qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Max Reitz <mreitz@redhat.com>
To: qemu-devel@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>,
	Stefan Hajnoczi <stefanha@redhat.com>,
	Max Reitz <mreitz@redhat.com>
Subject: [Qemu-devel] [PATCH 7/8] block/qcow2: Speed up zero cluster expansion
Date: Fri, 25 Jul 2014 20:07:44 +0200	[thread overview]
Message-ID: <1406311665-2814-8-git-send-email-mreitz@redhat.com> (raw)
In-Reply-To: <1406311665-2814-1-git-send-email-mreitz@redhat.com>

Actually, we do not need to allocate a new data cluster for every zero
cluster to be expanded: It is completely sufficient to rely on qcow2's
COW part and instead create a single zero cluster and reuse it as much
as possible.

Signed-off-by: Max Reitz <mreitz@redhat.com>
---
 block/qcow2-cluster.c | 119 ++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 92 insertions(+), 27 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index 905beb6..867db03 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -1558,6 +1558,9 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
     BDRVQcowState *s = bs->opaque;
     bool is_active_l1 = (l1_table == s->l1_table);
     uint64_t *l2_table = NULL;
+    int64_t zeroed_cluster_offset = 0;
+    int zeroed_cluster_refcount = 0;
+    int last_zeroed_cluster_l1i = 0, last_zeroed_cluster_l2i = 0;
     int ret;
     int i, j;
 
@@ -1617,47 +1620,79 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                     continue;
                 }
 
-                offset = qcow2_alloc_clusters(bs, s->cluster_size);
-                if (offset < 0) {
-                    ret = offset;
-                    goto fail;
+                if (zeroed_cluster_offset) {
+                    zeroed_cluster_refcount += l2_refcount;
+                    if (zeroed_cluster_refcount > 0xffff) {
+                        zeroed_cluster_refcount = 0;
+                        zeroed_cluster_offset = 0;
+                    }
                 }
+                if (!zeroed_cluster_offset) {
+                    offset = qcow2_alloc_clusters(bs, s->cluster_size);
+                    if (offset < 0) {
+                        ret = offset;
+                        goto fail;
+                    }
 
-                if (l2_refcount > 1) {
-                    /* For shared L2 tables, set the refcount accordingly (it is
-                     * already 1 and needs to be l2_refcount) */
-                    ret = qcow2_update_cluster_refcount(bs,
-                            offset >> s->cluster_bits, l2_refcount - 1,
-                            QCOW2_DISCARD_OTHER);
+                    ret = qcow2_pre_write_overlap_check(bs, 0, offset,
+                                                        s->cluster_size);
+                    if (ret < 0) {
+                        qcow2_free_clusters(bs, offset, s->cluster_size,
+                                            QCOW2_DISCARD_OTHER);
+                        goto fail;
+                    }
+
+                    ret = bdrv_write_zeroes(bs->file, offset / BDRV_SECTOR_SIZE,
+                                            s->cluster_sectors, 0);
                     if (ret < 0) {
                         qcow2_free_clusters(bs, offset, s->cluster_size,
                                             QCOW2_DISCARD_OTHER);
                         goto fail;
                     }
+
+                    if (l2_refcount > 1) {
+                        ret = qcow2_update_cluster_refcount(bs,
+                                offset >> s->cluster_bits, l2_refcount - 1,
+                                QCOW2_DISCARD_OTHER);
+                        if (ret < 0) {
+                            qcow2_free_clusters(bs, offset, s->cluster_size,
+                                                QCOW2_DISCARD_OTHER);
+                            goto fail;
+                        }
+                    }
+
+                    zeroed_cluster_offset = offset;
+                    zeroed_cluster_refcount = l2_refcount;
+                } else {
+                    ret = qcow2_update_cluster_refcount(bs,
+                            zeroed_cluster_offset >> s->cluster_bits,
+                            l2_refcount, QCOW2_DISCARD_OTHER);
+                    if (ret < 0) {
+                        goto fail;
+                    }
                 }
+
+                offset = zeroed_cluster_offset;
+                last_zeroed_cluster_l1i = i;
+                last_zeroed_cluster_l2i = j;
             }
 
-            ret = qcow2_pre_write_overlap_check(bs, 0, offset, s->cluster_size);
-            if (ret < 0) {
-                if (!preallocated) {
-                    qcow2_free_clusters(bs, offset, s->cluster_size,
-                                        QCOW2_DISCARD_ALWAYS);
+            if (preallocated) {
+                ret = qcow2_pre_write_overlap_check(bs, 0, offset,
+                                                    s->cluster_size);
+                if (ret < 0) {
+                    goto fail;
                 }
-                goto fail;
-            }
 
