diff options
Diffstat (limited to 'fs/pidfs.c')
| -rw-r--r-- | fs/pidfs.c | 681 |
1 files changed, 598 insertions, 83 deletions
diff --git a/fs/pidfs.c b/fs/pidfs.c index 618abb1fa1b8..c1f0a067be40 100644 --- a/fs/pidfs.c +++ b/fs/pidfs.c @@ -1,5 +1,6 @@ // SPDX-License-Identifier: GPL-2.0 #include <linux/anon_inodes.h> +#include <linux/exportfs.h> #include <linux/file.h> #include <linux/fs.h> #include <linux/cgroup.h> @@ -19,10 +20,125 @@ #include <linux/time_namespace.h> #include <linux/utsname.h> #include <net/net_namespace.h> +#include <linux/coredump.h> #include "internal.h" #include "mount.h" +static struct kmem_cache *pidfs_cachep __ro_after_init; + +/* + * Stashes information that userspace needs to access even after the + * process has been reaped. + */ +struct pidfs_exit_info { + __u64 cgroupid; + __s32 exit_code; + __u32 coredump_mask; +}; + +struct pidfs_inode { + struct pidfs_exit_info __pei; + struct pidfs_exit_info *exit_info; + struct inode vfs_inode; +}; + +static inline struct pidfs_inode *pidfs_i(struct inode *inode) +{ + return container_of(inode, struct pidfs_inode, vfs_inode); +} + +static struct rb_root pidfs_ino_tree = RB_ROOT; + +#if BITS_PER_LONG == 32 +static inline unsigned long pidfs_ino(u64 ino) +{ + return lower_32_bits(ino); +} + +/* On 32 bit the generation number are the upper 32 bits. */ +static inline u32 pidfs_gen(u64 ino) +{ + return upper_32_bits(ino); +} + +#else + +/* On 64 bit simply return ino. */ +static inline unsigned long pidfs_ino(u64 ino) +{ + return ino; +} + +/* On 64 bit the generation number is 0. */ +static inline u32 pidfs_gen(u64 ino) +{ + return 0; +} +#endif + +static int pidfs_ino_cmp(struct rb_node *a, const struct rb_node *b) +{ + struct pid *pid_a = rb_entry(a, struct pid, pidfs_node); + struct pid *pid_b = rb_entry(b, struct pid, pidfs_node); + u64 pid_ino_a = pid_a->ino; + u64 pid_ino_b = pid_b->ino; + + if (pid_ino_a < pid_ino_b) + return -1; + if (pid_ino_a > pid_ino_b) + return 1; + return 0; +} + +void pidfs_add_pid(struct pid *pid) +{ + static u64 pidfs_ino_nr = 2; + + /* + * On 64 bit nothing special happens. The 64bit number assigned + * to struct pid is the inode number. + * + * On 32 bit the 64 bit number assigned to struct pid is split + * into two 32 bit numbers. The lower 32 bits are used as the + * inode number and the upper 32 bits are used as the inode + * generation number. + * + * On 32 bit pidfs_ino() will return the lower 32 bit. When + * pidfs_ino() returns zero a wrap around happened. When a + * wraparound happens the 64 bit number will be incremented by 2 + * so inode numbering starts at 2 again. + * + * On 64 bit comparing two pidfds is as simple as comparing + * inode numbers. + * + * When a wraparound happens on 32 bit multiple pidfds with the + * same inode number are likely to exist (This isn't a problem + * since before pidfs pidfds used the anonymous inode meaning + * all pidfds had the same inode number.). Userspace can + * reconstruct the 64 bit identifier by retrieving both the + * inode number and the inode generation number to compare or + * use file handles. + */ + if (pidfs_ino(pidfs_ino_nr) == 0) + pidfs_ino_nr += 2; + + pid->ino = pidfs_ino_nr; + pid->stashed = NULL; + pidfs_ino_nr++; + + write_seqcount_begin(&pidmap_lock_seq); + rb_find_add_rcu(&pid->pidfs_node, &pidfs_ino_tree, pidfs_ino_cmp); + write_seqcount_end(&pidmap_lock_seq); +} + +void