blob: 8e90b7cfc43c54d5d29208dd6519bab46aae093e [file] [log] [blame]
/* Copyright (C) 2016-2021 Free Software Foundation, Inc.
Contributed by Martin Sebor <msebor@redhat.com>.
This file is part of GCC.
GCC is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation; either version 3, or (at your option) any later
version.
GCC is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3. If not see
<http://www.gnu.org/licenses/>. */
/* This file implements the printf-return-value pass. The pass does
two things: 1) it analyzes calls to formatted output functions like
sprintf looking for possible buffer overflows and calls to bounded
functions like snprintf for early truncation (and under the control
of the -Wformat-length option issues warnings), and 2) under the
control of the -fprintf-return-value option it folds the return
value of safe calls into constants, making it possible to eliminate
code that depends on the value of those constants.
For all functions (bounded or not) the pass uses the size of the
destination object. That means that it will diagnose calls to
snprintf not on the basis of the size specified by the function's
second argument but rather on the basis of the size the first
argument points to (if possible). For bound-checking built-ins
like __builtin___snprintf_chk the pass uses the size typically
determined by __builtin_object_size and passed to the built-in
by the Glibc inline wrapper.
The pass handles all forms standard sprintf format directives,
including character, integer, floating point, pointer, and strings,
with the standard C flags, widths, and precisions. For integers
and strings it computes the length of output itself. For floating
point it uses MPFR to format known constants with up and down
rounding and uses the resulting range of output lengths. For
strings it uses the length of string literals and the sizes of
character arrays that a character pointer may point to as a bound
on the longest string. */
#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "backend.h"
#include "tree.h"
#include "gimple.h"
#include "tree-pass.h"
#include "ssa.h"
#include "gimple-fold.h"
#include "gimple-pretty-print.h"
#include "diagnostic-core.h"
#include "fold-const.h"
#include "gimple-iterator.h"
#include "tree-ssa.h"
#include "tree-object-size.h"
#include "tree-cfg.h"
#include "tree-ssa-propagate.h"
#include "calls.h"
#include "cfgloop.h"
#include "tree-scalar-evolution.h"
#include "tree-ssa-loop.h"
#include "intl.h"
#include "langhooks.h"
#include "attribs.h"
#include "builtins.h"
#include "pointer-query.h"
#include "stor-layout.h"
#include "realmpfr.h"
#include "target.h"
#include "cpplib.h"
#include "input.h"
#include "toplev.h"
#include "substring-locations.h"
#include "diagnostic.h"
#include "domwalk.h"
#include "alloc-pool.h"
#include "vr-values.h"
#include "tree-ssa-strlen.h"
#include "tree-dfa.h"
/* The likely worst case value of MB_LEN_MAX for the target, large enough
for UTF-8. Ideally, this would be obtained by a target hook if it were
to be used for optimization but it's good enough as is for warnings. */
#define target_mb_len_max() 6
/* The maximum number of bytes a single non-string directive can result
in. This is the result of printf("%.*Lf", INT_MAX, -LDBL_MAX) for
LDBL_MAX_10_EXP of 4932. */
#define IEEE_MAX_10_EXP 4932
#define target_dir_max() (target_int_max () + IEEE_MAX_10_EXP + 2)
namespace {
/* Set to the warning level for the current function which is equal
either to warn_format_trunc for bounded functions or to
warn_format_overflow otherwise. */
static int warn_level;
/* The minimum, maximum, likely, and unlikely maximum number of bytes
of output either a formatting function or an individual directive
can result in. */
struct result_range
{
/* The absolute minimum number of bytes. The result of a successful
conversion is guaranteed to be no less than this. (An erroneous
conversion can be indicated by MIN > HOST_WIDE_INT_MAX.) */
unsigned HOST_WIDE_INT min;
/* The likely maximum result that is used in diagnostics. In most
cases MAX is the same as the worst case UNLIKELY result. */
unsigned HOST_WIDE_INT max;
/* The likely result used to trigger diagnostics. For conversions
that result in a range of bytes [MIN, MAX], LIKELY is somewhere
in that range. */
unsigned HOST_WIDE_INT likely;
/* In rare cases (e.g., for multibyte characters) UNLIKELY gives
the worst cases maximum result of a directive. In most cases
UNLIKELY == MAX. UNLIKELY is used to control the return value
optimization but not in diagnostics. */
unsigned HOST_WIDE_INT unlikely;
};
/* Return the value of INT_MIN for the target. */
static inline HOST_WIDE_INT
target_int_min ()
{
return tree_to_shwi (TYPE_MIN_VALUE (integer_type_node));
}
/* Return the value of INT_MAX for the target. */
static inline unsigned HOST_WIDE_INT
target_int_max ()
{
return tree_to_uhwi (TYPE_MAX_VALUE (integer_type_node));
}
/* Return the value of SIZE_MAX for the target. */
static inline unsigned HOST_WIDE_INT
target_size_max ()
{
return tree_to_uhwi (TYPE_MAX_VALUE (size_type_node));
}
/* A straightforward mapping from the execution character set to the host
character set indexed by execution character. */
static char target_to_host_charmap[256];
/* Initialize a mapping from the execution character set to the host
character set. */
static bool
init_target_to_host_charmap ()
{
/* If the percent sign is non-zero the mapping has already been
initialized. */
if (target_to_host_charmap['%'])
return true;
/* Initialize the target_percent character (done elsewhere). */
if (!init_target_chars ())
return false;
/* The subset of the source character set used by printf conversion
specifications (strictly speaking, not all letters are used but
they are included here for the sake of simplicity). The dollar
sign must be included even though it's not in the basic source
character set. */
const char srcset[] = " 0123456789!\"#%&'()*+,-./:;<=>?[\\]^_{|}~$"
"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";
/* Set the mapping for all characters to some ordinary value (i,e.,
not none used in printf conversion specifications) and overwrite
those that are used by conversion specifications with their
corresponding values. */
memset (target_to_host_charmap + 1, '?', sizeof target_to_host_charmap - 1);
/* Are the two sets of characters the same? */
bool all_same_p = true;
for (const char *pc = srcset; *pc; ++pc)
{
/* Slice off the high end bits in case target characters are
signed. All values are expected to be non-nul, otherwise
there's a problem. */
if (unsigned char tc = lang_hooks.to_target_charset (*pc))
{
target_to_host_charmap[tc] = *pc;
if (tc != *pc)
all_same_p = false;
}
else
return false;
}
/* Set the first element to a non-zero value if the mapping
is 1-to-1, otherwise leave it clear (NUL is assumed to be
the same in both character sets). */
target_to_host_charmap[0] = all_same_p;
return true;
}
/* Return the host source character corresponding to the character
CH in the execution character set if one exists, or some innocuous
(non-special, non-nul) source character otherwise. */
static inline unsigned char
target_to_host (unsigned char ch)
{
return target_to_host_charmap[ch];
}
/* Convert an initial substring of the string TARGSTR consisting of
characters in the execution character set into a string in the
source character set on the host and store up to HOSTSZ characters
in the buffer pointed to by HOSTR. Return HOSTR. */
static const char*
target_to_host (char *hostr, size_t hostsz, const char *targstr)
{
/* Make sure the buffer is reasonably big. */
gcc_assert (hostsz > 4);
/* The interesting subset of source and execution characters are
the same so no conversion is necessary. However, truncate
overlong strings just like the translated strings are. */
if (target_to_host_charmap['\0'] == 1)
{
size_t len = strlen (targstr);
if (len >= hostsz)
{
memcpy (hostr, targstr, hostsz - 4);
strcpy (hostr + hostsz - 4, "...");
}
else
memcpy (hostr, targstr, len + 1);
return hostr;
}
/* Convert the initial substring of TARGSTR to the corresponding
characters in the host set, appending "..." if TARGSTR is too
long to fit. Using the static buffer assumes the function is
not called in between sequence points (which it isn't). */
for (char *ph = hostr; ; ++targstr)
{
*ph++ = target_to_host (*targstr);
if (!*targstr)
break;
if (size_t (ph - hostr) == hostsz)
{
strcpy (ph - 4, "...");
break;
}
}
return hostr;
}
/* Convert the sequence of decimal digits in the execution character
starting at *PS to a HOST_WIDE_INT, analogously to strtol. Return
the result and set *PS to one past the last converted character.
On range error set ERANGE to the digit that caused it. */
static inline HOST_WIDE_INT
target_strtowi (const char **ps, const char **erange)
{
unsigned HOST_WIDE_INT val = 0;
for ( ; ; ++*ps)
{
unsigned char c = target_to_host (**ps);
if (ISDIGIT (c))
{
c -= '0';
/* Check for overflow. */
if (val > ((unsigned HOST_WIDE_INT) HOST_WIDE_INT_MAX - c) / 10LU)
{
val = HOST_WIDE_INT_MAX;
*erange = *ps;
/* Skip the remaining digits. */
do
c = target_to_host (*++*ps);
while (ISDIGIT (c));
break;
}
else
val = val * 10 + c;
}
else
break;
}
return val;
}
/* Given FORMAT, set *PLOC to the source location of the format string
and return the format string if it is known or null otherwise. */
static const char*
get_format_string (tree format, location_t *ploc)
{
*ploc = EXPR_LOC_OR_LOC (format, input_location);
return c_getstr (format);
}
/* For convenience and brevity, shorter named entrypoints of
format_string_diagnostic_t::emit_warning_va and
format_string_diagnostic_t::emit_warning_n_va.
These have to be functions with the attribute so that exgettext
works properly. */
static bool
ATTRIBUTE_GCC_DIAG (5, 6)
fmtwarn (const substring_loc &fmt_loc, location_t param_loc,
const char *corrected_substring, opt_code opt,
const char *gmsgid, ...)
{
format_string_diagnostic_t diag (fmt_loc, NULL, param_loc, NULL,
corrected_substring);
va_list ap;
va_start (ap, gmsgid);
bool warned = diag.emit_warning_va (opt, gmsgid, &ap);
va_end (ap);
return warned;
}
static bool
ATTRIBUTE_GCC_DIAG (6, 8) ATTRIBUTE_GCC_DIAG (7, 8)
fmtwarn_n (const substring_loc &fmt_loc, location_t param_loc,
const char *corrected_substring, opt_code opt,
unsigned HOST_WIDE_INT n,
const char *singular_gmsgid, const char *plural_gmsgid, ...)
{
format_string_diagnostic_t diag (fmt_loc, NULL, param_loc, NULL,
corrected_substring);
va_list ap;
va_start (ap, plural_gmsgid);
bool warned = diag.emit_warning_n_va (opt, n, singular_gmsgid, plural_gmsgid,
&ap);
va_end (ap);
return warned;
}
/* Format length modifiers. */
enum format_lengths
{
FMT_LEN_none,
FMT_LEN_hh, // char argument
FMT_LEN_h, // short
FMT_LEN_l, // long
FMT_LEN_ll, // long long
FMT_LEN_L, // long double (and GNU long long)
FMT_LEN_z, // size_t
FMT_LEN_t, // ptrdiff_t
FMT_LEN_j // intmax_t
};
/* Description of the result of conversion either of a single directive
or the whole format string. */
class fmtresult
{
public:
/* Construct a FMTRESULT object with all counters initialized
to MIN. KNOWNRANGE is set when MIN is valid. */
fmtresult (unsigned HOST_WIDE_INT min = HOST_WIDE_INT_MAX)
: argmin (), argmax (), dst_offset (HOST_WIDE_INT_MIN), nonstr (),
knownrange (min < HOST_WIDE_INT_MAX),
mayfail (), nullp ()
{
range.min = min;
range.max = min;
range.likely = min;
range.unlikely = min;
}
/* Construct a FMTRESULT object with MIN, MAX, and LIKELY counters.
KNOWNRANGE is set when both MIN and MAX are valid. */
fmtresult (unsigned HOST_WIDE_INT min, unsigned HOST_WIDE_INT max,
unsigned HOST_WIDE_INT likely = HOST_WIDE_INT_MAX)
: argmin (), argmax (), dst_offset (HOST_WIDE_INT_MIN), nonstr (),
knownrange (min < HOST_WIDE_INT_MAX && max < HOST_WIDE_INT_MAX),
mayfail (), nullp ()
{
range.min = min;
range.max = max;
range.likely = max < likely ? min : likely;
range.unlikely = max;
}
/* Adjust result upward to reflect the RANGE of values the specified
width or precision is known to be in. */
fmtresult& adjust_for_width_or_precision (const HOST_WIDE_INT[2],
tree = NULL_TREE,
unsigned = 0, unsigned = 0);
/* Return the maximum number of decimal digits a value of TYPE
formats as on output. */
static unsigned type_max_digits (tree, int);
/* The range a directive's argument is in. */
tree argmin, argmax;
/* The starting offset into the destination of the formatted function
call of the %s argument that points into (aliases with) the same
destination array. */
HOST_WIDE_INT dst_offset;
/* The minimum and maximum number of bytes that a directive
results in on output for an argument in the range above. */
result_range range;
/* Non-nul when the argument of a string directive is not a nul
terminated string. */
tree nonstr;
/* True when the range above is obtained from a known value of
a directive's argument or its bounds and not the result of
heuristics that depend on warning levels. */
bool knownrange;
/* True for a directive that may fail (such as wide character
directives). */
bool mayfail;
/* True when the argument is a null pointer. */
bool nullp;
};
/* Adjust result upward to reflect the range ADJUST of values the
specified width or precision is known to be in. When non-null,
TYPE denotes the type of the directive whose result is being
adjusted, BASE gives the base of the directive (octal, decimal,
or hex), and ADJ denotes the additional adjustment to the LIKELY
counter that may need to be added when ADJUST is a range. */
fmtresult&
fmtresult::adjust_for_width_or_precision (const HOST_WIDE_INT adjust[2],
tree type /* = NULL_TREE */,
unsigned base /* = 0 */,
unsigned adj /* = 0 */)
{
bool minadjusted = false;
/* Adjust the minimum and likely counters. */
if (adjust[0] >= 0)
{
if (range.min < (unsigned HOST_WIDE_INT)adjust[0])
{
range.min = adjust[0];
minadjusted = true;
}
/* Adjust the likely counter. */
if (range.likely < range.min)
range.likely = range.min;
}
else if (adjust[0] == target_int_min ()
&& (unsigned HOST_WIDE_INT)adjust[1] == target_int_max ())
knownrange = false;
/* Adjust the maximum counter. */
if (adjust[1] > 0)
{
if (range.max < (unsigned HOST_WIDE_INT)adjust[1])
{
range.max = adjust[1];
/* Set KNOWNRANGE if both the minimum and maximum have been
adjusted. Otherwise leave it at what it was before. */
knownrange = minadjusted;
}
}
if (warn_level > 1 && type)
{
/* For large non-constant width or precision whose range spans
the maximum number of digits produced by the directive for
any argument, set the likely number of bytes to be at most
the number digits plus other adjustment determined by the
caller (one for sign or two for the hexadecimal "0x"
prefix). */
unsigned dirdigs = type_max_digits (type, base);
if (adjust[0] < dirdigs && dirdigs < adjust[1]
&& range.likely < dirdigs)
range.likely = dirdigs + adj;
}
else if (range.likely < (range.min ? range.min : 1))
{
/* Conservatively, set LIKELY to at least MIN but no less than
1 unless MAX is zero. */
range.likely = (range.min
? range.min
: range.max && (range.max < HOST_WIDE_INT_MAX
|| warn_level > 1) ? 1 : 0);
}
/* Finally adjust the unlikely counter to be at least as large as
the maximum. */
if (range.unlikely < range.max)
range.unlikely = range.max;
return *this;
}
/* Return the maximum number of digits a value of TYPE formats in
BASE on output, not counting base prefix . */
unsigned
fmtresult::type_max_digits (tree type, int base)
{
unsigned prec = TYPE_PRECISION (type);
switch (base)
{
case 8:
return (prec + 2) / 3;
case 10:
/* Decimal approximation: yields 3, 5, 10, and 20 for precision
of 8, 16, 32, and 64 bits. */
return prec * 301 / 1000 + 1;
case 16:
return prec / 4;
}
gcc_unreachable ();
}
static bool
get_int_range (tree, gimple *, HOST_WIDE_INT *, HOST_WIDE_INT *,
bool, HOST_WIDE_INT, range_query *);
struct call_info;
/* Description of a format directive. A directive is either a plain
string or a conversion specification that starts with '%'. */
struct directive
{
directive (const call_info *inf, unsigned dno)
: info (inf), dirno (dno), argno (), beg (), len (), flags (),
width (), prec (), modifier (), specifier (), arg (), fmtfunc ()
{ }
/* Reference to the info structure describing the call that this
directive is a part of. */
const call_info *info;
/* The 1-based directive number (for debugging). */
unsigned dirno;
/* The zero-based argument number of the directive's argument ARG in
the function's argument list. */
unsigned argno;
/* The first character of the directive and its length. */
const char *beg;
size_t len;
/* A bitmap of flags, one for each character. */
unsigned flags[256 / sizeof (int)];
/* The range of values of the specified width, or -1 if not specified. */
HOST_WIDE_INT width[2];
/* The range of values of the specified precision, or -1 if not
specified. */
HOST_WIDE_INT prec[2];
/* Length modifier. */
format_lengths modifier;
/* Format specifier character. */
char specifier;
/* The argument of the directive or null when the directive doesn't
take one or when none is available (such as for vararg functions). */
tree arg;
/* Format conversion function that given a directive and an argument
returns the formatting result. */
fmtresult (*fmtfunc) (const directive &, tree, range_query *);
/* Return True when the format flag CHR has been used. */
bool get_flag (char chr) const
{
unsigned char c = chr & 0xff;
return (flags[c / (CHAR_BIT * sizeof *flags)]
& (1U << (c % (CHAR_BIT * sizeof *flags))));
}
/* Make a record of the format flag CHR having been used. */
void set_flag (char chr)
{
unsigned char c = chr & 0xff;
flags[c / (CHAR_BIT * sizeof *flags)]
|= (1U << (c % (CHAR_BIT * sizeof *flags)));
}
/* Reset the format flag CHR. */
void clear_flag (char chr)
{
unsigned char c = chr & 0xff;
flags[c / (CHAR_BIT * sizeof *flags)]
&= ~(1U << (c % (CHAR_BIT * sizeof *flags)));
}
/* Set both bounds of the width range to VAL. */
void set_width (HOST_WIDE_INT val)
{
width[0] = width[1] = val;
}
/* Set the width range according to ARG, with both bounds being
no less than 0. For a constant ARG set both bounds to its value
or 0, whichever is greater. For a non-constant ARG in some range
set width to its range adjusting each bound to -1 if it's less.
For an indeterminate ARG set width to [0, INT_MAX]. */
void set_width (tree arg, range_query *);
/* Set both bounds of the precision range to VAL. */
void set_precision (HOST_WIDE_INT val)
{
prec[0] = prec[1] = val;
}
/* Set the precision range according to ARG, with both bounds being
no less than -1. For a constant ARG set both bounds to its value
or -1 whichever is greater. For a non-constant ARG in some range
set precision to its range adjusting each bound to -1 if it's less.
For an indeterminate ARG set precision to [-1, INT_MAX]. */
void set_precision (tree arg, range_query *query);
/* Return true if both width and precision are known to be
either constant or in some range, false otherwise. */
bool known_width_and_precision () const
{
return ((width[1] < 0
|| (unsigned HOST_WIDE_INT)width[1] <= target_int_max ())
&& (prec[1] < 0
|| (unsigned HOST_WIDE_INT)prec[1] < target_int_max ()));
}
};
/* The result of a call to a formatted function. */
struct format_result
{
format_result ()
: range (), aliases (), alias_count (), knownrange (), posunder4k (),
floating (), warned () { /* No-op. */ }
~format_result ()
{
XDELETEVEC (aliases);
}
/* Range of characters written by the formatted function.
Setting the minimum to HOST_WIDE_INT_MAX disables all
length tracking for the remainder of the format string. */
result_range range;
struct alias_info
{
directive dir; /* The directive that aliases the destination. */
HOST_WIDE_INT offset; /* The offset at which it aliases it. */
result_range range; /* The raw result of the directive. */
};
/* An array of directives whose pointer argument aliases a part
of the destination object of the formatted function. */
alias_info *aliases;
unsigned alias_count;
/* True when the range above is obtained from known values of
directive arguments, or bounds on the amount of output such
as width and precision, and not the result of heuristics that
depend on warning levels. It's used to issue stricter diagnostics
in cases where strings of unknown lengths are bounded by the arrays
they are determined to refer to. KNOWNRANGE must not be used for
the return value optimization. */
bool knownrange;
/* True if no individual directive could fail or result in more than
4095 bytes of output (the total NUMBER_CHARS_{MIN,MAX} might be
greater). Implementations are not required to handle directives
that produce more than 4K bytes (leading to undefined behavior)
and so when one is found it disables the return value optimization.
