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 BF726C7EE29 for ; Thu, 8 Jun 2023 18:35:58 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S235116AbjFHSf5 (ORCPT ); Thu, 8 Jun 2023 14:35:57 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:47416 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231643AbjFHSfz (ORCPT ); Thu, 8 Jun 2023 14:35:55 -0400 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 3D9F12695 for ; Thu, 8 Jun 2023 11:35:11 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1686249310; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=a2bQcHG+zKEW1dxMHjPekLRCcGESsdy0BSM0jY0k8Cc=; b=hwQbDVor8XKgG0xx9aQDhZP8Df6yB4shQMbuQLYMpled/0iEBo6ijoCCrmB3evySB6GqTL w7QYs6YI3m/v8qsHLPzyJ/DBxPS2soBbTkKWWXMvdwXyTzLLlcqhApZd1kMBzpWTLldegc ciXYxS2oT/jtZH16MDKsUDSKqg9FkLo= Received: from mail-wr1-f69.google.com (mail-wr1-f69.google.com [209.85.221.69]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-460-oQeRYHy3MpGI5ykZ8ZPifw-1; Thu, 08 Jun 2023 14:35:09 -0400 X-MC-Unique: oQeRYHy3MpGI5ykZ8ZPifw-1 Received: by mail-wr1-f69.google.com with SMTP id ffacd0b85a97d-30ae4ed92eeso422300f8f.1 for ; Thu, 08 Jun 2023 11:35:09 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20221208; t=1686249308; x=1688841308; h=in-reply-to:content-transfer-encoding:content-disposition :mime-version:references:message-id:subject:cc:to:from:date :x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=a2bQcHG+zKEW1dxMHjPekLRCcGESsdy0BSM0jY0k8Cc=; b=hNzi2sWqdR8z3zEaomOODzfK49EPlbThiIUrQ/PbVbonTFTEL5Aov7l1CD3NZ4PEcb a7FQLUw3NuCMoSD+ur6Ya8Xv8JA/Lb4ERTFVn+UvLm6RIrDlsMg7Faxv5bqQJtau47/+ +a0ewVuQtGpKCm1lkPMFjuPFge1Db+bOYLTn0l7k2mvCa7/UvVgaErWny8eVt0SU04nI kSanaQgZeVN9ghi36bHHs2EOjrrJXjCpxxpGzbRWwi6VYRk97PlcAGSYGWSlcwdcteUN 96Qvcybiz6oc98sIywL3uCxgyHmv3bdwDVfrhgUvWvyXzgTWS2zuwpbah1fIkye4QLr8 nWhw== X-Gm-Message-State: AC+VfDyzNMIUqhYAGpoX3kmW6SLpUwKFEs5Ykxcnmue8L8u2qtx4xXF8 HYRtMHA0x3qNN6jsCiRzzTKoWud8o+IAZOAE5mW1g3DVlZM6fFrd+2UyD0a7cwM7YejraX68B+X 3yRbyDVEmYR0L2rJPZHnA5pkHRyqI2ih9 X-Received: by 2002:a5d:408b:0:b0:30a:e675:7e92 with SMTP id o11-20020a5d408b000000b0030ae6757e92mr6996783wrp.24.1686249307974; Thu, 08 Jun 2023 11:35:07 -0700 (PDT) X-Google-Smtp-Source: ACHHUZ7ebde3/2zvOHcahGfGkNtXLeveVr4UNWK45EvEkw55IL0kPC4bYFpc4NDrXh0K7WqjuxGaMQ== X-Received: by 2002:a5d:408b:0:b0:30a:e675:7e92 with SMTP id o11-20020a5d408b000000b0030ae6757e92mr6996773wrp.24.1686249307738; Thu, 08 Jun 2023 11:35:07 -0700 (PDT) Received: from redhat.com ([2.55.4.169]) by smtp.gmail.com with ESMTPSA id t18-20020adfeb92000000b003093a412310sm2249080wrn.92.2023.06.08.11.35.05 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 08 Jun 2023 11:35:07 -0700 (PDT) Date: Thu, 8 Jun 2023 14:35:03 -0400 From: "Michael S. Tsirkin" To: Dragos Tatulea Cc: "jasowang@redhat.com" , "xuanzhuo@linux.alibaba.com" , "virtualization@lists.linux-foundation.org" , Eli Cohen , Saeed Mahameed , "linux-kernel@vger.kernel.org" Subject: Re: [PATCH] vdpa/mlx5: Support interrupt bypassing Message-ID: <20230608143457-mutt-send-email-mst@kernel.org> References: <20230607190007.290505-1-dtatulea@nvidia.com> <118eac3c96665e3eac3397eef127606200f4aa50.camel@nvidia.