From patchwork Tue May 16 19:02:31 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Stefan Hajnoczi X-Patchwork-Id: 13243552 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from lists.xenproject.org (lists.xenproject.org [192.237.175.120]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 0D45BC7EE24 for ; Tue, 16 May 2023 19:07:59 +0000 (UTC) Received: from list by lists.xenproject.org with outflank-mailman.535498.833308 (Exim 4.92) (envelope-from ) id 1pz01L-0001hq-JU; Tue, 16 May 2023 19:07:39 +0000 X-Outflank-Mailman: Message body and most headers restored to incoming version Received: by outflank-mailman (output) from mailman id 535498.833308; Tue, 16 May 2023 19:07:39 +0000 Received: from localhost ([127.0.0.1] helo=lists.xenproject.org) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1pz01L-0001hf-GN; Tue, 16 May 2023 19:07:39 +0000 Received: by outflank-mailman (input) for mailman id 535498; Tue, 16 May 2023 19:07:37 +0000 Received: from se1-gles-flk1-in.inumbo.com ([94.247.172.50] helo=se1-gles-flk1.inumbo.com) by lists.xenproject.org with esmtp (Exim 4.92) (envelope-from ) id 1pyzxS-0002US-Ps for xen-devel@lists.xenproject.org; Tue, 16 May 2023 19:03:38 +0000 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.129.124]) by se1-gles-flk1.inumbo.com (Halon) with ESMTPS id 5c007aab-f41c-11ed-8611-37d641c3527e; Tue, 16 May 2023 21:03:36 +0200 (CEST) Received: from mimecast-mx02.redhat.com (mx3-rdu2.redhat.com [66.187.233.73]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-319-5SgmAg8zMuCHjVmCq-BmUA-1; Tue, 16 May 2023 15:03:32 -0400 Received: from smtp.corp.redhat.com (int-mx07.intmail.prod.int.rdu2.redhat.com [10.11.54.7]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id 7590B38107A8; Tue, 16 May 2023 19:03:31 +0000 (UTC) Received: from localhost (unknown [10.39.192.44]) by smtp.corp.redhat.com (Postfix) with ESMTP id D9BB414171C1; Tue, 16 May 2023 19:03:30 +0000 (UTC) X-BeenThere: xen-devel@lists.xenproject.org List-Id: Xen developer discussion List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: xen-devel-bounces@lists.xenproject.org Precedence: list Sender: "Xen-devel" X-Inumbo-ID: 5c007aab-f41c-11ed-8611-37d641c3527e DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1684263815; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=6NKS2h08c7ICfd8e4ndbZUq0n9RkVrcwRBS+UKLWGJE=; b=Nt/2QcSwA1S41n94usC3c/n1p8RIkwKiT7ubHWFmHyAA58FZMMSYvCEjMsGinJKU/h728g VnfQTUuZY/cOvU6XokphJ/ZFJzRIG53ETUl/kww3iEs1NnqgaL25yK3ZjELRwxOupoQvZO KXViQ3m2nk+GXkbfcO818g8v+BI5mu8= X-MC-Unique: 5SgmAg8zMuCHjVmCq-BmUA-1 From: Stefan Hajnoczi To: qemu-devel@nongnu.org Cc: Coiby Xu , Marcel Apfelbaum , "Michael S. Tsirkin" , "Richard W.M. Jones" , Peter Xu , xen-devel@lists.xenproject.org, Kevin Wolf , Ronnie Sahlberg , Stefano Stabellini , Richard Henderson , Stefan Hajnoczi , Julia Suvorova , Hanna Reitz , Leonardo Bras , eesposit@redhat.com, Fam Zheng , Aarushi Mehta , David Woodhouse , Xie Yongji , Stefano Garzarella , qemu-block@nongnu.org, Eduardo Habkost , Paul Durrant , Stefan Weil , Anthony Perard , =?utf-8?q?Daniel_P=2E_Berrang=C3=A9?= , Peter Lieven , Paolo Bonzini , =?utf-8?q?P?= =?utf-8?q?hilippe_Mathieu-Daud=C3=A9?= , Juan Quintela Subject: [PATCH v6 13/20] block/export: rewrite vduse-blk drain code Date: Tue, 16 May 2023 15:02:31 -0400 Message-Id: <20230516190238.8401-14-stefanha@redhat.com> In-Reply-To: <20230516190238.8401-1-stefanha@redhat.com> References: <20230516190238.8401-1-stefanha@redhat.com> MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.1 on 10.11.54.7 vduse_blk_detach_ctx() waits for in-flight requests using AIO_WAIT_WHILE(). This is not allowed according to a comment in bdrv_set_aio_context_commit(): /* * Take the old AioContex when detaching it from bs. * At this point, new_context lock is already acquired, and we are now * also taking old_context. This is safe as long as bdrv_detach_aio_context * does not call AIO_POLL_WHILE(). */ Use this opportunity to rewrite the drain code in vduse-blk: - Use the BlockExport refcount so that vduse_blk_exp_delete() is only called when there are no more requests in flight. - Implement .drained_poll() so in-flight request coroutines are stopped by the time .bdrv_detach_aio_context() is called. - Remove AIO_WAIT_WHILE() from vduse_blk_detach_ctx() to solve the .bdrv_detach_aio_context() constraint violation. It's no longer needed due to the previous changes. - Always handle the VDUSE file descriptor, even in drained sections. The VDUSE file descriptor doesn't submit I/O, so it's safe to handle it in drained sections. This ensures that the VDUSE kernel code gets a fast response. - Suspend virtqueue fd handlers in .drained_begin() and resume them in .drained_end(). This eliminates the need for the aio_set_fd_handler(is_external=true) flag, which is being removed from QEMU. This is a long list but splitting it into individual commits would probably lead to git bisect failures - the changes are all related. Signed-off-by: Stefan Hajnoczi Reviewed-by: Kevin Wolf --- block/export/vduse-blk.c | 132 +++++++++++++++++++++++++++------------ 1 file changed, 93 insertions(+), 39 deletions(-) diff --git a/block/export/vduse-blk.c b/block/export/vduse-blk.c index b53ef39da0..a25556fe04 100644 --- a/block/export/vduse-blk.c +++ b/block/export/vduse-blk.c @@ -31,7 +31,8 @@ typedef struct VduseBlkExport { VduseDev *dev; uint16_t num_queues; char *recon_file; - unsigned int inflight; + unsigned int inflight; /* atomic */ + bool vqs_started; } VduseBlkExport; typedef struct VduseBlkReq { @@ -41,13 +42,24 @@ typedef struct VduseBlkReq { static void vduse_blk_inflight_inc(VduseBlkExport *vblk_exp) { - vblk_exp->inflight++; + if (qatomic_fetch_inc(&vblk_exp->inflight) == 0) { + /* Prevent export from being deleted */ + aio_context_acquire(vblk_exp->export.ctx); + blk_exp_ref(&vblk_exp->export); + aio_context_release(vblk_exp->export.ctx); + } } static void vduse_blk_inflight_dec(VduseBlkExport *vblk_exp) { - if (--vblk_exp->inflight == 0) { + if (qatomic_fetch_dec(&vblk_exp->inflight) == 1) { + /* Wake