[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_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;
	u64 orig_addr;
	struct list_head free_list_node;
};

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

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;
	bool dma_need_sync;
};

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;
	struct list_head free_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;
	u8 cached_need_wakeup;
	bool uses_need_wakeup;
	bool dma_need_sync;
	bool unaligned;
	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->orig_addr = 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);
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);
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;
}

void xp_dma_sync_for_cpu_slow(struct xdp_buff_xsk *xskb);
static inline void xp_dma_sync_for_cpu(struct xdp_buff_xsk *xskb)
{
	xp_dma_sync_for_cpu_slow(xskb);
}

void xp_dma_sync_for_device_slow(struct xsk_buff_pool *pool, dma_addr_t dma,
				 size_t size);
static inline void xp_dma_sync_for_device(struct xsk_buff_pool *pool,
					  dma_addr_t dma, size_t size)
{
	if (!pool->dma_need_sync)
		return;

	xp_dma_sync_for_device_slow(pool, dma, size);
}

/* 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_cnt &&
	       !(pool->dma_pages[addr >> PAGE_SHIFT] & XSK_NEXT_PG_CONTIG_MASK);
}

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)
{
	u64 offset = xskb->xdp.data - xskb->xdp.data_hard_start;

	offset += xskb->pool->headroom;
	if (!xskb->pool->unaligned)
		return xskb->orig_addr + offset;
	return xskb->orig_addr + (offset << XSK_UNALIGNED_BUF_OFFSET_SHIFT);
}

