From: Stefan Hajnoczi <stefanha@redhat.com>
To: qemu-devel@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>,
Anthony Liguori <aliguori@us.ibm.com>,
Stefan Hajnoczi <stefanha@redhat.com>
Subject: [Qemu-devel] [PATCH 19/23] qcow2: Add refcount update reason to all callers
Date: Mon, 24 Jun 2013 11:10:31 +0200 [thread overview]
Message-ID: <1372065035-19601-20-git-send-email-stefanha@redhat.com> (raw)
In-Reply-To: <1372065035-19601-1-git-send-email-stefanha@redhat.com>
From: Kevin Wolf <kwolf@redhat.com>
This adds a refcount update reason to all callers of update_refcounts(),
so that a follow-up patch can use this information to decide whether
clusters that reach a refcount of 0 should be discarded in the image
file.
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
block/qcow2-cluster.c | 19 +++++++++++------
block/qcow2-refcount.c | 55 +++++++++++++++++++++++++++++++-------------------
block/qcow2-snapshot.c | 6 ++++--
block/qcow2.c | 3 ++-
block/qcow2.h | 16 ++++++++++++---
5 files changed, 66 insertions(+), 33 deletions(-)
diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index 76f30e5..3191d6b 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -98,14 +98,16 @@ int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
goto fail;
}
g_free(s->l1_table);
- qcow2_free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t));
+ qcow2_free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t),
+ QCOW2_DISCARD_OTHER);
s->l1_table_offset = new_l1_table_offset;
s->l1_table = new_l1_table;
s->l1_size = new_l1_size;
return 0;
fail:
g_free(new_l1_table);
- qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2);
+ qcow2_free_clusters(bs, new_l1_table_offset, new_l1_size2,
+ QCOW2_DISCARD_OTHER);
return ret;
}
@@ -548,7 +550,8 @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
/* Then decrease the refcount of the old table */
if (l2_offset) {
- qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t));
+ qcow2_free_clusters(bs, l2_offset, s->l2_size * sizeof(uint64_t),
+ QCOW2_DISCARD_OTHER);
}
}
@@ -715,10 +718,14 @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
/*
* If this was a COW, we need to decrease the refcount of the old cluster.
* Also flush bs->file to get the right order for L2 and refcount update.
+ *
+ * Don't discard clusters that reach a refcount of 0 (e.g. compressed
+ * clusters), the next write will reuse them anyway.
*/
if (j != 0) {
for (i = 0; i < j; i++) {
- qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1);
+ qcow2_free_any_clusters(bs, be64_to_cpu(old_cluster[i]), 1,
+ QCOW2_DISCARD_NEVER);
}
}
@@ -1339,7 +1346,7 @@ static int discard_single_l2(BlockDriverState *bs, uint64_t offset,
l2_table[l2_index + i] = cpu_to_be64(0);
/* Then decrease the refcount */
- qcow2_free_any_clusters(bs, old_offset, 1);
+ qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
}
ret = qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table);
@@ -1415,7 +1422,7 @@ static int zero_single_l2(BlockDriverState *bs, uint64_t offset,
qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
if (old_offset & QCOW_OFLAG_COMPRESSED) {
l2_table[l2_index + i] = cpu_to_be64(QCOW_OFLAG_ZERO);
- qcow2_free_any_clusters(bs, old_offset, 1);
+ qcow2_free_any_clusters(bs, old_offset, 1, QCOW2_DISCARD_REQUEST);
} else {
l2_table[l2_index + i] |= cpu_to_be64(QCOW_OFLAG_ZERO);
}
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index b32738f..6d35e49 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -29,7 +29,7 @@
static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size);
static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
int64_t offset, int64_t length,
- int addend);
+ int addend, enum qcow2_discard_type type);
/*********************************************************/
@@ -235,7 +235,8 @@ static int alloc_refcount_block(BlockDriverState *bs,
