mirrors/cosmopolitan
1
0
Fork 0
mirror of https://github.com/jart/cosmopolitan.git synced 2025-02-01 03:53:33 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
cosmopolitan/test/libc/stdio/posix_spawn_test.c

196 lines
7.8 KiB
C
Raw Normal View History

Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
/*-*- 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. │
╚─────────────────────────────────────────────────────────────────────────────*/
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
#include "libc/atomic.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
#include "libc/calls/calls.h"
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
#include "libc/calls/state.internal.h"
#include "libc/calls/struct/sigaction.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
#include "libc/dce.h"
#include "libc/fmt/conv.h"
Revamp posix_spawn() implementation
2022-10-12 12:26:58 +00:00
#include "libc/intrin/kprintf.h"
Clean up some code
2022-08-20 19:32:51 +00:00
#include "libc/intrin/safemacros.internal.h"
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
#include "libc/mem/gc.h"
#include "libc/mem/mem.h"
Unbloat build config - 10.5% reduction of o//depend dependency graph - 8.8% reduction in latency of make command - Fix issue with temporary file cleanup There's a new -w option in compile.com that turns off the recent Landlock output path workaround for "good commands" which do not unlink() the output file like GNU tooling does. Our new GNU Make unveil sandboxing appears to have zero overhead in the grand scheme of things. Full builds are pretty fast since the only thing that's actually slowed us down is probably libcxx make -j16 MODE=rel RL: took 85,732,063µs wall time RL: ballooned to 323,612kb in size RL: needed 828,560,521µs cpu (11% kernel) RL: caused 39,080,670 page faults (99% memcpy) RL: 350,073 context switches (72% consensual) RL: performed 0 reads and 11,494,960 write i/o operations pledge() and unveil() no longer consider ENOSYS to be an error. These functions have also been added to Python's cosmo module. This change also removes some WIN32 APIs and System Five magnums which we're not using and it's doubtful anyone else would be too
2022-08-10 08:32:17 +00:00
#include "libc/runtime/internal.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
#include "libc/runtime/runtime.h"
Revamp posix_spawn() implementation
2022-10-12 12:26:58 +00:00
#include "libc/stdio/posix_spawn.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
#include "libc/stdio/stdio.h"
#include "libc/sysv/consts/auxv.h"
#include "libc/sysv/consts/o.h"
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
#include "libc/sysv/consts/sig.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
#include "libc/testlib/ezbench.h"
#include "libc/testlib/testlib.h"
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
#include "libc/thread/thread.h"
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
Greatly expand system() shell code features The cosmopolitan command interpreter now has 13 builtin commands, variable support, support for ; / && / || syntax, asynchronous support, and plenty of unit tests with bug fixes. This change fixes a bug in posix_spawn() with null envp arg. strace logging now uses atomic writes for scatter functions. Breaking change renaming GetCpuCount() to _getcpucount(). TurfWar is now updated to use the new token bucket algorithm. WIN32 affinity masks now inherit across fork() and execve().
2022-10-12 04:06:27 +00:00
char testlib_enable_tmp_setup_teardown;
Release pledge.com v1.6
2022-08-10 15:11:05 +00:00
__attribute__((__constructor__)) static void init(void) {
Greatly expand system() shell code features The cosmopolitan command interpreter now has 13 builtin commands, variable support, support for ; / && / || syntax, asynchronous support, and plenty of unit tests with bug fixes. This change fixes a bug in posix_spawn() with null envp arg. strace logging now uses atomic writes for scatter functions. Breaking change renaming GetCpuCount() to _getcpucount(). TurfWar is now updated to use the new token bucket algorithm. WIN32 affinity masks now inherit across fork() and execve().
2022-10-12 04:06:27 +00:00
switch (atoi(nulltoempty(getenv("THE_DOGE")))) {
case 42:
exit(42);
default:
break;
Release pledge.com v1.6
2022-08-10 15:11:05 +00:00
}
Unbloat build config - 10.5% reduction of o//depend dependency graph - 8.8% reduction in latency of make command - Fix issue with temporary file cleanup There's a new -w option in compile.com that turns off the recent Landlock output path workaround for "good commands" which do not unlink() the output file like GNU tooling does. Our new GNU Make unveil sandboxing appears to have zero overhead in the grand scheme of things. Full builds are pretty fast since the only thing that's actually slowed us down is probably libcxx make -j16 MODE=rel RL: took 85,732,063µs wall time RL: ballooned to 323,612kb in size RL: needed 828,560,521µs cpu (11% kernel) RL: caused 39,080,670 page faults (99% memcpy) RL: 350,073 context switches (72% consensual) RL: performed 0 reads and 11,494,960 write i/o operations pledge() and unveil() no longer consider ENOSYS to be an error. These functions have also been added to Python's cosmo module. This change also removes some WIN32 APIs and System Five magnums which we're not using and it's doubtful anyone else would be too
2022-08-10 08:32:17 +00:00
}
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
TEST(posix_spawn, test) {
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
int rc, ws, pid;
Fix bugs with recent change This change makes further effort towards improving our poll() implementation on the New Technology. The stdin worker didn't work out so well for Python so it's not being used for now. System call tracing with the --strace flag should now be less noisy now on Windows unless you modify the strace.internal.h defines to turn on some optional ones that are most useful for debugging the system call wrappers.
