Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 05:48:47 +01:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request from GHSA-p5mv-gjc5-mwqv

Browse Source 
* always use calloc

clamp n_kv on failure to read a kv

* ggml : alternative ctx->header.n_kv update

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
Georgi Gerganov 2024-04-26 10:41:53 +03:00 committed by GitHub
parent d4a9afc100
commit 83b72cb086
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 18 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
ggml.c
View File

@ -20614,7 +20614,7 @@ static void gguf_free_kv(struct gguf_kv * kv) {
} }
struct gguf_context * gguf_init_empty(void) { struct gguf_context * gguf_init_empty(void) {
struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context)); struct gguf_context * ctx = GGML_CALLOC(1, sizeof(struct gguf_context));
memcpy(ctx->header.magic, GGUF_MAGIC, sizeof(ctx->header.magic)); memcpy(ctx->header.magic, GGUF_MAGIC, sizeof(ctx->header.magic));
ctx->header.version = GGUF_VERSION; ctx->header.version = GGUF_VERSION;
@ -20659,7 +20659,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
bool ok = true; bool ok = true;
struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context)); struct gguf_context * ctx = GGML_CALLOC(1, sizeof(struct gguf_context));
// read the header // read the header
{ {
@ -20696,9 +20696,13 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
// read the kv pairs // read the kv pairs
{ {
ctx->kv = GGML_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv)); const uint64_t n_kv = ctx->header.n_kv;
for (uint64_t i = 0; i < ctx->header.n_kv; ++i) { // header.n_kv will hold the actual value of pairs that were successfully read in the loop below
ctx->header.n_kv = 0;
ctx->kv = GGML_CALLOC(n_kv, sizeof(struct gguf_kv));
for (uint64_t i = 0; i < n_kv; ++i) {
struct gguf_kv * kv = &ctx->kv[i]; struct gguf_kv * kv = &ctx->kv[i];
//fprintf(stderr, "%s: reading kv %d\n", __func__, i); //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
@ -20747,7 +20751,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
return NULL; return NULL;
} }
kv->value.arr.data = GGML_MALLOC(kv->value.arr.n * gguf_type_size(kv->value.arr.type)); kv->value.arr.data = GGML_CALLOC(kv->value.arr.n, gguf_type_size(kv->value.arr.type));
ok = ok && gguf_fread_el(file, kv->value.arr.data, kv->value.arr.n * gguf_type_size(kv->value.arr.type), &offset); ok = ok && gguf_fread_el(file, kv->value.arr.data, kv->value.arr.n * gguf_type_size(kv->value.arr.type), &offset);
} break; } break;
@ -20761,7 +20765,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
return NULL; return NULL;
} }
kv->value.arr.data = GGML_MALLOC(kv->value.arr.n * sizeof(struct gguf_str)); kv->value.arr.data = GGML_CALLOC(kv->value.arr.n, sizeof(struct gguf_str));
for (uint64_t j = 0; j < kv->value.arr.n; ++j) { for (uint64_t j = 0; j < kv->value.arr.n; ++j) {
ok = ok && gguf_fread_str(file, &((struct gguf_str *) kv->value.arr.data)[j], &offset); ok = ok && gguf_fread_str(file, &((struct gguf_str *) kv->value.arr.data)[j], &offset);
@ -20777,6 +20781,8 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
if (!ok) { if (!ok) {
break; break;
} }
ctx->header.n_kv++;
} }
if (!ok) { if (!ok) {
@ -20789,7 +20795,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
