mirrors/cosmopolitan
1
0
Fork 0
mirror of https://github.com/jart/cosmopolitan.git synced 2025-01-31 11:37:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
cosmopolitan/libc/intrin/memset.c
Justine Tunney 791f79fcb3
Make improvements 
- We now serialize the file descriptor table when spawning / executing
  processes on Windows. This means you can now inherit more stuff than
  just standard i/o. It's needed by bash, which duplicates the console
  to file descriptor #255. We also now do a better job serializing the
  environment variables, so you're less likely to encounter E2BIG when
  using your bash shell. We also no longer coerce environ to uppercase

- execve() on Windows now remotely controls its parent process to make
  them spawn a replacement for itself. Then it'll be able to terminate
  immediately once the spawn succeeds, without having to linger around
  for the lifetime as a shell process for proxying the exit code. When
  process worker thread running in the parent sees the child die, it's
  given a handle to the new child, to replace it in the process table.

- execve() and posix_spawn() on Windows will now provide CreateProcess
  an explicit handle list. This allows us to remove handle locks which
  enables better fork/spawn concurrency, with seriously correct thread
  safety. Other codebases like Go use the same technique. On the other
  hand fork() still favors the conventional WIN32 inheritence approach
  which can be a little bit messy, but is *controlled* by guaranteeing
  perfectly clean slates at both the spawning and execution boundaries

- sigset_t is now 64 bits. Having it be 128 bits was a mistake because
  there's no reason to use that and it's only supported by FreeBSD. By
  using the system word size, signal mask manipulation on Windows goes
  very fast. Furthermore @asyncsignalsafe funcs have been rewritten on
  Windows to take advantage of signal masking, now that it's much more
  pleasant to use.

- All the overlapped i/o code on Windows has been rewritten for pretty
  good signal and cancelation safety. We're now able to ensure overlap
  data structures are cleaned up so long as you don't longjmp() out of
  out of a signal handler that interrupted an i/o operation. Latencies
  are also improved thanks to the removal of lots of "busy wait" code.
  Waits should be optimal for everything except poll(), which shall be
  the last and final demon we slay in the win32 i/o horror show.

- getrusage() on Windows is now able to report RUSAGE_CHILDREN as well
  as RUSAGE_SELF, thanks to aggregation in the process manager thread.
2023-10-08 08:59:53 -07:00

