From nobody Tue Feb 10 05:10:00 2026 Delivered-To: importer@patchew.org Received-SPF: pass (zoho.com: domain of gnu.org designates 208.118.235.17 as permitted sender) client-ip=208.118.235.17; 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 208.118.235.17 as permitted sender) smtp.mailfrom=qemu-devel-bounces+importer=patchew.org@nongnu.org Return-Path: Received: from lists.gnu.org (lists.gnu.org [208.118.235.17]) by mx.zohomail.com with SMTPS id 1500263680720632.5160060141161; Sun, 16 Jul 2017 20:54:40 -0700 (PDT) Received: from localhost ([::1]:47757 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dWx7H-0005Is-HE for importer@patchew.org; Sun, 16 Jul 2017 23:54:39 -0400 Received: from eggs.gnu.org ([2001:4830:134:3::10]:42433) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1dWx1Q-0000VX-QI for qemu-devel@nongnu.org; Sun, 16 Jul 2017 23:48:38 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1dWx1P-0004Xi-OD for qemu-devel@nongnu.org; Sun, 16 Jul 2017 23:48:36 -0400 Received: from mx1.redhat.com ([209.132.183.28]:46672) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1dWx1P-0004XH-FL for qemu-devel@nongnu.org; Sun, 16 Jul 2017 23:48:35 -0400 Received: from smtp.corp.redhat.com (int-mx02.intmail.prod.int.phx2.redhat.com [10.5.11.12]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 621012F86C4; Mon, 17 Jul 2017 03:48:34 +0000 (UTC) Received: from lemon.redhat.com (ovpn-12-94.pek2.redhat.com [10.72.12.94]) by smtp.corp.redhat.com (Postfix) with ESMTP id 1E99D70112; Mon, 17 Jul 2017 03:48:32 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com 621012F86C4 Authentication-Results: ext-mx05.extmail.prod.ext.phx2.redhat.com; dmarc=none (p=none dis=none) header.from=redhat.com Authentication-Results: ext-mx05.extmail.prod.ext.phx2.redhat.com; spf=pass smtp.mailfrom=famz@redhat.com DKIM-Filter: OpenDKIM Filter v2.11.0 mx1.redhat.com 621012F86C4 From: Fam Zheng To: qemu-devel@nongnu.org Date: Mon, 17 Jul 2017 11:48:13 +0800 Message-Id: <20170717034825.1524-4-famz@redhat.com> In-Reply-To: <20170717034825.1524-1-famz@redhat.com> References: <20170717034825.1524-1-famz@redhat.com> X-Scanned-By: MIMEDefang 2.79 on 10.5.11.12 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.29]); Mon, 17 Jul 2017 03:48:34 +0000 (UTC) X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] [fuzzy] X-Received-From: 209.132.183.28 Subject: [Qemu-devel] [PULL 03/15] vdi: make it thread-safe X-BeenThere: qemu-devel@nongnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Peter Maydell Errors-To: qemu-devel-bounces+importer=patchew.org@nongnu.org Sender: "Qemu-devel" X-ZohoMail: RSF_0 Z_629925259 SPT_0 Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" From: Paolo Bonzini The VirtualBox driver is using a mutex to order all allocating writes, but it is not protecting accesses to the bitmap because they implicitly happen under the AioContext mutex. Change this to use a CoRwlock explicitly. Reviewed-by: Eric Blake Reviewed-by: Stefan Hajnoczi Reviewed-by: Fam Zheng Signed-off-by: Paolo Bonzini Message-Id: <20170629132749.997-4-pbonzini@redhat.com> Signed-off-by: Fam Zheng --- block/vdi.c | 48 ++++++++++++++++++++++++------------------------ 1 file changed, 24 insertions(+), 24 deletions(-) diff --git a/block/vdi.c b/block/vdi.c index 2b6e8fa..8da5dfc 100644 --- a/block/vdi.c +++ b/block/vdi.c @@ -172,7 +172,7 @@ typedef struct { /* VDI header (converted to host endianness). */ VdiHeader header; =20 - CoMutex write_lock; + CoRwlock bmap_lock; =20 Error *migration_blocker; } BDRVVdiState; @@ -485,7 +485,7 @@ static int vdi_open(BlockDriverState *bs, QDict *option= s, int flags, goto fail_free_bmap; } =20 - qemu_co_mutex_init(&s->write_lock); + qemu_co_rwlock_init(&s->bmap_lock); =20 return 0; =20 @@ -557,7 +557,9 @@ vdi_co_preadv(BlockDriverState *bs, uint64_t offset, ui= nt64_t bytes, n_bytes, offset); =20 /* prepare next AIO request */ + qemu_co_rwlock_rdlock(&s->bmap_lock); bmap_entry =3D le32_to_cpu(s->bmap[block_index]); + qemu_co_rwlock_unlock(&s->bmap_lock); if (!VDI_IS_ALLOCATED(bmap_entry)) { /* Block not allocated, return zeros, no need to wait. */ qemu_iovec_memset(qiov, bytes_done, 0, n_bytes); @@ -595,6 +597,7 @@ vdi_co_pwritev(BlockDriverState *bs, uint64_t offset, u= int64_t bytes, uint32_t block_index; uint32_t offset_in_block; uint32_t n_bytes; + uint64_t data_offset; uint32_t bmap_first =3D VDI_UNALLOCATED; uint32_t bmap_last =3D VDI_UNALLOCATED; uint8_t *block =3D NULL; @@ -614,10 +617,19 @@ vdi_co_pwritev(BlockDriverState *bs, uint64_t offset,= uint64_t bytes, n_bytes, offset); =20 /* prepare next AIO request */ + qemu_co_rwlock_rdlock(&s->bmap_lock); bmap_entry =3D le32_to_cpu(s->bmap[block_index]); if (!VDI_IS_ALLOCATED(bmap_entry)) { /* Allocate new block and write to it. */ uint64_t data_offset; + qemu_co_rwlock_upgrade(&s->bmap_lock); + bmap_entry =3D le32_to_cpu(s->bmap[block_index]); + if (VDI_IS_ALLOCATED(bmap_entry)) { + /* A concurrent allocation did the work for us. */ + qemu_co_rwlock_downgrade(&s->bmap_lock); + goto nonallocating_write; + } + bmap_entry =3D s->header.blocks_allocated; s->bmap[block_index] =3D cpu_to_le32(bmap_entry); s->header.blocks_allocated++; @@ -635,30 +647,18 @@ vdi_co_pwritev(BlockDriverState *bs, uint64_t offset,= uint64_t bytes, memset(block + offset_in_block + n_bytes, 0, s->block_size - n_bytes - offset_in_block); =20 - /* Note that this coroutine does not yield anywhere from readi= ng the - * bmap entry until here, so in regards to all the coroutines = trying - * to write to this cluster, the one doing the allocation will - * always be the first to try to acquire the lock. - * Therefore, it is also the first that will actually be able = to - * acquire the lock and thus the padded cluster is written bef= ore - * the other coroutines can write to the affected area. */ - qemu_co_mutex_lock(&s->write_lock); + /* Write the new block under CoRwLock write-side protection, + * so this full-cluster write does not overlap a partial write + * of the same cluster, issued from the "else" branch. + */ ret =3D bdrv_pwrite(bs->file, data_offset, block, s->block_siz= e); - qemu_co_mutex_unlock(&s->write_lock); + qemu_co_rwlock_unlock(&s->bmap_lock); } else { - uint64_t data_offset =3D s->header.offset_data + - (uint64_t)bmap_entry * s->block_size + - offset_in_block; - qemu_co_mutex_lock(&s->write_lock); - /* This lock is only used to make sure the following write ope= ration - * is executed after the write issued by the coroutine allocat= ing - * this cluster, therefore we do not need to keep it locked. - * As stated above, the allocating coroutine will always try t= o lock - * the mutex before all the other concurrent accesses to that - * cluster, therefore at this point we can be absolutely certa= in - * that that write operation has returned (there may be other = writes - * in flight, but they do not concern this very operation). */ - qemu_co_mutex_unlock(&s->write_lock); +nonallocating_write: + data_offset =3D s->header.offset_data + + (uint64_t)bmap_entry * s->block_size + + offset_in_block; + qemu_co_rwlock_unlock(&s->bmap_lock); =20 qemu_iovec_reset(&local_qiov); qemu_iovec_concat(&local_qiov, qiov, bytes_done, n_bytes); --=20 2.9.4