Try   HackMD

The clone() syscall

This section aims to explain how the clone() syscall is executed when the systme is under the NUMA architechture.

System Call Handler for clone()

  • If the call is made by the glibc wrapper clone(), the clone syscall handler is used
  • If syscall(SYS_clone3, ...) is used (no wrapper provided by glibc), the clone3() syscall handler is used.

Both path ends in a kernel_clone() call.

#ifdef __ARCH_WANT_SYS_CLONE #ifdef CONFIG_CLONE_BACKWARDS SYSCALL_DEFINE5(clone, unsigned long, clone_flags, unsigned long, newsp, int __user *, parent_tidptr, unsigned long, tls, int __user *, child_tidptr) #elif defined(CONFIG_CLONE_BACKWARDS2) SYSCALL_DEFINE5(clone, unsigned long, newsp, unsigned long, clone_flags, int __user *, parent_tidptr, int __user *, child_tidptr, unsigned long, tls) #elif defined(CONFIG_CLONE_BACKWARDS3) SYSCALL_DEFINE6(clone, unsigned long, clone_flags, unsigned long, newsp, int, stack_size, int __user *, parent_tidptr, int __user *, child_tidptr, unsigned long, tls) #else SYSCALL_DEFINE5(clone, unsigned long, clone_flags, unsigned long, newsp, int __user *, parent_tidptr, int __user *, child_tidptr, unsigned long, tls) #endif { struct kernel_clone_args args = { .flags = (lower_32_bits(clone_flags) & ~CSIGNAL), .pidfd = parent_tidptr, .child_tid = child_tidptr, .parent_tid = parent_tidptr, .exit_signal = (lower_32_bits(clone_flags) & CSIGNAL), .stack = newsp, .tls = tls, }; return kernel_clone(&args); } #endif
#ifdef __ARCH_WANT_SYS_CLONE3 /** * sys_clone3 - create a new process with specific properties * @uargs: argument structure * @size: size of @uargs * * clone3() is the extensible successor to clone()/clone2(). * It takes a struct as argument that is versioned by its size. * * Return: On success, a positive PID for the child process. * On error, a negative errno number. */ SYSCALL_DEFINE2(clone3, struct clone_args __user *, uargs, size_t, size) { int err; struct kernel_clone_args kargs; pid_t set_tid[MAX_PID_NS_LEVEL]; kargs.set_tid = set_tid; err = copy_clone_args_from_user(&kargs, uargs, size); if (err) return err; if (!clone3_args_valid(&kargs)) return -EINVAL; return kernel_clone(&kargs); } #endif

Create a brand new task

After some initial setup, the function copy_process(NULL, trace, NUMA_NO_NODE, args); is called.

