cosmopolitan/libc/thread/sem_open.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

313 lines
9.9 KiB
C

/*-*- 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 2022 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/calls/blockcancel.internal.h"
#include "libc/calls/calls.h"
#include "libc/calls/struct/stat.h"
#include "libc/dce.h"
#include "libc/errno.h"
#include "libc/intrin/atomic.h"
#include "libc/intrin/strace.internal.h"
#include "libc/limits.h"
#include "libc/mem/mem.h"
#include "libc/runtime/runtime.h"
#include "libc/str/str.h"
#include "libc/sysv/consts/at.h"
#include "libc/sysv/consts/map.h"
#include "libc/sysv/consts/o.h"
#include "libc/sysv/consts/prot.h"
#include "libc/sysv/errfuns.h"
#include "libc/thread/semaphore.h"
#include "libc/thread/thread.h"
#include "libc/thread/tls.h"
static struct Semaphores {
pthread_once_t once;
pthread_mutex_t lock;
struct Semaphore {
struct Semaphore *next;
sem_t *sem;
char *path;
bool dead;
int refs;
} * list;
} g_semaphores;
static void sem_open_lock(void) {
pthread_mutex_lock(&g_semaphores.lock);
}
static void sem_open_unlock(void) {
pthread_mutex_unlock(&g_semaphores.lock);
}
static void sem_open_wipe(void) {
pthread_mutex_init(&g_semaphores.lock, 0);
}
static void sem_open_setup(void) {
sem_open_wipe();
pthread_atfork(sem_open_lock, sem_open_unlock, sem_open_wipe);
}
static void sem_open_init(void) {
pthread_once(&g_semaphores.once, sem_open_setup);
}
static sem_t *sem_open_impl(const char *path, int oflag, unsigned mode,
unsigned value) {
int fd;
sem_t *sem;
struct stat st;
oflag |= O_RDWR | O_CLOEXEC;
if ((fd = openat(AT_FDCWD, path, oflag, mode)) == -1) {
return SEM_FAILED;
}
npassert(!fstat(fd, &st));
if (st.st_size < 4096 && ftruncate(fd, 4096) == -1) {
npassert(!close(fd));
return SEM_FAILED;
}
sem = mmap(0, 4096, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
if (sem != MAP_FAILED) {
atomic_store_explicit(&sem->sem_value, value, memory_order_relaxed);
sem->sem_magic = SEM_MAGIC_NAMED;
sem->sem_dev = st.st_dev;
sem->sem_ino = st.st_ino;
sem->sem_pshared = true;
} else {
sem = SEM_FAILED;
}
npassert(!close(fd));
return sem;
}
static struct Semaphore *sem_open_find(const char *path) {
struct Semaphore *s;
for (s = g_semaphores.list; s; s = s->next) {
if (!strcmp(path, s->path)) {
return s;
}
}
return 0;
}
static struct Semaphore *sem_open_reopen(const char *path) {
int e = errno;
struct stat st;
struct Semaphore *s;
for (s = g_semaphores.list; s; s = s->next) {
if (!s->dead && //
!strcmp(path, s->path)) {
if (!fstatat(AT_FDCWD, path, &st, 0) && //
st.st_dev == s->sem->sem_dev && //
st.st_ino == s->sem->sem_ino) {
return s;
} else {
errno = e;
s->dead = true;
}
}
}
return 0;
}
static struct Semaphore *sem_open_get(const sem_t *sem,
struct Semaphore ***out_prev) {
struct Semaphore *s, **p;
for (p = &g_semaphores.list, s = *p; s; p = &s->next, s = s->next) {
if (s && sem == s->sem) {
*out_prev = p;
return s;
}
}
return 0;
}
/**
* Initializes and opens named semaphore.
*
* This function tracks open semaphore objects within a process. When a
* process calls sem_open() multiple times with the same name, then the
* same shared memory address will be returned, unless it was unlinked.
