[linux-block.git] / include / net / xsk_buff_pool.h

/* SPDX-License-Identifier: GPL-2.0 */
/* Copyright(c) 2020 Intel Corporation. */

#ifndef XSK_BUFF_POOL_H_
#define XSK_BUFF_POOL_H_

#include <linux/if_xdp.h>
#include <linux/types.h>
#include <linux/dma-mapping.h>
#include <linux/bpf.h>
#include <net/xdp.h>

struct xsk_buff_pool;
struct xdp_rxq_info;
struct xsk_cb_desc;
struct xsk_queue;
struct xdp_desc;
struct xdp_umem;
struct xdp_sock;
struct device;
struct page;

#define XSK_PRIV_MAX 24

struct xdp_buff_xsk {
	struct xdp_buff xdp;
	u8 cb[XSK_PRIV_MAX];
	dma_addr_t dma;
	dma_addr_t frame_dma;
	struct xsk_buff_pool *pool;
	struct list_head list_node;
} __aligned_largest;

#define XSK_CHECK_PRIV_TYPE(t) BUILD_BUG_ON(sizeof(t) > offsetofend(struct xdp_buff_xsk, cb))
#define XSK_TX_COMPL_FITS(t) BUILD_BUG_ON(sizeof(struct xsk_tx_metadata_compl) > sizeof(t))

struct xsk_dma_map {
	dma_addr_t *dma_pages;
	struct device *dev;
	struct net_device *netdev;
	refcount_t users;
	struct list_head list; /* Protected by the RTNL_LOCK */
	u32 dma_pages_cnt;
};

struct xsk_buff_pool {
	/* Members only used in the control path first. */
	struct device *dev;
	struct net_device *netdev;
	struct list_head xsk_tx_list;
	/* Protects modifications to the xsk_tx_list */
	spinlock_t xsk_tx_list_lock;
	refcount_t users;
	struct xdp_umem *umem;
	struct work_struct work;
	/* Protects generic receive in shared and non-shared umem mode. */
	spinlock_t rx_lock;
	struct list_head free_list;
	struct list_head xskb_list;
	u32 heads_cnt;
	u16 queue_id;

	/* Data path members as close to free_heads at the end as possible. */
	struct xsk_queue *fq ____cacheline_aligned_in_smp;
	struct xsk_queue *cq;
	/* For performance reasons, each buff pool has its own array of dma_pages
	 * even when they are identical.
	 */
	dma_addr_t *dma_pages;
	struct xdp_buff_xsk *heads;
	struct xdp_desc *tx_descs;
	u64 chunk_mask;
	u64 addrs_cnt;
	u32 free_list_cnt;
	u32 dma_pages_cnt;
	u32 free_heads_cnt;
	u32 headroom;
	u32 chunk_size;
	u32 chunk_shift;
	u32 frame_len;
	u32 xdp_zc_max_segs;
	u8 tx_metadata_len; /* inherited from umem */
	u8 cached_need_wakeup;
	bool uses_need_wakeup;
	bool unaligned;
	bool tx_sw_csum;
	void *addrs;
	/* Mutual exclusion of the completion ring in the SKB mode. Two cases to protect:
	 * NAPI TX thread and sendmsg error paths in the SKB destructor callback and when
	 * sockets share a single cq when the same netdev and queue id is shared.
	 */
	spinlock_t cq_lock;
	struct xdp_buff_xsk *free_heads[];
};

/* Masks for xdp_umem_page flags.
 * The low 12-bits of the addr will be 0 since this is the page address, so we
 * can use them for flags.
 */
#define XSK_NEXT_PG_CONTIG_SHIFT 0
#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)