/* * Ok, this is the main fork-routine. * * It copies the process, and if successful kick-starts * it and waits for it to finish using the VM if required. * * args->exit_signal is expected to be checked for sanity by the caller. */ pid_t kernel_clone(struct kernel_clone_args *args) { u64 clone_flags = args->flags; struct completion vfork; struct pid *pid; struct task_struct *p; int trace = 0; pid_t nr; /* * For legacy clone() calls, CLONE_PIDFD uses the parent_tid argument * to return the pidfd. Hence, CLONE_PIDFD and CLONE_PARENT_SETTID are * mutually exclusive. With clone3() CLONE_PIDFD has grown a separate * field in struct clone_args and it still doesn't make sense to have * them both point at the same memory location. Performing this check * here has the advantage that we don't need to have a separate helper * to check for legacy clone(). */ if ((clone_flags & CLONE_PIDFD) && (clone_flags & CLONE_PARENT_SETTID) && (args->pidfd == args->parent_tid)) return -EINVAL; /* * Determine whether and which event to report to ptracer. When * called from kernel_thread or CLONE_UNTRACED is explicitly * requested, no event is reported; otherwise, report if the event * for the type of forking is enabled. */ if (!(clone_flags & CLONE_UNTRACED)) { if (clone_flags & CLONE_VFORK) trace = PTRACE_EVENT_VFORK; else if (args->exit_signal != SIGCHLD) trace = PTRACE_EVENT_CLONE; else trace = PTRACE_EVENT_FORK; if (likely(!ptrace_event_enabled(current, trace))) trace = 0; } p = copy_process(NULL, trace, NUMA_NO_NODE, args); add_latent_entropy(); if (IS_ERR(p)) return PTR_ERR(p); /* * Do this prior waking up the new thread - the thread pointer * might get invalid after that point, if the thread exits quickly. */ trace_sched_process_fork(current, p); pid = get_task_pid(p, PIDTYPE_PID); nr = pid_vnr(pid); if (clone_flags & CLONE_PARENT_SETTID) put_user(nr, args->parent_tid); if (clone_flags & CLONE_VFORK) { p->vfork_done = &vfork; init_completion(&vfork); get_task_struct(p); } if (IS_ENABLED(CONFIG_LRU_GEN_WALKS_MMU) && !(clone_flags & CLONE_VM)) { /* lock the task to synchronize with memcg migration */ task_lock(p); lru_gen_add_mm(p->mm); task_unlock(p); } wake_up_new_task(p); /* forking complete and child started to run, tell ptracer */ if (unlikely(trace)) ptrace_event_pid(trace, pid); if (clone_flags & CLONE_VFORK) { if (!wait_for_vfork_done(p, &vfork)) ptrace_event_pid(PTRACE_EVENT_VFORK_DONE, pid); } put_pid(pid); return nr; }

