Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Arvind Yadav]

On Wednesday 02 May 2018 03:27 PM, Johan Hovold wrote:

On Wed, May 02, 2018 at 03:15:05PM +0530, Arvind Yadav wrote:


On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:

Replace the manual validity checks for the GPIO with the
gpio_is_valid().

Signed-off-by: Arvind Yadav 
---
chnage in v2 :
   Returning invalid gpio as error instead of -ENODEV.

   drivers/staging/greybus/arche-platform.c | 6 +++---
   1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/greybus/arche-platform.c 
b/drivers/staging/greybus/arche-platform.c
index 83254a7..c3a7da5 100644
--- a/drivers/staging/greybus/arche-platform.c
+++ b/drivers/staging/greybus/arche-platform.c
@@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
"svc,reset-gpio",
0);
-   if (arche_pdata->svc_reset_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
dev_err(dev, "failed to get reset-gpio\n");
return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.

We need to check gpio validity. If we are using of_get_named_gpio() or
not.  of_get_name_gpio() will read a device node and fetch the value.
But it'll not check that gpio is valid or not valid.

No, I believe you're mistaken here. of_get_named_gpio() does not return
an arbitrary gpio number, unlike what you could possibly find in
legacy board files and for which the gpio_is_valid() helper made sense.

Johan

Yes, You are coorect. It'll read gpio form gpio device node. Which means
it'll read from device tree node. without finding a valid entry. It'll 
return

an error.

~arvind

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Arvind Yadav]

On Wednesday 02 May 2018 03:27 PM, Johan Hovold wrote:

On Wed, May 02, 2018 at 03:15:05PM +0530, Arvind Yadav wrote:


On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:

Replace the manual validity checks for the GPIO with the
gpio_is_valid().

Signed-off-by: Arvind Yadav 
---
chnage in v2 :
   Returning invalid gpio as error instead of -ENODEV.

   drivers/staging/greybus/arche-platform.c | 6 +++---
   1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/greybus/arche-platform.c 
b/drivers/staging/greybus/arche-platform.c
index 83254a7..c3a7da5 100644
--- a/drivers/staging/greybus/arche-platform.c
+++ b/drivers/staging/greybus/arche-platform.c
@@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
"svc,reset-gpio",
0);
-   if (arche_pdata->svc_reset_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
dev_err(dev, "failed to get reset-gpio\n");
return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.

We need to check gpio validity. If we are using of_get_named_gpio() or
not.  of_get_name_gpio() will read a device node and fetch the value.
But it'll not check that gpio is valid or not valid.

No, I believe you're mistaken here. of_get_named_gpio() does not return
an arbitrary gpio number, unlike what you could possibly find in
legacy board files and for which the gpio_is_valid() helper made sense.

Johan

Yes, You are coorect. It'll read gpio form gpio device node. Which means
it'll read from device tree node. without finding a valid entry. It'll 
return

an error.

~arvind

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Johan Hovold]

On Wed, May 02, 2018 at 03:15:05PM +0530, Arvind Yadav wrote:

> On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:
> > On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:
> >> Replace the manual validity checks for the GPIO with the
> >> gpio_is_valid().
> >>
> >> Signed-off-by: Arvind Yadav 
> >> ---
> >> chnage in v2 :
> >>   Returning invalid gpio as error instead of -ENODEV.
> >>
> >>   drivers/staging/greybus/arche-platform.c | 6 +++---
> >>   1 file changed, 3 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/staging/greybus/arche-platform.c 
> >> b/drivers/staging/greybus/arche-platform.c
> >> index 83254a7..c3a7da5 100644
> >> --- a/drivers/staging/greybus/arche-platform.c
> >> +++ b/drivers/staging/greybus/arche-platform.c
> >> @@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
> >> *pdev)
> >>arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
> >>"svc,reset-gpio",
> >>0);
> >> -  if (arche_pdata->svc_reset_gpio < 0) {
> >> +  if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
> >>dev_err(dev, "failed to get reset-gpio\n");
> >>return arche_pdata->svc_reset_gpio;
> >
> > I'm sorry, but I don't this change is desirable. of_get_named_gpio()
> > returns a valid gpio number or a negative errno, so there's no need to
> > use the legacy gpio_is_valid() helper here.
> >
> > If you grep for of_get_named_gpio() you'll find that some drivers indeed
> > use that helper this way, but they are in a clear minority.
> >
> > And ultimately, we want to move to using gpio descriptors anyway.
> 
> We need to check gpio validity. If we are using of_get_named_gpio() or
> not.  of_get_name_gpio() will read a device node and fetch the value.
> But it'll not check that gpio is valid or not valid.

No, I believe you're mistaken here. of_get_named_gpio() does not return
an arbitrary gpio number, unlike what you could possibly find in
legacy board files and for which the gpio_is_valid() helper made sense.

