cosmopolitan/ape/ape-m1.c
2024-05-02 23:21:43 -07:00

1127 lines
36 KiB
C

/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:2;tab-width:8;coding:utf-8 -*-│
│ vi: set et 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 <assert.h>
#include <dispatch/dispatch.h>
#include <dlfcn.h>
#include <errno.h>
#include <fcntl.h>
#include <libkern/OSCacheControl.h>
#include <limits.h>
#include <pthread.h>
#include <semaphore.h>
#include <signal.h>
#include <stdarg.h>
#include <stdint.h>
#include <string.h>
#include <sys/mman.h>
#include <sys/random.h>
#include <sys/sysctl.h>
#include <sys/types.h>
#include <sys/uio.h>
#include <time.h>
#include <unistd.h>
#define pagesz 16384
/* maximum path size that cosmo can take */
#define PATHSIZE (PATH_MAX < 1024 ? PATH_MAX : 1024)
#define SYSLIB_MAGIC ('s' | 'l' << 8 | 'i' << 16 | 'b' << 24)
#define SYSLIB_VERSION 10 /* sync with libc/runtime/syslib.internal.h */
struct Syslib {
int magic;
int version;
long (*fork)(void);
long (*pipe)(int[2]);
long (*clock_gettime)(int, struct timespec *);
long (*nanosleep)(const struct timespec *, struct timespec *);
long (*mmap)(void *, size_t, int, int, int, off_t);
int (*pthread_jit_write_protect_supported_np)(void);
void (*pthread_jit_write_protect_np)(int);
void (*sys_icache_invalidate)(void *, size_t);
int (*pthread_create)(pthread_t *, const pthread_attr_t *, void *(*)(void *),
void *);
void (*pthread_exit)(void *);
int (*pthread_kill)(pthread_t, int);
int (*pthread_sigmask)(int, const sigset_t *, sigset_t *);
int (*pthread_setname_np)(const char *);
dispatch_semaphore_t (*dispatch_semaphore_create)(long);
long (*dispatch_semaphore_signal)(dispatch_semaphore_t);
long (*dispatch_semaphore_wait)(dispatch_semaphore_t, dispatch_time_t);
dispatch_time_t (*dispatch_walltime)(const struct timespec *, int64_t);
/* v2 (2023-09-10) */
pthread_t (*pthread_self)(void);
void (*dispatch_release)(dispatch_semaphore_t);
long (*raise)(int);
int (*pthread_join)(pthread_t, void **);
void (*pthread_yield_np)(void);
int pthread_stack_min;
int sizeof_pthread_attr_t;
int (*pthread_attr_init)(pthread_attr_t *);
int (*pthread_attr_destroy)(pthread_attr_t *);
int (*pthread_attr_setstacksize)(pthread_attr_t *, size_t);
int (*pthread_attr_setguardsize)(pthread_attr_t *, size_t);
/* v4 (2023-09-19) */
void (*exit)(int);
long (*close)(int);
long (*munmap)(void *, size_t);
long (*openat)(int, const char *, int, int);
long (*write)(int, const void *, size_t);
long (*read)(int, void *, size_t);
long (*sigaction)(int, const struct sigaction *, struct sigaction *);
long (*pselect)(int, fd_set *, fd_set *, fd_set *, const struct timespec *,
const sigset_t *);
long (*mprotect)(void *, size_t, int);
/* v5 (2023-10-09) */
long (*sigaltstack)(const stack_t *, stack_t *);
long (*getentropy)(void *, size_t);
long (*sem_open)(const char *, int, uint16_t, unsigned);
long (*sem_unlink)(const char *);
long (*sem_close)(int *);
long (*sem_post)(int *);
long (*sem_wait)(int *);
long (*sem_trywait)(int *);
long (*getrlimit)(int, struct rlimit *);
long (*setrlimit)(int, const struct rlimit *);
/* v6 (2023-11-03) */
void *(*dlopen)(const char *, int);
void *(*dlsym)(void *, const char *);
int (*dlclose)(void *);
char *(*dlerror)(void);
/* MANDATORY (cosmo runtime won't load if version < 8)
---------------------------------------------------
OPTIONAL (cosmo lib should check __syslib->version) */
/* v9 (2024-01-31) */
int (*pthread_cpu_number_np)(size_t *);
/* v10 (2024-05-02) */
long (*sysctl)(int *, u_int, void *, size_t *, void *, size_t);
long (*sysctlbyname)(const char *, void *, size_t *, void *, size_t);
long (*sysctlnametomib)(const char *, int *, size_t *);
};
#define ELFCLASS32 1
#define ELFDATA2LSB 1
#define EM_AARCH64 183
#define ET_EXEC 2
#define ET_DYN 3
#define PT_LOAD 1
#define PT_DYNAMIC 2
#define PT_INTERP 3
#define EI_CLASS 4
#define EI_DATA 5
