From mboxrd@z Thu Jan 1 00:00:00 1970 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 vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 83AFCC4167E for ; Wed, 9 Feb 2022 06:18:00 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S231168AbiBIGRy (ORCPT ); Wed, 9 Feb 2022 01:17:54 -0500 Received: from gmail-smtp-in.l.google.com ([23.128.96.19]:34720 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S241467AbiBIGLY (ORCPT ); Wed, 9 Feb 2022 01:11:24 -0500 Received: from out30-132.freemail.mail.aliyun.com (out30-132.freemail.mail.aliyun.com [115.124.30.132]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 17388C002146; Tue, 8 Feb 2022 22:09:14 -0800 (PST) X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R161e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=e01e04357;MF=xuanzhuo@linux.alibaba.com;NM=1;PH=DS;RN=11;SR=0;TI=SMTPD_---0V3zwKp0_1644386916; Received: from localhost(mailfrom:xuanzhuo@linux.alibaba.com fp:SMTPD_---0V3zwKp0_1644386916) by smtp.aliyun-inc.com(127.0.0.1); Wed, 09 Feb 2022 14:08:37 +0800 Message-ID: <1644386755.446152-1-xuanzhuo@linux.alibaba.com> Subject: Re: [PATCH v3 13/17] virtio_pci: queue_reset: support VIRTIO_F_RING_RESET Date: Wed, 9 Feb 2022 14:05:55 +0800 From: Xuan Zhuo To: Jason Wang Cc: "Michael S. Tsirkin" , "David S. Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend , bpf@vger.kernel.org, virtualization , netdev References: <20220126073533.44994-1-xuanzhuo@linux.alibaba.com> <20220126073533.44994-14-xuanzhuo@linux.alibaba.com> <0908a9f6-562d-fab5-39c3-2f0125acc80e@redhat.com> <1644220750.6834595-1-xuanzhuo@linux.alibaba.com> <1644305735.2620988-1-xuanzhuo@linux.alibaba.com> In-Reply-To: Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org On Wed, 9 Feb 2022 13:44:06 +0800, Jason Wang wrote: > On Tue, Feb 8, 2022 at 3:44 PM Xuan Zhuo wro= te: > > > > On Tue, 8 Feb 2022 10:55:37 +0800, Jason Wang wro= te: > > > On Mon, Feb 7, 2022 at 4:19 PM Xuan Zhuo = wrote: > > > > > > > > On Mon, 7 Feb 2022 14:57:13 +0800, Jason Wang = wrote: > > > > > > > > > > =E5=9C=A8 2022/1/26 =E4=B8=8B=E5=8D=883:35, Xuan Zhuo =E5=86=99= =E9=81=93: > > > > > > This patch implements virtio pci support for QUEUE RESET. > > > > > > > > > > > > Performing reset on a queue is divided into two steps: > > > > > > > > > > > > 1. reset_vq: reset one vq > > > > > > 2. enable_reset_vq: re-enable the reset queue > > > > > > > > > > > > In the first step, these tasks will be completed: > > > > > > 1. notify the hardware queue to reset > > > > > > 2. recycle the buffer from vq > > > > > > 3. release the ring of the vq > > > > > > > > > > > > The process of enable reset vq: > > > > > > vp_modern_enable_reset_vq() > > > > > > vp_enable_reset_vq() > > > > > > __vp_setup_vq() > > > > > > setup_vq() > > > > > > vring_setup_virtqueue() > > > > > > > > > > > > In this process, we added two parameters, vq and num, and final= ly passed > > > > > > them to vring_setup_virtqueue(). And reuse the original info a= nd vq. > > > > > > > > > > > > Signed-off-by: Xuan Zhuo > > > > > > --- > > > > > > drivers/virtio/virtio_pci_common.c | 36 +++++++---- > > > > > > drivers/virtio/virtio_pci_common.h | 5 ++ > > > > > > drivers/virtio/virtio_pci_modern.c | 100 ++++++++++++++++++++= +++++++++ > > > > > > 3 files changed, 128 insertions(+), 13 deletions(-) > > > > > > > > > > > > diff --git a/drivers/virtio/virtio_pci_common.c b/drivers/virti= o/virtio_pci_common.c > > > > > > index c02936d29a31..ad21638fbf66 100644 > > > > > > --- a/drivers/virtio/virtio_pci_common.c > > > > > > +++ b/drivers/virtio/virtio_pci_common.c > > > > > > @@ -205,23 +205,28 @@ static int vp_request_msix_vectors(struct= virtio_device *vdev, int nvectors, > > > > > > return err; > > > > > > } > > > > > > > > > > > > -static struct virtqueue *vp_setup_vq(struct virtio_device *vde= v, unsigned index, > > > > > > - void (*callback)(struct virtqu= eue *vq), > > > > > > - const char *name, > > > > > > - bool ctx, > > > > > > - u16 msix_vec, u16 num) > > > > > > +struct virtqueue *vp_setup_vq(struct virtio_device *vdev, unsi= gned index, > > > > > > + void (*callback)(struct virtqueue *vq= ), > > > > > > + const char *name, > > > > > > + bool ctx, > > > > > > + u16 msix_vec, u16 num) > > > > > > { > > > > > > struct virtio_pci_device *vp_dev =3D to_vp_device(vdev); > > > > > > - struct virtio_pci_vq_info *info =3D kmalloc(sizeof *info, G= FP_KERNEL); > > > > > > + struct virtio_pci_vq_info *info; > > > > > > struct virtqueue *vq; > > > > > > unsigned long flags; > > > > > > > > > > > > - /* fill out our structure that represents an active queue */ > > > > > > - if (!info) > > > > > > - return ERR_PTR(-ENOMEM); > > > > > > + info =3D vp_dev->vqs[index]; > > > > > > + if (!info) { > > > > > > + info =3D kzalloc(sizeof *info, GFP_KERNEL); > > > > > > + > > > > > > + /* fill out our structure that represents an active= queue */ > > > > > > + if (!info) > > > > > > + return ERR_PTR(-ENOMEM); > > > > > > + } > > > > > > > > > > > > vq =3D vp_dev->setup_vq(vp_dev, info, index, callback, name= , ctx, > > > > > > - msix_vec, NULL, num); > > > > > > + msix_vec, info->vq, num); > > > > > > if (IS_ERR(vq)) > > > > > > goto out_info; > > > > > > > > > > > > @@ -238,6 +243,9 @@ static struct virtqueue *vp_setup_vq(struct= virtio_device *vdev, unsigned index, > > > > > > return vq; > > > > > > > > > > > > out_info: > > > > > > + if (info->vq && info->vq->reset) > > > > > > + return vq; > > > > > > + > > > > > > kfree(info); > > > > > > return vq; > > > > > > } > > > > > > @@ -248,9 +256,11 @@ static void vp_del_vq(struct virtqueue *vq) > > > > > > struct virtio_pci_vq_info *info =3D vp_dev->vqs[vq->index]; > > > > > > unsigned long flags; > > > > > > > > > > > > - spin_lock_irqsave(&vp_dev->lock, flags); > > > > > > - list_del(&info->node); > > > > > > - spin_unlock_irqrestore(&vp_dev->lock, flags); > > > > > > + if (!vq->reset) { > > > > > > + spin_lock_irqsave(&vp_dev->lock, flags); > > > > > > + list_del(&info->node); > > > > > > + spin_unlock_irqrestore(&vp_dev->lock, flags); > > > > > > + } > > > > > > > > > > > > vp_dev->del_vq(info); > > > > > > kfree(info); > > > > > > diff --git a/drivers/virtio/virtio_pci_common.h b/drivers/virti= o/virtio_pci_common.h > > > > > > index 65db92245e41..c1d15f7c0be4 100644 > > > > > > --- a/drivers/virtio/virtio_pci_common.h > > > > > > +++ b/drivers/virtio/virtio_pci_common.h > > > > > > @@ -119,6 +119,11 @@ int vp_find_vqs(struct virtio_device *vdev= , unsigned nvqs, > > > > > > struct virtqueue *vqs[], vq_callback_t *callbacks[], > > > > > > const char * const names[], const bool *ctx, > > > > > > struct irq_affinity *desc); > > > > > > +struct virtqueue *vp_setup_vq(struct virtio_device *vdev, unsi= gned index, > > > > > > + void (*callback)(struct virtqueue *vq= ), > > > > > > + const char *name, > > > > > > + bool ctx, > > > > > > + u16 msix_vec, u16 num); > > > > > > const char *vp_bus_name(struct virtio_device *vdev); > > > > > > > > > > > > /* Setup the affinity for a virtqueue: > > > > > > diff --git a/drivers/virtio/virtio_pci_modern.c b/drivers/virti= o/virtio_pci_modern.c > > > > > > index 2ce58de549de..6789411169e4 100644 > > > > > > --- a/drivers/virtio/virtio_pci_modern.c > > > > > > +++ b/drivers/virtio/virtio_pci_modern.c > > > > > > @@ -34,6 +34,9 @@ static void vp_transport_features(struct virt= io_device *vdev, u64 features) > > > > > > if ((features & BIT_ULL(VIRTIO_F_SR_IOV)) && > > > > > > pci_find_ext_capability(pci_dev, PCI_EXT_CA= P_ID_SRIOV)) > > > > > > __virtio_set_bit(vdev, VIRTIO_F_SR_IOV); > > > > > > + > > > > > > + if (features & BIT_ULL(VIRTIO_F_RING_RESET)) > > > > > > + __virtio_set_bit(vdev, VIRTIO_F_RING_RESET); > > > > > > } > > > > > > > > > > > > /* virtio config->finalize_features() implementation */ > > > > > > @@ -176,6 +179,94 @@ static void vp_reset(struct virtio_device = *vdev) > > > > > > vp_disable_cbs(vdev); > > > > > > } > > > > > > > > > > > > +static int vp_modern_reset_vq(struct virtio_reset_vq *param) > > > > > > +{ > > > > > > + struct virtio_pci_device *vp_dev =3D to_vp_device(param->vd= ev); > > > > > > + struct virtio_pci_modern_device *mdev =3D &vp_dev->mdev; > > > > > > + struct virtio_pci_vq_info *info; > > > > > > + u16 msix_vec, queue_index; > > > > > > + unsigned long flags; > > > > > > + void *buf; > > > > > > + > > > > > > + if (!virtio_has_feature(param->vdev, VIRTIO_F_RING_RESET)) > > > > > > + return -ENOENT; > > > > > > + > > > > > > + queue_index =3D param->queue_index; > > > > > > + > > > > > > + vp_modern_set_queue_reset(mdev, queue_index); > > > > > > + > > > > > > + /* After write 1 to queue reset, the driver MUST wait for a= read of > > > > > > + * queue reset to return 1. > > > > > > + */ > > > > > > + while (vp_modern_get_queue_reset(mdev, queue_index) !=3D 1) > > > > > > + msleep(1); > > > > > > > > > > > > > > > Is this better to move this logic into vp_modern_set_queue_reset(= )? > > > > > > > > > > > > > OK. > > > > > > > > > > > > > > > + > > > > > > + info =3D vp_dev->vqs[queue_index]; > > > > > > + msix_vec =3D info->msix_vector; > > > > > > + > > > > > > + /* Disable VQ callback. */ > > > > > > + if (vp_dev->per_vq_vectors && msix_vec !=3D VIRTIO_MSI_NO_V= ECTOR) > > > > > > + disable_irq(pci_irq_vector(vp_dev->pci_dev, msix_ve= c)); > > > > > > > > > > > > > > > How about the INTX case where irq is shared? I guess we need to d= isable > > > > > and enable the irq as well. > > > > > > > > For the INTX scenario, I don't think we need to disable/enable the = irq. But I do > > > > have a mistake, I should put the following list_del(&info->node) he= re, so that > > > > when the irq comes, it will no longer operate this