From: Mark Brown broonie@linaro.org
Systems with the common clock API need clk_prepare() as well as the enable step.
Signed-off-by: Mark Brown broonie@linaro.org --- drivers/usb/phy/phy-nop.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/usb/phy/phy-nop.c b/drivers/usb/phy/phy-nop.c index f52b7f8..6988c15 100644 --- a/drivers/usb/phy/phy-nop.c +++ b/drivers/usb/phy/phy-nop.c @@ -80,7 +80,7 @@ static int nop_init(struct usb_phy *phy) }
if (!IS_ERR(nop->clk)) - clk_enable(nop->clk); + clk_prepare_enable(nop->clk);
if (!IS_ERR(nop->reset)) { /* De-assert RESET */ @@ -102,7 +102,7 @@ static void nop_shutdown(struct usb_phy *phy) }
if (!IS_ERR(nop->clk)) - clk_disable(nop->clk); + clk_disable_unprepare(nop->clk);
if (!IS_ERR(nop->vcc)) { if (regulator_disable(nop->vcc))
On Thu, Aug 08, 2013 at 12:35:04PM +0100, Mark Brown wrote:
From: Mark Brown broonie@linaro.org
Systems with the common clock API need clk_prepare() as well as the enable step.
Signed-off-by: Mark Brown broonie@linaro.org
clk_prepare() is done on probe()... -ECONFUSED
On Fri, Aug 09, 2013 at 05:38:57PM +0300, Felipe Balbi wrote:
On Thu, Aug 08, 2013 at 12:35:04PM +0100, Mark Brown wrote:
Systems with the common clock API need clk_prepare() as well as the enable step.
clk_prepare() is done on probe()... -ECONFUSED
Ah, so it is, missed that - though it seems like it'd be as well to just leave it to the time the clock is needed given that we're not in atomic context. Is there any great reason for doing it on probe?
On Fri, Aug 09, 2013 at 03:57:27PM +0100, Mark Brown wrote:
On Fri, Aug 09, 2013 at 05:38:57PM +0300, Felipe Balbi wrote:
On Thu, Aug 08, 2013 at 12:35:04PM +0100, Mark Brown wrote:
Systems with the common clock API need clk_prepare() as well as the enable step.
clk_prepare() is done on probe()... -ECONFUSED
Ah, so it is, missed that - though it seems like it'd be as well to just leave it to the time the clock is needed given that we're not in atomic context. Is there any great reason for doing it on probe?
dunno, Roger added that code originally.
Hi Mark,
On 08/09/2013 05:57 PM, Mark Brown wrote:
On Fri, Aug 09, 2013 at 05:38:57PM +0300, Felipe Balbi wrote:
On Thu, Aug 08, 2013 at 12:35:04PM +0100, Mark Brown wrote:
Systems with the common clock API need clk_prepare() as well as the enable step.
clk_prepare() is done on probe()... -ECONFUSED
Ah, so it is, missed that - though it seems like it'd be as well to just leave it to the time the clock is needed given that we're not in atomic context. Is there any great reason for doing it on probe?
No great reason. So I agree with you.
cheers, -roger
linaro-kernel@lists.linaro.org
-
Felipe Balbi -
Mark Brown -
Roger Quadros