/* AF_XDP core. */
struct xsk_buff_pool *xp_create_and_assign_umem(struct xdp_sock *xs,
						struct xdp_umem *umem);
int xp_assign_dev(struct xsk_buff_pool *pool, struct net_device *dev,
		  u16 queue_id, u16 flags);
int xp_assign_dev_shared(struct xsk_buff_pool *pool, struct xdp_sock *umem_xs,
			 struct net_device *dev, u16 queue_id);
int xp_alloc_tx_descs(struct xsk_buff_pool *pool, struct xdp_sock *xs);
void xp_destroy(struct xsk_buff_pool *pool);
void xp_get_pool(struct xsk_buff_pool *pool);
bool xp_put_pool(struct xsk_buff_pool *pool);
void xp_clear_dev(struct xsk_buff_pool *pool);
void xp_add_xsk(struct xsk_buff_pool *pool, struct xdp_sock *xs);
void xp_del_xsk(struct xsk_buff_pool *pool, struct xdp_sock *xs);

/* AF_XDP, and XDP core. */
void xp_free(struct xdp_buff_xsk *xskb);

static inline void xp_init_xskb_addr(struct xdp_buff_xsk *xskb, struct xsk_buff_pool *pool,
				     u64 addr)
{
	xskb->xdp.data_hard_start = pool->addrs + addr + pool->headroom;
}

static inline void xp_init_xskb_dma(struct xdp_buff_xsk *xskb, struct xsk_buff_pool *pool,
				    dma_addr_t *dma_pages, u64 addr)
{
	xskb->frame_dma = (dma_pages[addr >> PAGE_SHIFT] & ~XSK_NEXT_PG_CONTIG_MASK) +
		(addr & ~PAGE_MASK);
	xskb->dma = xskb->frame_dma + pool->headroom + XDP_PACKET_HEADROOM;
}

/* AF_XDP ZC drivers, via xdp_sock_buff.h */
void xp_set_rxq_info(struct xsk_buff_pool *pool, struct xdp_rxq_info *rxq);
void xp_fill_cb(struct xsk_buff_pool *pool, struct xsk_cb_desc *desc);
int xp_dma_map(struct xsk_buff_pool *pool, struct device *dev,
	       unsigned long attrs, struct page **pages, u32 nr_pages);
void xp_dma_unmap(struct xsk_buff_pool *pool, unsigned long attrs);
struct xdp_buff *xp_alloc(struct xsk_buff_pool *pool);
u32 xp_alloc_batch(struct xsk_buff_pool *pool, struct xdp_buff **xdp, u32 max);
bool xp_can_alloc(struct xsk_buff_pool *pool, u32 count);
void *xp_raw_get_data(struct xsk_buff_pool *pool, u64 addr);
dma_addr_t xp_raw_get_dma(struct xsk_buff_pool *pool, u64 addr);

struct xdp_desc_ctx {
	dma_addr_t dma;
	struct xsk_tx_metadata *meta;
};

struct xdp_desc_ctx xp_raw_get_ctx(const struct xsk_buff_pool *pool, u64 addr);

static inline dma_addr_t xp_get_dma(struct xdp_buff_xsk *xskb)
{
	return xskb->dma;
}

static inline dma_addr_t xp_get_frame_dma(struct xdp_buff_xsk *xskb)
{
	return xskb->frame_dma;
}

static inline void xp_dma_sync_for_cpu(struct xdp_buff_xsk *xskb)
{
	dma_sync_single_for_cpu(xskb->pool->dev, xskb->dma,
				xskb->pool->frame_len,
				DMA_BIDIRECTIONAL);
}

static inline void xp_dma_sync_for_device(struct xsk_buff_pool *pool,
					  dma_addr_t dma, size_t size)
{
	dma_sync_single_for_device(pool->dev, dma, size, DMA_BIDIRECTIONAL);
}