#define PF_X 1
#define PF_W 2
#define PF_R 4
#define AT_PHDR 3
#define AT_PHENT 4
#define AT_PHNUM 5
#define AT_PAGESZ 6
#define AT_BASE 7
#define AT_FLAGS 8
#define AT_FLAGS_PRESERVE_ARGV0_BIT 0
#define AT_FLAGS_PRESERVE_ARGV0 (1 << AT_FLAGS_PRESERVE_ARGV0_BIT)
#define AT_ENTRY 9
#define AT_UID 11
#define AT_EUID 12
#define AT_GID 13
#define AT_EGID 14
#define AT_HWCAP 16
#define AT_HWCAP2 16
#define AT_SECURE 23
#define AT_RANDOM 25
#define AT_EXECFN 31
#define AUXV_WORDS 31
/* from the xnu codebase */
#define _COMM_PAGE_START_ADDRESS 0x0000000FFFFFC000ul
#define _COMM_PAGE_APRR_SUPPORT (_COMM_PAGE_START_ADDRESS + 0x10C)
#define _COMM_PAGE_APRR_WRITE_ENABLE (_COMM_PAGE_START_ADDRESS + 0x110)
#define _COMM_PAGE_APRR_WRITE_DISABLE (_COMM_PAGE_START_ADDRESS + 0x118)
#define Min(X, Y) ((Y) > (X) ? (X) : (Y))
#define Max(X, Y) ((Y) < (X) ? (X) : (Y))
#define READ32(S) \
((unsigned)(255 & (S)[3]) << 030 | (unsigned)(255 & (S)[2]) << 020 | \
(unsigned)(255 & (S)[1]) << 010 | (unsigned)(255 & (S)[0]) << 000)
#define READ64(S) \
((unsigned long)(255 & (S)[7]) << 070 | \
(unsigned long)(255 & (S)[6]) << 060 | \
(unsigned long)(255 & (S)[5]) << 050 | \
(unsigned long)(255 & (S)[4]) << 040 | \
(unsigned long)(255 & (S)[3]) << 030 | \
(unsigned long)(255 & (S)[2]) << 020 | \
(unsigned long)(255 & (S)[1]) << 010 | \
(unsigned long)(255 & (S)[0]) << 000)
struct ElfEhdr {
unsigned char e_ident[16];
unsigned short e_type;
unsigned short e_machine;
unsigned e_version;
unsigned long e_entry;
unsigned long e_phoff;
unsigned long e_shoff;
unsigned e_flags;
unsigned short e_ehsize;
unsigned short e_phentsize;
unsigned short e_phnum;
unsigned short e_shentsize;
unsigned short e_shnum;
unsigned short e_shstrndx;
};
struct ElfPhdr {
unsigned p_type;
unsigned p_flags;
unsigned long p_offset;
unsigned long p_vaddr;
unsigned long p_paddr;
unsigned long p_filesz;
unsigned long p_memsz;
unsigned long p_align;
};
union ElfEhdrBuf {
struct ElfEhdr ehdr;
char buf[8192];
};
union ElfPhdrBuf {
struct ElfPhdr phdr;
char buf[1024];
};
struct PathSearcher {
int literally;
int indirect;
unsigned long namelen;
const char *name;
const char *syspath;
char path[PATHSIZE];
};
struct ApeLoader {
struct PathSearcher ps;
union ElfPhdrBuf phdr;
struct Syslib lib;
char rando[16];
};
static unsigned long StrLen(const char *s) {
unsigned long n = 0;
while (*s++)
++n;
return n;
}
static int StrCmp(const char *l, const char *r) {
unsigned long i = 0;
while (l[i] == r[i] && r[i])
++i;
return (l[i] & 255) - (r[i] & 255);
}
static const char *BaseName(const char *s) {
const char *b = strrchr(s, '/');
return b ? b + 1 : s;
}
static char *GetEnv(char **p, const char *s) {
unsigned long i, j;
if (p) {
for (i = 0; p[i]; ++i) {
for (j = 0;; ++j) {
if (!s[j]) {
if (p[i][j] == '=') {
return p[i] + j + 1;
}
break;
}
if (s[j] != p[i][j]) {
break;
}
}
}
}
return 0;
}
static char *Utoa(char p[21], unsigned long x) {
char t;
unsigned long i, a, b;
i = 0;
do {
p[i++] = x % 10 + '0';
x = x / 10;
} while (x > 0);
p[i] = '\0';
if (i) {
for (a = 0, b = i - 1; a < b; ++a, --b) {
t = p[a];
p[a] = p[b];
p[b] = t;
}
}
return p + i;
}
static char *Itoa(char p[21], long x) {
if (x < 0)
*p++ = '-', x = -(unsigned long)x;
return Utoa(p, x);
}
static void Emit(const char *s) {
write(2, s, StrLen(s));
}
static long Print(int fd, const char *s, ...) {
int c;
unsigned n;
char b[512];
__builtin_va_list va;
__builtin_va_start(va, s);
for (n = 0; s; s = __builtin_va_arg(va, const char *)) {
while ((c = *s++)) {
if (n < sizeof(b)) {
b[n++] = c;
}
}
}
__builtin_va_end(va);
return write(fd, b, n);
}
static int GetIndirectOffset(const char *arg0) {
char *tail = strrchr(arg0, '.');
if (tail && !StrCmp(tail + 1, "ape")) {
return tail - arg0;
}
return 0;
}
static void Perror(const char *thing, long rc, const char *reason) {
char ibuf[21];
ibuf[0] = 0;
if (rc)
Itoa(ibuf, -rc);
