1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
|
/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */
/*
* Copyright 2015-2020 Amazon.com, Inc. or its affiliates. All rights reserved.
*/
#ifndef ENA_ETH_COM_H_
#define ENA_ETH_COM_H_
#include "ena_com.h"
/* head update threshold in units of (queue size / ENA_COMP_HEAD_THRESH) */
#define ENA_COMP_HEAD_THRESH 4
struct ena_com_tx_ctx {
struct ena_com_tx_meta ena_meta;
struct ena_com_buf *ena_bufs;
/* For LLQ, header buffer - pushed to the device mem space */
void *push_header;
enum ena_eth_io_l3_proto_index l3_proto;
enum ena_eth_io_l4_proto_index l4_proto;
u16 num_bufs;
u16 req_id;
/* For regular queue, indicate the size of the header
* For LLQ, indicate the size of the pushed buffer
*/
u16 header_len;
u8 meta_valid;
u8 tso_enable;
u8 l3_csum_enable;
u8 l4_csum_enable;
u8 l4_csum_partial;
u8 df; /* Don't fragment */
};
struct ena_com_rx_ctx {
struct ena_com_rx_buf_info *ena_bufs;
enum ena_eth_io_l3_proto_index l3_proto;
enum ena_eth_io_l4_proto_index l4_proto;
bool l3_csum_err;
bool l4_csum_err;
u8 l4_csum_checked;
/* fragmented packet */
bool frag;
u32 hash;
u16 descs;
int max_bufs;
u8 pkt_offset;
};
int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
struct ena_com_tx_ctx *ena_tx_ctx,
int *nb_hw_desc);
int ena_com_rx_pkt(struct ena_com_io_cq *io_cq,
struct ena_com_io_sq *io_sq,
struct ena_com_rx_ctx *ena_rx_ctx);
int ena_com_add_single_rx_desc(struct ena_com_io_sq *io_sq,
struct ena_com_buf *ena_buf,
u16 req_id);
bool ena_com_cq_empty(struct ena_com_io_cq *io_cq);
static inline void ena_com_unmask_intr(struct ena_com_io_cq *io_cq,
struct ena_eth_io_intr_reg *intr_reg)
{
writel(intr_reg->intr_control, io_cq->unmask_reg);
}
static inline int ena_com_free_q_entries(struct ena_com_io_sq *io_sq)
{
u16 tail, next_to_comp, cnt;
next_to_comp = io_sq->next_to_comp;
tail = io_sq->tail;
cnt = tail - next_to_comp;
return io_sq->q_depth - 1 - cnt;
}
/* Check if the submission queue has enough space to hold required_buffers */
static inline bool ena_com_sq_have_enough_space(struct ena_com_io_sq *io_sq,
u16 required_buffers)
{
int temp;
if (io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST)
return ena_com_free_q_entries(io_sq) >= required_buffers;
/* This calculation doesn't need to be 100% accurate. So to reduce
* the calculation overhead just Subtract 2 lines from the free descs
* (one for the header line and one to compensate the devision
* down calculation.
