summaryrefslogtreecommitdiffstats
path: root/src/vnet/bier/bier_output.c
blob: 01eeffe475c71d2d40d7593f9afbbab9376f37d7 (plain)
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
/*
 * Copyright (c) 2016 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 <vnet/buffer.h>

#include <vnet/bier/bier_fmask.h>
#include <vnet/bier/bier_hdr_inlines.h>
#include <vlib/vlib.h>

static char * bier_output_error_strings[] = {
#define bier_error(n,s) s,
#include <vnet/bier/bier_output_error.def>
#undef bier_error
};

/*
 * Keep these values sematically the same as BIER output
 */
#define foreach_bier_output_next                \
    _(DROP, "bier-drop")

typedef enum {
#define _(s,n) BIER_OUTPUT_NEXT_##s,
    foreach_bier_output_next
#undef _
    BIER_OUTPUT_N_NEXT,
} bier_output_next_t;

typedef enum {
#define bier_error(n,s) BIER_OUTPUT_ERROR_##n,
#include <vnet/bier/bier_output_error.def>
#undef bier_error
    BIER_OUTPUT_N_ERROR,
} bier_output_error_t;

/**
 * Forward declaration
 */
vlib_node_registration_t bier_output_node;

/**
 * @brief Packet trace recoed for a BIER output
 */
typedef struct bier_output_trace_t_
{
    u32 next_index;
    index_t bfm_index;
} bier_output_trace_t;

static uword
bier_output (vlib_main_t * vm,
             vlib_node_runtime_t * node,
             vlib_frame_t * from_frame)
{
    u32 n_left_from, next_index, * from, * to_next;

    from = vlib_frame_vector_args (from_frame);
    n_left_from = from_frame->n_vectors;

    // vnet_buffer(b0)->sw_if_index[VLIB_TX] = d0->tx_fib_index;

    /*
     * objection your honour! speculation!
     */
    next_index = node->cached_next_index;

    while (n_left_from > 0)
    {
        u32 n_left_to_next;

        vlib_get_next_frame (vm, node, next_index,
                             to_next, n_left_to_next);

        while (n_left_from > 0 && n_left_to_next > 0)
        {
            bier_output_next_t next0;
            bier_bit_string_t bbs;
            vlib_buffer_t * b0;
            bier_fmask_t *bfm0;
            mpls_label_t *h0;
            bier_hdr_t *bh0;
            u32 bfmi0;
            u32 bi0;

            bi0 = from[0];
            to_next[0] = bi0;
            from += 1;
            to_next += 1;
            n_left_from -= 1;
            n_left_to_next -= 1;

            b0 = vlib_get_buffer (vm, bi0);
            bh0 = vlib_buffer_get_current (b0);
            bier_bit_string_init_from_hdr(bh0, &bbs);

            /*
             * In the BIER Lookup node we squirelled away the
             * BIER fmask index as the adj index
             */
            bfmi0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX];
            bfm0 = bier_fmask_get(bfmi0);

            /*
             * perform the logical AND of the packet's mask with
             * that of the fmask objects, to reset the bits that
             * are only on the shortest path the the fmask NH.
             */
            bier_bit_string_logical_and_string(
                &bfm0->bfm_bits.bfmb_input_reset_string,
                &bbs);

            /*
             * this is the last time we touch the BIER header
             * so flip to network order
             */
            bier_hdr_hton(bh0);

            /*
             * paint the BIER peer's label
             */
            if (!(bfm0->bfm_flags & BIER_FMASK_FLAG_DISP))
            {
                /*
                 * since a BIFT value and a MPLS label are formated the
                 * same, this painting works OK.
                 */
                vlib_buffer_advance(b0, -(word)sizeof(mpls_label_t));
                h0 = vlib_buffer_get_current(b0);
                
                h0[0] = bfm0->bfm_label;

