From patchwork Thu Oct 28 10:49:17 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Xuan Zhuo X-Patchwork-Id: 12589827 X-Patchwork-Delegate: kuba@kernel.org 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 mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 57014C433EF for ; Thu, 28 Oct 2021 10:49:24 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 3821361106 for ; Thu, 28 Oct 2021 10:49:24 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230135AbhJ1Kvt (ORCPT ); Thu, 28 Oct 2021 06:51:49 -0400 Received: from out30-43.freemail.mail.aliyun.com ([115.124.30.43]:34140 "EHLO out30-43.freemail.mail.aliyun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230077AbhJ1Kvs (ORCPT ); Thu, 28 Oct 2021 06:51:48 -0400 X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R201e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=e01e01424;MF=xuanzhuo@linux.alibaba.com;NM=1;PH=DS;RN=6;SR=0;TI=SMTPD_---0Uu-KRUZ_1635418159; Received: from localhost(mailfrom:xuanzhuo@linux.alibaba.com fp:SMTPD_---0Uu-KRUZ_1635418159) by smtp.aliyun-inc.com(127.0.0.1); Thu, 28 Oct 2021 18:49:20 +0800 From: Xuan Zhuo To: virtualization@lists.linux-foundation.org, netdev@vger.kernel.org Cc: "Michael S. Tsirkin" , Jason Wang , "David S. Miller" , Jakub Kicinski Subject: [PATCH v2 1/3] virtio: cache indirect desc for split Date: Thu, 28 Oct 2021 18:49:17 +0800 Message-Id: <20211028104919.3393-2-xuanzhuo@linux.alibaba.com> X-Mailer: git-send-email 2.31.0 In-Reply-To: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> References: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org In the case of using indirect, indirect desc must be allocated and released each time, which increases a lot of cpu overhead. Here, a cache is added for indirect. If the number of indirect desc to be applied for is less than VIRT_QUEUE_CACHE_DESC_NUM, the desc array with the size of VIRT_QUEUE_CACHE_DESC_NUM is fixed and cached for reuse. Signed-off-by: Xuan Zhuo --- drivers/virtio/virtio.c | 6 +++ drivers/virtio/virtio_ring.c | 77 ++++++++++++++++++++++++++++++++---- include/linux/virtio.h | 14 +++++++ 3 files changed, 89 insertions(+), 8 deletions(-) diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c index 0a5b54034d4b..1047149ac2a4 100644 --- a/drivers/virtio/virtio.c +++ b/drivers/virtio/virtio.c @@ -431,6 +431,12 @@ bool is_virtio_device(struct device *dev) } EXPORT_SYMBOL_GPL(is_virtio_device); +void virtio_set_desc_cache(struct virtio_device *dev, u32 thr) +{ + dev->desc_cache_thr = thr; +} +EXPORT_SYMBOL_GPL(virtio_set_desc_cache); + void unregister_virtio_device(struct virtio_device *dev) { int index = dev->index; /* save for after device release */ diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index dd95dfd85e98..0ebcd4f12d3b 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -117,6 +117,15 @@ struct vring_virtqueue { /* Hint for event idx: already triggered no need to disable. */ bool event_triggered; + /* desc cache threshold + * 0 - disable desc cache + * > 0 - enable desc cache. As the threshold of the desc cache. + */ + u32 desc_cache_thr; + + /* desc cache chain */ + struct list_head desc_cache; + union { /* Available for split ring */ struct { @@ -423,7 +432,53 @@ static unsigned int vring_unmap_one_split(const struct vring_virtqueue *vq, return extra[i].next; } -static struct vring_desc *alloc_indirect_split(struct virtqueue *_vq, +static void desc_cache_free(struct list_head *head) +{ + struct list_head *n, *pos; + + BUILD_BUG_ON(sizeof(struct list_head) > sizeof(struct vring_desc)); + BUILD_BUG_ON(sizeof(struct list_head) > sizeof(struct vring_packed_desc)); + + list_for_each_prev_safe(pos, n, head) + kfree(pos); +} + +static void __desc_cache_put(struct vring_virtqueue *vq, + struct list_head *node, int n) +{ + if (n <= vq->desc_cache_thr) + list_add(node, &vq->desc_cache); + else + kfree(node); +} + +#define desc_cache_put(vq, desc, n) \ + __desc_cache_put(vq, (struct list_head *)desc, n) + +static void *desc_cache_get(struct vring_virtqueue *vq, + int size, int n, gfp_t gfp) +{ + struct list_head *node; + + if (n > vq->desc_cache_thr) + return kmalloc_array(n, size, gfp); + + if (!list_empty(&vq->desc_cache)) { + node = vq->desc_cache.next; + list_del(node); + return node; + } + + return kmalloc_array(vq->desc_cache_thr, size, gfp); +} + +#define _desc_cache_get(vq, n, gfp, tp) \ + ((tp *)desc_cache_get(vq, (sizeof(tp)), n, gfp)) + +#define desc_cache_get_split(vq, n, gfp) \ + _desc_cache_get(vq, n, gfp, struct vring_desc) + +static struct vring_desc *alloc_indirect_split(struct vring_virtqueue *vq, unsigned int total_sg, gfp_t gfp) { @@ -437,12 +492,12 @@ static struct vring_desc *alloc_indirect_split(struct virtqueue *_vq, */ gfp &= ~__GFP_HIGHMEM; - desc = kmalloc_array(total_sg, sizeof(struct vring_desc), gfp); + desc = desc_cache_get_split(vq, total_sg, gfp); if (!desc) return NULL; for (i = 0; i < total_sg; i++) - desc[i].next = cpu_to_virtio16(_vq->vdev, i + 1); + desc[i].next = cpu_to_virtio16(vq->vq.vdev, i + 1); return desc; } @@ -508,7 +563,7 @@ static inline int virtqueue_add_split(struct virtqueue *_vq, head = vq->free_head; if (virtqueue_use_indirect(_vq, total_sg)) - desc = alloc_indirect_split(_vq, total_sg, gfp); + desc = alloc_indirect_split(vq, total_sg, gfp); else { desc = NULL; WARN_ON_ONCE(total_sg > vq->split.vring.num && !vq->indirect); @@ -652,7 +707,7 @@ static inline int virtqueue_add_split(struct virtqueue *_vq, } if (indirect) - kfree(desc); + desc_cache_put(vq, desc, total_sg); END_USE(vq); return -ENOMEM; @@ -717,7 +772,7 @@ static void detach_buf_split(struct vring_virtqueue *vq, unsigned int head, if (vq->indirect) { struct vring_desc *indir_desc = vq->split.desc_state[head].indir_desc; - u32 len; + u32 len, n; /* Free the indirect table, if any, now that it's unmapped. */ if (!indir_desc) @@ -729,10 +784,12 @@ static void detach_buf_split(struct vring_virtqueue *vq, unsigned int head, VRING_DESC_F_INDIRECT)); BUG_ON(len == 0 || len % sizeof(struct vring_desc)); - for (j = 0; j < len / sizeof(struct vring_desc); j++) + n = len / sizeof(struct vring_desc); + + for (j = 0; j < n; j++) vring_unmap_one_split_indirect(vq, &indir_desc[j]); - kfree(indir_desc); + desc_cache_put(vq, indir_desc, n); vq->split.desc_state[head].indir_desc = NULL; } else if (ctx) { *ctx = vq->split.desc_state[head].indir_desc; @@ -2199,6 +2256,9 @@ struct virtqueue *__vring_new_virtqueue(unsigned int index, vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC) && !context; vq->event = virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX); + vq->desc_cache_thr = vdev->desc_cache_thr; + + INIT_LIST_HEAD(&vq->desc_cache); if (virtio_has_feature(vdev, VIRTIO_F_ORDER_PLATFORM)) vq->weak_barriers = false; @@ -2329,6 +2389,7 @@ void vring_del_virtqueue(struct