X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=format_check.c;h=4c1d6701f9067cfff8f4c8dcb01654c9de27383d;hb=287127a9ff31adbada4945a2e82cd4ab57786b85;hp=d666d1a7eab39eea864937e1600c023f23a2143e;hpb=1c8c56918e13a7fc27f635c92d1500eb856469b1;p=cparser

diff --git a/format_check.c b/format_check.c
index d666d1a..4c1d670 100644
--- a/format_check.c
+++ b/format_check.c
@@ -1,6 +1,6 @@
 /*
  * This file is part of cparser.
- * Copyright (C) 2007-2008 Matthias Braun <matze@braunis.de>
+ * Copyright (C) 2007-2009 Matthias Braun <matze@braunis.de>
  *
  * This program is free software; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License
@@ -17,13 +17,18 @@
  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
  * 02111-1307, USA.
  */
+#include <config.h>
+
 #include <ctype.h>
-#include <wctype.h>
 
+#include "adt/strutil.h"
+#include "adt/util.h"
 #include "format_check.h"
 #include "symbol_t.h"
 #include "ast_t.h"
+#include "entity_t.h"
 #include "diagnostic.h"
+#include "parser.h"
 #include "types.h"
 #include "type_t.h"
 #include "warning.h"
@@ -59,6 +64,13 @@ typedef enum format_length_modifier_t {
 	FMT_MOD_I64
 } format_length_modifier_t;
 
+typedef struct format_spec_t {
+	const char    *name;     /**< name of the function */
+	format_kind_t  fmt_kind; /**< kind */
+	unsigned       fmt_idx;  /**< index of the format string */
+	unsigned       arg_idx;  /**< index of the first argument */
+} format_spec_t;
+
 static const char* get_length_modifier_name(const format_length_modifier_t mod)
 {
 	static const char* const names[] = {
@@ -78,137 +90,83 @@ static const char* get_length_modifier_name(const format_length_modifier_t mod)
 		[FMT_MOD_I32]  = "I32",
 		[FMT_MOD_I64]  = "I64"
 	};
-	assert(mod < sizeof(names) / sizeof(*names));
+	assert((size_t)mod < lengthof(names));
 	return names[mod];
 }
 
 static void warn_invalid_length_modifier(const source_position_t *pos,
                                          const format_length_modifier_t mod,
-                                         const wchar_rep_t conversion)
+                                         const utf32 conversion)
 {
-	warningf(pos,
-		"invalid length modifier '%s' for conversion specifier '%%%c'",
-		get_length_modifier_name(mod), conversion
-	);
-}
-
-typedef struct vchar_t vchar_t;
-struct vchar_t {
-	const void *string;   /**< the string */
-	size_t     position;  /**< current position */
-	size_t     size;      /**< size of the string */
-
-	/** return the first character of the string and setthe position to 0. */
-	unsigned (*first)(vchar_t *self);
-	/** return the next character of the string */
-	unsigned (*next)(vchar_t *self);
-	/** return non_zero if the given character is a digit */
-	int (*is_digit)(unsigned vchar);
-};
-
-static unsigned string_first(vchar_t *self) {
-	self->position = 0;
-	const string_t *string = self->string;
-	return string->begin[0];
-}
-
-static unsigned string_next(vchar_t *self) {
-	++self->position;
-	const string_t *string = self->string;
-	return string->begin[self->position];
-}
-
-static int string_isdigit(unsigned vchar) {
-	return isdigit(vchar);
-}
-
-static unsigned wstring_first(vchar_t *self) {
-	self->position = 0;
-	const wide_string_t *wstring = self->string;
-	return wstring->begin[0];
-}
-
-static unsigned wstring_next(vchar_t *self) {
-	++self->position;
-	const wide_string_t *wstring = self->string;
-	return wstring->begin[self->position];
-}
-
-static int wstring_isdigit(unsigned vchar) {
-	return iswdigit(vchar);
-}
-
-static bool atend(vchar_t *self) {
-	return self->position + 1 == self->size;
+	char const *const lmod = get_length_modifier_name(mod);
+	warningf(WARN_FORMAT, pos, "invalid length modifier '%s' for conversion specifier '%%%c'", lmod, conversion);
 }
 
 /**
- * Check printf-style format.
+ * Check printf-style format. Returns number of expected arguments.
  */
-static void check_format_arguments(const call_argument_t *arg, unsigned idx_fmt,
-		unsigned idx_param)
+static int internal_check_printf_format(const expression_t *fmt_expr,
+                                        const call_argument_t *arg,
+                                        const format_spec_t *spec)
 {
-	const call_argument_t *fmt_arg;
-	unsigned idx = 0;
-
-	/* find format arg */
-	for(idx = 0; idx < idx_fmt; ++idx)
-		arg = arg->next;
-	fmt_arg = arg;
-
-	const expression_t *fmt_expr = fmt_arg->expression;
-	if (fmt_expr->kind == EXPR_UNARY_CAST_IMPLICIT) {
+	while (fmt_expr->kind == EXPR_UNARY_CAST) {
 		fmt_expr = fmt_expr->unary.value;
 	}
 