Similarly, directives that can fail (such as wide character
directives) disable the optimization. */
bool posunder4k;
/* True when a floating point directive has been seen in the format
string. */
bool floating;
/* True when an intermediate result has caused a warning. Used to
avoid issuing duplicate warnings while finishing the processing
of a call. WARNED also disables the return value optimization. */
bool warned;
/* Preincrement the number of output characters by 1. */
format_result& operator++ ()
{
return *this += 1;
}
/* Postincrement the number of output characters by 1. */
format_result operator++ (int)
{
format_result prev (*this);
*this += 1;
return prev;
}
/* Increment the number of output characters by N. */
format_result& operator+= (unsigned HOST_WIDE_INT);
/* Add a directive to the sequence of those with potentially aliasing
arguments. */
void append_alias (const directive &, HOST_WIDE_INT, const result_range &);
private:
/* Not copyable or assignable. */
format_result (format_result&);
void operator= (format_result&);
};
format_result&
format_result::operator+= (unsigned HOST_WIDE_INT n)
{
gcc_assert (n < HOST_WIDE_INT_MAX);
if (range.min < HOST_WIDE_INT_MAX)
range.min += n;
if (range.max < HOST_WIDE_INT_MAX)
range.max += n;
if (range.likely < HOST_WIDE_INT_MAX)
range.likely += n;
if (range.unlikely < HOST_WIDE_INT_MAX)
range.unlikely += n;
return *this;
}
void
format_result::append_alias (const directive &d, HOST_WIDE_INT off,
const result_range &resrng)
{
unsigned cnt = alias_count + 1;
alias_info *ar = XNEWVEC (alias_info, cnt);
for (unsigned i = 0; i != alias_count; ++i)
ar[i] = aliases[i];
ar[alias_count].dir = d;
ar[alias_count].offset = off;
ar[alias_count].range = resrng;
XDELETEVEC (aliases);
alias_count = cnt;
aliases = ar;
}
/* Return the logarithm of X in BASE. */
static int
ilog (unsigned HOST_WIDE_INT x, int base)
{
int res = 0;
do
{
++res;
x /= base;
} while (x);
return res;
}
/* Return the number of bytes resulting from converting into a string
the INTEGER_CST tree node X in BASE with a minimum of PREC digits.
PLUS indicates whether 1 for a plus sign should be added for positive
numbers, and PREFIX whether the length of an octal ('O') or hexadecimal
('0x') prefix should be added for nonzero numbers. Return -1 if X cannot
be represented. */
static HOST_WIDE_INT
tree_digits (tree x, int base, HOST_WIDE_INT prec, bool plus, bool prefix)
{
unsigned HOST_WIDE_INT absval;
HOST_WIDE_INT res;
if (TYPE_UNSIGNED (TREE_TYPE (x)))
{
if (tree_fits_uhwi_p (x))
{
absval = tree_to_uhwi (x);
res = plus;
}
else
return -1;
}
else
{
if (tree_fits_shwi_p (x))
{
HOST_WIDE_INT i = tree_to_shwi (x);
if (HOST_WIDE_INT_MIN == i)
{
/* Avoid undefined behavior due to negating a minimum. */
absval = HOST_WIDE_INT_MAX;
res = 1;
}
else if (i < 0)
{
absval = -i;
res = 1;
}
else
{
absval = i;
res = plus;
}
}
else
return -1;
}
int ndigs = ilog (absval, base);
res += prec < ndigs ? ndigs : prec;
/* Adjust a non-zero value for the base prefix, either hexadecimal,
or, unless precision has resulted in a leading zero, also octal. */
if (prefix && absval && (base == 16 || prec <= ndigs))
{
if (base == 8)
res += 1;
else if (base == 16)
res += 2;
}
return res;
}
/* Description of a call to a formatted function. */
struct call_info
{
/* Function call statement. */
gimple *callstmt;
/* Function called. */
tree func;
/* Called built-in function code. */
built_in_function fncode;
/* The "origin" of the destination pointer argument, which is either
the DECL of the destination buffer being written into or a pointer
that points to it, plus some offset. */
tree dst_origin;
/* For a destination pointing to a struct array member, the offset of
the member. */
HOST_WIDE_INT dst_field;
/* The offset into the destination buffer. */
HOST_WIDE_INT dst_offset;
/* Format argument and format string extracted from it. */
tree format;
const char *fmtstr;
/* The location of the format argument. */
location_t fmtloc;
/* The destination object size for __builtin___xxx_chk functions
typically determined by __builtin_object_size, or -1 if unknown. */
unsigned HOST_WIDE_INT objsize;
/* Number of the first variable argument. */
unsigned HOST_WIDE_INT argidx;
/* True for functions like snprintf that specify the size of
the destination, false for others like sprintf that don't. */
bool bounded;
/* True for bounded functions like snprintf that specify a zero-size
buffer as a request to compute the size of output without actually
writing any. NOWRITE is cleared in response to the %n directive
which has side-effects similar to writing output. */
bool nowrite;
/* Return true if the called function's return value is used. */
bool retval_used () const
{
return gimple_get_lhs (callstmt);
}
/* Return the warning option corresponding to the called function. */
opt_code warnopt () const
{
return bounded ? OPT_Wformat_truncation_ : OPT_Wformat_overflow_;
}
/* Return true for calls to file formatted functions. */
bool is_file_func () const
{
return (fncode == BUILT_IN_FPRINTF
|| fncode == BUILT_IN_FPRINTF_CHK
|| fncode == BUILT_IN_FPRINTF_UNLOCKED
|| fncode == BUILT_IN_VFPRINTF
|| fncode == BUILT_IN_VFPRINTF_CHK);
}
/* Return true for calls to string formatted functions. */
bool is_string_func () const
{
return (fncode == BUILT_IN_SPRINTF
|| fncode == BUILT_IN_SPRINTF_CHK
|| fncode == BUILT_IN_SNPRINTF
|| fncode == BUILT_IN_SNPRINTF_CHK
|| fncode == BUILT_IN_VSPRINTF
|| fncode == BUILT_IN_VSPRINTF_CHK
|| fncode == BUILT_IN_VSNPRINTF
|| fncode == BUILT_IN_VSNPRINTF_CHK);
}
};
void
directive::set_width (tree arg, range_query *query)
{
get_int_range (arg, info->callstmt, width, width + 1, true, 0, query);
}
void
directive::set_precision (tree arg, range_query *query)
{
get_int_range (arg, info->callstmt, prec, prec + 1, false, -1, query);
}
/* Return the result of formatting a no-op directive (such as '%n'). */
static fmtresult
format_none (const directive &, tree, range_query *)
{
fmtresult res (0);
return res;
}
/* Return the result of formatting the '%%' directive. */
static fmtresult
format_percent (const directive &, tree, range_query *)
{
fmtresult res (1);
return res;
}
/* Compute intmax_type_node and uintmax_type_node similarly to how
tree.c builds size_type_node. */
static void
build_intmax_type_nodes (tree *pintmax, tree *puintmax)
{
if (strcmp (UINTMAX_TYPE, "unsigned int") == 0)
{
*pintmax = integer_type_node;
*puintmax = unsigned_type_node;
}
else if (strcmp (UINTMAX_TYPE, "long unsigned int") == 0)
{
*pintmax = long_integer_type_node;
*puintmax = long_unsigned_type_node;
}
else if (strcmp (UINTMAX_TYPE, "long long unsigned int") == 0)
{
*pintmax = long_long_integer_type_node;
*puintmax = long_long_unsigned_type_node;
}
else
{
for (int i = 0; i < NUM_INT_N_ENTS; i++)
if (int_n_enabled_p[i])
{
char name[50], altname[50];
sprintf (name, "__int%d unsigned", int_n_data[i].bitsize);
sprintf (altname, "__int%d__ unsigned", int_n_data[i].bitsize);
if (strcmp (name, UINTMAX_TYPE) == 0
|| strcmp (altname, UINTMAX_TYPE) == 0)
{
*pintmax = int_n_trees[i].signed_type;
*puintmax = int_n_trees[i].unsigned_type;
return;
}
}
gcc_unreachable ();
}
}
/* Determine the range [*PMIN, *PMAX] that the expression ARG is
in and that is representable in type int.
Return true when the range is a subrange of that of int.
When ARG is null it is as if it had the full range of int.
When ABSOLUTE is true the range reflects the absolute value of
the argument. When ABSOLUTE is false, negative bounds of
the determined range are replaced with NEGBOUND. */
static bool
get_int_range (tree arg, gimple *stmt,
HOST_WIDE_INT *pmin, HOST_WIDE_INT *pmax,
bool absolute, HOST_WIDE_INT negbound,
range_query *query)
{
/* The type of the result. */
const_tree type = integer_type_node;
bool knownrange = false;
if (!arg)
{
*pmin = tree_to_shwi (TYPE_MIN_VALUE (type));
*pmax = tree_to_shwi (TYPE_MAX_VALUE (type));
}
else if (TREE_CODE (arg) == INTEGER_CST
&& TYPE_PRECISION (TREE_TYPE (arg)) <= TYPE_PRECISION (type))
{
/* For a constant argument return its value adjusted as specified
by NEGATIVE and NEGBOUND and return true to indicate that the
result is known. */
*pmin = tree_fits_shwi_p (arg) ? tree_to_shwi (arg) : tree_to_uhwi (arg);
*pmax = *pmin;
knownrange = true;
}
else
{
/* True if the argument's range cannot be determined. */
bool unknown = true;
tree argtype = TREE_TYPE (arg);
/* Ignore invalid arguments with greater precision that that
of the expected type (e.g., in sprintf("%*i", 12LL, i)).
They will have been detected and diagnosed by -Wformat and
so it's not important to complicate this code to try to deal
with them again. */
if (TREE_CODE (arg) == SSA_NAME
&& INTEGRAL_TYPE_P (argtype)
&& TYPE_PRECISION (argtype) <= TYPE_PRECISION (type))
{
/* Try to determine the range of values of the integer argument. */
value_range vr;
query->range_of_expr (vr, arg, stmt);
if (!vr.undefined_p () && !vr.varying_p ())
{
HOST_WIDE_INT type_min
= (TYPE_UNSIGNED (argtype)
? tree_to_uhwi (TYPE_MIN_VALUE (argtype))
: tree_to_shwi (TYPE_MIN_VALUE (argtype)));
HOST_WIDE_INT type_max = tree_to_uhwi (TYPE_MAX_VALUE (argtype));
tree type = TREE_TYPE (arg);
tree tmin = wide_int_to_tree (type, vr.lower_bound ());
tree tmax = wide_int_to_tree (type, vr.upper_bound ());
*pmin = TREE_INT_CST_LOW (tmin);
*pmax = TREE_INT_CST_LOW (tmax);
if (*pmin < *pmax)
{
/* Return true if the adjusted range is a subrange of
the full range of the argument's type. *PMAX may
be less than *PMIN when the argument is unsigned
and its upper bound is in excess of TYPE_MAX. In
that (invalid) case disregard the range and use that
of the expected type instead. */
knownrange = type_min < *pmin || *pmax < type_max;
unknown = false;
}
}
}
/* Handle an argument with an unknown range as if none had been
provided. */
if (unknown)
return get_int_range (NULL_TREE, NULL, pmin, pmax, absolute,
negbound, query);
}
/* Adjust each bound as specified by ABSOLUTE and NEGBOUND. */
if (absolute)
{
if (*pmin < 0)
{
if (*pmin == *pmax)
*pmin = *pmax = -*pmin;
else
{
/* Make sure signed overlow is avoided. */
gcc_assert (*pmin != HOST_WIDE_INT_MIN);
HOST_WIDE_INT tmp = -*pmin;
*pmin = 0;
if (*pmax < tmp)
*pmax = tmp;
}
}
}
else if (*pmin < negbound)
*pmin = negbound;
return knownrange;
}
/* With the range [*ARGMIN, *ARGMAX] of an integer directive's actual
argument, due to the conversion from either *ARGMIN or *ARGMAX to
the type of the directive's formal argument it's possible for both
to result in the same number of bytes or a range of bytes that's
less than the number of bytes that would result from formatting
some other value in the range [*ARGMIN, *ARGMAX]. This can be
determined by checking for the actual argument being in the range
of the type of the directive. If it isn't it must be assumed to
take on the full range of the directive's type.
