Loading...
1/* Internal procfs definitions
2 *
3 * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
4 * Written by David Howells (dhowells@redhat.com)
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 */
11
12#include <linux/proc_fs.h>
13#include <linux/proc_ns.h>
14#include <linux/spinlock.h>
15#include <linux/atomic.h>
16#include <linux/binfmts.h>
17
18struct ctl_table_header;
19struct mempolicy;
20
21/*
22 * This is not completely implemented yet. The idea is to
23 * create an in-memory tree (like the actual /proc filesystem
24 * tree) of these proc_dir_entries, so that we can dynamically
25 * add new files to /proc.
26 *
27 * parent/subdir are used for the directory structure (every /proc file has a
28 * parent, but "subdir" is empty for all non-directory entries).
29 * subdir_node is used to build the rb tree "subdir" of the parent.
30 */
31struct proc_dir_entry {
32 unsigned int low_ino;
33 umode_t mode;
34 nlink_t nlink;
35 kuid_t uid;
36 kgid_t gid;
37 loff_t size;
38 const struct inode_operations *proc_iops;
39 const struct file_operations *proc_fops;
40 struct proc_dir_entry *parent;
41 struct rb_root subdir;
42 struct rb_node subdir_node;
43 void *data;
44 atomic_t count; /* use count */
45 atomic_t in_use; /* number of callers into module in progress; */
46 /* negative -> it's going away RSN */
47 struct completion *pde_unload_completion;
48 struct list_head pde_openers; /* who did ->open, but not ->release */
49 spinlock_t pde_unload_lock; /* proc_fops checks and pde_users bumps */
50 u8 namelen;
51 char name[];
52};
53
54union proc_op {
55 int (*proc_get_link)(struct dentry *, struct path *);
56 int (*proc_show)(struct seq_file *m,
57 struct pid_namespace *ns, struct pid *pid,
58 struct task_struct *task);
59};
60
61struct proc_inode {
62 struct pid *pid;
63 int fd;
64 union proc_op op;
65 struct proc_dir_entry *pde;
66 struct ctl_table_header *sysctl;
67 struct ctl_table *sysctl_entry;
68 const struct proc_ns_operations *ns_ops;
69 struct inode vfs_inode;
70};
71
72/*
73 * General functions
74 */
75static inline struct proc_inode *PROC_I(const struct inode *inode)
76{
77 return container_of(inode, struct proc_inode, vfs_inode);
78}
79
80static inline struct proc_dir_entry *PDE(const struct inode *inode)
81{
82 return PROC_I(inode)->pde;
83}
84
85static inline void *__PDE_DATA(const struct inode *inode)
86{
87 return PDE(inode)->data;
88}
89
90static inline struct pid *proc_pid(struct inode *inode)
91{
92 return PROC_I(inode)->pid;
93}
94
95static inline struct task_struct *get_proc_task(struct inode *inode)
96{
97 return get_pid_task(proc_pid(inode), PIDTYPE_PID);
98}
99
100static inline int task_dumpable(struct task_struct *task)
101{
102 int dumpable = 0;
103 struct mm_struct *mm;
104
105 task_lock(task);
106 mm = task->mm;
107 if (mm)
108 dumpable = get_dumpable(mm);
109 task_unlock(task);
110 if (dumpable == SUID_DUMP_USER)
111 return 1;
112 return 0;
113}
114
115static inline unsigned name_to_int(const struct qstr *qstr)
116{
117 const char *name = qstr->name;
118 int len = qstr->len;
119 unsigned n = 0;
120
121 if (len > 1 && *name == '0')
122 goto out;
123 while (len-- > 0) {
124 unsigned c = *name++ - '0';
125 if (c > 9)
126 goto out;
127 if (n >= (~0U-9)/10)
128 goto out;
129 n *= 10;
130 n += c;
131 }
132 return n;
133out:
134 return ~0U;
135}
136
137/*
138 * Offset of the first process in the /proc root directory..
