[PATCH 1/2] usb: bdc: Fix an error handling path in 'bdc_probe()' when no suitable DMA config is available

[PATCH 1/2] usb: bdc: Fix an error handling path in 'bdc_probe()' when no suitable DMA config is available

[Christophe JAILLET]

If no suitable DMA configuration is available, a previous 'bdc_phy_init()'
call must be undone by a corresponding 'bdc_phy_exit()' call.

Branch to the existing error handling path instead of returning
directly.

Fixes: cc29d4f67757 ("usb: bdc: Add support for USB phy")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
checkpatch.pl warns that:
   WARNING: ENOTSUPP is not a SUSV4 error code, prefer EOPNOTSUPP
   #29: FILE: drivers/usb/gadget/udc/bdc/bdc_core.c:563:
   +			ret = -ENOTSUPP;
I've never seen this warning before and don't want to make a blind fix for that.
Let me know if I should fix it or not.
---
drivers/usb/gadget/udc/bdc/bdc_core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/gadget/udc/bdc/bdc_core.c b/drivers/usb/gadget/udc/bdc/bdc_core.c
index 0bef6b3f049b..251db57e51fa 100644
--- a/drivers/usb/gadget/udc/bdc/bdc_core.c
+++ b/drivers/usb/gadget/udc/bdc/bdc_core.c
@@ -560,7 +560,8 @@ static int bdc_probe(struct platform_device *pdev)
 		if (ret) {
 			dev_err(dev,
 				"No suitable DMA config available, abort\n");
-			return -ENOTSUPP;
+			ret = -ENOTSUPP;
+			goto phycleanup;
 		}
 		dev_dbg(dev, "Using 32-bit address\n");
 	}
-- 
2.30.2

[PATCH 2/2] usb: bdc: Fix a resource leak in the error handling path of 'bdc_probe()'

[Christophe JAILLET]

If an error occurs after a successful 'clk_prepare_enable()' call, it must
be undone by a corresponding 'clk_disable_unprepare()' call.
This call is already present in the remove function.

Add this call in the error handling path and reorder the code so that the
'clk_prepare_enable()' call happens later in the function.
The goal is to have as much managed resources functions as possible
before the 'clk_prepare_enable()' call in order to keep the error handling
path simple.

While at it, remove the now unneeded 'clk' variable.

Fixes: c87dca047849 ("usb: bdc: Add clock enable for new chips with a separate BDC clock")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
---
Review with care.
I don't like shuffling code like that because of possible side effect.
Moving the code related to this clk looks fine to me, but who knows...
---
 drivers/usb/gadget/udc/bdc/bdc_core.c | 27 +++++++++++++--------------
 1 file changed, 13 insertions(+), 14 deletions(-)

diff --git a/drivers/usb/gadget/udc/bdc/bdc_core.c b/drivers/usb/gadget/udc/bdc/bdc_core.c
index 251db57e51fa..fa1a3908ec3b 100644
--- a/drivers/usb/gadget/udc/bdc/bdc_core.c
+++ b/drivers/usb/gadget/udc/bdc/bdc_core.c
@@ -488,27 +488,14 @@ static int bdc_probe(struct platform_device *pdev)
 	int irq;
 	u32 temp;
 	struct device *dev = &pdev->dev;
-	struct clk *clk;
 	int phy_num;
 
 	dev_dbg(dev, "%s()\n", __func__);
 
-	clk = devm_clk_get_optional(dev, "sw_usbd");
-	if (IS_ERR(clk))
-		return PTR_ERR(clk);
-
-	ret = clk_prepare_enable(clk);
-	if (ret) {
-		dev_err(dev, "could not enable clock\n");
-		return ret;
-	}
-
 	bdc = devm_kzalloc(dev, sizeof(*bdc), GFP_KERNEL);
 	if (!bdc)
 		return -ENOMEM;
 
-	bdc->clk = clk;
-
 	bdc->regs = devm_platform_ioremap_resource(pdev, 0);
 	if (IS_ERR(bdc->regs))
 		return PTR_ERR(bdc->regs);
@@ -545,10 +532,20 @@ static int bdc_probe(struct platform_device *pdev)
 		}
 	}
 
