]> git.proxmox.com Git - mirror_ubuntu-kernels.git/blob - net/xdp/xsk_queue.h
HID: logitech-dj: fix spelling in printk
[mirror_ubuntu-kernels.git] / net / xdp / xsk_queue.h
1 /* SPDX-License-Identifier: GPL-2.0 */
2 /* XDP user-space ring structure
3 * Copyright(c) 2018 Intel Corporation.
4 */
5
6 #ifndef _LINUX_XSK_QUEUE_H
7 #define _LINUX_XSK_QUEUE_H
8
9 #include <linux/types.h>
10 #include <linux/if_xdp.h>
11 #include <net/xdp_sock.h>
12
13 #define RX_BATCH_SIZE 16
14 #define LAZY_UPDATE_THRESHOLD 128
15
16 struct xdp_ring {
17 u32 producer ____cacheline_aligned_in_smp;
18 u32 consumer ____cacheline_aligned_in_smp;
19 };
20
21 /* Used for the RX and TX queues for packets */
22 struct xdp_rxtx_ring {
23 struct xdp_ring ptrs;
24 struct xdp_desc desc[0] ____cacheline_aligned_in_smp;
25 };
26
27 /* Used for the fill and completion queues for buffers */
28 struct xdp_umem_ring {
29 struct xdp_ring ptrs;
30 u64 desc[0] ____cacheline_aligned_in_smp;
31 };
32
33 struct xsk_queue {
34 u64 chunk_mask;
35 u64 size;
36 u32 ring_mask;
37 u32 nentries;
38 u32 prod_head;
39 u32 prod_tail;
40 u32 cons_head;
41 u32 cons_tail;
42 struct xdp_ring *ring;
43 u64 invalid_descs;
44 };
45
46 /* Common functions operating for both RXTX and umem queues */
47
48 static inline u64 xskq_nb_invalid_descs(struct xsk_queue *q)
49 {
50 return q ? q->invalid_descs : 0;
51 }
52
53 static inline u32 xskq_nb_avail(struct xsk_queue *q, u32 dcnt)
54 {
55 u32 entries = q->prod_tail - q->cons_tail;
56
57 if (entries == 0) {
58 /* Refresh the local pointer */
59 q->prod_tail = READ_ONCE(q->ring->producer);
60 entries = q->prod_tail - q->cons_tail;
61 }
62
63 return (entries > dcnt) ? dcnt : entries;
64 }
65
66 static inline u32 xskq_nb_free(struct xsk_queue *q, u32 producer, u32 dcnt)
67 {
68 u32 free_entries = q->nentries - (producer - q->cons_tail);
69
70 if (free_entries >= dcnt)
71 return free_entries;
72
73 /* Refresh the local tail pointer */
74 q->cons_tail = READ_ONCE(q->ring->consumer);
75 return q->nentries - (producer - q->cons_tail);
76 }
77
78 /* UMEM queue */
79
80 static inline bool xskq_is_valid_addr(struct xsk_queue *q, u64 addr)
81 {
82 if (addr >= q->size) {
83 q->invalid_descs++;
84 return false;
85 }
86
87 return true;
88 }
89
90 static inline u64 *xskq_validate_addr(struct xsk_queue *q, u64 *addr)
91 {
92 while (q->cons_tail != q->cons_head) {
93 struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
94 unsigned int idx = q->cons_tail & q->ring_mask;
95
96 *addr = READ_ONCE(ring->desc[idx]) & q->chunk_mask;
97 if (xskq_is_valid_addr(q, *addr))
98 return addr;
99
100 q->cons_tail++;
101 }
102
103 return NULL;
104 }
105
106 static inline u64 *xskq_peek_addr(struct xsk_queue *q, u64 *addr)
107 {
108 if (q->cons_tail == q->cons_head) {
109 WRITE_ONCE(q->ring->consumer, q->cons_tail);
110 q->cons_head = q->cons_tail + xskq_nb_avail(q, RX_BATCH_SIZE);
111
112 /* Order consumer and data */
113 smp_rmb();
114 }
115
116 return xskq_validate_addr(q, addr);
117 }
118
119 static inline void xskq_discard_addr(struct xsk_queue *q)
120 {
121 q->cons_tail++;
122 }
123
124 static inline int xskq_produce_addr(struct xsk_queue *q, u64 addr)
125 {
126 struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
127
128 if (xskq_nb_free(q, q->prod_tail, 1) == 0)
129 return -ENOSPC;
130
131 ring->desc[q->prod_tail++ & q->ring_mask] = addr;
