/* seitan - Syscall Expressive Interpreter, Transformer and Notifier * * operations.c - Execution of bytecode operations * * Copyright 2023 Red Hat GmbH * Authors: Alice Frosi * Stefano Brivio */ #define _GNU_SOURCE #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "gluten.h" #include "operations.h" static bool is_cookie_valid(int notifyFd, uint64_t id) { return ioctl(notifyFd, SECCOMP_IOCTL_NOTIF_ID_VALID, &id) == 0; } static int send_target(const struct seccomp_notif_resp *resp, int notifyfd) { if (!is_cookie_valid(notifyfd, resp->id)) { fprintf(stderr, "the response id isn't valid\ncheck if the targets has already terminated\n"); return -1; } if (ioctl(notifyfd, SECCOMP_IOCTL_NOTIF_SEND, resp) < 0) { if (errno != EINPROGRESS) { perror("sending the response"); return -1; } } return 0; } static int send_inject_target(const struct seccomp_notif_addfd *resp, int notifyfd) { if (!is_cookie_valid(notifyfd, resp->id)) { fprintf(stderr, "the response id isn't valid\ncheck if the targets has already terminated\n"); return -1; } if (ioctl(notifyfd, SECCOMP_IOCTL_NOTIF_ADDFD, resp) < 0) { if (errno != EINPROGRESS) { perror("sending the response"); return -1; } } return 0; } static void proc_ns_name(unsigned i, char *ns) { switch (i) { case NS_CGROUP: snprintf(ns, PATH_MAX + 1, "cgroup"); break; case NS_IPC: snprintf(ns, PATH_MAX + 1, "ipc"); break; case NS_NET: snprintf(ns, PATH_MAX + 1, "net"); break; case NS_MOUNT: snprintf(ns, PATH_MAX + 1, "mnt"); break; case NS_PID: snprintf(ns, PATH_MAX + 1, "pid"); break; case NS_USER: snprintf(ns, PATH_MAX + 1, "user"); break; case NS_UTS: snprintf(ns, PATH_MAX + 1, "uts"); break; case NS_TIME: snprintf(ns, PATH_MAX + 1, "time"); break; default: fprintf(stderr, "unrecognized namespace index %d\n", i); } } static int set_namespaces(const struct op_call *a, int tpid) { char path[PATH_MAX + 1]; char ns_name[PATH_MAX / 2]; struct ns_spec ns; int fd; unsigned int i; for (i = 0, ns = (a->context).ns[i]; i < sizeof(enum ns_type); i++, ns = (a->context).ns[i]) { proc_ns_name(i, ns_name); switch (ns.type) { case NS_NONE: continue; case NS_SPEC_TARGET: snprintf(path, sizeof(path), "/proc/%d/ns/%s", tpid, ns_name); break; case NS_SPEC_PID: snprintf(path, sizeof(path), "/proc/%d/ns/%s", ns.id.pid, ns_name); break; case NS_SPEC_PATH: snprintf(path, sizeof(path), "%s", ns.id.path); break; } if ((fd = open(path, O_CLOEXEC)) < 0) { fprintf(stderr, "open for file %s: %s", path, strerror(errno)); return -1; } if (setns(fd, 0) != 0) { perror("setns"); return -1; } } return 0; } static int execute_syscall(void *args) { struct arg_clone *a = (struct arg_clone *)args; const struct op_call *c = a->args; if (set_namespaces(a->args, a->pid) < 0) { exit(EXIT_FAILURE); } /* execute syscall */ a->ret = syscall(c->nr, c->args[0], c->args[1], c->args[2], c->args[3], c->args[4], c->args[5]); a->err = errno; if (a->ret < 0) { perror("syscall"); exit(EXIT_FAILURE); } exit(0); } static int op_load(struct seccomp_notif *req, int notifier, struct gluten *g, struct op_load *load) { char path[PATH_MAX]; int fd, ret = 0; snprintf(path, sizeof(path), "/proc/%d/mem", req->pid); if ((fd = open(path, O_RDONLY | O_CLOEXEC)) < 0) { perror("open mem"); return -1; } /* * Avoid the TOCTOU and check if the read mappings are still valid */ if (!is_cookie_valid(notifier, req->id)) { fprintf(stderr, "the seccomp request isn't valid anymore\n"); ret = -1; goto out; } memcpy(gluten_write_ptr(g, load->dst), gluten_ptr(&req->data, g, load->src), load->size); out: close(fd); return ret; } static int resolve_fd(void *data, struct op_resolvedfd *resfd, pid_t pid) { char