2023-10-09 12:50:47

by Ma Ke

Subject: [PATCH v2] ata: sata_mv: aspeed: fix value check in mv_platform_probe()

In mv_platform_probe(), check the return value of clk_prepare_enable()
and return the error code if clk_prepare_enable() returns an
unexpected value.

Signed-off-by: Ma Ke <[email protected]>
---
drivers/ata/sata_mv.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c
index 45e48d653c60..df3a02e7a50b 100644
--- a/drivers/ata/sata_mv.c
+++ b/drivers/ata/sata_mv.c
@@ -4125,8 +4125,10 @@ static int mv_platform_probe(struct platform_device *pdev)
hpriv->clk = clk_get(&pdev->dev, NULL);
if (IS_ERR(hpriv->clk))
dev_notice(&pdev->dev, "cannot get optional clkdev\n");
- else
- clk_prepare_enable(hpriv->clk);
+ else {
+ rc = clk_prepare_enable(hpriv->clk);
+ goto err;
+ }

for (port = 0; port < n_ports; port++) {
char port_number[16];
--
2.37.2

2023-10-09 14:23:10

by Niklas Cassel

Subject: Re: [PATCH v2] ata: sata_mv: aspeed: fix value check in mv_platform_probe()

On Mon, Oct 09, 2023 at 08:49:55PM +0800, Ma Ke wrote:
> In mv_platform_probe(), check the return value of clk_prepare_enable()
> and return the error code if clk_prepare_enable() returns an
> unexpected value.
>
> Signed-off-by: Ma Ke <[email protected]>
> ---
> drivers/ata/sata_mv.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c
> index 45e48d653c60..df3a02e7a50b 100644
> --- a/drivers/ata/sata_mv.c
> +++ b/drivers/ata/sata_mv.c
> @@ -4125,8 +4125,10 @@ static int mv_platform_probe(struct platform_device *pdev)
> hpriv->clk = clk_get(&pdev->dev, NULL);
> if (IS_ERR(hpriv->clk))
> dev_notice(&pdev->dev, "cannot get optional clkdev\n");
> - else
> - clk_prepare_enable(hpriv->clk);
> + else {
> + rc = clk_prepare_enable(hpriv->clk);
> + goto err;
> + }
>
> for (port = 0; port < n_ports; port++) {
> char port_number[16];
> --
> 2.37.2
>

Reviewed-by: Niklas Cassel <[email protected]>

2023-10-09 22:34:06

by Damien Le Moal

Subject: Re: [PATCH v2] ata: sata_mv: aspeed: fix value check in mv_platform_probe()

On 10/9/23 21:49, Ma Ke wrote:
> In mv_platform_probe(), check the return value of clk_prepare_enable()
> and return the error code if clk_prepare_enable() returns an
> unexpected value.
>
> Signed-off-by: Ma Ke <[email protected]>
> ---
> drivers/ata/sata_mv.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c
> index 45e48d653c60..df3a02e7a50b 100644
> --- a/drivers/ata/sata_mv.c
> +++ b/drivers/ata/sata_mv.c
> @@ -4125,8 +4125,10 @@ static int mv_platform_probe(struct platform_device *pdev)
> hpriv->clk = clk_get(&pdev->dev, NULL);
> if (IS_ERR(hpriv->clk))
> dev_notice(&pdev->dev, "cannot get optional clkdev\n");
> - else
> - clk_prepare_enable(hpriv->clk);
> + else {
> + rc = clk_prepare_enable(hpriv->clk);
> + goto err;

This is wrong. The goto err should only be done if there is an error.
Did you even test your own patch ???

> + }
>
> for (port = 0; port < n_ports; port++) {
> char port_number[16];

--
Damien Le Moal
Western Digital Research

2023-10-10 09:07:36

by Niklas Cassel

Subject: Re: [PATCH v2] ata: sata_mv: aspeed: fix value check in mv_platform_probe()

On Tue, Oct 10, 2023 at 07:33:49AM +0900, Damien Le Moal wrote:
> On 10/9/23 21:49, Ma Ke wrote:
> > In mv_platform_probe(), check the return value of clk_prepare_enable()
> > and return the error code if clk_prepare_enable() returns an
> > unexpected value.
> >
> > Signed-off-by: Ma Ke <[email protected]>
> > ---
> > drivers/ata/sata_mv.c | 6 ++++--
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/ata/sata_mv.c b/drivers/ata/sata_mv.c
> > index 45e48d653c60..df3a02e7a50b 100644
> > --- a/drivers/ata/sata_mv.c
> > +++ b/drivers/ata/sata_mv.c
> > @@ -4125,8 +4125,10 @@ static int mv_platform_probe(struct platform_device *pdev)
> > hpriv->clk = clk_get(&pdev->dev, NULL);
> > if (IS_ERR(hpriv->clk))
> > dev_notice(&pdev->dev, "cannot get optional clkdev\n");
> > - else
> > - clk_prepare_enable(hpriv->clk);
> > + else {
> > + rc = clk_prepare_enable(hpriv->clk);
> > + goto err;
>
> This is wrong. The goto err should only be done if there is an error.
> Did you even test your own patch ???

The "if (rc)" was there in V1....

I should have caught this in my review of V2.

BTW, if we want to be compliant with the coding kernel standard,
you should also have braces on the if (IS_ERR()),
since there are braces on the else-clause.


Kind regards,
Niklas