-	vchar_t vchar;
-	if (fmt_expr->kind == EXPR_WIDE_STRING_LITERAL) {
-		vchar.string   = &fmt_expr->wide_string.value;
-		vchar.size     = fmt_expr->wide_string.value.size;
-		vchar.first    = wstring_first;
-		vchar.next     = wstring_next;
-		vchar.is_digit = wstring_isdigit;
-	} else if (fmt_expr->kind == EXPR_STRING_LITERAL) {
-		vchar.string   = &fmt_expr->string.value;
-		vchar.size     = fmt_expr->string.value.size;
-		vchar.first    = string_first;
-		vchar.next     = string_next;
-		vchar.is_digit = string_isdigit;
-	} else {
-		return;
+	/*
+	 * gettext results in expressions like (X ? "format_string" : Y)
+	 * we assume the left part is the format string
+	 */
+	if (fmt_expr->kind == EXPR_CONDITIONAL) {
+		conditional_expression_t const *const c = &fmt_expr->conditional;
+		expression_t             const *      t = c->true_expression;
+		if (t == NULL)
+			t = c->condition;
+		int const nt = internal_check_printf_format(t,                   arg, spec);
+		int const nf = internal_check_printf_format(c->false_expression, arg, spec);
+		return nt > nf ? nt : nf;
 	}
-	/* find the real args */
-	for(; idx < idx_param; ++idx)
-		arg = arg->next;
+
+	if (fmt_expr->kind != EXPR_STRING_LITERAL)
+		return -1;
+
+	const char *string = fmt_expr->string_literal.value.begin;
+	size_t      size   = fmt_expr->string_literal.value.size;
+	const char *c      = string;
 
 	const source_position_t *pos = &fmt_expr->base.source_position;
-	unsigned fmt = vchar.first(&vchar);
-	for (; fmt != '\0'; fmt = vchar.next(&vchar)) {
+	unsigned num_fmt  = 0;
+	unsigned num_args = 0;
+	char     fmt;
+	for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
 		if (fmt != '%')
 			continue;
-		fmt = vchar.next(&vchar);
+		fmt = *(++c);
 
 		if (fmt == '%')
 			continue;
 
+		++num_fmt;
+		++num_args;
+
 		format_flags_t fmt_flags = FMT_FLAG_NONE;
 		if (fmt == '0') {
-			fmt = vchar.next(&vchar);
+			fmt = *(++c);
 			fmt_flags |= FMT_FLAG_ZERO;
 		}
 
 		/* argument selector or minimum field width */
-		if (vchar.is_digit(fmt)) {
+		if (isdigit(fmt)) {
 			do {
-				fmt = vchar.next(&vchar);
-			} while (vchar.is_digit(fmt));
+				fmt = *(++c);
+			} while (isdigit(fmt));
 
 			/* digit string was ... */
 			if (fmt == '$') {
 				/* ... argument selector */
 				fmt_flags = FMT_FLAG_NONE; /* reset possibly set 0-flag */
 				/* TODO implement */
-				return;
+				return -1;
 			}
 			/* ... minimum field width */
 		} else {
@@ -223,14 +181,14 @@ static void check_format_arguments(const call_argument_t *arg, unsigned idx_fmt,
 
 					case ' ':
 						if (fmt_flags & FMT_FLAG_PLUS) {
-							warningf(pos, "' ' is overridden by prior '+' in conversion specification");
+							warningf(WARN_FORMAT, pos, "' ' is overridden by prior '+' in conversion specification %u", num_fmt);
 						}
 						flag = FMT_FLAG_SPACE;
 						break;
 
 					case '+':
 						if (fmt_flags & FMT_FLAG_SPACE) {
-							warningf(pos, "'+' overrides prior ' ' in conversion specification");
+							warningf(WARN_FORMAT, pos, "'+' overrides prior ' ' in conversion specification %u", num_fmt);
 						}
 						flag = FMT_FLAG_PLUS;
 						break;
@@ -238,50 +196,56 @@ static void check_format_arguments(const call_argument_t *arg, unsigned idx_fmt,
 					default: goto break_fmt_flags;
 				}
 				if (fmt_flags & flag) {
-					warningf(pos, "repeated flag '%c' in conversion specification", (char)fmt);
+					warningf(WARN_FORMAT, pos, "repeated flag '%c' in conversion specification %u", (char)fmt, num_fmt);
 				}
 				fmt_flags |= flag;
-				fmt = vchar.next(&vchar);
+				fmt = *(++c);
 			}
 break_fmt_flags:
 
 			/* minimum field width */
 			if (fmt == '*') {
-				fmt = vchar.next(&vchar);
+				++num_args;
+				fmt = *(++c);
 				if (arg == NULL) {
-					warningf(pos, "missing argument for '*' field width in conversion specification");
-					return;
+					warningf(WARN_FORMAT, pos, "missing argument for '*' field width in conversion specification %u", num_fmt);
+					return -1;
 				}
 				const type_t *const arg_type = arg->expression->base.type;
 				if (arg_type != type_int) {
-					warningf(pos, "argument for '*' field width in conversion specification is not an 'int', but an '%T'", arg_type);
+					warningf(WARN_FORMAT, pos, "argument for '*' field width in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 				}
 				arg = arg->next;
 			} else {
-				while (vchar.is_digit(fmt)) {
-					fmt = vchar.next(&vchar);
+				while (isdigit(fmt)) {
+					fmt = *(++c);
 				}
 			}
 		}
 
 		/* precision */
 		if (fmt == '.') {
-			fmt = vchar.next(&vchar);
+			if (fmt_flags & FMT_FLAG_ZERO) {
+				warningf(WARN_FORMAT, pos, "'0' flag ignored with precision in conversion specification %u", num_fmt);
+			}
+
+			++num_args;
+			fmt = *(++c);
 			if (fmt == '*') {
-				fmt = vchar.next(&vchar);
+				fmt = *(++c);
 				if (arg == NULL) {
-					warningf(pos, "missing argument for '*' precision in conversion specification");
-					return;
+					warningf(WARN_FORMAT, pos, "missing argument for '*' precision in conversion specification %u", num_fmt);
+					return -1;
 				}
 				const type_t *const arg_type = arg->expression->base.type;
 				if (arg_type != type_int) {
-					warningf(pos, "argument for '*' precision in conversion specification is not an 'int', but an '%T'", arg_type);
+					warningf(WARN_FORMAT, pos, "argument for '*' precision in conversion specification %u is not an 'int', but an '%T'", num_fmt, arg_type);
 				}
 				arg = arg->next;
 			} else {
 				/* digit string may be omitted */
-				while (vchar.is_digit(fmt)) {
-					fmt = vchar.next(&vchar);
+				while (isdigit(fmt)) {
+					fmt = *(++c);
 				}
 			}
 		}
@@ -290,9 +254,9 @@ break_fmt_flags:
 		format_length_modifier_t fmt_mod;
 		switch (fmt) {
 			case 'h':
-				fmt = vchar.next(&vchar);
+				fmt = *(++c);
 				if (fmt == 'h') {
-					fmt = vchar.next(&vchar);
+					fmt = *(++c);
 					fmt_mod = FMT_MOD_hh;
 				} else {
 					fmt_mod = FMT_MOD_h;
@@ -300,48 +264,48 @@ break_fmt_flags:
 				break;
 
 			case 'l':
-				fmt = vchar.next(&vchar);
+				fmt = *(++c);
 				if (fmt == 'l') {
-					fmt = vchar.next(&vchar);
+					fmt = *(++c);
 					fmt_mod = FMT_MOD_ll;
 				} else {
 					fmt_mod = FMT_MOD_l;
 				}
 				break;
 
-			case 'L': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_L;    break;
-			case 'j': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_j;    break;
-			case 't': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_t;    break;
-			case 'z': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_z;    break;
-			case 'q': fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_q;    break;
+			case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L;    break;
+			case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j;    break;
+			case 't': fmt = *(++c); fmt_mod = FMT_MOD_t;    break;
+			case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z;    break;
+			case 'q': fmt = *(++c); fmt_mod = FMT_MOD_q;    break;
 			/* microsoft mode */
 			case 'w':
 				if (c_mode & _MS) {
-					fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_w;
+					fmt = *(++c); fmt_mod = FMT_MOD_w;
 				} else {
 					fmt_mod = FMT_MOD_NONE;
 				}
 				break;
 			case 'I':
 				if (c_mode & _MS) {
-					fmt = vchar.next(&vchar); fmt_mod = FMT_MOD_I;
+					fmt = *(++c); fmt_mod = FMT_MOD_I;
 					if (fmt == '3') {
-						fmt = vchar.next(&vchar);
+						fmt = *(++c);
 						if (fmt == '2') {
-							fmt = vchar.next(&vchar);
+							fmt = *(++c);
 							fmt_mod = FMT_MOD_I32;
 						} else {
 							/* rewind */
-							--vchar.position;
+							fmt = *(--c);
 						}
 					} else if (fmt == '6') {
-						fmt = vchar.next(&vchar);
+						fmt = *(++c);
 						if (fmt == '4') {
-							fmt = vchar.next(&vchar);
+							fmt = *(++c);
 							fmt_mod = FMT_MOD_I64;
 						} else {
 							/* rewind */
-							--vchar.position;
+							fmt = *(--c);
 						}
 					}
 				} else {
@@ -354,7 +318,7 @@ break_fmt_flags:
 		}
 
 		if (fmt == '\0') {
-			warningf(pos, "dangling %% in format string");
+			warningf(WARN_FORMAT, pos, "dangling %% in format string");
 			break;
 		}
 
@@ -365,17 +329,17 @@ break_fmt_flags:
 			case 'd':
 			case 'i':
 				switch (fmt_mod) {
-					case FMT_MOD_NONE: expected_type = type_int;       break;
-					case FMT_MOD_hh:   expected_type = type_int;       break; /* TODO promoted signed char */
-					case FMT_MOD_h:    expected_type = type_int;       break; /* TODO promoted short */
-					case FMT_MOD_l:    expected_type = type_long;      break;
-					case FMT_MOD_ll:   expected_type = type_long_long; break;
-					case FMT_MOD_j:    expected_type = type_intmax_t;  break;
-					case FMT_MOD_z:    expected_type = type_ssize_t;   break;
-					case FMT_MOD_t:    expected_type = type_ptrdiff_t; break;
-					case FMT_MOD_I:    expected_type = type_ptrdiff_t; break;
-					case FMT_MOD_I32:  expected_type = type_int32;     break;
-					case FMT_MOD_I64:  expected_type = type_int64;     break;
+					case FMT_MOD_NONE: expected_type = type_int;         break;
+					case FMT_MOD_hh:   expected_type = type_signed_char; break;
+					case FMT_MOD_h:    expected_type = type_short;       break;
+					case FMT_MOD_l:    expected_type = type_long;        break;
+					case FMT_MOD_ll:   expected_type = type_long_long;   break;
+					case FMT_MOD_j:    expected_type = type_intmax_t;    break;
+					case FMT_MOD_z:    expected_type = type_ssize_t;     break;
+					case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
+					case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
+					case FMT_MOD_I32:  expected_type = type_int32;       break;
+					case FMT_MOD_I64:  expected_type = type_int64;       break;
 
 					default:
 						warn_invalid_length_modifier(pos, fmt_mod, fmt);
@@ -389,15 +353,14 @@ break_fmt_flags:
 			case 'x':
 				allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_HASH | FMT_FLAG_ZERO;
 				goto eval_fmt_mod_unsigned;
-				break;
 
 			case 'u':
 				allowed_flags = FMT_FLAG_MINUS | FMT_FLAG_ZERO;
 eval_fmt_mod_unsigned:
 				switch (fmt_mod) {
 					case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
-					case FMT_MOD_hh:   expected_type = type_int;                break; /* TODO promoted unsigned char */
-					case FMT_MOD_h:    expected_type = type_int;                break; /* TODO promoted unsigned short */
+					case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
+					case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
 					case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
 					case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
 					case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
@@ -463,7 +426,7 @@ eval_fmt_mod_unsigned:
 				}
 				expected_type = type_wchar_t_ptr;
 				expected_qual = TYPE_QUALIFIER_CONST;
-				allowed_flags = FMT_FLAG_NONE;
+				allowed_flags = FMT_FLAG_MINUS;
 				break;
 
 			case 's':
@@ -477,7 +440,7 @@ eval_fmt_mod_unsigned:
 						goto next_arg;
 				}
 				expected_qual = TYPE_QUALIFIER_CONST;
-				allowed_flags = FMT_FLAG_NONE;
+				allowed_flags = FMT_FLAG_MINUS;
 				break;
 
 			case 'p':
@@ -508,25 +471,46 @@ eval_fmt_mod_unsigned:
 				break;
 
 			default:
-				warningf(pos, "encountered unknown conversion specifier '%%%C'", (wint_t)fmt);
+				warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at position %u", fmt, num_fmt);
+				if (arg == NULL) {
+					goto too_few_args;
+				}
 				goto next_arg;
 		}
 
-		if ((fmt_flags & ~allowed_flags) != 0) {
-			/* TODO better warning message text */
-			warningf(pos, "invalid format flags in conversion specification");
+		format_flags_t wrong_flags = fmt_flags & ~allowed_flags;
+		if (wrong_flags != 0) {
+			char  wrong[8];
+			char *p = wrong;
+			if (wrong_flags & FMT_FLAG_HASH)  *p++ = '#';
+			if (wrong_flags & FMT_FLAG_ZERO)  *p++ = '0';
+			if (wrong_flags & FMT_FLAG_MINUS) *p++ = '-';
+			if (wrong_flags & FMT_FLAG_SPACE) *p++ = ' ';
+			if (wrong_flags & FMT_FLAG_PLUS)  *p++ = '+';
+			if (wrong_flags & FMT_FLAG_TICK)  *p++ = '\'';
+			*p = '\0';
+
+			warningf(WARN_FORMAT, pos, "invalid format flags \"%s\" in conversion specification %%%c at position %u", wrong, fmt, num_fmt);
 		}
 
 		if (arg == NULL) {
-			warningf(pos, "too few arguments for format string");
-			return;
+too_few_args:
+			warningf(WARN_FORMAT, pos, "too few arguments for format string");
+			return -1;
 		}
 
-		{	/* create a scope here to prevent warning about the jump to next_arg */
+		{ /* create a scope here to prevent warning about the jump to next_arg */
 			type_t *const arg_type           = arg->expression->base.type;
+			type_t *const arg_skip           = skip_typeref(arg_type);
 			type_t *const expected_type_skip = skip_typeref(expected_type);
+
+			if (fmt == 'p') {
+				/* allow any pointer type for %p, not just void */
+				if (is_type_pointer(arg_skip))
+					goto next_arg;
+			}
+
 			if (is_type_pointer(expected_type_skip)) {
-				type_t *const arg_skip = skip_typeref(arg_type);
 				if (is_type_pointer(arg_skip)) {
 					type_t *const exp_to = skip_typeref(expected_type_skip->pointer.points_to);
 					type_t *const arg_to = skip_typeref(arg_skip->pointer.points_to);
@@ -535,34 +519,422 @@ eval_fmt_mod_unsigned:
 						goto next_arg;
 					}
 				}
+			} else if (get_unqualified_type(arg_skip) == expected_type_skip) {
+				goto next_arg;
+			} else if (arg->expression->kind == EXPR_UNARY_CAST) {
+				expression_t const *const expr        = arg->expression->unary.value;
+				type_t             *const unprom_type = skip_typeref(expr->base.type);
+				if (get_unqualified_type(unprom_type) == expected_type_skip) {
+					goto next_arg;
+				}
+				if (expected_type_skip == type_unsigned_int && !is_type_signed(unprom_type)) {
+					goto next_arg;
+				}
+			}
+			if (is_type_valid(arg_skip)) {
+				source_position_t const *const apos = &arg->expression->base.source_position;
+				char              const *const mod  = get_length_modifier_name(fmt_mod);
+				warningf(WARN_FORMAT, apos, "conversion '%%%s%c' at position %u specifies type '%T' but the argument has type '%T'", mod, (char)fmt, num_fmt, expected_type, arg_type);
+			}
+		}
+next_arg:
+		arg = arg->next;
+	}
+	assert(fmt == '\0');
+	if (c+1 < string + size) {
+		warningf(WARN_FORMAT, pos, "format string contains '\\0'");
+	}
+	return num_args;
+}
+
+/**
+ * Check printf-style format.
+ */
+static void check_printf_format(call_argument_t const *arg,
+                                format_spec_t const *const spec)
+{
+	/* find format arg */
+	size_t idx = 0;
+	for (; idx < spec->fmt_idx; ++idx) {
+		if (arg == NULL)
+			return;
+		arg = arg->next;
+	}
+
+	expression_t const *const fmt_expr = arg->expression;
+
+	/* find the real args */
+	for (; idx < spec->arg_idx && arg != NULL; ++idx)
+		arg = arg->next;
+
+	int const num_fmt = internal_check_printf_format(fmt_expr, arg, spec);
+	if (num_fmt < 0)
+		return;
+
+	size_t num_args = 0;
+	for (; arg != NULL; arg = arg->next)
+		++num_args;
+	if (num_args > (size_t)num_fmt) {
+		source_position_t const *const pos = &fmt_expr->base.source_position;
+		warningf(WARN_FORMAT, pos, "%u argument%s but only %u format specifier%s", num_args, num_args != 1 ? "s" : "", num_fmt,  num_fmt  != 1 ? "s" : "");
+	}
+}
+
+/**
+ * Check scanf-style format.
+ */
+static void check_scanf_format(const call_argument_t *arg,
+                               const format_spec_t *spec)
+{
+	/* find format arg */
+	unsigned idx = 0;
+	for (; idx < spec->fmt_idx; ++idx) {
+		if (arg == NULL)
+			return;
+		arg = arg->next;
+	}
+
+	const expression_t *fmt_expr = arg->expression;
+	if (fmt_expr->kind == EXPR_UNARY_CAST) {
+		fmt_expr = fmt_expr->unary.value;
+	}
+
+	if (fmt_expr->kind != EXPR_STRING_LITERAL)
+		return;
+
+	const char *string = fmt_expr->string_literal.value.begin;
+	size_t      size   = fmt_expr->string_literal.value.size;
+	const char *c      = string;
+
+	/* find the real args */
+	for (; idx < spec->arg_idx && arg != NULL; ++idx)
+		arg = arg->next;
+
+	const source_position_t *pos = &fmt_expr->base.source_position;
+	unsigned num_fmt = 0;
+	char     fmt;
+	for (fmt = *c; fmt != '\0'; fmt = *(++c)) {
+		if (fmt != '%')
+			continue;
+		fmt = *(++c);
+		if (fmt == '%')
+			continue;
+
+		++num_fmt;
+
+		bool suppress_assignment = false;
+		if (fmt == '*') {
+			fmt = *++c;
+			suppress_assignment = true;
+		}
+
+		size_t width = 0;
+		if ('0' <= fmt && fmt <= '9') {
+			do {
+				width = width * 10 + (fmt - '0');
+				fmt   = *++c;
+			} while ('0' <= fmt && fmt <= '9');
+			if (width == 0) {
+				warningf(WARN_FORMAT, pos, "field width is zero at format %u", num_fmt);
+			}
+		}
+
+		/* look for length modifiers */
+		format_length_modifier_t fmt_mod = FMT_MOD_NONE;
+		switch (fmt) {
+		case 'h':
+			fmt = *(++c);
+			if (fmt == 'h') {
+				fmt = *(++c);
+				fmt_mod = FMT_MOD_hh;
+			} else {
+				fmt_mod = FMT_MOD_h;
+			}
+			break;
+
+		case 'l':
+			fmt = *(++c);
+			if (fmt == 'l') {
+				fmt = *(++c);
+				fmt_mod = FMT_MOD_ll;
 			} else {
-				if (get_unqualified_type(skip_typeref(arg_type)) == expected_type_skip) {
+				fmt_mod = FMT_MOD_l;
+			}
+			break;
+
+		case 'L': fmt = *(++c); fmt_mod = FMT_MOD_L; break;
+		case 'j': fmt = *(++c); fmt_mod = FMT_MOD_j; break;
+		case 't': fmt = *(++c); fmt_mod = FMT_MOD_t; break;
+		case 'z': fmt = *(++c); fmt_mod = FMT_MOD_z; break;
+		/* microsoft mode */
+		case 'w':
+			if (c_mode & _MS) {
+				fmt = *(++c);
+				fmt_mod = FMT_MOD_w;
+			}
+			break;
+		case 'I':
+			if (c_mode & _MS) {
+				fmt = *(++c);
+				fmt_mod = FMT_MOD_I;
+				if (fmt == '3') {
+					fmt = *(++c);
+					if (fmt == '2') {
+						fmt = *(++c);
+						fmt_mod = FMT_MOD_I32;
+					} else {
+						/* rewind */
+						fmt = *(--c);
+					}
+				} else if (fmt == '6') {
+					fmt = *(++c);
+					if (fmt == '4') {
+						fmt = *(++c);
+						fmt_mod = FMT_MOD_I64;
+					} else {
+						/* rewind */
+						fmt = *(--c);
+					}
+				}
+			}
+			break;
+		}
+
+		if (fmt == '\0') {
+			warningf(WARN_FORMAT, pos, "dangling %% with conversion specififer in format string");
+			break;
+		}
+
+		type_t *expected_type;
+		switch (fmt) {
+		case 'd':
+		case 'i':
+			switch (fmt_mod) {
+			case FMT_MOD_NONE: expected_type = type_int;         break;
+			case FMT_MOD_hh:   expected_type = type_signed_char; break;
+			case FMT_MOD_h:    expected_type = type_short;       break;
+			case FMT_MOD_l:    expected_type = type_long;        break;
+			case FMT_MOD_ll:   expected_type = type_long_long;   break;
+			case FMT_MOD_j:    expected_type = type_intmax_t;    break;
+			case FMT_MOD_z:    expected_type = type_ssize_t;     break;
+			case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
+			case FMT_MOD_I:    expected_type = type_ptrdiff_t;   break;
+			case FMT_MOD_I32:  expected_type = type_int32;       break;
+			case FMT_MOD_I64:  expected_type = type_int64;       break;
+
+			default:
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			break;
+
+		case 'o':
+		case 'X':
+		case 'x':
+		case 'u':
+			switch (fmt_mod) {
+			case FMT_MOD_NONE: expected_type = type_unsigned_int;       break;
+			case FMT_MOD_hh:   expected_type = type_unsigned_char;      break;
+			case FMT_MOD_h:    expected_type = type_unsigned_short;     break;
+			case FMT_MOD_l:    expected_type = type_unsigned_long;      break;
+			case FMT_MOD_ll:   expected_type = type_unsigned_long_long; break;
+			case FMT_MOD_j:    expected_type = type_uintmax_t;          break;
+			case FMT_MOD_z:    expected_type = type_size_t;             break;
+			case FMT_MOD_t:    expected_type = type_uptrdiff_t;         break;
+			case FMT_MOD_I:    expected_type = type_size_t;             break;
+			case FMT_MOD_I32:  expected_type = type_unsigned_int32;     break;
+			case FMT_MOD_I64:  expected_type = type_unsigned_int64;     break;
+
+			default:
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			break;
+
+		case 'A':
+		case 'a':
+		case 'E':
+		case 'e':
+		case 'F':
+		case 'f':
+		case 'G':
+		case 'g':
+			switch (fmt_mod) {
+			case FMT_MOD_l:    expected_type = type_double;      break;
+			case FMT_MOD_NONE: expected_type = type_float;       break;
+			case FMT_MOD_L:    expected_type = type_long_double; break;
+
+			default:
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			break;
+
+		case 'C':
+			if (fmt_mod != FMT_MOD_NONE) {
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			expected_type = type_wchar_t;
+			goto check_c_width;
+
+		case 'c': {
+			switch (fmt_mod) {
+			case FMT_MOD_NONE: expected_type = type_char;    break;
+			case FMT_MOD_l:    expected_type = type_wchar_t; break;
+			case FMT_MOD_w:    expected_type = type_wchar_t; break;
+
+			default:
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+
+check_c_width:
+			if (width == 0)
+				width = 1;
+			if (!suppress_assignment && arg != NULL) {
+				type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
+				if (is_type_array(type)       &&
+				    type->array.size_constant &&
+				    width > type->array.size) {
+					warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters at format %u", type, width, num_fmt);
+				}
+			}
+			break;
+		}
+
+		case 'S':
+			if (fmt_mod != FMT_MOD_NONE) {
