File size: 7,518 Bytes
9425aed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
/* GGUF v3 parser — zero libc

 * Windows: CreateFileA/MapViewOfFile  Linux: open/mmap

 * No malloc — VirtualAlloc / mmap anonymous for metadata

 */
#include <stdint.h>
#include <stddef.h>

#ifdef _WIN32
#  define WIN32_LEAN_AND_MEAN
#  include <windows.h>
#  define SOV_ALLOC(sz)     VirtualAlloc(NULL,(sz),MEM_COMMIT|MEM_RESERVE,PAGE_READWRITE)
#  define SOV_FREE(p,sz)    VirtualFree((p),0,MEM_RELEASE)
#else
#  include <sys/mman.h>
#  include <fcntl.h>
#  include <unistd.h>
#  include <sys/stat.h>
#  define SOV_ALLOC(sz)     mmap(NULL,(sz),PROT_READ|PROT_WRITE,MAP_PRIVATE|MAP_ANONYMOUS,-1,0)
#  define SOV_FREE(p,sz)    munmap((p),(sz))
#endif

#define GGUF_MAGIC   0x46554747u  /* "GGUF" */
#define GGUF_VERSION 3u

typedef enum {
    GGUF_TYPE_F32  = 0,
    GGUF_TYPE_F16  = 1,
    GGUF_TYPE_Q4_0 = 2,
    GGUF_TYPE_Q8_0 = 8,
    GGUF_TYPE_Q4_K = 12,
    GGUF_TYPE_BF16 = 32,
} gguf_type_t;

typedef struct {
    char     name[256];
    uint32_t n_dims;
    uint64_t dims[4];
    uint32_t type;
    uint64_t offset;  /* from tensor data block */
} gguf_tensor_info_t;

typedef struct {
#ifdef _WIN32
    HANDLE   fh, mh;
#else
    int      fd;
#endif
    void*    base;
    size_t   file_size;
    uint64_t tensor_data_offset;
    uint32_t tensor_count;
    gguf_tensor_info_t* tensors;
} gguf_context_t;

static int map_file(gguf_context_t* ctx, const char* path) {
#ifdef _WIN32
    ctx->fh = CreateFileA(path, GENERIC_READ, FILE_SHARE_READ, NULL,
                          OPEN_EXISTING, FILE_ATTRIBUTE_NORMAL, NULL);
    if (ctx->fh == INVALID_HANDLE_VALUE) return -1;
    LARGE_INTEGER sz; GetFileSizeEx(ctx->fh, &sz); ctx->file_size = (size_t)sz.QuadPart;
    ctx->mh = CreateFileMappingA(ctx->fh, NULL, PAGE_READONLY, 0, 0, NULL);
    if (!ctx->mh) { CloseHandle(ctx->fh); return -2; }
    ctx->base = MapViewOfFile(ctx->mh, FILE_MAP_READ, 0, 0, 0);
    if (!ctx->base) { CloseHandle(ctx->mh); CloseHandle(ctx->fh); return -3; }
#else
    ctx->fd = open(path, O_RDONLY);
    if (ctx->fd < 0) return -1;
    struct stat st; fstat(ctx->fd, &st); ctx->file_size = st.st_size;
    ctx->base = mmap(NULL, ctx->file_size, PROT_READ, MAP_PRIVATE, ctx->fd, 0);
    if (ctx->base == MAP_FAILED) { close(ctx->fd); return -2; }
#endif
    return 0;
}

static void unmap_file(gguf_context_t* ctx) {
#ifdef _WIN32
    if (ctx->base) UnmapViewOfFile(ctx->base);
    if (ctx->mh)   CloseHandle(ctx->mh);
    if (ctx->fh && ctx->fh != INVALID_HANDLE_VALUE) CloseHandle(ctx->fh);
#else
    if (ctx->base && ctx->base != MAP_FAILED) munmap(ctx->base, ctx->file_size);
    if (ctx->fd >= 0) close(ctx->fd);
#endif
}

/* Skip GGUF KV value of given type; return advanced pointer */
static const char* skip_kv_value(const char* p, uint32_t type) {
    switch (type) {
        case 0: case 1: return p + 1;   /* uint8/int8 */
        case 2: case 3: return p + 2;   /* uint16/int16 */
        case 4: case 5: return p + 4;   /* uint32/int32/float32 */
        case 6:         return p + 1;   /* bool */
        case 7: { uint64_t n=*(uint64_t*)p; return p + 8 + n; } /* string */
        case 8:         return p + 8;   /* uint64/int64/float64 */
        case 9: {       /* array */
            uint32_t et=*(uint32_t*)p; p+=4;
            uint64_t n=*(uint64_t*)p;  p+=8;
            for (uint64_t i=0;i<n;i++) p = skip_kv_value(p, et);
            return p;
        }
        default: return p + 8;
    }
}

