Tomasz Zawadzki d4dc859947 lvol: Add size info to get_lvol_stores RPC
Currently there is no way to know total space availible on lvol store or
how much is left after creating lvol bdevs.
Four new fields should were added to get_lvol_stores:
- total number of blocks
- number of free blocks left
- block size (currently always 4096, but should be known to user)
- cluster size

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I93dec2d4b2843f0ee51dc9883c8451cf55353f7b
Reviewed-on: https://review.gerrithub.io/381131
Tested-by: SPDK Automated Test System <sys_sgsw@intel.com>
Reviewed-by: Piotr Pelpliński <piotr.pelplinski@intel.com>
Reviewed-by: Maciej Szwed <maciej.szwed@intel.com>
Reviewed-by: Daniel Verkamp <daniel.verkamp@intel.com>
2017-10-05 16:49:56 -04:00

422 lines
10 KiB
C

/*-
* BSD LICENSE
*
* Copyright (c) Intel Corporation.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "spdk_internal/lvolstore.h"
#include "spdk_internal/log.h"
#include "spdk/string.h"
#include "spdk/io_channel.h"
/* Length of string returned from uuid_unparse() */
#define UUID_STRING_LEN 37
SPDK_LOG_REGISTER_TRACE_FLAG("lvol", SPDK_TRACE_LVOL)
static inline size_t
divide_round_up(size_t num, size_t divisor)
{
return (num + divisor - 1) / divisor;
}
static void
_lvs_init_cb(void *cb_arg, struct spdk_blob_store *bs, int lvserrno)
{
struct spdk_lvs_with_handle_req *lvs_req = cb_arg;
struct spdk_lvol_store *lvs = lvs_req->lvol_store;
if (lvserrno != 0) {
assert(bs == NULL);
SPDK_ERRLOG("Lvol store init failed: could not initialize blobstore\n");
free(lvs);
lvs = NULL;
} else {
assert(bs != NULL);
lvs->blobstore = bs;
TAILQ_INIT(&lvs->lvols);
lvs->total_blocks = (spdk_bs_get_cluster_size(bs) * spdk_bs_free_cluster_count(
bs)) / spdk_bs_get_page_size(bs);
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Lvol store initialized\n");
}
assert(lvs_req->cb_fn != NULL);
lvs_req->cb_fn(lvs_req->cb_arg, lvs, lvserrno);
free(lvs_req);
}
static void
spdk_setup_lvs_opts(struct spdk_bs_opts *bs_opts, struct spdk_lvs_opts *o)
{
spdk_bs_opts_init(bs_opts);
if (o) {
bs_opts->cluster_sz = o->cluster_sz;
} else {
bs_opts->cluster_sz = SPDK_LVS_OPTS_CLUSTER_SZ;
}
}
int
spdk_lvs_init(struct spdk_bs_dev *bs_dev, struct spdk_lvs_opts *o,
spdk_lvs_op_with_handle_complete cb_fn, void *cb_arg)
{
struct spdk_lvol_store *lvs;
struct spdk_lvs_with_handle_req *lvs_req;
struct spdk_bs_opts opts;
if (bs_dev == NULL) {
SPDK_ERRLOG("Blobstore device does not exist\n");
return -ENODEV;
}
spdk_setup_lvs_opts(&opts, o);
lvs = calloc(1, sizeof(*lvs));
if (!lvs) {
SPDK_ERRLOG("Cannot alloc memory for lvol store base pointer\n");
return -ENOMEM;
}
uuid_generate_time(lvs->uuid);
lvs_req = calloc(1, sizeof(*lvs_req));
if (!lvs_req) {
free(lvs);
SPDK_ERRLOG("Cannot alloc memory for lvol store request pointer\n");
return -ENOMEM;
}
lvs_req->cb_fn = cb_fn;
lvs_req->cb_arg = cb_arg;
lvs_req->lvol_store = lvs;
lvs->bs_dev = bs_dev;
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Initializing lvol store\n");
spdk_bs_init(bs_dev, &opts, _lvs_init_cb, lvs_req);
return 0;
}
static void
_lvs_unload_cb(void *cb_arg, int lvserrno)
{
struct spdk_lvs_req *lvs_req = cb_arg;
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Lvol store unloaded\n");