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			expected_type = type_wchar_t;
+			break;
+
+		case 's':
+		case '[': {
+			switch (fmt_mod) {
+				case FMT_MOD_NONE: expected_type = type_char;    break;
+				case FMT_MOD_l:    expected_type = type_wchar_t; break;
+				case FMT_MOD_w:    expected_type = type_wchar_t; break;
+
+				default:
+					warn_invalid_length_modifier(pos, fmt_mod, fmt);
 					goto next_arg;
+			}
+
+			if (!suppress_assignment &&
+			    width != 0           &&
+			    arg   != NULL) {
+				type_t *const type = skip_typeref(revert_automatic_type_conversion(arg->expression));
+				if (is_type_array(type)       &&
+				    type->array.size_constant &&
+				    width >= type->array.size) {
+					warningf(WARN_FORMAT, pos, "target buffer '%T' is too small for %u characters and \\0 at format %u", type, width, num_fmt);
 				}
 			}
-			if (is_type_valid(arg_type)) {
-				warningf(pos,
-					"argument type '%T' does not match conversion specifier '%%%s%c'",
-					arg_type, get_length_modifier_name(fmt_mod), (char)fmt);
+			break;
+		}
+
+		case 'p':
+			if (fmt_mod != FMT_MOD_NONE) {
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			expected_type = type_void;
+			break;
+
+		case 'n': {
+			if (suppress_assignment) {
+				warningf(WARN_FORMAT, pos, "conversion '%n' cannot be suppressed with '*' at format %u", num_fmt);
+			}
+
+			switch (fmt_mod) {
+			case FMT_MOD_NONE: expected_type = type_int;         break;
+			case FMT_MOD_hh:   expected_type = type_signed_char; break;
+			case FMT_MOD_h:    expected_type = type_short;       break;
+			case FMT_MOD_l:    expected_type = type_long;        break;
+			case FMT_MOD_ll:   expected_type = type_long_long;   break;
+			case FMT_MOD_j:    expected_type = type_intmax_t;    break;
+			case FMT_MOD_z:    expected_type = type_ssize_t;     break;
+			case FMT_MOD_t:    expected_type = type_ptrdiff_t;   break;
+
+			default:
+				warn_invalid_length_modifier(pos, fmt_mod, fmt);
+				goto next_arg;
+			}
+			break;
+		}
+
+		default:
+			warningf(WARN_FORMAT, pos, "encountered unknown conversion specifier '%%%c' at format %u", fmt, num_fmt);
+			if (suppress_assignment)
+				continue;
+			if (arg == NULL)
+				goto too_few_args;
+			goto next_arg;
+		}
+
+		if (suppress_assignment)
+			continue;
+
+		if (arg == NULL) {
+too_few_args:
+			warningf(WARN_FORMAT, pos, "too few arguments for format string");
+			return;
+		}
+
+		{ /* create a scope here to prevent warning about the jump to next_arg */
+			type_t *const arg_type           = arg->expression->base.type;
+			type_t *const arg_skip           = skip_typeref(arg_type);
+			type_t *const expected_type_skip = skip_typeref(expected_type);
+
+			if (! is_type_pointer(arg_skip))
+				goto error_arg_type;
+			type_t *const ptr_skip = skip_typeref(arg_skip->pointer.points_to);
+
+			if (fmt == 'p') {
+				/* allow any pointer type for %p, not just void */
+				if (is_type_pointer(ptr_skip))
+					goto next_arg;
+			}
+
+			/* do NOT allow const or restrict, all other should be ok */
+			if (ptr_skip->base.qualifiers & (TYPE_QUALIFIER_CONST | TYPE_QUALIFIER_VOLATILE))
+				goto error_arg_type;
+			type_t *const unqual_ptr = get_unqualified_type(ptr_skip);
+			if (unqual_ptr == expected_type_skip) {
+				goto next_arg;
+			} else if (expected_type_skip == type_char) {
+				/* char matches with unsigned char AND signed char */
+				if (unqual_ptr == type_signed_char || unqual_ptr == type_unsigned_char)
+					goto next_arg;
+			}
+error_arg_type:
+			if (is_type_valid(arg_skip)) {
+				source_position_t const *const apos = &arg->expression->base.source_position;
+				char              const *const mod  = get_length_modifier_name(fmt_mod);
+				warningf(WARN_FORMAT, apos, "conversion '%%%s%c' at position %u specifies type '%T*' but the argument has type '%T'", mod, (char)fmt, num_fmt, expected_type, arg_type);
 			}
 		}
 next_arg:
 		arg = arg->next;
 	}
-	if (!atend(&vchar)) {
-		warningf(pos, "format string contains NUL");
+	assert(fmt == '\0');
+	if (c+1 < string + size) {
+		warningf(WARN_FORMAT, pos, "format string contains '\\0'");
 	}
 	if (arg != NULL) {
-		warningf(pos, "too many arguments for format string");
+		unsigned num_args = num_fmt;
+		while (arg != NULL) {
+			++num_args;
+			arg = arg->next;
+		}
+		warningf(WARN_FORMAT, pos, "%u argument%s but only %u format specifier%s", num_args, num_args != 1 ? "s" : "", num_fmt, num_fmt != 1 ? "s" : "");
 	}
 }
 
