMTIME_PM precondition for these code to avoid
break non runtime pm case?

#ifndef CONFIG_PM_RUNTIME
	clk_disable_unprepare(imx_data->clk_per);
	clk_disable_unprepare(imx_data->clk_ipg);
	clk_disable_unprepare(imx_data->clk_ahb);
#endif

> I'm looking at sdhci-pxav3 driver.  Why can it call
> clk_disable_unprepare(pltfm_host->clk) in .remove() with runtime PM
> support in there?  Is there anything about runtime PM not being done
> correctly for our driver?
> 

I checked the code, it seems it calls pm_runtime_get_sync(&pdev->dev);
before clk_disable_*.
That means the clock is already enabled,
Without it, i wonder it may have the same problem.

static int sdhci_pxav3_remove(struct platform_device *pdev)
{
...
        pm_runtime_get_sync(&pdev->dev);
        sdhci_remove_host(host, 1);
        pm_runtime_disable(&pdev->dev);

        clk_disable_unprepare(pltfm_host->clk);
        clk_put(pltfm_host->clk);
..
}

Regards
Dong Aisheng

> Shawn
> 
> > -
> >  	sdhci_pltfm_free(pdev);
> >  
> >  	return 0;
> > -- 
> > 1.7.2.rc3
> > 
> > 
>