132
133 /* Order producer and data */
134 smp_wmb();
135
136 WRITE_ONCE(q->ring->producer, q->prod_tail);
137 return 0;
138 }
139
140 static inline int xskq_produce_addr_lazy(struct xsk_queue *q, u64 addr)
141 {
142 struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
143
144 if (xskq_nb_free(q, q->prod_head, LAZY_UPDATE_THRESHOLD) == 0)
145 return -ENOSPC;
146
147 ring->desc[q->prod_head++ & q->ring_mask] = addr;
148 return 0;
149 }
150
151 static inline void xskq_produce_flush_addr_n(struct xsk_queue *q,
152 u32 nb_entries)
153 {
154 /* Order producer and data */
155 smp_wmb();
156
157 q->prod_tail += nb_entries;
158 WRITE_ONCE(q->ring->producer, q->prod_tail);
159 }
160
161 static inline int xskq_reserve_addr(struct xsk_queue *q)
162 {
163 if (xskq_nb_free(q, q->prod_head, 1) == 0)
164 return -ENOSPC;
165
166 q->prod_head++;
167 return 0;
168 }
169
170 /* Rx/Tx queue */
171
172 static inline bool xskq_is_valid_desc(struct xsk_queue *q, struct xdp_desc *d)
173 {
174 if (!xskq_is_valid_addr(q, d->addr))
175 return false;
176
177 if (((d->addr + d->len) & q->chunk_mask) !=
178 (d->addr & q->chunk_mask)) {
179 q->invalid_descs++;
180 return false;
181 }
182
183 return true;
184 }
185
186 static inline struct xdp_desc *xskq_validate_desc(struct xsk_queue *q,
187 struct xdp_desc *desc)
188 {
189 while (q->cons_tail != q->cons_head) {
190 struct xdp_rxtx_ring *ring = (struct xdp_rxtx_ring *)q->ring;
191 unsigned int idx = q->cons_tail & q->ring_mask;
192
193 *desc = READ_ONCE(ring->desc[idx]);
194 if (xskq_is_valid_desc(q, desc))
195 return desc;
196
197 q->cons_tail++;
198 }
199
200 return NULL;
201 }
202
203 static inline struct xdp_desc *xskq_peek_desc(struct xsk_queue *q,
204 struct xdp_desc *desc)
205 {
206 if (q->cons_tail == q->cons_head) {
207 WRITE_ONCE(q->ring->consumer, q->cons_tail);
208 q->cons_head = q->cons_tail + xskq_nb_avail(q, RX_BATCH_SIZE);
209
210 /* Order consumer and data */
211 smp_rmb();
212 }
213
214 return xskq_validate_desc(q, desc);
215 }
216
217 static inline void xskq_discard_desc(struct xsk_queue *q)
218 {
219 q->cons_tail++;
220 }
221
222 static inline int xskq_produce_batch_desc(struct xsk_queue *q,
223 u64 addr, u32 len)
224 {
225 struct xdp_rxtx_ring *ring = (struct xdp_rxtx_ring *)q->ring;
226 unsigned int idx;
227
228 if (xskq_nb_free(q, q->prod_head, 1) == 0)
229 return -ENOSPC;
230
231 idx = (q->prod_head++) & q->ring_mask;
232 ring->desc[idx].addr = addr;
233 ring->desc[idx].len = len;
234
235 return 0;
236 }
237
238 static inline void xskq_produce_flush_desc(struct xsk_queue *q)
239 {
240 /* Order producer and data */
241 smp_wmb();
242
243 q->prod_tail = q->prod_head,
244 WRITE_ONCE(q->ring->producer, q->prod_tail);
245 }
246
247 static inline bool xskq_full_desc(struct xsk_queue *q)
248 {
249 return xskq_nb_avail(q, q->nentries) == q->nentries;
250 }
251
252 static inline bool xskq_empty_desc(struct xsk_queue *q)
253 {
254 return xskq_nb_free(q, q->prod_tail, q->nentries) == q->nentries;
255 }
256
257 void xskq_set_umem(struct xsk_queue *q, u64 size, u64 chunk_mask);
258 struct xsk_queue *xskq_create(u32 nentries, bool umem_queue);
259 void xskq_destroy(struct xsk_queue *q_ops);
260
261 /* Executed by the core when the entire UMEM gets freed */
262 void xsk_reuseq_destroy(struct xdp_umem *umem);
263
264 #endif /* _LINUX_XSK_QUEUE_H */