Print(2, "ape error: ", thing, ": ", reason, rc ? " failed w/ errno " : "",
ibuf, "\n", 0l);
}
__attribute__((__noreturn__)) static void Pexit(const char *c, int failed,
const char *s) {
Perror(c, failed, s);
_exit(127);
}
static char AccessCommand(struct PathSearcher *ps, unsigned long pathlen) {
if (pathlen + 1 + ps->namelen + 1 > sizeof(ps->path)) {
return 0;
}
if (pathlen && ps->path[pathlen - 1] != '/')
ps->path[pathlen++] = '/';
memmove(ps->path + pathlen, ps->name, ps->namelen);
ps->path[pathlen + ps->namelen] = 0;
return !access(ps->path, X_OK);
}
static char SearchPath(struct PathSearcher *ps) {
const char *p;
unsigned long i;
for (p = ps->syspath;;) {
for (i = 0; p[i] && p[i] != ':'; ++i) {
if (i < sizeof(ps->path)) {
ps->path[i] = p[i];
}
}
if (AccessCommand(ps, i)) {
return 1;
} else if (p[i] == ':') {
p += i + 1;
} else {
return 0;
}
}
}
static char FindCommand(struct PathSearcher *ps) {
ps->path[0] = 0;
/* paths are always 100% taken literally when a slash exists
$ ape foo/bar.com arg1 arg2
we don't run files in the current directory
$ ape foo.com arg1 arg2
unless $PATH has an empty string entry, e.g.
$ expert PATH=":/bin"
$ ape foo.com arg1 arg2
however we will execute this
$ ape - foo.com foo.com arg1 arg2
because cosmo's execve needs it */
if (ps->literally || memchr(ps->name, '/', ps->namelen)) {
return AccessCommand(ps, 0);
}
/* otherwise search for name on $PATH */
return SearchPath(ps);
}
static char *Commandv(struct PathSearcher *ps, const char *name,
const char *syspath) {
ps->syspath = syspath ? syspath : "/bin:/usr/local/bin:/usr/bin";
ps->name = name;
if (!(ps->namelen = ps->indirect ? ps->indirect : StrLen(ps->name)))
return 0;
if (ps->namelen + 1 > sizeof(ps->path))
return 0;
if (FindCommand(ps)) {
return ps->path;
} else {
return 0;
}
}
static void pthread_jit_write_protect_np_workaround(int enabled) {
int count_start = 8192;
volatile int count = count_start;
unsigned long *addr, val, val2, reread = -1;
addr = (unsigned long *)(!enabled ? _COMM_PAGE_APRR_WRITE_ENABLE
: _COMM_PAGE_APRR_WRITE_DISABLE);
switch (*(volatile unsigned char *)_COMM_PAGE_APRR_SUPPORT) {
case 1:
do {
val = *addr;
reread = -1;
__asm__ volatile("msr\tS3_4_c15_c2_7,%0\n"
"isb\tsy\n"
: /* no outputs */
: "r"(val)
: "memory");
val2 = *addr;
__asm__ volatile("mrs\t%0,S3_4_c15_c2_7\n"
: "=r"(reread)
: /* no inputs */
: "memory");
if (val2 == reread) {
return;
}
usleep(10);
} while (count-- > 0);
break;
case 3:
do {
val = *addr;
reread = -1;
__asm__ volatile("msr\tS3_6_c15_c1_5,%0\n"
"isb\tsy\n"
: /* no outputs */
: "r"(val)
: "memory");
val2 = *addr;
__asm__ volatile("mrs\t%0,S3_6_c15_c1_5\n"
: "=r"(reread)
: /* no inputs */
: "memory");
if (val2 == reread) {
return;
}
usleep(10);
} while (count-- > 0);
break;
default:
pthread_jit_write_protect_np(enabled);
return;
}
Pexit("ape", 0, "failed to set jit write protection");
}
__attribute__((__noinline__)) static long sysret(long rc) {
return rc == -1 ? -errno : rc;
}
static long sys_fork(void) {
return sysret(fork());
}
static long sys_close(int fd) {
return sysret(close(fd));
}
static long sys_raise(int sig) {
return sysret(raise(sig));
}
static long sys_pipe(int pfds[2]) {
return sysret(pipe(pfds));
}
static long sys_munmap(void *addr, size_t size) {
return sysret(munmap(addr, size));
}
static long sys_read(int fd, void *data, size_t size) {
return sysret(read(fd, data, size));
}
static long sys_mprotect(void *data, size_t size, int prot) {
return sysret(mprotect(data, size, prot));
}
static long sys_sigaltstack(const stack_t *ss, stack_t *oss) {
return sysret(sigaltstack(ss, oss));
}
static long sys_getentropy(void *buf, size_t buflen) {
return sysret(getentropy(buf, buflen));
}
static long sys_sem_open(const char *name, int oflags, mode_t mode,
unsigned value) {
return sysret((long)sem_open(name, oflags, mode, value));
}
static long sys_sem_unlink(const char *name) {