Return true when the range has been adjusted to the full range
of DIRTYPE, and false otherwise. */
static bool
adjust_range_for_overflow (tree dirtype, tree *argmin, tree *argmax)
{
tree argtype = TREE_TYPE (*argmin);
unsigned argprec = TYPE_PRECISION (argtype);
unsigned dirprec = TYPE_PRECISION (dirtype);
/* If the actual argument and the directive's argument have the same
precision and sign there can be no overflow and so there is nothing
to adjust. */
if (argprec == dirprec && TYPE_SIGN (argtype) == TYPE_SIGN (dirtype))
return false;
/* The logic below was inspired/lifted from the CONVERT_EXPR_CODE_P
branch in the extract_range_from_unary_expr function in tree-vrp.c. */
if (TREE_CODE (*argmin) == INTEGER_CST
&& TREE_CODE (*argmax) == INTEGER_CST
&& (dirprec >= argprec
|| integer_zerop (int_const_binop (RSHIFT_EXPR,
int_const_binop (MINUS_EXPR,
*argmax,
*argmin),
size_int (dirprec)))))
{
*argmin = force_fit_type (dirtype, wi::to_widest (*argmin), 0, false);
*argmax = force_fit_type (dirtype, wi::to_widest (*argmax), 0, false);
/* If *ARGMIN is still less than *ARGMAX the conversion above
is safe. Otherwise, it has overflowed and would be unsafe. */
if (tree_int_cst_le (*argmin, *argmax))
return false;
}
*argmin = TYPE_MIN_VALUE (dirtype);
*argmax = TYPE_MAX_VALUE (dirtype);
return true;
}
/* Return a range representing the minimum and maximum number of bytes
that the format directive DIR will output for any argument given
the WIDTH and PRECISION (extracted from DIR). This function is
used when the directive argument or its value isn't known. */
static fmtresult
format_integer (const directive &dir, tree arg, range_query *query)
{
tree intmax_type_node;
tree uintmax_type_node;
/* Base to format the number in. */
int base;
/* True when a conversion is preceded by a prefix indicating the base
of the argument (octal or hexadecimal). */
bool maybebase = dir.get_flag ('#');
/* True when a signed conversion is preceded by a sign or space. */
bool maybesign = false;
/* True for signed conversions (i.e., 'd' and 'i'). */
bool sign = false;
switch (dir.specifier)
{
case 'd':
case 'i':
/* Space and '+' are only meaningful for signed conversions. */
maybesign = dir.get_flag (' ') | dir.get_flag ('+');
sign = true;
base = 10;
break;
case 'u':
base = 10;
break;
case 'o':
base = 8;
break;
case 'X':
case 'x':
base = 16;
break;
default:
gcc_unreachable ();
}
/* The type of the "formal" argument expected by the directive. */
tree dirtype = NULL_TREE;
/* Determine the expected type of the argument from the length
modifier. */
switch (dir.modifier)
{
case FMT_LEN_none:
if (dir.specifier == 'p')
dirtype = ptr_type_node;
else
dirtype = sign ? integer_type_node : unsigned_type_node;
break;
case FMT_LEN_h:
dirtype = sign ? short_integer_type_node : short_unsigned_type_node;
break;
case FMT_LEN_hh:
dirtype = sign ? signed_char_type_node : unsigned_char_type_node;
break;
case FMT_LEN_l:
dirtype = sign ? long_integer_type_node : long_unsigned_type_node;
break;
case FMT_LEN_L:
case FMT_LEN_ll:
dirtype = (sign
? long_long_integer_type_node
: long_long_unsigned_type_node);
break;
case FMT_LEN_z:
dirtype = signed_or_unsigned_type_for (!sign, size_type_node);
break;
case FMT_LEN_t:
dirtype = signed_or_unsigned_type_for (!sign, ptrdiff_type_node);
break;
case FMT_LEN_j:
build_intmax_type_nodes (&intmax_type_node, &uintmax_type_node);
dirtype = sign ? intmax_type_node : uintmax_type_node;
break;
default:
return fmtresult ();
}
/* The type of the argument to the directive, either deduced from
the actual non-constant argument if one is known, or from
the directive itself when none has been provided because it's
a va_list. */
tree argtype = NULL_TREE;
if (!arg)
{
/* When the argument has not been provided, use the type of
the directive's argument as an approximation. This will
result in false positives for directives like %i with
arguments with smaller precision (such as short or char). */
argtype = dirtype;
}
else if (TREE_CODE (arg) == INTEGER_CST)
{
/* When a constant argument has been provided use its value
rather than type to determine the length of the output. */
fmtresult res;
if ((dir.prec[0] <= 0 && dir.prec[1] >= 0) && integer_zerop (arg))
{
/* As a special case, a precision of zero with a zero argument
results in zero bytes except in base 8 when the '#' flag is
specified, and for signed conversions in base 8 and 10 when
either the space or '+' flag has been specified and it results
in just one byte (with width having the normal effect). This
must extend to the case of a specified precision with
an unknown value because it can be zero. */
res.range.min = ((base == 8 && dir.get_flag ('#')) || maybesign);
if (res.range.min == 0 && dir.prec[0] != dir.prec[1])
{
res.range.max = 1;
res.range.likely = 1;
}
else
{
res.range.max = res.range.min;
res.range.likely = res.range.min;
}
}
else
{
/* Convert the argument to the type of the directive. */
arg = fold_convert (dirtype, arg);
res.range.min = tree_digits (arg, base, dir.prec[0],
maybesign, maybebase);
if (dir.prec[0] == dir.prec[1])
res.range.max = res.range.min;
else
res.range.max = tree_digits (arg, base, dir.prec[1],
maybesign, maybebase);
res.range.likely = res.range.min;
res.knownrange = true;
}
res.range.unlikely = res.range.max;
/* Bump up the counters if WIDTH is greater than LEN. */
res.adjust_for_width_or_precision (dir.width, dirtype, base,
(sign | maybebase) + (base == 16));
/* Bump up the counters again if PRECision is greater still. */
res.adjust_for_width_or_precision (dir.prec, dirtype, base,
(sign | maybebase) + (base == 16));
return res;
}
else if (INTEGRAL_TYPE_P (TREE_TYPE (arg))
|| TREE_CODE (TREE_TYPE (arg)) == POINTER_TYPE)
/* Determine the type of the provided non-constant argument. */
argtype = TREE_TYPE (arg);
else
/* Don't bother with invalid arguments since they likely would
have already been diagnosed, and disable any further checking
of the format string by returning [-1, -1]. */
return fmtresult ();
fmtresult res;
/* Using either the range the non-constant argument is in, or its
type (either "formal" or actual), create a range of values that
constrain the length of output given the warning level. */
tree argmin = NULL_TREE;
tree argmax = NULL_TREE;
if (arg
&& TREE_CODE (arg) == SSA_NAME
&& INTEGRAL_TYPE_P (argtype))
{
/* Try to determine the range of values of the integer argument
(range information is not available for pointers). */
value_range vr;
query->range_of_expr (vr, arg, dir.info->callstmt);
if (!vr.varying_p () && !vr.undefined_p ())
{
argmin = wide_int_to_tree (TREE_TYPE (arg), vr.lower_bound ());
argmax = wide_int_to_tree (TREE_TYPE (arg), vr.upper_bound ());
/* Set KNOWNRANGE if the argument is in a known subrange
of the directive's type and neither width nor precision
is unknown. (KNOWNRANGE may be reset below). */
res.knownrange
= ((!tree_int_cst_equal (TYPE_MIN_VALUE (dirtype), argmin)
|| !tree_int_cst_equal (TYPE_MAX_VALUE (dirtype), argmax))
&& dir.known_width_and_precision ());
res.argmin = argmin;
res.argmax = argmax;
}
else
{
/* The argument here may be the result of promoting the actual
argument to int. Try to determine the type of the actual
argument before promotion and narrow down its range that
way. */
gimple *def = SSA_NAME_DEF_STMT (arg);
if (is_gimple_assign (def))
{
tree_code code = gimple_assign_rhs_code (def);
if (code == INTEGER_CST)
{
arg = gimple_assign_rhs1 (def);
return format_integer (dir, arg, query);
}
if (code == NOP_EXPR)
{
tree type = TREE_TYPE (gimple_assign_rhs1 (def));
if (INTEGRAL_TYPE_P (type)
|| TREE_CODE (type) == POINTER_TYPE)
argtype = type;
}
}
}
}
if (!argmin)
{
if (TREE_CODE (argtype) == POINTER_TYPE)
{
argmin = build_int_cst (pointer_sized_int_node, 0);
argmax = build_all_ones_cst (pointer_sized_int_node);
}
else
{
argmin = TYPE_MIN_VALUE (argtype);
argmax = TYPE_MAX_VALUE (argtype);
}
}
/* Clear KNOWNRANGE if the range has been adjusted to the maximum
of the directive. If it has been cleared then since ARGMIN and/or
ARGMAX have been adjusted also adjust the corresponding ARGMIN and
ARGMAX in the result to include in diagnostics. */
if (adjust_range_for_overflow (dirtype, &argmin, &argmax))
{
res.knownrange = false;
res.argmin = argmin;
res.argmax = argmax;
}
/* Recursively compute the minimum and maximum from the known range. */
if (TYPE_UNSIGNED (dirtype) || tree_int_cst_sgn (argmin) >= 0)
{
/* For unsigned conversions/directives or signed when
the minimum is positive, use the minimum and maximum to compute
the shortest and longest output, respectively. */
res.range.min = format_integer (dir, argmin, query).range.min;
res.range.max = format_integer (dir, argmax, query).range.max;
}
else if (tree_int_cst_sgn (argmax) < 0)
{
/* For signed conversions/directives if maximum is negative,
use the minimum as the longest output and maximum as the
shortest output. */
res.range.min = format_integer (dir, argmax, query).range.min;
res.range.max = format_integer (dir, argmin, query).range.max;
}
else
{
/* Otherwise, 0 is inside of the range and minimum negative. Use 0
as the shortest output and for the longest output compute the
length of the output of both minimum and maximum and pick the
longer. */
unsigned HOST_WIDE_INT max1
= format_integer (dir, argmin, query).range.max;
unsigned HOST_WIDE_INT max2
= format_integer (dir, argmax, query).range.max;
res.range.min
= format_integer (dir, integer_zero_node, query).range.min;
res.range.max = MAX (max1, max2);
}
/* If the range is known, use the maximum as the likely length. */
if (res.knownrange)
res.range.likely = res.range.max;
else
{
/* Otherwise, use the minimum. Except for the case where for %#x or
%#o the minimum is just for a single value in the range (0) and
for all other values it is something longer, like 0x1 or 01.