#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;
	15	struct xsk_queue;
	16	struct xdp_desc;
1742b3d5	17	struct xdp_umem;
1c1efc2a	18	struct xdp_sock;
2b43470a BT	19	struct device;
	20	struct page;
	21
94ecc5ca THJ	22	#define XSK_PRIV_MAX 24
94ecc5ca THJ	23
2b43470a BT	24	struct xdp_buff_xsk {
2b43470a BT	25	struct xdp_buff xdp;
94ecc5ca	26	u8 cb[XSK_PRIV_MAX];
2b43470a BT	27	dma_addr_t dma;
	28	dma_addr_t frame_dma;
	29	struct xsk_buff_pool *pool;
2b43470a BT	30	u64 orig_addr;
	31	struct list_head free_list_node;
	32	};
	33
94ecc5ca THJ	34	#define XSK_CHECK_PRIV_TYPE(t) BUILD_BUG_ON(sizeof(t) > offsetofend(struct xdp_buff_xsk, cb))
94ecc5ca THJ	35
921b6869 MK	36	struct xsk_dma_map {
	37	dma_addr_t *dma_pages;
	38	struct device *dev;
	39	struct net_device *netdev;
	40	refcount_t users;
	41	struct list_head list; /* Protected by the RTNL_LOCK */
	42	u32 dma_pages_cnt;
	43	bool dma_need_sync;
	44	};
	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;
26062b18	56	struct list_head free_list;
8ef4e27e MK	57	u32 heads_cnt;
	58	u16 queue_id;
	59
	60	/* Data path members as close to free_heads at the end as possible. */
	61	struct xsk_queue *fq ____cacheline_aligned_in_smp;
	62	struct xsk_queue *cq;
921b6869 MK	63	/* For performance reasons, each buff pool has its own array of dma_pages
	64	* even when they are identical.
	65	*/
26062b18 BT	66	dma_addr_t *dma_pages;
26062b18 BT	67	struct xdp_buff_xsk *heads;
d1bc532e	68	struct xdp_desc *tx_descs;
26062b18 BT	69	u64 chunk_mask;
	70	u64 addrs_cnt;
	71	u32 free_list_cnt;
	72	u32 dma_pages_cnt;
26062b18 BT	73	u32 free_heads_cnt;
	74	u32 headroom;
	75	u32 chunk_size;
94033cd8	76	u32 chunk_shift;
26062b18	77	u32 frame_len;
c2d3d6a4 MK	78	u8 cached_need_wakeup;
c2d3d6a4 MK	79	bool uses_need_wakeup;
91d5b702	80	bool dma_need_sync;
26062b18 BT	81	bool unaligned;
26062b18 BT	82	void *addrs;
f09ced40 MK	83	/* Mutual exclusion of the completion ring in the SKB mode. Two cases to protect:
	84	* NAPI TX thread and sendmsg error paths in the SKB destructor callback and when
	85	* sockets share a single cq when the same netdev and queue id is shared.
	86	*/
	87	spinlock_t cq_lock;
26062b18 BT	88	struct xdp_buff_xsk *free_heads[];
	89	};
	90
94033cd8 MK	91	/* Masks for xdp_umem_page flags.
	92	* The low 12-bits of the addr will be 0 since this is the page address, so we
	93	* can use them for flags.
	94	*/
	95	#define XSK_NEXT_PG_CONTIG_SHIFT 0
	96	#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)
	97
2b43470a	98	/* AF_XDP core. */
1c1efc2a MK	99	struct xsk_buff_pool xp_create_and_assign_umem(struct xdp_sock xs,
	100	struct xdp_umem *umem);
	101	int xp_assign_dev(struct xsk_buff_pool pool, struct net_device dev,
	102	u16 queue_id, u16 flags);
60240bc2	103	int xp_assign_dev_shared(struct xsk_buff_pool pool, struct xdp_sock umem_xs,
b5aea28d	104	struct net_device *dev, u16 queue_id);
ba3beec2	105	int xp_alloc_tx_descs(struct xsk_buff_pool pool, struct xdp_sock xs);
2b43470a	106	void xp_destroy(struct xsk_buff_pool *pool);
1c1efc2a	107	void xp_get_pool(struct xsk_buff_pool *pool);
e5e1a4bc	108	bool xp_put_pool(struct xsk_buff_pool *pool);
1c1efc2a	109	void xp_clear_dev(struct xsk_buff_pool *pool);
a5aa8e52 MK	110	void xp_add_xsk(struct xsk_buff_pool pool, struct xdp_sock xs);
a5aa8e52 MK	111	void xp_del_xsk(struct xsk_buff_pool pool, struct xdp_sock xs);
2b43470a BT	112
	113	/* AF_XDP, and XDP core. */
	114	void xp_free(struct xdp_buff_xsk *xskb);
	115
94033cd8 MK	116	static inline void xp_init_xskb_addr(struct xdp_buff_xsk xskb, struct xsk_buff_pool pool,
	117	u64 addr)
	118	{
	119	xskb->orig_addr = addr;
	120	xskb->xdp.data_hard_start = pool->addrs + addr + pool->headroom;
	121	}
	122
	123	static inline void xp_init_xskb_dma(struct xdp_buff_xsk xskb, struct xsk_buff_pool pool,
