return test_bit(IMG_REQ_LAYERED, &img_request->flags) != 0;
}
+static enum obj_operation_type
+rbd_img_request_op_type(struct rbd_img_request *img_request)
+{
+ if (img_request_write_test(img_request))
+ return OBJ_OP_WRITE;
+ else if (img_request_discard_test(img_request))
+ return OBJ_OP_DISCARD;
+ else
+ return OBJ_OP_READ;
+}
+
static void
rbd_img_obj_request_read_callback(struct rbd_obj_request *obj_request)
{
* it to our originally-requested length.
*/
obj_request->xferred = obj_request->length;
+ /* discarding a non-existent object is not a problem */
+ if (obj_request->result == -ENOENT)
+ obj_request->result = 0;
obj_request_done_set(obj_request);
}
}
/*
- * Create a copyup osd request based on the information in the
- * object request supplied. A copyup request has three osd ops,
- * a copyup method call, a hint op, and a write op.
+ * Create a copyup osd request based on the information in the object
+ * request supplied. A copyup request has two or three osd ops, a
+ * copyup method call, potentially a hint op, and a write or truncate
+ * or zero op.
*/
static struct ceph_osd_request *
rbd_osd_req_create_copyup(struct rbd_obj_request *obj_request)
struct rbd_device *rbd_dev;
struct ceph_osd_client *osdc;
struct ceph_osd_request *osd_req;
+ int num_osd_ops = 3;
rbd_assert(obj_request_img_data_test(obj_request));
img_request = obj_request->img_request;
rbd_assert(img_request);
- rbd_assert(img_request_write_test(img_request));
+ rbd_assert(img_request_write_test(img_request) ||
+ img_request_discard_test(img_request));
- /* Allocate and initialize the request, for the three ops */
+ if (img_request_discard_test(img_request))
+ num_osd_ops = 2;
+
+ /* Allocate and initialize the request, for all the ops */
snapc = img_request->snapc;
rbd_dev = img_request->rbd_dev;
osdc = &rbd_dev->rbd_client->client->osdc;
- osd_req = ceph_osdc_alloc_request(osdc, snapc, 3, false, GFP_ATOMIC);
+ osd_req = ceph_osdc_alloc_request(osdc, snapc, num_osd_ops,
+ false, GFP_ATOMIC);
if (!osd_req)
return NULL; /* ENOMEM */
rbd_img_request_complete(img_request);
}
+/*
+ * Add individual osd ops to the given ceph_osd_request and prepare
+ * them for submission. num_ops is the current number of
+ * osd operations already to the object request.
+ */
+static void rbd_img_obj_request_fill(struct rbd_obj_request *obj_request,
+ struct ceph_osd_request *osd_request,
+ enum obj_operation_type op_type,
+ unsigned int num_ops)
+{
+ struct rbd_img_request *img_request = obj_request->img_request;
+ struct rbd_device *rbd_dev = img_request->rbd_dev;
+ u64 object_size = rbd_obj_bytes(&rbd_dev->header);
+ u64 offset = obj_request->offset;
+ u64 length = obj_request->length;
+ u64 img_end;
+ u16 opcode;
+
+ if (op_type == OBJ_OP_DISCARD) {
+ if (!offset && length == object_size &&
+ (!img_request_layered_test(img_request) ||
+ !obj_request_overlaps_parent(obj_request))) {
+ opcode = CEPH_OSD_OP_DELETE;
+ } else if ((offset + length == object_size)) {
+ opcode = CEPH_OSD_OP_TRUNCATE;
+ } else {
+ down_read(&rbd_dev->header_rwsem);
+ img_end = rbd_dev->header.image_size;
+ up_read(&rbd_dev->header_rwsem);
+
+ if (obj_request->img_offset + length == img_end)
+ opcode = CEPH_OSD_OP_TRUNCATE;
+ else
+ opcode = CEPH_OSD_OP_ZERO;
+ }
+ } else if (op_type == OBJ_OP_WRITE) {
+ opcode = CEPH_OSD_OP_WRITE;
+ osd_req_op_alloc_hint_init(osd_request, num_ops,
+ object_size, object_size);
+ num_ops++;
+ } else {
+ opcode = CEPH_OSD_OP_READ;
+ }
+
+ osd_req_op_extent_init(osd_request, num_ops, opcode, offset, length,
+ 0, 0);
+ if (obj_request->type == OBJ_REQUEST_BIO)
+ osd_req_op_extent_osd_data_bio(osd_request, num_ops,
+ obj_request->bio_list, length);
+ else if (obj_request->type == OBJ_REQUEST_PAGES)
+ osd_req_op_extent_osd_data_pages(osd_request, num_ops,
+ obj_request->pages, length,
+ offset & ~PAGE_MASK, false, false);
+
+ /* Discards are also writes */
+ if (op_type == OBJ_OP_WRITE || op_type == OBJ_OP_DISCARD)
+ rbd_osd_req_format_write(obj_request);
+ else
+ rbd_osd_req_format_read(obj_request);
+}
+
/*
* Split up an image request into one or more object requests, each
* to a different object. The "type" parameter indicates whether
unsigned int bio_offset = 0;
struct page **pages = NULL;
enum obj_operation_type op_type;
- u64 object_size = rbd_obj_bytes(&rbd_dev->header);
u64 img_offset;
- u64 img_end;
u64 resid;
- u16 opcode;
dout("%s: img %p type %d data_desc %p\n", __func__, img_request,
