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
18 struct ctl_table_header;
19 struct 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 */
31 struct 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
pde_is_permanent(const struct proc_dir_entry * pde)78 static inline bool pde_is_permanent(const struct proc_dir_entry *pde)
79 {
80 return pde->flags & PROC_ENTRY_PERMANENT;
81 }
82
pde_make_permanent(struct proc_dir_entry * pde)83 static inline void pde_make_permanent(struct proc_dir_entry *pde)
84 {
85 pde->flags |= PROC_ENTRY_PERMANENT;
86 }
87
pde_has_proc_read_iter(const struct proc_dir_entry * pde)88 static inline bool pde_has_proc_read_iter(const struct proc_dir_entry *pde)
89 {
90 return pde->flags & PROC_ENTRY_proc_read_iter;
91 }
92
pde_has_proc_compat_ioctl(const struct proc_dir_entry * pde)93 static inline bool pde_has_proc_compat_ioctl(const struct proc_dir_entry *pde)
94 {
95 #ifdef CONFIG_COMPAT
96 return pde->flags & PROC_ENTRY_proc_compat_ioctl;
97 #else
98 return false;
99 #endif
100 }
101
pde_has_proc_lseek(const struct proc_dir_entry * pde)102 static inline bool pde_has_proc_lseek(const struct proc_dir_entry *pde)
103 {
104 return pde->flags & PROC_ENTRY_proc_lseek;
105 }
106
107 extern struct kmem_cache *proc_dir_entry_cache;
108 void pde_free(struct proc_dir_entry *pde);
109
110 union proc_op {
111 int (*proc_get_link)(struct dentry *, struct path *);
112 int (*proc_show)(struct seq_file *m,
113 struct pid_namespace *ns, struct pid *pid,
114 struct task_struct *task);
115 int lsmid;
116 };
117
118 struct proc_inode {
119 struct pid *pid;
120 unsigned int fd;
121 union proc_op op;
122 struct proc_dir_entry *pde;
123 struct ctl_table_header *sysctl;
124 struct ctl_table *sysctl_entry;
125 struct hlist_node sibling_inodes;
126 const struct proc_ns_operations *ns_ops;
127 struct inode vfs_inode;
128 } __randomize_layout;
129
130 /*
131 * General functions
132 */
PROC_I(const struct inode * inode)133 static inline struct proc_inode *PROC_I(const struct inode *inode)
134 {
135 return container_of(inode, struct proc_inode, vfs_inode);
136 }
137
PDE(const struct inode * inode)138 static inline struct proc_dir_entry *PDE(const struct inode *inode)
139 {
140 return PROC_I(inode)->pde;
141 }
142
proc_pid(const struct inode * inode)143 static inline struct pid *proc_pid(const struct inode *inode)
144 {
145 return PROC_I(inode)->pid;
146 }
147
get_proc_task(const struct inode * inode)148 static inline struct task_struct *get_proc_task(const struct inode *inode)
149 {
150 return get_pid_task(proc_pid(inode), PIDTYPE_PID);
151 }
152
153 void task_dump_owner(struct task_struct *task, umode_t mode,
154 kuid_t *ruid, kgid_t *rgid);
155
156 unsigned name_to_int(const struct qstr *qstr);
157 /*
158 * Offset of the first process in the /proc root directory..
159 */
160 #define FIRST_PROCESS_ENTRY 256
161
162 /* Worst case buffer size needed for holding an integer. */
163 #define PROC_NUMBUF 13
164
165 /**
166 * folio_precise_page_mapcount() - Number of mappings of this folio page.
167 * @folio: The folio.
168 * @page: The page.
169 *
170 * The number of present user page table entries that reference this page
171 * as tracked via the RMAP: either referenced directly (PTE) or as part of
172 * a larger area that covers this page (e.g., PMD).
173 *
174 * Use this function only for the calculation of existing statistics
175 * (USS, PSS, mapcount_max) and for debugging purposes (/proc/kpagecount).
176 *
177 * Do not add new users.
178 *
179 * Returns: The number of mappings of this folio page. 0 for
180 * folios that are not mapped to user space or are not tracked via the RMAP
181 * (e.g., shared zeropage).