2022-04-16 17:40:23 +00:00
char *prog = GetProgramExecutableName();
GetZipCdir: prevent integer underflow. posix_spawn_test.c: stop attempting to load zipos from /usr/bin/ape. (#758)
2023-03-05 22:29:38 +00:00
char *args[] = {prog, NULL};
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
char *envs[] = {"THE_DOGE=42", NULL};
EXPECT_EQ(0, posix_spawn(&pid, prog, NULL, NULL, args, envs));
EXPECT_NE(-1, waitpid(pid, &ws, 0));
EXPECT_TRUE(WIFEXITED(ws));
EXPECT_EQ(42, WEXITSTATUS(ws));
}
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
TEST(posix_spawn, pipe) {
Greatly expand system() shell code features The cosmopolitan command interpreter now has 13 builtin commands, variable support, support for ; / && / || syntax, asynchronous support, and plenty of unit tests with bug fixes. This change fixes a bug in posix_spawn() with null envp arg. strace logging now uses atomic writes for scatter functions. Breaking change renaming GetCpuCount() to _getcpucount(). TurfWar is now updated to use the new token bucket algorithm. WIN32 affinity masks now inherit across fork() and execve().
2022-10-12 04:06:27 +00:00
char buf[10];
int p[2], pid, status;
const char *pn = "./echo.com";
posix_spawn_file_actions_t fa;
testlib_extract("/zip/echo.com", "echo.com", 0755);
ASSERT_SYS(0, 0, pipe(p));
ASSERT_EQ(0, posix_spawn_file_actions_init(&fa));
ASSERT_EQ(0, posix_spawn_file_actions_addclose(&fa, p[0]));
ASSERT_EQ(0, posix_spawn_file_actions_adddup2(&fa, p[1], 1));
ASSERT_EQ(0, posix_spawn_file_actions_addclose(&fa, p[1]));
ASSERT_EQ(0, posix_spawnp(&pid, pn, &fa, 0, (char *[]){pn, "hello", 0}, 0));
ASSERT_SYS(0, 0, close(p[1]));
ASSERT_SYS(0, pid, waitpid(pid, &status, 0));
ASSERT_SYS(0, 6, read(p[0], buf, sizeof(buf)));
ASSERT_SYS(0, 0, close(p[0]));
ASSERT_EQ(0, posix_spawn_file_actions_destroy(&fa));
}
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
_Thread_local atomic_int gotsome;
void OhMyGoth(int sig) {
++gotsome;
}
// time for a vfork() clone() signal bloodbath
TEST(posix_spawn, torture) {
int n = 10;
int ws, pid;
short flags;
sigset_t allsig;
posix_spawnattr_t attr;
posix_spawn_file_actions_t fa;
signal(SIGUSR2, OhMyGoth);
sigfillset(&allsig);
if (!fileexists("echo.com")) {
testlib_extract("/zip/echo.com", "echo.com", 0755);
}
// XXX: NetBSD doesn't seem to let us set the scheduler to itself ;_;
ASSERT_EQ(0, posix_spawnattr_init(&attr));
ASSERT_EQ(0, posix_spawnattr_setflags(
Make improvements - Improve compatibility with Blink virtual machine - Add non-POSIX APIs for joining threads and signal masks - Never ever use anything except 32-bit integers for atomics - Add some `#undef` statements to workaround `ctags` problems
2022-11-11 05:52:47 +00:00
&attr, POSIX_SPAWN_SETSIGDEF | POSIX_SPAWN_SETSIGDEF |
POSIX_SPAWN_SETPGROUP | POSIX_SPAWN_SETSIGMASK |
Implement pthread_atfork() If threads are being used, then fork() will now acquire and release and runtime locks so that fork() may be safely used from threads. This also makes vfork() thread safe, because pthread mutexes will do nothing when the process is a child of vfork(). More torture tests have been written to confirm this all works like a charm. Additionally: - Invent hexpcpy() api - Rename nsync_malloc_() to kmalloc() - Complete posix named semaphore implementation - Make pthread_create() asynchronous signal safe - Add rm, rmdir, and touch to command interpreter builtins - Invent sigisprecious() and modify sigset functions to use it - Add unit tests for posix_spawn() attributes and fix its bugs One unresolved problem is the reclaiming of *NSYNC waiter memory in the forked child processes, within apps which have threads waiting on locks
