From nobody Mon Feb 9 10:29:48 2026 Delivered-To: importer@patchew.org Received-SPF: pass (zoho.com: domain of gnu.org designates 209.51.188.17 as permitted sender) client-ip=209.51.188.17; envelope-from=qemu-devel-bounces+importer=patchew.org@nongnu.org; helo=lists.gnu.org; Authentication-Results: mx.zohomail.com; spf=pass (zoho.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=fail(p=none dis=none) header.from=virtuozzo.com ARC-Seal: i=1; a=rsa-sha256; t=1568656950; cv=none; d=zoho.com; s=zohoarc; b=JBA+i/tlntddrmqgbEFhoGFsIDJJU3LBnM3gP9ARETUrdxDWa0BFKscx+tyHWXN/TODjiKjEIoWyf+Eav/LVOvpNiboXXK4lRPgYOHBwasa4jcqq7FgIHi9bEGKk+TH8O+m95nPUf5RVrZfZSxBSEnVfnNI1DOZMSyYvC7hYhcA= ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=zoho.com; s=zohoarc; t=1568656950; h=Content-Transfer-Encoding:Cc:Date:From:In-Reply-To:List-Subscribe:List-Post:List-Id:List-Archive:List-Help:List-Unsubscribe:MIME-Version:Message-ID:References:Sender:Subject:To:ARC-Authentication-Results; bh=xwEdEVp20vloxWAyD8ZqCzXp07mfOnX43hDbwacxLNI=; b=Ji/Cz0N1oeNTmxvyLg+JV0GSgFk+1NH3u5/DATQjBrqYahELsbXBBKxL4PoO2YH6rJ7+t8iTGxNwIF85r9QzGMJ8E6bkQJAIS6NMN1i+2Bmh8IqTFrBLXjGmnCpC33pObBs0oBf3lXLi1qrNWJPXxBiR+ccLX1nVpL3K0jfzl/Q= ARC-Authentication-Results: i=1; mx.zoho.com; spf=pass (zoho.com: domain of gnu.org designates 209.51.188.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org; dmarc=fail header.from= (p=none dis=none) header.from= Return-Path: Received: from lists.gnu.org (209.51.188.17 [209.51.188.17]) by mx.zohomail.com with SMTPS id 1568656950671322.02855367232985; Mon, 16 Sep 2019 11:02:30 -0700 (PDT) Received: from localhost ([::1]:38318 helo=lists1p.gnu.org) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1i9vKR-0005PT-SK for importer@patchew.org; Mon, 16 Sep 2019 14:02:23 -0400 Received: from eggs.gnu.org ([2001:470:142:3::10]:33417) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1i9vC8-0006Xu-Ey for qemu-devel@nongnu.org; Mon, 16 Sep 2019 13:53:55 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1i9vC3-0004v3-Tr for qemu-devel@nongnu.org; Mon, 16 Sep 2019 13:53:48 -0400 Received: from relay.sw.ru ([185.231.240.75]:36812) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1i9vBp-0004nh-RQ; Mon, 16 Sep 2019 13:53:30 -0400 Received: from [10.94.3.0] (helo=kvm.qa.sw.ru) by relay.sw.ru with esmtp (Exim 4.92) (envelope-from ) id 1i9vBm-0004Nc-RW; Mon, 16 Sep 2019 20:53:26 +0300 From: Vladimir Sementsov-Ogievskiy To: qemu-block@nongnu.org Date: Mon, 16 Sep 2019 20:53:24 +0300 Message-Id: <20190916175324.18478-6-vsementsov@virtuozzo.com> X-Mailer: git-send-email 2.21.0 In-Reply-To: <20190916175324.18478-1-vsementsov@virtuozzo.com> References: <20190916175324.18478-1-vsementsov@virtuozzo.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-detected-operating-system: by eggs.gnu.org: GNU/Linux 3.x X-Received-From: 185.231.240.75 Subject: [Qemu-devel] [PATCH v5 5/5] block/qcow2: introduce parallel subrequest handling in read and write X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.23 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: kwolf@redhat.com, den@openvz.org, vsementsov@virtuozzo.com, qemu-devel@nongnu.org, mreitz@redhat.com Errors-To: qemu-devel-bounces+importer=patchew.org@nongnu.org Sender: "Qemu-devel" Content-Type: text/plain; charset="utf-8" It improves performance for fragmented qcow2 images. Signed-off-by: Vladimir Sementsov-Ogievskiy --- block/qcow2.h | 3 ++ block/qcow2.c | 125 ++++++++++++++++++++++++++++++++++++++++----- block/trace-events | 1 + 3 files changed, 117 insertions(+), 12 deletions(-) diff --git a/block/qcow2.h b/block/qcow2.h index a488d761ff..f51f478e34 100644 --- a/block/qcow2.h +++ b/block/qcow2.h @@ -65,6 +65,9 @@ #define QCOW2_MAX_BITMAPS 65535 #define QCOW2_MAX_BITMAP_DIRECTORY_SIZE (1024 * QCOW2_MAX_BITMAPS) =20 +/* Maximum of parallel sub-request per guest request */ +#define QCOW2_MAX_WORKERS 8 + /* indicate that the refcount of the referenced cluster is exactly one. */ #define QCOW_OFLAG_COPIED (1ULL << 63) /* indicate that the cluster is compressed (they never have the copied fla= g) */ diff --git a/block/qcow2.c b/block/qcow2.c index 164b22e4a2..7961c05783 100644 --- a/block/qcow2.c +++ b/block/qcow2.c @@ -41,6 +41,7 @@ #include "qapi/qobject-input-visitor.h" #include "qapi/qapi-visit-block-core.h" #include "crypto.h" +#include "block/aio_task.h" =20 /* Differences with QCOW: @@ -2025,6 +2026,60 @@ fail: return ret; } =20 +typedef struct Qcow2AioTask { + AioTask task; + + BlockDriverState *bs; + QCow2ClusterType cluster_type; /* only for read */ + uint64_t file_cluster_offset; + uint64_t offset; + uint64_t bytes; + QEMUIOVector *qiov; + uint64_t qiov_offset; + QCowL2Meta *l2meta; /* only for write */ +} Qcow2AioTask; + +static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task); +static coroutine_fn int qcow2_add_task(BlockDriverState *bs, + AioTaskPool *pool, + AioTaskFunc func, + QCow2ClusterType cluster_type, + uint64_t file_cluster_offset, + uint64_t offset, + uint64_t bytes, + QEMUIOVector *qiov, + size_t qiov_offset, + QCowL2Meta *l2meta) +{ + Qcow2AioTask local_task; + Qcow2AioTask *task =3D pool ? g_new(Qcow2AioTask, 1) : &local_task; + + *task =3D (Qcow2AioTask) { + .task.func =3D func, + .bs =3D bs, + .cluster_type =3D cluster_type, + .qiov =3D qiov, + .file_cluster_offset =3D file_cluster_offset, + .offset =3D offset, + .bytes =3D bytes, + .qiov_offset =3D qiov_offset, + .l2meta =3D l2meta, + }; + + trace_qcow2_add_task(qemu_coroutine_self(), bs, pool, + func =3D=3D qcow2_co_preadv_task_entry ? "read" := "write", + cluster_type, file_cluster_offset, offset, bytes, + qiov, qiov_offset); + + if (!pool) { + return func(&task->task); + } + + aio_task_pool_start_task(pool, &task->task); + + return 0; +} + static coroutine_fn int qcow2_co_preadv_task(BlockDriverState *bs, QCow2ClusterType cluster_type, uint64_t file_cluster_offset, @@ -2074,18 +2129,28 @@ static coroutine_fn int qcow2_co_preadv_task(BlockD= riverState *bs, g_assert_not_reached(); } =20 +static coroutine_fn int qcow2_co_preadv_task_entry(AioTask *task) +{ + Qcow2AioTask *t =3D container_of(task, Qcow2AioTask, task); + + assert(!t->l2meta); + + return qcow2_co_preadv_task(t->bs, t->cluster_type, t->file_cluster_of= fset, + t->offset, t->bytes, t->qiov, t->qiov_offs= et); +} + static coroutine_fn int qcow2_co_preadv_part(BlockDriverState *bs, uint64_t offset, uint64_t byt= es, QEMUIOVector *qiov, size_t qiov_offset, int flags) { BDRVQcow2State *s =3D bs->opaque; - int ret; + int ret =3D 0; unsigned int cur_bytes; /* number of bytes in current iteration */ uint64_t cluster_offset =3D 0; + AioTaskPool *aio =3D NULL; =20 - while (bytes !=3D 0) { - + while (bytes !=3D 0 && aio_task_pool_status(aio) =3D=3D 0) { /* prepare next request */ cur_bytes =3D MIN(bytes, INT_MAX); if (s->crypto) { @@ -2097,7 +2162,7 @@ static coroutine_fn int qcow2_co_preadv_part(BlockDri= verState *bs, ret =3D qcow2_get_cluster_offset(bs, offset, &cur_bytes, &cluster_= offset); qemu_co_mutex_unlock(&s->lock); if (ret < 0) { - return ret; + goto out; } =20 if (ret =3D=3D QCOW2_CLUSTER_ZERO_PLAIN || @@ -2106,11 +2171,14 @@ static coroutine_fn int qcow2_co_preadv_part(BlockD= riverState *bs, { qemu_iovec_memset(qiov, qiov_offset, 0, cur_bytes); } else { - ret =3D qcow2_co_preadv_task(bs, ret, - cluster_offset, offset, cur_bytes, - qiov, qiov_offset); + if (!aio && cur_bytes !