*
* @param name is arbitrary string that begins with a slash character
* @param oflag can have any of:
* - `O_CREAT` to create the named semaphore if it doesn't exist,
* in which case two additional arguments must be supplied
* - `O_EXCL` to raise `EEXIST` if semaphore already exists
* @param mode is octal mode bits, required if `oflag & O_CREAT`
* @param value is initial of semaphore, required if `oflag & O_CREAT`
* @return semaphore object which needs sem_close(), or SEM_FAILED w/ errno
* @raise ENOSPC if file system is full when `name` would be `O_CREAT`ed
* @raise EINVAL if `oflag` has bits other than `O_CREAT | O_EXCL`
* @raise EINVAL if `value` is negative or exceeds `SEM_VALUE_MAX`
* @raise EEXIST if `O_CREAT|O_EXCL` is used and semaphore exists
* @raise EACCES if we didn't have permission to create semaphore
* @raise EACCES if recreating open semaphore pending an unlink
* @raise EMFILE if process `RLIMIT_NOFILE` has been reached
* @raise ENFILE if system-wide file limit has been reached
* @raise ENOMEM if we require more vespene gas
* @raise EINTR if signal handler was called
*/
sem_t *sem_open(const char *name, int oflag, ...) {
sem_t *sem;
va_list va;
const char *path;
struct Semaphore *s;
char pathbuf[PATH_MAX];
unsigned mode = 0, value = 0;
if (oflag & ~(O_CREAT | O_EXCL)) {
einval();
return SEM_FAILED;
}
if (oflag & O_CREAT) {
va_start(va, oflag);
mode = va_arg(va, unsigned);
value = va_arg(va, unsigned);
va_end(va);
if (value > SEM_VALUE_MAX) {
einval();
return SEM_FAILED;
}
}
if (!(path = sem_path_np(name, pathbuf, sizeof(pathbuf)))) {
return SEM_FAILED;
}
BLOCK_CANCELATION;
sem_open_init();
sem_open_lock();
if ((s = sem_open_reopen(path))) {
if (s->sem->sem_lazydelete) {
if (oflag & O_CREAT) {
eacces();
} else {
enoent();
}
sem = SEM_FAILED;
} else if (~oflag & O_EXCL) {
sem = s->sem;
atomic_fetch_add_explicit(&sem->sem_prefs, 1, memory_order_acq_rel);
++s->refs;
} else {
eexist();
sem = SEM_FAILED;
}
} else if ((s = calloc(1, sizeof(struct Semaphore)))) {
if ((s->path = strdup(path))) {
if ((sem = sem_open_impl(path, oflag, mode, value)) != SEM_FAILED) {
atomic_fetch_add_explicit(&sem->sem_prefs, 1, memory_order_acq_rel);
s->next = g_semaphores.list;
s->sem = sem;
s->refs = 1;
g_semaphores.list = s;
} else {
free(s->path);
free(s);
}
} else {
free(s);
sem = SEM_FAILED;
}
} else {
sem = SEM_FAILED;
}
sem_open_unlock();
ALLOW_CANCELATION;
return sem;
}
/**
* Closes named semaphore.
*
* Calling sem_close() on a semaphore not created by sem_open() has
* undefined behavior. Using `sem` after calling sem_close() from either
* the current process or forked processes sharing the same address is
* also undefined behavior. If any threads in this process or forked
* children are currently blocked on `sem` then calling sem_close() has
* undefined behavior.
*
* @param sem was created with sem_open()
* @return 0 on success, or -1 w/ errno
*/
int sem_close(sem_t *sem) {
int prefs;
bool unmap, delete;
struct Semaphore *s, **p;
npassert(sem->sem_magic == SEM_MAGIC_NAMED);
sem_open_init();
sem_open_lock();
npassert((s = sem_open_get(sem, &p)));
prefs = atomic_fetch_add_explicit(&sem->sem_prefs, -1, memory_order_acq_rel);
npassert(s->refs > 0);
if ((unmap = !--s->refs)) {
npassert(prefs > 0);
delete = sem->sem_lazydelete && prefs == 1;
*p = s->next;
} else {
npassert(prefs > 1);
delete = false;
}
sem_open_unlock();
if (unmap) {
npassert(!munmap(sem, 4096));
}
if (delete) {
unlink(s->path);
}
if (unmap) {
free(s->path);
free(s);
}
return 0;
}
/**
* Removes named semaphore.
*
* This causes the file resource to be deleted. If processes have this
* semaphore currently opened, then on platforms like Windows deletion
* may be postponed until the last process calls sem_close().
*
* @param name can be absolute path or should be component w/o slashes
* @return 0 on success, or -1 w/ errno
* @raise ACCESS if Windows is being fussy about deleting open files
* @raise EPERM if pledge() is in play w/o `cpath` promise
* @raise ENOENT if named semaphore doesn't exist
* @raise EACCES if permission is denied
* @raise ENAMETOOLONG if too long
*/
int sem_unlink(const char *name) {
const char *path;
int rc, e = errno;
struct Semaphore *s;
char pathbuf[PATH_MAX];
if (!(path = sem_path_np(name, pathbuf, sizeof(pathbuf)))) return -1;
if ((rc = unlink(path)) == -1 && IsWindows() && errno == EACCES) {
sem_open_init();
sem_open_lock();
if ((s = sem_open_find(path))) {
s->sem->sem_lazydelete = true;
errno = e;
rc = 0;
}
sem_open_unlock();
}
return rc;
}