From: "Benoît Canet" <benoit@irqsave.net>
To: qemu-devel@nongnu.org
Cc: kwolf@redhat.com, "Benoît Canet" <benoit@irqsave.net>,
stefanha@redhat.com
Subject: [Qemu-devel] [RFC V7 19/32] block: Add qcow2_dedup format and image creation code.
Date: Fri, 15 Mar 2013 15:49:33 +0100 [thread overview]
Message-ID: <1363358986-8360-20-git-send-email-benoit@irqsave.net> (raw)
In-Reply-To: <1363358986-8360-1-git-send-email-benoit@irqsave.net>
Also modify qemu-io-test.
Signed-off-by: Benoit Canet <benoit@irqsave.net>
---
block/qcow2.c | 185 +++++++++++++++++++++++++++++++++++++++---
include/block/block_int.h | 1 +
tests/qemu-iotests/common.rc | 3 +-
3 files changed, 175 insertions(+), 14 deletions(-)
diff --git a/block/qcow2.c b/block/qcow2.c
index 1210780..9032dfc 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1263,7 +1263,8 @@ static int preallocate(BlockDriverState *bs)
static int qcow2_create2(const char *filename, int64_t total_size,
const char *backing_file, const char *backing_format,
int flags, size_t cluster_size, int prealloc,
- QEMUOptionParameter *options, int version)
+ QEMUOptionParameter *options, int version,
+ bool dedup, uint8_t hash_algo)
{
/* Calculate cluster_bits */
int cluster_bits;
@@ -1291,7 +1292,8 @@ static int qcow2_create2(const char *filename, int64_t total_size,
*/
BlockDriverState* bs;
QCowHeader header;
- uint8_t* refcount_table;
+ uint8_t *tables;
+ int size;
int ret;
ret = bdrv_create_file(filename, options);
@@ -1333,10 +1335,11 @@ static int qcow2_create2(const char *filename, int64_t total_size,
goto out;
}
- /* Write an empty refcount table */
- refcount_table = g_malloc0(cluster_size);
- ret = bdrv_pwrite(bs, cluster_size, refcount_table, cluster_size);
- g_free(refcount_table);
+ /* Write an empty refcount table + extra space for dedup table if needed */
+ size = dedup ? 2 : 1;
+ tables = g_malloc0(size * cluster_size);
+ ret = bdrv_pwrite(bs, cluster_size, tables, size * cluster_size);
+ g_free(tables);
if (ret < 0) {
goto out;
@@ -1347,7 +1350,7 @@ static int qcow2_create2(const char *filename, int64_t total_size,
/*
* And now open the image and make it consistent first (i.e. increase the
* refcount of the cluster that is occupied by the header and the refcount
- * table)
+ * table and the eventual dedup table)
*/
BlockDriver* drv = bdrv_find_format("qcow2");
assert(drv != NULL);
@@ -1357,7 +1360,8 @@ static int qcow2_create2(const char *filename, int64_t total_size,
goto out;
}
- ret = qcow2_alloc_clusters(bs, 2 * cluster_size);
+ size++; /* Add a cluster for the header */
+ ret = qcow2_alloc_clusters(bs, size * cluster_size);
if (ret < 0) {
goto out;
@@ -1367,11 +1371,31 @@ static int qcow2_create2(const char *filename, int64_t total_size,
}
/* Okay, now that we have a valid image, let's give it the right size */
+ BDRVQcowState *s = bs->opaque;
ret = bdrv_truncate(bs, total_size * BDRV_SECTOR_SIZE);
if (ret < 0) {
goto out;
}
+ if (dedup) {
+ s->has_dedup = true;
+ s->dedup_table_offset = cluster_size * 2;
+ s->dedup_table_size = cluster_size / sizeof(uint64_t);
+ s->dedup_hash_algo = hash_algo;
+
+ ret = qcow2_set_incompat_feature(bs, QCOW2_INCOMPAT_DEDUP);
+ if (ret < 0) {
+ goto out;
+ }
+
+ s->dedup_status = DEDUP_STATUS_STARTED;
+ ret = qcow2_update_header(bs);
+ s->dedup_status = DEDUP_STATUS_STOPPED;
+ if (ret < 0) {
+ goto out;
+ }
+ }
+
/* Want a backing file? There you go.*/
if (backing_file) {
