llama.cpp provides LLM inference in C/C++. Prior to b3427, llama.cpp contains a null pointer dereference in ggufinitfrom_file. This vulnerability is fixed in b3427.