From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andy Shevchenko Subject: Re: [PATCH v4] i2c: designware-platdrv: fix unbalanced clk enable and prepare Date: Mon, 25 Apr 2016 12:07:15 +0300 Message-ID: <1461575235.17131.3.camel@linux.intel.com> References: <1461314971-5944-1-git-send-email-jszhang@marvell.com> <571A2E43.9030109@linux.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from mga04.intel.com ([192.55.52.120]:44057 "EHLO mga04.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753522AbcDYJGU (ORCPT ); Mon, 25 Apr 2016 05:06:20 -0400 In-Reply-To: <571A2E43.9030109@linux.intel.com> Sender: linux-i2c-owner@vger.kernel.org List-Id: linux-i2c@vger.kernel.org To: Jarkko Nikula , Jisheng Zhang , mika.westerberg@linux.intel.com, wsa@the-dreams.de Cc: linux-i2c@vger.kernel.org, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org On Fri, 2016-04-22 at 16:59 +0300, Jarkko Nikula wrote: > Hi >=20 > On 04/22/2016 11:49 AM, Jisheng Zhang wrote: > >=20 > > If i2c_dw_probe() fails, we should disable and unprepare the clock, > > otherwise the clock enable and prepare is left unbalanced. > >=20 > > In dw_i2c_plat_remove(), we'd better to not rely on runtime PM to > > disable and unprepare the clock since CONFIG_PM may be disabled whe= n > > configuring the kernel. So we explicitly disable and unprepare the > > clock in dw_i2c_plat_remove() rather than implicitly rely on > > pm_runtime_put_sync(). To keep the device usage count balanced, we > > call pm_runtime_put_noidle() to decrease the usage count. > >=20 > > Signed-off-by: Jisheng Zhang > > --- > > =C2=A0 Since v3: > > =C2=A0=C2=A0=C2=A0- use runtime PM rather than rpm in commit msg > > =C2=A0=C2=A0=C2=A0- remove duplicated "(" in commit msg > >=20 > > =C2=A0 Since v2: > > =C2=A0=C2=A0=C2=A0- s/clk/clock > > =C2=A0=C2=A0=C2=A0- describe why use pm_runtime_put_noidle() > >=20 > > =C2=A0 Since v1: > > =C2=A0=C2=A0=C2=A0- fix commit msg: "not rely on rpm" rather than "= rely on rpm" > > =C2=A0=C2=A0=C2=A0- call i2c_dw_plat_prepare_clk after pm_rumtime_d= isable() > > =C2=A0 drivers/i2c/busses/i2c-designware-platdrv.c | 16 ++++++++++-= ----- > > =C2=A0 1 file changed, 10 insertions(+), 6 deletions(-) > >=20 > > diff --git a/drivers/i2c/busses/i2c-designware-platdrv.c > > b/drivers/i2c/busses/i2c-designware-platdrv.c > > index d656657..a771781 100644 > > --- a/drivers/i2c/busses/i2c-designware-platdrv.c > > +++ b/drivers/i2c/busses/i2c-designware-platdrv.c > > @@ -253,8 +253,11 @@ static int dw_i2c_plat_probe(struct > > platform_device *pdev) > > =C2=A0=C2=A0 } > >=20 > > =C2=A0=C2=A0 r =3D i2c_dw_probe(dev); > > - if (r && !dev->pm_runtime_disabled) > > - pm_runtime_disable(&pdev->dev); > > + if (r) { > > + if (!dev->pm_runtime_disabled) > > + pm_runtime_disable(&pdev->dev); > > + i2c_dw_plat_prepare_clk(dev, false); > > + } > >=20 > > =C2=A0=C2=A0 return r; > > =C2=A0 } > > @@ -264,15 +267,16 @@ static int dw_i2c_plat_remove(struct > > platform_device *pdev) > > =C2=A0=C2=A0 struct dw_i2c_dev *dev =3D platform_get_drvdata(pdev); > >=20 > > =C2=A0=C2=A0 pm_runtime_get_sync(&pdev->dev); > > + pm_runtime_dont_use_autosuspend(&pdev->dev); > > + if (!dev->pm_runtime_disabled) > > + pm_runtime_disable(&pdev->dev); > > + pm_runtime_put_noidle(&pdev->dev); > >=20 > > =C2=A0=C2=A0 i2c_del_adapter(&dev->adapter); > >=20 > > =C2=A0=C2=A0 i2c_dw_disable(dev); > >=20 > > - pm_runtime_dont_use_autosuspend(&pdev->dev); > > - pm_runtime_put_sync(&pdev->dev); > > - if (!dev->pm_runtime_disabled) > > - pm_runtime_disable(&pdev->dev); > > + i2c_dw_plat_prepare_clk(dev, false); > >=20 > This feels a bit an invasive change to me for unbalanced clock=C2=A0 > enable/disable and I noticed this changes semantics how=C2=A0 > drivers/acpi/acpi_lpss.c devices are shutdown when removing the > driver.=C2=A0 > Although I didn't notice does it cause any regression. >=20 > Before patch: > 1. drivers/base/dd.c: __device_release_driver() > =C2=A0=C2=A0=C2=A0=C2=A0- pm_runtime_get_sync() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0-> acpi_device_set_power(D0) > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0acpi_lpss_resto= re_ctx() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0dw_i2c_plat_res= ume() > 2. dw_i2c_plat_remove() > =C2=A0=C2=A0=C2=A0=C2=A0- pm_runtime_dont_use_autosuspend() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pm_runtime_put_sync() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0-> dw_i2c_plat_suspend() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0acpi_lpss_save_= ctx() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0acpi_device_set= _power(D3) > 3. __device_release_driver() continue > =C2=A0=C2=A0=C2=A0=C2=A0- dev->pm_domain->dismiss(dev) > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0-> acpi_lpss_dismiss() ... -> acp= i_device_set_power(D3) >=20 > After patch: > 1. drivers/base/dd.c: __device_release_driver() > =C2=A0 - pm_runtime_get_sync() > =C2=A0=C2=A0=C2=A0=C2=A0-> acpi_device_set_power(D0) > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0acpi_lpss_restore_ctx() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0dw_i2c_plat_resume() > 2. dw_i2c_plat_remove() > =C2=A0=C2=A0=C2=A0=C2=A0- pm_runtime_dont_use_autosuspend() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0pm_runtime_put_noidle() > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0* no device suspending and acpi_l= pss_save_ctx() > 3. __device_release_driver() continue > =C2=A0=C2=A0=C2=A0=C2=A0- dev->pm_domain->dismiss(dev) > =C2=A0=C2=A0=C2=A0=C2=A0-> acpi_lpss_dismiss() ... -> acpi_device_set= _power(D3) > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0* powers down here >=20 > So after patch there is no acpi_lpss_save_ctx() call but I don't see=C2= =A0 > does it cause any issue here. Maybe it's better to track clock only.=C2= =A0 > What you think Andy? Now it looks like two fixes in one patch. From the commit message I didn't get the relation between change runtime PM call (one to the other) and clock (un)preparation. --=20 Andy Shevchenko Intel Finland Oy