2022-10-16 19:05:08 +00:00
(IsNetbsd() ? 0 : POSIX_SPAWN_SETSCHEDULER)));
ASSERT_EQ(0, posix_spawnattr_setsigmask(&attr, &allsig));
ASSERT_EQ(0, posix_spawnattr_setsigdefault(&attr, &allsig));
ASSERT_EQ(0, posix_spawn_file_actions_init(&fa));
ASSERT_EQ(0, posix_spawn_file_actions_addclose(&fa, 0));
ASSERT_EQ(
0, posix_spawn_file_actions_addopen(&fa, 1, "/dev/null", O_WRONLY, 0644));
ASSERT_EQ(0, posix_spawn_file_actions_adddup2(&fa, 1, 0));
for (int i = 0; i < n; ++i) {
char *volatile zzz = malloc(13);
volatile int fd = open("/dev/null", O_WRONLY);
char *args[] = {"./echo.com", NULL};
char *envs[] = {NULL};
raise(SIGUSR2);
ASSERT_EQ(0, posix_spawn(&pid, "./echo.com", &fa, &attr, args, envs));
ASSERT_FALSE(__vforked);
ASSERT_NE(-1, waitpid(pid, &ws, 0));
ASSERT_TRUE(WIFEXITED(ws));
ASSERT_EQ(0, WEXITSTATUS(ws));
close(fd);
free(zzz);
}
// TODO(jart): Why does it deliver 2x the signals sometimes?
ASSERT_NE(0, gotsome);
ASSERT_EQ(0, posix_spawn_file_actions_destroy(&fa));
ASSERT_EQ(0, posix_spawnattr_destroy(&attr));
}
void *Torturer(void *arg) {
posix_spawn_torture();
return 0;
}
TEST(posix_spawn, agony) {
int i, n = 3;
pthread_t *t = _gc(malloc(sizeof(pthread_t) * n));
testlib_extract("/zip/echo.com", "echo.com", 0755);
for (i = 0; i < n; ++i) {
ASSERT_EQ(0, pthread_create(t + i, 0, Torturer, 0));
}
for (i = 0; i < n; ++i) {
ASSERT_EQ(0, pthread_join(t[i], 0));
}
}
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
/*
* BEST LINUX FORK+EXEC+EXIT+WAIT PERFORMANCE
* The fastest it can go with fork() is 40µs
* The fastest it can go with vfork() is 26µs
*/
void BenchmarkProcessLifecycle(void) {
int rc, ws, pid;
char *prog = "/tmp/tiny64";
char *args[] = {"tiny64", NULL};
char *envs[] = {NULL};
ASSERT_EQ(0, posix_spawn(&pid, prog, NULL, NULL, args, envs));
ASSERT_NE(-1, waitpid(pid, &ws, 0));
ASSERT_TRUE(WIFEXITED(ws));
ASSERT_EQ(42, WEXITSTATUS(ws));
}
const char kTinyLinuxExit[128] = {
0x7f, 0x45, 0x4c, 0x46, 0x02, 0x01, 0x01, 0x00, // ⌂ELF☻☺☺ 
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //         
0x02, 0x00, 0x3e, 0x00, 0x01, 0x00, 0x00, 0x00, // ☻ > ☺   
Fix writev() on the New Technology (#117)
2021-03-09 19:21:08 +00:00
0x78, 0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x00, // x @     
Add posix_spawn (#114)
2021-03-08 04:14:07 +00:00
0x40, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // @       
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //         
0x00, 0x00, 0x00, 0x00, 0x40, 0x00, 0x38, 0x00, //     @ 8 
0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // ☺       
0x01, 0x00, 0x00, 0x00, 0x05, 0x00, 0x00, 0x00, // ☺   ♣   
0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //         
0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x00, //   @     
0x00, 0x00, 0x40, 0x00, 0x00, 0x00, 0x00, 0x00, //   @     
0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // Ç       
0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // Ç       
0x00, 0x10, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //  ►      
0x6a, 0x2a, 0x5f, 0x6a, 0x3c, 0x58, 0x0f, 0x05, // j*_j<X☼♣
};
Unbloat the build This change resurrects ae5d06dc53b4356966e36a4dcf40cb192e06618f
2022-08-11 07:15:29 +00:00
/* BENCH(spawn, bench) { */
/* int fd; */
/* if (IsLinux()) { */
/* fd = open("/tmp/tiny64", O_CREAT | O_TRUNC | O_WRONLY, 0755); */
/* write(fd, kTinyLinuxExit, 128); */
/* close(fd); */
/* EZBENCH2("spawn", donothing, BenchmarkProcessLifecycle()); */
/* unlink("/tmp/tiny64"); */
/* } */
/* } */
Reference in a new issue Copy permalink