[PATCH net v2] net: stmmac: add clk_prepare_enable() error handling

[PATCH net v2] net: stmmac: add clk_prepare_enable() error handling

[Pavel Zhigulin]

The driver previously ignored the return value of 'clk_prepare_enable()'
for both the CSR clock and the PCLK in 'stmmac_probe_config_dt()' function.

Add 'clk_prepare_enable()' return value checks.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: bfab27a146ed ("stmmac: add the experimental PCI support")
Signed-off-by: Pavel Zhigulin <Pavel.Zhigulin@kaspersky.com>
---
v2: Fix 'ret' value initialization after build bot notification.
v1: https://lore.kernel.org/all/20251113134009.79440-1-Pavel.Zhigulin@kaspersky.com/

 drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
index 27bcaae07a7f..8f9eb9683d2b 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
@@ -632,7 +632,9 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
 			dev_warn(&pdev->dev, "Cannot get CSR clock\n");
 			plat->stmmac_clk = NULL;
 		}
-		clk_prepare_enable(plat->stmmac_clk);
+		rc = clk_prepare_enable(plat->stmmac_clk);
+		if (rc < 0)
+			dev_warn(&pdev->dev, "Cannot enable CSR clock: %d\n", rc);
 	}

 	plat->pclk = devm_clk_get_optional(&pdev->dev, "pclk");
@@ -640,7 +642,12 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
 		ret = plat->pclk;
 		goto error_pclk_get;
 	}
-	clk_prepare_enable(plat->pclk);
+	rc = clk_prepare_enable(plat->pclk);
+	if (rc < 0) {
+		ret = ERR_PTR(rc);
+		dev_err(&pdev->dev, "Cannot enable pclk: %d\n", rc);
+		goto error_pclk_get;
+	}

 	/* Fall-back to main clock in case of no PTP ref is passed */
 	plat->clk_ptp_ref = devm_clk_get(&pdev->dev, "ptp_ref");
--
2.43.0

Re: [PATCH net v2] net: stmmac: add clk_prepare_enable() error handling

[Paolo Abeni]

On 11/14/25 3:23 PM, Pavel Zhigulin wrote:
> The driver previously ignored the return value of 'clk_prepare_enable()'
> for both the CSR clock and the PCLK in 'stmmac_probe_config_dt()' function.
> 
> Add 'clk_prepare_enable()' return value checks.
> 
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
> 
> Fixes: bfab27a146ed ("stmmac: add the experimental PCI support")
> Signed-off-by: Pavel Zhigulin <Pavel.Zhigulin@kaspersky.com>
> ---
> v2: Fix 'ret' value initialization after build bot notification.
> v1: https://lore.kernel.org/all/20251113134009.79440-1-Pavel.Zhigulin@kaspersky.com/
> 
>  drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 11 +++++++++--
>  1 file changed, 9 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> index 27bcaae07a7f..8f9eb9683d2b 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> @@ -632,7 +632,9 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
>  			dev_warn(&pdev->dev, "Cannot get CSR clock\n");
>  			plat->stmmac_clk = NULL;
>  		}
> -		clk_prepare_enable(plat->stmmac_clk);
> +		rc = clk_prepare_enable(plat->stmmac_clk);
> +		if (rc < 0)
> +			dev_warn(&pdev->dev, "Cannot enable CSR clock: %d\n", rc);
>  	}
> 
>  	plat->pclk = devm_clk_get_optional(&pdev->dev, "pclk");
> @@ -640,7 +642,12 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
>  		ret = plat->pclk;
>  		goto error_pclk_get;
>  	}
> -	clk_prepare_enable(plat->pclk);
> +	rc = clk_prepare_enable(plat->pclk);
> +	if (rc < 0) {
> +		ret = ERR_PTR(rc);
> +		dev_err(&pdev->dev, "Cannot enable pclk: %d\n", rc);
> +		goto error_pclk_get;
> +	}

It looks like the driver is supposed to handle the
IS_ERR_OR_NULL(plat->pclk) condition. This check could cause regression
on existing setup currently failing to initialize the (optional) clock
and still being functional.

I *think* we are better off without the added checks.

/P

Re: [PATCH net v2] net: stmmac: add clk_prepare_enable() error handling

[Paolo Abeni]