virtqueue *_vq) if (!vq->packed_ring) { kfree(vq->split.desc_state); kfree(vq->split.desc_extra); + desc_cache_free(&vq->desc_cache); } kfree(vq); } diff --git a/include/linux/virtio.h b/include/linux/virtio.h index 41edbc01ffa4..bda6f9853e97 100644 --- a/include/linux/virtio.h +++ b/include/linux/virtio.h @@ -118,6 +118,7 @@ struct virtio_device { struct list_head vqs; u64 features; void *priv; + u32 desc_cache_thr; }; static inline struct virtio_device *dev_to_virtio(struct device *_dev) @@ -130,6 +131,19 @@ int register_virtio_device(struct virtio_device *dev); void unregister_virtio_device(struct virtio_device *dev); bool is_virtio_device(struct device *dev); +/** + * virtio_set_desc_cache - set virtio ring desc cache threshold + * + * virtio will cache the allocated indirect desc. + * + * This function must be called before find_vqs. + * + * @thr: + * 0 - disable desc cache + * > 0 - enable desc cache. As the threshold of the desc cache. + */ +void virtio_set_desc_cache(struct virtio_device *dev, u32 thr); + void virtio_break_device(struct virtio_device *dev); void virtio_config_changed(struct virtio_device *dev); From patchwork Thu Oct 28 10:49:18 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Xuan Zhuo X-Patchwork-Id: 12589829 X-Patchwork-Delegate: kuba@kernel.org 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 mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id D8269C433FE for ; Thu, 28 Oct 2021 10:49:24 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id C035F61106 for ; Thu, 28 Oct 2021 10:49:24 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230153AbhJ1Kvu (ORCPT ); Thu, 28 Oct 2021 06:51:50 -0400 Received: from out30-133.freemail.mail.aliyun.com ([115.124.30.133]:42025 "EHLO out30-133.freemail.mail.aliyun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230126AbhJ1Kvt (ORCPT ); Thu, 28 Oct 2021 06:51:49 -0400 X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R641e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=e01e04423;MF=xuanzhuo@linux.alibaba.com;NM=1;PH=DS;RN=6;SR=0;TI=SMTPD_---0Uu.SeFY_1635418160; Received: from localhost(mailfrom:xuanzhuo@linux.alibaba.com fp:SMTPD_---0Uu.SeFY_1635418160) by smtp.aliyun-inc.com(127.0.0.1); Thu, 28 Oct 2021 18:49:20 +0800 From: Xuan Zhuo To: virtualization@lists.linux-foundation.org, netdev@vger.kernel.org Cc: "Michael S. Tsirkin" , Jason Wang , "David S. Miller" , Jakub Kicinski Subject: [PATCH v2 2/3] virtio: cache indirect desc for packed Date: Thu, 28 Oct 2021 18:49:18 +0800 Message-Id: <20211028104919.3393-3-xuanzhuo@linux.alibaba.com> X-Mailer: git-send-email 2.31.0 In-Reply-To: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> References: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org In the case of using indirect, indirect desc must be allocated and released each time, which increases a lot of cpu overhead. Here, a cache is added for indirect. If the number of indirect desc to be applied for is less than VIRT_QUEUE_CACHE_DESC_NUM, the desc array with the size of VIRT_QUEUE_CACHE_DESC_NUM is fixed and cached for reuse. Signed-off-by: Xuan Zhuo --- drivers/virtio/virtio_ring.c | 29 +++++++++++++++++++---------- 1 file changed, 19 insertions(+), 10 deletions(-) diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 0ebcd4f12d3b..e6d1985a87a8 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1089,7 +1089,11 @@ static void vring_unmap_desc_packed(const struct vring_virtqueue *vq, } } -static struct vring_packed_desc *alloc_indirect_packed(unsigned int total_sg, +#define desc_cache_get_packed(vq, n, gfp) \ + _desc_cache_get(vq, n, gfp, struct vring_packed_desc) + +static struct vring_packed_desc *alloc_indirect_packed(struct vring_virtqueue *vq, + unsigned int total_sg, gfp_t gfp) { struct vring_packed_desc *desc; @@ -1101,7 +1105,7 @@ static struct vring_packed_desc *alloc_indirect_packed(unsigned int total_sg, */ gfp &= ~__GFP_HIGHMEM; - desc = kmalloc_array(total_sg, sizeof(struct vring_packed_desc), gfp); + desc = desc_cache_get_packed(vq, total_sg, gfp); return desc; } @@ -1121,7 +1125,7 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq, dma_addr_t addr; head = vq->packed.next_avail_idx; - desc = alloc_indirect_packed(total_sg, gfp); + desc = alloc_indirect_packed(vq, total_sg, gfp); if (unlikely(vq->vq.num_free < 1)) { pr_debug("Can't add buf len 1 - avail = 0\n"); @@ -1212,7 +1216,7 @@ static int virtqueue_add_indirect_packed(struct vring_virtqueue *vq, for (i = 0; i < err_idx; i++) vring_unmap_desc_packed(vq, &desc[i]); - kfree(desc); + desc_cache_put(vq, desc, total_sg); END_USE(vq); return -ENOMEM; @@ -1437,20 +1441,22 @@ static void detach_buf_packed(struct vring_virtqueue *vq, } if (vq->indirect) { - u32 len; + u32 len, n; /* Free the indirect table, if any, now that it's unmapped. */ desc = state->indir_desc; if (!desc) return; + len = vq->packed.desc_extra[id].len; + n = len / sizeof(struct vring_packed_desc); + if (vq->use_dma_api) { - len = vq->packed.desc_extra[id].len; - for (i = 0; i < len / sizeof(struct vring_packed_desc); - i++) + for (i = 0; i < n; i++) vring_unmap_desc_packed(vq, &desc[i]); } - kfree(desc); + + desc_cache_put(vq, desc, n); state->indir_desc = NULL; } else if (ctx) { *ctx = state->indir_desc; @@ -1768,6 +1774,9 @@ static struct virtqueue *vring_create_virtqueue_packed( vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC) && !context; vq->event = virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX); + vq->desc_cache_thr = vdev->desc_cache_thr; + + INIT_LIST_HEAD(&vq->desc_cache); if (virtio_has_feature(vdev, VIRTIO_F_ORDER_PLATFORM)) vq->weak_barriers = false; @@ -2389,8 +2398,8 @@ void vring_del_virtqueue(struct virtqueue *_vq) if (!vq->packed_ring) { kfree(vq->split.desc_state); kfree(vq->split.desc_extra); - desc_cache_free(&vq->desc_cache); } + desc_cache_free(&vq->desc_cache); kfree(vq); } EXPORT_SYMBOL_GPL(vring_del_virtqueue); From patchwork Thu Oct 28 10:49:19 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Xuan Zhuo X-Patchwork-Id: 12589831 X-Patchwork-Delegate: kuba@kernel.org 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 mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 6CB91C433F5 for ; Thu, 28 Oct 2021 10:49:25 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 552FB610CB for ; Thu, 28 Oct 2021 10:49:25 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230155AbhJ1Kvv (ORCPT ); Thu, 28 Oct 2021 06:51:51 -0400 Received: from out30-132.freemail.mail.aliyun.com ([115.124.30.132]:44079 "EHLO out30-132.freemail.mail.aliyun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230124AbhJ1Kvt (ORCPT ); Thu, 28 Oct 2021 06:51:49 -0400 X-Alimail-AntiSpam: AC=PASS;BC=-1|-1;BR=01201311R181e4;CH=green;DM=||false|;DS=||;FP=0|-1|-1|-1|0|-1|-1|-1;HT=e01e04426;MF=xuanzhuo@linux.alibaba.com;NM=1;PH=DS;RN=6;SR=0;TI=SMTPD_---0Uu-eygn_1635418160; Received: from localhost(mailfrom:xuanzhuo@linux.alibaba.com fp:SMTPD_---0Uu-eygn_1635418160) by smtp.aliyun-inc.com(127.0.0.1); Thu, 28 Oct 2021 18:49:21 +0800 From: Xuan Zhuo To: virtualization@lists.linux-foundation.org, netdev@vger.kernel.org Cc: "Michael S. Tsirkin" , Jason Wang , "David S. Miller" , Jakub Kicinski Subject: [PATCH v2 3/3] virtio-net: enable virtio indirect cache Date: Thu, 28 Oct 2021 18:49:19 +0800 Message-Id: <20211028104919.3393-4-xuanzhuo@linux.alibaba.com> X-Mailer: git-send-email 2.31.0 In-Reply-To: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> References: <20211028104919.3393-1-xuanzhuo@linux.alibaba.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org If the VIRTIO_RING_F_INDIRECT_DESC negotiation succeeds, and the number of sgs used for sending packets is greater than 1. We must constantly call __kmalloc/kfree to allocate/release desc. In the case of extremely fast package delivery, the overhead cannot be ignored: 27.46% [kernel] [k] virtqueue_add 16.66% [kernel] [k] detach_buf_split 16.51% [kernel] [k] virtnet_xsk_xmit 14.04% [kernel] [k] virtqueue_add_outbuf 5.18% [kernel] [k] __kmalloc 4.08% [kernel] [k] kfree 2.80% [kernel] [k] virtqueue_get_buf_ctx 2.22% [kernel] [k] xsk_tx_peek_desc 2.08% [kernel] [k] memset_erms 0.83% [kernel] [k] virtqueue_kick_prepare 0.76% [kernel] [k] virtnet_xsk_run 0.62% [kernel] [k] __free_old_xmit_ptr 0.60% [kernel] [k] vring_map_one_sg 0.53% [kernel] [k] native_apic_mem_write 0.46% [kernel] [k] sg_next 0.43% [kernel] [k] sg_init_table 0.41% [kernel] [k] kmalloc_slab Compared to not using virtio indirect cache, virtio-net can get a 16% performance improvement when using indirect desc cache. In the test case, the CPU where the package is sent has reached 100%. The following are the PPS in two cases: indirect desc cache | no cache 3074658 | 2685132 3111866 | 2666118 3152527 | 2653632 3125867 | 2669820 3027147 | 2644464 3069211 | 2669777 3038522 | 2675645 3034507 | 2671302 3102257 | 2685504 3083712 | 2692800 3051771 | 2676928 3080684 | 2695040 3147816 | 2720876 3123887 | 2705492 3180963 | 2699520 3191579 | 2676480 3161670 | 2686272 3189768 | 2692588 3174272 | 2686692 3143434 | 2682416 Signed-off-by: Xuan Zhuo Reported-by: kernel test robot --- drivers/net/virtio_net.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index 4ad25a8b0870..e1ade176ab46 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c @@ -31,6 +31,13 @@ module_param(csum, bool, 0444); module_param(gso, bool, 0444); module_param(napi_tx, bool, 0644); +/** + * Because virtio desc cache will increase memory overhead, users can turn it + * off or select an acceptable value. The maximum value is 2 + MAX_SKB_FRAGS. + */ +static u32 virtio_desc_cache_thr = 4; +module_param(virtio_desc_cache_thr, uint, 0644); + /* FIXME: MTU in config. */ #define GOOD_PACKET_LEN (ETH_HLEN + VLAN_HLEN + ETH_DATA_LEN) #define GOOD_COPY_LEN 128 @@ -3214,6 +3221,11 @@ static int virtnet_probe(struct virtio_device *vdev) vi->curr_queue_pairs = num_online_cpus(); vi->max_queue_pairs = max_queue_pairs; + if (virtio_desc_cache_thr > 2 + MAX_SKB_FRAGS) + virtio_set_desc_cache(vdev, 2 + MAX_SKB_FRAGS); + else + virtio_set_desc_cache(vdev, virtio_desc_cache_thr); + /* Allocate/initialize the rx/tx queues, and invoke find_vqs */ err = init_vqs(vi); if (err)