// read the tensor infos // read the tensor infos
{ {
ctx->infos = GGML_MALLOC(ctx->header.n_tensors * sizeof(struct gguf_tensor_info)); ctx->infos = GGML_CALLOC(ctx->header.n_tensors, sizeof(struct gguf_tensor_info));
for (uint64_t i = 0; i < ctx->header.n_tensors; ++i) { for (uint64_t i = 0; i < ctx->header.n_tensors; ++i) {
struct gguf_tensor_info * info = &ctx->infos[i]; struct gguf_tensor_info * info = &ctx->infos[i];
@ -20810,6 +20816,7 @@ struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_p
ok = ok && gguf_fread_el (file, &info->type, sizeof(info->type), &offset); ok = ok && gguf_fread_el (file, &info->type, sizeof(info->type), &offset);
ok = ok && gguf_fread_el (file, &info->offset, sizeof(info->offset), &offset); ok = ok && gguf_fread_el (file, &info->offset, sizeof(info->offset), &offset);
// TODO: return an error instead of crashing with GGML_ASSERT
gguf_tensor_info_sanitize(info); gguf_tensor_info_sanitize(info);
if (!ok) { if (!ok) {
@ -21291,7 +21298,7 @@ void gguf_set_arr_data(struct gguf_context * ctx, const char * key, enum gguf_ty
ctx->kv[idx].type = GGUF_TYPE_ARRAY; ctx->kv[idx].type = GGUF_TYPE_ARRAY;
ctx->kv[idx].value.arr.type = type; ctx->kv[idx].value.arr.type = type;
ctx->kv[idx].value.arr.n = n; ctx->kv[idx].value.arr.n = n;
ctx->kv[idx].value.arr.data = GGML_MALLOC(n*gguf_type_size(type)); ctx->kv[idx].value.arr.data = GGML_CALLOC(n, gguf_type_size(type));
memcpy(ctx->kv[idx].value.arr.data, data, n*gguf_type_size(type)); memcpy(ctx->kv[idx].value.arr.data, data, n*gguf_type_size(type));
} }
@ -21301,7 +21308,7 @@ void gguf_set_arr_str(struct gguf_context * ctx, const char * key, const char **
ctx->kv[idx].type = GGUF_TYPE_ARRAY; ctx->kv[idx].type = GGUF_TYPE_ARRAY;
ctx->kv[idx].value.arr.type = GGUF_TYPE_STRING; ctx->kv[idx].value.arr.type = GGUF_TYPE_STRING;
ctx->kv[idx].value.arr.n = n; ctx->kv[idx].value.arr.n = n;
ctx->kv[idx].value.arr.data = GGML_MALLOC(n*sizeof(struct gguf_str)); ctx->kv[idx].value.arr.data = GGML_CALLOC(n, sizeof(struct gguf_str));
for (int i = 0; i < n; i++) { for (int i = 0; i < n; i++) {
struct gguf_str * str = &((struct gguf_str *)ctx->kv[idx].value.arr.data)[i]; struct gguf_str * str = &((struct gguf_str *)ctx->kv[idx].value.arr.data)[i];
str->n = strlen(data[i]); str->n = strlen(data[i]);
@ -21328,7 +21335,7 @@ void gguf_set_kv(struct gguf_context * ctx, struct gguf_context * src) {
case GGUF_TYPE_ARRAY: case GGUF_TYPE_ARRAY:
{ {
if (src->kv[i].value.arr.type == GGUF_TYPE_STRING) { if (src->kv[i].value.arr.type == GGUF_TYPE_STRING) {
const char ** data = GGML_MALLOC(src->kv[i].value.arr.n*sizeof(char *)); const char ** data = GGML_CALLOC(src->kv[i].value.arr.n, sizeof(char *));
for (uint32_t j = 0; j < src->kv[i].value.arr.n; j++) { for (uint32_t j = 0; j < src->kv[i].value.arr.n; j++) {
data[j] = ((struct gguf_str *)src->kv[i].value.arr.data)[j].data; data[j] = ((struct gguf_str *)src->kv[i].value.arr.data)[j].data;
} }
@ -21416,7 +21423,7 @@ struct gguf_buf {
static struct gguf_buf gguf_buf_init(size_t size) { static struct gguf_buf gguf_buf_init(size_t size) {
struct gguf_buf buf = { struct gguf_buf buf = {
/*buf.data =*/ size == 0 ? NULL : GGML_MALLOC(size), /*buf.data =*/ size == 0 ? NULL : GGML_CALLOC(1, size),
/*buf.size =*/ size, /*buf.size =*/ size,
/*buf.offset =*/ 0, /*buf.offset =*/ 0,
}; };