fdpath[PATH_MAX], buf[PATH_MAX]; char *path = (char *)((uint16_t *)data + resfd->path_off); int *fd = (int *)((uint16_t *)data + resfd->fd_off); ssize_t nbytes; snprintf(fdpath, PATH_MAX, "/proc/%d/fd/%d", pid, *fd); if ((nbytes = readlink(fdpath, buf, resfd->path_size)) < 0) { fprintf(stderr, "error reading %s\n", fdpath); perror("readlink"); return -1; } if (strcmp(path, buf) == 0) return 0; else return 1; } int do_call(struct arg_clone *c) { char stack[STACK_SIZE]; pid_t child; c->ret = -1; c->err = 0; /* Create a process that will be moved to the namespace */ child = clone(execute_syscall, stack + sizeof(stack), CLONE_FILES | CLONE_VM | SIGCHLD, (void *)c); if (child == -1) { perror("clone"); return -1; } wait(NULL); return 0; } static void set_inject_fields(uint64_t id, struct gluten *g, const struct op_inject *a, struct seccomp_notif_addfd *resp) { resp->flags = SECCOMP_ADDFD_FLAG_SETFD; resp->id = id; memcpy(&resp->newfd, gluten_ptr(NULL, g, a->new_fd), sizeof(resp->newfd)); memcpy(&resp->srcfd, gluten_ptr(NULL, g, a->new_fd), sizeof(resp->srcfd)); resp->newfd_flags = 0; } static int op_cmp(struct seccomp_notif *req, int notifier, struct gluten *g, struct op_cmp *op) { int res = memcmp(gluten_ptr(&req->data, g, op->x), gluten_ptr(&req->data, g, op->y), op->size); enum op_cmp_type cmp = op->cmp; (void)notifier; if ((res == 0 && (cmp == CMP_EQ || cmp == CMP_LE || cmp == CMP_GE)) || (res < 0 && (cmp == CMP_LT || cmp == CMP_LE)) || (res > 0 && (cmp == CMP_GT || cmp == CMP_GE))) return op->jmp; return -1; } int do_operations(struct gluten *g, struct op *ops, struct seccomp_notif *req, unsigned int n_ops, int notifyfd) { struct seccomp_notif_addfd resp_fd; struct seccomp_notif_resp resp; struct arg_clone c; unsigned int i; struct op *op; int ret; for (i = 0, op = ops; i < n_ops; i++, op++) { switch (op->type) { case OP_CALL: resp.id = req->id; resp.val = 0; resp.flags = 0; c.args = &ops[i].op.call; c.pid = req->pid; if (do_call(&c) == -1) { resp.error = -1; if (send_target(&resp, notifyfd) == -1) return -1; } if (c.err != 0) { resp.error = -1; if (send_target(&resp, notifyfd) == -1) return c.err; } /* * The result of the call needs to be save as * reference */ if (ops[i].op.call.has_ret) { memcpy(gluten_write_ptr(g, op->op.call.ret), &c.ret, sizeof(c.ret)); } break; case OP_BLOCK: resp.id = req->id; resp.val = 0; resp.flags = 0; resp.error = ops[i].op.block.error; if (send_target(&resp, notifyfd) == -1) return -1; break; case OP_RETURN: resp.id = req->id; resp.flags = 0; resp.error = 0; memcpy(&resp.val, gluten_ptr(&req->data, g, op->op.ret.val), sizeof(resp.val)); if (send_target(&resp, notifyfd) == -1) return -1; break; case OP_CONT: resp.id = req->id; resp.flags = SECCOMP_USER_NOTIF_FLAG_CONTINUE; resp.error = 0; resp.val = 0; if (send_target(&resp, notifyfd) == -1) return -1; break; case OP_INJECT_A: set_inject_fields(req->id, g, &ops[i].op.inject, &resp_fd); resp_fd.flags |= SECCOMP_ADDFD_FLAG_SEND; if (send_inject_target(&resp_fd, notifyfd) == -1) return -1; break; case OP_INJECT: set_inject_fields(req->id, g, &ops[i].op.inject, &resp_fd); if (send_inject_target(&resp_fd, notifyfd) == -1) return -1; break; case OP_LOAD: if (op_load(req, notifyfd, g, &op->op.load)) return -1; break; case OP_END: return 0; case OP_CMP: ret = op_cmp(req, notifyfd, g, (struct op_cmp *)op); if (ret == -1) return -1; i = ret; break; case OP_RESOLVEDFD: ret = resolve_fd(g->data, &ops[i].op.resfd, req->pid); if (ret == -1) return -1; else if (ret == 1) i = ops[i].op.resfd.jmp; break; default: fprintf(stderr, "unknown operation %d \n", ops[i].type); } } return 0; }