Re: [Qemu-devel] [RFC PATCH 4/6] scsi-disk: allow customization of the lun

[Christoph Hellwig]

>      case REPORT_LUNS:
> +        assert(!s->lun);

Besides REPORT_LUNS really belonging into the core code as mentioned before
the assert seems dangerous to me.  What protects a guest from issuing a
REPORT LUNS for a non-zero LUN and hitting this assert?  Note that SPC
explicitly allows sending REPORT LUNS to either LUN 0 or the well known
LUN if it exists, even if at least Linux doesn't make use of the latter
yet.