From: Guangliang Zhao <lucienc...@gmail.com>
It could only handle the read and write operations now,
extend it for the coming discard support.
Signed-off-by: Guangliang Zhao <lucienc...@gmail.com>
Reviewed-by: Josh Durgin <josh.dur...@inktank.com>
Reviewed-by: Alex Elder <el...@linaro.org>
---
drivers/block/rbd.c | 94 ++++++++++++++++++++++++++++++++-------------------
1 file changed, 60 insertions(+), 34 deletions(-)
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 282bfcd..c90a60a 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -209,6 +209,11 @@ enum obj_request_type {
OBJ_REQUEST_NODATA, OBJ_REQUEST_BIO, OBJ_REQUEST_PAGES
};
+enum obj_operation_type {
+ OBJ_OP_WRITE,
+ OBJ_OP_READ,
+};
+
enum obj_req_flags {
OBJ_REQ_DONE, /* completion flag: not done = 0, done = 1 */
OBJ_REQ_IMG_DATA, /* object usage: standalone = 0, image = 1 */
@@ -715,6 +720,18 @@ static int parse_rbd_opts_token(char *c, void *private)
return 0;
}
+static char* obj_op_name(enum obj_operation_type op_type)
+{
+ switch (op_type) {
+ case OBJ_OP_READ:
+ return "read";
+ case OBJ_OP_WRITE:
+ return "write";
+ default:
+ return "invalid op code";
+ }
+}
+
/*
* Get a ceph client with specific addr and configuration, if one does
* not exist create it. Either way, ceph_opts is consumed by this
@@ -1727,7 +1744,7 @@ static void rbd_osd_req_format_write(struct
rbd_obj_request *obj_request)
*/
static struct ceph_osd_request *rbd_osd_req_create(
struct rbd_device *rbd_dev,
- bool write_request,
+ enum obj_operation_type op_type,
unsigned int num_ops,
struct rbd_obj_request *obj_request)
{
@@ -1735,16 +1752,14 @@ static struct ceph_osd_request *rbd_osd_req_create(
struct ceph_osd_client *osdc;
struct ceph_osd_request *osd_req;
- if (obj_request_img_data_test(obj_request)) {
+ if (obj_request_img_data_test(obj_request) && op_type == OBJ_OP_WRITE) {
struct rbd_img_request *img_request = obj_request->img_request;
- rbd_assert(write_request ==
- img_request_write_test(img_request));
- if (write_request)
- snapc = img_request->snapc;
+ rbd_assert(img_request_write_test(img_request));
+ snapc = img_request->snapc;
}
- rbd_assert(num_ops == 1 || (write_request && num_ops == 2));
+ rbd_assert(num_ops == 1 || ((op_type == OBJ_OP_WRITE) && num_ops == 2));
/* Allocate and initialize the request, for the num_ops ops */
@@ -1754,7 +1769,7 @@ static struct ceph_osd_request *rbd_osd_req_create(
if (!osd_req)
return NULL; /* ENOMEM */
- if (write_request)
+ if (op_type == OBJ_OP_WRITE)
osd_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK;
else
osd_req->r_flags = CEPH_OSD_FLAG_READ;
@@ -1961,7 +1976,7 @@ static bool rbd_dev_parent_get(struct rbd_device *rbd_dev)
static struct rbd_img_request *rbd_img_request_create(
struct rbd_device *rbd_dev,
u64 offset, u64 length,
- bool write_request,
+ enum obj_operation_type op_type,
struct ceph_snap_context *snapc)
{
struct rbd_img_request *img_request;
@@ -1975,7 +1990,7 @@ static struct rbd_img_request *rbd_img_request_create(
img_request->offset = offset;
img_request->length = length;
img_request->flags = 0;
- if (write_request) {
+ if (op_type == OBJ_OP_WRITE) {
img_request_write_set(img_request);
img_request->snapc = snapc;
} else {
@@ -1992,8 +2007,7 @@ static struct rbd_img_request *rbd_img_request_create(
kref_init(&img_request->kref);
dout("%s: rbd_dev %p %s %llu/%llu -> img %p\n", __func__, rbd_dev,
- write_request ? "write" : "read", offset, length,
- img_request);
+ obj_op_name(op_type), offset, length, img_request);
return img_request;
}
@@ -2034,7 +2048,7 @@ static struct rbd_img_request *rbd_parent_request_create(
rbd_dev = obj_request->img_request->rbd_dev;
parent_request = rbd_img_request_create(rbd_dev->parent, img_offset,
- length, false, NULL);
+ length, OBJ_OP_READ, NULL);
if (!parent_request)
return NULL;
@@ -2075,11 +2089,14 @@ static bool rbd_img_obj_end_request(struct
rbd_obj_request *obj_request)
result = obj_request->result;
if (result) {
struct rbd_device *rbd_dev = img_request->rbd_dev;
+ enum obj_operation_type op_type;
+
+ op_type = img_request_write_test(img_request) ? OBJ_OP_WRITE :
+ OBJ_OP_READ;
rbd_warn(rbd_dev, "%s %llx at %llx (%llx)\n",
- img_request_write_test(img_request) ? "write" : "read",
- obj_request->length, obj_request->img_offset,
- obj_request->offset);
+ obj_op_name(op_type), obj_request->length,
+ obj_request->img_offset, obj_request->offset);
rbd_warn(rbd_dev, " result %d xferred %x\n",
result, xferred);
if (!img_request->result)
@@ -2157,10 +2174,10 @@ static int rbd_img_request_fill(struct rbd_img_request
*img_request,