171 lines
7.8 KiB
C
Raw Blame History

/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:2;tab-width:8;coding:utf-8 -*-│
│vi: set net ft=c ts=2 sts=2 sw=2 fenc=utf-8 :vi│
╞══════════════════════════════════════════════════════════════════════════════╡
│ Copyright 2021 Justine Alexandra Roberts Tunney │
│ │
│ Permission to use, copy, modify, and/or distribute this software for │
│ any purpose with or without fee is hereby granted, provided that the │
│ above copyright notice and this permission notice appear in all copies. │
│ │
│ THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL │
│ WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED │
│ WARRANTIES OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE │
│ AUTHOR BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL │
│ DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR │
│ PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER │
│ TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR │
│ PERFORMANCE OF THIS SOFTWARE. │
╚─────────────────────────────────────────────────────────────────────────────*/
#include "libc/assert.h"
#include "libc/dce.h"
#include "libc/nexgen32e/nexgen32e.h"
#include "libc/nexgen32e/x86feature.h"
#include "libc/str/str.h"
#ifndef __aarch64__
#ifndef __chibicc__
typedef char xmm_t __attribute__((__vector_size__(16), __aligned__(1)));
typedef long long xmm_a __attribute__((__vector_size__(16), __aligned__(16)));
static void *memset_sse(char *p, char c, size_t n) {
xmm_t v = {c, c, c, c, c, c, c, c, c, c, c, c, c, c, c, c};
if (n <= 32) {
*(xmm_t *)(p + n - 16) = v;
*(xmm_t *)p = v;
} else {
do {
n -= 32;
*(xmm_t *)(p + n) = v;
*(xmm_t *)(p + n + 16) = v;
} while (n > 32);
*(xmm_t *)(p + 16) = v;
*(xmm_t *)p = v;
}
return p;
}
#endif
#if defined(__x86_64__) && !defined(__chibicc__)
_Microarchitecture("avx") static void *memset_avx(char *p, char c, size_t n) {
char *t;
xmm_t v = {c, c, c, c, c, c, c, c, c, c, c, c, c, c, c, c};
if (n <= 32) {
*(xmm_t *)(p + n - 16) = v;
*(xmm_t *)p = v;
} else if (n >= 1024 && X86_HAVE(ERMS)) {
asm("rep stosb" : "=D"(t), "+c"(n), "=m"(*(char(*)[n])p) : "0"(p), "a"(c));
} else {
if (n < kHalfCache3 || !kHalfCache3) {
do {
n -= 32;
*(xmm_t *)(p + n) = v;
*(xmm_t *)(p + n + 16) = v;
} while (n > 32);
} else {
while ((uintptr_t)(p + n) & 15) {
p[--n] = c;
}
do {
n -= 32;
__builtin_ia32_movntdq((xmm_a *)(p + n), (xmm_a)v);
__builtin_ia32_movntdq((xmm_a *)(p + n + 16), (xmm_a)v);
} while (n > 32);
asm("sfence");
}
*(xmm_t *)(p + 16) = v;
*(xmm_t *)p = v;
}
return p;
}
#endif /* __x86_64__ */
/**
* Sets memory.
*
* memset n=0 992 picoseconds
* memset n=1 992 ps/byte 984 mb/s
* memset n=2 330 ps/byte 2,952 mb/s
* memset n=3 330 ps/byte 2,952 mb/s
* memset n=4 165 ps/byte 5,904 mb/s
* memset n=7 94 ps/byte 10,333 mb/s
* memset n=8 124 ps/byte 7,872 mb/s
* memset n=15 66 ps/byte 14,761 mb/s
* memset n=16 62 ps/byte 15,745 mb/s
* memset n=31 32 ps/byte 30,506 mb/s
* memset n=32 20 ps/byte 47,236 mb/s
* memset n=63 26 ps/byte 37,198 mb/s
* memset n=64 20 ps/byte 47,236 mb/s
* memset n=127 23 ps/byte 41,660 mb/s
* memset n=128 12 ps/byte 75,578 mb/s
* memset n=255 18 ps/byte 53,773 mb/s
* memset n=256 12 ps/byte 75,578 mb/s
* memset n=511 17 ps/byte 55,874 mb/s
* memset n=512 12 ps/byte 75,578 mb/s
* memset n=1023 16 ps/byte 58,080 mb/s
* memset n=1024 11 ps/byte 86,375 mb/s
* memset n=2047 9 ps/byte 101 gb/s
* memset n=2048 8 ps/byte 107 gb/s
* memset n=4095 8 ps/byte 113 gb/s
* memset n=4096 8 ps/byte 114 gb/s
* memset n=8191 7 ps/byte 126 gb/s
* memset n=8192 7 ps/byte 126 gb/s
* memset n=16383 7 ps/byte 133 gb/s
* memset n=16384 7 ps/byte 131 gb/s
* memset n=32767 14 ps/byte 69,246 mb/s
* memset n=32768 6 ps/byte 138 gb/s
* memset n=65535 15 ps/byte 62,756 mb/s
* memset n=65536 15 ps/byte 62,982 mb/s
* memset n=131071 18 ps/byte 52,834 mb/s
* memset n=131072 15 ps/byte 62,023 mb/s
* memset n=262143 15 ps/byte 61,169 mb/s
* memset n=262144 16 ps/byte 61,011 mb/s
* memset n=524287 16 ps/byte 60,633 mb/s
* memset n=524288 16 ps/byte 57,902 mb/s
* memset n=1048575 16 ps/byte 60,405 mb/s
* memset n=1048576 16 ps/byte 58,754 mb/s
* memset n=2097151 16 ps/byte 59,329 mb/s
* memset n=2097152 16 ps/byte 58,729 mb/s
* memset n=4194303 16 ps/byte 59,329 mb/s
* memset n=4194304 16 ps/byte 59,262 mb/s
* memset n=8388607 16 ps/byte 59,530 mb/s
* memset n=8388608 16 ps/byte 60,205 mb/s
*
* @param p is memory address
* @param c is masked with 255 and used as repeated byte
* @param n is byte length
* @return p
* @asyncsignalsafe
*/
void *memset(void *p, int c, size_t n) {
char *b;
uint32_t u;
uint64_t x;
b = p;
if (n <= 16) {
if (n >= 8) {
x = 0x0101010101010101ul * (c & 255);
__builtin_memcpy(b, &x, 8);
__builtin_memcpy(b + n - 8, &x, 8);
} else if (n >= 4) {
u = 0x01010101u * (c & 255);
__builtin_memcpy(b, &u, 4);
__builtin_memcpy(b + n - 4, &u, 4);
} else if (n) {
do {
asm volatile("" ::: "memory");
b[--n] = c;
} while (n);
}
return b;
#if defined(__x86_64__) && !defined(__chibicc__)
} else if (IsTiny()) {
asm("rep stosb" : "+D"(b), "+c"(n), "=m"(*(char(*)[n])b) : "a"(c));
return p;
} else if (X86_HAVE(AVX)) {
return memset_avx(b, c, n);
#endif
} else {
return memset_sse(b, c, n);
}
}
#endif /* __aarch64__ */
Reference in a new issue View git blame Copy permalink