Add bounds checking for output tensor buffer in wasi-nn llama.cpp

The get_output function copies LLM output into output_tensor->buf
without checking against output_tensor->size, allowing writes
past the buffer when the model generates output longer than the
caller-provided buffer. Add size checks for both the metadata
path and the token output loop.

Author: sumleo

core/iwasm/libraries/wasi-nn/src/wasi_nn_llamacpp.c

@@ -623,8 +623,11 @@ get_output(void *ctx, graph_execution_context exec_ctx, uint32_t index,
             printf("%s\n", output_metadata);
         }
 
-        memcpy(output_tensor->buf, output_metadata, strlen(output_metadata));
-        *output_tensor_size = strlen(output_metadata);
+        size_t metadata_len = strlen(output_metadata);
+        if (metadata_len > output_tensor->size)
+            metadata_len = output_tensor->size;
+        memcpy(output_tensor->buf, output_metadata, metadata_len);
+        *output_tensor_size = metadata_len;
         return success;
     }
 
@@ -643,8 +646,11 @@ get_output(void *ctx, graph_execution_context exec_ctx, uint32_t index,
             printf("%s", buf);
         }
 
-        memcpy(output_tensor->buf + end_pos, buf, strlen(buf));
-        end_pos += strlen(buf);
+        size_t piece_len = strlen(buf);
+        if (end_pos + piece_len > output_tensor->size)
+            break;
+        memcpy(output_tensor->buf + end_pos, buf, piece_len);
+        end_pos += piece_len;
     }
 
     if (backend_ctx->config.stream_stdout) {