pidfs_remove_pid(struct pid *pid) +{ + write_seqcount_begin(&pidmap_lock_seq); + rb_erase(&pid->pidfs_node, &pidfs_ino_tree); + write_seqcount_end(&pidmap_lock_seq); +} + #ifdef CONFIG_PROC_FS /** * pidfd_show_fdinfo - print information about a pidfd @@ -96,36 +212,64 @@ static void pidfd_show_fdinfo(struct seq_file *m, struct file *f) static __poll_t pidfd_poll(struct file *file, struct poll_table_struct *pts) { struct pid *pid = pidfd_pid(file); - bool thread = file->f_flags & PIDFD_THREAD; struct task_struct *task; __poll_t poll_flags = 0; poll_wait(file, &pid->wait_pidfd, pts); /* - * Depending on PIDFD_THREAD, inform pollers when the thread - * or the whole thread-group exits. + * Don't wake waiters if the thread-group leader exited + * prematurely. They either get notified when the last subthread + * exits or not at all if one of the remaining subthreads execs + * and assumes the struct pid of the old thread-group leader. */ guard(rcu)(); task = pid_task(pid, PIDTYPE_PID); if (!task) poll_flags = EPOLLIN | EPOLLRDNORM | EPOLLHUP; - else if (task->exit_state && (thread || thread_group_empty(task))) + else if (task->exit_state && !delay_group_leader(task)) poll_flags = EPOLLIN | EPOLLRDNORM; return poll_flags; } -static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long arg) +static inline bool pid_in_current_pidns(const struct pid *pid) +{ + const struct pid_namespace *ns = task_active_pid_ns(current); + + if (ns->level <= pid->level) + return pid->numbers[ns->level].ns == ns; + + return false; +} + +static __u32 pidfs_coredump_mask(unsigned long mm_flags) +{ + switch (__get_dumpable(mm_flags)) { + case SUID_DUMP_USER: + return PIDFD_COREDUMP_USER; + case SUID_DUMP_ROOT: + return PIDFD_COREDUMP_ROOT; + case SUID_DUMP_DISABLE: + return PIDFD_COREDUMP_SKIP; + default: + WARN_ON_ONCE(true); + } + + return 0; +} + +static long pidfd_info(struct file *file, unsigned int cmd, unsigned long arg) { struct pidfd_info __user *uinfo = (struct pidfd_info __user *)arg; + struct inode *inode = file_inode(file); + struct pid *pid = pidfd_pid(file); size_t usize = _IOC_SIZE(cmd); struct pidfd_info kinfo = {}; + struct pidfs_exit_info *exit_info; struct user_namespace *user_ns; + struct task_struct *task; const struct cred *c; __u64 mask; -#ifdef CONFIG_CGROUPS - struct cgroup *cgrp; -#endif if (!uinfo) return -EINVAL; @@ -135,10 +279,53 @@ static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long if (copy_from_user(&mask, &uinfo->mask, sizeof(mask))) return -EFAULT; + /* + * Restrict information retrieval to tasks within the caller's pid + * namespace hierarchy. + */ + if (!pid_in_current_pidns(pid)) + return -ESRCH; + + if (mask & PIDFD_INFO_EXIT) { + exit_info = READ_ONCE(pidfs_i(inode)->exit_info); + if (exit_info) { + kinfo.mask |= PIDFD_INFO_EXIT; +#ifdef CONFIG_CGROUPS + kinfo.cgroupid = exit_info->cgroupid; + kinfo.mask |= PIDFD_INFO_CGROUPID; +#endif + kinfo.exit_code = exit_info->exit_code; + } + } + + if (mask & PIDFD_INFO_COREDUMP) { + kinfo.mask |= PIDFD_INFO_COREDUMP; + kinfo.coredump_mask = READ_ONCE(pidfs_i(inode)->__pei.coredump_mask); + } + + task = get_pid_task(pid, PIDTYPE_PID); + if (!task) { + /* + * If the task has already been reaped, only exit + * information is available + */ + if (!