ret = bdrv_change_backing_file(bs, backing_file, backing_format);
@@ -1397,15 +1421,41 @@ out:
return ret;
}
+static int qcow2_warn_if_version_3_is_needed(int version,
+ bool has_feature,
+ const char *feature)
+{
+ if (version < 3 && has_feature) {
+ fprintf(stderr, "%s only supported with compatibility "
+ "level 1.1 and above (use compat=1.1 or greater)\n",
+ feature);
+ return -EINVAL;
+ }
+ return 0;
+}
+
+static int8_t qcow2_get_dedup_hash_algo(char *value)
+{
+ if (!value || !strcmp(value, "sha256")) {
+ return QCOW_HASH_SHA256;
+ }
+
+ error_printf("Unsupported deduplication hash algorithm.\n");
+ return -EINVAL;
+}
+
static int qcow2_create(const char *filename, QEMUOptionParameter *options)
{
const char *backing_file = NULL;
const char *backing_fmt = NULL;
uint64_t sectors = 0;
int flags = 0;
+ int ret;
size_t cluster_size = DEFAULT_CLUSTER_SIZE;
int prealloc = 0;
int version = 2;
+ bool dedup = false;
+ int8_t hash_algo = 0;
/* Read out options */
while (options && options->name) {
@@ -1443,6 +1493,13 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options)
}
} else if (!strcmp(options->name, BLOCK_OPT_LAZY_REFCOUNTS)) {
flags |= options->value.n ? BLOCK_FLAG_LAZY_REFCOUNTS : 0;
+ } else if (!strcmp(options->name, BLOCK_OPT_DEDUP)) {
+ hash_algo = qcow2_get_dedup_hash_algo(options->value.s);
+ if (hash_algo < 0) {
+ return hash_algo;
+ }
+ dedup = true;
+ version = 3;
}
options++;
}
@@ -1453,14 +1510,22 @@ static int qcow2_create(const char *filename, QEMUOptionParameter *options)
return -EINVAL;
}
- if (version < 3 && (flags & BLOCK_FLAG_LAZY_REFCOUNTS)) {
- fprintf(stderr, "Lazy refcounts only supported with compatibility "
- "level 1.1 and above (use compat=1.1 or greater)\n");
- return -EINVAL;
+ ret = qcow2_warn_if_version_3_is_needed(version,
+ flags & BLOCK_FLAG_LAZY_REFCOUNTS,
+ "Lazy refcounts");
+ if (ret < 0) {
+ return ret;
+ }
+ ret = qcow2_warn_if_version_3_is_needed(version,
+ dedup,
+ "Deduplication");
+ if (ret < 0) {
+ return ret;
}
return qcow2_create2(filename, sectors, backing_file, backing_fmt, flags,
- cluster_size, prealloc, options, version);
+ cluster_size, prealloc, options, version,
+ dedup, hash_algo);
}
static int qcow2_make_empty(BlockDriverState *bs)
@@ -1766,6 +1831,51 @@ static QEMUOptionParameter qcow2_create_options[] = {
{ NULL }
};
+static QEMUOptionParameter qcow2_dedup_create_options[] = {
+ {
+ .name = BLOCK_OPT_SIZE,
+ .type = OPT_SIZE,
+ .help = "Virtual disk size"
+ },
+ {
+ .name = BLOCK_OPT_BACKING_FILE,
+ .type = OPT_STRING,
+ .help = "File name of a base image"
+ },
+ {
+ .name = BLOCK_OPT_BACKING_FMT,
+ .type = OPT_STRING,
+ .help = "Image format of the base image"
+ },
+ {
+ .name = BLOCK_OPT_ENCRYPT,
+ .type = OPT_FLAG,
+ .help = "Encrypt the image"
+ },
+ {
+ .name = BLOCK_OPT_CLUSTER_SIZE,
+ .type = OPT_SIZE,
+ .help = "qcow2 cluster size",
+ .value = { .n = DEFAULT_DEDUP_CLUSTER_SIZE },
+ },
+ {
+ .name = BLOCK_OPT_PREALLOC,
+ .type = OPT_STRING,
+ .help = "Preallocation mode (allowed values: off, metadata)"
+ },
+ {
+ .name = BLOCK_OPT_LAZY_REFCOUNTS,
+ .type = OPT_FLAG,
+ .help = "Postpone refcount updates",
+ },
+ {
+ .name = BLOCK_OPT_DEDUP,
+ .type = OPT_STRING,
+ .help = "Deduplication",
+ },
+ { NULL }
+};
+
static BlockDriver bdrv_qcow2 = {
.format_name = "qcow2",
.instance_size = sizeof(BDRVQcowState),
@@ -1805,9 +1915,58 @@ static BlockDriver bdrv_qcow2 = {
.bdrv_check = qcow2_check,
};
