Re: [PATCH 1/3 linux-next] udf: fix ioctl errors

[Jan Kara]

On Tue 24-01-17 21:48:34, Fabian Frederick wrote:
> Currently, lsattr for instance in udf directory gives
> "udf: Invalid argument While reading flags on ..."
> 
> This patch returns -ENOIOCTLCMD
> when command is unknown to have more accurate message like this:
> "Inappropriate ioctl for device While reading flags on ..."
> 
> As suggested by Jan Kara, if arg is NULL with a correct ioctl,
> we return -VM_FAULT_SIGBUS to report error.

You cannot return -VM_FAULT_SIGBUS from ioctl handler! Just look how it is
defined - it is an internal error code to the fault handling and in this
case would just get passed as is to userspace. Which suggests you didn't
test this check... Anyway, I've taken the patch and just removed the change
of the error code. Thanks.

								Honza

> 
> Signed-off-by: Fabian Frederick <fabf@xxxxxxxxx>
> ---
>  fs/udf/file.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/udf/file.c b/fs/udf/file.c
> index dbcb3a4a..d44b3cb 100644
> --- a/fs/udf/file.c
> +++ b/fs/udf/file.c
> @@ -184,9 +184,10 @@ long udf_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>  		goto out;
>  	}
>  
> -	if (!arg) {
> +	if (!arg && ((cmd == UDF_GETVOLIDENT) || (cmd == UDF_GETEASIZE) ||
> +		     (cmd == UDF_RELOCATE_BLOCKS) || (cmd == UDF_GETEABLOCK))) {
>  		udf_debug("invalid argument to udf_ioctl\n");
> -		result = -EINVAL;
> +		result = -VM_FAULT_SIGBUS;
>  		goto out;
>  	}
>  
> @@ -220,6 +221,8 @@ long udf_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>  				      UDF_I(inode)->i_ext.i_data,
>  				      UDF_I(inode)->i_lenEAttr) ? -EFAULT : 0;
>  		goto out;
> +	default:
> +		return -ENOIOCTLCMD;
>  	}
>  
>  out:
> -- 
> 2.9.3
> 
> 
-- 
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR