Re: dm: dm_blk_ioctl(): implement failover for SG_IO on dm-multipath

[Paolo Bonzini <pbonzini@xxxxxxxxxx>]

On 29/04/21 10:33, Martin Wilck wrote:
> On Wed, 2021-04-28 at 15:54 -0400, Mike Snitzer wrote:
> > @@ -626,32 +626,16 @@ static int dm_sg_io_ioctl(struct block_device
> > *bdev, fmode_t mode,
> >                  }
> >   
> >                  if (rhdr.info & SG_INFO_CHECK) {
> > -                       /*
> > -                        * See if this is a target or path error.
> > -                        * Compare blk_path_error(),
> > scsi_result_to_blk_status(),
> > -                        * blk_errors[].
> > -                        */
> > -                       switch (rhdr.host_status) {
> > -                       case DID_OK:
> > -                               if (scsi_status_is_good(rhdr.status))
> > -                                       rc = 0;
> > -                               break;
> > -                       case DID_TARGET_FAILURE:
> > -                               rc = -EREMOTEIO;
> > -                               goto out;
> > -                       case DID_NEXUS_FAILURE:
> > -                               rc = -EBADE;
> > -                               goto out;
> > -                       case DID_ALLOC_FAILURE:
> > -                               rc = -ENOSPC;
> > -                               goto out;
> > -                       case DID_MEDIUM_ERROR:
> > -                               rc = -ENODATA;
> > -                               goto out;
> > -                       default:
> > -                               /* Everything else is a path error */
> > +                       blk_status_t sts =
> > scsi_result_to_blk_status(rhdr.host_status, NULL);
>
> This change makes dm_mod depend on scsi_mod.
> Would you seriously prefer that over a re-implementation of the logic?

You could just make it an inline function too.

Paolo