Use the length for value 1 in that case instead as the likely
length. */
res.range.likely = res.range.min;
if (maybebase
&& base != 10
&& (tree_int_cst_sgn (argmin) < 0 || tree_int_cst_sgn (argmax) > 0))
{
if (res.range.min == 1)
res.range.likely += base == 8 ? 1 : 2;
else if (res.range.min == 2
&& base == 16
&& (dir.width[0] == 2 || dir.prec[0] == 2))
++res.range.likely;
}
}
res.range.unlikely = res.range.max;
res.adjust_for_width_or_precision (dir.width, dirtype, base,
(sign | maybebase) + (base == 16));
res.adjust_for_width_or_precision (dir.prec, dirtype, base,
(sign | maybebase) + (base == 16));
return res;
}
/* Return the number of bytes that a format directive consisting of FLAGS,
PRECision, format SPECification, and MPFR rounding specifier RNDSPEC,
would result for argument X under ideal conditions (i.e., if PREC
weren't excessive). MPFR 3.1 allocates large amounts of memory for
values of PREC with large magnitude and can fail (see MPFR bug #21056).
This function works around those problems. */
static unsigned HOST_WIDE_INT
get_mpfr_format_length (mpfr_ptr x, const char *flags, HOST_WIDE_INT prec,
char spec, char rndspec)
{
char fmtstr[40];
HOST_WIDE_INT len = strlen (flags);
fmtstr[0] = '%';
memcpy (fmtstr + 1, flags, len);
memcpy (fmtstr + 1 + len, ".*R", 3);
fmtstr[len + 4] = rndspec;
fmtstr[len + 5] = spec;
fmtstr[len + 6] = '\0';
spec = TOUPPER (spec);
if (spec == 'E' || spec == 'F')
{
/* For %e, specify the precision explicitly since mpfr_sprintf
does its own thing just to be different (see MPFR bug 21088). */
if (prec < 0)
prec = 6;
}
else
{
/* Avoid passing negative precisions with larger magnitude to MPFR
to avoid exposing its bugs. (A negative precision is supposed
to be ignored.) */
if (prec < 0)
prec = -1;
}
HOST_WIDE_INT p = prec;
if (spec == 'G' && !strchr (flags, '#'))
{
/* For G/g without the pound flag, precision gives the maximum number
of significant digits which is bounded by LDBL_MAX_10_EXP, or, for
a 128 bit IEEE extended precision, 4932. Using twice as much here
should be more than sufficient for any real format. */
if ((IEEE_MAX_10_EXP * 2) < prec)
prec = IEEE_MAX_10_EXP * 2;
p = prec;
}
else
{
/* Cap precision arbitrarily at 1KB and add the difference
(if any) to the MPFR result. */
if (prec > 1024)
p = 1024;
}
len = mpfr_snprintf (NULL, 0, fmtstr, (int)p, x);
/* Handle the unlikely (impossible?) error by returning more than
the maximum dictated by the function's return type. */
if (len < 0)
return target_dir_max () + 1;
/* Adjust the return value by the difference. */
if (p < prec)
len += prec - p;
return len;
}
/* Return the number of bytes to format using the format specifier
SPEC and the precision PREC the largest value in the real floating
TYPE. */
static unsigned HOST_WIDE_INT
format_floating_max (tree type, char spec, HOST_WIDE_INT prec)
{
machine_mode mode = TYPE_MODE (type);
/* IBM Extended mode. */
if (MODE_COMPOSITE_P (mode))
mode = DFmode;
/* Get the real type format description for the target. */
const real_format *rfmt = REAL_MODE_FORMAT (mode);
REAL_VALUE_TYPE rv;
real_maxval (&rv, 0, mode);
/* Convert the GCC real value representation with the precision
of the real type to the mpfr_t format with the GCC default
round-to-nearest mode. */
mpfr_t x;
mpfr_init2 (x, rfmt->p);
mpfr_from_real (x, &rv, MPFR_RNDN);
/* Return a value one greater to account for the leading minus sign. */
unsigned HOST_WIDE_INT r
= 1 + get_mpfr_format_length (x, "", prec, spec, 'D');
mpfr_clear (x);
return r;
}
/* Return a range representing the minimum and maximum number of bytes
that the directive DIR will output for any argument. PREC gives
the adjusted precision range to account for negative precisions
meaning the default 6. This function is used when the directive
argument or its value isn't known. */
static fmtresult
format_floating (const directive &dir, const HOST_WIDE_INT prec[2])
{
tree type;
switch (dir.modifier)
{
case FMT_LEN_l:
case FMT_LEN_none:
type = double_type_node;
break;
case FMT_LEN_L:
type = long_double_type_node;
break;
case FMT_LEN_ll:
type = long_double_type_node;
break;
default:
return fmtresult ();
}
/* The minimum and maximum number of bytes produced by the directive. */
fmtresult res;
/* The minimum output as determined by flags. It's always at least 1.
When plus or space are set the output is preceded by either a sign
or a space. */
unsigned flagmin = (1 /* for the first digit */
+ (dir.get_flag ('+') | dir.get_flag (' ')));
/* The minimum is 3 for "inf" and "nan" for all specifiers, plus 1
for the plus sign/space with the '+' and ' ' flags, respectively,
unless reduced below. */
res.range.min = 2 + flagmin;
/* When the pound flag is set the decimal point is included in output
regardless of precision. Whether or not a decimal point is included
otherwise depends on the specification and precision. */
bool radix = dir.get_flag ('#');
switch (dir.specifier)
{
case 'A':
case 'a':
{
HOST_WIDE_INT minprec = 6 + !radix /* decimal point */;
if (dir.prec[0] <= 0)
minprec = 0;
else if (dir.prec[0] > 0)
minprec = dir.prec[0] + !radix /* decimal point */;
res.range.likely = (2 /* 0x */
+ flagmin
+ radix
+ minprec
+ 3 /* p+0 */);
res.range.max = format_floating_max (type, 'a', prec[1]);
/* The unlikely maximum accounts for the longest multibyte
decimal point character. */
res.range.unlikely = res.range.max;
if (dir.prec[1] > 0)
res.range.unlikely += target_mb_len_max () - 1;
break;
}
case 'E':
case 'e':
{
/* Minimum output attributable to precision and, when it's
non-zero, decimal point. */
HOST_WIDE_INT minprec = prec[0] ? prec[0] + !radix : 0;
/* The likely minimum output is "[-+]1.234567e+00" regardless
of the value of the actual argument. */
res.range.likely = (flagmin
+ radix
+ minprec
+ 2 /* e+ */ + 2);
res.range.max = format_floating_max (type, 'e', prec[1]);
/* The unlikely maximum accounts for the longest multibyte
decimal point character. */
if (dir.prec[0] != dir.prec[1]
|| dir.prec[0] == -1 || dir.prec[0] > 0)
res.range.unlikely = res.range.max + target_mb_len_max () -1;
else
res.range.unlikely = res.range.max;
break;
}
case 'F':
case 'f':
{
/* Minimum output attributable to precision and, when it's non-zero,
decimal point. */
HOST_WIDE_INT minprec = prec[0] ? prec[0] + !radix : 0;
/* For finite numbers (i.e., not infinity or NaN) the lower bound
when precision isn't specified is 8 bytes ("1.23456" since
precision is taken to be 6). When precision is zero, the lower
bound is 1 byte (e.g., "1"). Otherwise, when precision is greater
than zero, then the lower bound is 2 plus precision (plus flags).
But in all cases, the lower bound is no greater than 3. */
unsigned HOST_WIDE_INT min = flagmin + radix + minprec;
if (min < res.range.min)
res.range.min = min;
/* Compute the upper bound for -TYPE_MAX. */
res.range.max = format_floating_max (type, 'f', prec[1]);
/* The minimum output with unknown precision is a single byte
(e.g., "0") but the more likely output is 3 bytes ("0.0"). */
if (dir.prec[0] < 0 && dir.prec[1] > 0)
res.range.likely = 3;
else
res.range.likely = min;
/* The unlikely maximum accounts for the longest multibyte
decimal point character. */
if (dir.prec[0] != dir.prec[1]
|| dir.prec[0] == -1 || dir.prec[0] > 0)
res.range.unlikely = res.range.max + target_mb_len_max () - 1;
break;
}
case 'G':
case 'g':
{
/* The %g output depends on precision and the exponent of
the argument. Since the value of the argument isn't known
the lower bound on the range of bytes (not counting flags
or width) is 1 plus radix (i.e., either "0" or "0." for
"%g" and "%#g", respectively, with a zero argument). */
unsigned HOST_WIDE_INT min = flagmin + radix;
if (min < res.range.min)
res.range.min = min;
char spec = 'g';
HOST_WIDE_INT maxprec = dir.prec[1];
if (radix && maxprec)
{
/* When the pound flag (radix) is set, trailing zeros aren't
trimmed and so the longest output is the same as for %e,
except with precision minus 1 (as specified in C11). */
spec = 'e';
if (maxprec > 0)
--maxprec;
else if (maxprec < 0)
maxprec = 5;
}
else
maxprec = prec[1];
res.range.max = format_floating_max (type, spec, maxprec);
/* The likely output is either the maximum computed above
minus 1 (assuming the maximum is positive) when precision
is known (or unspecified), or the same minimum as for %e
(which is computed for a non-negative argument). Unlike
for the other specifiers above the likely output isn't
the minimum because for %g that's 1 which is unlikely. */
if (dir.prec[1] < 0
|| (unsigned HOST_WIDE_INT)dir.prec[1] < target_int_max ())
res.range.likely = res.range.max - 1;
else
{
HOST_WIDE_INT minprec = 6 + !radix /* decimal point */;
res.range.likely = (flagmin
+ radix
+ minprec
+ 2 /* e+ */ + 2);
}
/* The unlikely maximum accounts for the longest multibyte
decimal point character. */
res.range.unlikely = res.range.max + target_mb_len_max () - 1;
break;
}
default:
return fmtresult ();
}
/* Bump up the byte counters if WIDTH is greater. */
res.adjust_for_width_or_precision (dir.width);
return res;
}
/* Return a range representing the minimum and maximum number of bytes
that the directive DIR will write on output for the floating argument
ARG. */
static fmtresult
format_floating (const directive &dir, tree arg, range_query *)
{
HOST_WIDE_INT prec[] = { dir.prec[0], dir.prec[1] };
tree type = (dir.modifier == FMT_LEN_L || dir.modifier == FMT_LEN_ll
? long_double_type_node : double_type_node);
/* For an indeterminate precision the lower bound must be assumed
to be zero. */
if (TOUPPER (dir.specifier) == 'A')
{
/* Get the number of fractional decimal digits needed to represent
the argument without a loss of accuracy. */
unsigned fmtprec
= REAL_MODE_FORMAT (TYPE_MODE (type))->p;
/* The precision of the IEEE 754 double format is 53.