assert(lvs_req->cb_fn != NULL);
lvs_req->cb_fn(lvs_req->cb_arg, lvserrno);
free(lvs_req);
}
int
spdk_lvs_unload(struct spdk_lvol_store *lvs, spdk_lvs_op_complete cb_fn,
void *cb_arg)
{
struct spdk_lvs_req *lvs_req;
if (lvs == NULL) {
SPDK_ERRLOG("Lvol store is NULL\n");
return -ENODEV;
}
lvs_req = calloc(1, sizeof(*lvs_req));
if (!lvs_req) {
SPDK_ERRLOG("Cannot alloc memory for lvol store request pointer\n");
return -ENOMEM;
}
lvs_req->cb_fn = cb_fn;
lvs_req->cb_arg = cb_arg;
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Unloading lvol store\n");
spdk_bs_unload(lvs->blobstore, _lvs_unload_cb, lvs_req);
free(lvs);
return 0;
}
static void
_spdk_lvol_return_to_caller(void *cb_arg, int lvolerrno)
{
struct spdk_lvol_with_handle_req *req = cb_arg;
assert(req->cb_fn != NULL);
req->cb_fn(req->cb_arg, req->lvol, lvolerrno);
free(req);
}
static void
_spdk_lvs_destruct_cb(void *cb_arg, int lvserrno)
{
struct spdk_lvs_req *req = cb_arg;
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Lvol store bdev deleted\n");
if (req->cb_fn != NULL)
req->cb_fn(req->cb_arg, lvserrno);
free(req);
}
static void
_spdk_lvol_close_blob_cb(void *cb_arg, int lvolerrno)
{
struct spdk_lvol *lvol = cb_arg;
if (lvolerrno < 0) {
SPDK_ERRLOG("Could not close blob on lvol\n");
free(lvol->name);
free(lvol);
return;
}
TAILQ_REMOVE(&lvol->lvol_store->lvols, lvol, link);
if (lvol->lvol_store->destruct_req && TAILQ_EMPTY(&lvol->lvol_store->lvols)) {
spdk_lvs_unload(lvol->lvol_store, _spdk_lvs_destruct_cb, lvol->lvol_store->destruct_req);
}
free(lvol->name);
free(lvol);
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Blob closed on lvol\n");
}
static void
_spdk_lvol_delete_blob_cb(void *cb_arg, int lvolerrno)
{
struct spdk_lvol *lvol = cb_arg;
struct spdk_blob_store *bs = lvol->lvol_store->blobstore;
if (lvolerrno < 0) {
SPDK_ERRLOG("Could not delete blob on lvol\n");
free(lvol->name);
free(lvol);
return;
}
SPDK_INFOLOG(SPDK_TRACE_LVOL, "Blob closed on lvol\n");
spdk_bs_md_delete_blob(bs, lvol->blob_id, _spdk_lvol_close_blob_cb, lvol);
}
static void
_spdk_lvol_create_open_cb(void *cb_arg, struct spdk_blob *blob, int lvolerrno)
{
struct spdk_lvol_with_handle_req *req = cb_arg;
spdk_blob_id blob_id = spdk_blob_get_id(blob);
struct spdk_lvol *lvol = req->lvol;
char uuid[UUID_STRING_LEN];
if (lvolerrno < 0) {
free(lvol);
goto invalid;
}
lvol->blob = blob;
lvol->blob_id = blob_id;
uuid_unparse(lvol->lvol_store->uuid, uuid);
lvol->name = spdk_sprintf_alloc("%s_%"PRIu64, uuid, (uint64_t)blob_id);
if (!lvol->name) {
spdk_bs_md_close_blob(&blob, _spdk_lvol_delete_blob_cb, lvol);
goto invalid;
}
lvolerrno = spdk_bs_md_resize_blob(blob, lvol->num_clusters);
if (lvolerrno < 0) {
spdk_bs_md_close_blob(&blob, _spdk_lvol_delete_blob_cb, lvol);
goto invalid;
}
TAILQ_INSERT_TAIL(&lvol->lvol_store->lvols, lvol, link);
spdk_bs_md_sync_blob(blob, _spdk_lvol_return_to_caller, req);
return;
invalid:
assert(req->cb_fn != NULL);
req->cb_fn(req->cb_arg, NULL, lvolerrno);
free(req);
return;
}
static void
_spdk_lvol_create_cb(void *cb_arg, spdk_blob_id blobid, int lvolerrno)
{
struct spdk_lvol_with_handle_req *req = cb_arg;
struct spdk_blob_store *bs;
if (lvolerrno < 0) {
free(req->lvol);
