Skip to contentSkip to conversation

mfd: omap-usb-tll: handle clk_prepare return code in usbtll_omap_probe

PATCH In review Karol Przybylski 4 participants 8 messages Nov 6, 2024 – Nov 12, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mfd: omap-usb-tll: handle clk_prepare return code in usbtll_omap_probe

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +6 -2
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
 	struct device				*dev =  &pdev->dev;
 	struct usbtll_omap			*tll;
 	void __iomem				*base;
-	int					i, nch, ver;
+	int					i, nch, ver, err;
 
 	dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
 		if (IS_ERR(tll->ch_clk[i]))
 			dev_dbg(dev, "can't get clock : %s\n", clkname);
 		else
-			clk_prepare(tll->ch_clk[i]);
+			err = clk_prepare(tll->ch_clk[i]);
+			if (err) {
+				dev_err(dev, "Unable to prepare clock\n");
+				return err;
+	}
 	}
 
 	pm_runtime_put_sync(dev);
Signature · 1 line
2.34.1
#1 lore raw
Quoted 27 lines · Karol Przybylski

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
  drivers/mfd/omap-usb-tll.c | 8 ++++++--
  1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
  	struct device				*dev =  &pdev->dev;
  	struct usbtll_omap			*tll;
  	void __iomem				*base;
-	int					i, nch, ver;
+	int					i, nch, ver, err;
 
  	dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 254

Quoted 4 lines
drivers/mfd/omap-usb-tll.c +0 -1
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
  		if (IS_ERR(tll->ch_clk[i]))
  			dev_dbg(dev, "can't get clock : %s\n", clkname);
  		else
-			clk_prepare(tll->ch_clk[i]);

Braces for the conditional don't looks right.

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 257

Quoted 4 lines
drivers/mfd/omap-usb-tll.c +4 -0
+			err = clk_prepare(tll->ch_clk[i]);
+			if (err) {
+				dev_err(dev, "Unable to prepare clock\n");
+				return err;

Did you check to see if callers handle this new error return in this path?

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 258

Quoted 1 line
drivers/mfd/omap-usb-tll.c +1 -0
+	}

Same here

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 260

Quoted 2 lines
drivers/mfd/omap-usb-tll.c +0 -0
  	}
 

Same here

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 261

Quoted 1 line
drivers/mfd/omap-usb-tll.c +0 -0
  	pm_runtime_put_sync(dev);

thanks,
-- Shuah

maintainer
#2 lore raw
In reply to Karol Przybylski

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 27 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
 	struct device				*dev =  &pdev->dev;
 	struct usbtll_omap			*tll;
 	void __iomem				*base;
-	int					i, nch, ver;
+	int					i, nch, ver, err;
 
 	dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 2 lines
drivers/mfd/omap-usb-tll.c +0 -0
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
 		if (IS_ERR(tll->ch_clk[i]))
 			dev_dbg(dev, "can't get clock : %s\n", clkname);

if you add more intensive error checking, then why is this error ignored and not returned?

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 255

Quoted 4 lines
drivers/mfd/omap-usb-tll.c +2 -1
 		else
-			clk_prepare(tll->ch_clk[i]);
+			err = clk_prepare(tll->ch_clk[i]);
+			if (err) {

unnatural braces, if (err) is not in the else branch ?!

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 261

Quoted 6 lines
drivers/mfd/omap-usb-tll.c +3 -0
+				dev_err(dev, "Unable to prepare clock\n");
+				return err;
+	}
 	}
 
 	pm_runtime_put_sync(dev);

and this one is not called if you return early.

Regards,
Andreas

author
#3 lore raw
Quoted 28 lines · Andreas Kemnade

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 24 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
      struct device                           *dev =  &pdev->dev;
      struct usbtll_omap                      *tll;
      void __iomem                            *base;
-     int                                     i, nch, ver;
+     int                                     i, nch, ver, err;
 
Quoted 8 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +0 -0
      dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
              if (IS_ERR(tll->ch_clk[i]))
                      dev_dbg(dev, "can't get clock : %s\n", clkname);

if you add more intensive error checking, then why is this error ignored and not returned?

