quantize : fix missing 'noreturn' (-Wmissing-noreturn)

CMakeLists.txt

@@ -431,6 +431,7 @@ if (LLAMA_ALL_WARNINGS)
         set(cxx_flags
             ${warning_flags}
             -Wmissing-declarations
+            -Wmissing-noreturn
         )
         if (CMAKE_CXX_COMPILER_ID MATCHES "Clang")  # clang++ only
             set(cxx_flags ${cxx_flags} -Wmissing-prototypes)

Makefile

@@ -176,7 +176,7 @@ endif # LLAMA_DISABLE_LOGS
 WARN_FLAGS    = -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function
 MK_CFLAGS    += $(WARN_FLAGS) -Wdouble-promotion -Wshadow -Wstrict-prototypes -Wpointer-arith -Wmissing-prototypes \
 				-Werror=implicit-int
-MK_CXXFLAGS  += $(WARN_FLAGS) -Wmissing-declarations
+MK_CXXFLAGS  += $(WARN_FLAGS) -Wmissing-declarations -Wmissing-noreturn
 
 # TODO(cebtenzzre): remove this once PR #2632 gets merged
 TTFS_CXXFLAGS = $(CXXFLAGS) -Wno-missing-declarations

examples/quantize/quantize.cpp

@@ -71,6 +71,7 @@ static bool try_parse_ftype(const std::string & ftype_str_in, llama_ftype & ftyp
 // usage:
 //  ./quantize [--allow-requantize] [--leave-output-tensor] models/llama/ggml-model.gguf [models/llama/ggml-model-quant.gguf] type [nthreads]
 //
+[[noreturn]]
 static void usage(const char * executable) {
     printf("usage: %s [--help] [--allow-requantize] [--leave-output-tensor] model-f32.gguf [model-quant.gguf] type [nthreads]\n\n", executable);
     printf("  --allow-requantize: Allows requantizing tensors that have already been quantized. Warning: This can severely reduce quality compared to quantizing from 16bit or 32bit\n");