gnh1201/wasm-micro-runtime
1
0
Fork 0
mirror of https://github.com/bytecodealliance/wasm-micro-runtime.git synced 2025-09-07 10:22:52 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

wasi_nn_llamacpp.c: fix buffer overruns in set_input (#4420)

Browse Source 
note: for some reasons, wasmedge seems to ignore type/dimensions
for the input of ggml. some user code relies on it.
cf. https://github.com/second-state/WasmEdge-WASINN-examples/issues/196

note: despite the comment in our code, the input doesn't seem
nul-terminated.
This commit is contained in:
YAMAMOTO Takashi 2025-06-27 12:51:03 +09:00 committed by GitHub
parent 2372a472aa
commit d6876f1e9f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
core/iwasm/libraries/wasi-nn/src/wasi_nn_llamacpp.c
View File

@ -384,18 +384,18 @@ set_input(void *ctx, graph_execution_context exec_ctx, uint32_t index,
tensor *wasi_nn_tensor) tensor *wasi_nn_tensor)
{ {
struct LlamaContext *backend_ctx = (struct LlamaContext *)ctx; struct LlamaContext *backend_ctx = (struct LlamaContext *)ctx;
// tensor->data is the prompt string. ends with \0 // tensor->data is the prompt string.
char *prompt_text = (char *)wasi_nn_tensor->data.buf; char *prompt_text = (char *)wasi_nn_tensor->data.buf;
uint32_t prompt_text_len = wasi_nn_tensor->data.size;
#ifndef NDEBUG #ifndef NDEBUG
NN_DBG_PRINTF("--------------------------------------------------"); NN_DBG_PRINTF("--------------------------------------------------");
NN_DBG_PRINTF("prompt_text: %s", prompt_text); NN_DBG_PRINTF("prompt_text: %.*s", (int)prompt_text_len, prompt_text);
NN_DBG_PRINTF("--------------------------------------------------"); NN_DBG_PRINTF("--------------------------------------------------");
#endif #endif
// tokenize the prompt // tokenize the prompt
uint32_t n_token_max = llama_n_ctx(backend_ctx->ctx); uint32_t n_token_max = llama_n_ctx(backend_ctx->ctx);
uint32_t prompt_text_len = strlen(prompt_text);
if (backend_ctx->prompt == NULL) { if (backend_ctx->prompt == NULL) {
backend_ctx->prompt = calloc(n_token_max, sizeof(llama_token)); backend_ctx->prompt = calloc(n_token_max, sizeof(llama_token));