Thank you for the feedback. It does seem that elevated error checking is not the way
to go in this case. Do you think it would be good to add a similar statement instead of
my initial changes? It would look something like this:

+               else {
                        err = clk_prepare(tll->ch_clk[i]);

+ if (err)
+ dev_dbg(dev, "clock prepare error for:
%s\n", clkname);
+ }

Quoted 16 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 255

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +2 -1
              else
-                     clk_prepare(tll->ch_clk[i]);
+                     err = clk_prepare(tll->ch_clk[i]);
+                     if (err) {

unnatural braces, if (err) is not in the else branch ?!

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 261

Quoted 6 lines
drivers/mfd/omap-usb-tll.c +3 -0
+                             dev_err(dev, "Unable to prepare clock\n");
+                             return err;
+     }
      }
 
      pm_runtime_put_sync(dev);

and this one is not called if you return early.

Regards,
Andreas

maintainer
#4 lore raw

Am Thu, 7 Nov 2024 12:12:52 +0100
schrieb Karol P <[email protected]>:

Quoted 38 lines · Andreas Kemnade
Quoted 28 lines

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 24 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
      struct device                           *dev =  &pdev->dev;
      struct usbtll_omap                      *tll;
      void __iomem                            *base;
-     int                                     i, nch, ver;
+     int                                     i, nch, ver, err;
 
Quoted 8 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +0 -0
      dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
              if (IS_ERR(tll->ch_clk[i]))
                      dev_dbg(dev, "can't get clock : %s\n", clkname);  

if you add more intensive error checking, then why is this error ignored and not returned?

Quoted 3 lines

Thank you for the feedback. It does seem that elevated error checking is not the way
to go in this case.

As far as I can see everything checks ch_clk[i] for validity before usage. Also clk_enable() called later is checked which would catch clk_prepare() failures, if there were even possible here.

So the only question which I am not 100% sure about is whether having ch_clk sparsly populated is normal operation. If that is the case, then more error checking is not useful. If not, then it might let us better sleep. As said as far as I can see errors are catched later.

@Roger: what is your opintion towards this?

BTW: If you do this kind of work, you could also use W=1 or CONFIG_WERROR during compiling to catch easy things. At least I see new compile warnings with your patch.

Regards,
Andreas

maintainer
#5 lore raw

Hi,

Quoted 49 lines · Andreas Kemnade

Am Thu, 7 Nov 2024 12:12:52 +0100
schrieb Karol P <[email protected]>:

Quoted 38 lines · Andreas Kemnade
Quoted 28 lines

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 24 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
      struct device                           *dev =  &pdev->dev;
      struct usbtll_omap                      *tll;
      void __iomem                            *base;
-     int                                     i, nch, ver;
+     int                                     i, nch, ver, err;
 
Quoted 8 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +0 -0
      dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
              if (IS_ERR(tll->ch_clk[i]))
                      dev_dbg(dev, "can't get clock : %s\n", clkname);  

if you add more intensive error checking, then why is this error ignored and not returned?

Quoted 3 lines

Thank you for the feedback. It does seem that elevated error checking is not the way
to go in this case.

As far as I can see everything checks ch_clk[i] for validity before usage. Also clk_enable() called later is checked which would catch clk_prepare() failures, if there were even possible here.

Quoted 6 lines

So the only question which I am not 100% sure about is whether having ch_clk sparsly populated is normal operation. If that is the case, then more error checking is not useful. If not, then it might let us better sleep. As said as far as I can see errors are catched later.

@Roger: what is your opintion towards this?

I don't see usb_tll_hs_usb_ch?_clk in any of the OMAP device trees. Could it be that they are optional?
If so then we could convert it to devm_clk_get_optional()?

While at that, maybe the device tree binding could also be updated and converted to yaml.

Quoted 7 lines

BTW: If you do this kind of work, you could also use W=1 or CONFIG_WERROR during compiling to catch easy things. At least I see new compile warnings with your patch.

Regards,
Andreas

Signature · 2 lines
cheers,
-roger
maintainer
#6 lore raw

Am Mon, 11 Nov 2024 16:41:47 +0200
schrieb Roger Quadros <[email protected]>:

Quoted 57 lines

Hi,

Quoted 54 lines · Andreas Kemnade

Am Thu, 7 Nov 2024 12:12:52 +0100
schrieb Karol P <[email protected]>:

Quoted 38 lines · Andreas Kemnade
Quoted 28 lines

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 24 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
      struct device                           *dev =  &pdev->dev;
      struct usbtll_omap                      *tll;
      void __iomem                            *base;
-     int                                     i, nch, ver;
+     int                                     i, nch, ver, err;
 
Quoted 8 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +0 -0
      dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
              if (IS_ERR(tll->ch_clk[i]))
                      dev_dbg(dev, "can't get clock : %s\n", clkname);    

if you add more intensive error checking, then why is this error ignored and not returned?