} else {
/* Described somewhere else. This can recurse at most twice before we
* arrive at a block that describes itself. */
- ret = update_refcount(bs, new_block, s->cluster_size, 1);
+ ret = update_refcount(bs, new_block, s->cluster_size, 1,
+ QCOW2_DISCARD_NEVER);
if (ret < 0) {
goto fail_block;
}
@@ -399,7 +400,8 @@ static int alloc_refcount_block(BlockDriverState *bs,
/* Free old table. Remember, we must not change free_cluster_index */
uint64_t old_free_cluster_index = s->free_cluster_index;
- qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t));
+ qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t),
+ QCOW2_DISCARD_OTHER);
s->free_cluster_index = old_free_cluster_index;
ret = load_refcount_block(bs, new_block, (void**) refcount_block);
@@ -420,7 +422,7 @@ fail_block:
/* XXX: cache several refcount block clusters ? */
static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs,
- int64_t offset, int64_t length, int addend)
+ int64_t offset, int64_t length, int addend, enum qcow2_discard_type type)
{
BDRVQcowState *s = bs->opaque;
int64_t start, last, cluster_offset;
@@ -506,7 +508,8 @@ fail:
*/
if (ret < 0) {
int dummy;
- dummy = update_refcount(bs, offset, cluster_offset - offset, -addend);
+ dummy = update_refcount(bs, offset, cluster_offset - offset, -addend,
+ QCOW2_DISCARD_NEVER);
(void)dummy;
}
@@ -522,12 +525,14 @@ fail:
*/
static int update_cluster_refcount(BlockDriverState *bs,
int64_t cluster_index,
- int addend)
+ int addend,
+ enum qcow2_discard_type type)
{
BDRVQcowState *s = bs->opaque;
int ret;
- ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend);
+ ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend,
+ type);
if (ret < 0) {
return ret;
}
@@ -579,7 +584,7 @@ int64_t qcow2_alloc_clusters(BlockDriverState *bs, int64_t size)
return offset;
}
- ret = update_refcount(bs, offset, size, 1);
+ ret = update_refcount(bs, offset, size, 1, QCOW2_DISCARD_NEVER);
if (ret < 0) {
return ret;
}
@@ -611,7 +616,8 @@ int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
old_free_cluster_index = s->free_cluster_index;
s->free_cluster_index = cluster_index + i;
- ret = update_refcount(bs, offset, i << s->cluster_bits, 1);
+ ret = update_refcount(bs, offset, i << s->cluster_bits, 1,
+ QCOW2_DISCARD_NEVER);
if (ret < 0) {
return ret;
}
@@ -649,7 +655,8 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
if (free_in_cluster == 0)
s->free_byte_offset = 0;
if ((offset & (s->cluster_size - 1)) != 0)
- update_cluster_refcount(bs, offset >> s->cluster_bits, 1);
+ update_cluster_refcount(bs, offset >> s->cluster_bits, 1,
+ QCOW2_DISCARD_NEVER);
} else {
offset = qcow2_alloc_clusters(bs, s->cluster_size);
if (offset < 0) {
@@ -659,7 +666,8 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
if ((cluster_offset + s->cluster_size) == offset) {
/* we are lucky: contiguous data */
offset = s->free_byte_offset;
- update_cluster_refcount(bs, offset >> s->cluster_bits, 1);
+ update_cluster_refcount(bs, offset >> s->cluster_bits, 1,
+ QCOW2_DISCARD_NEVER);
s->free_byte_offset += size;
} else {
s->free_byte_offset = offset;
@@ -676,12 +684,13 @@ int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
}
void qcow2_free_clusters(BlockDriverState *bs,
- int64_t offset, int64_t size)
+ int64_t offset, int64_t size,
+ enum qcow2_discard_type type)
{
int ret;
BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_FREE);
- ret = update_refcount(bs, offset, size, -1);
+ ret = update_refcount(bs, offset, size, -1, type);
if (ret < 0) {
fprintf(stderr, "qcow2_free_clusters failed: %s\n", strerror(-ret));
/* TODO Remember the clusters to free them later and avoid leaking */
@@ -692,8 +701,8 @@ void qcow2_free_clusters(BlockDriverState *bs,
* Free a cluster using its L2 entry (handles clusters of all types, e.g.
