[PATCH net-next v3 0/4] net: Avoid the memory waste in some Ethernet drivers

[PATCH net-next v3 0/4] net: Avoid the memory waste in some Ethernet drivers

[Kevin Hao]

Hi,

v3:
  - Adjust patch 1 and 2 according to Alexander's suggestion.
  - Add Tested-by from Subbaraya.
  - Add Reviewed-by from Ioana.

v2:
  - Inline page_frag_alloc() and {netdev,napi}_alloc_frag()
  - Adopt Vlastimil's suggestion and add his Acked-by

In the current implementation of napi_alloc_frag(), it doesn't have any
align guarantee for the returned buffer address. We would have to use
some ugly workarounds to make sure that we can get a align buffer
address for some Ethernet drivers. This patch series tries to introduce
some helper functions to make sure that an align buffer is returned.
Then we can drop the ugly workarounds and avoid the unnecessary memory
waste.

Kevin Hao (4):
  mm: page_frag: Introduce page_frag_alloc_align()
  net: Introduce {netdev,napi}_alloc_frag_align()
  net: octeontx2: Use napi_alloc_frag_align() to avoid the memory waste
  net: dpaa2: Use napi_alloc_frag_align() to avoid the memory waste

 .../net/ethernet/freescale/dpaa2/dpaa2-eth.c  |  3 +-
 .../marvell/octeontx2/nic/otx2_common.c       |  3 +-
 include/linux/gfp.h                           | 12 +++++--
 include/linux/skbuff.h                        | 36 +++++++++++++++++--
 mm/page_alloc.c                               |  8 +++--
 net/core/skbuff.c                             | 26 ++++++--------
 6 files changed, 61 insertions(+), 27 deletions(-)

-- 
2.29.2

[PATCH net-next v3 1/4] mm: page_frag: Introduce page_frag_alloc_align()

[Kevin Hao]

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 <haokexin@gmail.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
---
v3: Use align mask as suggested by Alexander.

 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 53caa9846854..52cd415b436c 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 ad3ed3ec4dd5..3583c6accd88 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 @@ void *page_frag_alloc(struct page_frag_cache *nc,
 	}
 
 	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.
-- 
2.29.2

Re: [PATCH net-next v3 1/4] mm: page_frag: Introduce page_frag_alloc_align()

[Alexander Duyck]

On Thu, Feb 4, 2021 at 3:06 AM Kevin Hao <haokexin@gmail.com> wrote:
>
> 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 <haokexin@gmail.com>
> Acked-by: Vlastimil Babka <vbabka@suse.cz>
> ---
> v3: Use align mask as suggested by Alexander.
>
>  include/linux/gfp.h | 12 ++++++++++--
>  mm/page_alloc.c     |  8 +++++---
>  2 files changed, 15 insertions(+), 5 deletions(-)

Looks good to me.

Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>

[PATCH net-next v3 2/4] net: Introduce {netdev,napi}_alloc_frag_align()

[Kevin Hao]

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 <haokexin@gmail.com>
---
v3: Use align mask and refactor the {netdev,napi}_alloc_frag_align() as
    suggested by Alexander.

 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 9313b5aaf45b..c875b36c43fc 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 2af12f7e170c..063b365ce5b2 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
-- 
2.29.2

Re: [PATCH net-next v3 2/4] net: Introduce {netdev,napi}_alloc_frag_align()

[Alexander Duyck]

On Thu, Feb 4, 2021 at 3:06 AM Kevin Hao <haokexin@gmail.com> wrote:
>
> 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 <haokexin@gmail.com>
> ---
> v3: Use align mask and refactor the {netdev,napi}_alloc_frag_align() as
>     suggested by Alexander.
>
>  include/linux/skbuff.h | 36 ++++++++++++++++++++++++++++++++++--
>  net/core/skbuff.c      | 26 ++++++++++----------------
>  2 files changed, 44 insertions(+), 18 deletions(-)

Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>

[PATCH net-next v3 3/4] net: octeontx2: Use napi_alloc_frag_align() to avoid the memory waste

[Kevin Hao]

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 <haokexin@gmail.com>
Tested-by: Subbaraya Sundeep <sbhatta@marvell.com>
---
v3: Add Tested-by from Subbaraya.

 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))) {
-- 
2.29.2

[PATCH net-next v3 4/4] net: dpaa2: Use napi_alloc_frag_align() to avoid the memory waste

[Kevin Hao]

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 <haokexin@gmail.com>
Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com>
---
v3: Add Reviewed-by from Ioana.

 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 41e225baf571..882b32a04f5e 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);
-- 
2.29.2

Re: [PATCH net-next v3 0/4] net: Avoid the memory waste in some Ethernet drivers

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (refs/heads/master):

On Thu,  4 Feb 2021 18:56:34 +0800 you wrote:
> Hi,
> 
> v3:
>   - Adjust patch 1 and 2 according to Alexander's suggestion.
>   - Add Tested-by from Subbaraya.
>   - Add Reviewed-by from Ioana.
> 
> [...]

Here is the summary with links:
  - [net-next,v3,1/4] mm: page_frag: Introduce page_frag_alloc_align()
    https://git.kernel.org/netdev/net-next/c/b358e2122b9d
  - [net-next,v3,2/4] net: Introduce {netdev,napi}_alloc_frag_align()
    https://git.kernel.org/netdev/net-next/c/3f6e687dff39
  - [net-next,v3,3/4] net: octeontx2: Use napi_alloc_frag_align() to avoid the memory waste
    https://git.kernel.org/netdev/net-next/c/1b041601c798
  - [net-next,v3,4/4] net: dpaa2: Use napi_alloc_frag_align() to avoid the memory waste
    https://git.kernel.org/netdev/net-next/c/d0dfbb9912d9

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html