| Message ID | 20250226-dwc3-refactor-v4-3-4415e7111e49@oss.qualcomm.com |
|---|---|
| State | New |
| Headers | show |
| Series | [v4,1/7] dt-bindings: usb: Introduce qcom,snps-dwc3 | expand |
On Thu, Feb 27, 2025 at 06:24:17AM +0200, Dmitry Baryshkov wrote: > On Wed, Feb 26, 2025 at 04:17:50PM -0800, Bjorn Andersson wrote: > > When the core is integrated with glue, it's reasonable to assume that > > the glue driver will have to touch the IP before/after the core takes > > the hardware out and into reset. As such the glue must own these > > resources and be allowed to turn them on/off outside the core's > > handling. > > > > Allow the platform or glue layer to indicate if the core logic for > > clocks and resets should be skipped to deal with this. > > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@oss.qualcomm.com> > > --- > > drivers/usb/dwc3/core.c | 19 +++++++++++-------- > > drivers/usb/dwc3/glue.h | 1 + > > 2 files changed, 12 insertions(+), 8 deletions(-) > > > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c > > index d9f0a6782d36..aecdde8dc999 100644 > > --- a/drivers/usb/dwc3/core.c > > +++ b/drivers/usb/dwc3/core.c > > @@ -2328,6 +2330,7 @@ static int dwc3_probe(struct platform_device *pdev) > > > > probe_data.dwc = dwc; > > probe_data.res = res; > > + probe_data.ignore_clocks_and_resets = false; > > Isn't it a default value? > I like the false because it's explicit, but I have no strong attachment to it. That said, it's not the default value, because probe_data isn't zero-initialized. That would however make sense to do, in order to avoid surprises in the future when probe_data grows. Regards, Bjorn > > > > return dwc3_core_probe(&probe_data); > > } > > diff --git a/drivers/usb/dwc3/glue.h b/drivers/usb/dwc3/glue.h > > index e73cfc466012..1ddb451bdbd0 100644 > > --- a/drivers/usb/dwc3/glue.h > > +++ b/drivers/usb/dwc3/glue.h > > @@ -17,6 +17,7 @@ > > struct dwc3_probe_data { > > struct dwc3 *dwc; > > struct resource *res; > > + bool ignore_clocks_and_resets; > > }; > > > > int dwc3_core_probe(const struct dwc3_probe_data *data); > > > > -- > > 2.45.2 > > > > -- > With best wishes > Dmitry
On Thu, Feb 27, 2025 at 09:55:29AM -0600, Bjorn Andersson wrote: > On Thu, Feb 27, 2025 at 06:24:17AM +0200, Dmitry Baryshkov wrote: > > On Wed, Feb 26, 2025 at 04:17:50PM -0800, Bjorn Andersson wrote: > > > When the core is integrated with glue, it's reasonable to assume that > > > the glue driver will have to touch the IP before/after the core takes > > > the hardware out and into reset. As such the glue must own these > > > resources and be allowed to turn them on/off outside the core's > > > handling. > > > > > > Allow the platform or glue layer to indicate if the core logic for > > > clocks and resets should be skipped to deal with this. > > > > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@oss.qualcomm.com> > > > --- > > > drivers/usb/dwc3/core.c | 19 +++++++++++-------- > > > drivers/usb/dwc3/glue.h | 1 + > > > 2 files changed, 12 insertions(+), 8 deletions(-) > > > > > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c > > > index d9f0a6782d36..aecdde8dc999 100644 > > > --- a/drivers/usb/dwc3/core.c > > > +++ b/drivers/usb/dwc3/core.c > > > @@ -2328,6 +2330,7 @@ static int dwc3_probe(struct platform_device *pdev) > > > > > > probe_data.dwc = dwc; > > > probe_data.res = res; > > > + probe_data.ignore_clocks_and_resets = false; > > > > Isn't it a default value? > > > > I like the false because it's explicit, but I have no strong attachment > to it. I'm more biased to the 'make unusal cases stand out', which means the defaults can go away to highlight non-default cases. > > That said, it's not the default value, because probe_data isn't > zero-initialized. That would however make sense to do, in order to avoid > surprises in the future when probe_data grows. :-) > > Regards, > Bjorn > > > > > > > return dwc3_core_probe(&probe_data); > > > } > > > diff --git a/drivers/usb/dwc3/glue.h b/drivers/usb/dwc3/glue.h > > > index e73cfc466012..1ddb451bdbd0 100644 > > > --- a/drivers/usb/dwc3/glue.h > > > +++ b/drivers/usb/dwc3/glue.h > > > @@ -17,6 +17,7 @@ > > > struct dwc3_probe_data { > > > struct dwc3 *dwc; > > > struct resource *res; > > > + bool ignore_clocks_and_resets; > > > }; > > > > > > int dwc3_core_probe(const struct dwc3_probe_data *data); > > > > > > -- > > > 2.45.2 > > > > > > > -- > > With best wishes > > Dmitry