* normal cluster, compressed cluster, etc.)
*/
-void qcow2_free_any_clusters(BlockDriverState *bs,
- uint64_t l2_entry, int nb_clusters)
+void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
+ int nb_clusters, enum qcow2_discard_type type)
{
BDRVQcowState *s = bs->opaque;
@@ -705,12 +714,12 @@ void qcow2_free_any_clusters(BlockDriverState *bs,
s->csize_mask) + 1;
qcow2_free_clusters(bs,
(l2_entry & s->cluster_offset_mask) & ~511,
- nb_csectors * 512);
+ nb_csectors * 512, type);
}
break;
case QCOW2_CLUSTER_NORMAL:
qcow2_free_clusters(bs, l2_entry & L2E_OFFSET_MASK,
- nb_clusters << s->cluster_bits);
+ nb_clusters << s->cluster_bits, type);
break;
case QCOW2_CLUSTER_UNALLOCATED:
case QCOW2_CLUSTER_ZERO:
@@ -785,7 +794,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
int ret;
ret = update_refcount(bs,
(offset & s->cluster_offset_mask) & ~511,
- nb_csectors * 512, addend);
+ nb_csectors * 512, addend,
+ QCOW2_DISCARD_SNAPSHOT);
if (ret < 0) {
goto fail;
}
@@ -795,7 +805,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
} else {
uint64_t cluster_index = (offset & L2E_OFFSET_MASK) >> s->cluster_bits;
if (addend != 0) {
- refcount = update_cluster_refcount(bs, cluster_index, addend);
+ refcount = update_cluster_refcount(bs, cluster_index, addend,
+ QCOW2_DISCARD_SNAPSHOT);
} else {
refcount = get_refcount(bs, cluster_index);
}
@@ -827,7 +838,8 @@ int qcow2_update_snapshot_refcount(BlockDriverState *bs,
if (addend != 0) {
- refcount = update_cluster_refcount(bs, l2_offset >> s->cluster_bits, addend);
+ refcount = update_cluster_refcount(bs, l2_offset >> s->cluster_bits, addend,
+ QCOW2_DISCARD_SNAPSHOT);
} else {
refcount = get_refcount(bs, l2_offset >> s->cluster_bits);
}
@@ -1253,7 +1265,8 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
if (num_fixed) {
ret = update_refcount(bs, i << s->cluster_bits, 1,
- refcount2 - refcount1);
+ refcount2 - refcount1,
+ QCOW2_DISCARD_ALWAYS);
if (ret >= 0) {
(*num_fixed)++;
continue;
diff --git a/block/qcow2-snapshot.c b/block/qcow2-snapshot.c
index 992a5c8..0caac90 100644
--- a/block/qcow2-snapshot.c
+++ b/block/qcow2-snapshot.c
@@ -262,7 +262,8 @@ static int qcow2_write_snapshots(BlockDriverState *bs)
}
/* free the old snapshot table */
- qcow2_free_clusters(bs, s->snapshots_offset, s->snapshots_size);
+ qcow2_free_clusters(bs, s->snapshots_offset, s->snapshots_size,
+ QCOW2_DISCARD_SNAPSHOT);
s->snapshots_offset = snapshots_offset;
s->snapshots_size = snapshots_size;
return 0;
@@ -569,7 +570,8 @@ int qcow2_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
if (ret < 0) {
return ret;
}
- qcow2_free_clusters(bs, sn.l1_table_offset, sn.l1_size * sizeof(uint64_t));
+ qcow2_free_clusters(bs, sn.l1_table_offset, sn.l1_size * sizeof(uint64_t),
+ QCOW2_DISCARD_SNAPSHOT);
/* must update the copied flag on the current cluster offsets */
ret = qcow2_update_snapshot_refcount(bs, s->l1_table_offset, s->l1_size, 0);
diff --git a/block/qcow2.c b/block/qcow2.c
index 0fa5cb2..e28ea47 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -1196,7 +1196,8 @@ static int preallocate(BlockDriverState *bs)
ret = qcow2_alloc_cluster_link_l2(bs, meta);
if (ret < 0) {
- qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters);
+ qcow2_free_any_clusters(bs, meta->alloc_offset, meta->nb_clusters,
+ QCOW2_DISCARD_NEVER);
return ret;
}
diff --git a/block/qcow2.h b/block/qcow2.h
index 6959c6a..64a6479 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -129,6 +129,15 @@ enum {
QCOW2_COMPAT_FEAT_MASK = QCOW2_COMPAT_LAZY_REFCOUNTS,