The precision of all other GCC binary double formats
is 56 or less. */
unsigned maxprec = fmtprec <= 56 ? 13 : 15;
/* For %a, leave the minimum precision unspecified to let
MFPR trim trailing zeros (as it and many other systems
including Glibc happen to do) and set the maximum
precision to reflect what it would be with trailing zeros
present (as Solaris and derived systems do). */
if (dir.prec[1] < 0)
{
/* Both bounds are negative implies that precision has
not been specified. */
prec[0] = maxprec;
prec[1] = -1;
}
else if (dir.prec[0] < 0)
{
/* With a negative lower bound and a non-negative upper
bound set the minimum precision to zero and the maximum
to the greater of the maximum precision (i.e., with
trailing zeros present) and the specified upper bound. */
prec[0] = 0;
prec[1] = dir.prec[1] < maxprec ? maxprec : dir.prec[1];
}
}
else if (dir.prec[0] < 0)
{
if (dir.prec[1] < 0)
{
/* A precision in a strictly negative range is ignored and
the default of 6 is used instead. */
prec[0] = prec[1] = 6;
}
else
{
/* For a precision in a partly negative range, the lower bound
must be assumed to be zero and the new upper bound is the
greater of 6 (the default precision used when the specified
precision is negative) and the upper bound of the specified
range. */
prec[0] = 0;
prec[1] = dir.prec[1] < 6 ? 6 : dir.prec[1];
}
}
if (!arg
|| TREE_CODE (arg) != REAL_CST
|| !useless_type_conversion_p (type, TREE_TYPE (arg)))
return format_floating (dir, prec);
/* The minimum and maximum number of bytes produced by the directive. */
fmtresult res;
/* Get the real type format description for the target. */
const REAL_VALUE_TYPE *rvp = TREE_REAL_CST_PTR (arg);
const real_format *rfmt = REAL_MODE_FORMAT (TYPE_MODE (TREE_TYPE (arg)));
if (!real_isfinite (rvp))
{
/* The format for Infinity and NaN is "[-]inf"/"[-]infinity"
and "[-]nan" with the choice being implementation-defined
but not locale dependent. */
bool sign = dir.get_flag ('+') || real_isneg (rvp);
res.range.min = 3 + sign;
res.range.likely = res.range.min;
res.range.max = res.range.min;
/* The unlikely maximum is "[-/+]infinity" or "[-/+][qs]nan".
For NaN, the C/POSIX standards specify two formats:
"[-/+]nan"
and
"[-/+]nan(n-char-sequence)"
No known printf implementation outputs the latter format but AIX
outputs QNaN and SNaN for quiet and signalling NaN, respectively,
so the unlikely maximum reflects that. */
res.range.unlikely = sign + (real_isinf (rvp) ? 8 : 4);
/* The range for infinity and NaN is known unless either width
or precision is unknown. Width has the same effect regardless
of whether the argument is finite. Precision is either ignored
(e.g., Glibc) or can have an effect on the short vs long format
such as inf/infinity (e.g., Solaris). */
res.knownrange = dir.known_width_and_precision ();
/* Adjust the range for width but ignore precision. */
res.adjust_for_width_or_precision (dir.width);
return res;
}
char fmtstr [40];
char *pfmt = fmtstr;
/* Append flags. */
for (const char *pf = "-+ #0"; *pf; ++pf)
if (dir.get_flag (*pf))
*pfmt++ = *pf;
*pfmt = '\0';
{
/* Set up an array to easily iterate over. */
unsigned HOST_WIDE_INT* const minmax[] = {
&res.range.min, &res.range.max
};
for (int i = 0; i != sizeof minmax / sizeof *minmax; ++i)
{
/* Convert the GCC real value representation with the precision
of the real type to the mpfr_t format rounding down in the
first iteration that computes the minimum and up in the second
that computes the maximum. This order is arbitrary because
rounding in either direction can result in longer output. */
mpfr_t mpfrval;
mpfr_init2 (mpfrval, rfmt->p);
mpfr_from_real (mpfrval, rvp, i ? MPFR_RNDU : MPFR_RNDD);
/* Use the MPFR rounding specifier to round down in the first
iteration and then up. In most but not all cases this will
result in the same number of bytes. */
char rndspec = "DU"[i];
/* Format it and store the result in the corresponding member
of the result struct. */
*minmax[i] = get_mpfr_format_length (mpfrval, fmtstr, prec[i],
dir.specifier, rndspec);
mpfr_clear (mpfrval);
}
}
/* Make sure the minimum is less than the maximum (MPFR rounding
in the call to mpfr_snprintf can result in the reverse. */
if (res.range.max < res.range.min)
{
unsigned HOST_WIDE_INT tmp = res.range.min;
res.range.min = res.range.max;
res.range.max = tmp;
}
/* The range is known unless either width or precision is unknown. */
res.knownrange = dir.known_width_and_precision ();
/* For the same floating point constant, unless width or precision
is unknown, use the longer output as the likely maximum since
with round to nearest either is equally likely. Otherwise, when
precision is unknown, use the greater of the minimum and 3 as
the likely output (for "0.0" since zero precision is unlikely). */
if (res.knownrange)
res.range.likely = res.range.max;
else if (res.range.min < 3
&& dir.prec[0] < 0
&& (unsigned HOST_WIDE_INT)dir.prec[1] == target_int_max ())
res.range.likely = 3;
else
res.range.likely = res.range.min;
res.range.unlikely = res.range.max;
if (res.range.max > 2 && (prec[0] != 0 || prec[1] != 0))
{
/* Unless the precision is zero output longer than 2 bytes may
include the decimal point which must be a single character
up to MB_LEN_MAX in length. This is overly conservative
since in some conversions some constants result in no decimal
point (e.g., in %g). */
res.range.unlikely += target_mb_len_max () - 1;
}
res.adjust_for_width_or_precision (dir.width);
return res;
}
/* Return a FMTRESULT struct set to the lengths of the shortest and longest
strings referenced by the expression STR, or (-1, -1) when not known.
Used by the format_string function below. */
static fmtresult
get_string_length (tree str, gimple *stmt, unsigned eltsize,
range_query *query)
{
if (!str)
return fmtresult ();
/* Try to determine the dynamic string length first.
Set MAXBOUND to an arbitrary non-null non-integer node as a request
to have it set to the length of the longest string in a PHI. */
c_strlen_data lendata = { };
lendata.maxbound = str;
if (eltsize == 1)
get_range_strlen_dynamic (str, stmt, &lendata, query);
else
{
/* Determine the length of the shortest and longest string referenced
by STR. Strings of unknown lengths are bounded by the sizes of
arrays that subexpressions of STR may refer to. Pointers that
aren't known to point any such arrays result in LENDATA.MAXLEN
set to SIZE_MAX. */
get_range_strlen (str, &lendata, eltsize);
}
/* If LENDATA.MAXBOUND is not equal to .MINLEN it corresponds to the bound
of the largest array STR refers to, if known, or it's set to SIZE_MAX
otherwise. */
/* Return the default result when nothing is known about the string. */
if ((lendata.maxbound && !tree_fits_uhwi_p (lendata.maxbound))
|| !tree_fits_uhwi_p (lendata.maxlen))
{
fmtresult res;
res.nonstr = lendata.decl;
return res;
}
unsigned HOST_WIDE_INT lenmax = tree_to_uhwi (max_object_size ()) - 2;
if (integer_zerop (lendata.minlen)
&& (!lendata.maxbound || lenmax <= tree_to_uhwi (lendata.maxbound))
&& lenmax <= tree_to_uhwi (lendata.maxlen))
{
fmtresult res;
res.nonstr = lendata.decl;
return res;
}
HOST_WIDE_INT min
= (tree_fits_uhwi_p (lendata.minlen)
? tree_to_uhwi (lendata.minlen)
: 0);
HOST_WIDE_INT max
= (lendata.maxbound && tree_fits_uhwi_p (lendata.maxbound)
? tree_to_uhwi (lendata.maxbound)
: HOST_WIDE_INT_M1U);
const bool unbounded = integer_all_onesp (lendata.maxlen);
/* Set the max/likely counters to unbounded when a minimum is known
but the maximum length isn't bounded. This implies that STR is
a conditional expression involving a string of known length and
an expression of unknown/unbounded length. */
if (min
&& (unsigned HOST_WIDE_INT)min < HOST_WIDE_INT_M1U
&& unbounded)
max = HOST_WIDE_INT_M1U;
/* get_range_strlen() returns the target value of SIZE_MAX for
strings of unknown length. Bump it up to HOST_WIDE_INT_M1U
which may be bigger. */
if ((unsigned HOST_WIDE_INT)min == target_size_max ())
min = HOST_WIDE_INT_M1U;
if ((unsigned HOST_WIDE_INT)max == target_size_max ())
max = HOST_WIDE_INT_M1U;
fmtresult res (min, max);
res.nonstr = lendata.decl;
/* Set RES.KNOWNRANGE to true if and only if all strings referenced
by STR are known to be bounded (though not necessarily by their
actual length but perhaps by their maximum possible length). */
if (res.range.max < target_int_max ())
{
res.knownrange = true;
/* When the length of the longest string is known and not
excessive use it as the likely length of the string(s). */
res.range.likely = res.range.max;
}
else
{
/* When the upper bound is unknown (it can be zero or excessive)
set the likely length to the greater of 1. If MAXBOUND is
known, also reset the length of the lower bound to zero. */
res.range.likely = res.range.min ? res.range.min : warn_level > 1;
if (lendata.maxbound && !integer_all_onesp (lendata.maxbound))
res.range.min = 0;
}
res.range.unlikely = unbounded ? HOST_WIDE_INT_MAX : res.range.max;
return res;
}
/* Return the minimum and maximum number of characters formatted
by the '%c' format directives and its wide character form for
the argument ARG. ARG can be null (for functions such as
vsprinf). */
static fmtresult
format_character (const directive &dir, tree arg, range_query *query)
{
fmtresult res;
res.knownrange = true;
if (dir.specifier == 'C'
|| dir.modifier == FMT_LEN_l)
{
/* A wide character can result in as few as zero bytes. */
res.range.min = 0;
HOST_WIDE_INT min, max;
if (get_int_range (arg, dir.info->callstmt, &min, &max, false, 0, query))
{
if (min == 0 && max == 0)
{
/* The NUL wide character results in no bytes. */
res.range.max = 0;
res.range.likely = 0;
res.range.unlikely = 0;
}
else if (min >= 0 && min < 128)
{
/* Be conservative if the target execution character set
is not a 1-to-1 mapping to the source character set or
if the source set is not ASCII. */
bool one_2_one_ascii
= (target_to_host_charmap[0] == 1 && target_to_host ('a') == 97);
/* A wide character in the ASCII range most likely results
in a single byte, and only unlikely in up to MB_LEN_MAX. */
res.range.max = one_2_one_ascii ? 1 : target_mb_len_max ();;
res.range.likely = 1;
res.range.unlikely = target_mb_len_max ();
res.mayfail = !one_2_one_ascii;
}
else
{
/* A wide character outside the ASCII range likely results
in up to two bytes, and only unlikely in up to MB_LEN_MAX. */
res.range.max = target_mb_len_max ();
res.range.likely = 2;
res.range.unlikely = res.range.max;
/* Converting such a character may fail. */
res.mayfail = true;
}
}
else
{
/* An unknown wide character is treated the same as a wide
character outside the ASCII range. */
res.range.max = target_mb_len_max ();
res.range.likely = 2;
res.range.unlikely = res.range.max;
res.mayfail = true;
}
}
else
{
/* A plain '%c' directive. Its output is exactly 1. */
res.range.min = res.range.max = 1;
res.range.likely = res.range.unlikely = 1;
res.knownrange = true;
}
/* Bump up the byte counters if WIDTH is greater. */
return res.adjust_for_width_or_precision (dir.width);
}
/* Determine the offset *INDEX of the first byte of an array element of
TYPE (possibly recursively) into which the byte offset OFF points.