/* Masks for xdp_umem_page flags.
 * The low 12-bits of the addr will be 0 since this is the page address, so we
 * can use them for flags.
 */
#define XSK_NEXT_PG_CONTIG_SHIFT 0
#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)

static inline bool xp_desc_crosses_non_contig_pg(struct xsk_buff_pool *pool,
						 u64 addr, u32 len)
{
	bool cross_pg = (addr & (PAGE_SIZE - 1)) + len > PAGE_SIZE;

	if (likely(!cross_pg))
		return false;

	return pool->dma_pages &&
	       !(pool->dma_pages[addr >> PAGE_SHIFT] & XSK_NEXT_PG_CONTIG_MASK);
}

static inline bool xp_mb_desc(const struct xdp_desc *desc)
{
	return desc->options & XDP_PKT_CONTD;
}

static inline u64 xp_aligned_extract_addr(struct xsk_buff_pool *pool, u64 addr)
{
	return addr & pool->chunk_mask;
}

static inline u64 xp_unaligned_extract_addr(u64 addr)
{
	return addr & XSK_UNALIGNED_BUF_ADDR_MASK;
}

static inline u64 xp_unaligned_extract_offset(u64 addr)
{
	return addr >> XSK_UNALIGNED_BUF_OFFSET_SHIFT;
}

static inline u64 xp_unaligned_add_offset_to_addr(u64 addr)
{
	return xp_unaligned_extract_addr(addr) +
		xp_unaligned_extract_offset(addr);
}

static inline u32 xp_aligned_extract_idx(struct xsk_buff_pool *pool, u64 addr)
{
	return xp_aligned_extract_addr(pool, addr) >> pool->chunk_shift;
}

static inline void xp_release(struct xdp_buff_xsk *xskb)
{
	if (xskb->pool->unaligned)
		xskb->pool->free_heads[xskb->pool->free_heads_cnt++] = xskb;
}

static inline u64 xp_get_handle(struct xdp_buff_xsk *xskb,
				struct xsk_buff_pool *pool)
{
	u64 orig_addr = xskb->xdp.data - pool->addrs;
	u64 offset;

	if (!pool->unaligned)
		return orig_addr;

	offset = xskb->xdp.data - xskb->xdp.data_hard_start;
	offset += pool->headroom;
	orig_addr -= offset;
	return orig_addr + (offset << XSK_UNALIGNED_BUF_OFFSET_SHIFT);
}

static inline bool xp_tx_metadata_enabled(const struct xsk_buff_pool *pool)
{
	return pool->tx_metadata_len > 0;
}

