Re: [PATCH 19/24] scsi_ioctl: return error code when blk_map_user() fails

[Bart Van Assche]

On 10/21/19 2:53 AM, Hannes Reinecke wrote:
When failing to map the user buffer we should return the actual
error code, avoiding the usage of DRIVER_ERROR.
Signed-off-by: Hannes Reinecke <h...@suse.de>
---
  block/scsi_ioctl.c | 7 ++++---
  1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/block/scsi_ioctl.c b/block/scsi_ioctl.c
index f5e0ad65e86a..1ab1b8d9641c 100644
--- a/block/scsi_ioctl.c
+++ b/block/scsi_ioctl.c
@@ -485,9 +485,10 @@ int sg_scsi_ioctl(struct request_queue *q, struct gendisk 
*disk, fmode_t mode,
                break;
        }
  
-	if (bytes && blk_rq_map_kern(q, rq, buffer, bytes, GFP_NOIO)) {
-               err = DRIVER_ERROR << 24;
-               goto error;
+       if (bytes) {
+               err = blk_rq_map_kern(q, rq, buffer, bytes, GFP_NOIO);
+               if (err)
+                       goto error;
        }
  
  	blk_execute_rq(q, disk, rq, 0);

Since sg_scsi_ioctl() is used to implement SCSI_IOCTL_SEND_COMMAND, does 
this patch change the ABI between user space and kernel in a 
backwards-incompatible way?

Thanks,

Bart.