=3D bytes) { + aio =3D aio_task_pool_new(QCOW2_MAX_WORKERS); + } + ret =3D qcow2_add_task(bs, aio, qcow2_co_preadv_task_entry, re= t, + cluster_offset, offset, cur_bytes, + qiov, qiov_offset, NULL); if (ret < 0) { - return ret; + goto out; } } =20 @@ -2119,7 +2187,16 @@ static coroutine_fn int qcow2_co_preadv_part(BlockDr= iverState *bs, qiov_offset +=3D cur_bytes; } =20 - return 0; +out: + if (aio) { + aio_task_pool_wait_all(aio); + if (ret =3D=3D 0) { + ret =3D aio_task_pool_status(aio); + } + g_free(aio); + } + + return ret; } =20 /* Check if it's possible to merge a write request with the writing of @@ -2324,6 +2401,17 @@ out_locked: return ret; } =20 +static coroutine_fn int qcow2_co_pwritev_task_entry(AioTask *task) +{ + Qcow2AioTask *t =3D container_of(task, Qcow2AioTask, task); + + assert(!t->cluster_type); + + return qcow2_co_pwritev_task(t->bs, t->file_cluster_offset, + t->offset, t->bytes, t->qiov, t->qiov_off= set, + t->l2meta); +} + static coroutine_fn int qcow2_co_pwritev_part( BlockDriverState *bs, uint64_t offset, uint64_t bytes, QEMUIOVector *qiov, size_t qiov_offset, int flags) @@ -2334,10 +2422,11 @@ static coroutine_fn int qcow2_co_pwritev_part( unsigned int cur_bytes; /* number of sectors in current iteration */ uint64_t cluster_offset; QCowL2Meta *l2meta =3D NULL; + AioTaskPool *aio =3D NULL; =20 trace_qcow2_writev_start_req(qemu_coroutine_self(), offset, bytes); =20 - while (bytes !=3D 0) { + while (bytes !=3D 0 && aio_task_pool_status(aio) =3D=3D 0) { =20 l2meta =3D NULL; =20 @@ -2369,8 +2458,12 @@ static coroutine_fn int qcow2_co_pwritev_part( =20 qemu_co_mutex_unlock(&s->lock); =20 - ret =3D qcow2_co_pwritev_task(bs, cluster_offset, offset, cur_byte= s, - qiov, qiov_offset, l2meta); + if (!aio && cur_bytes !=3D bytes) { + aio =3D aio_task_pool_new(QCOW2_MAX_WORKERS); + } + ret =3D qcow2_add_task(bs, aio, qcow2_co_pwritev_task_entry, 0, + cluster_offset, offset, cur_bytes, + qiov, qiov_offset, l2meta); l2meta =3D NULL; /* l2meta is consumed by qcow2_co_pwritev_task() = */ if (ret < 0) { goto fail_nometa; @@ -2391,6 +2484,14 @@ out_locked: qemu_co_mutex_unlock(&s->lock); =20 fail_nometa: + if (aio) { + aio_task_pool_wait_all(aio); + if (ret =3D=3D 0) { + ret =3D aio_task_pool_status(aio); + } + g_free(aio); + } + trace_qcow2_writev_done_req(qemu_coroutine_self(), ret); =20 return ret; diff --git a/block/trace-events b/block/trace-events index 04209f058d..3aa27e6b1e 100644 --- a/block/trace-events +++ b/block/trace-events @@ -62,6 +62,7 @@ file_paio_submit(void *acb, void *opaque, int64_t offset,= int count, int type) " file_copy_file_range(void *bs, int src, int64_t src_off, int dst, int64_t = dst_off, int64_t bytes, int flags, int64_t ret) "bs %p src_fd %d offset %"P= RIu64" dst_fd %d offset %"PRIu64" bytes %"PRIu64" flags %d ret %"PRId64 =20 # qcow2.c +qcow2_add_task(void *co, void *bs, void *pool, const char *action, int clu= ster_type, uint64_t file_cluster_offset, uint64_t offset, uint64_t bytes, v= oid *qiov, size_t qiov_offset) "co %p bs %p pool %p: %s: cluster_type %d fi= le_cluster_offset %" PRIu64 " offset %" PRIu64 " bytes %" PRIu64 " qiov %p = qiov_offset %zu" qcow2_writev_start_req(void *co, int64_t offset, int bytes) "co %p offset = 0x%" PRIx64 " bytes %d" qcow2_writev_done_req(void *co, int ret) "co %p ret %d" qcow2_writev_start_part(void *co) "co %p" --=20 2.21.0