cosmopolitan/third_party/python/Modules/_decimal/libmpdec/crt.c
Justine Tunney 957c61cbbf
Release Cosmopolitan v3.3
This change upgrades to GCC 12.3 and GNU binutils 2.42. The GNU linker
appears to have changed things so that only a single de-duplicated str
table is present in the binary, and it gets placed wherever the linker
wants, regardless of what the linker script says. To cope with that we
need to stop using .ident to embed licenses. As such, this change does
significant work to revamp how third party licenses are defined in the
codebase, using `.section .notice,"aR",@progbits`.

This new GCC 12.3 toolchain has support for GNU indirect functions. It
lets us support __target_clones__ for the first time. This is used for
optimizing the performance of libc string functions such as strlen and
friends so far on x86, by ensuring AVX systems favor a second codepath
that uses VEX encoding. It shaves some latency off certain operations.
It's a useful feature to have for scientific computing for the reasons
explained by the test/libcxx/openmp_test.cc example which compiles for
fifteen different microarchitectures. Thanks to the upgrades, it's now
also possible to use newer instruction sets, such as AVX512FP16, VNNI.

Cosmo now uses the %gs register on x86 by default for TLS. Doing it is
helpful for any program that links `cosmo_dlopen()`. Such programs had
to recompile their binaries at startup to change the TLS instructions.
That's not great, since it means every page in the executable needs to
be faulted. The work of rewriting TLS-related x86 opcodes, is moved to
fixupobj.com instead. This is great news for MacOS x86 users, since we
previously needed to morph the binary every time for that platform but
now that's no longer necessary. The only platforms where we need fixup
of TLS x86 opcodes at runtime are now Windows, OpenBSD, and NetBSD. On
Windows we morph TLS to point deeper into the TIB, based on a TlsAlloc
assignment, and on OpenBSD/NetBSD we morph %gs back into %fs since the
kernels do not allow us to specify a value for the %gs register.

OpenBSD users are now required to use APE Loader to run Cosmo binaries
and assimilation is no longer possible. OpenBSD kernel needs to change
to allow programs to specify a value for the %gs register, or it needs
to stop marking executable pages loaded by the kernel as mimmutable().

This release fixes __constructor__, .ctor, .init_array, and lastly the
.preinit_array so they behave the exact same way as glibc.

We no longer use hex constants to define math.h symbols like M_PI.
2024-02-20 13:27:59 -08:00