Copying The Process

/* * This creates a new process as a copy of the old one, * but does not actually start it yet. * * It copies the registers, and all the appropriate * parts of the process environment (as per the clone * flags). The actual kick-off is left to the caller. */ __latent_entropy struct task_struct *copy_process( struct pid *pid, int trace, int node, struct kernel_clone_args *args) { int pidfd = -1, retval; struct task_struct *p; struct multiprocess_signals delayed; struct file *pidfile = NULL; const u64 clone_flags = args->flags; struct nsproxy *nsp = current->nsproxy; /* * CLONE_* flags processing */ // ... /* * Force any signals received before this point to be delivered * before the fork happens. Collect up signals sent to multiple * processes that happen during the fork and delay them so that * they appear to happen after the fork. */ // ... retval = -ENOMEM; p = dup_task_struct(current, node); if (!p) goto fork_out; p->flags &= ~PF_KTHREAD; if (args->kthread) p->flags |= PF_KTHREAD; if (args->user_worker) { /* * Mark us a user worker, and block any signal that isn't * fatal or STOP */ p->flags |= PF_USER_WORKER; siginitsetinv(&p->blocked, sigmask(SIGKILL)|sigmask(SIGSTOP)); } if (args->io_thread) p->flags |= PF_IO_WORKER; if (args->name) strscpy_pad(p->comm, args->name, sizeof(p->comm)); p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? args->child_tid : NULL; /* * Clear TID on mm_release()? */ p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? args->child_tid : NULL; ftrace_graph_init_task(p); rt_mutex_init_task(p); lockdep_assert_irqs_enabled(); #ifdef CONFIG_PROVE_LOCKING DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled); #endif retval = copy_creds(p, clone_flags); if (retval < 0) goto bad_fork_free; retval = -EAGAIN; if (is_rlimit_overlimit(task_ucounts(p), UCOUNT_RLIMIT_NPROC, rlimit(RLIMIT_NPROC))) { if (p->real_cred->user != INIT_USER && !capable(CAP_SYS_RESOURCE) && !capable(CAP_SYS_ADMIN)) goto bad_fork_cleanup_count; } current->flags &= ~PF_NPROC_EXCEEDED; /* * If multiple threads are within copy_process(), then this check * triggers too late. This doesn't hurt, the check is only there * to stop root fork bombs. */ retval = -EAGAIN; if (data_race(nr_threads >= max_threads)) goto bad_fork_cleanup_count; delayacct_tsk_init(p); /* Must remain after dup_task_struct() */ p->flags &= ~(PF_SUPERPRIV | PF_WQ_WORKER | PF_IDLE | PF_NO_SETAFFINITY); p->flags |= PF_FORKNOEXEC; INIT_LIST_HEAD(&p->children); INIT_LIST_HEAD(&p->sibling); rcu_copy_process(p); p->vfork_done = NULL; spin_lock_init(&p->alloc_lock); init_sigpending(&p->pending); p->utime = p->stime = p->gtime = 0; #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME p->utimescaled = p->stimescaled = 0; #endif prev_cputime_init(&p->prev_cputime); #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN seqcount_init(&p->vtime.seqcount); p->vtime.starttime = 0; p->vtime.state = VTIME_INACTIVE; #endif #ifdef CONFIG_IO_URING p->io_uring = NULL; #endif p->default_timer_slack_ns = current->timer_slack_ns; #ifdef CONFIG_PSI p->psi_flags = 0; #endif task_io_accounting_init(&p->ioac); acct_clear_integrals(p); posix_cputimers_init(&p->posix_cputimers); p->io_context = NULL; audit_set_context(p, NULL); cgroup_fork(p); if (args->kthread) { if (!set_kthread_struct(p)) goto bad_fork_cleanup_delayacct; } #ifdef CONFIG_NUMA p->mempolicy = mpol_dup(p->mempolicy); if (IS_ERR(p->mempolicy)) { retval = PTR_ERR(p->mempolicy); p->mempolicy = NULL; goto bad_fork_cleanup_delayacct; } #endif #ifdef CONFIG_CPUSETS p->cpuset_mem_spread_rotor = NUMA_NO_NODE; p->cpuset_slab_spread_rotor = NUMA_NO_NODE; seqcount_spinlock_init(&p->mems_allowed_seq, &p->alloc_lock); #endif #ifdef CONFIG_TRACE_IRQFLAGS memset(&p->irqtrace, 0, sizeof(p->irqtrace)); p->irqtrace.hardirq_disable_ip = _THIS_IP_; p->irqtrace.softirq_enable_ip = _THIS_IP_; p->softirqs_enabled = 1; p->softirq_context = 0; #endif p->pagefault_disabled = 0; #ifdef CONFIG_LOCKDEP lockdep_init_task(p); #endif #ifdef CONFIG_DEBUG_MUTEXES p->blocked_on = NULL; /* not blocked yet */ #endif #ifdef CONFIG_BCACHE p->sequential_io = 0; p->sequential_io_avg = 0; #endif #ifdef CONFIG_BPF_SYSCALL RCU_INIT_POINTER(p->bpf_storage, NULL); p->bpf_ctx = NULL; #endif /* Perform scheduler related setup. Assign this task to a CPU. */ retval = sched_fork(clone_flags, p); if (retval) goto bad_fork_cleanup_policy; retval = perf_event_init_task(p, clone_flags); if (retval) goto bad_fork_cleanup_policy; retval = audit_alloc(p); if (retval) goto bad_fork_cleanup_perf; /* copy all the process information */ shm_init_task(p); retval = security_task_alloc(p, clone_flags); if (retval) goto bad_fork_cleanup_audit; retval = copy_semundo(clone_flags, p); if (retval) goto bad_fork_cleanup_security; retval = copy_files(clone_flags, p, args->no_files); if (retval) goto bad_fork_cleanup_semundo; retval = copy_fs(clone_flags, p); if (retval) goto bad_fork_cleanup_files; retval = copy_sighand(clone_flags, p); if (retval) goto bad_fork_cleanup_fs; retval = copy_signal(clone_flags, p); if (retval) goto bad_fork_cleanup_sighand; retval = copy_mm(clone_flags, p); if (retval) goto bad_fork_cleanup_signal; retval = copy_namespaces(clone_flags, p); if (retval) goto bad_fork_cleanup_mm; retval = copy_io(clone_flags, p); if (retval) goto bad_fork_cleanup_namespaces; retval = copy_thread(p, args); if (retval) goto bad_fork_cleanup_io; stackleak_task_init(p); if (pid != &init_struct_pid) { pid = alloc_pid(p->nsproxy->pid_ns_for_children, args->set_tid, args->set_tid_size); if (IS_ERR(pid)) { retval = PTR_ERR(pid); goto bad_fork_cleanup_thread; } } /* * This has to happen after we've potentially unshared the file * descriptor table (so that the pidfd doesn't leak into the child * if the fd table isn't shared). */ if (clone_flags & CLONE_PIDFD) { int flags = (clone_flags & CLONE_THREAD) ? PIDFD_THREAD : 0; /* Note that no task has been attached to @pid yet. */ retval = __pidfd_prepare(pid, flags, &pidfile); if (retval < 0) goto bad_fork_free_pid; pidfd = retval; retval = put_user(pidfd, args->pidfd); if (retval) goto bad_fork_put_pidfd; } #ifdef CONFIG_BLOCK p->plug = NULL; #endif futex_init_task(p); /* * sigaltstack should be cleared when sharing the same VM */ if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM) sas_ss_reset(p); /* * Syscall tracing and stepping should be turned off in the * child regardless of CLONE_PTRACE. */ user_disable_single_step(p); clear_task_syscall_work(p, SYSCALL_TRACE); #if defined(CONFIG_GENERIC_ENTRY) || defined(TIF_SYSCALL_EMU) clear_task_syscall_work(p, SYSCALL_EMU); #endif clear_tsk_latency_tracing(p); /* ok, now we should be set up.. */ p->pid = pid_nr(pid); if (clone_flags & CLONE_THREAD) { p->group_leader = current->group_leader; p->tgid = current->tgid; } else { p->group_leader = p; p->tgid = p->pid; } p->nr_dirtied = 0; p->nr_dirtied_pause = 128 >> (PAGE_SHIFT - 10); p->dirty_paused_when = 0; p->pdeath_signal = 0; p->task_works = NULL; clear_posix_cputimers_work(p); #ifdef CONFIG_KRETPROBES p->kretprobe_instances.first = NULL; #endif #ifdef CONFIG_RETHOOK p->rethooks.first = NULL; #endif /* * Ensure that the cgroup subsystem policies allow the new process to be * forked. It should be noted that the new process's css_set can be changed * between here and cgroup_post_fork() if an organisation operation is in * progress. */ retval = cgroup_can_fork(p, args); if (retval) goto bad_fork_put_pidfd; /* * Now that the cgroups are pinned, re-clone the parent cgroup and put * the new task on the correct runqueue. All this *before* the task * becomes visible. * * This isn't part of ->can_fork() because while the re-cloning is * cgroup specific, it unconditionally needs to place the task on a * runqueue. */ sched_cgroup_fork(p, args); /* * From this point on we must avoid any synchronous user-space * communication until we take the tasklist-lock. In particular, we do * not want user-space to be able to predict the process start-time by * stalling fork(2) after we recorded the start_time but before it is * visible to the system. */ p->start_time = ktime_get_ns(); p->start_boottime = ktime_get_boottime_ns(); /* * Make it visible to the rest of the system, but dont wake it up yet. * Need tasklist lock for parent etc handling! */ write_lock_irq(&tasklist_lock); /* CLONE_PARENT re-uses the old parent */ if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) { p->real_parent = current->real_parent; p->parent_exec_id = current->parent_exec_id; if (clone_flags & CLONE_THREAD) p->exit_signal = -1; else p->exit_signal = current->group_leader->exit_signal; } else { p->real_parent = current; p->parent_exec_id = current->self_exec_id; p->exit_signal = args->exit_signal; } klp_copy_process(p); sched_core_fork(p); spin_lock(&current->sighand->siglock); rv_task_fork(p); rseq_fork(p, clone_flags); /* Don't start children in a dying pid namespace */ if (unlikely(!(ns_of_pid(pid)->pid_allocated & PIDNS_ADDING))) { retval = -ENOMEM; goto bad_fork_cancel_cgroup; } /* Let kill terminate clone/fork in the middle */ if (fatal_signal_pending(current)) { retval = -EINTR; goto bad_fork_cancel_cgroup; } /* No more failure paths after this point. */ /* * Copy seccomp details explicitly here, in case they were changed * before holding sighand lock. */ copy_seccomp(p); init_task_pid_links(p); if (likely(p->pid)) { ptrace_init_task(p, (clone_flags & CLONE_PTRACE) || trace); init_task_pid(p, PIDTYPE_PID, pid); if (thread_group_leader(p)) { init_task_pid(p, PIDTYPE_TGID, pid); init_task_pid(p, PIDTYPE_PGID, task_pgrp(current)); init_task_pid(p, PIDTYPE_SID, task_session(current)); if (is_child_reaper(pid)) { ns_of_pid(pid)->child_reaper = p; p->signal->flags |= SIGNAL_UNKILLABLE; } p->signal->shared_pending.signal = delayed.signal; p->signal->tty = tty_kref_get(current->signal->tty); /* * Inherit has_child_subreaper flag under the same * tasklist_lock with adding child to the process tree * for propagate_has_child_subreaper optimization. */ p->signal->has_child_subreaper = p->real_parent->signal->has_child_subreaper || p->real_parent->signal->is_child_subreaper; list_add_tail(&p->sibling, &p->real_parent->children); list_add_tail_rcu(&p->tasks, &init_task.tasks); attach_pid(p, PIDTYPE_TGID); attach_pid(p, PIDTYPE_PGID); attach_pid(p, PIDTYPE_SID); __this_cpu_inc(process_counts); } else { current->signal->nr_threads++; current->signal->quick_threads++; atomic_inc(&current->signal->live); refcount_inc(&current->signal->sigcnt); task_join_group_stop(p); list_add_tail_rcu(&p->thread_node, &p->signal->thread_head); } attach_pid(p, PIDTYPE_PID); nr_threads++; } total_forks++; hlist_del_init(&delayed.node); spin_unlock(&current->sighand->siglock); syscall_tracepoint_update(p); write_unlock_irq(&tasklist_lock); if (pidfile) fd_install(pidfd, pidfile); proc_fork_connector(p); sched_post_fork(p); cgroup_post_fork(p, args); perf_event_fork(p); trace_task_newtask(p, clone_flags); uprobe_copy_process(p, clone_flags); user_events_fork(p, clone_flags); copy_oom_score_adj(clone_flags, p); return p; /* * Tons of bad fork exiting failure paths, reached via goto's. */ // ... return ERR_PTR(retval); }

