Paul Eggert authored
etags had undefined behavior if input files, lines, tags, etc., had more than INT_MAX bytes. Clean up the usage of integer types to fix the overflow errors I found. * admin/merge-gnulib (GNULIB_MODULES): Add mempcpy. * lib-src/etags.c: Include inttypes.h, intprops.h. (memcpyz): New function. Use it to simplify several occurrences of memcpy followed by storing a trailing '\0'. (xnew): Use xnmalloc, to catch overflow on integer multiplication. (xrnew): Change last arg to multiplier. The type is not needed. All callers changed. (node, lineno, charno, linecharno, invalidcharno, make_tag): (pfnote, add_node, number_len, C_symtype, lbz, Makefile_targets) (readline): Use intmax_t for line numbers and character positions, instead of int or long. (linebuffer, make_tag, pfnote, total_size_of_entries, put_entry) (in_word_set, C_symtype, token, cstack, pushclass_above): (popclass_above, write_classname, consider_token, C_entries) (Ruby_functions, Makefile_targets, Lua_functions, TeX_commands) (TeX_decode_env, erlang_func, erlang_attribute, erlang_atom) (substitute, regex_tag_multiline, nocase_tail, readline_interval) (readline, savenstr, concat, etags_getcwd, relative_filename) (linebuffer_setlen): Use ptrdiff_t for object sizes, instead of int or long or unsigned or size_t. (write_classname, C_entries): Avoid sprintf, as the result could exceed INT_MAX bytes and then behavior goes haywire. (main): Use int, instead of unsigned, for argv counts. (get_language_from_filename): Use bool for boolean. (Ruby_functions): Prefer strcpy to memcpy when copying "=". (linebuffer_setlen): Use ‘if’ instead of ‘while’. (memory_full, xnmalloc, xnrealloc): New functions. (xmalloc): Use memory_full, and take a ptrdiff_t instead of a size_t. (xrealloc): Remove; no longer needed. * lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate. * lib/mempcpy.c, m4/mempcpy.m4: New files, copied from Gnulib.