[RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[Krzysztof Kozlowski]

The power_supply_get_by_phandle() on error returns ENODEV or NULL.
The driver later expects obtained pointer to power supply to be
valid or NULL. If it is not NULL then it dereferences it in
bq2415x_notifier_call() which would lead to dereferencing ENODEV-value
pointer.

Properly handle the power_supply_get_by_phandle() error case and abort probe.

Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
Fixes: faffd234cf85 ("bq2415x_charger: Add DT support")
Cc: <stable@vger.kernel.org>
---
 drivers/power/bq2415x_charger.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/power/bq2415x_charger.c b/drivers/power/bq2415x_charger.c
index e384844a1ae1..d9c457217b6a 100644
--- a/drivers/power/bq2415x_charger.c
+++ b/drivers/power/bq2415x_charger.c
@@ -1579,8 +1579,13 @@ static int bq2415x_probe(struct i2c_client *client,
 	if (np) {
 		bq->notify_psy = power_supply_get_by_phandle(np, "ti,usb-charger-detection");
 
-		if (!bq->notify_psy)
+		if (!bq->notify_psy) {
 			return -EPROBE_DEFER;
+		} else if (IS_ERR(bq->notify_psy)) {
+			dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
+			ret = PTR_ERR(bq->notify_psy);
+			goto error_2;
+		}
 	}
 	else if (pdata->notify_device)
 		bq->notify_psy = power_supply_get_by_name(pdata->notify_device);
-- 
1.9.1

Re: [RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[Sebastian Reichel]

[-- Attachment #1: Type: text/plain, Size: 1870 bytes --]

Hi Krzysztof,

On Tue, Oct 14, 2014 at 10:32:53AM +0200, Krzysztof Kozlowski wrote:
> The power_supply_get_by_phandle() on error returns ENODEV or NULL.
> The driver later expects obtained pointer to power supply to be
> valid or NULL. If it is not NULL then it dereferences it in
> bq2415x_notifier_call() which would lead to dereferencing ENODEV-value
> pointer.
> 
> Properly handle the power_supply_get_by_phandle() error case and abort probe.
> 
> Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
> Fixes: faffd234cf85 ("bq2415x_charger: Add DT support")
> Cc: <stable@vger.kernel.org>

That looks valid, but I guess this should change one more thing:

> ---
>  drivers/power/bq2415x_charger.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/power/bq2415x_charger.c b/drivers/power/bq2415x_charger.c
> index e384844a1ae1..d9c457217b6a 100644
> --- a/drivers/power/bq2415x_charger.c
> +++ b/drivers/power/bq2415x_charger.c
> @@ -1579,8 +1579,13 @@ static int bq2415x_probe(struct i2c_client *client,
>  	if (np) {
>  		bq->notify_psy = power_supply_get_by_phandle(np, "ti,usb-charger-detection");
>  
> -		if (!bq->notify_psy)
> +		if (!bq->notify_psy) {
>  			return -EPROBE_DEFER;

this also needs to goto error_2. I guess something as the following
is needed:

if (IS_ERR_OR_NULL(bq->notify_psy)) {
    if (bq->notify_psy)
        dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
    ret = PTR_ERR(bq->notify_psy);
    goto error_2;
}

> +		} else if (IS_ERR(bq->notify_psy)) {
> +			dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
> +			ret = PTR_ERR(bq->notify_psy);
> +			goto error_2;
> +		}
>  	}
>  	else if (pdata->notify_device)
>  		bq->notify_psy = power_supply_get_by_name(pdata->notify_device);

-- Sebastian

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[Krzysztof Kozlowski]

On śro, 2014-10-15 at 11:05 +0200, Sebastian Reichel wrote:
> Hi Krzysztof,
> 
> On Tue, Oct 14, 2014 at 10:32:53AM +0200, Krzysztof Kozlowski wrote:
> > The power_supply_get_by_phandle() on error returns ENODEV or NULL.
> > The driver later expects obtained pointer to power supply to be
> > valid or NULL. If it is not NULL then it dereferences it in
> > bq2415x_notifier_call() which would lead to dereferencing ENODEV-value
> > pointer.
> > 
> > Properly handle the power_supply_get_by_phandle() error case and abort probe.
> > 
> > Signed-off-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>
> > Fixes: faffd234cf85 ("bq2415x_charger: Add DT support")
> > Cc: <stable@vger.kernel.org>
> 
> That looks valid, but I guess this should change one more thing:
> 
> > ---
> >  drivers/power/bq2415x_charger.c | 7 ++++++-
> >  1 file changed, 6 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/power/bq2415x_charger.c b/drivers/power/bq2415x_charger.c
> > index e384844a1ae1..d9c457217b6a 100644
> > --- a/drivers/power/bq2415x_charger.c
> > +++ b/drivers/power/bq2415x_charger.c
> > @@ -1579,8 +1579,13 @@ static int bq2415x_probe(struct i2c_client *client,
> >  	if (np) {
> >  		bq->notify_psy = power_supply_get_by_phandle(np, "ti,usb-charger-detection");
> >  
> > -		if (!bq->notify_psy)
> > +		if (!bq->notify_psy) {
> >  			return -EPROBE_DEFER;
> 
> this also needs to goto error_2. I guess something as the following
> is needed:
> 
> if (IS_ERR_OR_NULL(bq->notify_psy)) {
>     if (bq->notify_psy)
>         dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
>     ret = PTR_ERR(bq->notify_psy);
>     goto error_2;
> }

So you do not want to defer the probe? What if notified charger will
come online after this probe?

