From: Fam Zheng <famz@redhat.com>
To: qemu-devel@nongnu.org
Cc: Kevin Wolf <kwolf@redhat.com>,
qemu-block@nongnu.org, Jeff Cody <jcody@redhat.com>,
Stefan Hajnoczi <stefanha@redhat.com>,
pbonzini@redhat.com, jsnow@redhat.com, wangxiaolong@ucloud.cn
Subject: [Qemu-devel] [PATCH v3 1/6] mirror: Do zero write on target if sectors not allocated
Date: Tue, 12 May 2015 19:48:30 +0800 [thread overview]
Message-ID: <1431431315-32079-2-git-send-email-famz@redhat.com> (raw)
In-Reply-To: <1431431315-32079-1-git-send-email-famz@redhat.com>
If guest discards a source cluster during mirror, we would want to
skip the read-write steps.
Source and target may have different "can_write_zeroes_with_unmap"
values and different discard granularity. It's important to replicate
the exact zeroing on the destination, so bdrv_aio_discard is only safe
if source has can_write_zeroes_with_unmap=false.
Signed-off-by: Fam Zheng <famz@redhat.com>
---
block/mirror.c | 53 +++++++++++++++++++++++++++++++++++++++++++++++------
1 file changed, 47 insertions(+), 6 deletions(-)
diff --git a/block/mirror.c b/block/mirror.c
index 58f391a..1a1d997 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -57,6 +57,10 @@ typedef struct MirrorBlockJob {
int in_flight;
int sectors_in_flight;
int ret;
+ /* Source driver can_write_zeroes_with_unmap. */
+ bool source_may_unmap;
+ /* Target driver can_write_zeroes_with_unmap. */
+ bool target_may_unmap;
} MirrorBlockJob;
typedef struct MirrorOp {
@@ -163,6 +167,7 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
int64_t end, sector_num, next_chunk, next_sector, hbitmap_next_sector;
uint64_t delay_ns = 0;
MirrorOp *op;
+ int pnum;
s->sector_num = hbitmap_iter_next(&s->hbi);
if (s->sector_num < 0) {
@@ -289,8 +294,32 @@ static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
s->in_flight++;
s->sectors_in_flight += nb_sectors;
trace_mirror_one_iteration(s, sector_num, nb_sectors);
- bdrv_aio_readv(source, sector_num, &op->qiov, nb_sectors,
- mirror_read_complete, op);
+
+ if (!bdrv_is_allocated_above(source, NULL, sector_num,
+ nb_sectors, &pnum)) {
+ op->nb_sectors = pnum;
+ if (s->source_may_unmap) {
+ /*
+ * Source unallocated sectors have zero data. We can't discard
+ * target even if s->target_may_unmap, because the discard
+ * granularity may be different.
+ */
+ bdrv_aio_write_zeroes(s->target, sector_num, op->nb_sectors,
+ s->target_may_unmap ? BDRV_REQ_MAY_UNMAP : 0,
+ mirror_write_complete,
+ op);
+ } else {
+ /*
+ * Source has irrelevant data in unmapped sectors, it's safe to
+ * discard target.
+ * */
+ bdrv_aio_discard(s->target, sector_num, op->nb_sectors,
+ mirror_write_complete, op);
+ }
+ } else {
+ bdrv_aio_readv(source, sector_num, &op->qiov, nb_sectors,
+ mirror_read_complete, op);
+ }
return delay_ns;
}
@@ -399,6 +428,22 @@ static void coroutine_fn mirror_run(void *opaque)
length = DIV_ROUND_UP(s->bdev_length, s->granularity);
s->in_flight_bitmap = bitmap_new(length);
+ ret = bdrv_get_info(bs, &bdi);
+ if (ret < 0) {
+ /* Safe side. */
+ s->source_may_unmap = true;
+ } else {
+ s->source_may_unmap = bdi.can_write_zeroes_with_unmap;
+ }
+
+ ret = bdrv_get_info(s->target, &bdi);
+ if (ret < 0) {
+ /* Safe side. */
+ s->target_may_unmap = false;
+ } else {
+ s->target_may_unmap = bdi.can_write_zeroes_with_unmap;
+ }
+
/* If we have no backing file yet in the destination, we cannot let
* the destination do COW. Instead, we copy sectors around the
* dirty data if needed. We need a bitmap to do that.
@@ -406,10 +451,6 @@ static void coroutine_fn mirror_run(void *opaque)
bdrv_get_backing_filename(s->target, backing_filename,
sizeof(backing_filename));
if (backing_filename[0] && !s->target->backing_hd) {
- ret = bdrv_get_info(s->target, &bdi);
- if (ret < 0) {
- goto immediate_exit;
- }
if (s->granularity < bdi.cluster_size) {
s->buf_size = MAX(s->buf_size, bdi.cluster_size);
s->cow_bitmap = bitmap_new(length);
--
2.4.0
next prev parent reply other threads:[~2015-05-12 11:48 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-05-12 11:48 [Qemu-devel] [PATCH v3 0/6] block: Mirror discarded sectors Fam Zheng
2015-05-12 11:48 ` Fam Zheng [this message]
2015-05-14 10:30 ` [Qemu-devel] [PATCH v3 1/6] mirror: Do zero write on target if sectors not allocated Paolo Bonzini
2015-05-14 13:52 ` Fam Zheng
2015-05-12 11:48 ` [Qemu-devel] [PATCH v3 2/6] block: Fix dirty bitmap in bdrv_co_discard Fam Zheng
2015-05-14 10:31 ` Paolo Bonzini
2015-05-12 11:48 ` [Qemu-devel] [PATCH v3 3/6] block: Remove bdrv_reset_dirty Fam Zheng
2015-05-12 11:48 ` [Qemu-devel] [PATCH v3 4/6] qemu-iotests: Make block job methods common Fam Zheng
2015-05-14 10:31 ` Paolo Bonzini
2015-05-12 11:48 ` [Qemu-devel] [PATCH v3 5/6] qemu-iotests: Add test case for mirror with unmap Fam Zheng
2015-05-14 10:37 ` Paolo Bonzini
2015-05-14 13:50 ` Fam Zheng
2015-05-12 11:48 ` [Qemu-devel] [PATCH v3 6/6] iotests: Use event_wait in wait_ready Fam Zheng
2015-05-14 6:27 ` [Qemu-devel] [PATCH v3 0/6] block: Mirror discarded sectors Fam Zheng
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=1431431315-32079-2-git-send-email-famz@redhat.com \
--to=famz@redhat.com \
--cc=jcody@redhat.com \
--cc=jsnow@redhat.com \
--cc=kwolf@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=stefanha@redhat.com \
--cc=wangxiaolong@ucloud.cn \
/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).