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
|
#include <sys/prctl.h>
#include <unistd.h>
#include <netinet/ip.h>
#include <netinet/ip6.h>
#include <linux/tcp.h>
#include <linux/netfilter.h> // for NF_ACCEPT
#include <libnetfilter_queue/libnetfilter_queue.h>
#include <linux/if_tun.h>
#include <MESA/MESA_prof_load.h>
#include <bpf_obj.h>
#include <tfe_utils.h>
#include <tfe_cmsg.h>
#include <proxy.h>
#include "io_uring.h"
#include "tfe_packet_io_fs.h"
#include "tfe_tcp_restore.h"
#include "acceptor_kni_v4.h"
#include "tap.h"
#include "tfe_packet_io.h"
#include "tfe_session_table.h"
static int tfe_tap_read_per_thread(int tap_fd, char *buff, int buff_size, void *logger)
{
int ret = read(tap_fd, buff, buff_size);
if (ret < 0)
{
if (errno != EWOULDBLOCK && errno != EAGAIN)
{
TFE_LOG_ERROR(g_default_logger, "%s: unable to read data from tapfd %d, aborting: %s", LOG_TAG_PKTIO, tap_fd, strerror(errno));
}
}
return ret;
}
void acceptor_ctx_destory(struct acceptor_kni_v4 * ctx)
{
if (ctx)
{
packet_io_destory(ctx->io);
packet_io_fs_destory(ctx->packet_io_fs);
free(ctx);
ctx = NULL;
}
return;
}
struct acceptor_kni_v4 *acceptor_ctx_create(const char *profile)
{
struct acceptor_kni_v4 *ctx = ALLOC(struct acceptor_kni_v4, 1);
MESA_load_profile_int_def(profile, "PACKET_IO", "firewall_sids", (int *)&(ctx->firewall_sids), 1000);
MESA_load_profile_int_def(profile, "PACKET_IO", "proxy_sids", (int *)&(ctx->proxy_sids), 1001);
MESA_load_profile_int_def(profile, "PACKET_IO", "service_chaining_sids", (int *)&(ctx->sce_sids), 1002);
MESA_load_profile_int_def(profile, "PACKET_IO", "packet_io_threads", (int *)&(ctx->nr_worker_threads), 8);
MESA_load_profile_uint_range(profile, "PACKET_IO", "packet_io_cpu_affinity_mask", TFE_THREAD_MAX, (unsigned int *)ctx->cpu_affinity_mask);
ctx->nr_worker_threads = MIN(ctx->nr_worker_threads, TFE_THREAD_MAX);
CPU_ZERO(&ctx->coremask);
for (int i = 0; i < ctx->nr_worker_threads; i++)
{
int cpu_id = ctx->cpu_affinity_mask[i];
CPU_SET(cpu_id, &ctx->coremask);
}
ctx->io = packet_io_create(profile, ctx->nr_worker_threads, &ctx->coremask);
if (ctx->io == NULL)
{
goto error_out;
}
ctx->packet_io_fs = packet_io_fs_create();
if (ctx->packet_io_fs == NULL)
{
goto error_out;
}
return ctx;
error_out:
acceptor_ctx_destory(ctx);
return NULL;
}
static void *worker_thread_cycle(void *arg)
{
struct packet_io_thread_ctx *thread_ctx = (struct packet_io_thread_ctx *)arg;
struct packet_io *handle = thread_ctx->ref_io;
int pkg_len = 0;
char thread_name[16];
int n_pkt_recv_from_nf = 0;
int n_pkt_recv_from_tap = 0;
int n_pkt_recv_from_tap_c = 0;
int n_pkt_recv_from_tap_s = 0;
snprintf(thread_name, sizeof(thread_name), "kni:worker-%d", thread_ctx->thread_index);
prctl(PR_SET_NAME, (unsigned long long)thread_name, NULL, NULL, NULL);
if (packet_io_thread_init(handle, thread_ctx) != 0)
{
goto error_out;
}
if (is_enable_iouring(handle)) {
io_uring_register_read_callback(thread_ctx->tap_ctx->io_uring_fd, handle_raw_packet_from_tap, thread_ctx);
io_uring_register_read_callback(thread_ctx->tap_ctx->io_uring_c, handle_decryption_packet_from_tap, thread_ctx);
io_uring_register_read_callback(thread_ctx->tap_ctx->io_uring_s, handle_decryption_packet_from_tap, thread_ctx);
}
else {
thread_ctx->tap_ctx->buff_size = 3000;
thread_ctx->tap_ctx->buff = ALLOC(char, thread_ctx->tap_ctx->buff_size);
}
