| Message ID | 1374830310-21765-4-git-send-email-peter.chen@freescale.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, Jul 26, 2013 at 05:18:19PM +0800, Peter Chen wrote: > - If devm_regulator_get returns -EPROBE_DEFER, we also return > -EPROBE_DEFER to wait regulator being ready later. > - If devm_regulator_get returns -ENODEV, we think there is > no "vbus-supply" node at DT, it means this board doesn't need > vbus control. > - If devm_regulator_get returns other error values, it means > there are something wrong for getting this regulator. > > Signed-off-by: Peter Chen <peter.chen@freescale.com> > --- > drivers/usb/chipidea/ci_hdrc_imx.c | 14 ++++++++++++-- > 1 files changed, 12 insertions(+), 2 deletions(-) > > diff --git a/drivers/usb/chipidea/ci_hdrc_imx.c b/drivers/usb/chipidea/ci_hdrc_imx.c > index d06355e..0ced8c1 100644 > --- a/drivers/usb/chipidea/ci_hdrc_imx.c > +++ b/drivers/usb/chipidea/ci_hdrc_imx.c > @@ -144,8 +144,18 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev) > > /* Get the vbus regulator */ > pdata.reg_vbus = devm_regulator_get(&pdev->dev, "vbus"); > - if (IS_ERR(pdata.reg_vbus)) > - pdata.reg_vbus = NULL; > + if (PTR_ERR(pdata.reg_vbus) == -EPROBE_DEFER) { > + ret = -EPROBE_DEFER; > + goto err_clk; > + } else if (PTR_ERR(pdata.reg_vbus) == -ENODEV) { > + pdata.reg_vbus = NULL; /* no vbus regualator is needed */ > + } else if (IS_ERR(pdata.reg_vbus)) { > + dev_err(&pdev->dev, > + "Getting regulator error: %ld\n", > + PTR_ERR(pdata.reg_vbus)); > + ret = PTR_ERR(pdata.reg_vbus); > + goto err_clk; > + } > > if (!pdev->dev.dma_mask) > pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask; > -- This is wrong, you should squash that into the previous patch. And as already mentioned, this can probably go into core.c as well. Pick up the habit *not* to change code in one series which another patch of the same series introduced. This only adds *dusty* unused history in the patchstack that nobody needs. A *clean* and *coherent* series with discrete patches is much easier to review and will get accepted much faster. Michael
On Tue, Jul 30, 2013 at 12:47:17AM +0200, Michael Grzeschik wrote: > On Fri, Jul 26, 2013 at 05:18:19PM +0800, Peter Chen wrote: > > - If devm_regulator_get returns -EPROBE_DEFER, we also return > > -EPROBE_DEFER to wait regulator being ready later. > > - If devm_regulator_get returns -ENODEV, we think there is > > no "vbus-supply" node at DT, it means this board doesn't need > > vbus control. > > - If devm_regulator_get returns other error values, it means > > there are something wrong for getting this regulator. > > > > Signed-off-by: Peter Chen <peter.chen@freescale.com> > > --- > > drivers/usb/chipidea/ci_hdrc_imx.c | 14 ++++++++++++-- > > 1 files changed, 12 insertions(+), 2 deletions(-) > > > > diff --git a/drivers/usb/chipidea/ci_hdrc_imx.c b/drivers/usb/chipidea/ci_hdrc_imx.c > > index d06355e..0ced8c1 100644 > > --- a/drivers/usb/chipidea/ci_hdrc_imx.c > > +++ b/drivers/usb/chipidea/ci_hdrc_imx.c > > @@ -144,8 +144,18 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev) > > > > /* Get the vbus regulator */ > > pdata.reg_vbus = devm_regulator_get(&pdev->dev, "vbus"); > > - if (IS_ERR(pdata.reg_vbus)) > > - pdata.reg_vbus = NULL; > > + if (PTR_ERR(pdata.reg_vbus) == -EPROBE_DEFER) { > > + ret = -EPROBE_DEFER; > > + goto err_clk; > > + } else if (PTR_ERR(pdata.reg_vbus) == -ENODEV) { > > + pdata.reg_vbus = NULL; /* no vbus regualator is needed */ > > + } else if (IS_ERR(pdata.reg_vbus)) { > > + dev_err(&pdev->dev, > > + "Getting regulator error: %ld\n", > > + PTR_ERR(pdata.reg_vbus)); > > + ret = PTR_ERR(pdata.reg_vbus); > > + goto err_clk; > > + } > > > > if (!pdev->dev.dma_mask) > > pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask; > > -- > > This is wrong, you should squash that into the previous patch. And > as already mentioned, this can probably go into core.c as well. > > Pick up the habit *not* to change code in one series which another patch > of the same series introduced. This only adds *dusty* unused history in the > patchstack that nobody needs. A *clean* and *coherent* series with discrete > patches is much easier to review and will get accepted much faster. > My rule is do ONE thing at ONE patch, is it not correct? Previous one[2/14]: Remove the vbus operation at imx glue layer. This one [3/14]: Fix a bug that vbus may be gotten delay (EPROBE_DEFER), and vbus is valid at this case.
diff --git a/drivers/usb/chipidea/ci_hdrc_imx.c b/drivers/usb/chipidea/ci_hdrc_imx.c index d06355e..0ced8c1 100644 --- a/drivers/usb/chipidea/ci_hdrc_imx.c +++ b/drivers/usb/chipidea/ci_hdrc_imx.c @@ -144,8 +144,18 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev) /* Get the vbus regulator */ pdata.reg_vbus = devm_regulator_get(&pdev->dev, "vbus"); - if (IS_ERR(pdata.reg_vbus)) - pdata.reg_vbus = NULL; + if (PTR_ERR(pdata.reg_vbus) == -EPROBE_DEFER) { + ret = -EPROBE_DEFER; + goto err_clk; + } else if (PTR_ERR(pdata.reg_vbus) == -ENODEV) { + pdata.reg_vbus = NULL; /* no vbus regualator is needed */ + } else if (IS_ERR(pdata.reg_vbus)) { + dev_err(&pdev->dev, + "Getting regulator error: %ld\n", + PTR_ERR(pdata.reg_vbus)); + ret = PTR_ERR(pdata.reg_vbus); + goto err_clk; + } if (!pdev->dev.dma_mask) pdev->dev.dma_mask = &pdev->dev.coherent_dma_mask;
- If devm_regulator_get returns -EPROBE_DEFER, we also return -EPROBE_DEFER to wait regulator being ready later. - If devm_regulator_get returns -ENODEV, we think there is no "vbus-supply" node at DT, it means this board doesn't need vbus control. - If devm_regulator_get returns other error values, it means there are something wrong for getting this regulator. Signed-off-by: Peter Chen <peter.chen@freescale.com> --- drivers/usb/chipidea/ci_hdrc_imx.c | 14 ++++++++++++-- 1 files changed, 12 insertions(+), 2 deletions(-)