#endif /* XSK_BUFF_POOL_H_ */
Commit	Line	Data
2b43470a BT	1	/* SPDX-License-Identifier: GPL-2.0 */
	2	/* Copyright(c) 2020 Intel Corporation. */
	3
	4	#ifndef XSK_BUFF_POOL_H_
	5	#define XSK_BUFF_POOL_H_
	6
26062b18	7	#include <linux/if_xdp.h>
2b43470a BT	8	#include <linux/types.h>
2b43470a BT	9	#include <linux/dma-mapping.h>
94033cd8	10	#include <linux/bpf.h>
2b43470a BT	11	#include <net/xdp.h>
	12
	13	struct xsk_buff_pool;
	14	struct xdp_rxq_info;
b4e352ff	15	struct xsk_cb_desc;
2b43470a BT	16	struct xsk_queue;
2b43470a BT	17	struct xdp_desc;
1742b3d5	18	struct xdp_umem;
1c1efc2a	19	struct xdp_sock;
2b43470a BT	20	struct device;
	21	struct page;
	22
94ecc5ca THJ	23	#define XSK_PRIV_MAX 24
94ecc5ca THJ	24
2b43470a BT	25	struct xdp_buff_xsk {
2b43470a BT	26	struct xdp_buff xdp;
94ecc5ca	27	u8 cb[XSK_PRIV_MAX];
2b43470a BT	28	dma_addr_t dma;
	29	dma_addr_t frame_dma;
	30	struct xsk_buff_pool *pool;
30ec2c1b	31	struct list_head list_node;
ca5c9494	32	} __aligned_largest;
2b43470a	33
94ecc5ca	34	#define XSK_CHECK_PRIV_TYPE(t) BUILD_BUG_ON(sizeof(t) > offsetofend(struct xdp_buff_xsk, cb))
48eb03dd	35	#define XSK_TX_COMPL_FITS(t) BUILD_BUG_ON(sizeof(struct xsk_tx_metadata_compl) > sizeof(t))
94ecc5ca	36
921b6869 MK	37	struct xsk_dma_map {
	38	dma_addr_t *dma_pages;
	39	struct device *dev;
	40	struct net_device *netdev;
	41	refcount_t users;
	42	struct list_head list; /* Protected by the RTNL_LOCK */
	43	u32 dma_pages_cnt;
921b6869 MK	44	};
921b6869 MK	45
26062b18	46	struct xsk_buff_pool {
8ef4e27e MK	47	/* Members only used in the control path first. */
	48	struct device *dev;
	49	struct net_device *netdev;
	50	struct list_head xsk_tx_list;
	51	/* Protects modifications to the xsk_tx_list */
	52	spinlock_t xsk_tx_list_lock;
	53	refcount_t users;
	54	struct xdp_umem *umem;
	55	struct work_struct work;
a1356ac7	56	/* Protects generic receive in shared and non-shared umem mode. */
a1356ac7	57	spinlock_t rx_lock;
26062b18	58	struct list_head free_list;
24ea5012	59	struct list_head xskb_list;
8ef4e27e MK	60	u32 heads_cnt;
	61	u16 queue_id;
	62
	63	/* Data path members as close to free_heads at the end as possible. */
	64	struct xsk_queue *fq ____cacheline_aligned_in_smp;
	65	struct xsk_queue *cq;
921b6869 MK	66	/* For performance reasons, each buff pool has its own array of dma_pages
	67	* even when they are identical.
	68	*/
26062b18 BT	69	dma_addr_t *dma_pages;
26062b18 BT	70	struct xdp_buff_xsk *heads;
d1bc532e	71	struct xdp_desc *tx_descs;
26062b18 BT	72	u64 chunk_mask;
	73	u64 addrs_cnt;
	74	u32 free_list_cnt;
	75	u32 dma_pages_cnt;
26062b18 BT	76	u32 free_heads_cnt;
	77	u32 headroom;
	78	u32 chunk_size;
94033cd8	79	u32 chunk_shift;
26062b18	80	u32 frame_len;
6e126872	81	u32 xdp_zc_max_segs;
341ac980	82	u8 tx_metadata_len; /* inherited from umem */
c2d3d6a4 MK	83	u8 cached_need_wakeup;
c2d3d6a4 MK	84	bool uses_need_wakeup;
26062b18	85	bool unaligned;
11614723	86	bool tx_sw_csum;
26062b18	87	void *addrs;
f09ced40 MK	88	/* Mutual exclusion of the completion ring in the SKB mode. Two cases to protect:
	89	* NAPI TX thread and sendmsg error paths in the SKB destructor callback and when