-static const struct {
-	const char    *name;
-	format_kind_t  fmt_kind;
-	unsigned       fmt_idx;
-	unsigned       arg_idx;
-} builtin_table[] = {
+static const format_spec_t builtin_table[] = {
 	{ "printf",        FORMAT_PRINTF,   0, 1 },
 	{ "wprintf",       FORMAT_PRINTF,   0, 1 },
 	{ "sprintf",       FORMAT_PRINTF,   1, 2 },
@@ -613,33 +985,37 @@ static const struct {
 
 void check_format(const call_expression_t *const call)
 {
-	if (!warning.format)
+	if (!is_warn_on(WARN_FORMAT))
 		return;
 
 	const expression_t *const func_expr = call->function;
 	if (func_expr->kind != EXPR_REFERENCE)
 		return;
 
-	const declaration_t   *const decl = func_expr->reference.declaration;
-	const call_argument_t *      arg  = call->arguments;
-
-	if(false) {
-		/* the declaration has a GNU format attribute, check it */
-	} else {
-		/*
-		 * For some functions we always check the format, even if it was not specified.
-		 * This allows to check format even in MS mode or without header included.
-		 */
-		const char            *const name = decl->symbol->string;
-		for(size_t i = 0; i < sizeof(builtin_table) / sizeof(builtin_table[0]); ++i) {
-			if(strcmp(name, builtin_table[i].name) == 0) {
-				if(builtin_table[i].fmt_kind == FORMAT_PRINTF) {
-					check_format_arguments(arg,
-					                       builtin_table[i].fmt_idx,
-					                       builtin_table[i].arg_idx);
-				}
+	const entity_t        *const entity = func_expr->reference.entity;
+	const call_argument_t *      arg    = call->arguments;
+
+	/*
+	 * For some functions we always check the format, even if it was not
+	 * specified. This allows to check format even in MS mode or without
+	 * header included.
+	 */
+	const char *const name = entity->base.symbol->string;
+	for (size_t i = 0; i < lengthof(builtin_table); ++i) {
+		if (streq(name, builtin_table[i].name)) {
+			switch (builtin_table[i].fmt_kind) {
+			case FORMAT_PRINTF:
+				check_printf_format(arg, &builtin_table[i]);
+				break;
+			case FORMAT_SCANF:
+				check_scanf_format(arg, &builtin_table[i]);
+				break;
+			case FORMAT_STRFTIME:
+			case FORMAT_STRFMON:
+				/* TODO: implement other cases */
 				break;
 			}
+			break;
 		}
 	}
 }