On success set *INDEX to the offset of the first byte and return type.
Otherwise, if no such element can be found, return null. */
static tree
array_elt_at_offset (tree type, HOST_WIDE_INT off, HOST_WIDE_INT *index)
{
gcc_assert (TREE_CODE (type) == ARRAY_TYPE);
tree eltype = type;
while (TREE_CODE (TREE_TYPE (eltype)) == ARRAY_TYPE)
eltype = TREE_TYPE (eltype);
if (TYPE_MODE (TREE_TYPE (eltype)) != TYPE_MODE (char_type_node))
eltype = TREE_TYPE (eltype);
if (eltype == type)
{
*index = 0;
return type;
}
HOST_WIDE_INT typsz = int_size_in_bytes (type);
HOST_WIDE_INT eltsz = int_size_in_bytes (eltype);
if (off < typsz * eltsz)
{
*index = (off / eltsz) * eltsz;
return TREE_CODE (eltype) == ARRAY_TYPE ? TREE_TYPE (eltype) : eltype;
}
return NULL_TREE;
}
/* Determine the offset *INDEX of the first byte of a struct member of TYPE
(possibly recursively) into which the byte offset OFF points. On success
set *INDEX to the offset of the first byte and return true. Otherwise,
if no such member can be found, return false. */
static bool
field_at_offset (tree type, HOST_WIDE_INT off, HOST_WIDE_INT *index)
{
gcc_assert (RECORD_OR_UNION_TYPE_P (type));
for (tree fld = TYPE_FIELDS (type); fld; fld = TREE_CHAIN (fld))
{
if (TREE_CODE (fld) != FIELD_DECL || DECL_ARTIFICIAL (fld))
continue;
tree fldtype = TREE_TYPE (fld);
HOST_WIDE_INT fldoff = int_byte_position (fld);
/* If the size is not available the field is a flexible array
member. Treat this case as success. */
tree typesize = TYPE_SIZE_UNIT (fldtype);
HOST_WIDE_INT fldsize = (tree_fits_uhwi_p (typesize)
? tree_to_uhwi (typesize)
: off);
if (fldoff + fldsize < off)
continue;
if (TREE_CODE (fldtype) == ARRAY_TYPE)
{
HOST_WIDE_INT idx = 0;
if (tree ft = array_elt_at_offset (fldtype, off, &idx))
fldtype = ft;
else
break;
*index += idx;
fldoff -= idx;
off -= idx;
}
if (RECORD_OR_UNION_TYPE_P (fldtype))
{
*index += fldoff;
return field_at_offset (fldtype, off - fldoff, index);
}
*index += fldoff;
return true;
}
return false;
}
/* For an expression X of pointer type, recursively try to find the same
origin (object or pointer) as Y it references and return such an X.
When X refers to a struct member, set *FLDOFF to the offset of the
member from the beginning of the "most derived" object. */
static tree
get_origin_and_offset (tree x, HOST_WIDE_INT *fldoff, HOST_WIDE_INT *off)
{
if (!x)
return NULL_TREE;
switch (TREE_CODE (x))
{
case ADDR_EXPR:
x = TREE_OPERAND (x, 0);
return get_origin_and_offset (x, fldoff, off);
case ARRAY_REF:
{
tree offset = TREE_OPERAND (x, 1);
HOST_WIDE_INT idx = (tree_fits_uhwi_p (offset)
? tree_to_uhwi (offset) : HOST_WIDE_INT_MAX);
tree eltype = TREE_TYPE (x);
if (TREE_CODE (eltype) == INTEGER_TYPE)
{
if (off)
*off = idx;
}
else if (idx < HOST_WIDE_INT_MAX)
*fldoff += idx * int_size_in_bytes (eltype);
else
*fldoff = idx;
x = TREE_OPERAND (x, 0);
return get_origin_and_offset (x, fldoff, NULL);
}
case MEM_REF:
if (off)
{
tree offset = TREE_OPERAND (x, 1);
*off = (tree_fits_uhwi_p (offset)
? tree_to_uhwi (offset) : HOST_WIDE_INT_MAX);
}
x = TREE_OPERAND (x, 0);
if (off)
{
tree xtype
= (TREE_CODE (x) == ADDR_EXPR
? TREE_TYPE (TREE_OPERAND (x, 0)) : TREE_TYPE (TREE_TYPE (x)));
/* The byte offset of the most basic struct member the byte
offset *OFF corresponds to, or for a (multidimensional)
array member, the byte offset of the array element. */
HOST_WIDE_INT index = 0;
if ((RECORD_OR_UNION_TYPE_P (xtype)
&& field_at_offset (xtype, *off, &index))
|| (TREE_CODE (xtype) == ARRAY_TYPE
&& TREE_CODE (TREE_TYPE (xtype)) == ARRAY_TYPE
&& array_elt_at_offset (xtype, *off, &index)))
{
*fldoff += index;
*off -= index;
}
}
return get_origin_and_offset (x, fldoff, NULL);
case COMPONENT_REF:
{
tree fld = TREE_OPERAND (x, 1);
*fldoff += int_byte_position (fld);
get_origin_and_offset (fld, fldoff, off);
x = TREE_OPERAND (x, 0);
return get_origin_and_offset (x, fldoff, off);
}
case SSA_NAME:
{
gimple *def = SSA_NAME_DEF_STMT (x);
if (is_gimple_assign (def))
{
tree_code code = gimple_assign_rhs_code (def);
if (code == ADDR_EXPR)
{
x = gimple_assign_rhs1 (def);
return get_origin_and_offset (x, fldoff, off);
}
if (code == POINTER_PLUS_EXPR)
{
tree offset = gimple_assign_rhs2 (def);
if (off)
*off = (tree_fits_uhwi_p (offset)
? tree_to_uhwi (offset) : HOST_WIDE_INT_MAX);
x = gimple_assign_rhs1 (def);
return get_origin_and_offset (x, fldoff, NULL);
}
else if (code == VAR_DECL)
{
x = gimple_assign_rhs1 (def);
return get_origin_and_offset (x, fldoff, off);
}
}
else if (gimple_nop_p (def) && SSA_NAME_VAR (x))
x = SSA_NAME_VAR (x);
}
default:
break;
}
return x;
}
/* If ARG refers to the same (sub)object or array element as described
by DST and DST_FLD, return the byte offset into the struct member or
array element referenced by ARG. Otherwise return HOST_WIDE_INT_MIN
to indicate that ARG and DST do not refer to the same object. */
static HOST_WIDE_INT
alias_offset (tree arg, tree dst, HOST_WIDE_INT dst_fld)
{
/* See if the argument refers to the same base object as the destination
of the formatted function call, and if so, try to determine if they
can alias. */
if (!arg || !dst || !ptr_derefs_may_alias_p (arg, dst))
return HOST_WIDE_INT_MIN;
/* The two arguments may refer to the same object. If they both refer
to a struct member, see if the members are one and the same. */
HOST_WIDE_INT arg_off = 0, arg_fld = 0;
tree arg_orig = get_origin_and_offset (arg, &arg_fld, &arg_off);
if (arg_orig == dst && arg_fld == dst_fld)
return arg_off;
return HOST_WIDE_INT_MIN;
}
/* Return the minimum and maximum number of characters formatted
by the '%s' format directive and its wide character form for
the argument ARG. ARG can be null (for functions such as
vsprinf). */
static fmtresult
format_string (const directive &dir, tree arg, range_query *query)
{
fmtresult res;
if (warn_restrict)
{
/* See if ARG might alias the destination of the call with
DST_ORIGIN and DST_FIELD. If so, store the starting offset
so that the overlap can be determined for certain later,
when the amount of output of the call (including subsequent
directives) has been computed. Otherwise, store HWI_MIN. */
res.dst_offset = alias_offset (arg, dir.info->dst_origin,
dir.info->dst_field);
}
/* Compute the range the argument's length can be in. */
int count_by = 1;
if (dir.specifier == 'S' || dir.modifier == FMT_LEN_l)
{
/* Get a node for a C type that will be the same size
as a wchar_t on the target. */
tree node = get_typenode_from_name (MODIFIED_WCHAR_TYPE);
/* Now that we have a suitable node, get the number of
bytes it occupies. */
count_by = int_size_in_bytes (node);
gcc_checking_assert (count_by == 2 || count_by == 4);
}
fmtresult slen = get_string_length (arg, dir.info->callstmt, count_by, query);
if (slen.range.min == slen.range.max
&& slen.range.min < HOST_WIDE_INT_MAX)
{
/* The argument is either a string constant or it refers
to one of a number of strings of the same length. */
/* A '%s' directive with a string argument with constant length. */
res.range = slen.range;
if (dir.specifier == 'S'
|| dir.modifier == FMT_LEN_l)
{
/* In the worst case the length of output of a wide string S
is bounded by MB_LEN_MAX * wcslen (S). */
res.range.max *= target_mb_len_max ();
res.range.unlikely = res.range.max;
/* It's likely that the total length is not more that
2 * wcslen (S).*/
res.range.likely = res.range.min * 2;
if (dir.prec[1] >= 0
&& (unsigned HOST_WIDE_INT)dir.prec[1] < res.range.max)
{
res.range.max = dir.prec[1];
res.range.likely = dir.prec[1];
res.range.unlikely = dir.prec[1];
}
if (dir.prec[0] < 0 && dir.prec[1] > -1)
res.range.min = 0;
else if (dir.prec[0] >= 0)
res.range.likely = dir.prec[0];
/* Even a non-empty wide character string need not convert into
any bytes. */
res.range.min = 0;
/* A non-empty wide character conversion may fail. */
if (slen.range.max > 0)
res.mayfail = true;
}
else
{
res.knownrange = true;
if (dir.prec[0] < 0 && dir.prec[1] > -1)
res.range.min = 0;
else if ((unsigned HOST_WIDE_INT)dir.prec[0] < res.range.min)
res.range.min = dir.prec[0];
if ((unsigned HOST_WIDE_INT)dir.prec[1] < res.range.max)
{
res.range.max = dir.prec[1];
res.range.likely = dir.prec[1];
res.range.unlikely = dir.prec[1];
}
}
}
else if (arg && integer_zerop (arg))
{
/* Handle null pointer argument. */
fmtresult res (0);
res.nullp = true;
return res;
}
else
{
/* For a '%s' and '%ls' directive with a non-constant string (either
one of a number of strings of known length or an unknown string)
the minimum number of characters is lesser of PRECISION[0] and
the length of the shortest known string or zero, and the maximum
is the lesser of the length of the longest known string or
PTRDIFF_MAX and PRECISION[1]. The likely length is either
the minimum at level 1 and the greater of the minimum and 1
at level 2. This result is adjust upward for width (if it's
specified). */
if (dir.specifier == 'S'
|| dir.modifier == FMT_LEN_l)