Johan

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Johan Hovold]

On Wed, May 02, 2018 at 03:15:05PM +0530, Arvind Yadav wrote:

> On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:
> > On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:
> >> Replace the manual validity checks for the GPIO with the
> >> gpio_is_valid().
> >>
> >> Signed-off-by: Arvind Yadav 
> >> ---
> >> chnage in v2 :
> >>   Returning invalid gpio as error instead of -ENODEV.
> >>
> >>   drivers/staging/greybus/arche-platform.c | 6 +++---
> >>   1 file changed, 3 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/staging/greybus/arche-platform.c 
> >> b/drivers/staging/greybus/arche-platform.c
> >> index 83254a7..c3a7da5 100644
> >> --- a/drivers/staging/greybus/arche-platform.c
> >> +++ b/drivers/staging/greybus/arche-platform.c
> >> @@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
> >> *pdev)
> >>arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
> >>"svc,reset-gpio",
> >>0);
> >> -  if (arche_pdata->svc_reset_gpio < 0) {
> >> +  if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
> >>dev_err(dev, "failed to get reset-gpio\n");
> >>return arche_pdata->svc_reset_gpio;
> >
> > I'm sorry, but I don't this change is desirable. of_get_named_gpio()
> > returns a valid gpio number or a negative errno, so there's no need to
> > use the legacy gpio_is_valid() helper here.
> >
> > If you grep for of_get_named_gpio() you'll find that some drivers indeed
> > use that helper this way, but they are in a clear minority.
> >
> > And ultimately, we want to move to using gpio descriptors anyway.
> 
> We need to check gpio validity. If we are using of_get_named_gpio() or
> not.  of_get_name_gpio() will read a device node and fetch the value.
> But it'll not check that gpio is valid or not valid.

No, I believe you're mistaken here. of_get_named_gpio() does not return
an arbitrary gpio number, unlike what you could possibly find in
legacy board files and for which the gpio_is_valid() helper made sense.

Johan

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Arvind Yadav]

On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:

Replace the manual validity checks for the GPIO with the
gpio_is_valid().

Signed-off-by: Arvind Yadav 
---
chnage in v2 :
  Returning invalid gpio as error instead of -ENODEV.

  drivers/staging/greybus/arche-platform.c | 6 +++---
  1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/greybus/arche-platform.c 
b/drivers/staging/greybus/arche-platform.c
index 83254a7..c3a7da5 100644
--- a/drivers/staging/greybus/arche-platform.c
+++ b/drivers/staging/greybus/arche-platform.c
@@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
"svc,reset-gpio",
0);
-   if (arche_pdata->svc_reset_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
dev_err(dev, "failed to get reset-gpio\n");
return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.


We need to check gpio validity. If we are using of_get_named_gpio() or not.
of_get_name_gpio() will read a device node and fetch the value. But 
it'll not

check that gpio is valid or not valid.



}
@@ -468,7 +468,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_sysboot_gpio = of_get_named_gpio(np,
  "svc,sysboot-gpio",
  0);
-   if (arche_pdata->svc_sysboot_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_sysboot_gpio)) {
dev_err(dev, "failed to get sysboot gpio\n");
return arche_pdata->svc_sysboot_gpio;
}
@@ -487,7 +487,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_refclk_req = of_get_named_gpio(np,
"svc,refclk-req-gpio",
0);
-   if (arche_pdata->svc_refclk_req < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_refclk_req)) {
dev_err(dev, "failed to get svc clock-req gpio\n");
return arche_pdata->svc_refclk_req;
}

But if this were to be changed, you'd need to update also the fourth
call to of_get_named_gpio() in this function.


Sorry, I missed it. Thanks for point out. I will add check for this.

~arvind


Thanks,
Johan

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Arvind Yadav]

On Wednesday 02 May 2018 02:13 PM, Johan Hovold wrote:

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:

Replace the manual validity checks for the GPIO with the
gpio_is_valid().

Signed-off-by: Arvind Yadav 
---
chnage in v2 :
  Returning invalid gpio as error instead of -ENODEV.

  drivers/staging/greybus/arche-platform.c | 6 +++---
  1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/staging/greybus/arche-platform.c 
b/drivers/staging/greybus/arche-platform.c
index 83254a7..c3a7da5 100644
--- a/drivers/staging/greybus/arche-platform.c
+++ b/drivers/staging/greybus/arche-platform.c
@@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
"svc,reset-gpio",
0);
-   if (arche_pdata->svc_reset_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
dev_err(dev, "failed to get reset-gpio\n");
return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.


We need to check gpio validity. If we are using of_get_named_gpio() or not.
of_get_name_gpio() will read a device node and fetch the value. But 
it'll not

check that gpio is valid or not valid.