TFE_LOG_INFO(g_default_logger, "%s: worker thread %d is running", "LOG_TAG_KNI", thread_ctx->thread_index);
while(1) {
n_pkt_recv_from_nf = packet_io_polling_nf_interface(handle, thread_ctx->thread_index, thread_ctx);
if (is_enable_iouring(handle)) {
n_pkt_recv_from_tap = io_uring_peek_ready_entrys(thread_ctx->tap_ctx->io_uring_fd);
n_pkt_recv_from_tap_c = io_uring_peek_ready_entrys(thread_ctx->tap_ctx->io_uring_c);
n_pkt_recv_from_tap_s = io_uring_peek_ready_entrys(thread_ctx->tap_ctx->io_uring_s);
}
else {
if ((pkg_len = tfe_tap_read_per_thread(thread_ctx->tap_ctx->tap_fd, thread_ctx->tap_ctx->buff, thread_ctx->tap_ctx->buff_size, g_default_logger)) > 0)
{
handle_raw_packet_from_tap(thread_ctx->tap_ctx->buff, pkg_len, thread_ctx);
}
if ((pkg_len = tfe_tap_read_per_thread(thread_ctx->tap_ctx->tap_c, thread_ctx->tap_ctx->buff, thread_ctx->tap_ctx->buff_size, g_default_logger)) > 0)
{
handle_decryption_packet_from_tap(thread_ctx->tap_ctx->buff, pkg_len, thread_ctx);
}
if ((pkg_len = tfe_tap_read_per_thread(thread_ctx->tap_ctx->tap_s, thread_ctx->tap_ctx->buff, thread_ctx->tap_ctx->buff_size, g_default_logger)) > 0)
{
handle_decryption_packet_from_tap(thread_ctx->tap_ctx->buff, pkg_len, thread_ctx);
}
}
if (n_pkt_recv_from_nf == 0 && n_pkt_recv_from_tap == 0 && n_pkt_recv_from_tap_c == 0 && n_pkt_recv_from_tap_s == 0)
{
packet_io_thread_wait(handle, thread_ctx, -1);
}
if (__atomic_fetch_add(&thread_ctx->session_table_need_reset, 0, __ATOMIC_RELAXED) > 0)
{
session_table_reset(thread_ctx->session_table);
__atomic_fetch_and(&thread_ctx->session_table_need_reset, 0, __ATOMIC_RELAXED);
}
}
error_out:
TFE_LOG_ERROR(g_default_logger, "%s: worker thread %d exiting", LOG_TAG_SCE, thread_ctx->thread_index);
return (void *)NULL;
}
void acceptor_kni_v4_destroy(struct acceptor_kni_v4 *ctx)
{
if (ctx)
{
packet_io_destory(ctx->io);
free(ctx);
ctx = NULL;
}
return;
}
struct acceptor_kni_v4 *acceptor_kni_v4_create(struct tfe_proxy *proxy, const char *profile, void *logger)
{
struct acceptor_kni_v4 *acceptor_ctx = acceptor_ctx_create(profile);
if (acceptor_ctx == NULL)
goto error_out;
acceptor_ctx->ref_proxy = proxy;
for (int i = 0; i < acceptor_ctx->nr_worker_threads; i++) {
acceptor_ctx->work_threads[i].tid = 0;
acceptor_ctx->work_threads[i].thread_index = i;
acceptor_ctx->work_threads[i].ref_acceptor_ctx = acceptor_ctx;
acceptor_ctx->work_threads[i].tap_ctx = tfe_tap_ctx_create(&acceptor_ctx->work_threads[i]);
if (acceptor_ctx->work_threads[i].tap_ctx == NULL)
goto error_out;
acceptor_ctx->work_threads[i].session_table = session_table_create();
acceptor_ctx->work_threads[i].ref_io = acceptor_ctx->io;
acceptor_ctx->work_threads[i].ref_proxy = proxy;
acceptor_ctx->work_threads[i].ret_fs_state = acceptor_ctx->packet_io_fs;
acceptor_ctx->work_threads[i].ref_acceptor_ctx = acceptor_ctx;
acceptor_ctx->work_threads[i].session_table_need_reset = 0;
}
for (int i = 0; i < acceptor_ctx->nr_worker_threads; i++) {
struct packet_io_thread_ctx *thread_ctx = &acceptor_ctx->work_threads[i];
if (pthread_create(&thread_ctx->tid, NULL, worker_thread_cycle, (void *)thread_ctx) < 0)
{
goto error_out;
}
}
return acceptor_ctx;
error_out:
for (int i = 0; i < acceptor_ctx->nr_worker_threads; i++) {
tfe_tap_ctx_destory(acceptor_ctx->work_threads[i].tap_ctx);
session_table_destory(acceptor_ctx->work_threads[i].session_table);
}
acceptor_kni_v4_destroy(acceptor_ctx);
return NULL;
}
|