{
/* A wide character converts to as few as zero bytes. */
slen.range.min = 0;
if (slen.range.max < target_int_max ())
slen.range.max *= target_mb_len_max ();
if (slen.range.likely < target_int_max ())
slen.range.likely *= 2;
if (slen.range.likely < target_int_max ())
slen.range.unlikely *= target_mb_len_max ();
/* A non-empty wide character conversion may fail. */
if (slen.range.max > 0)
res.mayfail = true;
}
res.range = slen.range;
if (dir.prec[0] >= 0)
{
/* Adjust the minimum to zero if the string length is unknown,
or at most the lower bound of the precision otherwise. */
if (slen.range.min >= target_int_max ())
res.range.min = 0;
else if ((unsigned HOST_WIDE_INT)dir.prec[0] < slen.range.min)
res.range.min = dir.prec[0];
/* Make both maxima no greater than the upper bound of precision. */
if ((unsigned HOST_WIDE_INT)dir.prec[1] < slen.range.max
|| slen.range.max >= target_int_max ())
{
res.range.max = dir.prec[1];
res.range.unlikely = dir.prec[1];
}
/* If precision is constant, set the likely counter to the lesser
of it and the maximum string length. Otherwise, if the lower
bound of precision is greater than zero, set the likely counter
to the minimum. Otherwise set it to zero or one based on
the warning level. */
if (dir.prec[0] == dir.prec[1])
res.range.likely
= ((unsigned HOST_WIDE_INT)dir.prec[0] < slen.range.max
? dir.prec[0] : slen.range.max);
else if (dir.prec[0] > 0)
res.range.likely = res.range.min;
else
res.range.likely = warn_level > 1;
}
else if (dir.prec[1] >= 0)
{
res.range.min = 0;
if ((unsigned HOST_WIDE_INT)dir.prec[1] < slen.range.max)
res.range.max = dir.prec[1];
res.range.likely = dir.prec[1] ? warn_level > 1 : 0;
if ((unsigned HOST_WIDE_INT)dir.prec[1] < slen.range.unlikely)
res.range.unlikely = dir.prec[1];
}
else if (slen.range.min >= target_int_max ())
{
res.range.min = 0;
res.range.max = HOST_WIDE_INT_MAX;
/* At level 1 strings of unknown length are assumed to be
empty, while at level 1 they are assumed to be one byte
long. */
res.range.likely = warn_level > 1;
res.range.unlikely = HOST_WIDE_INT_MAX;
}
else
{
/* A string of unknown length unconstrained by precision is
assumed to be empty at level 1 and just one character long
at higher levels. */
if (res.range.likely >= target_int_max ())
res.range.likely = warn_level > 1;
}
}
/* If the argument isn't a nul-terminated string and the number
of bytes on output isn't bounded by precision, set NONSTR. */
if (slen.nonstr && slen.range.min < (unsigned HOST_WIDE_INT)dir.prec[0])
res.nonstr = slen.nonstr;
/* Bump up the byte counters if WIDTH is greater. */
return res.adjust_for_width_or_precision (dir.width);
}
/* Format plain string (part of the format string itself). */
static fmtresult
format_plain (const directive &dir, tree, range_query *)
{
fmtresult res (dir.len);
return res;
}
/* Return true if the RESULT of a directive in a call describe by INFO
should be diagnosed given the AVAILable space in the destination. */
static bool
should_warn_p (const call_info &info,
const result_range &avail, const result_range &result)
{
if (result.max <= avail.min)
{
/* The least amount of space remaining in the destination is big
enough for the longest output. */
return false;
}
if (info.bounded)
{
if (warn_format_trunc == 1 && result.min <= avail.max
&& info.retval_used ())
{
/* The likely amount of space remaining in the destination is big
enough for the least output and the return value is used. */
return false;
}
if (warn_format_trunc == 1 && result.likely <= avail.likely
&& !info.retval_used ())
{
/* The likely amount of space remaining in the destination is big
enough for the likely output and the return value is unused. */
return false;
}
if (warn_format_trunc == 2
&& result.likely <= avail.min
&& (result.max <= avail.min
|| result.max > HOST_WIDE_INT_MAX))
{
/* The minimum amount of space remaining in the destination is big
enough for the longest output. */
return false;
}
}
else
{
if (warn_level == 1 && result.likely <= avail.likely)
{
/* The likely amount of space remaining in the destination is big
enough for the likely output. */
return false;
}
if (warn_level == 2
&& result.likely <= avail.min
&& (result.max <= avail.min
|| result.max > HOST_WIDE_INT_MAX))
{
/* The minimum amount of space remaining in the destination is big
enough for the longest output. */
return false;
}
}
return true;
}
/* At format string location describe by DIRLOC in a call described
by INFO, issue a warning for a directive DIR whose output may be
in excess of the available space AVAIL_RANGE in the destination
given the formatting result FMTRES. This function does nothing
except decide whether to issue a warning for a possible write
past the end or truncation and, if so, format the warning.
Return true if a warning has been issued. */
static bool
maybe_warn (substring_loc &dirloc, location_t argloc,
const call_info &info,
const result_range &avail_range, const result_range &res,
const directive &dir)
{
if (!should_warn_p (info, avail_range, res))
return false;
/* A warning will definitely be issued below. */
/* The maximum byte count to reference in the warning. Larger counts
imply that the upper bound is unknown (and could be anywhere between
RES.MIN + 1 and SIZE_MAX / 2) are printed as "N or more bytes" rather
than "between N and X" where X is some huge number. */
unsigned HOST_WIDE_INT maxbytes = target_dir_max ();
/* True when there is enough room in the destination for the least
amount of a directive's output but not enough for its likely or
maximum output. */
bool maybe = (res.min <= avail_range.max
&& (avail_range.min < res.likely
|| (res.max < HOST_WIDE_INT_MAX
&& avail_range.min < res.max)));
/* Buffer for the directive in the host character set (used when
the source character set is different). */
char hostdir[32];
if (avail_range.min == avail_range.max)
{
/* The size of the destination region is exact. */
unsigned HOST_WIDE_INT navail = avail_range.max;
if (target_to_host (*dir.beg) != '%')
{
/* For plain character directives (i.e., the format string itself)
but not others, point the caret at the first character that's
past the end of the destination. */
if (navail < dir.len)
dirloc.set_caret_index (dirloc.get_caret_idx () + navail);
}
if (*dir.beg == '\0')
{
/* This is the terminating nul. */
gcc_assert (res.min == 1 && res.min == res.max);
return fmtwarn (dirloc, UNKNOWN_LOCATION, NULL, info.warnopt (),
info.bounded
? (maybe
? G_("%qE output may be truncated before the "
"last format character")
: G_("%qE output truncated before the last "
"format character"))
: (maybe
? G_("%qE may write a terminating nul past the "
"end of the destination")
: G_("%qE writing a terminating nul past the "
"end of the destination")),
info.func);
}
if (res.min == res.max)
{
const char *d = target_to_host (hostdir, sizeof hostdir, dir.beg);
if (!info.bounded)
return fmtwarn_n (dirloc, argloc, NULL, info.warnopt (), res.min,
"%<%.*s%> directive writing %wu byte into a "
"region of size %wu",
"%<%.*s%> directive writing %wu bytes into a "
"region of size %wu",
(int) dir.len, d, res.min, navail);
else if (maybe)
return fmtwarn_n (dirloc, argloc, NULL, info.warnopt (), res.min,
"%<%.*s%> directive output may be truncated "
"writing %wu byte into a region of size %wu",
"%<%.*s%> directive output may be truncated "
"writing %wu bytes into a region of size %wu",
(int) dir.len, d, res.min, navail);
else
return fmtwarn_n (dirloc, argloc, NULL, info.warnopt (), res.min,
"%<%.*s%> directive output truncated writing "
"%wu byte into a region of size %wu",
"%<%.*s%> directive output truncated writing "
"%wu bytes into a region of size %wu",
(int) dir.len, d, res.min, navail);
}
if (res.min == 0 && res.max < maxbytes)
return fmtwarn (dirloc, argloc, NULL,
info.warnopt (),
info.bounded
? (maybe
? G_("%<%.*s%> directive output may be truncated "
"writing up to %wu bytes into a region of "
"size %wu")
: G_("%<%.*s%> directive output truncated writing "
"up to %wu bytes into a region of size %wu"))
: G_("%<%.*s%> directive writing up to %wu bytes "
"into a region of size %wu"), (int) dir.len,
target_to_host (hostdir, sizeof hostdir, dir.beg),
res.max, navail);
if (res.min == 0 && maxbytes <= res.max)
/* This is a special case to avoid issuing the potentially
confusing warning:
writing 0 or more bytes into a region of size 0. */
return fmtwarn (dirloc, argloc, NULL, info.warnopt (),
info.bounded
? (maybe
? G_("%<%.*s%> directive output may be truncated "
"writing likely %wu or more bytes into a "
"region of size %wu")
: G_("%<%.*s%> directive output truncated writing "
"likely %wu or more bytes into a region of "
"size %wu"))
: G_("%<%.*s%> directive writing likely %wu or more "
"bytes into a region of size %wu"), (int) dir.len,
target_to_host (hostdir, sizeof hostdir, dir.beg),
res.likely, navail);
if (res.max < maxbytes)
return fmtwarn (dirloc, argloc,