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
|
/*
*------------------------------------------------------------------
* Copyright (c) 2018 Cisco and/or its affiliates.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at:
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*------------------------------------------------------------------
*/
#include <vlib/vlib.h>
#include <vlib/unix/unix.h>
#include <vlib/pci/pci.h>
#include <vnet/ethernet/ethernet.h>
#include <vnet/devices/devices.h>
#include <vmxnet3/vmxnet3.h>
static_always_inline void
vmxnet3_tx_comp_ring_advance_next (vmxnet3_txq_t * txq)
{
vmxnet3_tx_comp_ring *comp_ring = &txq->tx_comp_ring;
comp_ring->next++;
if (PREDICT_FALSE (comp_ring->next == txq->size))
{
comp_ring->next = 0;
comp_ring->gen ^= VMXNET3_TXCF_GEN;
}
}
static_always_inline void
vmxnet3_tx_ring_advance_produce (vmxnet3_txq_t * txq)
{
txq->tx_ring.produce++;
if (PREDICT_FALSE (txq->tx_ring.produce == txq->size))
{
txq->tx_ring.produce = 0;
txq->tx_ring.gen ^= VMXNET3_TXF_GEN;
}
}
static_always_inline void
vmxnet3_tx_ring_advance_consume (vmxnet3_txq_t * txq)
{
txq->tx_ring.consume++;
txq->tx_ring.consume &= txq->size - 1;
}
static_always_inline void
vmxnet3_txq_release (vlib_main_t * vm, vmxnet3_device_t * vd,
vmxnet3_txq_t * txq)
{
vmxnet3_tx_comp *tx_comp;
vmxnet3_tx_comp_ring *comp_ring;
comp_ring = &txq->tx_comp_ring;
tx_comp = &txq->tx_comp[comp_ring->next];
while ((tx_comp->flags & VMXNET3_TXCF_GEN) == comp_ring->gen)
{
u16 eop_idx = tx_comp->index & VMXNET3_TXC_INDEX;
u32 bi0 = txq->tx_ring.bufs[txq->tx_ring.consume];
vlib_buffer_free_one (vm, bi0);
while (txq->tx_ring.consume != eop_idx)
{
vmxnet3_tx_ring_advance_consume (txq);
}
vmxnet3_tx_ring_advance_consume (txq);
vmxnet3_tx_comp_ring_advance_next (txq);
tx_comp = &txq->tx_comp[comp_ring->next];
}
}
static_always_inline u16
vmxnet3_tx_ring_space_left (vmxnet3_txq_t * txq)
{
u16 count;
count = (txq->tx_ring.consume - txq->tx_ring.produce - 1);
/* Wrapped? */
if (txq->tx_ring.produce >= txq->tx_ring.consume)
count += txq->size;
return count;
}
VNET_DEVICE_CLASS_TX_FN (vmxnet3_device_class) (vlib_main_t * vm,
vlib_node_runtime_t * node,
vlib_frame_t * frame)
{
vmxnet3_main_t *vmxm = &vmxnet3_main;
vnet_interface_output_runtime_t *rd = (void *) node->runtime_data;
vmxnet3_device_t *vd = pool_elt_at_index (vmxm->devices, rd->dev_instance);
u32 *buffers = vlib_frame_vector_args (frame);
u32 bi0;
vlib_buffer_t *b0;
vmxnet3_tx_desc *txd = 0;
u32 desc_idx, generation, first_idx;
u16 space_left;
u16 n_left = frame->n_vectors;
vmxnet3_txq_t *txq;
u16 qid = vm->thread_index % vd->num_tx_queues, produce;
if (PREDICT_FALSE (!(vd->flags & VMXNET3_DEVICE_F_LINK_UP)))
{
vlib_buffer_free (vm, buffers, n_left);
vlib_error_count (vm, node->node_index, VMXNET3_TX_ERROR_LINK_DOWN,
n_left);
return (0);
}
txq = vec_elt_at_index (vd->txqs, qid);
clib_spinlock_lock_if_init (&txq->lock);
vmxnet3_txq_release (vm, vd, txq);
produce = txq->tx_ring.produce;
while (PREDICT_TRUE (n_left))
{
u16 space_needed = 1, i;
vlib_buffer_t *b;
bi0 = buffers[0];
b0 = vlib_get_buffer (vm, bi0);
b = b0;
space_left = vmxnet3_tx_ring_space_left (txq);
while (b->flags & VLIB_BUFFER_NEXT_PRESENT)
{
u32 next_buffer = b->next_buffer;
b = vlib_get_buffer (vm, next_buffer);
space_needed++;
}
if (PREDICT_FALSE (space_left < space_needed))
{
vmxnet3_txq_release (vm, vd, txq);
space_left = vmxnet3_tx_ring_space_left (txq);
if (PREDICT_FALSE (space_left < space_needed))
{
vlib_buffer_free_one (vm, bi0);
vlib_error_count (vm, node->node_index,
VMXNET3_TX_ERROR_NO_FREE_SLOTS, 1);
buffers++;
n_left--;
/*
* Drop this packet. But we may have enough room for the next
* packet
*/
continue;
}
}
/*
* Toggle the generation bit for SOP fragment to avoid device starts
* reading incomplete packet
*/
generation = txq->tx_ring.gen ^ VMXNET3_TXF_GEN;
first_idx = txq->tx_ring.produce;
for (i = 0; i < space_needed; i++)
{
b0 = vlib_get_buffer (vm, bi0);
VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b0);
desc_idx = txq->tx_ring.produce;
vmxnet3_tx_ring_advance_produce (txq);
txq->tx_ring.bufs[desc_idx] = bi0;
txd = &txq->tx_desc[desc_idx];
txd->address = vlib_buffer_get_current_pa (vm, b0);
txd->flags[0] = generation | b0->current_length;
generation = txq->tx_ring.gen;
txd->flags[1] = 0;
bi0 = b0->next_buffer;
}
txd->flags[1] = VMXNET3_TXF_CQ | VMXNET3_TXF_EOP;
asm volatile ("":::"memory");
/*
* Now toggle back the generation bit for the first segment.
* Device can start reading the packet
*/
txq->tx_desc[first_idx].flags[0] ^= VMXNET3_TXF_GEN;
buffers++;
n_left--;
}
if (PREDICT_TRUE (produce != txq->tx_ring.produce))
vmxnet3_reg_write_inline (vd, 0, txq->reg_txprod, txq->tx_ring.produce);
clib_spinlock_unlock_if_init (&txq->lock);
return (frame->n_vectors - n_left);
}
/*
* fd.io coding-style-patch-verification: ON
*
* Local Variables:
* eval: (c-set-style "gnu")
* End:
*/
|