182 */
folio_precise_page_mapcount(struct folio * folio,struct page * page)183 static inline int folio_precise_page_mapcount(struct folio *folio,
184 struct page *page)
185 {
186 int mapcount = atomic_read(&page->_mapcount) + 1;
187
188 if (page_mapcount_is_type(mapcount))
189 mapcount = 0;
190 if (folio_test_large(folio))
191 mapcount += folio_entire_mapcount(folio);
192
193 return mapcount;
194 }
195
196 /*
197 * array.c
198 */
199 extern const struct file_operations proc_tid_children_operations;
200
201 extern void proc_task_name(struct seq_file *m, struct task_struct *p,
202 bool escape);
203 extern int proc_tid_stat(struct seq_file *, struct pid_namespace *,
204 struct pid *, struct task_struct *);
205 extern int proc_tgid_stat(struct seq_file *, struct pid_namespace *,
206 struct pid *, struct task_struct *);
207 extern int proc_pid_status(struct seq_file *, struct pid_namespace *,
208 struct pid *, struct task_struct *);
209 extern int proc_pid_statm(struct seq_file *, struct pid_namespace *,
210 struct pid *, struct task_struct *);
211
212 /*
213 * base.c
214 */
215 extern const struct dentry_operations pid_dentry_operations;
216 extern int pid_getattr(struct mnt_idmap *, const struct path *,
217 struct kstat *, u32, unsigned int);
218 extern int proc_setattr(struct mnt_idmap *, struct dentry *,
219 struct iattr *);
220 extern void proc_pid_evict_inode(struct proc_inode *);
221 extern struct inode *proc_pid_make_inode(struct super_block *, struct task_struct *, umode_t);
222 extern void pid_update_inode(struct task_struct *, struct inode *);
223 extern int pid_delete_dentry(const struct dentry *);
224 extern int proc_pid_readdir(struct file *, struct dir_context *);
225 struct dentry *proc_pid_lookup(struct dentry *, unsigned int);
226 extern loff_t mem_lseek(struct file *, loff_t, int);
227
228 /* Lookups */
229 typedef struct dentry *instantiate_t(struct dentry *,
230 struct task_struct *, const void *);
231 bool proc_fill_cache(struct file *, struct dir_context *, const char *, unsigned int,
232 instantiate_t, struct task_struct *, const void *);
233
234 /*
235 * generic.c
236 */
237 struct proc_dir_entry *proc_create_reg(const char *name, umode_t mode,
238 struct proc_dir_entry **parent, void *data);
239 struct proc_dir_entry *proc_register(struct proc_dir_entry *dir,
240 struct proc_dir_entry *dp);
241 extern struct dentry *proc_lookup(struct inode *, struct dentry *, unsigned int);
242 struct dentry *proc_lookup_de(struct inode *, struct dentry *, struct proc_dir_entry *);
243 extern int proc_readdir(struct file *, struct dir_context *);
244 int proc_readdir_de(struct file *, struct dir_context *, struct proc_dir_entry *);
245
pde_get(struct proc_dir_entry * pde)246 static inline void pde_get(struct proc_dir_entry *pde)
247 {
248 refcount_inc(&pde->refcnt);
249 }
250 extern void pde_put(struct proc_dir_entry *);
251
is_empty_pde(const struct proc_dir_entry * pde)252 static inline bool is_empty_pde(const struct proc_dir_entry *pde)
253 {
254 return S_ISDIR(pde->mode) && !pde->proc_iops;
255 }
256 extern ssize_t proc_simple_write(struct file *, const char __user *, size_t, loff_t *);
257
258 /*
259 * inode.c
260 */
261 struct pde_opener {
262 struct list_head lh;
263 struct file *file;
264 bool closing;
265 struct completion *c;
266 } __randomize_layout;
267 extern const struct inode_operations proc_link_inode_operations;
268 extern const struct inode_operations proc_pid_link_inode_operations;
269 extern const struct super_operations proc_sops;
270
271 void proc_init_kmemcache(void);
272 void proc_invalidate_siblings_dcache(struct hlist_head *inodes, spinlock_t *lock);
273 void set_proc_pid_nlink(void);
