https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121404
Bug ID: 121404
Summary: to_chars with hex format produces same output for
different long double inputs on powerpc64le
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: libstdc++
Assignee: unassigned at gcc dot gnu.org
Reporter: cassio.neri at gmail dot com
Target Milestone: ---
I added this test to trunk (be377ef9ce3eb870ffd18ac02dabe32260dfcb6f):
// { dg-do run { target c++14 } }
// { dg-require-string-conversions "" }
#include <charconv>
#include <cstdio>
#include <iostream>
#include <testsuite_hooks.h>
int main() {
long double const x1 = 1.0l + 1.0e-17l;
long double const x2 = 1.0l + .99999999999999899e-17l;
VERIFY(x1 != x2);
char str1[80] = {};
auto const [ptr1, ec1] = std::to_chars(str1, str1 + sizeof(str1) - 1,
x1, std::chars_format::hex);
auto const size1 = ptr1 - str1;
VERIFY(ec1 == std::errc{});
char str2[80] = {};
auto const [ptr2, ec2] = std::to_chars(str2, str2 + sizeof(str2) - 1,
x2, std::chars_format::hex);
auto const size2 = ptr2 - str2;
VERIFY(ec2 == std::errc{});
std::printf("%La\n%La\n", x1, x2);
std::cerr << str1 << '\n' << str2 << '\n';
VERIFY(size1 != size2 || memcmp(str1, str2, size1) != 0);
}
Only the last VERIFY failed so that x1 and x2 are different but str1 and str2
are equal strings. Furtheremore, the results should be the same obtained by
printf with "a" conversion specifier but without "0x". It turns out that printf
produces different strings for x1 and x2. The output is:
0x1.00000000000000b877aa3236a4b8p+0
0x1.00000000000000b877aa3236a48p+0
1.00000000000000b877aa3236a48p+0
1.00000000000000b877aa3236a48p+0
/home/cassio/gcc/src/libstdc++-v3/testsuite/20_util/to_chars/6.cc:28: int
main(): Assertion 'size1 != size2 || memcmp(str1, str2, size1) != 0' failed.
FAIL: 20_util/to_chars/6.cc -std=gnu++17 execution test
FWIW: As I mentioned, in bug 121403, I'm working on to_chars to replace Ryu
with Tejú Jaguá. However, for hex format to_chars doesn't use Ryu so replacing
it won't help. The problem is elsewhere.