The main part. At line 139 of this snippet, the function also sets up the memory policy for the newly forked task, which happens at line 39 of this snippet (dup_task_struct(current, node), with node = NUMA_NO_NODE, and orig pointed to the task issued the system call, current).

dup_task_struct():
static struct task_struct *dup_task_struct(struct task_struct *orig, int node) { struct task_struct *tsk; int err; if (node == NUMA_NO_NODE) node = tsk_fork_get_node(orig); tsk = alloc_task_struct_node(node); if (!tsk) return NULL; err = arch_dup_task_struct(tsk, orig); if (err) goto free_tsk; err = alloc_thread_stack_node(tsk, node); if (err) goto free_tsk; #ifdef CONFIG_THREAD_INFO_IN_TASK refcount_set(&tsk->stack_refcount, 1); #endif account_kernel_stack(tsk, 1); err = scs_prepare(tsk, node); if (err) goto free_stack; #ifdef CONFIG_SECCOMP /* * We must handle setting up seccomp filters once we're under * the sighand lock in case orig has changed between now and * then. Until then, filter must be NULL to avoid messing up * the usage counts on the error path calling free_task. */ tsk->seccomp.filter = NULL; #endif setup_thread_stack(tsk, orig); clear_user_return_notifier(tsk); clear_tsk_need_resched(tsk); set_task_stack_end_magic(tsk); clear_syscall_work_syscall_user_dispatch(tsk); #ifdef CONFIG_STACKPROTECTOR tsk->stack_canary = get_random_canary(); #endif if (orig->cpus_ptr == &orig->cpus_mask) tsk->cpus_ptr = &tsk->cpus_mask; dup_user_cpus_ptr(tsk, orig, node); /* * One for the user space visible state that goes away when reaped. * One for the scheduler. */ refcount_set(&tsk->rcu_users, 2); /* One for the rcu users */ refcount_set(&tsk->usage, 1); #ifdef CONFIG_BLK_DEV_IO_TRACE tsk->btrace_seq = 0; #endif tsk->splice_pipe = NULL; tsk->task_frag.page = NULL; tsk->wake_q.next = NULL; tsk->worker_private = NULL; kcov_task_init(tsk); kmsan_task_create(tsk); kmap_local_fork(tsk); #ifdef CONFIG_FAULT_INJECTION tsk->fail_nth = 0; #endif #ifdef CONFIG_BLK_CGROUP tsk->throttle_disk = NULL; tsk->use_memdelay = 0; #endif #ifdef CONFIG_ARCH_HAS_CPU_PASID tsk->pasid_activated = 0; #endif #ifdef CONFIG_MEMCG tsk->active_memcg = NULL; #endif #ifdef CONFIG_CPU_SUP_INTEL tsk->reported_split_lock = 0; #endif #ifdef CONFIG_SCHED_MM_CID tsk->mm_cid = -1; tsk->last_mm_cid = -1; tsk->mm_cid_active = 0; tsk->migrate_from_cpu = -1; #endif return tsk; free_stack: exit_task_stack_account(tsk); free_thread_stack(tsk); free_tsk: free_task_struct(tsk); return NULL; }

