Re: [PATCH] Allow ATA_ passthrough command through sg.

[Douglas Gilbert <dgilbert@interlog.com>]

Strange, my utilities and packages like smartmontools
have been sending them through for at least three years.

I presume this patch allows non-root users to send these
commands. Might there be security implications if ATA
WRITE commands are sent through? Non-root users would
still need permissions on the sg device node (e.g.
/dev/sg1).

Doug Gilbert


Gwendal Grignou wrote:
> We can already send ATA specific commands using /dev/sd device files.
> This patch allow to use /dev/sg device files as well.
> 
> Signed-off-by: Gwendal Grignou <gwendal@google.com>
> ---
>  block/scsi_ioctl.c |    6 ++++++
>  1 files changed, 6 insertions(+), 0 deletions(-)
> 
> diff --git a/block/scsi_ioctl.c b/block/scsi_ioctl.c
> index 82a0ca2..93fa53e 100644
> --- a/block/scsi_ioctl.c
> +++ b/block/scsi_ioctl.c
> @@ -186,6 +186,12 @@ void blk_set_cmd_filter_defaults(struct blk_cmd_filter *filter)
>  	__set_bit(GPCMD_LOAD_UNLOAD, filter->write_ok);
>  	__set_bit(GPCMD_SET_STREAMING, filter->write_ok);
>  	__set_bit(GPCMD_SET_READ_AHEAD, filter->write_ok);
> +
> +	/* ATA Passthrough */
> +	__set_bit(ATA_12, filter->read_ok);
> +	__set_bit(ATA_12, filter->write_ok);
> +	__set_bit(ATA_16, filter->read_ok);
> +	__set_bit(ATA_16, filter->write_ok);
>  }
>  EXPORT_SYMBOL_GPL(blk_set_cmd_filter_defaults);
>