On Wed, Feb 26, 2025 at 04:17:50PM -0800, Bjorn Andersson wrote: > When the core is integrated with glue, it's reasonable to assume that > the glue driver will have to touch the IP before/after the core takes > the hardware out and into reset. As such the glue must own these > resources and be allowed to turn them on/off outside the core's > handling. > > Allow the platform or glue layer to indicate if the core logic for > clocks and resets should be skipped to deal with this. > > Signed-off-by: Bjorn Andersson <bjorn.andersson@oss.qualcomm.com> > --- > drivers/usb/dwc3/core.c | 19 +++++++++++-------- > drivers/usb/dwc3/glue.h | 1 + > 2 files changed, 12 insertions(+), 8 deletions(-) > > diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c > index d9f0a6782d36..aecdde8dc999 100644 > --- a/drivers/usb/dwc3/core.c > +++ b/drivers/usb/dwc3/core.c > @@ -2187,15 +2187,17 @@ int dwc3_core_probe(const struct dwc3_probe_data *data) > if (IS_ERR(dwc->usb_psy)) > return dev_err_probe(dev, PTR_ERR(dwc->usb_psy), "couldn't get usb power supply\n"); > > - dwc->reset = devm_reset_control_array_get_optional_shared(dev); > - if (IS_ERR(dwc->reset)) { > - ret = PTR_ERR(dwc->reset); > - goto err_put_psy; > - } > + if (!data->ignore_clocks_and_resets) { > + dwc->reset = devm_reset_control_array_get_optional_shared(dev); > + if (IS_ERR(dwc->reset)) { > + ret = PTR_ERR(dwc->reset); > + goto err_put_psy; > + } > > - ret = dwc3_get_clocks(dwc); > - if (ret) > - goto err_put_psy; > + ret = dwc3_get_clocks(dwc); > + if (ret) > + goto err_put_psy; > + } > > ret = reset_control_deassert(dwc->reset); > if (ret) > @@ -2328,6 +2330,7 @@ static int dwc3_probe(struct platform_device *pdev) > > probe_data.dwc = dwc; > probe_data.res = res; > + probe_data.ignore_clocks_and_resets = false; > > return dwc3_core_probe(&probe_data); > } > diff --git a/drivers/usb/dwc3/glue.h b/drivers/usb/dwc3/glue.h > index e73cfc466012..1ddb451bdbd0 100644 > --- a/drivers/usb/dwc3/glue.h > +++ b/drivers/usb/dwc3/glue.h > @@ -17,6 +17,7 @@ > struct dwc3_probe_data { > struct dwc3 *dwc; > struct resource *res; > + bool ignore_clocks_and_resets; Nit: suggest add kernel doc comments for each fields Reviewed-by: Frank Li <Frank.Li@nxp.com> > }; > > int dwc3_core_probe(const struct dwc3_probe_data *data); > > -- > 2.45.2 >
diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c index d9f0a6782d36..aecdde8dc999 100644 --- a/drivers/usb/dwc3/core.c +++ b/drivers/usb/dwc3/core.c @@ -2187,15 +2187,17 @@ int dwc3_core_probe(const struct dwc3_probe_data *data) if (IS_ERR(dwc->usb_psy)) return dev_err_probe(dev, PTR_ERR(dwc->usb_psy), "couldn't get usb power supply\n"); - dwc->reset = devm_reset_control_array_get_optional_shared(dev); - if (IS_ERR(dwc->reset)) { - ret = PTR_ERR(dwc->reset); - goto err_put_psy; - } + if (!data->ignore_clocks_and_resets) { + dwc->reset = devm_reset_control_array_get_optional_shared(dev); + if (IS_ERR(dwc->reset)) { + ret = PTR_ERR(dwc->reset); + goto err_put_psy; + } - ret = dwc3_get_clocks(dwc); - if (ret) - goto err_put_psy; + ret = dwc3_get_clocks(dwc); + if (ret) + goto err_put_psy; + } ret = reset_control_deassert(dwc->reset); if (ret) @@ -2328,6 +2330,7 @@ static int dwc3_probe(struct platform_device *pdev) probe_data.dwc = dwc; probe_data.res = res; + probe_data.ignore_clocks_and_resets = false; return dwc3_core_probe(&probe_data); } diff --git a/drivers/usb/dwc3/glue.h b/drivers/usb/dwc3/glue.h index e73cfc466012..1ddb451bdbd0 100644 --- a/drivers/usb/dwc3/glue.h +++ b/drivers/usb/dwc3/glue.h @@ -17,6 +17,7 @@ struct dwc3_probe_data { struct dwc3 *dwc; struct resource *res; + bool ignore_clocks_and_resets; }; int dwc3_core_probe(const struct dwc3_probe_data *data);
When the core is integrated with glue, it's reasonable to assume that the glue driver will have to touch the IP before/after the core takes the hardware out and into reset. As such the glue must own these resources and be allowed to turn them on/off outside the core's handling. Allow the platform or glue layer to indicate if the core logic for clocks and resets should be skipped to deal with this. Signed-off-by: Bjorn Andersson <bjorn.andersson@oss.qualcomm.com> --- drivers/usb/dwc3/core.c | 19 +++++++++++-------- drivers/usb/dwc3/glue.h | 1 + 2 files changed, 12 insertions(+), 8 deletions(-)