Second idea - now I think my change is not compatible with bindings
(documentation) and could break booting of existing boards which do not
provide the "ti,usb-charger-detection" property. For example
arch/arm/boot/dts/omap3-n900.dts

The "ti,usb-charger-detection" property is marked as optional but my
patch actually makes it required.

It should be rather like this:

if (IS_ERR(bq->notify_psy)) {
    bq->notify_psy = NULL;
    dev_info(&client->dev, "no 'ti,usb-charger-detection' property \n");
} else if (!bq->notify_psy) {
    ret = -EPROBE_DEFER;
    goto error_2;
}

What do you think?

Krzysztof

> > +		} else if (IS_ERR(bq->notify_psy)) {
> > +			dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
> > +			ret = PTR_ERR(bq->notify_psy);
> > +			goto error_2;
> > +		}
> >  	}
> >  	else if (pdata->notify_device)
> >  		bq->notify_psy = power_supply_get_by_name(pdata->notify_device);
> 
> -- Sebastian

Re: [RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[Sebastian Reichel]

[-- Attachment #1: Type: text/plain, Size: 1594 bytes --]

Hi,

On Wed, Oct 15, 2014 at 01:11:31PM +0200, Krzysztof Kozlowski wrote:
> > [...] I guess something as the following is needed:
> > 
> > if (IS_ERR_OR_NULL(bq->notify_psy)) {
> >     if (bq->notify_psy)
> >         dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
> >     ret = PTR_ERR(bq->notify_psy);
> >     goto error_2;
> > }
> 
> So you do not want to defer the probe? What if notified charger will
> come online after this probe?

No, but resources must be freed fore the EPROBE_DEFER case, too.
You are right, though, that my snipped does not setup the
EPROBE_DEFER return value correctly.

> Second idea - now I think my change is not compatible with bindings
> (documentation) and could break booting of existing boards which do not
> provide the "ti,usb-charger-detection" property. For example
> arch/arm/boot/dts/omap3-n900.dts
>
> The "ti,usb-charger-detection" property is marked as optional but my
> patch actually makes it required.
> 
> It should be rather like this:
> 
> if (IS_ERR(bq->notify_psy)) {
>     bq->notify_psy = NULL;
>     dev_info(&client->dev, "no 'ti,usb-charger-detection' property \n");
> } else if (!bq->notify_psy) {
>     ret = -EPROBE_DEFER;
>     goto error_2;
> }
>
> What do you think?

I thing the dev_info call should include the error code returned,
since its not propagated further:

dev_info(&client->dev, "no 'ti,usb-charger-detection' property (err=%d)\n",
         PTR_ERR(by->notify_psy));

Apart from that it looks fine to me. Can you send a v2?

> [...]

-- Sebastian

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [RFT PATCH] power: bq2415x_charger: Properly handle ENODEV from power_supply_get_by_phandle

[Krzysztof Kozlowski]

On śro, 2014-10-15 at 15:58 +0200, Sebastian Reichel wrote:
> Hi,
> 
> On Wed, Oct 15, 2014 at 01:11:31PM +0200, Krzysztof Kozlowski wrote:
> > > [...] I guess something as the following is needed:
> > > 
> > > if (IS_ERR_OR_NULL(bq->notify_psy)) {
> > >     if (bq->notify_psy)
> > >         dev_err(&client->dev, "no 'ti,usb-charger-detection' property\n");
> > >     ret = PTR_ERR(bq->notify_psy);
> > >     goto error_2;
> > > }
> > 
> > So you do not want to defer the probe? What if notified charger will
> > come online after this probe?
> 
> No, but resources must be freed fore the EPROBE_DEFER case, too.
> You are right, though, that my snipped does not setup the
> EPROBE_DEFER return value correctly.
> 
> > Second idea - now I think my change is not compatible with bindings
> > (documentation) and could break booting of existing boards which do not
> > provide the "ti,usb-charger-detection" property. For example
> > arch/arm/boot/dts/omap3-n900.dts
> >
> > The "ti,usb-charger-detection" property is marked as optional but my
> > patch actually makes it required.
> > 
> > It should be rather like this:
> > 
> > if (IS_ERR(bq->notify_psy)) {
> >     bq->notify_psy = NULL;
> >     dev_info(&client->dev, "no 'ti,usb-charger-detection' property \n");
> > } else if (!bq->notify_psy) {
> >     ret = -EPROBE_DEFER;
> >     goto error_2;
> > }
> >
> > What do you think?
> 
> I thing the dev_info call should include the error code returned,
> since its not propagated further:
> 
> dev_info(&client->dev, "no 'ti,usb-charger-detection' property (err=%d)\n",
>          PTR_ERR(by->notify_psy));
> 
> Apart from that it looks fine to me. Can you send a v2?

Sure, I'll send fixed version.

Best regards,
Krzysztof
> 
> > [...]
> 
> -- Sebastian