	90	* sockets share a single cq when the same netdev and queue id is shared.
	91	*/
	92	spinlock_t cq_lock;
26062b18 BT	93	struct xdp_buff_xsk *free_heads[];
	94	};
	95
94033cd8 MK	96	/* Masks for xdp_umem_page flags.
	97	* The low 12-bits of the addr will be 0 since this is the page address, so we
	98	* can use them for flags.
	99	*/
	100	#define XSK_NEXT_PG_CONTIG_SHIFT 0
	101	#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)
	102
2b43470a	103	/* AF_XDP core. */
1c1efc2a MK	104	struct xsk_buff_pool xp_create_and_assign_umem(struct xdp_sock xs,
	105	struct xdp_umem *umem);
	106	int xp_assign_dev(struct xsk_buff_pool pool, struct net_device dev,
	107	u16 queue_id, u16 flags);
60240bc2	108	int xp_assign_dev_shared(struct xsk_buff_pool pool, struct xdp_sock umem_xs,
b5aea28d	109	struct net_device *dev, u16 queue_id);
ba3beec2	110	int xp_alloc_tx_descs(struct xsk_buff_pool pool, struct xdp_sock xs);
2b43470a	111	void xp_destroy(struct xsk_buff_pool *pool);
1c1efc2a	112	void xp_get_pool(struct xsk_buff_pool *pool);
e5e1a4bc	113	bool xp_put_pool(struct xsk_buff_pool *pool);
1c1efc2a	114	void xp_clear_dev(struct xsk_buff_pool *pool);
a5aa8e52 MK	115	void xp_add_xsk(struct xsk_buff_pool pool, struct xdp_sock xs);
a5aa8e52 MK	116	void xp_del_xsk(struct xsk_buff_pool pool, struct xdp_sock xs);
2b43470a BT	117
	118	/* AF_XDP, and XDP core. */
	119	void xp_free(struct xdp_buff_xsk *xskb);
	120
94033cd8 MK	121	static inline void xp_init_xskb_addr(struct xdp_buff_xsk xskb, struct xsk_buff_pool pool,
	122	u64 addr)
	123	{
94033cd8 MK	124	xskb->xdp.data_hard_start = pool->addrs + addr + pool->headroom;
	125	}
	126
	127	static inline void xp_init_xskb_dma(struct xdp_buff_xsk xskb, struct xsk_buff_pool pool,
	128	dma_addr_t *dma_pages, u64 addr)
	129	{
	130	xskb->frame_dma = (dma_pages[addr >> PAGE_SHIFT] & ~XSK_NEXT_PG_CONTIG_MASK) +
	131	(addr & ~PAGE_MASK);
	132	xskb->dma = xskb->frame_dma + pool->headroom + XDP_PACKET_HEADROOM;
	133	}
	134
2b43470a BT	135	/* AF_XDP ZC drivers, via xdp_sock_buff.h */
2b43470a BT	136	void xp_set_rxq_info(struct xsk_buff_pool pool, struct xdp_rxq_info rxq);
b4e352ff	137	void xp_fill_cb(struct xsk_buff_pool pool, struct xsk_cb_desc desc);
2b43470a BT	138	int xp_dma_map(struct xsk_buff_pool pool, struct device dev,
	139	unsigned long attrs, struct page **pages, u32 nr_pages);
	140	void xp_dma_unmap(struct xsk_buff_pool *pool, unsigned long attrs);
	141	struct xdp_buff xp_alloc(struct xsk_buff_pool pool);
47e4075d	142	u32 xp_alloc_batch(struct xsk_buff_pool pool, struct xdp_buff *xdp, u32 max);
2b43470a BT	143	bool xp_can_alloc(struct xsk_buff_pool *pool, u32 count);
	144	void xp_raw_get_data(struct xsk_buff_pool pool, u64 addr);
	145	dma_addr_t xp_raw_get_dma(struct xsk_buff_pool *pool, u64 addr);
23d9324a AL	146
	147	struct xdp_desc_ctx {
	148	dma_addr_t dma;
	149	struct xsk_tx_metadata *meta;
	150	};
	151
	152	struct xdp_desc_ctx xp_raw_get_ctx(const struct xsk_buff_pool *pool, u64 addr);
	153
26062b18 BT	154	static inline dma_addr_t xp_get_dma(struct xdp_buff_xsk *xskb)
	155	{
	156	return xskb->dma;
	157	}