return sysret(sem_unlink(name));
}
static long sys_sem_close(sem_t *sem) {
return sysret(sem_close(sem));
}
static long sys_sem_post(sem_t *sem) {
return sysret(sem_post(sem));
}
static long sys_sem_wait(sem_t *sem) {
return sysret(sem_wait(sem));
}
static long sys_sem_trywait(sem_t *sem) {
return sysret(sem_trywait(sem));
}
static long sys_getrlimit(int which, struct rlimit *rlim) {
return sysret(getrlimit(which, rlim));
}
static long sys_setrlimit(int which, const struct rlimit *rlim) {
return sysret(setrlimit(which, rlim));
}
static long sys_write(int fd, const void *data, size_t size) {
return sysret(write(fd, data, size));
}
static long sys_clock_gettime(int clock, struct timespec *ts) {
return sysret(clock_gettime((clockid_t)clock, ts));
}
static long sys_openat(int fd, const char *path, int flags, int mode) {
return sysret(openat(fd, path, flags, mode));
}
static long sys_nanosleep(const struct timespec *req, struct timespec *rem) {
return sysret(nanosleep(req, rem));
}
static long sys_mmap(void *addr, size_t size, int prot, int flags, int fd,
off_t off) {
return sysret((long)mmap(addr, size, prot, flags, fd, off));
}
static long sys_sigaction(int sig, const struct sigaction *act,
struct sigaction *oact) {
return sysret(sigaction(sig, act, oact));
}
static long sys_pselect(int nfds, fd_set *readfds, fd_set *writefds,
fd_set *errorfds, const struct timespec *timeout,
const sigset_t *sigmask) {
return sysret(pselect(nfds, readfds, writefds, errorfds, timeout, sigmask));
}
static long sys_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
void *newp, size_t newlen) {
return sysret(sysctl(name, namelen, oldp, oldlenp, newp, newlen));
}
static long sys_sysctlbyname(const char *name, void *oldp, size_t *oldlenp,
void *newp, size_t newlen) {
return sysret(sysctlbyname(name, oldp, oldlenp, newp, newlen));
}
static long sys_sysctlnametomib(const char *name, int *mibp, size_t *sizep) {
return sysret(sysctlnametomib(name, mibp, sizep));
}
__attribute__((__noreturn__)) static void Spawn(const char *exe, int fd,
long *sp, struct ElfEhdr *e,
struct ElfPhdr *p,
struct Syslib *lib,
char *path) {
long rc;
int prot;
int flags;
int found_entry;
unsigned long dynbase;
unsigned long virtmin, virtmax;
unsigned long a, b, c, d, i, j;
/* validate elf */
found_entry = 0;
virtmin = virtmax = 0;
for (i = 0; i < e->e_phnum; ++i) {
if (p[i].p_type != PT_LOAD) {
continue;
}
if (p[i].p_filesz > p[i].p_memsz) {
Pexit(exe, 0, "ELF p_filesz exceeds p_memsz");
}
if ((p[i].p_flags & (PF_W | PF_X)) == (PF_W | PF_X)) {
Pexit(exe, 0, "Apple Silicon doesn't allow RWX memory");
}
if ((p[i].p_vaddr & (pagesz - 1)) != (p[i].p_offset & (pagesz - 1))) {
Pexit(exe, 0, "ELF p_vaddr incongruent w/ p_offset modulo 16384");
}
if (p[i].p_vaddr + p[i].p_memsz < p[i].p_vaddr ||
p[i].p_vaddr + p[i].p_memsz + (pagesz - 1) < p[i].p_vaddr) {
Pexit(exe, 0, "ELF p_vaddr + p_memsz overflow");
}
if (p[i].p_offset + p[i].p_filesz < p[i].p_offset ||
p[i].p_offset + p[i].p_filesz + (pagesz - 1) < p[i].p_offset) {
Pexit(exe, 0, "ELF p_offset + p_filesz overflow");
}
a = p[i].p_vaddr & -pagesz;
b = (p[i].p_vaddr + p[i].p_memsz + (pagesz - 1)) & -pagesz;
for (j = i + 1; j < e->e_phnum; ++j) {
if (p[j].p_type != PT_LOAD)
continue;
c = p[j].p_vaddr & -pagesz;
d = (p[j].p_vaddr + p[j].p_memsz + (pagesz - 1)) & -pagesz;
if (Max(a, c) < Min(b, d)) {
Pexit(exe, 0, "ELF segments overlap each others virtual memory");
}
}
if (p[i].p_flags & PF_X) {
if (p[i].p_vaddr <= e->e_entry &&
e->e_entry < p[i].p_vaddr + p[i].p_memsz) {
found_entry = 1;
}
}
if (p[i].p_vaddr < virtmin) {
virtmin = p[i].p_vaddr;
}
if (p[i].p_vaddr + p[i].p_memsz > virtmax) {
virtmax = p[i].p_vaddr + p[i].p_memsz;
}
}
if (!found_entry) {
Pexit(exe, 0, "ELF entrypoint not found in PT_LOAD with PF_X");
}
/* choose loading address for dynamic elf executables
that maintains relative distances between segments */
