/* magiskinit.c - Pre-init Magisk support * * This code has to be compiled statically to work properly. * * To unify Magisk support for both legacy "normal" devices and new skip_initramfs devices, * magisk binary compilation is split into two parts - first part only compiles "magisk". * The python build script will load the magisk main binary and compress with lzma2, dumping * the results into "dump.h". The "magisk" binary is embedded into this binary, and will * get extracted to the overlay folder along with init.magisk.rc. * * This tool does all pre-init operations to setup a Magisk environment, which pathces rootfs * on the fly, providing fundamental support such as init, init.rc, and sepolicy patching. * * Magiskinit is also responsible for constructing a proper rootfs on skip_initramfs devices. * On skip_initramfs devices, it will parse kernel cmdline, mount sysfs, parse through * uevent files to make the system (or vendor if available) block device node, then copy * rootfs files from system. * * This tool will be replaced with the real init to continue the boot process, but hardlinks are * preserved as it also provides CLI for sepolicy patching (magiskpolicy) */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "magisk.h" #include "utils.h" #include "magiskpolicy.h" #include "magiskrc.h" #include "dump.h" // #define VLOG(fmt, ...) printf(fmt, __VA_ARGS__) /* Enable to debug */ #define VLOG(fmt, ...) struct cmdline { int skip_initramfs; char slot[3]; }; struct device { dev_t major; dev_t minor; char devname[32]; char partname[32]; char path[64]; }; extern policydb_t *policydb; static void parse_cmdline(struct cmdline *cmd) { char *tok; char buffer[4096]; mkdir("/proc", 0555); mount("proc", "/proc", "proc", 0, NULL); int fd = open("/proc/cmdline", O_RDONLY | O_CLOEXEC); ssize_t size = read(fd, buffer, sizeof(buffer)); buffer[size] = '\0'; close(fd); umount("/proc"); tok = strtok(buffer, " "); cmd->skip_initramfs = 0; cmd->slot[0] = '\0'; while (tok != NULL) { if (strncmp(tok, "androidboot.slot_suffix", 23) == 0) { sscanf(tok, "androidboot.slot_suffix=%s", cmd->slot); } else if (strncmp(tok, "androidboot.slot", 16) == 0) { cmd->slot[0] = '_'; sscanf(tok, "androidboot.slot=%s", cmd->slot + 1); } else if (strcmp(tok, "skip_initramfs") == 0) { cmd->skip_initramfs = 1; } tok = strtok(NULL, " "); } } static void parse_device(struct device *dev, char *uevent) { dev->partname[0] = '\0'; char *tok; tok = strtok(uevent, "\n"); while (tok != NULL) { if (strncmp(tok, "MAJOR", 5) == 0) { sscanf(tok, "MAJOR=%ld", (long*) &dev->major); } else if (strncmp(tok, "MINOR", 5) == 0) { sscanf(tok, "MINOR=%ld", (long*) &dev->minor); } else if (strncmp(tok, "DEVNAME", 7) == 0) { sscanf(tok, "DEVNAME=%s", dev->devname); } else if (strncmp(tok, "PARTNAME", 8) == 0) { sscanf(tok, "PARTNAME=%s", dev->partname); } tok = strtok(NULL, "\n"); } VLOG("%s [%s] (%u, %u)\n", dev->devname, dev->partname, (unsigned) dev->major, (unsigned) dev->minor); } static int setup_block(struct device *dev, const char *partname) { char buffer[1024], path[128]; struct dirent *entry; DIR *dir = opendir("/sys/dev/block"); if (dir == NULL) return 1; int found = 0; while ((entry = readdir(dir))) { if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) continue; snprintf(path, sizeof(path), "/sys/dev/block/%s/uevent", entry->d_name); int fd = open(path, O_RDONLY | O_CLOEXEC); ssize_t size = read(fd, buffer, sizeof(buffer)); buffer[size] = '\0'; close(fd); parse_device(dev, buffer); if (strcmp(dev->partname, partname) == 0) { snprintf(dev->path, sizeof(dev->path), "/dev/block/%s", dev->devname); found = 1; break; } } closedir(dir); if (!found) return 