	158
	159	static inline dma_addr_t xp_get_frame_dma(struct xdp_buff_xsk *xskb)
	160	{
	161	return xskb->frame_dma;
	162	}
	163
26062b18 BT	164	static inline void xp_dma_sync_for_cpu(struct xdp_buff_xsk *xskb)
26062b18 BT	165	{
163943ac AL	166	dma_sync_single_for_cpu(xskb->pool->dev, xskb->dma,
	167	xskb->pool->frame_len,
	168	DMA_BIDIRECTIONAL);
26062b18 BT	169	}
26062b18 BT	170
26062b18 BT	171	static inline void xp_dma_sync_for_device(struct xsk_buff_pool *pool,
	172	dma_addr_t dma, size_t size)
	173	{
163943ac	174	dma_sync_single_for_device(pool->dev, dma, size, DMA_BIDIRECTIONAL);
26062b18 BT	175	}
	176
	177	/* Masks for xdp_umem_page flags.
	178	* The low 12-bits of the addr will be 0 since this is the page address, so we
	179	* can use them for flags.
	180	*/
	181	#define XSK_NEXT_PG_CONTIG_SHIFT 0
	182	#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)
	183
	184	static inline bool xp_desc_crosses_non_contig_pg(struct xsk_buff_pool *pool,
	185	u64 addr, u32 len)
	186	{
	187	bool cross_pg = (addr & (PAGE_SIZE - 1)) + len > PAGE_SIZE;
	188
2f996198 MK	189	if (likely(!cross_pg))
	190	return false;
	191
6ec7be9a	192	return pool->dma_pages &&
d769ccaf	193	!(pool->dma_pages[addr >> PAGE_SHIFT] & XSK_NEXT_PG_CONTIG_MASK);
26062b18 BT	194	}
26062b18 BT	195
dcf3827c	196	static inline bool xp_mb_desc(const struct xdp_desc *desc)
1b725b0c MF	197	{
	198	return desc->options & XDP_PKT_CONTD;
	199	}
	200
26062b18 BT	201	static inline u64 xp_aligned_extract_addr(struct xsk_buff_pool *pool, u64 addr)
	202	{
	203	return addr & pool->chunk_mask;
	204	}
	205
	206	static inline u64 xp_unaligned_extract_addr(u64 addr)
	207	{
	208	return addr & XSK_UNALIGNED_BUF_ADDR_MASK;
	209	}
	210
	211	static inline u64 xp_unaligned_extract_offset(u64 addr)
	212	{
	213	return addr >> XSK_UNALIGNED_BUF_OFFSET_SHIFT;
	214	}
	215
	216	static inline u64 xp_unaligned_add_offset_to_addr(u64 addr)
	217	{
	218	return xp_unaligned_extract_addr(addr) +
	219	xp_unaligned_extract_offset(addr);
	220	}
2b43470a	221
94033cd8 MK	222	static inline u32 xp_aligned_extract_idx(struct xsk_buff_pool *pool, u64 addr)
	223	{
	224	return xp_aligned_extract_addr(pool, addr) >> pool->chunk_shift;
	225	}
	226
	227	static inline void xp_release(struct xdp_buff_xsk *xskb)
	228	{
	229	if (xskb->pool->unaligned)
	230	xskb->pool->free_heads[xskb->pool->free_heads_cnt++] = xskb;
	231	}
	232
bea14124 MF	233	static inline u64 xp_get_handle(struct xdp_buff_xsk *xskb,
bea14124 MF	234	struct xsk_buff_pool *pool)
94033cd8	235	{
bea14124 MF	236	u64 orig_addr = xskb->xdp.data - pool->addrs;
bea14124 MF	237	u64 offset;
94033cd8	238
bea14124 MF	239	if (!pool->unaligned)
	240	return orig_addr;
	241
	242	offset = xskb->xdp.data - xskb->xdp.data_hard_start;
bea14124	243	offset += pool->headroom;
bf20af07	244	orig_addr -= offset;
bea14124	245	return orig_addr + (offset << XSK_UNALIGNED_BUF_OFFSET_SHIFT);
94033cd8 MK	246	}
94033cd8 MK	247
48eb03dd SF	248	static inline bool xp_tx_metadata_enabled(const struct xsk_buff_pool *pool)
	249	{
	250	return pool->tx_metadata_len > 0;
	251	}
	252
2b43470a	253	#endif /* XSK_BUFF_POOL_H_ */