fixup

src/llama-mmap.h

@@ -21,7 +21,7 @@ struct llama_file {
     size_t tell() const;
     size_t size() const;
 
-    int file_id() const;
+    int file_id() const; // fileno overload
 
     void seek(size_t offset, int whence) const;
 

src/llama-model-loader.cpp

@@ -659,7 +659,7 @@ llama_model_loader::llama_model_loader(
 
             LLAMA_LOG_INFO("%s: additional %d GGUFs metadata loaded.\n",  __func__, n_split - 1);
         }
-    } else if (file) {
+    } else if (file != nullptr) {
         struct ggml_context * ctx = NULL;
         struct gguf_init_params params = {
             /*.no_alloc = */ true,

src/llama.cpp

@@ -891,9 +891,21 @@ static struct llama_model * llama_model_load_from_file_impl(
         std::vector<std::string> & splits,
         FILE * file,
         struct llama_model_params params) {
-    if (metadata == nullptr && path_model.empty() && !file) {
-        LLAMA_LOG_ERROR("%s: no model source provided\n", __func__);
-        return nullptr;
+    {
+        int n_sources_defined = 0;
+        if (metadata != nullptr) {
+            n_sources_defined++;
+        }
+        if (!path_model.empty()) {
+            n_sources_defined++;
+        }
+        if (file != nullptr) {
+            n_sources_defined++;
+        }
+        if (n_sources_defined != 1) {
+            LLAMA_LOG_ERROR("%s: exactly one out metadata, path_model, and file must be defined\n", __func__);
+            return nullptr;
+        }
     }
     ggml_time_init();
 

tests/CMakeLists.txt

@@ -240,7 +240,6 @@ llama_build_and_test(test-gguf.cpp)
 llama_build_and_test(test-backend-ops.cpp)
 
 llama_build_and_test(test-model-load-cancel.cpp LABEL "model")
-llama_build_and_test(test-model-load-fd.cpp     LABEL "model")
 llama_build_and_test(test-autorelease.cpp       LABEL "model")
 llama_build_and_test(test-backend-sampler.cpp   LABEL "model")
 

tests/test-model-load-fd.cpp

@@ -1,54 +0,0 @@
-#include "llama.h"
-#include "get-model.h"
-
-#include <cstdio>
-#include <cstdlib>
-
-#ifdef _WIN32
-int main(int /*argc*/, char ** /*argv*/) {
-    fprintf(stderr, "skipping on Windows\n");
-    return EXIT_SUCCESS;
-}
-#else
-#    include <fcntl.h>
-#    include <unistd.h>
-
-int main(int argc, char ** argv) {
-    auto * model_path = get_model_or_exit(argc, argv);
-
-    llama_backend_init();
-
-    const int fd = open(model_path, O_RDONLY);
-    if (fd < 0) {
-        fprintf(stderr, "failed to open %s\n", model_path);
-        return EXIT_FAILURE;
-    }
-
-    FILE * f = fdopen(dup(fd), "rb");
-    close(fd);
-    if (!f) {
-        fprintf(stderr, "failed to fdopen\n");
-        return EXIT_FAILURE;
-    }
-
-    auto params = llama_model_default_params();
-    params.use_mmap = true;
-    params.vocab_only = true;
-
-    struct llama_model * model = llama_model_load_from_file_ptr(f, params);
-    fclose(f);
-
-    if (model == nullptr) {
-        fprintf(stderr, "load from file pointer failed\n");
-        return EXIT_FAILURE;
-    }
-
-    const int n_vocab = llama_vocab_n_tokens(llama_model_get_vocab(model));
-    fprintf(stderr, "loaded %d tokens via file pointer\n", n_vocab);
-
-    llama_model_free(model);
-    llama_backend_free();
-
-    return n_vocab > 0 ? EXIT_SUCCESS : EXIT_FAILURE;
-}
-#endif // _WIN32