152 lines
6 KiB
C

/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:4;tab-width:8;coding:utf-8 -*-│
│ vi: set et ft=c ts=4 sts=4 sw=4 fenc=utf-8 :vi │
╞══════════════════════════════════════════════════════════════════════════════╡
│ Copyright (c) 2008-2016 Stefan Krah. All rights reserved. │
│ │
│ Redistribution and use in source and binary forms, with or without │
│ modification, are permitted provided that the following conditions │
│ are met: │
│ │
│ 1. Redistributions of source code must retain the above copyright │
│ notice, this list of conditions and the following disclaimer. │
│ │
│ 2. Redistributions in binary form must reproduce the above copyright │
│ notice, this list of conditions and the following disclaimer in │
│ the documentation and/or other materials provided with the │
│ distribution. │
│ │
│ THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS "AS IS" AND │
│ ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE │
│ IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR │
│ PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS │
│ BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, │
│ OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT │
│ OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR │
│ BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, │
│ WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE │
│ OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, │
│ EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. │
╚─────────────────────────────────────────────────────────────────────────────*/
#include "third_party/python/Modules/_decimal/libmpdec/crt.h"
#include "third_party/python/Modules/_decimal/libmpdec/mpdecimal.h"
#include "third_party/python/Modules/_decimal/libmpdec/numbertheory.h"
#include "third_party/python/Modules/_decimal/libmpdec/umodarith.h"
__static_yoink("libmpdec_notice");
/* Bignum: Chinese Remainder Theorem, extends the maximum transform length. */
/* Multiply P1P2 by v, store result in w. */
static inline void
_crt_mulP1P2_3(mpd_uint_t w[3], mpd_uint_t v)
{
mpd_uint_t hi1, hi2, lo;
_mpd_mul_words(&hi1, &lo, LH_P1P2, v);
w[0] = lo;
_mpd_mul_words(&hi2, &lo, UH_P1P2, v);
lo = hi1 + lo;
if (lo < hi1) hi2++;
w[1] = lo;
w[2] = hi2;
}
/* Add 3 words from v to w. The result is known to fit in w. */
static inline void
_crt_add3(mpd_uint_t w[3], mpd_uint_t v[3])
{
mpd_uint_t carry;
mpd_uint_t s;
s = w[0] + v[0];
carry = (s < w[0]);
w[0] = s;
s = w[1] + (v[1] + carry);
carry = (s < w[1]);
w[1] = s;
w[2] = w[2] + (v[2] + carry);
}
/* Divide 3 words in u by v, store result in w, return remainder. */
static inline mpd_uint_t
_crt_div3(mpd_uint_t *w, const mpd_uint_t *u, mpd_uint_t v)
{
mpd_uint_t r1 = u[2];
mpd_uint_t r2;
if (r1 < v) {
w[2] = 0;
}
else {
_mpd_div_word(&w[2], &r1, u[2], v); /* GCOV_NOT_REACHED */
}
_mpd_div_words(&w[1], &r2, r1, u[1], v);
_mpd_div_words(&w[0], &r1, r2, u[0], v);
return r1;
}
/*
* Chinese Remainder Theorem:
* Algorithm from Joerg Arndt, "Matters Computational",
* Chapter 37.4.1 [http://www.jjj.de/fxt/]
*
* See also Knuth, TAOCP, Volume 2, 4.3.2, exercise 7.
*/
/*
* CRT with carry: x1, x2, x3 contain numbers modulo p1, p2, p3. For each
* triple of members of the arrays, find the unique z modulo p1*p2*p3, with
* zmax = p1*p2*p3 - 1.
*
* In each iteration of the loop, split z into result[i] = z % MPD_RADIX
* and carry = z / MPD_RADIX. Let N be the size of carry[] and cmax the
* maximum carry.
*
* Limits for the 32-bit build:
*
* N = 2**96
* cmax = 7711435591312380274
*
* Limits for the 64 bit build:
*
* N = 2**192
* cmax = 627710135393475385904124401220046371710
*
* The following statements hold for both versions:
*
* 1) cmax + zmax < N, so the addition does not overflow.
*
* 2) (cmax + zmax) / MPD_RADIX == cmax.
*
* 3) If c <= cmax, then c_next = (c + zmax) / MPD_RADIX <= cmax.
*/
void
crt3(mpd_uint_t *x1, mpd_uint_t *x2, mpd_uint_t *x3, mpd_size_t rsize)
{
mpd_uint_t p1 = mpd_moduli[P1];
mpd_uint_t umod;
mpd_uint_t a1, a2, a3;
mpd_uint_t s;
mpd_uint_t z[3], t[3];
mpd_uint_t carry[3] = {0,0,0};
mpd_uint_t hi, lo;
mpd_size_t i;
for (i = 0; i < rsize; i++) {
a1 = x1[i];
a2 = x2[i];
a3 = x3[i];
SETMODULUS(P2);
s = ext_submod(a2, a1, umod);
s = MULMOD(s, INV_P1_MOD_P2);
_mpd_mul_words(&hi, &lo, s, p1);
lo = lo + a1;
if (lo < a1) hi++;
SETMODULUS(P3);
s = dw_submod(a3, hi, lo, umod);
s = MULMOD(s, INV_P1P2_MOD_P3);
z[0] = lo;
z[1] = hi;
z[2] = 0;
_crt_mulP1P2_3(t, s);
_crt_add3(z, t);
_crt_add3(carry, z);
x1[i] = _crt_div3(carry, carry, MPD_RADIX);
}
assert(carry[0] == 0 && carry[1] == 0 && carry[2] == 0);
}