if (e->e_type == ET_DYN) {
rc = sys_mmap(0, virtmax - virtmin, PROT_NONE, MAP_PRIVATE | MAP_ANONYMOUS,
-1, 0);
if (rc < 0)
Pexit(exe, rc, "pie mmap");
dynbase = rc;
if (dynbase & (pagesz - 1)) {
Pexit(exe, 0, "OS mmap incongruent w/ AT_PAGESZ");
}
if (dynbase + virtmin < dynbase) {
Pexit(exe, 0, "ELF dynamic base overflow");
}
} else {
dynbase = 0;
}
/* load elf */
for (i = 0; i < e->e_phnum; ++i) {
void *addr;
unsigned long size;
if (p[i].p_type != PT_LOAD)
continue;
/* configure mapping */
prot = 0;
flags = MAP_FIXED | MAP_PRIVATE;
if (p[i].p_flags & PF_R)
prot |= PROT_READ;
if (p[i].p_flags & PF_W)
prot |= PROT_WRITE;
if (p[i].p_flags & PF_X)
prot |= PROT_EXEC;
/* load from file */
if (p[i].p_filesz) {
int prot1, prot2;
unsigned long wipe;
prot1 = prot;
prot2 = prot;
/* when we ask the system to map the interval [vaddr,vaddr+filesz)
it might schlep extra file content into memory on both the left
and the righthand side. that's because elf doesn't require that
either side of the interval be aligned on the system page size.
normally we can get away with ignoring these junk bytes. but if
the segment defines bss memory (i.e. memsz > filesz) then we'll
need to clear the extra bytes in the page, if they exist. since
we can't do that if we're mapping a read-only page, we can just
map it with write permissions and call mprotect on it afterward */
a = p[i].p_vaddr + p[i].p_filesz; /* end of file content */
b = (a + (pagesz - 1)) & -pagesz; /* first pure bss page */
c = p[i].p_vaddr + p[i].p_memsz; /* end of segment data */
wipe = Min(b - a, c - a);
if (wipe && (~prot1 & PROT_WRITE)) {
prot1 = PROT_READ | PROT_WRITE;
}
addr = (void *)(dynbase + (p[i].p_vaddr & -pagesz));
size = (p[i].p_vaddr & (pagesz - 1)) + p[i].p_filesz;
if (prot1 & PROT_EXEC) {
#ifdef SIP_DISABLED
/* if sip is disabled then we can load the executable segments
off the binary into memory without needing to copy anything
which provides considerably better performance for building */
rc = sys_mmap(addr, size, prot1, flags, fd, p[i].p_offset & -pagesz);
if (rc < 0) {
if (rc == -EPERM) {
Pexit(exe, rc,
"map(exec) failed because SIP (System Integrity Protection) "
"is enabled, but APE Loader was compiled with SIP_DISABLED");
} else {
Pexit(exe, rc, "map(exec) failed");
}
}
#else
/* the issue is that if sip is enabled then, attempting to map
it with exec permission will cause xnu to phone home a hash
of the entire file to apple intelligence as a one time cost
which is literally minutes for executables holding big data
since there's no public apple api for detecting sip we read
as the default strategy which is slow but it works for both */
rc = sys_mmap(addr, size, (prot1 = PROT_READ | PROT_WRITE),
MAP_PRIVATE | MAP_FIXED | MAP_ANONYMOUS, -1, 0);
if (rc < 0)
Pexit(exe, rc, "prog mmap anon");
rc = pread(fd, addr, p[i].p_filesz, p[i].p_offset & -pagesz);
if (rc != p[i].p_filesz)
Pexit(exe, -errno, "prog pread");
#endif
} else {
rc = sys_mmap(addr, size, prot1, flags, fd, p[i].p_offset & -pagesz);
if (rc < 0)
Pexit(exe, rc, "prog mmap");
}
if (wipe)
memset((void *)(dynbase + a), 0, wipe);
if (prot2 != prot1) {
rc = sys_mprotect(addr, size, prot2);
if (rc < 0)
Pexit(exe, rc, "prog mprotect");
}
/* allocate extra bss */
if (c > b) {
flags |= MAP_ANONYMOUS;
rc = sys_mmap((void *)(dynbase + b), c - b, prot, flags, -1, 0);
if (rc < 0)
Pexit(exe, rc, "extra bss mmap");
}
} else {
/* allocate pure bss */
addr = (void *)(dynbase + (p[i].p_vaddr & -pagesz));
size = (p[i].p_vaddr & (pagesz - 1)) + p[i].p_memsz;
flags |= MAP_ANONYMOUS;
rc = sys_mmap(addr, size, prot, flags, -1, 0);
if (rc < 0)
Pexit(exe, rc, "bss mmap");
}
}
/* finish up */
close(fd);
register long *x0 __asm__("x0") = sp;
register char *x2 __asm__("x2") = path;
register int x3 __asm__("x3") = 8; /* _HOSTXNU */
register struct Syslib *x15 __asm__("x15") = lib;