1; mkdir("/dev", 0755); mkdir("/dev/block", 0755); mknod(dev->path, S_IFBLK | 0600, makedev(dev->major, dev->minor)); return 0; } static void *patch_init_rc(char *data, uint32_t *size) { int injected = 0; char *new_data = malloc(*size + 23); char *old_data = data; uint32_t pos = 0; for (char *tok = strsep(&old_data, "\n"); tok; tok = strsep(&old_data, "\n")) { if (!injected && strncmp(tok, "import", 6) == 0) { if (strstr(tok, "init.magisk.rc")) { injected = 1; } else { strcpy(new_data + pos, "import /init.magisk.rc\n"); pos += 23; injected = 1; } } else if (strstr(tok, "selinux.reload_policy")) { continue; } // Copy the line strcpy(new_data + pos, tok); pos += strlen(tok); new_data[pos++] = '\n'; } *size = pos; return new_data; } static void patch_ramdisk() { void *addr; size_t size; mmap_rw("/init", &addr, &size); for (int i = 0; i < size; ++i) { if (memcmp(addr + i, "/system/etc/selinux/plat_sepolicy.cil", 37) == 0) { memcpy(addr + i, "/system/etc/selinux/plat_sepolicy.xxx", 37); break; } } munmap(addr, size); mmap_rw("/init.rc", &addr, &size); uint32_t new_size = size; void *init_rc = patch_init_rc(addr, &new_size); munmap(addr, size); int fd = open("/init.rc", O_WRONLY | O_TRUNC | O_CLOEXEC); write(fd, init_rc, new_size); close(fd); free(init_rc); } static int strend(const char *s1, const char *s2) { int l1 = strlen(s1); int l2 = strlen(s2); return strcmp(s1 + l1 - l2, s2); } static int compile_cil() { DIR *dir; struct dirent *entry; char path[128]; struct cil_db *db = NULL; sepol_policydb_t *pdb = NULL; void *addr; size_t size; cil_db_init(&db); cil_set_mls(db, 1); cil_set_target_platform(db, SEPOL_TARGET_SELINUX); cil_set_policy_version(db, POLICYDB_VERSION_XPERMS_IOCTL); cil_set_attrs_expand_generated(db, 0); // plat mmap_ro("/system/etc/selinux/plat_sepolicy.cil", &addr, &size); cil_add_file(db, "/system/etc/selinux/plat_sepolicy.cil", addr, size); munmap(addr, size); // mapping char plat[10]; int fd = open("/vendor/etc/selinux/plat_sepolicy_vers.txt", O_RDONLY | O_CLOEXEC); if (fd > 0) { plat[read(fd, plat, sizeof(plat)) - 1] = '\0'; snprintf(path, sizeof(path), "/system/etc/selinux/mapping/%s.cil", plat); mmap_ro(path, &addr, &size); cil_add_file(db, path, addr, size); munmap(addr, size); close(fd); } dir = opendir("/vendor/etc/selinux"); while ((entry = readdir(dir))) { if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) continue; if (strend(entry->d_name, ".cil") == 0) { snprintf(path, sizeof(path), "/vendor/etc/selinux/%s", entry->d_name); mmap_ro(path, &addr, &size); cil_add_file(db, path, addr, size); munmap(addr, size); } } closedir(dir); cil_compile(db); cil_build_policydb(db, &pdb); cil_db_destroy(&db); policydb = &pdb->p; return 0; } static int verify_precompiled() { DIR *dir; struct dirent *entry; int fd; char sys_sha[70], ven_sha[70]; dir = opendir("/vendor/etc/selinux"); while ((entry = readdir(dir))) { if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) continue; if (strend(entry->d_name, ".sha256") == 0) { fd = openat(dirfd(dir), entry->d_name, O_RDONLY | O_CLOEXEC); ven_sha[read(fd, ven_sha, sizeof(ven_sha))] = '\0'; close(fd); break; } } closedir(dir); dir = opendir("/system/etc/selinux"); while ((entry = readdir(dir))) { if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) continue; if (strend(entry->d_name, ".sha256") == 0) { fd = openat(dirfd(dir), entry->d_name, O_RDONLY | O_CLOEXEC); sys_sha[read(fd, sys_sha, sizeof(sys_sha))] = '\0'; close(fd); break; } } closedir(dir); return strcmp(sys_sha, ven_sha); } static void patch_sepolicy() { if (access("/sepolicy", R_OK) == 0) { load_policydb("/sepolicy"); } else if (access("/vendor/etc/selinux/precompiled_sepolicy", R_OK) == 0 && verify_precompiled() == 0) { load_policydb("/vendor/etc/selinux/precompiled_sepolicy"); } else if (access("/system/etc/selinux/plat_sepolicy.cil", R_OK) == 0) { compile_cil(); } sepol_med_rules(); dump_policydb("/sepolicy"); } #define BUFSIZE (1 << 20) static int unxz(const void *buf, size_t size, int fd) { lzma_stream strm = LZMA_STREAM_INIT; if (lzma_auto_decoder(&strm, UINT64_MAX, 0) != LZMA_OK) return 1; lzma_ret ret = 0; void *out = malloc(BUFSIZE); strm.next_in = buf; strm.avail_in = size; do { strm.next_out = out; strm.avail_out = BUFSIZE; ret = lzma_code(&strm, LZMA_RUN); write(fd, out, BUFSIZE - strm.avail_out); } while (strm.avail_out == 0 && ret == LZMA_OK); free(out); lzma_end(&strm); if (ret != LZMA_OK && ret != LZMA_STREAM_END) return 1; return 0; } static int dump_magisk(const char *path, mode_t mode) { unlink(path); int fd = creat(path, mode); int ret = unxz(magisk_dump, sizeof(magisk_dump), fd); close(fd); return ret; } static int dump_magiskrc(const char *path, mode_t mode) { int fd = creat(path, mode); write(fd, magiskrc, sizeof(magiskrc)); close(fd); return 0; } int main(int argc, char *argv[]) { if (strcmp(basename(argv[0]), "magiskpolicy") == 0 || strcmp(basename(argv[0]), "supolicy") == 0) return magiskpolicy_main(argc, argv); if (argc > 1 && (strcmp(argv[1], "magiskpolicy") == 0 || strcmp(argv[1], "supolicy") == 0)) return magiskpolicy_main(argc - 1, argv + 1); umask(0); if (argc > 1) { if (strcmp(argv[2], "magisk") == 0) return dump_magisk(argv[3], 0755); else if (strcmp(argv[2], "magiskrc") == 0) return dump_magiskrc(argv[3], 0755); } // Extract and link files mkdir("/overlay", 0000); dump_magiskrc("/overlay/init.magisk.rc", 0750); mkdir("/overlay/sbin", 0755); dump_magisk("/overlay/sbin/magisk", 0755); mkdir("/overlay/root", 0755); link("/init", "/overlay/root/magiskinit"); symlink("/root/magiskinit", "/overlay/root/magiskpolicy"); symlink("/root/magiskinit", "/overlay/root/supolicy"); struct cmdline cmd; parse_cmdline(&cmd); VLOG("cmdline: skip_initramfs=[%d] slot_suffix=[%s]\n", cmd.skip_initramfs, cmd.slot); int root = open("/", O_RDONLY | O_CLOEXEC); if (cmd.skip_initramfs) { // Exclude overlay folder excl_list = (char *[]) { "overlay", ".backup", NULL }; // Clear rootfs frm_rf(root); mkdir("/sys", 0755); mount("sysfs", "/sys", "sysfs", 0, NULL); char partname[32]; snprintf(partname, sizeof(partname), "system%s", cmd.slot); struct device dev; setup_block(&dev, partname); mkdir("/system_root", 0755); mount(dev.path, "/system_root", "ext4", MS_RDONLY, NULL); int system_root = open("/system_root", O_RDONLY | O_CLOEXEC); // Exclude system folder excl_list = (char *[]) { "system", NULL }; clone_dir(system_root, root); mkdir("/system", 0755); mount("/system_root/system", "/system", NULL, MS_BIND, NULL); snprintf(partname, sizeof(partname), "vendor%s", cmd.slot); // We need to mount independent vendor partition if (setup_block(&dev, partname) == 0) mount(dev.path, "/vendor", "ext4", MS_RDONLY, NULL); close(system_root); } else { // Revert original init binary unlink("/init"); link("/.backup/init", "/init"); } int overlay = open("/overlay", O_RDONLY | O_CLOEXEC); mv_dir(overlay, root); patch_ramdisk(); patch_sepolicy(); // Clean up rmdir("/overlay"); umount("/vendor"); close(overlay); close(root); if (fork() == 0) { // Fork a new process for full patch setsid(); sepol_allow("su", ALL, ALL, ALL); while (access(PATCHSTART, W_OK) == -1) usleep(500); /* Wait 0.5ms */ unlink(PATCHSTART); dump_policydb(SELINUX_LOAD); close(open(PATCHDONE, O_RDONLY | O_CREAT, 0)); destroy_policydb(); } else { // Finally, give control back! execv("/init", argv); } return 0; }