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <118eac3c96665e3eac3397eef127606200f4aa50.camel@nvidia.com> Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Jun 08, 2023 at 04:25:55PM +0000, Dragos Tatulea wrote: > On Wed, 2023-06-07 at 22:00 +0300, Dragos Tatulea via Virtualization wrote: > > From: Eli Cohen > > > > Add support for generation of interrupts from the device directly to the > > VM to the VCPU thus avoiding the overhead on the host CPU. > > > > When supported, the driver will attempt to allocate vectors for each > > data virtqueue. If a vector for a virtqueue cannot be provided it will > > use the QP mode where notifications go through the driver. > > > > In addition, we add a shutdown callback to make sure allocated > > interrupts are released in case of shutdown to allow clean shutdown. > > > > Signed-off-by: Eli Cohen > > Signed-off-by: Saeed Mahameed > > > Just realized that this patch should have been marked as a v3. Let me know if I > should resend it. no need. > > --- > >  drivers/vdpa/mlx5/net/mlx5_vnet.c | 165 ++++++++++++++++++++++++++++-- > >  drivers/vdpa/mlx5/net/mlx5_vnet.h |  15 +++ > >  2 files changed, 171 insertions(+), 9 deletions(-) > > > > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c > > b/drivers/vdpa/mlx5/net/mlx5_vnet.c > > index 279ac6a558d2..9138ef2fb2c8 100644 > > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c > > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c > > @@ -83,6 +83,7 @@ struct mlx5_vq_restore_info { > >         u64 driver_addr; > >         u16 avail_index; > >         u16 used_index; > > +       struct msi_map map; > >         bool ready; > >         bool restore; > >  }; > > @@ -118,6 +119,7 @@ struct mlx5_vdpa_virtqueue { > >         u16 avail_idx; > >         u16 used_idx; > >         int fw_state; > > +       struct msi_map map; > >   > >         /* keep last in the struct */ > >         struct mlx5_vq_restore_info ri; > > @@ -808,6 +810,13 @@ static bool counters_supported(const struct mlx5_vdpa_dev > > *mvdev) > >                BIT_ULL(MLX5_OBJ_TYPE_VIRTIO_Q_COUNTERS); > >  } > >   > > +static bool msix_mode_supported(struct mlx5_vdpa_dev *mvdev) > > +{ > > +       return MLX5_CAP_DEV_VDPA_EMULATION(mvdev->mdev, event_mode) & > > +               (1 << MLX5_VIRTIO_Q_EVENT_MODE_MSIX_MODE) && > > +               pci_msix_can_alloc_dyn(mvdev->mdev->pdev); > > +} > > + > >  static int create_virtqueue(struct mlx5_vdpa_net *ndev, struct > > mlx5_vdpa_virtqueue *mvq) > >  { > >         int inlen = MLX5_ST_SZ_BYTES(create_virtio_net_q_in); > > @@ -849,9 +858,15 @@ static int create_virtqueue(struct mlx5_vdpa_net *ndev, > > struct mlx5_vdpa_virtque > >         if (vq_is_tx(mvq->index)) > >                 MLX5_SET(virtio_net_q_object, obj_context, tisn_or_qpn, ndev- > > >res.tisn); > >   > > -       MLX5_SET(virtio_q, vq_ctx, event_mode, > > MLX5_VIRTIO_Q_EVENT_MODE_QP_MODE); > > +       if (mvq->map.virq) { > > +               MLX5_SET(virtio_q, vq_ctx, event_mode, > > MLX5_VIRTIO_Q_EVENT_MODE_MSIX_MODE); > > +               MLX5_SET(virtio_q, vq_ctx, event_qpn_or_msix, mvq->map.index); > > +       } else { > > +               MLX5_SET(virtio_q, vq_ctx, event_mode, > > MLX5_VIRTIO_Q_EVENT_MODE_QP_MODE); > > +               MLX5_SET(virtio_q, vq_ctx, event_qpn_or_msix, mvq- > > >fwqp.mqp.qpn); > > +       } > > + > >         MLX5_SET(virtio_q, vq_ctx, queue_index, mvq->index); > > -       MLX5_SET(virtio_q, vq_ctx, event_qpn_or_msix, mvq->fwqp.mqp.qpn); > >         MLX5_SET(virtio_q, vq_ctx, queue_size, mvq->num_ent); > >         MLX5_SET(virtio_q, vq_ctx, virtio_version_1_0, > >                  !!