AIO_WAIT_WHILE() */ aio_wait_kick(); + + /* Now the export can be deleted */ + aio_context_acquire(vblk_exp->export.ctx); + blk_exp_unref(&vblk_exp->export); + aio_context_release(vblk_exp->export.ctx); } } @@ -124,8 +136,12 @@ static void vduse_blk_enable_queue(VduseDev *dev, VduseVirtq *vq) { VduseBlkExport *vblk_exp = vduse_dev_get_priv(dev); + if (!vblk_exp->vqs_started) { + return; /* vduse_blk_drained_end() will start vqs later */ + } + aio_set_fd_handler(vblk_exp->export.ctx, vduse_queue_get_fd(vq), - true, on_vduse_vq_kick, NULL, NULL, NULL, vq); + false, on_vduse_vq_kick, NULL, NULL, NULL, vq); /* Make sure we don't miss any kick afer reconnecting */ eventfd_write(vduse_queue_get_fd(vq), 1); } @@ -133,9 +149,14 @@ static void vduse_blk_enable_queue(VduseDev *dev, VduseVirtq *vq) static void vduse_blk_disable_queue(VduseDev *dev, VduseVirtq *vq) { VduseBlkExport *vblk_exp = vduse_dev_get_priv(dev); + int fd = vduse_queue_get_fd(vq); - aio_set_fd_handler(vblk_exp->export.ctx, vduse_queue_get_fd(vq), - true, NULL, NULL, NULL, NULL, NULL); + if (fd < 0) { + return; + } + + aio_set_fd_handler(vblk_exp->export.ctx, fd, false, + NULL, NULL, NULL, NULL, NULL); } static const VduseOps vduse_blk_ops = { @@ -152,42 +173,19 @@ static void on_vduse_dev_kick(void *opaque) static void vduse_blk_attach_ctx(VduseBlkExport *vblk_exp, AioContext *ctx) { - int i; - aio_set_fd_handler(vblk_exp->export.ctx, vduse_dev_get_fd(vblk_exp->dev), - true, on_vduse_dev_kick, NULL, NULL, NULL, + false, on_vduse_dev_kick, NULL, NULL, NULL, vblk_exp->dev); - for (i = 0; i < vblk_exp->num_queues; i++) { - VduseVirtq *vq = vduse_dev_get_queue(vblk_exp->dev, i); - int fd = vduse_queue_get_fd(vq); - - if (fd < 0) { - continue; - } - aio_set_fd_handler(vblk_exp->export.ctx, fd, true, - on_vduse_vq_kick, NULL, NULL, NULL, vq); - } + /* Virtqueues are handled by vduse_blk_drained_end() */ } static void vduse_blk_detach_ctx(VduseBlkExport *vblk_exp) { - int i; - - for (i = 0; i < vblk_exp->num_queues; i++) { - VduseVirtq *vq = vduse_dev_get_queue(vblk_exp->dev, i); - int fd = vduse_queue_get_fd(vq); - - if (fd < 0) { - continue; - } - aio_set_fd_handler(vblk_exp->export.ctx, fd, - true, NULL, NULL, NULL, NULL, NULL); - } aio_set_fd_handler(vblk_exp->export.ctx, vduse_dev_get_fd(vblk_exp->dev), - true, NULL, NULL, NULL, NULL, NULL); + false, NULL, NULL, NULL, NULL, NULL); - AIO_WAIT_WHILE(vblk_exp->export.ctx, vblk_exp->inflight > 0); + /* Virtqueues are handled by vduse_blk_drained_begin() */ } @@ -220,8 +218,55 @@ static void vduse_blk_resize(void *opaque) (char *)&config.capacity); } +static void vduse_blk_stop_virtqueues(VduseBlkExport *vblk_exp) +{ + for (uint16_t i = 0; i < vblk_exp->num_queues; i++) { + VduseVirtq *vq = vduse_dev_get_queue(vblk_exp->dev, i); + vduse_blk_disable_queue(vblk_exp->dev, vq); + } + + vblk_exp->vqs_started = false; +} + +static void vduse_blk_start_virtqueues(VduseBlkExport *vblk_exp) +{ + vblk_exp->vqs_started = true; + + for (uint16_t i = 0; i < vblk_exp->num_queues; i++) { + VduseVirtq *vq = vduse_dev_get_queue(vblk_exp->dev, i); + vduse_blk_enable_queue(vblk_exp->dev, vq); + } +} + +static void vduse_blk_drained_begin(void *opaque) +{ + BlockExport *exp = opaque; + VduseBlkExport *vblk_exp = container_of(exp, VduseBlkExport, export); + + vduse_blk_stop_virtqueues(vblk_exp); +} + +static void vduse_blk_drained_end(void *opaque) +{ + BlockExport *exp = opaque; + VduseBlkExport *vblk_exp = container_of(exp, VduseBlkExport, export); + + vduse_blk_start_virtqueues(vblk_exp); +} + +static bool vduse_blk_drained_poll(void *opaque) +{ + BlockExport *exp = opaque; + VduseBlkExport *vblk_exp = container_of(exp, VduseBlkExport, export); + + return qatomic_read(&vblk_exp->inflight) > 0; +} + static const BlockDevOps vduse_block_ops = { - .resize_cb = vduse_blk_resize, + .resize_cb = vduse_blk_resize, + .drained_begin = vduse_blk_drained_begin, + .drained_end = vduse_blk_drained_end, + .drained_poll = vduse_blk_drained_poll, }; static int vduse_blk_exp_create(BlockExport *exp, BlockExportOptions *opts, @@ -268,6 +313,7 @@ static int vduse_blk_exp_create(BlockExport *exp, BlockExportOptions *opts, vblk_exp->handler.serial = g_strdup(vblk_opts->serial ?: ""); vblk_exp->handler.logical_block_size = logical_block_size; vblk_exp->handler.writable = opts->writable; + vblk_exp->vqs_started = true; config.capacity = cpu_to_le64(blk_getlength(exp->blk) >> VIRTIO_BLK_SECTOR_BITS); @@ -322,14 +368,20 @@ static int vduse_blk_exp_create(BlockExport *exp, BlockExportOptions *opts, vduse_dev_setup_queue(vblk_exp->dev, i, queue_size); } - aio_set_fd_handler(exp->ctx, vduse_dev_get_fd(vblk_exp->dev), true, + aio_set_fd_handler(exp->ctx, vduse_dev_get_fd(vblk_exp->dev), false, on_vduse_dev_kick, NULL, NULL, NULL, vblk_exp->dev); blk_add_aio_context_notifier(exp->blk, blk_aio_attached, blk_aio_detach, vblk_exp); - blk_set_dev_ops(exp->blk, &vduse_block_ops, exp); + /* + * We handle draining ourselves using an in-flight counter and by disabling + * virtqueue fd handlers. Do not queue BlockBackend requests, they need to + * complete so the in-flight counter reaches zero. + */ + blk_set_disable_request_queuing(exp->blk, true); + return 0; err: vduse_dev_destroy(vblk_exp->dev); @@ -344,6 +396,9 @@ static void vduse_blk_exp_delete(BlockExport *exp) VduseBlkExport *vblk_exp = container_of(exp, VduseBlkExport, export); int ret; + assert(qatomic_read(&vblk_exp->inflight) == 0); + + vduse_blk_detach_ctx(vblk_exp); blk_remove_aio_context_notifier(exp->blk, blk_aio_attached, blk_aio_detach, vblk_exp); ret = vduse_dev_destroy(vblk_exp->dev); @@ -354,13 +409,12 @@ static void vduse_blk_exp_delete(BlockExport *exp) g_free(vblk_exp->handler.serial); } +/* Called with exp->ctx acquired */ static void vduse_blk_exp_request_shutdown(BlockExport *exp) { VduseBlkExport *vblk_exp = container_of(exp, VduseBlkExport, export); - aio_context_acquire(vblk_exp->export.ctx); - vduse_blk_detach_ctx(vblk_exp); - aio_context_acquire(vblk_exp->export.ctx); + vduse_blk_stop_virtqueues(vblk_exp); } const BlockExportDriver blk_exp_vduse_blk = {