(mask & PIDFD_INFO_EXIT)) + return -ESRCH; + + goto copy_out; + } + c = get_task_cred(task); if (!c) return -ESRCH; + if (!(kinfo.mask & PIDFD_INFO_COREDUMP)) { + task_lock(task); + if (task->mm) + kinfo.coredump_mask = pidfs_coredump_mask(task->mm->flags); + task_unlock(task); + } + /* Unconditionally return identifiers and credentials, the rest only on request */ user_ns = current_user_ns(); @@ -154,11 +341,15 @@ static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long put_cred(c); #ifdef CONFIG_CGROUPS - rcu_read_lock(); - cgrp = task_dfl_cgroup(task); - kinfo.cgroupid = cgroup_id(cgrp); - kinfo.mask |= PIDFD_INFO_CGROUPID; - rcu_read_unlock(); + if (!kinfo.cgroupid) { + struct cgroup *cgrp; + + rcu_read_lock(); + cgrp = task_dfl_cgroup(task); + kinfo.cgroupid = cgroup_id(cgrp); + kinfo.mask |= PIDFD_INFO_CGROUPID; + rcu_read_unlock(); + } #endif /* @@ -178,33 +369,72 @@ static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long if (kinfo.pid == 0 || kinfo.tgid == 0 || (kinfo.ppid == 0 && kinfo.pid != 1)) return -ESRCH; +copy_out: /* * If userspace and the kernel have the same struct size it can just * be copied. If userspace provides an older struct, only the bits that * userspace knows about will be copied. If userspace provides a new * struct, only the bits that the kernel knows about will be copied. */ - if (copy_to_user(uinfo, &kinfo, min(usize, sizeof(kinfo)))) - return -EFAULT; + return copy_struct_to_user(uinfo, usize, &kinfo, sizeof(kinfo), NULL); +} - return 0; +static bool pidfs_ioctl_valid(unsigned int cmd) +{ + switch (cmd) { + case FS_IOC_GETVERSION: + case PIDFD_GET_CGROUP_NAMESPACE: + case PIDFD_GET_IPC_NAMESPACE: + case PIDFD_GET_MNT_NAMESPACE: + case PIDFD_GET_NET_NAMESPACE: + case PIDFD_GET_PID_FOR_CHILDREN_NAMESPACE: + case PIDFD_GET_TIME_NAMESPACE: + case PIDFD_GET_TIME_FOR_CHILDREN_NAMESPACE: + case PIDFD_GET_UTS_NAMESPACE: + case PIDFD_GET_USER_NAMESPACE: + case PIDFD_GET_PID_NAMESPACE: + return true; + } + + /* Extensible ioctls require some more careful checks. */ + switch (_IOC_NR(cmd)) { + case _IOC_NR(PIDFD_GET_INFO): + /* + * Try to prevent performing a pidfd ioctl when someone + * erronously mistook the file descriptor for a pidfd. + * This is not perfect but will catch most cases. + */ + return (_IOC_TYPE(cmd) == _IOC_TYPE(PIDFD_GET_INFO)); + } + + return false; } static long pidfd_ioctl(struct file *file, unsigned int cmd, unsigned long arg) { struct task_struct *task __free(put_task) = NULL; struct nsproxy *nsp __free(put_nsproxy) = NULL; - struct pid *pid = pidfd_pid(file); struct ns_common *ns_common = NULL; struct pid_namespace *pid_ns; - task = get_pid_task(pid, PIDTYPE_PID); - if (!task) - return -ESRCH; + if (!pidfs_ioctl_valid(cmd)) + return -ENOIOCTLCMD; + + if (cmd == FS_IOC_GETVERSION) { + if (!arg) + return -EINVAL; + + __u32 __user *argp = (__u32 __user *)arg; + return put_user(file_inode(file)->i_generation, argp); + } /* Extensible IOCTL that does not open namespace FDs, take a shortcut */ if (_IOC_NR(cmd) == _IOC_NR(PIDFD_GET_INFO)) - return pidfd_info(task, cmd, arg); + return pidfd_info(file, cmd, arg); + + task = get_pid_task(pidfd_pid(file), PIDTYPE_PID); + if (!task) + return -ESRCH; if (arg) return -EINVAL; @@ -316,42 +546,76 @@ struct pid *pidfd_pid(const struct file *file) return file_inode(file)->i_private; } -static struct vfsmount *pidfs_mnt __ro_after_init; - -#if BITS_PER_LONG == 32 /* - * Provide a fallback mechanism for 32-bit systems so processes remain - * reliably comparable by inode number even on those systems. + * We're called from release_task(). We know there's at least one + * reference to struct pid being held that won't be released until the + * task has been reaped which cannot happen until we're out of + * release_task(). + * + * If this struct pid is referred to by a pidfd then + * stashed_dentry_get() will return the dentry and inode for that struct + * pid. Since we've taken a reference on it there's now an additional + * reference from the exit path on it. Which is fine. We're going to put + * it again in a second and we know that the pid is kept alive anyway. + * + * Worst case is that we've filled in the info and immediately free the + * dentry and inode afterwards since the pidfd has been closed. Since + * pidfs_exit() currently is placed after exit_task_work() we know that + * it cannot be us aka the exiting task holding a pidfd to ourselves. */ -static DEFINE_IDA(pidfd_inum_ida); - -static int pidfs_inum(struct pid *pid, unsigned long *ino) +void pidfs_exit(struct task_struct *tsk) { - int ret; + struct dentry *dentry; - ret = ida_alloc_range(&pidfd_inum_ida, RESERVED_PIDS + 1, - UINT_MAX, GFP_ATOMIC); - if (ret < 0) - return -ENOSPC; + might_sleep(); - *ino = ret; - return 0; -} + dentry = stashed_dentry_get(&task_pid(tsk)->stashed); + if (dentry) { + struct inode *inode = d_inode(dentry); + struct pidfs_exit_info *exit_info = &pidfs_i(inode)->__pei; +#ifdef CONFIG_CGROUPS + struct cgroup *cgrp; -static inline void pidfs_free_inum(unsigned long ino) -{ - if (ino > 0) - ida_free(&pidfd_inum_ida, ino); + rcu_read_lock(); + cgrp = task_dfl_cgroup(tsk); + exit_info->cgroupid = cgroup_id(cgrp); + rcu_read_unlock(); +#endif + exit_info->exit_code = tsk->exit_code; + + /* Ensure that PIDFD_GET_INFO sees either all or nothing. */ + smp_store_release(&pidfs_i(inode)->exit_info, &pidfs_i(inode)->__pei); + dput(dentry); + } } -#else -static inline int pidfs_inum(struct pid *pid, unsigned long *ino) + +#ifdef CONFIG_COREDUMP +void pidfs_coredump(const struct coredump_params *cprm) { - *ino = pid->ino; - return 0; + struct pid *pid = cprm->pid; + struct pidfs_exit_info *exit_info; + struct dentry *dentry; + struct inode *inode; + __u32 coredump_mask = 0; + + dentry = pid->stashed; + if (WARN_ON_ONCE(!dentry)) + return; + + inode = d_inode(dentry); + exit_info = &pidfs_i(inode)->__pei; + /* Note how we were coredumped. */ + coredump_mask = pidfs_coredump_mask(cprm->mm_flags); + /* Note that we actually did coredump. */ + coredump_mask |= PIDFD_COREDUMPED; + /* If coredumping is set to skip we should never end up here. */ + VFS_WARN_ON_ONCE(coredump_mask & PIDFD_COREDUMP_SKIP); + smp_store_release(&exit_info->coredump_mask, coredump_mask); } -#define pidfs_free_inum(ino) ((void)(ino)) #endif +static struct vfsmount *pidfs_mnt __ro_after_init; + /* * The vfs falls back to simple_setattr() if i_op->setattr() isn't * implemented. Let's reject it completely until we have a clean @@ -360,36 +624,14 @@ static inline int pidfs_inum(struct pid *pid, unsigned long *ino) static int pidfs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, struct