(ndev->mvdev.actual_features & > > BIT_ULL(VIRTIO_F_VERSION_1))); > > @@ -1194,6 +1209,56 @@ static void counter_set_dealloc(struct mlx5_vdpa_net > > *ndev, struct mlx5_vdpa_vir > >                 mlx5_vdpa_warn(&ndev->mvdev, "dealloc counter set 0x%x\n", > > mvq->counter_set_id); > >  } > >   > > +static irqreturn_t mlx5_vdpa_int_handler(int irq, void *priv) > > +{ > > +       struct vdpa_callback *cb = priv; > > + > > +       if (cb->callback) > > +               return cb->callback(cb->private); > > + > > +       return IRQ_HANDLED; > > +} > > + > > +static void alloc_vector(struct mlx5_vdpa_net *ndev, > > +                        struct mlx5_vdpa_virtqueue *mvq) > > +{ > > +       struct mlx5_vdpa_irq_pool *irqp = &ndev->irqp; > > +       struct mlx5_vdpa_irq_pool_entry *ent; > > +       int err; > > +       int i; > > + > > +       for (i = 0; i < irqp->num_ent; i++) { > > +               ent = &irqp->entries[i]; > > +               if (!ent->used) { > > +                       snprintf(ent->name, MLX5_VDPA_IRQ_NAME_LEN, "%s-vq- > > %d", > > +                                dev_name(&ndev->mvdev.vdev.dev), mvq->index); > > +                       ent->dev_id = &ndev->event_cbs[mvq->index]; > > +                       err = request_irq(ent->map.virq, > > mlx5_vdpa_int_handler, 0, > > +                                         ent->name, ent->dev_id); > > +                       if (err) > > +                               return; > > + > > +                       ent->used = true; > > +                       mvq->map = ent->map; > > +                       return; > > +               } > > +       } > > +} > > + > > +static void dealloc_vector(struct mlx5_vdpa_net *ndev, > > +                          struct mlx5_vdpa_virtqueue *mvq) > > +{ > > +       struct mlx5_vdpa_irq_pool *irqp = &ndev->irqp; > > +       int i; > > + > > +       for (i = 0; i < irqp->num_ent; i++) > > +               if (mvq->map.virq == irqp->entries[i].map.virq) { > > +                       free_irq(mvq->map.virq, irqp->entries[i].dev_id); > > +                       irqp->entries[i].used = false; > > +                       return; > > +               } > > +} > > + > >  static int setup_vq(struct mlx5_vdpa_net *ndev, struct mlx5_vdpa_virtqueue > > *mvq) > >  { > >         u16 idx = mvq->index; > > @@ -1223,27 +1288,31 @@ static int setup_vq(struct mlx5_vdpa_net *ndev, struct > > mlx5_vdpa_virtqueue *mvq) > >   > >         err = counter_set_alloc(ndev, mvq); > >         if (err) > > -               goto err_counter; > > +               goto err_connect; > >   > > +       alloc_vector(ndev, mvq); > >         err = create_virtqueue(ndev, mvq); > >         if (err) > > -               goto err_connect; > > +               goto err_vq; > >   > >         if (mvq->ready) { > >                 err = modify_virtqueue(ndev, mvq, > > MLX5_VIRTIO_NET_Q_OBJECT_STATE_RDY); > >                 if (err) { > >                         mlx5_vdpa_warn(&ndev->mvdev, "failed to modify to > > ready vq idx %d(%d)\n", > >                                        idx, err); > > -                       goto err_connect; > > +                       goto err_modify; > >                 } > >         } > >   > >         mvq->initialized = true; > >         return 0; > >   > > -err_connect: > > +err_modify: > > +       destroy_virtqueue(ndev, mvq); > > +err_vq: > > +       dealloc_vector(ndev, mvq); > >         