                ((char*)h0)[3]= vnet_buffer(b0)->mpls.ttl - 1;
            }

            /*
             * setup next graph node
             */
            next0 = bfm0->bfm_dpo.dpoi_next_node;
            vnet_buffer(b0)->ip.adj_index[VLIB_TX] = bfm0->bfm_dpo.dpoi_index;

            if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED))
            {
                bier_output_trace_t *tr;

                tr = vlib_add_trace (vm, node, b0, sizeof (*tr));
                tr->next_index = next0;
                tr->bfm_index = bfmi0;
            }

            vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
                                             to_next, n_left_to_next,
                                             bi0, next0);
        }

        vlib_put_next_frame (vm, node, next_index, n_left_to_next);
    }

    vlib_node_increment_counter (vm, bier_output_node.index,
                                 BIER_OUTPUT_ERROR_NONE,
                                 from_frame->n_vectors);
    return (from_frame->n_vectors);
}

static u8 *
format_bier_output_trace (u8 * s, va_list * args)
{
    CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
    CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
    bier_output_trace_t * t = va_arg (*args, bier_output_trace_t *);

    s = format (s, " next [%d], BFM index %d",
                t->next_index, t->bfm_index);
    return s;
}

VLIB_REGISTER_NODE (bier_output_node) = {
    .function = bier_output,
    .name = "bier-output",
    /* Takes a vector of packets. */
    .vector_size = sizeof (u32),

    .n_errors = BIER_OUTPUT_N_ERROR,
    .error_strings = bier_output_error_strings,

    .n_next_nodes = BIER_OUTPUT_N_NEXT,
    .next_nodes = {
        [BIER_OUTPUT_NEXT_DROP] = "bier-drop",
    },

