move fname to impl

Author: rgerganov

src/llama-mmap.cpp

@@ -159,7 +159,7 @@ struct llama_file::impl {
         }
     }
 #else
-    impl(const char * fname, const char * mode) {
+    impl(const char * fname, const char * mode) : fname(fname) {
         fp = ggml_fopen(fname, mode);
         if (fp == NULL) {
             throw std::runtime_error(format("failed to open %s: %s", fname, strerror(errno)));
@@ -237,15 +237,17 @@ struct llama_file::impl {
     }
 #endif
 
+    const char * fname;
     FILE * fp;
     size_t size;
 };
 
-llama_file::llama_file(const char * fname, const char * mode) : fname(fname), pimpl(std::make_unique<impl>(fname, mode)) {}
+llama_file::llama_file(const char * fname, const char * mode) : pimpl(std::make_unique<impl>(fname, mode)) {}
 llama_file::~llama_file() = default;
 
 size_t llama_file::tell() const { return pimpl->tell(); }
 size_t llama_file::size() const { return pimpl->size; }
+const char * llama_file::fname() const { return pimpl->fname; }
 
 int llama_file::file_id() const {
 #ifdef _WIN32

src/llama-mmap.h

@@ -19,6 +19,7 @@ struct llama_file {
     size_t tell() const;
     size_t size() const;
 
+    const char * fname() const;
     int file_id() const; // fileno overload
 
     void seek(size_t offset, int whence) const;
@@ -28,7 +29,6 @@ struct llama_file {
 
     void write_raw(const void * ptr, size_t len) const;
     void write_u32(uint32_t val) const;
-    const char * fname;
 
 private:
     struct impl;

src/llama-model-loader.cpp

@@ -1039,7 +1039,7 @@ bool llama_model_loader::load_all_data(
                 mmap_used.second = std::max(mmap_used.second, weight->offs + n_size);
             } else {
                 const auto & file = files.at(weight->idx);
-                if (!load_tensor(cur, file->fname, weight->offs, 0, n_size)) {
+                if (!load_tensor(cur, file->fname(), weight->offs, 0, n_size)) {
                     ggml_backend_tensor_set(cur, data, 0, n_size);
                 }
             }