*/
temp = required_buffers / io_sq->llq_info.descs_per_entry + 2;
return ena_com_free_q_entries(io_sq) > temp;
}
static inline bool ena_com_meta_desc_changed(struct ena_com_io_sq *io_sq,
struct ena_com_tx_ctx *ena_tx_ctx)
{
if (!ena_tx_ctx->meta_valid)
return false;
return !!memcmp(&io_sq->cached_tx_meta,
&ena_tx_ctx->ena_meta,
sizeof(struct ena_com_tx_meta));
}
static inline bool is_llq_max_tx_burst_exists(struct ena_com_io_sq *io_sq)
{
return (io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) &&
io_sq->llq_info.max_entries_in_tx_burst > 0;
}
static inline bool ena_com_is_doorbell_needed(struct ena_com_io_sq *io_sq,
struct ena_com_tx_ctx *ena_tx_ctx)
{
struct ena_com_llq_info *llq_info;
int descs_after_first_entry;
int num_entries_needed = 1;
u16 num_descs;
if (!is_llq_max_tx_burst_exists(io_sq))
return false;
llq_info = &io_sq->llq_info;
num_descs = ena_tx_ctx->num_bufs;
if (llq_info->disable_meta_caching ||
unlikely(ena_com_meta_desc_changed(io_sq, ena_tx_ctx)))
++num_descs;
if (num_descs > llq_info->descs_num_before_header) {
descs_after_first_entry = num_descs - llq_info->descs_num_before_header;
num_entries_needed += DIV_ROUND_UP(descs_after_first_entry,
llq_info->descs_per_entry);
}
pr_debug("Queue: %d num_descs: %d num_entries_needed: %d\n", io_sq->qid,
num_descs, num_entries_needed);
return num_entries_needed > io_sq->entries_in_tx_burst_left;
}
static inline int ena_com_write_sq_doorbell(struct ena_com_io_sq *io_sq)
{
u16 max_entries_in_tx_burst = io_sq->llq_info.max_entries_in_tx_burst;
u16 tail = io_sq->tail;
pr_debug("Write submission queue doorbell for queue: %d tail: %d\n",
io_sq->qid, tail);
writel(tail, io_sq->db_addr);
if (is_llq_max_tx_burst_exists(io_sq)) {
pr_debug("Reset available entries in tx burst for queue %d to %d\n",
io_sq->qid, max_entries_in_tx_burst);
io_sq->entries_in_tx_burst_left = max_entries_in_tx_burst;
}
return 0;
}
static inline int ena_com_update_dev_comp_head(struct ena_com_io_cq *io_cq)
{
u16 unreported_comp, head;
bool need_update;
if (unlikely(io_cq->cq_head_db_reg)) {
head = io_cq->head;
unreported_comp = head - io_cq->last_head_update;
need_update = unreported_comp > (io_cq->q_depth / ENA_COMP_HEAD_THRESH);
if (unlikely(need_update)) {
pr_debug("Write completion queue doorbell for queue %d: head: %d\n",
io_cq->qid, head);
writel(head, io_cq->cq_head_db_reg);
io_cq->last_head_update = head;
}
}
return 0;
}
static inline void ena_com_update_numa_node(struct ena_com_io_cq *io_cq,
u8 numa_node)
{
struct ena_eth_io_numa_node_cfg_reg numa_cfg;
if (!io_cq->numa_node_cfg_reg)
return;
numa_cfg.numa_cfg = (numa_node & ENA_ETH_IO_NUMA_NODE_CFG_REG_NUMA_MASK)
| ENA_ETH_IO_NUMA_NODE_CFG_REG_ENABLED_MASK;
writel(numa_cfg.numa_cfg, io_cq->numa_node_cfg_reg);
}
static inline void ena_com_comp_ack(struct ena_com_io_sq *io_sq, u16 elem)
{
io_sq->next_to_comp += elem;
}
static inline void ena_com_cq_inc_head(struct ena_com_io_cq *io_cq)
{
io_cq->head++;
/* Switch phase bit in case of wrap around */
if (unlikely((io_cq->head & (io_cq->q_depth - 1)) == 0))
io_cq->phase ^= 1;
}
static inline int ena_com_tx_comp_req_id_get(struct ena_com_io_cq *io_cq,
u16 *req_id)
{
u8 expected_phase, cdesc_phase;
struct ena_eth_io_tx_cdesc *cdesc;
u16 masked_head;
masked_head = io_cq->head & (io_cq->q_depth - 1);
expected_phase = io_cq->phase;
cdesc = (struct ena_eth_io_tx_cdesc *)
((uintptr_t)io_cq->cdesc_addr.virt_addr +
(masked_head * io_cq->cdesc_entry_size_in_bytes));
/* When the current completion descriptor phase isn't the same as the
* expected, it mean that the device still didn't update
* this completion.
*/
cdesc_phase = READ_ONCE(cdesc->flags) & ENA_ETH_IO_TX_CDESC_PHASE_MASK;
if (cdesc_phase != expected_phase)
return -EAGAIN;
dma_rmb();
*req_id = READ_ONCE(cdesc->req_id);
if (unlikely(*req_id >= io_cq->q_depth)) {
pr_err("Invalid req id %d\n", cdesc->req_id);
return -EINVAL;
}
ena_com_cq_inc_head(io_cq);
return 0;
}
#endif /* ENA_ETH_COM_H_ */
|