struct rbd_device *rbd_dev = img_request->rbd_dev;
struct rbd_obj_request *obj_request = NULL;
struct rbd_obj_request *next_obj_request;
- bool write_request = img_request_write_test(img_request);
struct bio *bio_list = NULL;
unsigned int bio_offset = 0;
struct page **pages = NULL;
+ enum obj_operation_type op_type;
u64 img_offset;
u64 resid;
u16 opcode;
@@ -2168,7 +2185,6 @@ static int rbd_img_request_fill(struct rbd_img_request
*img_request,
dout("%s: img %p type %d data_desc %p\n", __func__, img_request,
(int)type, data_desc);
- opcode = write_request ? CEPH_OSD_OP_WRITE : CEPH_OSD_OP_READ;
img_offset = img_request->offset;
resid = img_request->length;
rbd_assert(resid > 0);
@@ -2230,15 +2246,21 @@ static int rbd_img_request_fill(struct rbd_img_request
*img_request,
pages += page_count;
}
- osd_req = rbd_osd_req_create(rbd_dev, write_request,
- (write_request ? 2 : 1),
- obj_request);
+ if (img_request_write_test(img_request)) {
+ op_type = OBJ_OP_WRITE;
+ opcode = CEPH_OSD_OP_WRITE;
+ } else {
+ op_type = OBJ_OP_READ;
+ opcode = CEPH_OSD_OP_READ;
+ }
+
+ osd_req = rbd_osd_req_create(rbd_dev, op_type, obj_request);
if (!osd_req)
goto out_unwind;
obj_request->osd_req = osd_req;
obj_request->callback = rbd_img_obj_callback;
- if (write_request) {
+ if (op_type == OBJ_OP_WRITE) {
osd_req_op_alloc_hint_init(osd_req, which,
rbd_obj_bytes(&rbd_dev->header),
rbd_obj_bytes(&rbd_dev->header));
@@ -2255,7 +2277,7 @@ static int rbd_img_request_fill(struct rbd_img_request
*img_request,
obj_request->pages, length,
offset & ~PAGE_MASK, false, false);
- if (write_request)
+ if (op_type == OBJ_OP_WRITE)
rbd_osd_req_format_write(obj_request);
else
rbd_osd_req_format_read(obj_request);
@@ -2625,7 +2647,7 @@ static int rbd_img_obj_exists_submit(struct
rbd_obj_request *obj_request)
rbd_assert(obj_request->img_request);
rbd_dev = obj_request->img_request->rbd_dev;
- stat_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+ stat_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_READ, 1,
stat_request);
if (!stat_request->osd_req)
goto out;
@@ -2848,7 +2870,7 @@ static int rbd_obj_notify_ack_sync(struct rbd_device
*rbd_dev, u64 notify_id)
return -ENOMEM;
ret = -ENOMEM;
- obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+ obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_READ, 1,
obj_request);
if (!obj_request->osd_req)
goto out;
@@ -2912,7 +2934,7 @@ static int __rbd_dev_header_watch_sync(struct rbd_device
*rbd_dev, bool start)
if (!obj_request)
goto out_cancel;
- obj_request->osd_req = rbd_osd_req_create(rbd_dev, true, 1,
+ obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_WRITE, 1,
obj_request);
if (!obj_request->osd_req)
goto out_cancel;
@@ -3021,7 +3043,7 @@ static int rbd_obj_method_sync(struct rbd_device *rbd_dev,
obj_request->pages = pages;
obj_request->page_count = page_count;
- obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+ obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_READ, 1,
obj_request);
if (!obj_request->osd_req)
goto out;
@@ -3077,7 +3099,7 @@ static void rbd_request_fn(struct request_queue *q)
int result;
while ((rq = blk_fetch_request(q))) {
- bool write_request = rq_data_dir(rq) == WRITE;
+ enum obj_operation_type op_type;
struct rbd_img_request *img_request;
struct ceph_snap_context *snapc = NULL;
u64 offset;
@@ -3106,9 +3128,14 @@ static void rbd_request_fn(struct request_queue *q)
spin_unlock_irq(q->queue_lock);
- /* Disallow writes to a read-only device */
+ if (rq->cmd_flags & REQ_WRITE)
+ op_type = OBJ_OP_WRITE;
+ else
+ op_type = OBJ_OP_READ;
+
+ /* Only allow reads to a read-only device */
- if (write_request) {
+ if (op_type != OBJ_OP_READ) {
result = -EROFS;
if (read_only)
goto end_request;
@@ -3153,7 +3180,7 @@ static void rbd_request_fn(struct request_queue *q)
result = -ENOMEM;
img_request = rbd_img_request_create(rbd_dev, offset, length,
- write_request, snapc);
+ op_type, snapc);
if (!img_request)
goto end_request;
@@ -3169,8 +3196,7 @@ end_request:
spin_lock_irq(q->queue_lock);
if (result < 0) {
rbd_warn(rbd_dev, "%s %llx at %llx result %d\n",
- write_request ? "write" : "read",
- length, offset, result);
+ obj_op_name(op_type), length, offset, result);
if (snapc)
ceph_put_snap_context(snapc);
@@ -3267,7 +3293,7 @@ static int rbd_obj_read_sync(struct rbd_device *rbd_dev,
obj_request->pages = pages;
obj_request->page_count = page_count;
- obj_request->osd_req = rbd_osd_req_create(rbd_dev, false, 1,
+ obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_READ, 1,
obj_request);
if (!obj_request->osd_req)
goto out;
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