counter_set_dealloc(ndev, mvq); > > -err_counter: > > +err_connect: > >         qp_destroy(ndev, &mvq->vqqp); > >  err_vqqp: > >         qp_destroy(ndev, &mvq->fwqp); > > @@ -1288,6 +1357,7 @@ static void teardown_vq(struct mlx5_vdpa_net *ndev, > > struct mlx5_vdpa_virtqueue * > >   > >         suspend_vq(ndev, mvq); > >         destroy_virtqueue(ndev, mvq); > > +       dealloc_vector(ndev, mvq); > >         counter_set_dealloc(ndev, mvq); > >         qp_destroy(ndev, &mvq->vqqp); > >         qp_destroy(ndev, &mvq->fwqp); > > @@ -2505,6 +2575,7 @@ static int save_channel_info(struct mlx5_vdpa_net *ndev, > > struct mlx5_vdpa_virtqu > >         ri->desc_addr = mvq->desc_addr; > >         ri->device_addr = mvq->device_addr; > >         ri->driver_addr = mvq->driver_addr; > > +       ri->map = mvq->map; > >         ri->restore = true; > >         return 0; > >  } > > @@ -2549,6 +2620,7 @@ static void restore_channels_info(struct mlx5_vdpa_net > > *ndev) > >                 mvq->desc_addr = ri->desc_addr; > >                 mvq->device_addr = ri->device_addr; > >                 mvq->driver_addr = ri->driver_addr; > > +               mvq->map = ri->map; > >         } > >  } > >   > > @@ -2833,6 +2905,25 @@ static struct device *mlx5_get_vq_dma_dev(struct > > vdpa_device *vdev, u16 idx) > >         return mvdev->vdev.dma_dev; > >  } > >   > > +static void free_irqs(struct mlx5_vdpa_net *ndev) > > +{ > > +       struct mlx5_vdpa_irq_pool_entry *ent; > > +       int i; > > + > > +       if (!msix_mode_supported(&ndev->mvdev)) > > +               return; > > + > > +       if (!ndev->irqp.entries) > > +               return; > > + > > +       for (i = ndev->irqp.num_ent - 1; i >= 0; i--) { > > +               ent = ndev->irqp.entries + i; > > +               if (ent->map.virq) > > +                       pci_msix_free_irq(ndev->mvdev.mdev->pdev, ent->map); > > +       } > > +       kfree(ndev->irqp.entries); > > +} > > + > >  static void mlx5_vdpa_free(struct vdpa_device *vdev) > >  { > >         struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev); > > @@ -2848,6 +2939,7 @@ static void mlx5_vdpa_free(struct vdpa_device *vdev) > >                 mlx5_mpfs_del_mac(pfmdev, ndev->config.mac); > >         } > >         mlx5_vdpa_free_resources(&ndev->mvdev); > > +       free_irqs(ndev); > >         kfree(ndev->event_cbs); > >         kfree(ndev->vqs); > >  } > > @@ -2876,9 +2968,23 @@ static struct vdpa_notification_area > > mlx5_get_vq_notification(struct vdpa_device > >         return ret; > >  } > >   > > -static int mlx5_get_vq_irq(struct vdpa_device *vdv, u16 idx) > > +static int mlx5_get_vq_irq(struct vdpa_device *vdev, u16 idx) > >  { > > -       return -EOPNOTSUPP; > > +       struct mlx5_vdpa_dev *mvdev = to_mvdev(vdev); > > +       struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev); > > +       struct mlx5_vdpa_virtqueue *mvq; > > + > > +       if (!is_index_valid(mvdev, idx)) > > +               return -EINVAL; > > + > > +       if (is_ctrl_vq_idx(mvdev, idx)) > > +               return -EOPNOTSUPP; > > + > > +       mvq = &ndev->vqs[idx]; > > +       if (!mvq->map.virq) > > +               return -EOPNOTSUPP; > > + > > +       return mvq->map.virq; > >  } > >   > >  static u64 mlx5_vdpa_get_driver_features(struct vdpa_device *vdev) > > @@ -3155,6 +3261,34 @@ static int config_func_mtu(struct mlx5_core_dev *mdev, > > u16 mtu) > >         return