rbd: be picky about osd request status type
authorAlex Elder <elder@dreamhost.com>
Thu, 8 Nov 2012 14:01:39 +0000 (08:01 -0600)
committerAlex Elder <elder@inktank.com>
Thu, 17 Jan 2013 20:53:20 +0000 (14:53 -0600)
The result field in a ceph osd reply header is a signed 32-bit type,
but rbd code often casually uses int to represent it.

The following changes the types of variables that handle this result
value to be "s32" instead of "int" to be completely explicit about
it.  Only at the point we pass that result to __blk_end_request()
does the type get converted to the plain old int defined for that
interface.

There is almost certainly no binary impact of this change, but I
prefer to show the exact size and signedness of the value since we
know it.

Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Dan Mick <dan.mick@inktank.com>
drivers/block/rbd.c

index 85131de..8d93b6a 100644 (file)
@@ -171,7 +171,7 @@ struct rbd_client {
  */
 struct rbd_req_status {
        int done;
-       int rc;
+       s32 rc;
        u64 bytes;
 };
 
@@ -1053,13 +1053,13 @@ static void rbd_destroy_ops(struct ceph_osd_req_op *ops)
 static void rbd_coll_end_req_index(struct request *rq,
                                   struct rbd_req_coll *coll,
                                   int index,
-                                  int ret, u64 len)
+                                  s32 ret, u64 len)
 {
        struct request_queue *q;
        int min, max, i;
 
        dout("rbd_coll_end_req_index %p index %d ret %d len %llu\n",
-            coll, index, ret, (unsigned long long) len);
+            coll, index, (int)ret, (unsigned long long)len);
 
        if (!rq)
                return;
@@ -1080,7 +1080,7 @@ static void rbd_coll_end_req_index(struct request *rq,
                max++;
 
        for (i = min; i<max; i++) {
-               __blk_end_request(rq, coll->status[i].rc,
+               __blk_end_request(rq, (int)coll->status[i].rc,
                                  coll->status[i].bytes);
                coll->num_done++;
                kref_put(&coll->kref, rbd_coll_release);
@@ -1089,7 +1089,7 @@ static void rbd_coll_end_req_index(struct request *rq,
 }
 
 static void rbd_coll_end_req(struct rbd_request *rbd_req,
-                            int ret, u64 len)
+                            s32 ret, u64 len)
 {
        rbd_coll_end_req_index(rbd_req->rq,
                                rbd_req->coll, rbd_req->coll_index,
@@ -1129,7 +1129,7 @@ static int rbd_do_request(struct request *rq,
        if (!rbd_req) {
                if (coll)
                        rbd_coll_end_req_index(rq, coll, coll_index,
-                                              -ENOMEM, len);
+                                              (s32)-ENOMEM, len);
                return -ENOMEM;
        }
 
@@ -1206,7 +1206,7 @@ done_err:
        bio_chain_put(rbd_req->bio);
        ceph_osdc_put_request(osd_req);
 done_pages:
-       rbd_coll_end_req(rbd_req, ret, len);
+       rbd_coll_end_req(rbd_req, (s32)ret, len);
        kfree(rbd_req);
        return ret;
 }
@@ -1219,7 +1219,7 @@ static void rbd_req_cb(struct ceph_osd_request *osd_req, struct ceph_msg *msg)
        struct rbd_request *rbd_req = osd_req->r_priv;
        struct ceph_osd_reply_head *replyhead;
        struct ceph_osd_op *op;
-       __s32 rc;
+       s32 rc;
        u64 bytes;
        int read_op;
 
@@ -1227,14 +1227,14 @@ static void rbd_req_cb(struct ceph_osd_request *osd_req, struct ceph_msg *msg)
        replyhead = msg->front.iov_base;
        WARN_ON(le32_to_cpu(replyhead->num_ops) == 0);
        op = (void *)(replyhead + 1);
-       rc = le32_to_cpu(replyhead->result);
+       rc = (s32)le32_to_cpu(replyhead->result);
        bytes = le64_to_cpu(op->extent.length);
        read_op = (le16_to_cpu(op->op) == CEPH_OSD_OP_READ);
 
        dout("rbd_req_cb bytes=%llu readop=%d rc=%d\n",
                (unsigned long long) bytes, read_op, (int) rc);
 
-       if (rc == -ENOENT && read_op) {
+       if (rc == (s32)-ENOENT && read_op) {
                zero_bio_chain(rbd_req->bio, 0);
                rc = 0;
        } else if (rc == 0 && read_op && bytes < rbd_req->len) {
@@ -1679,7 +1679,8 @@ static void rbd_rq_fn(struct request_queue *q)
                                                bio_chain, coll, cur_seg);
                        else
                                rbd_coll_end_req_index(rq, coll, cur_seg,
-                                                      -ENOMEM, chain_size);
+                                                      (s32)-ENOMEM,
+                                                      chain_size);
                        size -= chain_size;
                        ofs += chain_size;