From b358e2122b9d7aa99f681d4edfafd999845d16ff Mon Sep 17 00:00:00 2001 From: Kevin Hao Date: Thu, 4 Feb 2021 18:56:35 +0800 Subject: [PATCH 1/4] mm: page_frag: Introduce page_frag_alloc_align() In the current implementation of page_frag_alloc(), it doesn't have any align guarantee for the returned buffer address. But for some hardwares they do require the DMA buffer to be aligned correctly, so we would have to use some workarounds like below if the buffers allocated by the page_frag_alloc() are used by these hardwares for DMA. buf = page_frag_alloc(really_needed_size + align); buf = PTR_ALIGN(buf, align); These codes seems ugly and would waste a lot of memories if the buffers are used in a network driver for the TX/RX. So introduce page_frag_alloc_align() to make sure that an aligned buffer address is returned. Signed-off-by: Kevin Hao Acked-by: Vlastimil Babka Reviewed-by: Alexander Duyck Signed-off-by: Jakub Kicinski --- include/linux/gfp.h | 12 ++++++++++-- mm/page_alloc.c | 8 +++++--- 2 files changed, 15 insertions(+), 5 deletions(-) diff --git a/include/linux/gfp.h b/include/linux/gfp.h index 6e479e9c48ce..80544d5c08e7 100644 --- a/include/linux/gfp.h +++ b/include/linux/gfp.h @@ -583,8 +583,16 @@ extern void free_pages(unsigned long addr, unsigned int order); struct page_frag_cache; extern void __page_frag_cache_drain(struct page *page, unsigned int count); -extern void *page_frag_alloc(struct page_frag_cache *nc, - unsigned int fragsz, gfp_t gfp_mask); +extern void *page_frag_alloc_align(struct page_frag_cache *nc, + unsigned int fragsz, gfp_t gfp_mask, + unsigned int align_mask); + +static inline void *page_frag_alloc(struct page_frag_cache *nc, + unsigned int fragsz, gfp_t gfp_mask) +{ + return page_frag_alloc_align(nc, fragsz, gfp_mask, ~0u); +} + extern void page_frag_free(void *addr); #define __free_page(page) __free_pages((page), 0) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 519a60d5b6f7..ef5070fed76b 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -5137,8 +5137,9 @@ void __page_frag_cache_drain(struct page *page, unsigned int count) } EXPORT_SYMBOL(__page_frag_cache_drain); -void *page_frag_alloc(struct page_frag_cache *nc, - unsigned int fragsz, gfp_t gfp_mask) +void *page_frag_alloc_align(struct page_frag_cache *nc, + unsigned int fragsz, gfp_t gfp_mask, + unsigned int align_mask) { unsigned int size = PAGE_SIZE; struct page *page; @@ -5190,11 +5191,12 @@ refill: } nc->pagecnt_bias--; + offset &= align_mask; nc->offset = offset; return nc->va + offset; } -EXPORT_SYMBOL(page_frag_alloc); +EXPORT_SYMBOL(page_frag_alloc_align); /* * Frees a page fragment allocated out of either a compound or order 0 page. From 3f6e687dff395da43b056c18150a423bc7bf5d14 Mon Sep 17 00:00:00 2001 From: Kevin Hao Date: Thu, 4 Feb 2021 18:56:36 +0800 Subject: [PATCH 2/4] net: Introduce {netdev,napi}_alloc_frag_align() In the current implementation of {netdev,napi}_alloc_frag(), it doesn't have any align guarantee for the returned buffer address, But for some hardwares they do require the DMA buffer to be aligned correctly, so we would have to use some workarounds like below if the buffers allocated by the {netdev,napi}_alloc_frag() are used by these hardwares for DMA. buf = napi_alloc_frag(really_needed_size + align); buf = PTR_ALIGN(buf, align); These codes seems ugly and would waste a lot of memories if the buffers are used in a network driver for the TX/RX. We have added the align support for the page_frag functions, so add the corresponding {netdev,napi}_frag functions. Signed-off-by: Kevin Hao Reviewed-by: Alexander Duyck Signed-off-by: Jakub Kicinski --- include/linux/skbuff.h | 36 ++++++++++++++++++++++++++++++++++-- net/core/skbuff.c | 26 ++++++++++---------------- 2 files changed, 44 insertions(+), 18 deletions(-) diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 0e42c53b8ca9..0a4e91a2f873 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -2818,7 +2818,26 @@ void skb_queue_purge(struct sk_buff_head *list); unsigned int skb_rbtree_purge(struct rb_root *root); -void *netdev_alloc_frag(unsigned int fragsz); +void *__netdev_alloc_frag_align(unsigned int fragsz, unsigned int align_mask); + +/** + * netdev_alloc_frag - allocate a page fragment + * @fragsz: fragment size + * + * Allocates a frag from a page for receive buffer. + * Uses GFP_ATOMIC allocations. + */ +static inline void *netdev_alloc_frag(unsigned int fragsz) +{ + return __netdev_alloc_frag_align(fragsz, ~0u); +} + +static inline void *netdev_alloc_frag_align(unsigned int fragsz, + unsigned int align) +{ + WARN_ON_ONCE(!is_power_of_2(align)); + return __netdev_alloc_frag_align(fragsz, -align); +} struct sk_buff *__netdev_alloc_skb(struct net_device *dev, unsigned int length, gfp_t gfp_mask); @@ -2877,7 +2896,20 @@ static inline void skb_free_frag(void *addr) page_frag_free(addr); } -void *napi_alloc_frag(unsigned int fragsz); +void *__napi_alloc_frag_align(unsigned int fragsz, unsigned int align_mask); + +static inline void *napi_alloc_frag(unsigned int fragsz) +{ + return __napi_alloc_frag_align(fragsz, ~0u); +} + +static inline void *napi_alloc_frag_align(unsigned int fragsz, + unsigned int align) +{ + WARN_ON_ONCE(!is_power_of_2(align)); + return __napi_alloc_frag_align(fragsz, -align); +} + struct sk_buff *__napi_alloc_skb(struct napi_struct *napi, unsigned int length, gfp_t gfp_mask); static inline struct sk_buff *napi_alloc_skb(struct napi_struct *napi, diff --git a/net/core/skbuff.c b/net/core/skbuff.c index 3787093239f5..d380c7b5a12d 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -374,29 +374,23 @@ struct napi_alloc_cache { static DEFINE_PER_CPU(struct page_frag_cache, netdev_alloc_cache); static DEFINE_PER_CPU(struct napi_alloc_cache, napi_alloc_cache); -static void *__napi_alloc_frag(unsigned int fragsz, gfp_t gfp_mask) +static void *__alloc_frag_align(unsigned int fragsz, gfp_t gfp_mask, + unsigned int align_mask) { struct napi_alloc_cache *nc = this_cpu_ptr(&napi_alloc_cache); - return page_frag_alloc(&nc->page, fragsz, gfp_mask); + return page_frag_alloc_align(&nc->page, fragsz, gfp_mask, align_mask); } -void *napi_alloc_frag(unsigned int fragsz) +void *__napi_alloc_frag_align(unsigned int fragsz, unsigned int align_mask) { fragsz = SKB_DATA_ALIGN(fragsz); - return __napi_alloc_frag(fragsz, GFP_ATOMIC); + return __alloc_frag_align(fragsz, GFP_ATOMIC, align_mask); } -EXPORT_SYMBOL(napi_alloc_frag); +EXPORT_SYMBOL(__napi_alloc_frag_align); -/** - * netdev_alloc_frag - allocate