    .format_trace = format_bier_output_trace,
};
\ while (len >= 32 + 16 - offset) { \ xmm0 = _mm_loadu_si128((const __m128i *)((const u8 *)src - offset + 0 * 16)); \ len -= 32; \ xmm1 = _mm_loadu_si128((const __m128i *)((const u8 *)src - offset + 1 * 16)); \ xmm2 = _mm_loadu_si128((const __m128i *)((const u8 *)src - offset + 2 * 16)); \ src = (const u8 *)src + 32; \ _mm_storeu_si128((__m128i *)((u8 *)dst + 0 * 16), _mm_alignr_epi8(xmm1, xmm0, offset)); \ _mm_storeu_si128((__m128i *)((u8 *)dst + 1 * 16), _mm_alignr_epi8(xmm2, xmm1, offset)); \ dst = (u8 *)dst + 32; \ } \ tmp = len; \ len = ((len - 16 + offset) & 31) + 16 - offset; \ tmp -= len; \ src = (const u8 *)src + tmp; \ dst = (u8 *)dst + tmp; \ } \ }) /** * Macro for copying unaligned block from one location to another, * 47 bytes leftover maximum, * locations should not overlap. * Use switch here because the aligning instruction requires immediate value for shift count. * Requirements: * - Store is aligned * - Load offset is <offset>, which must be within [1, 15] * - For <src>, make sure <offset> bit backwards & <16 - offset> bit forwards are available for loading * - <dst>, <src>, <len> must be variables * - __m128i <xmm0> ~ <xmm8> used in CLIB_MVUNALIGN_LEFT47_IMM must be pre-defined */ #define CLIB_MVUNALIGN_LEFT47(dst, src, len, offset) \ ({ \ switch (offset) { \ case 0x01: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x01); break; \ case 0x02: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x02); break; \ case 0x03: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x03); break; \ case 0x04: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x04); break; \ case 0x05: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x05); break; \ case 0x06: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x06); break; \ case 0x07: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x07); break; \ case 0x08: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x08); break; \ case 0x09: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x09); break; \ case 0x0A: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0A); break; \ case 0x0B: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0B); break; \ case 0x0C: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0C); break; \ case 0x0D: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0D); break; \ case 0x0E: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0E); break; \ case 0x0F: CLIB_MVUNALIGN_LEFT47_IMM(dst, src, n, 0x0F); break; \ default:; \ } \ }) static inline void * clib_memcpy_fast (void *dst, const void *src, size_t n) { __m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8; uword dstu = (uword) dst; uword srcu = (uword) src; void *ret = dst; size_t dstofss; size_t srcofs; /** * Copy less than 16 bytes */ if (n < 16) { if (n & 0x01) { *(u8 *) dstu = *(const u8 *) srcu; srcu = (uword) ((const u8 *) srcu + 1); dstu = (uword) ((u8 *) dstu + 1); } if (n & 0x02) { *(u16 *) dstu = *(const u16 *) srcu; srcu = (uword) ((const u16 *) srcu + 1); dstu = (uword) ((u16 *) dstu + 1); } if (n & 0x04) { *(u32 *) dstu = *(const u32 *) srcu; srcu = (uword) ((const u32 *) srcu + 1); dstu = (uword) ((u32 *) dstu + 1); } if (n & 0x08) { *(u64 *) dstu = *(const u64 *) srcu; } return ret; } /** * Fast way when copy size doesn't exceed 512 bytes */ if (n <= 32) { clib_mov16 ((u8 *) dst, (const u8 *) src); clib_mov16 ((u8 *) dst - 16 + n, (const u8 *) src - 16 + n); return ret; } if (n <= 48) { clib_mov32 ((u8 *) dst, (const u8 *) src); clib_mov16 ((u8 *) dst - 16 + n, (const u8 *) src - 16 + n); return ret; } if (n <= 64) { clib_mov32 ((u8 *) dst, (const u8 *) src); clib_mov16 ((u8 *) dst + 32, (const u8 *) src + 32); clib_mov16 ((u8 *) dst - 16 + n, (const u8 *) src - 16 + n); return ret; } if (n <= 128) { goto COPY_BLOCK_128_BACK15; } if (n <= 512) { if (n >= 256) { n -= 256; clib_mov128 ((u8 *) dst, (const u8 *) src); clib_mov128 ((u8 *) dst + 128, (const u8 *) src + 128); src = (const u8 *) src + 256; dst = (u8 *) dst + 256; } COPY_BLOCK_255_BACK15: if (n >= 128) { n -= 128; clib_mov128 ((u8 *) dst, (const u8 *) src); src = (const u8 *) src + 128; dst = (u8 *) dst + 128; } COPY_BLOCK_128_BACK15: if (n >= 64) { n -= 64; clib_mov64 ((u8 *) dst, (const u8 *) src); src = (const u8 *) src + 64; dst = (u8 *) dst + 64; } COPY_BLOCK_64_BACK15: if (n >= 32) { n -= 32; clib_mov32 ((u8 *) dst, (const u8 *) src); src = (const u8 *) src + 32; dst = (u8 *) dst + 32; } if (n > 16) { clib_mov16 ((u8 *) dst, (const u8 *) src); clib_mov16 ((u8 *) dst - 16 + n, (const u8 *) src - 16 + n); return ret; } if (n > 0) { clib_mov16 ((u8 *) dst - 16 + n, (const u8 *) src - 16 + n); } return ret; } /** * Make store aligned when copy size exceeds 512 bytes, * and make sure the first 15 bytes are copied, because * unaligned copy functions require up to 15 bytes * backwards access. */ dstofss = (uword) dst & 0x0F; if (dstofss > 0) { dstofss = 16 - dstofss + 16; n -= dstofss; clib_mov32 ((u8 *) dst, (const u8 *) src); src = (const u8 *) src + dstofss; dst = (u8 *) dst + dstofss; } srcofs = ((uword) src & 0x0F); /** * For aligned copy */ if (srcofs == 0) { /** * Copy 256-byte blocks */ for (; n >= 256; n -= 256) { clib_mov256 ((u8 *) dst, (const u8 *) src); dst = (u8 *) dst + 256; src = (const u8 *) src + 256; } /** * Copy whatever left */ goto COPY_BLOCK_255_BACK15; } /** * For copy with unaligned load */ CLIB_MVUNALIGN_LEFT47 (dst, src, n, srcofs); /** * Copy whatever left */ goto COPY_BLOCK_64_BACK15; } #undef CLIB_MVUNALIGN_LEFT47_IMM #undef CLIB_MVUNALIGN_LEFT47 #endif /* included_clib_memcpy_sse3_h */ /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */