From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by dpdk.org (Postfix) with ESMTP id A47BA1B3A4 for ; Thu, 4 Oct 2018 13:54:08 +0200 (CEST) Received: from smtp.corp.redhat.com (int-mx10.intmail.prod.int.phx2.redhat.com [10.5.11.25]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 03608307D855; Thu, 4 Oct 2018 11:54:08 +0000 (UTC) Received: from [10.36.112.13] (unknown [10.36.112.13]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 9B25A2010D64; Thu, 4 Oct 2018 11:54:03 +0000 (UTC) To: Jens Freimann , dev@dpdk.org Cc: tiwei.bie@intel.com, Gavin.Hu@arm.com References: <20181003131118.21491-1-jfreimann@redhat.com> <20181003131118.21491-2-jfreimann@redhat.com> From: Maxime Coquelin Message-ID: <458553bf-260a-d63d-46e3-69b8c676d73f@redhat.com> Date: Thu, 4 Oct 2018 13:54:00 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 MIME-Version: 1.0 In-Reply-To: <20181003131118.21491-2-jfreimann@redhat.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.84 on 10.5.11.25 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.48]); Thu, 04 Oct 2018 11:54:08 +0000 (UTC) Subject: Re: [dpdk-dev] [PATCH v7 1/8] net/virtio: vring init for packed queues X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 04 Oct 2018 11:54:09 -0000 On 10/03/2018 03:11 PM, Jens Freimann wrote: > Add and initialize descriptor data structures. > > To allow out of order processing a .next field was added to > struct vq_desc_extra because there is none in the packed virtqueue > descriptor itself. This is used to chain descriptors and process them > similiar to how it is handled for split virtqueues. > > Signed-off-by: Jens Freimann > --- > drivers/net/virtio/virtio_ethdev.c | 28 +++++++++------ > drivers/net/virtio/virtio_pci.h | 8 +++++ > drivers/net/virtio/virtio_ring.h | 55 +++++++++++++++++++++++++++--- > drivers/net/virtio/virtqueue.h | 13 ++++++- > 4 files changed, 88 insertions(+), 16 deletions(-) > > diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c > index b81df0a99..d6a1613dd 100644 > --- a/drivers/net/virtio/virtio_ethdev.c > +++ b/drivers/net/virtio/virtio_ethdev.c > @@ -299,19 +299,27 @@ virtio_init_vring(struct virtqueue *vq) > > PMD_INIT_FUNC_TRACE(); > > - /* > - * Reinitialise since virtio port might have been stopped and restarted > - */ > memset(ring_mem, 0, vq->vq_ring_size); > - vring_init(vr, size, ring_mem, VIRTIO_PCI_VRING_ALIGN); > + vring_init(vq->hw, vr, size, ring_mem, VIRTIO_PCI_VRING_ALIGN); > + > + vq->vq_free_cnt = vq->vq_nentries; > + memset(vq->vq_descx, 0, sizeof(struct vq_desc_extra) * vq->vq_nentries); > vq->vq_used_cons_idx = 0; > + vq->vq_avail_idx = 0; > vq->vq_desc_head_idx = 0; > - vq->vq_avail_idx = 0; > vq->vq_desc_tail_idx = (uint16_t)(vq->vq_nentries - 1); > - vq->vq_free_cnt = vq->vq_nentries; > - memset(vq->vq_descx, 0, sizeof(struct vq_desc_extra) * vq->vq_nentries); > + if (vtpci_packed_queue(vq->hw)) { > + uint16_t i; > + for(i = 0; i < size - 1; i++) { > + vq->vq_descx[i].next = i + 1; I would move it in a dedicated loop, and do it only if IN_ORDER hasn't been negotiated. Not for performance reason of course, but just to highlight that this extra stuff isn't needed with in-order. > + vq->vq_ring.desc_packed[i].index = i; I would use the vring_desc_init_packed function declared below instead. > + } Trailing space. > + vq->vq_ring.desc_packed[i].index = i; > + vq->vq_descx[i].next = VQ_RING_DESC_CHAIN_END; > + } else { > > - vring_desc_init(vr->desc, size); > + vring_desc_init_split(vr->desc, size); > + } > > /* > * Disable device(host) interrupting guest > @@ -386,7 +394,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx) > /* > * Reserve a memzone for vring elements > */ > - size = vring_size(vq_size, VIRTIO_PCI_VRING_ALIGN); > + size = vring_size(hw, vq_size, VIRTIO_PCI_VRING_ALIGN); > vq->vq_ring_size = RTE_ALIGN_CEIL(size, VIRTIO_PCI_VRING_ALIGN); > PMD_INIT_LOG(DEBUG, "vring_size: %d, rounded_vring_size: %d", > size, vq->vq_ring_size); > @@ -489,7 +497,7 @@ virtio_init_queue(struct rte_eth_dev *dev, uint16_t vtpci_queue_idx) > for (i = 0; i < vq_size; i++) { > struct vring_desc *start_dp = txr[i].tx_indir; > > - vring_desc_init(start_dp, RTE_DIM(txr[i].tx_indir)); > + vring_desc_init_split(start_dp, RTE_DIM(txr[i].tx_indir)); > > /* first indirect descriptor is always the tx header */ > start_dp->addr = txvq->virtio_net_hdr_mem > diff --git a/drivers/net/virtio/virtio_pci.h b/drivers/net/virtio/virtio_pci.h > index 58fdd3d45..90204d281 100644 > --- a/drivers/net/virtio/virtio_pci.h > +++ b/drivers/net/virtio/virtio_pci.h > @@ -113,6 +113,8 @@ struct virtnet_ctl; > > #define VIRTIO_F_VERSION_1 32 > #define VIRTIO_F_IOMMU_PLATFORM 33 > +#define VIRTIO_F_RING_PACKED 34 > +#define VIRTIO_F_IN_ORDER 35 Isn't that feature already declared? > > /* > * Some VirtIO feature bits (currently bits 28 through 31) are > @@ -314,6 +316,12 @@ vtpci_with_feature(struct virtio_hw *hw, uint64_t bit) > return (hw->guest_features & (1ULL << bit)) != 0; > } > > +static inline int > +vtpci_packed_queue(struct virtio_hw *hw) > +{ > + return vtpci_with_feature(hw, VIRTIO_F_RING_PACKED); > +} > + > /* > * Function declaration from virtio_pci.c > */ > diff --git a/drivers/net/virtio/virtio_ring.h b/drivers/net/virtio/virtio_ring.h > index 9e3c2a015..309069fdb 100644 > --- a/drivers/net/virtio/virtio_ring.h > +++ b/drivers/net/virtio/virtio_ring.h > @@ -54,11 +54,38 @@ struct vring_used { > struct vring_used_elem ring[0]; > }; > > +/* For support of packed virtqueues in Virtio 1.1 the format of descriptors > + * looks like this. > + */ > +struct vring_desc_packed { > + uint64_t addr; > + uint32_t len; > + uint16_t index; > + uint16_t flags; > +} __attribute__ ((aligned(16))); > + > +#define RING_EVENT_FLAGS_ENABLE 0x0 > +#define RING_EVENT_FLAGS_DISABLE 0x1 > +#define RING_EVENT_FLAGS_DESC 0x2 > +struct vring_packed_desc_event { > + uint16_t desc_event_off_wrap; > + uint16_t desc_event_flags; > +}; > + > struct vring { > unsigned int num; > - struct vring_desc *desc; > - struct vring_avail *avail; > - struct vring_used *used; > + union { > + struct vring_desc_packed *desc_packed; > + struct vring_desc *desc; > + }; > + union { > + struct vring_avail *avail; > + struct vring_packed_desc_event *driver_event; > + }; > + union { > + struct vring_used *used; > + struct vring_packed_desc_event *device_event; > + }; > }; > > /* The standard layout for the ring is a continuous chunk of memory which > @@ -95,10 +122,18 @@ struct vring { > #define vring_avail_event(vr) (*(uint16_t *)&(vr)->used->ring[(vr)->num]) > > static inline size_t > -vring_size(unsigned int num, unsigned long align) > +vring_size(struct virtio_hw *hw, unsigned int num, unsigned long align) > { > size_t size; > > + if (vtpci_packed_queue(hw)) { > + size = num * sizeof(struct vring_desc_packed); > + size += sizeof(struct vring_packed_desc_event); > + size = RTE_ALIGN_CEIL(size, align); > + size += sizeof(struct vring_packed_desc_event); > + return size; > + } > + > size = num * sizeof(struct vring_desc); > size += sizeof(struct vring_avail) + (num * sizeof(uint16_t)); > size = RTE_ALIGN_CEIL(size, align); > @@ -108,10 +143,20 @@ vring_size(unsigned int num, unsigned long align) > } > > static inline void > -vring_init(struct vring *vr, unsigned int num, uint8_t *p, > +vring_init(struct virtio_hw *hw, struct vring *vr, unsigned int num, uint8_t *p, > unsigned long align) > { > vr->num = num; > + if (vtpci_packed_queue(hw)) { > + vr->desc_packed = (struct vring_desc_packed *)p; > + vr->driver_event = (struct vring_packed_desc_event *)(p + > + num * sizeof(struct vring_desc_packed)); > + vr->device_event = (struct vring_packed_desc_event *) > + RTE_ALIGN_CEIL((uintptr_t)(vr->driver_event + > + sizeof(struct vring_packed_desc_event)), align); > + return; > + } > + As a general comment, I would find it cleaner to have dedicated functions for split and packed variants, like vring_init_split, vring_init_packed, etc... > vr->desc = (struct vring_desc *) p; > vr->avail = (struct vring_avail *) (p + > num * sizeof(struct vring_desc)); > diff --git a/drivers/net/virtio/virtqueue.h b/drivers/net/virtio/virtqueue.h > index 26518ed98..6a4f92b79 100644 > --- a/drivers/net/virtio/virtqueue.h > +++ b/drivers/net/virtio/virtqueue.h > @@ -161,6 +161,7 @@ struct virtio_pmd_ctrl { > struct vq_desc_extra { > void *cookie; > uint16_t ndescs; > + uint16_t next; > }; > > struct virtqueue { > @@ -245,9 +246,19 @@ struct virtio_tx_region { > __attribute__((__aligned__(16))); > }; > > +static inline void > +vring_desc_init_packed(struct vring *vr, int n) > +{ > + int i; > + for (i = 0; i < n; i++) { > + struct vring_desc_packed *desc = &vr->desc_packed[i]; > + desc->index = i; > + } > +} I see the split variant is also called to init the indirect tables. Do you confirm this isn't needed in the case of packed ring? > + > /* Chain all the descriptors in the ring with an END */ > static inline void > -vring_desc_init(struct vring_desc *dp, uint16_t n) > +vring_desc_init_split(struct vring_desc *dp, uint16_t n) > { > uint16_t i; > >