a page fragment - * @fragsz: fragment size - * - * Allocates a frag from a page for receive buffer. - * Uses GFP_ATOMIC allocations. - */ -void *netdev_alloc_frag(unsigned int fragsz) +void *__netdev_alloc_frag_align(unsigned int fragsz, unsigned int align_mask) { struct page_frag_cache *nc; void *data; @@ -404,15 +398,15 @@ void *netdev_alloc_frag(unsigned int fragsz) fragsz = SKB_DATA_ALIGN(fragsz); if (in_irq() || irqs_disabled()) { nc = this_cpu_ptr(&netdev_alloc_cache); - data = page_frag_alloc(nc, fragsz, GFP_ATOMIC); + data = page_frag_alloc_align(nc, fragsz, GFP_ATOMIC, align_mask); } else { local_bh_disable(); - data = __napi_alloc_frag(fragsz, GFP_ATOMIC); + data = __alloc_frag_align(fragsz, GFP_ATOMIC, align_mask); local_bh_enable(); } return data; } -EXPORT_SYMBOL(netdev_alloc_frag); +EXPORT_SYMBOL(__netdev_alloc_frag_align); /** * __netdev_alloc_skb - allocate an skbuff for rx on a specific device From 1b041601c798a1a6bb3a651ce17aefd41979a1e2 Mon Sep 17 00:00:00 2001 From: Kevin Hao Date: Thu, 4 Feb 2021 18:56:37 +0800 Subject: [PATCH 3/4] net: octeontx2: Use napi_alloc_frag_align() to avoid the memory waste The napi_alloc_frag_align() will guarantee that a correctly align buffer address is returned. So use this function to simplify the buffer alloc and avoid the unnecessary memory waste. Signed-off-by: Kevin Hao Tested-by: Subbaraya Sundeep Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c index 5ddedc3b754d..cbd68fa9f1d6 100644 --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c @@ -488,11 +488,10 @@ dma_addr_t __otx2_alloc_rbuf(struct otx2_nic *pfvf, struct otx2_pool *pool) dma_addr_t iova; u8 *buf; - buf = napi_alloc_frag(pool->rbsize + OTX2_ALIGN); + buf = napi_alloc_frag_align(pool->rbsize, OTX2_ALIGN); if (unlikely(!buf)) return -ENOMEM; - buf = PTR_ALIGN(buf, OTX2_ALIGN); iova = dma_map_single_attrs(pfvf->dev, buf, pool->rbsize, DMA_FROM_DEVICE, DMA_ATTR_SKIP_CPU_SYNC); if (unlikely(dma_mapping_error(pfvf->dev, iova))) { From d0dfbb9912d9477578f41c5200d7eac3da899dce Mon Sep 17 00:00:00 2001 From: Kevin Hao Date: Thu, 4 Feb 2021 18:56:38 +0800 Subject: [PATCH 4/4] net: dpaa2: Use napi_alloc_frag_align() to avoid the memory waste The napi_alloc_frag_align() will guarantee that a correctly align buffer address is returned. So use this function to simplify the buffer alloc and avoid the unnecessary memory waste. Signed-off-by: Kevin Hao Reviewed-by: Ioana Ciornei Signed-off-by: Jakub Kicinski --- drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c index c5ceebca8cae..19f74d4cbb4e 100644 --- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c +++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c @@ -764,12 +764,11 @@ static int dpaa2_eth_build_sg_fd(struct dpaa2_eth_priv *priv, /* Prepare the HW SGT structure */ sgt_buf_size = priv->tx_data_offset + sizeof(struct dpaa2_sg_entry) * num_dma_bufs; - sgt_buf = napi_alloc_frag(sgt_buf_size + DPAA2_ETH_TX_BUF_ALIGN); + sgt_buf = napi_alloc_frag_align(sgt_buf_size, DPAA2_ETH_TX_BUF_ALIGN); if (unlikely(!sgt_buf)) { err = -ENOMEM; goto sgt_buf_alloc_failed; } - sgt_buf = PTR_ALIGN(sgt_buf, DPAA2_ETH_TX_BUF_ALIGN); memset(sgt_buf, 0, sgt_buf_size); sgt = (struct dpaa2_sg_entry *)(sgt_buf + priv->tx_data_offset);