-            ret = bdrv_write_zeroes(bs->file, offset / BDRV_SECTOR_SIZE,
-                                    s->cluster_sectors, 0);
-            if (ret < 0) {
-                if (!preallocated) {
-                    qcow2_free_clusters(bs, offset, s->cluster_size,
-                                        QCOW2_DISCARD_ALWAYS);
+                ret = bdrv_write_zeroes(bs->file, offset / BDRV_SECTOR_SIZE,
+                                        s->cluster_sectors, 0);
+                if (ret < 0) {
+                    goto fail;
                 }
-                goto fail;
             }
 
-            if (l2_refcount == 1) {
-                l2_table[j] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
+            if (preallocated) {
+                l2_table[j] = cpu_to_be64(offset | (l2_entry & QCOW_OFLAG_COPIED));
             } else {
                 l2_table[j] = cpu_to_be64(offset);
             }
@@ -1670,8 +1705,8 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                 qcow2_cache_depends_on_flush(s->l2_table_cache);
             }
             ret = qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table);
+            l2_table = NULL;
             if (ret < 0) {
-                l2_table = NULL;
                 goto fail;
             }
         } else {
@@ -1697,6 +1732,36 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
         }
     }
 
+    /* Fix COPIED (only valid for active L2 tables) */
+    if (is_active_l1 && zeroed_cluster_refcount == 1) {
+        uint64_t l2_offset, l2_entry;
+
+        l2_offset = l1_table[last_zeroed_cluster_l1i] & L1E_OFFSET_MASK;
+        assert(l2_offset);
+
+        ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset,
+                              (void **)&l2_table);
+        if (ret < 0) {
+            goto fail;
+        }
+
+        l2_entry = be64_to_cpu(l2_table[last_zeroed_cluster_l2i]);
+
+        assert(!(l2_entry & QCOW_OFLAG_COPIED));
+        l2_entry |= QCOW_OFLAG_COPIED;
+
+        l2_table[last_zeroed_cluster_l2i] = cpu_to_be64(l2_entry);
+
+        qcow2_cache_entry_mark_dirty(s->l2_table_cache, l2_table);
+        qcow2_cache_depends_on_flush(s->l2_table_cache);
+
+        ret = qcow2_cache_put(bs, s->l2_table_cache, (void **)&l2_table);
+        l2_table = NULL;
+        if (ret < 0) {
+            goto fail;
+        }
+    }
+
     ret = 0;
 
 fail:
-- 
2.0.1

  parent reply	other threads:[~2014-07-25 18:07 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-07-25 18:07 [Qemu-devel] [PATCH 0/8] block/qcow2: Improve (?) zero cluster expansion Max Reitz
2014-07-25 18:07 ` [Qemu-devel] [PATCH 1/8] block: Add status callback to bdrv_amend_options() Max Reitz
2014-07-30 14:50   ` Eric Blake
2014-07-25 18:07 ` [Qemu-devel] [PATCH 2/8] qemu-img: Add progress output for amend Max Reitz
2014-07-30 14:55   ` Eric Blake
2014-07-30 20:20     ` Max Reitz
2014-07-25 18:07 ` [Qemu-devel] [PATCH 3/8] qemu-img: Fix insignifcant memleak Max Reitz
2014-07-30 14:56   ` Eric Blake
2014-07-25 18:07 ` [Qemu-devel] [PATCH 4/8] block/qcow2: Make get_refcount() global Max Reitz
2014-07-30 15:04   ` Eric Blake
2014-07-25 18:07 ` [Qemu-devel] [PATCH 5/8] block/qcow2: Implement status CB for amend Max Reitz
2014-07-30 15:23   ` Eric Blake
2014-07-25 18:07 ` [Qemu-devel] [PATCH 6/8] block/qcow2: Simplify shared L2 handling in amend Max Reitz
2014-07-30 15:36   ` Eric Blake
2014-07-25 18:07 ` Max Reitz [this message]
2014-07-30 16:14   ` [Qemu-devel] [PATCH 7/8] block/qcow2: Speed up zero cluster expansion Eric Blake
2014-07-30 20:31     ` Max Reitz
2014-07-30 20:31     ` Eric Blake
2014-07-30 20:41       ` Max Reitz
2014-07-25 18:07 ` [Qemu-devel] [PATCH 8/8] iotests: Expand test 061 Max Reitz

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=1406311665-2814-8-git-send-email-mreitz@redhat.com \
    --to=mreitz@redhat.com \
    --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).