[ggml/llama] fix warns as errors in CI build and temp set force C23

Author: GermanAizek

Makefile

@@ -257,9 +257,9 @@ endif
 # Compile flags
 #
 
-# keep standard at C11 and C++17
+# keep standard at C23 and C++17
 MK_CPPFLAGS  = -Iggml/include -Iggml/src -Iinclude -Isrc -Icommon -DGGML_USE_CPU
-MK_CFLAGS    = -std=c11   -fPIC
+MK_CFLAGS    = -std=c23   -fPIC
 MK_CXXFLAGS  = -std=c++17 -fPIC
 MK_NVCCFLAGS = -std=c++17
 
@@ -393,8 +393,8 @@ MK_CXXFLAGS += \
 	-Wmissing-noreturn
 
 ifeq ($(LLAMA_FATAL_WARNINGS),1)
-	MK_CFLAGS   += -Werror
-	MK_CXXFLAGS += -Werror
+	MK_CFLAGS   += -Werror -pedantic-errors -Wno-pedantic
+	MK_CXXFLAGS += -Werror -pedantic-errors -Wno-pedantic
 endif
 
 # this version of Apple ld64 is buggy

ggml/CMakeLists.txt

@@ -216,7 +216,7 @@ option(GGML_BUILD_EXAMPLES "ggml: build examples" ${GGML_STANDALONE})
 # dependencies
 #
 
-set(CMAKE_C_STANDARD 11)
+set(CMAKE_C_STANDARD 23)
 set(CMAKE_C_STANDARD_REQUIRED true)
 
 set(CMAKE_CXX_STANDARD 17)

ggml/src/CMakeLists.txt

@@ -28,7 +28,9 @@ endif()
 if (GGML_FATAL_WARNINGS)
     if (CMAKE_CXX_COMPILER_ID MATCHES "GNU" OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
         list(APPEND C_FLAGS   -Werror)
+        list(APPEND C_FLAGS   -pedantic-errors -Wno-pedantic)
         list(APPEND CXX_FLAGS -Werror)
+        list(APPEND CXX_FLAGS -pedantic-errors -Wno-pedantic)
     elseif (CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
         add_compile_options(/WX)
     endif()

ggml/src/ggml-quants.c

@@ -4996,7 +4996,7 @@ static bool validate_fp16(ggml_fp16_t f, size_t i) {
     }
 
 bool ggml_validate_row_data(enum ggml_type type, const void * data, size_t nbytes) {
-    if (type < 0 || type >= GGML_TYPE_COUNT) {
+    if (type >= GGML_TYPE_COUNT) {
         fprintf(stderr, "%s: invalid type %d\n", __func__, type);
         return false;
     }

ggml/src/ggml.c

@@ -1565,7 +1565,7 @@ static struct ggml_tensor * ggml_new_tensor_impl(
         struct ggml_tensor  * view_src,
         size_t                view_offs) {
 
-    GGML_ASSERT(type >= 0 && type < GGML_TYPE_COUNT);
+    GGML_ASSERT(type > 0 && type < GGML_TYPE_COUNT);
     GGML_ASSERT(n_dims >= 1 && n_dims <= GGML_MAX_DIMS);
 
     // find the base tensor and absolute offset

ggml/src/gguf.cpp

@@ -546,7 +546,7 @@ struct gguf_context * gguf_init_from_file_impl(FILE * file, struct gguf_init_par
             ok = ok && gr.read(info.t.type);
 
             // check that tensor type is within defined range
-            if (info.t.type < 0 || info.t.type >= GGML_TYPE_COUNT) {
+            if (info.t.type >= GGML_TYPE_COUNT) {
                 fprintf(stderr, "%s: tensor '%s' has invalid ggml type %d (%s)\n",
                     __func__, info.t.name, info.t.type, ggml_type_name(info.t.type));
                 ok = false;

src/llama-vocab.cpp

@@ -826,9 +826,9 @@ struct llm_tokenizer_ugm_session {
         }
 
         // initialize score_sum to -FLT_MAX so it will be always lower than sums of token scores
-        std::vector<struct best_tokenization> tokenization_results(input_len + 1, {vocab.token_unk(), 0, -FLT_MAX});
+        std::vector<struct best_tokenization> tokenization_results(input_len + 1, {0, vocab.token_unk(), -FLT_MAX});
         // at the beginning tokenization score is zero
-        tokenization_results[0] = { vocab.token_unk(), 0, 0 };
+        tokenization_results[0] = {  0, vocab.token_unk(), 0 };
 
         for (size_t input_offset = 0; input_offset < input_len;) {
             size_t prefix_offset = input_offset;
@@ -858,7 +858,7 @@ struct llm_tokenizer_ugm_session {
                     const double challenger_score = current_best.score_sum + token_score;
                     struct best_tokenization & current_champ = tokenization_results[prefix_offset];
                     if (challenger_score > current_champ.score_sum) {
-                        struct best_tokenization challenger = { token_id, input_offset, (float) challenger_score };
+                        struct best_tokenization challenger = { input_offset, token_id, (float) challenger_score };
                         current_champ = challenger;
                     }
                 }
@@ -872,7 +872,7 @@ struct llm_tokenizer_ugm_session {
                 prefix_offset = input_offset + n_utf8_code_units;
                 struct best_tokenization & current_champ = tokenization_results[prefix_offset];
                 if (challenger_score > current_champ.score_sum) {
-                    struct best_tokenization challenger = { vocab.token_unk(), input_offset, (float) challenger_score };
+                    struct best_tokenization challenger = { input_offset, vocab.token_unk(), (float) challenger_score };
                     current_champ = challenger;
                 }
             }