int gguf_load(const char* path, gguf_context_t** out) {
    gguf_context_t* ctx = (gguf_context_t*)SOV_ALLOC(sizeof(gguf_context_t));
    if (!ctx) return -1;
    /* zero-init */
    for (size_t i=0;i<sizeof(*ctx);i++) ((uint8_t*)ctx)[i]=0;

    if (map_file(ctx, path) != 0) { SOV_FREE(ctx, sizeof(*ctx)); return -2; }

    const uint8_t* b = (const uint8_t*)ctx->base;
    if (ctx->file_size < 24) goto fail;
    if (*(uint32_t*)(b+0) != GGUF_MAGIC)   goto fail;
    if (*(uint32_t*)(b+4) != GGUF_VERSION) goto fail;

    uint64_t n_tensors = *(uint64_t*)(b+8);
    uint64_t n_kv      = *(uint64_t*)(b+16);
    ctx->tensor_count  = (uint32_t)n_tensors;

    const char* p = (const char*)(b + 24);
    /* skip KV section */
    for (uint64_t i=0;i<n_kv;i++) {
        uint64_t klen = *(uint64_t*)p; p += 8 + klen; /* key */
        uint32_t vtype = *(uint32_t*)p; p += 4;
        p = skip_kv_value(p, vtype);
    }

    /* parse tensor info */
    size_t tsize = n_tensors * sizeof(gguf_tensor_info_t);
    ctx->tensors = (gguf_tensor_info_t*)SOV_ALLOC(tsize);
    if (!ctx->tensors) goto fail;
    for (size_t i=0;i<tsize;i++) ((uint8_t*)ctx->tensors)[i]=0;

    for (uint64_t i=0;i<n_tensors;i++) {
        uint64_t nlen = *(uint64_t*)p; p += 8;
        size_t copy = nlen < 255 ? nlen : 255;
        for (size_t j=0;j<copy;j++) ctx->tensors[i].name[j] = p[j];
        ctx->tensors[i].name[copy] = 0;
        p += nlen;
        ctx->tensors[i].n_dims = *(uint32_t*)p; p += 4;
        for (uint32_t d=0;d<ctx->tensors[i].n_dims;d++) {
            ctx->tensors[i].dims[d] = *(uint64_t*)p; p += 8;
        }
        ctx->tensors[i].type   = *(uint32_t*)p; p += 4;
        ctx->tensors[i].offset = *(uint64_t*)p; p += 8;
    }
    /* align tensor data to 32 bytes */
    size_t hdr_off = (size_t)(p - (const char*)b);
    ctx->tensor_data_offset = (hdr_off + 31) & ~31ULL;
    *out = ctx;
    return 0;

fail:
    unmap_file(ctx);
    if (ctx->tensors) SOV_FREE(ctx->tensors, ctx->tensor_count * sizeof(gguf_tensor_info_t));
    SOV_FREE(ctx, sizeof(*ctx));
    return -3;
}

const void* gguf_get_tensor(gguf_context_t* ctx, const char* name) {
    for (uint32_t i=0;i<ctx->tensor_count;i++) {
        const char* a = ctx->tensors[i].name, *b = name;
        while (*a && *b && *a == *b) { a++; b++; }
        if (!*a && !*b) {
            return (const uint8_t*)ctx->base + ctx->tensor_data_offset + ctx->tensors[i].offset;
        }
    }
    return 0;
}

/* Upload tensor to GPU via caller-supplied upload fn */
int gguf_upload_to_gpu(gguf_context_t* ctx, const char* name,

                       int (*h2d)(void* dst, const void* src, size_t sz),

                       void* gpu_dst) {
    for (uint32_t i=0;i<ctx->tensor_count;i++) {
        const char* a = ctx->tensors[i].name, *b = name;
        while (*a && *b && *a == *b) { a++; b++; }
        if (!*a && !*b) {
            const void* src = (const uint8_t*)ctx->base + ctx->tensor_data_offset + ctx->tensors[i].offset;
            /* compute byte size */
            size_t n = 1;
            for (uint32_t d=0;d<ctx->tensors[i].n_dims;d++) n *= (size_t)ctx->tensors[i].dims[d];
            size_t elem;
            switch (ctx->tensors[i].type) {
                case 0:  elem = 4; break;  /* F32 */
                case 1:  elem = 2; break;  /* F16 */
                case 32: elem = 2; break;  /* BF16 */
                case 2:  elem = 1; break;  /* Q4_0 ~0.5 byte/elem, approx 1 */
                case 8:  elem = 1; break;  /* Q8_0 */
                case 12: elem = 1; break;  /* Q4_K */
                default: elem = 1; break;
            }
            return h2d(gpu_dst, src, n * elem);
        }
    }
    return -1;
}

void gguf_free(gguf_context_t* ctx) {
    if (!ctx) return;
    if (ctx->tensors) SOV_FREE(ctx->tensors, ctx->tensor_count * sizeof(gguf_tensor_info_t));
    unmap_file(ctx);
    SOV_FREE(ctx, sizeof(*ctx));
}