CA-122162: immediately fail request when reaching 2TiB limit

When hitting the 2TiB limit, fail the request with ENOSPC instead of
EIO, because tapdisk retries requests that have failed with EIO hoping
that the request will utterly succeeded, which can never happen in our
case.

Signed-off-by: Thanos Makatos <[email protected]>
Reviewed-by: Keith Petley <[email protected]>

GitHub: closes #68 on xapi-project/blktap

drivers/block-vhd.c

@@ -1325,7 +1325,7 @@ reserve_new_block(struct vhd_state *s, uint32_t blk)
 		gap = (s->spp - ((s->next_db + s->bm_secs) % s->spp));
 
 	if (s->next_db + gap > UINT_MAX)
-		return (uint64_t)EIO << 32;
+		return (uint64_t)ENOSPC << 32;
 
 	s->bat.pbw_blk    = blk;
 	s->bat.pbw_offset = s->next_db + gap;
@@ -1511,7 +1511,7 @@ allocate_block(struct vhd_state *s, uint32_t blk)
 	}
 
 	if (next_db > UINT_MAX)
-		return -EIO;
+		return -ENOSPC;
 
 	s->next_db = next_db;