Since called with node == NUMA_NO_NODE, node = tak_for_get_node(orig) would definetely get called thus, decides the node where the task_struct of the newly created task would be allocated.

int tsk_fork_get_node(struct task_struct *tsk) { #ifdef CONFIG_NUMA if (tsk == kthreadd_task) return tsk->pref_node_fork; #endif return NUMA_NO_NODE; }

The kthreadd_task is the task of the kthread daemon, which is the ultimate parent of all kernel threads. However, since tsk here refers to the user space process issued the clone call, node remains set to NUMA_NO_NODE.

static inline struct task_struct *alloc_task_struct_node(int node) { return kmem_cache_alloc_node(task_struct_cachep, GFP_KERNEL, node); }
#define kmem_cache_alloc_node(...) \ alloc_hooks(kmem_cache_alloc_node_noprof(__VA_ARGS__))
void *kmem_cache_alloc_node_noprof(struct kmem_cache *s, gfp_t gfpflags, int node) { void *ret = slab_alloc_node(s, NULL, gfpflags, node, _RET_IP_, s->object_size); trace_kmem_cache_alloc(_RET_IP_, ret, s, gfpflags, node); return ret; } EXPORT_SYMBOL(kmem_cache_alloc_node_noprof);

Since task_struct is sort of a core data structure, there exists a slap cache managing the alloation of it, whose pointer is stored within task_struct_cachep.

With the existing slab cache pointer, the slab_alloc_node() function allocates the neccessary memory for the task_struct on the specified node, node (A.k.a., the same node as the CPU the task is running on, since node is set to NUMA_NO_NODES).

The call path goes as:

  • slab_alloc_node() __slab_alloc_node()
    At here, the slab allocator attempts to allocate the memory via the previous task_struct_cachep. If the obtained slab do not exist on the specified node, the check node_match(slab, node) will not be passed, thus the slow path is entered

Many other checks are performed to determine if the slow path is taken, but NUMA related info are focused here. The whole path is contained in

Image Not Showing Possible Reasons
  • The image file may be corrupted
  • The server hosting the image is unavailable
  • The image path is incorrect
  • The image format is not supported
Learn More →
linux/mm/slub.c.

  • ... __slab_alloc() ___slab_alloc()

Back to the copy_process() function with CONFIG_NUMA=y, memory policy are also setup for the newly created task:

#ifdef CONFIG_NUMA p->mempolicy = mpol_dup(p->mempolicy); if (IS_ERR(p->mempolicy)) { retval = PTR_ERR(p->mempolicy); p->mempolicy = NULL; goto bad_fork_cleanup_delayacct; } #endif

Memory Policy