139 */
140#define FIRST_PROCESS_ENTRY 256
141
142/* Worst case buffer size needed for holding an integer. */
143#define PROC_NUMBUF 13
144
145/*
146 * array.c
147 */
148extern const struct file_operations proc_tid_children_operations;
149
150extern int proc_tid_stat(struct seq_file *, struct pid_namespace *,
151 struct pid *, struct task_struct *);
152extern int proc_tgid_stat(struct seq_file *, struct pid_namespace *,
153 struct pid *, struct task_struct *);
154extern int proc_pid_status(struct seq_file *, struct pid_namespace *,
155 struct pid *, struct task_struct *);
156extern int proc_pid_statm(struct seq_file *, struct pid_namespace *,
157 struct pid *, struct task_struct *);
158
159/*
160 * base.c
161 */
162extern const struct dentry_operations pid_dentry_operations;
163extern int pid_getattr(struct vfsmount *, struct dentry *, struct kstat *);
164extern int proc_setattr(struct dentry *, struct iattr *);
165extern struct inode *proc_pid_make_inode(struct super_block *, struct task_struct *);
166extern int pid_revalidate(struct dentry *, unsigned int);
167extern int pid_delete_dentry(const struct dentry *);
168extern int proc_pid_readdir(struct file *, struct dir_context *);
169extern struct dentry *proc_pid_lookup(struct inode *, struct dentry *, unsigned int);
170extern loff_t mem_lseek(struct file *, loff_t, int);
171
172/* Lookups */
173typedef int instantiate_t(struct inode *, struct dentry *,
174 struct task_struct *, const void *);
175extern bool proc_fill_cache(struct file *, struct dir_context *, const char *, int,
176 instantiate_t, struct task_struct *, const void *);
177
178/*
179 * generic.c
180 */
181extern struct dentry *proc_lookup(struct inode *, struct dentry *, unsigned int);
182extern struct dentry *proc_lookup_de(struct proc_dir_entry *, struct inode *,
183 struct dentry *);
184extern int proc_readdir(struct file *, struct dir_context *);
185extern int proc_readdir_de(struct proc_dir_entry *, struct file *, struct dir_context *);
186
187static inline struct proc_dir_entry *pde_get(struct proc_dir_entry *pde)
188{
189 atomic_inc(&pde->count);
190 return pde;
191}
192extern void pde_put(struct proc_dir_entry *);
193
194static inline bool is_empty_pde(const struct proc_dir_entry *pde)
195{
196 return S_ISDIR(pde->mode) && !pde->proc_iops;
197}
198struct proc_dir_entry *proc_create_mount_point(const char *name);
199
200/*
201 * inode.c
202 */
203struct pde_opener {
204 struct file *file;
205 struct list_head lh;
206 int closing;
207 struct completion *c;
208};
209extern const struct inode_operations proc_link_inode_operations;
210
211extern const struct inode_operations proc_pid_link_inode_operations;
212
213extern void proc_init_inodecache(void);
214extern struct inode *proc_get_inode(struct super_block *, struct proc_dir_entry *);
215extern int proc_fill_super(struct super_block *);
216extern void proc_entry_rundown(struct proc_dir_entry *);
217
218/*
219 * proc_namespaces.c
220 */
221extern const struct inode_operations proc_ns_dir_inode_operations;
222extern const struct file_operations proc_ns_dir_operations;
223
224/*
225 * proc_net.c
226 */
227extern const struct file_operations proc_net_operations;
228extern const struct inode_operations proc_net_inode_operations;
229
230#ifdef CONFIG_NET
231extern int proc_net_init(void);
232#else
233static inline int proc_net_init(void) { return 0; }
234#endif
235
236/*
237 * proc_self.c
238 */
239extern int proc_setup_self(struct super_block *);
240
241/*
242 * proc_thread_self.c
243 */
244extern int proc_setup_thread_self(struct super_block *);
245extern void proc_thread_self_init(void);
246
247/*
248 * proc_sysctl.c
249 */
250#ifdef CONFIG_PROC_SYSCTL
251extern int proc_sys_init(void);
252extern void sysctl_head_put(struct ctl_table_header *);
253#else
254static inline void proc_sys_init(void) { }
255static inline void sysctl_head_put(struct ctl_table_header *head) { }
256#endif
257
258/*
259 * proc_tty.c
260 */
261#ifdef CONFIG_TTY
262extern void proc_tty_init(void);
263#else
264static inline void proc_tty_init(void) {}
265#endif
266
267/*
268 * root.c
269 */
270extern struct proc_dir_entry proc_root;
271
272extern void proc_self_init(void);
273extern int proc_remount(struct super_block *, int *, char *);
274
275/*
276 * task_[no]mmu.c
277 */
278struct proc_maps_private {
279 struct inode *inode;
280 struct task_struct *task;
281 struct mm_struct *mm;
282#ifdef CONFIG_MMU
283 struct vm_area_struct *tail_vma;
284#endif
285#ifdef CONFIG_NUMA
286 struct mempolicy *task_mempolicy;
287#endif
288};
289
290struct mm_struct *proc_mem_open(struct inode *inode, unsigned int mode);
291
292extern const struct file_operations proc_pid_maps_operations;
293extern const struct file_operations proc_tid_maps_operations;
294extern const struct file_operations proc_pid_numa_maps_operations;
295extern const struct file_operations proc_tid_numa_maps_operations;
296extern const struct file_operations proc_pid_smaps_operations;
297extern const struct file_operations proc_tid_smaps_operations;
298extern const struct file_operations proc_clear_refs_operations;
299extern const struct file_operations proc_pagemap_operations;
300
301extern unsigned long task_vsize(struct mm_struct *);
302extern unsigned long task_statm(struct mm_struct *,
303 unsigned long *, unsigned long *,
304 unsigned long *, unsigned long *);
305extern void task_mem(struct seq_file *, struct mm_struct *);
1/* SPDX-License-Identifier: GPL-2.0-or-later */
2/* Internal procfs definitions
3 *
4 * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
5 * Written by David Howells (dhowells@redhat.com)
6 */
7
8#include <linux/proc_fs.h>
9#include <linux/proc_ns.h>
10#include <linux/refcount.h>
11#include <linux/spinlock.h>
12#include <linux/atomic.h>
13#include <linux/binfmts.h>
14#include <linux/sched/coredump.h>
15#include <linux/sched/task.h>
16#include <linux/mm.h>
17
18struct ctl_table_header;
19struct mempolicy;
20
21/*
22 * This is not completely implemented yet. The idea is to
23 * create an in-memory tree (like the actual /proc filesystem
24 * tree) of these proc_dir_entries, so that we can dynamically
25 * add new files to /proc.
26 *
27 * parent/subdir are used for the directory structure (every /proc file has a
28 * parent, but "subdir" is empty for all non-directory entries).
29 * subdir_node is used to build the rb tree "subdir" of the parent.
30 */
31struct proc_dir_entry {
32 /*
33 * number of callers into module in progress;
34 * negative -> it's going away RSN
35 */
36 atomic_t in_use;
37 refcount_t refcnt;
38 struct list_head pde_openers; /* who did ->open, but not ->release */
39 /* protects ->pde_openers and all struct pde_opener instances */
40 spinlock_t pde_unload_lock;
41 struct completion *pde_unload_completion;
42 const struct inode_operations *proc_iops;
43 union {
44 const struct proc_ops *proc_ops;
45 const struct file_operations *proc_dir_ops;
46 };
47 const struct dentry_operations *proc_dops;
48 union {
49 const struct seq_operations *seq_ops;
50 int (*single_show)(struct seq_file *, void *);
51 };
52 proc_write_t write;
53 void *data;
54 unsigned int state_size;
55 unsigned int low_ino;
56 nlink_t nlink;
57 kuid_t uid;
58 kgid_t gid;
59 loff_t size;
60 struct proc_dir_entry *parent;
61 struct rb_root subdir;
62 struct rb_node subdir_node;
63 char *name;
64 umode_t mode;
65 u8 flags;
66 u8 namelen;
67 char inline_name[];
68} __randomize_layout;
69
70#define SIZEOF_PDE ( \
71 sizeof(struct proc_dir_entry) < 128 ? 128 : \
72 sizeof(struct proc_dir_entry) < 192 ? 192 : \
73 sizeof(struct proc_dir_entry) < 256 ? 256 : \
74 sizeof(struct proc_dir_entry) < 512 ? 512 : \
75 0)
76#define SIZEOF_PDE_INLINE_NAME (SIZEOF_PDE - sizeof(struct proc_dir_entry))
77
78static inline bool pde_is_permanent(const struct proc_dir_entry *pde)
79{
80 return pde->flags & PROC_ENTRY_PERMANENT;
81}
82
83static inline void pde_make_permanent(struct proc_dir_entry *pde)
84{
85 pde->flags |= PROC_ENTRY_PERMANENT;
86}
87
88extern struct kmem_cache *proc_dir_entry_cache;
89void pde_free(struct proc_dir_entry *pde);
90
91union proc_op {
92 int (*proc_get_link)(struct dentry *, struct path *);
93 int (*proc_show)(struct seq_file *m,
94 struct pid_namespace *ns, struct pid *pid,
95 struct task_struct *task);
96 int lsmid;
97};
98
99struct proc_inode {
100 struct pid *pid;
101 unsigned int fd;
102 union proc_op op;
103 struct proc_dir_entry *pde;
104 struct ctl_table_header *sysctl;
105 const struct ctl_table *sysctl_entry;
106 struct hlist_node sibling_inodes;
107 const struct proc_ns_operations *ns_ops;
108 struct inode vfs_inode;
109} __randomize_layout;
110
111/*
112 * General functions
113 */
114static inline struct proc_inode *PROC_I(const struct inode *inode)
115{
116 return container_of(inode, struct proc_inode, vfs_inode);
117}
118
119static inline struct proc_dir_entry *PDE(const struct inode *inode)
120{
121 return PROC_I(inode)->pde;
122}
123
124static inline struct pid *proc_pid(const struct inode *inode)
125{
126 return PROC_I(inode)->pid;
127}
128
129static inline struct task_struct *get_proc_task(const struct inode *inode)
130{
131 return get_pid_task(proc_pid(inode), PIDTYPE_PID);
132}
133
134void task_dump_owner(struct task_struct *task, umode_t mode,
135 kuid_t *ruid, kgid_t *rgid);
136
137unsigned name_to_int(const struct qstr *qstr);
138/*
139 * Offset of the first process in the /proc root directory..
140 */
141#define FIRST_PROCESS_ENTRY 256
142
143/* Worst case buffer size needed for holding an integer. */
144#define PROC_NUMBUF 13
145
146/**
147 * folio_precise_page_mapcount() - Number of mappings of this folio page.
148 * @folio: The folio.
149 * @page: The page.
150 *
151 * The number of present user page table entries that reference this page
152 * as tracked via the RMAP: either referenced directly (PTE) or as part of
153 * a larger area that covers this page (e.g., PMD).
154 *
155 * Use this function only for the calculation of existing statistics
156 * (USS, PSS, mapcount_max) and for debugging purposes (/proc/kpagecount).
157 *
158 * Do not add new users.
159 *
160 * Returns: The number of mappings of this folio page. 0 for
161 * folios that are not mapped to user space or are not tracked via the RMAP
162 * (e.g., shared zeropage).
163 */
164static inline int folio_precise_page_mapcount(struct folio *folio,
165 struct page *page)
166{
167 int mapcount = atomic_read(&page->_mapcount) + 1;
168
169 if (page_mapcount_is_type(mapcount))
170 mapcount = 0;
171 if (folio_test_large(folio))
172 mapcount += folio_entire_mapcount(folio);
173
174 return mapcount;
175}
176
177/*
178 * array.c
179 */
180extern const struct file_operations proc_tid_children_operations;
181
182extern void proc_task_name(struct seq_file *m, struct task_struct *p,
183 bool escape);
184extern int proc_tid_stat(struct seq_file *, struct pid_namespace *,
185 struct pid *, struct task_struct *);
186extern int proc_tgid_stat(struct seq_file *, struct pid_namespace *,
187 struct pid *, struct task_struct *);
188extern int proc_pid_status(struct seq_file *, struct pid_namespace *,
189 struct pid *, struct task_struct *);
190extern int proc_pid_statm(struct seq_file *, struct pid_namespace *,
191 struct pid *, struct task_struct *);
192
193/*
194 * base.c
195 */
196extern const struct dentry_operations pid_dentry_operations;
197extern int pid_getattr(struct mnt_idmap *, const struct path *,
198 struct kstat *, u32, unsigned int);
199extern int proc_setattr(struct mnt_idmap *, struct dentry *,
200 struct iattr *);
201extern void proc_pid_evict_inode(struct proc_inode *);
202extern struct inode *proc_pid_make_inode(struct super_block *, struct task_struct *, umode_t);
203extern void pid_update_inode(struct task_struct *, struct inode *);
204extern int pid_delete_dentry(const struct dentry *);
205extern int proc_pid_readdir(struct file *, struct dir_context *);
206struct dentry *proc_pid_lookup(struct dentry *, unsigned int);
207extern loff_t mem_lseek(struct file *, loff_t, int);
208
209/* Lookups */
210typedef struct dentry *instantiate_t(struct dentry *,
211 struct task_struct *, const void *);
212bool proc_fill_cache(struct file *, struct dir_context *, const char *, unsigned int,
213 instantiate_t, struct task_struct *, const void *);
214
215/*
216 * generic.c
217 */
218struct proc_dir_entry *proc_create_reg(const char *name, umode_t mode,
219 struct proc_dir_entry **parent, void *data);
220struct proc_dir_entry *proc_register(struct proc_dir_entry *dir,
221 struct proc_dir_entry *dp);
222extern struct dentry *proc_lookup(struct inode *, struct dentry *, unsigned int);
223struct dentry *proc_lookup_de(struct inode *, struct dentry *, struct proc_dir_entry *);
224extern int proc_readdir(struct file *, struct dir_context *);
225int proc_readdir_de(struct file *, struct dir_context *, struct proc_dir_entry *);
226
227static inline void pde_get(struct proc_dir_entry *pde)
228{
229 refcount_inc(&pde->refcnt);
230}
231extern void pde_put(struct proc_dir_entry *);
232
233static inline bool is_empty_pde(const struct proc_dir_entry *pde)
234{
235 return S_ISDIR(pde->mode) && !pde->proc_iops;
236}
237extern ssize_t proc_simple_write(struct file *, const char __user *, size_t, loff_t *);
238
239/*
240 * inode.c
241 */
242struct pde_opener {
243 struct list_head lh;
244 struct file *file;
245 bool closing;
246 struct completion *c;
247} __randomize_layout;
248extern const struct inode_operations proc_link_inode_operations;
249extern const struct inode_operations proc_pid_link_inode_operations;
250extern const struct super_operations proc_sops;
251
252void proc_init_kmemcache(void);
253void proc_invalidate_siblings_dcache(struct hlist_head *inodes, spinlock_t *lock);
254void set_proc_pid_nlink(void);
255extern struct inode *proc_get_inode(struct super_block *, struct proc_dir_entry *);
256extern void proc_entry_rundown(struct proc_dir_entry *);
257
258/*
259 * proc_namespaces.c
260 */
261extern const struct inode_operations proc_ns_dir_inode_operations;
262extern const struct file_operations proc_ns_dir_operations;
263
264/*
265 * proc_net.c
266 */
267extern const struct file_operations proc_net_operations;
268extern const struct inode_operations proc_net_inode_operations;
269
270#ifdef CONFIG_NET
271extern int proc_net_init(void);
272#else
273static inline int proc_net_init(void) { return 0; }
274#endif
275
276/*
277 * proc_self.c
278 */
279extern int proc_setup_self(struct super_block *);
280
281/*
282 * proc_thread_self.c
283 */
284extern int proc_setup_thread_self(struct super_block *);
285extern void proc_thread_self_init(void);
286
287/*
288 * proc_sysctl.c
289 */
290#ifdef CONFIG_PROC_SYSCTL
291extern int proc_sys_init(void);
292extern void proc_sys_evict_inode(struct inode *inode,
293 struct ctl_table_header *head);
294#else
295static inline void proc_sys_init(void) { }
296static inline void proc_sys_evict_inode(struct inode *inode,
297 struct ctl_table_header *head) { }
298#endif
299
300/*
301 * proc_tty.c
302 */
303#ifdef CONFIG_TTY
304extern void proc_tty_init(void);
305#else
306static inline void proc_tty_init(void) {}
307#endif
308
309/*
310 * root.c
311 */
312extern struct proc_dir_entry proc_root;
313
314extern void proc_self_init(void);
315
316/*
317 * task_[no]mmu.c
318 */
319struct mem_size_stats;
320struct proc_maps_private {
321 struct inode *inode;
322 struct task_struct *task;
323 struct mm_struct *mm;
324 struct vma_iterator iter;
325#ifdef CONFIG_NUMA
326 struct mempolicy *task_mempolicy;
327#endif
328} __randomize_layout;
329
330struct mm_struct *proc_mem_open(struct inode *inode, unsigned int mode);
331
332extern const struct file_operations proc_pid_maps_operations;
333extern const struct file_operations proc_pid_numa_maps_operations;
334extern const struct file_operations proc_pid_smaps_operations;
335extern const struct file_operations proc_pid_smaps_rollup_operations;
336extern const struct file_operations proc_clear_refs_operations;
337extern const struct file_operations proc_pagemap_operations;
338
339extern unsigned long task_vsize(struct mm_struct *);
340extern unsigned long task_statm(struct mm_struct *,
341 unsigned long *, unsigned long *,
342 unsigned long *, unsigned long *);
343extern void task_mem(struct seq_file *, struct mm_struct *);
344
345extern const struct dentry_operations proc_net_dentry_ops;
346static inline void pde_force_lookup(struct proc_dir_entry *pde)
347{
348 /* /proc/net/ entries can be changed under us by setns(CLONE_NEWNET) */
349 pde->proc_dops = &proc_net_dentry_ops;
350}
351
352/*
353 * Add a new procfs dentry that can't serve as a mountpoint. That should
354 * encompass anything that is ephemeral and can just disappear while the
355 * process is still around.
356 */
357static inline struct dentry *proc_splice_unmountable(struct inode *inode,
358 struct dentry *dentry, const struct dentry_operations *d_ops)
359{
360 d_set_d_op(dentry, d_ops);
361 dont_mount(dentry);
362 return d_splice_alias(inode, dentry);
363}