iattr *attr) { - return -EOPNOTSUPP; + return anon_inode_setattr(idmap, dentry, attr); } - -/* - * User space expects pidfs inodes to have no file type in st_mode. - * - * In particular, 'lsof' has this legacy logic: - * - * type = s->st_mode & S_IFMT; - * switch (type) { - * ... - * case 0: - * if (!strcmp(p, "anon_inode")) - * Lf->ntype = Ntype = N_ANON_INODE; - * - * to detect our old anon_inode logic. - * - * Rather than mess with our internal sane inode data, just fix it - * up here in getattr() by masking off the format bits. - */ static int pidfs_getattr(struct mnt_idmap *idmap, const struct path *path, struct kstat *stat, u32 request_mask, unsigned int query_flags) { - struct inode *inode = d_inode(path->dentry); - - generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat); - stat->mode &= ~S_IFMT; - return 0; + return anon_inode_getattr(idmap, path, stat, request_mask, query_flags); } static const struct inode_operations pidfs_inode_operations = { @@ -403,12 +645,32 @@ static void pidfs_evict_inode(struct inode *inode) clear_inode(inode); put_pid(pid); - pidfs_free_inum(inode->i_ino); +} + +static struct inode *pidfs_alloc_inode(struct super_block *sb) +{ + struct pidfs_inode *pi; + + pi = alloc_inode_sb(sb, pidfs_cachep, GFP_KERNEL); + if (!pi) + return NULL; + + memset(&pi->__pei, 0, sizeof(pi->__pei)); + pi->exit_info = NULL; + + return &pi->vfs_inode; +} + +static void pidfs_free_inode(struct inode *inode) +{ + kmem_cache_free(pidfs_cachep, pidfs_i(inode)); } static const struct super_operations pidfs_sops = { + .alloc_inode = pidfs_alloc_inode, .drop_inode = generic_delete_inode, .evict_inode = pidfs_evict_inode, + .free_inode = pidfs_free_inode, .statfs = simple_statfs, }; @@ -421,25 +683,193 @@ static char *pidfs_dname(struct dentry *dentry, char *buffer, int buflen) return dynamic_dname(buffer, buflen, "anon_inode:[pidfd]"); } -static const struct dentry_operations pidfs_dentry_operations = { - .d_delete = always_delete_dentry, +const struct dentry_operations pidfs_dentry_operations = { .d_dname = pidfs_dname, .d_prune = stashed_dentry_prune, }; +static int pidfs_encode_fh(struct inode *inode, u32 *fh, int *max_len, + struct inode *parent) +{ + const struct pid *pid = inode->i_private; + + if (*max_len < 2) { + *max_len = 2; + return FILEID_INVALID; + } + + *max_len = 2; + *(u64 *)fh = pid->ino; + return FILEID_KERNFS; +} + +static int pidfs_ino_find(const void *key, const struct rb_node *node) +{ + const u64 pid_ino = *(u64 *)key; + const struct pid *pid = rb_entry(node, struct pid, pidfs_node); + + if (pid_ino < pid->ino) + return -1; + if (pid_ino > pid->ino) + return 1; + return 0; +} + +/* Find a struct pid based on the inode number. */ +static struct pid *pidfs_ino_get_pid(u64 ino) +{ + struct pid *pid; + struct rb_node *node; + unsigned int seq; + + guard(rcu)(); + do { + seq = read_seqcount_begin(&pidmap_lock_seq); + node = rb_find_rcu(&ino, &pidfs_ino_tree, pidfs_ino_find); + if (node) + break; + } while (read_seqcount_retry(&pidmap_lock_seq, seq)); + + if (!node) + return NULL; + + pid = rb_entry(node, struct pid, pidfs_node); + + /* Within our pid namespace hierarchy? */ + if (pid_vnr(pid) == 0) + return NULL; + + return get_pid(pid); +} + +static struct dentry *pidfs_fh_to_dentry(struct super_block *sb, + struct fid *fid, int fh_len, + int fh_type) +{ + int ret; + u64 pid_ino; + struct path path; + struct pid *pid; + + if (fh_len < 2) + return NULL; + + switch (fh_type) { + case FILEID_KERNFS: + pid_ino = *(u64 *)fid; + break; + default: + return NULL; + } + + pid = pidfs_ino_get_pid(pid_ino); + if (!pid) + return NULL; + + ret = path_from_stashed(&pid->stashed, pidfs_mnt, pid, &path); + if (ret < 0) + return ERR_PTR(ret); + + mntput(path.mnt); + return path.dentry; +} + +/* + * Make sure that we reject any nonsensical flags that users pass via + * open_by_handle_at(). Note that PIDFD_THREAD is defined as O_EXCL, and + * PIDFD_NONBLOCK as O_NONBLOCK. + */ +#define VALID_FILE_HANDLE_OPEN_FLAGS \ + (O_RDONLY | O_WRONLY | O_RDWR | O_NONBLOCK | O_CLOEXEC | O_EXCL) + +static int pidfs_export_permission(struct handle_to_path_ctx *ctx, + unsigned int oflags) +{ + if (oflags & ~(VALID_FILE_HANDLE_OPEN_FLAGS | O_LARGEFILE)) + return -EINVAL; + + /* + * pidfd_ino_get_pid() will verify that the struct pid is part + * of the caller's pid namespace hierarchy. No further + * permission checks are needed. + */ + return 0; +} + +static inline bool pidfs_pid_valid(struct pid *pid, const struct path *path, + unsigned int flags) +{ + enum pid_type type; + + if (flags & PIDFD_STALE) + return true; + + /* + * Make sure that if a pidfd is created PIDFD_INFO_EXIT + * information will be available. So after an inode for the + * pidfd has been allocated perform another check that the pid + * is still alive. If it is exit information is available even + * if the task gets reaped before the pidfd is returned to + * userspace. The only exception are indicated by PIDFD_STALE: + * + * (1) The kernel is in the middle of task creation and thus no + * task linkage has been established yet. + * (2) The caller knows @pid has been registered in pidfs at a + * time when the task was still alive. + * + * In both cases exit information will have been reported. + */ + if (flags & PIDFD_THREAD) + type = PIDTYPE_PID; + else + type = PIDTYPE_TGID; + + /* + * Since pidfs_exit() is called before struct pid's task linkage + * is removed the case where the task got reaped but a dentry + * was already attached to struct pid and exit information was + * recorded and published can be handled correctly. + */ + if (unlikely(!pid_has_task(pid, type))) { + struct inode *inode = d_inode(path->dentry); + return !!READ_ONCE(pidfs_i(inode)->exit_info); + } + + return true; +} + +static struct file *pidfs_export_open(struct path *path, unsigned int oflags) +{ + if (!pidfs_pid_valid(d_inode(path->dentry)->i_private, path, oflags)) + return ERR_PTR(-ESRCH); + + /* + * Clear O_LARGEFILE as open_by_handle_at() forces it and raise + * O_RDWR as pidfds always are. + */ + oflags &= ~O_LARGEFILE; + return dentry_open(path, oflags | O_RDWR, current_cred()); +} + +static const struct export_operations pidfs_export_operations = { + .encode_fh = pidfs_encode_fh, + .fh_to_dentry = pidfs_fh_to_dentry, + .open = pidfs_export_open, + .permission = pidfs_export_permission, +}; + static int pidfs_init_inode(struct inode *inode, void *data) { + const struct pid *pid = data; + inode->i_private = data; - inode->i_flags |= S_PRIVATE; + inode->i_flags |= S_PRIVATE | S_ANON_INODE; inode->i_mode |= S_IRWXU; inode->i_op = &pidfs_inode_operations; inode->i_fop = &pidfs_file_operations; - /* - * Inode numbering for pidfs start at RESERVED_PIDS + 1. This - * avoids collisions with the root inode which is 1 for pseudo - * filesystems. - */ - return pidfs_inum(data, &inode->i_ino); + inode->i_ino = pidfs_ino(pid->ino); + inode->i_generation = pidfs_gen(pid->ino); + return 0; } static void pidfs_put_data(void *data) @@ -462,6 +892,7 @@ static int pidfs_init_fs_context(struct fs_context *fc) return -ENOMEM; ctx->ops = &pidfs_sops; + ctx->eops = &pidfs_export_operations; ctx->dops = &pidfs_dentry_operations; fc->s_fs_info = (void *)&pidfs_stashed_ops; return 0; @@ -475,22 +906,106 @@ static struct file_system_type pidfs_type = { struct file *pidfs_alloc_file(struct pid *pid, unsigned int flags) { - struct file *pidfd_file; - struct path path; + struct path path __free(path_put) = {}; int ret; + /* + * Ensure that PIDFD_STALE can be passed as a flag without + * overloading other uapi pidfd flags. + */ + BUILD_BUG_ON(PIDFD_STALE == PIDFD_THREAD); + BUILD_BUG_ON(PIDFD_STALE == PIDFD_NONBLOCK); + ret = path_from_stashed(&pid->stashed, pidfs_mnt, get_pid(pid), &path); if (ret < 0) return ERR_PTR(ret); + if (!pidfs_pid_valid(pid, &path, flags)) + return ERR_PTR(-ESRCH); + + flags &= ~PIDFD_STALE; + flags |= O_RDWR; pidfd_file = dentry_open(&path, flags, current_cred()); - path_put(&path); + /* Raise PIDFD_THREAD explicitly as do_dentry_open() strips it. */ + if (!IS_ERR(pidfd_file)) + pidfd_file->f_flags |= (flags & PIDFD_THREAD); + return pidfd_file; } +/** + * pidfs_register_pid - register a struct pid in pidfs + * @pid: pid to pin + * + * Register a struct pid in pidfs. Needs to be paired with + * pidfs_put_pid() to not risk leaking the pidfs dentry and inode. + * + * Return: On success zero, on error a negative error code is returned. + */ +int pidfs_register_pid(struct pid *pid) +{ + struct path path __free(path_put) = {}; + int ret; + + might_sleep(); + + if (!pid) + return 0; + + ret = path_from_stashed(&pid->stashed, pidfs_mnt, get_pid(pid), &path); + if (unlikely(ret)) + return ret; + /* Keep the dentry and only put the reference to the mount. */ + path.dentry = NULL; + return 0; +} + +/** + * pidfs_get_pid - pin a struct pid through pidfs + * @pid: pid to pin + * + * Similar to pidfs_register_pid() but only valid if the caller knows + * there's a reference to the @pid through a dentry already that can't + * go away. + */ +void pidfs_get_pid(struct pid *pid) +{ + if (!pid) + return; + WARN_ON_ONCE(!stashed_dentry_get(&pid->stashed)); +} + +/** + * pidfs_put_pid - drop a pidfs reference + * @pid: pid to drop + * + * Drop a reference to @pid via pidfs. This is only safe if the + * reference has been taken via pidfs_get_pid(). + */ +void pidfs_put_pid(struct pid *pid) +{ + might_sleep(); + + if (!pid) + return; + VFS_WARN_ON_ONCE(!pid->stashed); + dput(pid->stashed); +} + +static void pidfs_inode_init_once(void *data) +{ + struct pidfs_inode *pi = data; + + inode_init_once(&pi->vfs_inode); +} + void __init pidfs_init(void) { + pidfs_cachep = kmem_cache_create("pidfs_cache", sizeof(struct pidfs_inode), 0, + (SLAB_HWCACHE_ALIGN | SLAB_RECLAIM_ACCOUNT | + SLAB_ACCOUNT | SLAB_PANIC), + pidfs_inode_init_once); pidfs_mnt = kern_mount(&pidfs_type); if (IS_ERR(pidfs_mnt)) panic("Failed to mount pidfs pseudo filesystem"); |