On 11/18/25 3:30 PM, Paolo Abeni wrote:
> On 11/14/25 3:23 PM, Pavel Zhigulin wrote:
>> The driver previously ignored the return value of 'clk_prepare_enable()'
>> for both the CSR clock and the PCLK in 'stmmac_probe_config_dt()' function.
>>
>> Add 'clk_prepare_enable()' return value checks.
>>
>> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>>
>> Fixes: bfab27a146ed ("stmmac: add the experimental PCI support")
>> Signed-off-by: Pavel Zhigulin <Pavel.Zhigulin@kaspersky.com>
>> ---
>> v2: Fix 'ret' value initialization after build bot notification.
>> v1: https://lore.kernel.org/all/20251113134009.79440-1-Pavel.Zhigulin@kaspersky.com/
>>
>>  drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 11 +++++++++--
>>  1 file changed, 9 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
>> index 27bcaae07a7f..8f9eb9683d2b 100644
>> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
>> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
>> @@ -632,7 +632,9 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
>>  			dev_warn(&pdev->dev, "Cannot get CSR clock\n");
>>  			plat->stmmac_clk = NULL;
>>  		}
>> -		clk_prepare_enable(plat->stmmac_clk);
>> +		rc = clk_prepare_enable(plat->stmmac_clk);
>> +		if (rc < 0)
>> +			dev_warn(&pdev->dev, "Cannot enable CSR clock: %d\n", rc);
>>  	}
>>
>>  	plat->pclk = devm_clk_get_optional(&pdev->dev, "pclk");
>> @@ -640,7 +642,12 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
>>  		ret = plat->pclk;
>>  		goto error_pclk_get;
>>  	}
>> -	clk_prepare_enable(plat->pclk);
>> +	rc = clk_prepare_enable(plat->pclk);
>> +	if (rc < 0) {
>> +		ret = ERR_PTR(rc);
>> +		dev_err(&pdev->dev, "Cannot enable pclk: %d\n", rc);
>> +		goto error_pclk_get;
>> +	}
> 
> It looks like the driver is supposed to handle the
> IS_ERR_OR_NULL(plat->pclk) condition. This check could cause regression
> on existing setup currently failing to initialize the (optional) clock
> and still being functional.

I'm sorry, ENOCOFFEE above, ->pclk is not NULL nor ERR when
clk_prepare_enable() fails. Still I don't stmmac  code depending pclk
being successfully initialized, and the eventual regression looks like a
real possibility.

/P

Re: [PATCH net v2] net: stmmac: add clk_prepare_enable() error handling

[Russell King (Oracle)]

On Tue, Nov 18, 2025 at 03:30:09PM +0100, Paolo Abeni wrote:
> On 11/14/25 3:23 PM, Pavel Zhigulin wrote:
> > The driver previously ignored the return value of 'clk_prepare_enable()'
> > for both the CSR clock and the PCLK in 'stmmac_probe_config_dt()' function.
> > 
> > Add 'clk_prepare_enable()' return value checks.
> > 
> > Found by Linux Verification Center (linuxtesting.org) with SVACE.
> > 
> > Fixes: bfab27a146ed ("stmmac: add the experimental PCI support")
> > Signed-off-by: Pavel Zhigulin <Pavel.Zhigulin@kaspersky.com>
> > ---
> > v2: Fix 'ret' value initialization after build bot notification.
> > v1: https://lore.kernel.org/all/20251113134009.79440-1-Pavel.Zhigulin@kaspersky.com/
> > 
> >  drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c | 11 +++++++++--
> >  1 file changed, 9 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> > index 27bcaae07a7f..8f9eb9683d2b 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
> > @@ -632,7 +632,9 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
> >  			dev_warn(&pdev->dev, "Cannot get CSR clock\n");
> >  			plat->stmmac_clk = NULL;
> >  		}
> > -		clk_prepare_enable(plat->stmmac_clk);
> > +		rc = clk_prepare_enable(plat->stmmac_clk);
> > +		if (rc < 0)
> > +			dev_warn(&pdev->dev, "Cannot enable CSR clock: %d\n", rc);
> >  	}
> > 
> >  	plat->pclk = devm_clk_get_optional(&pdev->dev, "pclk");
> > @@ -640,7 +642,12 @@ stmmac_probe_config_dt(struct platform_device *pdev, u8 *mac)
> >  		ret = plat->pclk;
> >  		goto error_pclk_get;
> >  	}
> > -	clk_prepare_enable(plat->pclk);
> > +	rc = clk_prepare_enable(plat->pclk);
> > +	if (rc < 0) {
> > +		ret = ERR_PTR(rc);
> > +		dev_err(&pdev->dev, "Cannot enable pclk: %d\n", rc);
> > +		goto error_pclk_get;
> > +	}
> 
> It looks like the driver is supposed to handle the
> IS_ERR_OR_NULL(plat->pclk) condition. This check could cause regression
> on existing setup currently failing to initialize the (optional) clock
> and still being functional.
> 
> I *think* we are better off without the added checks.

Note that the clk API permits NULL as valid. CCF checks for this
in clk_prepare() and avoids returning an error:

        if (!clk)
                return 0;

Same check in clk_enable(). So if plat->pclk is NULL, then no error
will be returned.

Places that set plat->pclk:

stmmac_probe_config_dt() - checks for error-pointers and fails. This
will cause driver probe failure.

dwc_qos_probe() - uses stmmac_pltfr_find_clk() which returns the
clk from the bulk-get or NULL. These clocks will have been obtained
using devm_clk_bulk_get_all_enabled(), which I think will return an
error if any fail, which fails the driver probe.

So, I don't think plat->pclk can be an error-pointer here.

Therefore, I don't think there's any concern with error pointers
or NULL in plat->pclk.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!