	124	dma_addr_t *dma_pages, u64 addr)
	125	{
	126	xskb->frame_dma = (dma_pages[addr >> PAGE_SHIFT] & ~XSK_NEXT_PG_CONTIG_MASK) +
	127	(addr & ~PAGE_MASK);
	128	xskb->dma = xskb->frame_dma + pool->headroom + XDP_PACKET_HEADROOM;
	129	}
	130
2b43470a BT	131	/* AF_XDP ZC drivers, via xdp_sock_buff.h */
	132	void xp_set_rxq_info(struct xsk_buff_pool pool, struct xdp_rxq_info rxq);
	133	int xp_dma_map(struct xsk_buff_pool pool, struct device dev,
	134	unsigned long attrs, struct page **pages, u32 nr_pages);
	135	void xp_dma_unmap(struct xsk_buff_pool *pool, unsigned long attrs);
	136	struct xdp_buff xp_alloc(struct xsk_buff_pool pool);
47e4075d	137	u32 xp_alloc_batch(struct xsk_buff_pool pool, struct xdp_buff *xdp, u32 max);
2b43470a BT	138	bool xp_can_alloc(struct xsk_buff_pool *pool, u32 count);
	139	void xp_raw_get_data(struct xsk_buff_pool pool, u64 addr);
	140	dma_addr_t xp_raw_get_dma(struct xsk_buff_pool *pool, u64 addr);
26062b18 BT	141	static inline dma_addr_t xp_get_dma(struct xdp_buff_xsk *xskb)
	142	{
	143	return xskb->dma;
	144	}
	145
	146	static inline dma_addr_t xp_get_frame_dma(struct xdp_buff_xsk *xskb)
	147	{
	148	return xskb->frame_dma;
	149	}
	150
	151	void xp_dma_sync_for_cpu_slow(struct xdp_buff_xsk *xskb);
	152	static inline void xp_dma_sync_for_cpu(struct xdp_buff_xsk *xskb)
	153	{
26062b18 BT	154	xp_dma_sync_for_cpu_slow(xskb);
	155	}
	156
	157	void xp_dma_sync_for_device_slow(struct xsk_buff_pool *pool, dma_addr_t dma,
	158	size_t size);
	159	static inline void xp_dma_sync_for_device(struct xsk_buff_pool *pool,
	160	dma_addr_t dma, size_t size)
	161	{
91d5b702	162	if (!pool->dma_need_sync)
26062b18 BT	163	return;
	164
	165	xp_dma_sync_for_device_slow(pool, dma, size);
	166	}
	167
	168	/* Masks for xdp_umem_page flags.
	169	* The low 12-bits of the addr will be 0 since this is the page address, so we
	170	* can use them for flags.
	171	*/
	172	#define XSK_NEXT_PG_CONTIG_SHIFT 0
	173	#define XSK_NEXT_PG_CONTIG_MASK BIT_ULL(XSK_NEXT_PG_CONTIG_SHIFT)
	174
	175	static inline bool xp_desc_crosses_non_contig_pg(struct xsk_buff_pool *pool,
	176	u64 addr, u32 len)
	177	{
	178	bool cross_pg = (addr & (PAGE_SIZE - 1)) + len > PAGE_SIZE;
	179
2f996198 MK	180	if (likely(!cross_pg))
	181	return false;
	182
d769ccaf KC	183	return pool->dma_pages_cnt &&
d769ccaf KC	184	!(pool->dma_pages[addr >> PAGE_SHIFT] & XSK_NEXT_PG_CONTIG_MASK);
26062b18 BT	185	}
	186
	187	static inline u64 xp_aligned_extract_addr(struct xsk_buff_pool *pool, u64 addr)
	188	{
	189	return addr & pool->chunk_mask;
	190	}
	191
	192	static inline u64 xp_unaligned_extract_addr(u64 addr)
	193	{
	194	return addr & XSK_UNALIGNED_BUF_ADDR_MASK;
	195	}
	196
	197	static inline u64 xp_unaligned_extract_offset(u64 addr)
	198	{
	199	return addr >> XSK_UNALIGNED_BUF_OFFSET_SHIFT;
	200	}
	201
	202	static inline u64 xp_unaligned_add_offset_to_addr(u64 addr)
	203	{
	204	return xp_unaligned_extract_addr(addr) +
	205	xp_unaligned_extract_offset(addr);
	206	}
2b43470a	207
94033cd8 MK	208	static inline u32 xp_aligned_extract_idx(struct xsk_buff_pool *pool, u64 addr)
	209	{
	210	return xp_aligned_extract_addr(pool, addr) >> pool->chunk_shift;
	211	}
	212
	213	static inline void xp_release(struct xdp_buff_xsk *xskb)
	214	{
	215	if (xskb->pool->unaligned)
	216	xskb->pool->free_heads[xskb->pool->free_heads_cnt++] = xskb;
	217	}
	218
	219	static inline u64 xp_get_handle(struct xdp_buff_xsk *xskb)
	220	{
	221	u64 offset = xskb->xdp.data - xskb->xdp.data_hard_start;
	222
	223	offset += xskb->pool->headroom;
	224	if (!xskb->pool->unaligned)
	225	return xskb->orig_addr + offset;
	226	return xskb->orig_addr + (offset << XSK_UNALIGNED_BUF_OFFSET_SHIFT);
	227	}
	228
2b43470a	229	#endif /* XSK_BUFF_POOL_H_ */