+	bdc->clk = devm_clk_get_optional(dev, "sw_usbd");
+	if (IS_ERR(bdc->clk))
+		return PTR_ERR(bdc->clk);
+
+	ret = clk_prepare_enable(bdc->clk);
+	if (ret) {
+		dev_err(dev, "could not enable clock\n");
+		return ret;
+	}
+
 	ret = bdc_phy_init(bdc);
 	if (ret) {
 		dev_err(bdc->dev, "BDC phy init failure:%d\n", ret);
-		return ret;
+		goto disable_clk;
 	}
 
 	temp = bdc_readl(bdc->regs, BDC_BDCCAP1);
@@ -581,6 +578,8 @@ static int bdc_probe(struct platform_device *pdev)
 	bdc_hw_exit(bdc);
 phycleanup:
 	bdc_phy_exit(bdc);
+disable_clk:
+	clk_disable_unprepare(bdc->clk);
 	return ret;
 }
 
-- 
2.30.2

Re: [PATCH 1/2] usb: bdc: Fix an error handling path in 'bdc_probe()' when no suitable DMA config is available

[Florian Fainelli]

On 8/18/2021 9:32 PM, Christophe JAILLET wrote:
> If no suitable DMA configuration is available, a previous 'bdc_phy_init()'
> call must be undone by a corresponding 'bdc_phy_exit()' call.
> 
> Branch to the existing error handling path instead of returning
> directly.
> 
> Fixes: cc29d4f67757 ("usb: bdc: Add support for USB phy")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

Acked-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

Re: [PATCH 2/2] usb: bdc: Fix a resource leak in the error handling path of 'bdc_probe()'

[Florian Fainelli]

On 8/18/2021 9:32 PM, Christophe JAILLET wrote:
> If an error occurs after a successful 'clk_prepare_enable()' call, it must
> be undone by a corresponding 'clk_disable_unprepare()' call.
> This call is already present in the remove function.
> 
> Add this call in the error handling path and reorder the code so that the
> 'clk_prepare_enable()' call happens later in the function.
> The goal is to have as much managed resources functions as possible
> before the 'clk_prepare_enable()' call in order to keep the error handling
> path simple.
> 
> While at it, remove the now unneeded 'clk' variable.
> 
> Fixes: c87dca047849 ("usb: bdc: Add clock enable for new chips with a separate BDC clock")
> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>

Acked-by: Florian Fainelli <f.fainelli@gmail.com>

> ---
> Review with care.
> I don't like shuffling code like that because of possible side effect.
> Moving the code related to this clk looks fine to me, but who knows...

There are no register accesses until bdc_phy_init() gets called, so this 
looks fine to me. Al knows this code better than I do though, so it 
would be better to wait for his Acked-by tag.
-- 
Florian

Re: [PATCH 2/2] usb: bdc: Fix a resource leak in the error handling path of 'bdc_probe()'

[Alan Cooper]

On Fri, Aug 20, 2021 at 4:03 PM Florian Fainelli <f.fainelli@gmail.com> wrote:
>
>
>
> On 8/18/2021 9:32 PM, Christophe JAILLET wrote:
> > If an error occurs after a successful 'clk_prepare_enable()' call, it must
> > be undone by a corresponding 'clk_disable_unprepare()' call.
> > This call is already present in the remove function.
> >
> > Add this call in the error handling path and reorder the code so that the
> > 'clk_prepare_enable()' call happens later in the function.
> > The goal is to have as much managed resources functions as possible
> > before the 'clk_prepare_enable()' call in order to keep the error handling
> > path simple.
> >
> > While at it, remove the now unneeded 'clk' variable.
> >
> > Fixes: c87dca047849 ("usb: bdc: Add clock enable for new chips with a separate BDC clock")
> > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>
> Acked-by: Florian Fainelli <f.fainelli@gmail.com>
>
> > ---
> > Review with care.
> > I don't like shuffling code like that because of possible side effect.
> > Moving the code related to this clk looks fine to me, but who knows...
>
> There are no register accesses until bdc_phy_init() gets called, so this
> looks fine to me. Al knows this code better than I do though, so it
> would be better to wait for his Acked-by tag.

This is safe.
Acked-by: Al Cooper <alcooperx@gmail.com>

> --
> Florian