Quoted 3 lines

Thank you for the feedback. It does seem that elevated error checking is not the way
to go in this case.

As far as I can see everything checks ch_clk[i] for validity before usage. Also clk_enable() called later is checked which would catch clk_prepare() failures, if there were even possible here.

So the only question which I am not 100% sure about is whether having ch_clk sparsly populated is normal operation. If that is the case, then more error checking is not useful. If not, then it might let us better sleep. As said as far as I can see errors are catched later.

Quoted 6 lines
Quoted 1 line

@Roger: what is your opintion towards this?

I don't see usb_tll_hs_usb_ch?_clk in any of the OMAP device trees. Could it be that they are optional?
If so then we could convert it to devm_clk_get_optional()?

They live in drivers/clk/ti/clk-[54]4xx.c
But nothing about omap3. So apparently we can have valid use cases where these clocks are not available. So no real need more anything more than dev_dbg output here.

Regards,
Andreas

author
#7 lore raw
Quoted 67 lines · Andreas Kemnade

Am Mon, 11 Nov 2024 16:41:47 +0200
schrieb Roger Quadros <[email protected]>:

Quoted 57 lines

Hi,

Quoted 54 lines · Andreas Kemnade

Am Thu, 7 Nov 2024 12:12:52 +0100
schrieb Karol P <[email protected]>:

Quoted 38 lines · Andreas Kemnade
Quoted 28 lines

Am Wed, 6 Nov 2024 23:33:24 +0100
schrieb Karol Przybylski <[email protected]>:

Quoted 24 lines

clk_prepare() is called in usbtll_omap_probe to fill clk array. Return code is not checked, leaving possible error condition unhandled.

Added variable to hold return value from clk_prepare() and return statement when it's not successful.

Found in coverity scan, CID 1594680

---
 drivers/mfd/omap-usb-tll.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/mfd/omap-usb-tll.c +1 -1
index 0f7fdb99c809..28446b082c85 100644
@@ -202,7 +202,7 @@ static int usbtll_omap_probe(struct platform_device *pdev)
      struct device                           *dev =  &pdev->dev;
      struct usbtll_omap                      *tll;
      void __iomem                            *base;
-     int                                     i, nch, ver;
+     int                                     i, nch, ver, err;
 
Quoted 8 lines

drivers/mfd/omap-usb-tll.c static int usbtll_omap_probe(struct platform_device *pdev) · line 252

Quoted 5 lines
drivers/mfd/omap-usb-tll.c +0 -0
      dev_dbg(dev, "starting TI HSUSB TLL Controller\n");
 
@@ -251,7 +251,11 @@ static int usbtll_omap_probe(struct platform_device *pdev)
              if (IS_ERR(tll->ch_clk[i]))
                      dev_dbg(dev, "can't get clock : %s\n", clkname);

if you add more intensive error checking, then why is this error ignored and not returned?

Quoted 3 lines

Thank you for the feedback. It does seem that elevated error checking is not the way
to go in this case.

As far as I can see everything checks ch_clk[i] for validity before usage. Also clk_enable() called later is checked which would catch clk_prepare() failures, if there were even possible here.

So the only question which I am not 100% sure about is whether having ch_clk sparsly populated is normal operation. If that is the case, then more error checking is not useful. If not, then it might let us better sleep. As said as far as I can see errors are catched later.

Quoted 6 lines
Quoted 1 line

@Roger: what is your opintion towards this?

I don't see usb_tll_hs_usb_ch?_clk in any of the OMAP device trees. Could it be that they are optional?
If so then we could convert it to devm_clk_get_optional()?

Quoted 7 lines

They live in drivers/clk/ti/clk-[54]4xx.c
But nothing about omap3. So apparently we can have valid use cases where these clocks are not available. So no real need more anything more than dev_dbg output here.

Regards,
Andreas

Thanks, I will send in a new patch then. I can also take a look at updating device tree binding.

Best regards,
Karol