274 extern struct inode *proc_get_inode(struct super_block *, struct proc_dir_entry *);
275 extern void proc_entry_rundown(struct proc_dir_entry *);
276
277 /*
278 * proc_namespaces.c
279 */
280 extern const struct inode_operations proc_ns_dir_inode_operations;
281 extern const struct file_operations proc_ns_dir_operations;
282
283 /*
284 * proc_net.c
285 */
286 extern const struct file_operations proc_net_operations;
287 extern const struct inode_operations proc_net_inode_operations;
288
289 #ifdef CONFIG_NET
290 extern int proc_net_init(void);
291 #else
proc_net_init(void)292 static inline int proc_net_init(void) { return 0; }
293 #endif
294
295 /*
296 * proc_self.c
297 */
298 extern int proc_setup_self(struct super_block *);
299
300 /*
301 * proc_thread_self.c
302 */
303 extern int proc_setup_thread_self(struct super_block *);
304 extern void proc_thread_self_init(void);
305
306 /*
307 * proc_sysctl.c
308 */
309 #ifdef CONFIG_PROC_SYSCTL
310 extern int proc_sys_init(void);
311 extern void proc_sys_evict_inode(struct inode *inode,
312 struct ctl_table_header *head);
313 #else
proc_sys_init(void)314 static inline void proc_sys_init(void) { }
proc_sys_evict_inode(struct inode * inode,struct ctl_table_header * head)315 static inline void proc_sys_evict_inode(struct inode *inode,
316 struct ctl_table_header *head) { }
317 #endif
318
319 /*
320 * proc_tty.c
321 */
322 #ifdef CONFIG_TTY
323 extern void proc_tty_init(void);
324 #else
proc_tty_init(void)325 static inline void proc_tty_init(void) {}
326 #endif
327
328 /*
329 * root.c
330 */
331 extern struct proc_dir_entry proc_root;
332
333 extern void proc_self_init(void);
334
335 /*
336 * task_[no]mmu.c
337 */
338 struct mem_size_stats;
339 struct proc_maps_private {
340 struct inode *inode;
341 struct task_struct *task;
342 struct mm_struct *mm;
343 struct vma_iterator iter;
344 loff_t last_pos;
345 #ifdef CONFIG_PER_VMA_LOCK
346 bool mmap_locked;
347 struct vm_area_struct *locked_vma;
348 #endif
349 #ifdef CONFIG_NUMA
350 struct mempolicy *task_mempolicy;
351 #endif
352 } __randomize_layout;
353
354 struct mm_struct *proc_mem_open(struct inode *inode, unsigned int mode);
355
356 extern const struct file_operations proc_pid_maps_operations;
357 extern const struct file_operations proc_pid_numa_maps_operations;
358 extern const struct file_operations proc_pid_smaps_operations;
359 extern const struct file_operations proc_pid_smaps_rollup_operations;
360 extern const struct file_operations proc_clear_refs_operations;
361 extern const struct file_operations proc_pagemap_operations;
362
363 extern unsigned long task_vsize(struct mm_struct *);
364 extern unsigned long task_statm(struct mm_struct *,
365 unsigned long *, unsigned long *,
366 unsigned long *, unsigned long *);
367 extern void task_mem(struct seq_file *, struct mm_struct *);
368
369 extern const struct dentry_operations proc_net_dentry_ops;
pde_force_lookup(struct proc_dir_entry * pde)370 static inline void pde_force_lookup(struct proc_dir_entry *pde)
371 {
372 /* /proc/net/ entries can be changed under us by setns(CLONE_NEWNET) */
373 pde->proc_dops = &proc_net_dentry_ops;
374 }
375
376 /*
377 * Add a new procfs dentry that can't serve as a mountpoint. That should
378 * encompass anything that is ephemeral and can just disappear while the
379 * process is still around.
380 */
proc_splice_unmountable(struct inode * inode,struct dentry * dentry,const struct dentry_operations * d_ops)381 static inline struct dentry *proc_splice_unmountable(struct inode *inode,
382 struct dentry *dentry, const struct dentry_operations *d_ops)
383 {
384 d_set_d_op(dentry, d_ops);
385 dont_mount(dentry);
386 return d_splice_alias(inode, dentry);
387 }
388