register long x16 __asm__("x16") = e->e_entry;
__asm__ volatile("mov\tx1,#0\n\t"
"mov\tx4,#0\n\t"
"mov\tx5,#0\n\t"
"mov\tx6,#0\n\t"
"mov\tx7,#0\n\t"
"mov\tx8,#0\n\t"
"mov\tx9,#0\n\t"
"mov\tx10,#0\n\t"
"mov\tx11,#0\n\t"
"mov\tx12,#0\n\t"
"mov\tx13,#0\n\t"
"mov\tx14,#0\n\t"
"mov\tx17,#0\n\t"
"mov\tx19,#0\n\t"
"mov\tx20,#0\n\t"
"mov\tx21,#0\n\t"
"mov\tx22,#0\n\t"
"mov\tx23,#0\n\t"
"mov\tx24,#0\n\t"
"mov\tx25,#0\n\t"
"mov\tx26,#0\n\t"
"mov\tx27,#0\n\t"
"mov\tx28,#0\n\t"
"mov\tx29,#0\n\t"
"mov\tx30,#0\n\t"
"mov\tsp,x0\n\t"
"mov\tx0,#0\n\t"
"br\tx16"
: /* no outputs */
: "r"(x0), "r"(x2), "r"(x3), "r"(x15), "r"(x16)
: "memory");
__builtin_unreachable();
}
static const char *TryElf(struct ApeLoader *M, union ElfEhdrBuf *ebuf,
const char *exe, int fd, long *sp, long *auxv,
char *execfn) {
long i, rc;
unsigned size;
struct ElfEhdr *e;
struct ElfPhdr *p;
/* validate elf header */
e = &ebuf->ehdr;
if (READ32(ebuf->buf) != READ32("\177ELF")) {
return "didn't embed ELF magic";
}
if (e->e_ident[EI_CLASS] == ELFCLASS32) {
return "32-bit ELF isn't supported";
}
if (e->e_type != ET_EXEC && e->e_type != ET_DYN) {
return "ELF not ET_EXEC or ET_DYN";
}
if (e->e_machine != EM_AARCH64) {
return "couldn't find ELF header with ARM64 machine type";
}
if (e->e_phentsize != sizeof(struct ElfPhdr)) {
Pexit(exe, 0, "e_phentsize is wrong");
}
size = e->e_phnum;
if ((size *= sizeof(struct ElfPhdr)) > sizeof(M->phdr.buf)) {
Pexit(exe, 0, "too many ELF program headers");
}
/* read program headers */
rc = pread(fd, M->phdr.buf, size, ebuf->ehdr.e_phoff);
if (rc < 0)
return "failed to read ELF program headers";
if (rc != size)
return "truncated read of ELF program headers";
/* bail on recoverable program header errors */
p = &M->phdr.phdr;
for (i = 0; i < e->e_phnum; ++i) {
if (p[i].p_type == PT_INTERP) {
return "ELF has PT_INTERP which isn't supported";
}
if (p[i].p_type == PT_DYNAMIC) {
return "ELF has PT_DYNAMIC which isn't supported";
}
}
/* remove empty program headers */
for (i = 0; i < e->e_phnum;) {
if (p[i].p_type == PT_LOAD && !p[i].p_memsz) {
if (i + 1 < e->e_phnum) {
memmove(p + i, p + i + 1,
(e->e_phnum - (i + 1)) * sizeof(struct ElfPhdr));
}
--e->e_phnum;
} else {
++i;
}
}
/* merge adjacent loads that are contiguous with equal protection,
which prevents our program header overlap check from needlessly
failing later on; it also shaves away a microsecond of latency,
since every program header requires invoking at least 1 syscall */
for (i = 0; i + 1 < e->e_phnum;) {
if (p[i].p_type == PT_LOAD && p[i + 1].p_type == PT_LOAD &&
((p[i].p_flags & (PF_R | PF_W | PF_X)) ==
(p[i + 1].p_flags & (PF_R | PF_W | PF_X))) &&
((p[i].p_offset + p[i].p_filesz + (pagesz - 1)) & -pagesz) -
(p[i + 1].p_offset & -pagesz) <=
pagesz &&
((p[i].p_vaddr + p[i].p_memsz + (pagesz - 1)) & -pagesz) -
(p[i + 1].p_vaddr & -pagesz) <=
pagesz) {
p[i].p_memsz = (p[i + 1].p_vaddr + p[i + 1].p_memsz) - p[i].p_vaddr;
p[i].p_filesz = (p[i + 1].p_offset + p[i + 1].p_filesz) - p[i].p_offset;
if (i + 2 < e->e_phnum) {
memmove(p + i + 1, p + i + 2,
(e->e_phnum - (i + 2)) * sizeof(struct ElfPhdr));
}
--e->e_phnum;
} else {
++i;
}
}
/* simulate linux auxiliary values */
auxv[0] = AT_PHDR;
auxv[1] = (long)&M->phdr.phdr;
auxv[2] = AT_PHENT;
auxv[3] = ebuf->ehdr.e_phentsize;
auxv[4] = AT_PHNUM;
auxv[5] = ebuf->ehdr.e_phnum;
auxv[6] = AT_ENTRY;
auxv[7] = ebuf->ehdr.e_entry;
auxv[8] = AT_PAGESZ;
auxv[9] = pagesz;
auxv[10] = AT_FLAGS;
auxv[11] = M->ps.literally ? AT_FLAGS_PRESERVE_ARGV0 : 0;
auxv[12] = AT_UID;
auxv[13] = getuid();
auxv[14] = AT_EUID;
auxv[15] = geteuid();
auxv[16] = AT_GID;
auxv[17] = getgid();
auxv[18] = AT_EGID;
auxv[19] = getegid();
auxv[20] = AT_HWCAP;
auxv[21] = 0xffb3ffffu;
auxv[22] = AT_HWCAP2;
auxv[23] = 0x181;
auxv[24] = AT_SECURE;
auxv[25] = issetugid();
auxv[26] = AT_RANDOM;
auxv[27] = (long)M->rando;
auxv[28] = AT_EXECFN;
auxv[29] = (long)execfn;