};
+enum qcow2_discard_type {
+ QCOW2_DISCARD_NEVER = 0,
+ QCOW2_DISCARD_ALWAYS,
+ QCOW2_DISCARD_REQUEST,
+ QCOW2_DISCARD_SNAPSHOT,
+ QCOW2_DISCARD_OTHER,
+ QCOW2_DISCARD_MAX
+};
+
typedef struct Qcow2Feature {
uint8_t type;
uint8_t bit;
@@ -349,9 +358,10 @@ int qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset,
int nb_clusters);
int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size);
void qcow2_free_clusters(BlockDriverState *bs,
- int64_t offset, int64_t size);
-void qcow2_free_any_clusters(BlockDriverState *bs,
- uint64_t cluster_offset, int nb_clusters);
+ int64_t offset, int64_t size,
+ enum qcow2_discard_type type);
+void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry,
+ int nb_clusters, enum qcow2_discard_type type);
int qcow2_update_snapshot_refcount(BlockDriverState *bs,
int64_t l1_table_offset, int l1_size, int addend);
--
1.8.1.4
next prev parent reply other threads:[~2013-06-24 9:11 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-06-24 9:10 [Qemu-devel] [PULL 00/23] Block patches Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 01/23] ide: Add handler to ide_cmd_table Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 02/23] ide: Convert WIN_DSM to ide_cmd_table handler Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 03/23] ide: Convert WIN_IDENTIFY " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 04/23] ide: Convert cmd_nop commands " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 05/23] ide: Convert verify " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 06/23] ide: Convert read/write multiple " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 07/23] ide: Convert PIO read/write " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 08/23] ide: Convert DMA " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 09/23] ide: Convert READ NATIVE MAX ADDRESS " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 10/23] ide: Convert CHECK POWER MDOE " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 11/23] ide: Convert SET FEATURES " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 12/23] ide: Convert FLUSH CACHE " Stefan Hajnoczi
2013-07-03 21:41 ` Alex Williamson
2013-07-03 21:51 ` Alex Williamson
2013-07-04 7:58 ` Kevin Wolf
2013-06-24 9:10 ` [Qemu-devel] [PATCH 13/23] ide: Convert SEEK " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 14/23] ide: Convert ATAPI commands " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 15/23] ide: Convert CF-ATA " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 16/23] ide: Convert SMART " Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 17/23] ide: Clean up ide_exec_cmd() Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 18/23] Revert "block: Disable driver-specific options for 1.5" Stefan Hajnoczi
2013-06-24 9:10 ` Stefan Hajnoczi [this message]
2013-06-24 9:10 ` [Qemu-devel] [PATCH 20/23] qcow2: Options to enable discard for freed clusters Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 21/23] qcow2: Batch discards Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 22/23] block: Always enable discard on the protocol level Stefan Hajnoczi
2013-06-24 9:10 ` [Qemu-devel] [PATCH 23/23] vmdk: refuse to open higher version than supported Stefan Hajnoczi
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=1372065035-19601-20-git-send-email-stefanha@redhat.com \
--to=stefanha@redhat.com \
--cc=aliguori@us.ibm.com \
--cc=kwolf@redhat.com \
--cc=qemu-devel@nongnu.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.