Re: [PATCH v2 3/3] scsi: aic7xxx: return negative error codes in aic7770_probe()

[Christophe JAILLET]

Le 01/12/2023 à 03:59, Su Hui a écrit :
> aic7770_config() returns both negative and positive error code.
> it's better to make aic7770_probe() only return negative error codes.
>
> And the previous patch made ahc_linux_register_host() return negative error
> codes, which makes sure aic7770_probe() returns negative error codes.
>
> Signed-off-by: Su Hui <suhui@xxxxxxxxxxxx>
> ---
>   drivers/scsi/aic7xxx/aic7770_osm.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/scsi/aic7xxx/aic7770_osm.c b/drivers/scsi/aic7xxx/aic7770_osm.c
> index bdd177e3d762..a19cdd87c453 100644
> --- a/drivers/scsi/aic7xxx/aic7770_osm.c
> +++ b/drivers/scsi/aic7xxx/aic7770_osm.c
> @@ -87,17 +87,17 @@ aic7770_probe(struct device *dev)
>   	sprintf(buf, "ahc_eisa:%d", eisaBase >> 12);
>   	name = kstrdup(buf, GFP_ATOMIC);
>   	if (name == NULL)
> -		return (ENOMEM);
> +		return -ENOMEM;
>   	ahc = ahc_alloc(&aic7xxx_driver_template, name);
>   	if (ahc == NULL)

Unrelated to your fix, but 'name' is leaking here.

Also, kasprintf() could be used to avoid buf+sprintf()+kstrdup()

The GFP_ATOMIC in the allocation could certainly also be just a GFP_KERNEL.

CJ

> -		return (ENOMEM);
> +		return -ENOMEM;
>   	ahc->dev = dev;
>   	error = aic7770_config(ahc, aic7770_ident_table + edev->id.driver_data,
>   			       eisaBase);
>   	if (error != 0) {
>   		ahc->bsh.ioport = 0;
>   		ahc_free(ahc);
> -		return (error);
> +		return error < 0 ? error : -error;
>   	}
>   
>    	dev_set_drvdata(dev, ahc);