From: Shin'ichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
Write I/Os to conventional zones may have the range that spans across
zone boundaries. Such writes may cause I/O errors when its next zone
is a sequential zone.
To avoid such I/O errors, check for the cross over from a conventional
to a sequential zone. When the write crosses the boundary, shrink the
I/O length to fit within the first zone. If the offset is too close to
the end of the zone, wrap it around to the beginning of the same zone.
Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@xxxxxxx>
Signed-off-by: Dmitry Fomichev <dmitry.fomichev@xxxxxxx>
---
zbd.c | 26 ++++++++++++++++++++++++--
1 file changed, 24 insertions(+), 2 deletions(-)
diff --git a/zbd.c b/zbd.c
index c27efd54..8fb9c8cf 100644
--- a/zbd.c
+++ b/zbd.c
@@ -1563,9 +1563,31 @@ enum io_u_action zbd_adjust_block(struct thread_data *td, struct io_u *io_u)
zb = get_zone(f, zone_idx_b);
orig_zb = zb;
- /* Accept the I/O offset for conventional zones. */
- if (!zb->has_wp)
+ if (!zb->has_wp) {
+ /* Accept non-write I/Os for conventional zones. */
+ if (io_u->ddir != DDIR_WRITE)
+ return io_u_accept;
+ /*
+ * Make sure that writes to conventional zones
+ * don't cross over to any sequential zones.
+ */
+ if (!(zb + 1)->has_wp ||
+ io_u->offset + io_u->buflen <= (zb + 1)->start)
+ return io_u_accept;
+
+ if (io_u->offset + min_bs > (zb + 1)->start) {
+ dprint(FD_IO,
+ "%s: off=%llu + min_bs=%u > next zone %lu\n",
+ f->file_name, io_u->offset, min_bs,
+ (zb + 1)->start);
+ io_u->offset = zb->start + (zb + 1)->start - io_u->offset;
+ new_len = min(io_u->buflen, (zb + 1)->start - io_u->offset);
+ } else {
+ new_len = (zb + 1)->start - io_u->offset;
+ }
+ io_u->buflen = new_len / min_bs * min_bs;
return io_u_accept;
+ }
/*
* Accept the I/O offset for reads if reading beyond the write pointer
--
2.28.0