auxv[30] = 0;
/* we're now ready to load */
Spawn(exe, fd, sp, e, p, &M->lib, M->ps.path);
}
int main(int argc, char **argv, char **envp) {
unsigned i;
int c, n, fd, rc;
struct ApeLoader *M;
long *sp, *sp2, *auxv;
union ElfEhdrBuf *ebuf;
char *p, *pe, *exe, *prog, *execfn;
/* allocate loader memory in program's arg block */
n = sizeof(struct ApeLoader);
M = (struct ApeLoader *)__builtin_alloca(n);
/* expose apple libs */
M->lib.magic = SYSLIB_MAGIC;
M->lib.version = SYSLIB_VERSION;
M->lib.fork = sys_fork;
M->lib.pipe = sys_pipe;
M->lib.clock_gettime = sys_clock_gettime;
M->lib.nanosleep = sys_nanosleep;
M->lib.mmap = sys_mmap;
M->lib.pthread_jit_write_protect_supported_np =
pthread_jit_write_protect_supported_np;
M->lib.pthread_jit_write_protect_np = pthread_jit_write_protect_np_workaround;
M->lib.sys_icache_invalidate = sys_icache_invalidate;
M->lib.pthread_create = pthread_create;
M->lib.pthread_exit = pthread_exit;
M->lib.pthread_kill = pthread_kill;
M->lib.pthread_sigmask = pthread_sigmask;
M->lib.pthread_setname_np = pthread_setname_np;
M->lib.dispatch_semaphore_create = dispatch_semaphore_create;
M->lib.dispatch_semaphore_signal = dispatch_semaphore_signal;
M->lib.dispatch_semaphore_wait = dispatch_semaphore_wait;
M->lib.dispatch_walltime = dispatch_walltime;
M->lib.pthread_self = pthread_self;
M->lib.dispatch_release = dispatch_release;
M->lib.raise = sys_raise;
M->lib.pthread_join = pthread_join;
M->lib.pthread_yield_np = pthread_yield_np;
M->lib.pthread_stack_min = PTHREAD_STACK_MIN;
M->lib.sizeof_pthread_attr_t = sizeof(pthread_attr_t);
M->lib.pthread_attr_init = pthread_attr_init;
M->lib.pthread_attr_destroy = pthread_attr_destroy;
M->lib.pthread_attr_setstacksize = pthread_attr_setstacksize;
M->lib.pthread_attr_setguardsize = pthread_attr_setguardsize;
M->lib.exit = exit;
M->lib.close = sys_close;
M->lib.munmap = sys_munmap;
M->lib.openat = sys_openat;
M->lib.write = sys_write;
M->lib.read = sys_read;
M->lib.sigaction = sys_sigaction;
M->lib.pselect = sys_pselect;
M->lib.mprotect = sys_mprotect;
M->lib.sigaltstack = sys_sigaltstack;
M->lib.getentropy = sys_getentropy;
M->lib.sem_open = sys_sem_open;
M->lib.sem_unlink = sys_sem_unlink;
M->lib.sem_close = sys_sem_close;
M->lib.sem_post = sys_sem_post;
M->lib.sem_wait = sys_sem_wait;
M->lib.sem_trywait = sys_sem_trywait;
M->lib.getrlimit = sys_getrlimit;
M->lib.setrlimit = sys_setrlimit;
M->lib.dlopen = dlopen;
M->lib.dlsym = dlsym;
M->lib.dlclose = dlclose;
M->lib.dlerror = dlerror;
M->lib.pthread_cpu_number_np = pthread_cpu_number_np;
M->lib.sysctl = sys_sysctl;
M->lib.sysctlbyname = sys_sysctlbyname;
M->lib.sysctlnametomib = sys_sysctlnametomib;
/* getenv("_") is close enough to at_execfn */
execfn = 0;
for (i = 0; envp[i]; ++i) {
if (envp[i][0] == '_' && envp[i][1] == '=') {
execfn = envp[i] + 2;
}
}
prog = GetEnv(envp + i + 1, "executable_path");
if (!execfn) {
execfn = prog;
}
/* sneak the system five abi back out of args */
sp = (long *)(argv - 1);
auxv = (long *)(envp + i + 1);
/* create new bottom of stack for spawned program
system v abi aligns this on a 16-byte boundary
grows down the alloc by poking the guard pages */
n = (auxv - sp + AUXV_WORDS + 1) * sizeof(long);
sp2 = (long *)__builtin_alloca(n);
if ((long)sp2 & 15)
++sp2;
for (; n > 0; n -= pagesz) {
((char *)sp2)[n - 1] = 0;
}
memmove(sp2, sp, (auxv - sp) * sizeof(long));
argv = (char **)(sp2 + 1);
envp = (char **)(sp2 + 1 + argc + 1);
auxv = (long *)(envp + i + 1);
sp = sp2;
/* interpret command line arguments */
if ((M->ps.indirect = GetIndirectOffset(prog))) {
/* if called as $prog.ape, then strip off the .ape and run the
$prog. This allows you to use symlinks to trick the OS when
a native executable is required. For example, let's say you
want to use the APE binary /opt/cosmos/bin/bash as a system
shell in /etc/shells, or perhaps in shebang lines like e.g.