err; > >  } > >   > > +static void allocate_irqs(struct mlx5_vdpa_net *ndev) > > +{ > > +       struct mlx5_vdpa_irq_pool_entry *ent; > > +       int i; > > + > > +       if (!msix_mode_supported(&ndev->mvdev)) > > +               return; > > + > > +       if (!ndev->mvdev.mdev->pdev) > > +               return; > > + > > +       ndev->irqp.entries = kcalloc(ndev->mvdev.max_vqs, sizeof(*ndev- > > >irqp.entries), GFP_KERNEL); > > +       if (!ndev->irqp.entries) > > +               return; > > + > > + > > +       for (i = 0; i < ndev->mvdev.max_vqs; i++) { > > +               ent = ndev->irqp.entries + i; > > +               snprintf(ent->name, MLX5_VDPA_IRQ_NAME_LEN, "%s-vq-%d", > > +                        dev_name(&ndev->mvdev.vdev.dev), i); > > +               ent->map = pci_msix_alloc_irq_at(ndev->mvdev.mdev->pdev, > > MSI_ANY_INDEX, NULL); > > +               if (!ent->map.virq) > > +                       return; > > + > > +               ndev->irqp.num_ent++; > > +       } > > +} > > + > >  static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev *v_mdev, const char *name, > >                              const struct vdpa_dev_set_config *add_config) > >  { > > @@ -3233,6 +3367,7 @@ static int mlx5_vdpa_dev_add(struct vdpa_mgmt_dev > > *v_mdev, const char *name, > >         } > >   > >         init_mvqs(ndev); > > +       allocate_irqs(ndev); > >         init_rwsem(&ndev->reslock); > >         config = &ndev->config; > >   > > @@ -3413,6 +3548,17 @@ static void mlx5v_remove(struct auxiliary_device *adev) > >         kfree(mgtdev); > >  } > >   > > +static void mlx5v_shutdown(struct auxiliary_device *auxdev) > > +{ > > +       struct mlx5_vdpa_mgmtdev *mgtdev; > > +       struct mlx5_vdpa_net *ndev; > > + > > +       mgtdev = auxiliary_get_drvdata(auxdev); > > +       ndev = mgtdev->ndev; > > + > > +       free_irqs(ndev); > > +} > > + > >  static const struct auxiliary_device_id mlx5v_id_table[] = { > >         { .name = MLX5_ADEV_NAME ".vnet", }, > >         {}, > > @@ -3424,6 +3570,7 @@ static struct auxiliary_driver mlx5v_driver = { > >         .name = "vnet", > >         .probe = mlx5v_probe, > >         .remove = mlx5v_remove, > > +       .shutdown = mlx5v_shutdown, > >         .id_table = mlx5v_id_table, > >  }; > >   > > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.h > > b/drivers/vdpa/mlx5/net/mlx5_vnet.h > > index c90a89e1de4d..36c44d9fdd16 100644 > > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.h > > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.h > > @@ -26,6 +26,20 @@ static inline u16 key2vid(u64 key) > >         return (u16)(key >> 48) & 0xfff; > >  } > >   > > +#define MLX5_VDPA_IRQ_NAME_LEN 32 > > + > > +struct mlx5_vdpa_irq_pool_entry { > > +       struct msi_map map; > > +       bool used; > > +       char name[MLX5_VDPA_IRQ_NAME_LEN]; > > +       void *dev_id; > > +}; > > + > > +struct mlx5_vdpa_irq_pool { > > +       int num_ent; > > +       struct mlx5_vdpa_irq_pool_entry *entries; > > +}; > > + > >  struct mlx5_vdpa_net { > >         struct mlx5_vdpa_dev mvdev; > >         struct mlx5_vdpa_net_resources res; > > @@ -49,6 +63,7 @@ struct mlx5_vdpa_net { > >         struct vdpa_callback config_cb; > >         struct mlx5_vdpa_wq_ent cvq_ent; > >         struct hlist_head macvlan_hash[MLX5V_MACVLAN_SIZE]; > > +       struct mlx5_vdpa_irq_pool irqp; > >         struct dentry *debugfs; > >  }; > >   > >