}
@@ -468,7 +468,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_sysboot_gpio = of_get_named_gpio(np,
  "svc,sysboot-gpio",
  0);
-   if (arche_pdata->svc_sysboot_gpio < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_sysboot_gpio)) {
dev_err(dev, "failed to get sysboot gpio\n");
return arche_pdata->svc_sysboot_gpio;
}
@@ -487,7 +487,7 @@ static int arche_platform_probe(struct platform_device 
*pdev)
arche_pdata->svc_refclk_req = of_get_named_gpio(np,
"svc,refclk-req-gpio",
0);
-   if (arche_pdata->svc_refclk_req < 0) {
+   if (!gpio_is_valid(arche_pdata->svc_refclk_req)) {
dev_err(dev, "failed to get svc clock-req gpio\n");
return arche_pdata->svc_refclk_req;
}

But if this were to be changed, you'd need to update also the fourth
call to of_get_named_gpio() in this function.


Sorry, I missed it. Thanks for point out. I will add check for this.

~arvind


Thanks,
Johan

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Johan Hovold]

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:
> Replace the manual validity checks for the GPIO with the
> gpio_is_valid().
> 
> Signed-off-by: Arvind Yadav 
> ---
> chnage in v2 :
>  Returning invalid gpio as error instead of -ENODEV.
> 
>  drivers/staging/greybus/arche-platform.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/staging/greybus/arche-platform.c 
> b/drivers/staging/greybus/arche-platform.c
> index 83254a7..c3a7da5 100644
> --- a/drivers/staging/greybus/arche-platform.c
> +++ b/drivers/staging/greybus/arche-platform.c
> @@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
>   "svc,reset-gpio",
>   0);
> - if (arche_pdata->svc_reset_gpio < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
>   dev_err(dev, "failed to get reset-gpio\n");
>   return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.

>   }
> @@ -468,7 +468,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_sysboot_gpio = of_get_named_gpio(np,
> "svc,sysboot-gpio",
> 0);
> - if (arche_pdata->svc_sysboot_gpio < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_sysboot_gpio)) {
>   dev_err(dev, "failed to get sysboot gpio\n");
>   return arche_pdata->svc_sysboot_gpio;
>   }
> @@ -487,7 +487,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_refclk_req = of_get_named_gpio(np,
>   "svc,refclk-req-gpio",
>   0);
> - if (arche_pdata->svc_refclk_req < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_refclk_req)) {
>   dev_err(dev, "failed to get svc clock-req gpio\n");
>   return arche_pdata->svc_refclk_req;
>   }

But if this were to be changed, you'd need to update also the fourth
call to of_get_named_gpio() in this function.

Thanks,
Johan

Re: [PATCH v2] staging: greybus: Use gpio_is_valid()

[Johan Hovold]

On Sat, Apr 28, 2018 at 10:05:39AM +0530, Arvind Yadav wrote:
> Replace the manual validity checks for the GPIO with the
> gpio_is_valid().
> 
> Signed-off-by: Arvind Yadav 
> ---
> chnage in v2 :
>  Returning invalid gpio as error instead of -ENODEV.
> 
>  drivers/staging/greybus/arche-platform.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/staging/greybus/arche-platform.c 
> b/drivers/staging/greybus/arche-platform.c
> index 83254a7..c3a7da5 100644
> --- a/drivers/staging/greybus/arche-platform.c
> +++ b/drivers/staging/greybus/arche-platform.c
> @@ -448,7 +448,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_reset_gpio = of_get_named_gpio(np,
>   "svc,reset-gpio",
>   0);
> - if (arche_pdata->svc_reset_gpio < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_reset_gpio)) {
>   dev_err(dev, "failed to get reset-gpio\n");
>   return arche_pdata->svc_reset_gpio;

I'm sorry, but I don't this change is desirable. of_get_named_gpio()
returns a valid gpio number or a negative errno, so there's no need to
use the legacy gpio_is_valid() helper here.

If you grep for of_get_named_gpio() you'll find that some drivers indeed
use that helper this way, but they are in a clear minority.

And ultimately, we want to move to using gpio descriptors anyway.

>   }
> @@ -468,7 +468,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_sysboot_gpio = of_get_named_gpio(np,
> "svc,sysboot-gpio",
> 0);
> - if (arche_pdata->svc_sysboot_gpio < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_sysboot_gpio)) {
>   dev_err(dev, "failed to get sysboot gpio\n");
>   return arche_pdata->svc_sysboot_gpio;
>   }
> @@ -487,7 +487,7 @@ static int arche_platform_probe(struct platform_device 
> *pdev)
>   arche_pdata->svc_refclk_req = of_get_named_gpio(np,
>   "svc,refclk-req-gpio",
>   0);
> - if (arche_pdata->svc_refclk_req < 0) {
> + if (!gpio_is_valid(arche_pdata->svc_refclk_req)) {
>   dev_err(dev, "failed to get svc clock-req gpio\n");
>   return arche_pdata->svc_refclk_req;
>   }

But if this were to be changed, you'd need to update also the fourth
call to of_get_named_gpio() in this function.

Thanks,
Johan