(int)type, data_desc);
img_offset = img_request->offset;
resid = img_request->length;
rbd_assert(resid > 0);
+ op_type = rbd_img_request_op_type(img_request);
if (type == OBJ_REQUEST_BIO) {
bio_list = data_desc;
const char *object_name;
u64 offset;
u64 length;
- unsigned int which = 0;
object_name = rbd_segment_name(rbd_dev, img_offset);
if (!object_name)
pages += page_count;
}
- if (img_request_discard_test(img_request)) {
- op_type = OBJ_OP_DISCARD;
- if (!offset && (length == object_size)
- && (!img_request_layered_test(img_request) ||
- (rbd_dev->parent_overlap <=
- obj_request->img_offset))) {
- opcode = CEPH_OSD_OP_DELETE;
- } else if ((offset + length == object_size)) {
- opcode = CEPH_OSD_OP_TRUNCATE;
- } else {
- down_read(&rbd_dev->header_rwsem);
- img_end = rbd_dev->header.image_size;
- up_read(&rbd_dev->header_rwsem);
-
- if (obj_request->img_offset + length == img_end)
- opcode = CEPH_OSD_OP_TRUNCATE;
- else
- opcode = CEPH_OSD_OP_ZERO;
- }
- } else 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,
(op_type == OBJ_OP_WRITE) ? 2 : 1,
obj_request);
if (!osd_req)
goto out_unwind;
+
obj_request->osd_req = osd_req;
obj_request->callback = rbd_img_obj_callback;
- rbd_img_request_get(img_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));
- which++;
- }
-
- osd_req_op_extent_init(osd_req, which, opcode, offset, length,
- 0, 0);
- if (type == OBJ_REQUEST_BIO)
- osd_req_op_extent_osd_data_bio(osd_req, which,
- obj_request->bio_list, length);
- else if (type == OBJ_REQUEST_PAGES)
- osd_req_op_extent_osd_data_pages(osd_req, which,
- obj_request->pages, length,
- offset & ~PAGE_MASK, false, false);
+ obj_request->img_offset = img_offset;
- /* Discards are also writes */
- if (op_type == OBJ_OP_WRITE || op_type == OBJ_OP_DISCARD)
- rbd_osd_req_format_write(obj_request);
- else
- rbd_osd_req_format_read(obj_request);
+ rbd_img_obj_request_fill(obj_request, osd_req, op_type, 0);
- obj_request->img_offset = img_offset;
+ rbd_img_request_get(img_request);
img_offset += length;
resid -= length;
struct page **pages;
u32 page_count;
- rbd_assert(obj_request->type == OBJ_REQUEST_BIO);
+ rbd_assert(obj_request->type == OBJ_REQUEST_BIO ||
+ obj_request->type == OBJ_REQUEST_NODATA);
rbd_assert(obj_request_img_data_test(obj_request));
img_request = obj_request->img_request;
rbd_assert(img_request);
struct ceph_osd_client *osdc;
struct rbd_device *rbd_dev;
struct page **pages;
+ enum obj_operation_type op_type;
u32 page_count;
int img_result;
u64 parent_length;
- u64 offset;
- u64 length;
rbd_assert(img_request_child_test(img_request));
osd_req_op_cls_request_data_pages(osd_req, 0, pages, parent_length, 0,
false, false);
- /* Then the hint op */
-
- osd_req_op_alloc_hint_init(osd_req, 1, rbd_obj_bytes(&rbd_dev->header),
- rbd_obj_bytes(&rbd_dev->header));
-
- /* And the original write request op */
-
- offset = orig_request->offset;
- length = orig_request->length;
- osd_req_op_extent_init(osd_req, 2, CEPH_OSD_OP_WRITE,
- offset, length, 0, 0);
- if (orig_request->type == OBJ_REQUEST_BIO)
- osd_req_op_extent_osd_data_bio(osd_req, 2,
- orig_request->bio_list, length);
- else
- osd_req_op_extent_osd_data_pages(osd_req, 2,
- orig_request->pages, length,
- offset & ~PAGE_MASK, false, false);
+ /* Add the other op(s) */
- rbd_osd_req_format_write(orig_request);
+ op_type = rbd_img_request_op_type(orig_request->img_request);
+ rbd_img_obj_request_fill(orig_request, osd_req, op_type, 1);
/* All set, send it off. */
rbd_dev = img_request->rbd_dev;
/* Reads */
- if (!img_request_write_test(img_request))
+ if (!img_request_write_test(img_request) &&
+ !img_request_discard_test(img_request))
return true;
/* Non-layered writes */
queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
q->limits.discard_granularity = segment_size;
q->limits.discard_alignment = segment_size;
+ q->limits.max_discard_sectors = segment_size / SECTOR_SIZE;
+ q->limits.discard_zeroes_data = 1;
blk_queue_merge_bvec(q, rbd_merge_bvec);
disk->queue = q;
set_capacity(rbd_dev->disk, rbd_dev->mapping.size / SECTOR_SIZE);
set_disk_ro(rbd_dev->disk, rbd_dev->mapping.read_only);
- rbd_dev->rq_wq = alloc_workqueue("%s", 0, 0, rbd_dev->disk->disk_name);
+ rbd_dev->rq_wq = alloc_workqueue("%s", WQ_MEM_RECLAIM, 0,
+ rbd_dev->disk->disk_name);
if (!rbd_dev->rq_wq) {
ret = -ENOMEM;
goto err_out_mapping;
projects / firefly-linux-kernel-4.4.55.git / blobdiff