vq. > > > > > > > > In fact, for no INTX case, the disable_irq here is not necessary, a= ccording to > > > > the requirements of the spec, after reset, the device should not ge= nerate irq > > > > anymore. > > > > > > I may miss something but I don't think INTX differs from MSI from the > > > vq handler perspective. > > > > > > > Here just to prevent accidents. > > > > > > The problem is that if you don't disable/sync IRQ there could be a > > > race between the vq irq handler and the virtqueue_detach_unused_buf()? > > > > > > > > > > > > > > > > > > > > > > > + > > > > > > + while ((buf =3D virtqueue_detach_unused_buf(info->vq)) !=3D= NULL) > > > > > > + param->free_unused_cb(param, buf); > > > > > > > > > > > > > > > Any reason that we can't leave this logic to driver? (Or is there= any > > > > > operations that must be done before the following operations?) > > > > > > > > As you commented before, we merged free unused buf and reset queue. > > > > > > > > I think it's a good note, otherwise, we're going to > > > > > > > > 1. reset vq > > > > 2. free unused(by driver) > > > > 3. free ring of vq > > > > > > Rethink about this, I'd prefer to leave it to the driver for consiste= ncy. > > > > > > E.g the virtqueue_detach_unused_buf() is called by each driver nowday= s. > > > > In this case, go back to my previous design and add three helpers: > > > > int (*reset_vq)(); > > int (*free_reset_vq)(); > > So this is only needed if there are any transport specific operations. > But I don't see there's any. Yes, you are right. Performing reset on a queue is divided into four steps 1. virtio_config_ops->reset_vq(): reset one vq 2. recycle the buffer from vq by virtqueue_detach_unused_buf() 3. release the ring of the vq by vring_release_virtqueue() (new functio= n) 4. virtio_config_ops->enable_reset_vq(): re-enable the reset queue Thanks. > > Thanks > > > int (*enable_reset_vq)(); > > > > Thanks. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > + > > > > > > + /* delete vq */ > > > > > > + spin_lock_irqsave(&vp_dev->lock, flags); > > > > > > + list_del(&info->node); > > > > > > + spin_unlock_irqrestore(&vp_dev->lock, flags); > > > > > > + > > > > > > + INIT_LIST_HEAD(&info->node); > > > > > > + > > > > > > + if (vp_dev->msix_enabled) > > > > > > + vp_modern_queue_vector(mdev, info->vq->index, > > > > > > + VIRTIO_MSI_NO_VECTOR); > > > > > > > > > > > > > > > I wonder if this is a must. > > > > > > > > > > > > > > > > + > > > > > > + if (!mdev->notify_base) > > > > > > + pci_iounmap(mdev->pci_dev, > > > > > > + (void __force __iomem *)info->vq->priv); > > > > > > > > > > > > > > > Is this a must? what happens if we simply don't do this? > > > > > > > > > > > > > The purpose of these two operations is mainly to be symmetrical wit= h del_vq(). > > > > > > This is another question I want to ask. Any reason for calling > > > del_vq()? Is it because you need to exclude a vq from the vq handler? > > > > > > For any case, the MSI and notification stuff seems unnecessary. > > > > > > Thanks > > > > > > > > > > > > > > > > > > > > > > + > > > > > > + vring_reset_virtqueue(info->vq); > > > > > > + > > > > > > + return 0; > > > > > > +} > > > > > > + > > > > > > +static struct virtqueue *vp_modern_enable_reset_vq(struct virt= io_reset_vq *param) > > > > > > +{ > > > > > > + struct virtio_pci_device *vp_dev =3D to_vp_device(param->vd= ev); > > > > > > + struct virtio_pci_modern_device *mdev =3D &vp_dev->mdev; > > > > > > + struct virtio_pci_vq_info *info; > > > > > > + u16 msix_vec, queue_index; > > > > > > + struct virtqueue *vq; > > > > > > + > > > > > > + if (!virtio_has_feature(param->vdev, VIRTIO_F_RING_RESET)) > > > > > > + return ERR_PTR(-ENOENT); > > > > > > + > > > > > > + queue_index =3D param->queue_index; > > > > > > + > > > > > > + info =3D vp_dev->vqs[queue_index]; > > > > > > + > > > > > > + if (!info->vq->reset) > > > > > > + return ERR_PTR(-EPERM); > > > > > > + > > > > > > + /* check queue reset status */ > > > > > > + if (vp_modern_get_queue_reset(mdev, queue_index) !=3D 1) > > > > > > + return ERR_PTR(-EBUSY); > > > > > > + > > > > > > + vq =3D vp_setup_vq(param->vdev, queue_index, param->callbac= k, param->name, > > > > > > + param->ctx, info->msix_vector, param->ring= _num); > > > > > > + if (IS_ERR(vq)) > > > > > > + return vq; > > > > > > + > > > > > > + vp_modern_set_queue_enable(&vp_dev->mdev, vq->index, true); > > > > > > + > > > > > > + msix_vec =3D vp_dev->vqs[queue_index]->msix_vector; > > > > > > + if (vp_dev->per_vq_vectors && msix_vec !=3D VIRTIO_MSI_NO_V= ECTOR) > > > > > > + enable_irq(pci_irq_vector(vp_dev->pci_dev, msix_vec= )); > > > > > > > > > > > > > > > How about the INT-X case? > > > > > > > > Explained above. > > > > > > > > Thanks. > > > > > > > > > > > > > > Thanks > > > > > > > > > > > > > > > > + > > > > > > + return vq; > > > > > > +} > > > > > > + > > > > > > static u16 vp_config_vector(struct virtio_pci_device *vp_dev,= u16 vector) > > > > > > { > > > > > > return vp_modern_config_vector(&vp_dev->mdev, vector); > > > > > > @@ -284,6 +375,11 @@ static void del_vq(struct virtio_pci_vq_in= fo *info) > > > > > > struct virtio_pci_device *vp_dev =3D to_vp_device(vq->vdev); > > > > > > struct virtio_pci_modern_device *mdev =3D &vp_dev->mdev; > > > > > > > > > > > > + if (vq->reset) { > > > > > > + vring_del_virtqueue(vq); > > > > > > + return; > > > > > > + } > > > > > > + > > > > > > if (vp_dev->msix_enabled) > > > > > > vp_modern_queue_vector(mdev, vq->index, > > > > > > VIRTIO_MSI_NO_VECTOR); > > > > > > @@ -403,6 +499,8 @@ static const struct virtio_config_ops virti= o_pci_config_nodev_ops =3D { > > > > > > .set_vq_affinity =3D vp_set_vq_affinity, > > > > > > .get_vq_affinity =3D vp_get_vq_affinity, > > > > > > .get_shm_region =3D vp_get_shm_region, > > > > > > + .reset_vq =3D vp_modern_reset_vq, > > > > > > + .enable_reset_vq =3D vp_modern_enable_reset_vq, > > > > > > }; > > > > > > > > > > > > static const struct virtio_config_ops virtio_pci_config_ops = =3D { > > > > > > @@ -421,6 +519,8 @@ static const struct virtio_config_ops virti= o_pci_config_ops =3D { > > > > > > .set_vq_affinity =3D vp_set_vq_affinity, > > > > > > .get_vq_affinity =3D vp_get_vq_affinity, > > > > > > .get_shm_region =3D vp_get_shm_region, > > > > > > + .reset_vq =3D vp_modern_reset_vq, > > > > > > + .enable_reset_vq =3D vp_modern_enable_reset_vq, > > > > > > }; > > > > > > > > > > > > /* the PCI probing function */ > > > > > > > > > > > > > > >