assert(req->cb_fn != NULL);
req->cb_fn(req->cb_arg, NULL, lvolerrno);
free(req);
return;
}
bs = req->lvol->lvol_store->blobstore;
spdk_bs_md_open_blob(bs, blobid, _spdk_lvol_create_open_cb, req);
}
int
spdk_lvol_create(struct spdk_lvol_store *lvs, uint64_t sz,
spdk_lvol_op_with_handle_complete cb_fn, void *cb_arg)
{
struct spdk_lvol_with_handle_req *req;
struct spdk_blob_store *bs;
struct spdk_lvol *lvol;
uint64_t num_clusters, free_clusters;
if (lvs == NULL) {
SPDK_ERRLOG("lvol store does not exist\n");
return -ENODEV;
}
bs = lvs->blobstore;
num_clusters = divide_round_up(sz, spdk_bs_get_cluster_size(bs));
free_clusters = spdk_bs_free_cluster_count(bs);
if (num_clusters > free_clusters) {
SPDK_ERRLOG("Not enough free clusters left (%zu) on lvol store to add lvol %zu clusters\n",
free_clusters, num_clusters);
return -ENOMEM;
}
req = calloc(1, sizeof(*req));
if (!req) {
SPDK_ERRLOG("Cannot alloc memory for lvol request pointer\n");
return -ENOMEM;
}
req->cb_fn = cb_fn;
req->cb_arg = cb_arg;
lvol = calloc(1, sizeof(*lvol));
if (!lvol) {
free(req);
SPDK_ERRLOG("Cannot alloc memory for lvol base pointer\n");
return -ENOMEM;
}
lvol->lvol_store = lvs;
lvol->num_clusters = num_clusters;
lvol->close_only = false;
req->lvol = lvol;
spdk_bs_md_create_blob(lvs->blobstore, _spdk_lvol_create_cb, req);
return 0;
}
static void
_spdk_lvol_resize_cb(void *cb_arg, int lvolerrno)
{
struct spdk_lvol_req *req = cb_arg;
req->cb_fn(req->cb_arg, lvolerrno);
free(req);
}
int
spdk_lvol_resize(struct spdk_lvol *lvol, uint64_t sz,
spdk_lvol_op_complete cb_fn, void *cb_arg)
{
int rc;
struct spdk_blob *blob = lvol->blob;
struct spdk_lvol_store *lvs = lvol->lvol_store;
struct spdk_lvol_req *req;
uint64_t free_clusters = spdk_bs_free_cluster_count(lvs->blobstore);
uint64_t used_clusters = lvol->num_clusters;
uint64_t new_clusters = divide_round_up(sz, spdk_bs_get_cluster_size(lvs->blobstore));
/* Check if size of lvol increasing */
if (new_clusters > used_clusters) {
/* Check if there is enough clusters left to resize */
if (new_clusters - used_clusters > free_clusters) {
SPDK_ERRLOG("Not enough free clusters left on lvol store to resize lvol to %zu clusters\n", sz);
return -ENOMEM;
}
}
req = calloc(1, sizeof(*req));
if (!req) {
SPDK_ERRLOG("Cannot alloc memory for lvol request pointer\n");
return -ENOMEM;
}
req->cb_fn = cb_fn;
req->cb_arg = cb_arg;
rc = spdk_bs_md_resize_blob(blob, sz);
if (rc < 0) {
goto invalid;
}
lvol->num_clusters = new_clusters;
spdk_blob_md_set_xattr(blob, "length", &sz, sizeof(sz));
spdk_bs_md_sync_blob(blob, _spdk_lvol_resize_cb, req);
return rc;
invalid:
req->cb_fn(req->cb_arg, rc);
free(req);
return rc;
}
void
spdk_lvol_destroy(struct spdk_lvol *lvol)
{
if (lvol == NULL) {
SPDK_ERRLOG("lvol does not exist\n");
return;
}
TAILQ_REMOVE(&lvol->lvol_store->lvols, lvol, link);
spdk_bs_md_close_blob(&(lvol->blob), _spdk_lvol_delete_blob_cb, lvol);
}
void
spdk_lvol_close(struct spdk_lvol *lvol)
{
if (lvol == NULL) {
SPDK_ERRLOG("lvol does not exist\n");
return;
}
spdk_bs_md_close_blob(&(lvol->blob), _spdk_lvol_close_blob_cb, lvol);
}
struct spdk_io_channel *
spdk_lvol_get_io_channel(struct spdk_lvol *lvol)
{
return spdk_bs_alloc_io_channel(lvol->lvol_store->blobstore);
}