+/* As all the defined .create_options are passed to qcow2_create() even if
+ * the user does not specify them it's not possible to have a default 4KB
+ * cluster size for deduplication.
+ * For example it's impossible to make the difference between the 64KB cluster
+ * size default create option of qcow2 or a 64KB user specified cluster size.
+ * So we declare the qcow2_dedup format in order to be able to define
+ * deduplication specific create options.
+ * It will also help for qemu-io-test integration.
+ */
+static BlockDriver bdrv_qcow2_dedup = {
+ .format_name = "qcow2_dedup",
+ .instance_size = sizeof(BDRVQcowState),
+ .bdrv_probe = qcow2_probe,
+ .bdrv_open = qcow2_open,
+ .bdrv_close = qcow2_close,
+ .bdrv_reopen_prepare = qcow2_reopen_prepare,
+ .bdrv_create = qcow2_create,
+ .bdrv_co_is_allocated = qcow2_co_is_allocated,
+ .bdrv_set_key = qcow2_set_key,
+ .bdrv_make_empty = qcow2_make_empty,
+
+ .bdrv_co_readv = qcow2_co_readv,
+ .bdrv_co_writev = qcow2_co_writev,
+ .bdrv_co_flush_to_os = qcow2_co_flush_to_os,
+
+ .bdrv_co_write_zeroes = qcow2_co_write_zeroes,
+ .bdrv_co_discard = qcow2_co_discard,
+ .bdrv_truncate = qcow2_truncate,
+ .bdrv_write_compressed = qcow2_write_compressed,
+
+ .bdrv_snapshot_create = qcow2_snapshot_create,
+ .bdrv_snapshot_goto = qcow2_snapshot_goto,
+ .bdrv_snapshot_delete = qcow2_snapshot_delete,
+ .bdrv_snapshot_list = qcow2_snapshot_list,
+ .bdrv_snapshot_load_tmp = qcow2_snapshot_load_tmp,
+ .bdrv_get_info = qcow2_get_info,
+
+ .bdrv_save_vmstate = qcow2_save_vmstate,
+ .bdrv_load_vmstate = qcow2_load_vmstate,
+
+ .bdrv_change_backing_file = qcow2_change_backing_file,
+
+ .bdrv_invalidate_cache = qcow2_invalidate_cache,
+
+ .create_options = qcow2_dedup_create_options,
+ .bdrv_check = qcow2_check,
+};
+
static void bdrv_qcow2_init(void)
{
bdrv_register(&bdrv_qcow2);
+ bdrv_register(&bdrv_qcow2_dedup);
}
block_init(bdrv_qcow2_init);
diff --git a/include/block/block_int.h b/include/block/block_int.h
index eaad53e..62c72fc 100644
--- a/include/block/block_int.h
+++ b/include/block/block_int.h
@@ -57,6 +57,7 @@
#define BLOCK_OPT_COMPAT_LEVEL "compat"
#define BLOCK_OPT_LAZY_REFCOUNTS "lazy_refcounts"
#define BLOCK_OPT_ADAPTER_TYPE "adapter_type"
+#define BLOCK_OPT_DEDUP "dedup"
typedef struct BdrvTrackedRequest BdrvTrackedRequest;
diff --git a/tests/qemu-iotests/common.rc b/tests/qemu-iotests/common.rc
index e522d61..520083a 100644
--- a/tests/qemu-iotests/common.rc
+++ b/tests/qemu-iotests/common.rc
@@ -124,7 +124,8 @@ _make_test_img()
-e "s# compat='[^']*'##g" \
-e "s# compat6=\\(on\\|off\\)##g" \
-e "s# static=\\(on\\|off\\)##g" \
- -e "s# lazy_refcounts=\\(on\\|off\\)##g"
+ -e "s# lazy_refcounts=\\(on\\|off\\)##g" \
+ -e "s# dedup=\\('sha256'\\|'skein'\\|'sha3'\\)##g"
# Start an NBD server on the image file, which is what we'll be talking to
if [ $IMGPROTO = "nbd" ]; then
--
1.7.10.4
next prev parent reply other threads:[~2013-03-15 14:51 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-03-15 14:49 [Qemu-devel] [RFC V7 00/32] QCOW2 deduplication core functionality Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 01/32] qcow2: Add deduplication to the qcow2 specification Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 02/32] qmp: Add DedupStatus enum Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 03/32] qcow2: Add deduplication structures and fields Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 04/32] qcow2: Add qcow2_dedup_read_missing_and_concatenate Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 05/32] qcow2: Create a way to link to l2 tables when deduplicating Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 06/32] qcow2: Make qcow2_update_cluster_refcount public Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 07/32] qcow2: Add qcow2_dedup and related functions Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 08/32] qcow2: Add qcow2_dedup_store_new_hashes Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 09/32] qcow2: Do allocate on rewrite on the dedup case Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 10/32] qcow2: Implement qcow2_compute_cluster_hash Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 11/32] qcow2: Add qcow2_dedup_grow_table and use it Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 12/32] qcow2: Makes qcow2_alloc_cluster_link_l2 mark to deduplicate clusters Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 13/32] qcow2: make the deduplication forget a cluster hash when a cluster is to dedupe Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 14/32] qcow2: Create qcow2_is_cluster_to_dedup Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 15/32] qcow2: Load and save deduplication table header extension Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 16/32] qcow2: Extract qcow2_do_table_init Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 17/32] qcow2-cache: Allow to choose table size at creation Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 18/32] qcow2: Extract qcow2_set_incompat_feature and qcow2_clear_incompat_feature Benoît Canet
2013-03-15 14:49 ` Benoît Canet [this message]
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 20/32] qcow2: Drop hash for a given cluster when dedup makes refcount > 2^16/2 Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 21/32] qcow2: Remove hash when cluster is deleted Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 22/32] qcow2: Add qcow2_dedup_is_running to probe if dedup is running Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 23/32] qcow2: Integrate deduplication in qcow2_co_writev loop Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 24/32] qcow2: Serialize write requests when deduplication is activated Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 25/32] qcow2: Adapt checking of QCOW_OFLAG_COPIED for dedup Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 26/32] qcow2: Add check_dedup_l2 in order to check l2 of dedup table Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 27/32] qcow2: Add verification " Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 28/32] qcow2: Integrate SKEIN hash algorithm in deduplication Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 29/32] qcow: Set large dedup hash block size Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 30/32] qcow2: Add qcow2_dedup_init and qcow2_dedup_close Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 31/32] qcow2: Add qcow2_co_dedup_resume to restart deduplication Benoît Canet
2013-03-15 14:49 ` [Qemu-devel] [RFC V7 32/32] qcow2: Enable the deduplication feature Benoît Canet
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1363358986-8360-20-git-send-email-benoit@irqsave.net \
--to=benoit@irqsave.net \
--cc=kwolf@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=stefanha@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).