`#!/opt/cosmos/bin/bash`. That won't work with APE normally,
but it will if you say:
ln -sf /usr/local/bin/ape /opt/cosmos/bin/bash.ape
and then use #!/opt/cosmos/bin/bash.ape instead. */
M->ps.literally = 1;
argc = sp[0];
argv = (char **)(sp + 1);
} else if ((M->ps.literally = argc >= 3 && !StrCmp(argv[1], "-"))) {
/* if the first argument is a hyphen then we give the user the
power to change argv[0] or omit it entirely. most operating
systems don't permit the omission of argv[0] but we do, b/c
it's specified by ANSI X3.159-1988. */
prog = (char *)sp[3];
argc = sp[3] = sp[0] - 3;
argv = (char **)((sp += 3) + 1);
} else if (argc < 2) {
Emit("usage: ape PROG [ARGV1,ARGV2,...]\n"
" ape - PROG [ARGV0,ARGV1,...]\n"
" PROG.ape [ARGV1,ARGV2,...]\n"
"actually portable executable loader silicon 1.10\n"
"copyrights 2023 justine alexandra roberts tunney\n"
"https://justine.lol/ape.html\n");
_exit(1);
} else {
prog = (char *)sp[2];
argc = sp[1] = sp[0] - 1;
argv = (char **)((sp += 1) + 1);
}
/* allocate ephemeral memory for reading file */
n = sizeof(union ElfEhdrBuf);
ebuf = (union ElfEhdrBuf *)__builtin_alloca(n);
for (; n > 0; n -= pagesz) {
((char *)ebuf)[n - 1] = 0;
}
/* search for executable */
if (!(exe = Commandv(&M->ps, prog, GetEnv(envp, "PATH")))) {
Pexit(prog, 0, "not found (maybe chmod +x)");
} else if ((fd = sys_openat(AT_FDCWD, exe, O_RDONLY, 0)) < 0) {
Pexit(exe, fd, "open");
} else if ((rc = sys_read(fd, ebuf->buf, sizeof(ebuf->buf))) < 0) {
Pexit(exe, rc, "read");
} else if ((unsigned long)rc < sizeof(ebuf->ehdr)) {
Pexit(exe, 0, "too small");
}
pe = ebuf->buf + rc;
/* generate some hard random data */
if ((rc = sys_getentropy(M->rando, sizeof(M->rando))) < 0) {
Pexit(argv[0], rc, "getentropy");
}
/* ape intended behavior
1. if ape, will scan shell script for elf printf statements
2. shell script may have multiple lines producing elf headers
3. all elf printf lines must exist in the first 8192 bytes of file
4. elf program headers may appear anywhere in the binary */
if (READ64(ebuf->buf) == READ64("MZqFpD='") ||
READ64(ebuf->buf) == READ64("jartsr='") ||
READ64(ebuf->buf) == READ64("APEDBG='")) {
for (p = ebuf->buf; p < pe; ++p) {
if (READ64(p) != READ64("printf '")) {
continue;
}
for (i = 0, p += 8; p + 3 < pe && (c = *p++) != '\'';) {
if (c == '\\') {
if ('0' <= *p && *p <= '7') {
c = *p++ - '0';
if ('0' <= *p && *p <= '7') {
c *= 8;
c += *p++ - '0';
if ('0' <= *p && *p <= '7') {
c *= 8;
c += *p++ - '0';
}
}
}
}
ebuf->buf[i++] = c;
if (i >= sizeof(ebuf->buf)) {
break;
}
}
if (i >= sizeof(ebuf->ehdr)) {
TryElf(M, ebuf, exe, fd, sp, auxv, execfn);
}
}
}
Pexit(exe, 0, TryElf(M, ebuf, exe, fd, sp, auxv, execfn));
}