Commit | Line | Data |
---|---|---|
b2441318 | 1 | // SPDX-License-Identifier: GPL-2.0 |
1da177e4 LT |
2 | /* |
3 | * linux/fs/proc/base.c | |
4 | * | |
5 | * Copyright (C) 1991, 1992 Linus Torvalds | |
6 | * | |
7 | * proc base directory handling functions | |
8 | * | |
9 | * 1999, Al Viro. Rewritten. Now it covers the whole per-process part. | |
10 | * Instead of using magical inumbers to determine the kind of object | |
11 | * we allocate and fill in-core inodes upon lookup. They don't even | |
12 | * go into icache. We cache the reference to task_struct upon lookup too. | |
13 | * Eventually it should become a filesystem in its own. We don't use the | |
14 | * rest of procfs anymore. | |
e070ad49 ML |
15 | * |
16 | * | |
17 | * Changelog: | |
18 | * 17-Jan-2005 | |
19 | * Allan Bezerra | |
20 | * Bruna Moreira <bruna.moreira@indt.org.br> | |
21 | * Edjard Mota <edjard.mota@indt.org.br> | |
22 | * Ilias Biris <ilias.biris@indt.org.br> | |
23 | * Mauricio Lin <mauricio.lin@indt.org.br> | |
24 | * | |
25 | * Embedded Linux Lab - 10LE Instituto Nokia de Tecnologia - INdT | |
26 | * | |
27 | * A new process specific entry (smaps) included in /proc. It shows the | |
28 | * size of rss for each memory area. The maps entry lacks information | |
29 | * about physical memory size (rss) for each mapped file, i.e., | |
30 | * rss information for executables and library files. | |
31 | * This additional information is useful for any tools that need to know | |
32 | * about physical memory consumption for a process specific library. | |
33 | * | |
34 | * Changelog: | |
35 | * 21-Feb-2005 | |
36 | * Embedded Linux Lab - 10LE Instituto Nokia de Tecnologia - INdT | |
37 | * Pud inclusion in the page table walking. | |
38 | * | |
39 | * ChangeLog: | |
40 | * 10-Mar-2005 | |
41 | * 10LE Instituto Nokia de Tecnologia - INdT: | |
42 | * A better way to walks through the page table as suggested by Hugh Dickins. | |
43 | * | |
44 | * Simo Piiroinen <simo.piiroinen@nokia.com>: | |
45 | * Smaps information related to shared, private, clean and dirty pages. | |
46 | * | |
47 | * Paul Mundt <paul.mundt@nokia.com>: | |
48 | * Overall revision about smaps. | |
1da177e4 LT |
49 | */ |
50 | ||
7c0f6ba6 | 51 | #include <linux/uaccess.h> |
1da177e4 | 52 | |
1da177e4 LT |
53 | #include <linux/errno.h> |
54 | #include <linux/time.h> | |
55 | #include <linux/proc_fs.h> | |
56 | #include <linux/stat.h> | |
5995477a | 57 | #include <linux/task_io_accounting_ops.h> |
1da177e4 | 58 | #include <linux/init.h> |
16f7e0fe | 59 | #include <linux/capability.h> |
1da177e4 | 60 | #include <linux/file.h> |
9f3acc31 | 61 | #include <linux/fdtable.h> |
94f8f3b0 | 62 | #include <linux/generic-radix-tree.h> |
1da177e4 LT |
63 | #include <linux/string.h> |
64 | #include <linux/seq_file.h> | |
65 | #include <linux/namei.h> | |
6b3286ed | 66 | #include <linux/mnt_namespace.h> |
1da177e4 | 67 | #include <linux/mm.h> |
a63d83f4 | 68 | #include <linux/swap.h> |
b835996f | 69 | #include <linux/rcupdate.h> |
54354c6a | 70 | #include <linux/kallsyms.h> |
2ec220e2 | 71 | #include <linux/stacktrace.h> |
d85f50d5 | 72 | #include <linux/resource.h> |
5096add8 | 73 | #include <linux/module.h> |
1da177e4 LT |
74 | #include <linux/mount.h> |
75 | #include <linux/security.h> | |
76 | #include <linux/ptrace.h> | |
87ebdc00 | 77 | #include <linux/printk.h> |
efb1a57d | 78 | #include <linux/cache.h> |
a424316c | 79 | #include <linux/cgroup.h> |
1da177e4 LT |
80 | #include <linux/cpuset.h> |
81 | #include <linux/audit.h> | |
5addc5dd | 82 | #include <linux/poll.h> |
1651e14e | 83 | #include <linux/nsproxy.h> |
8ac773b4 | 84 | #include <linux/oom.h> |
3cb4a0bb | 85 | #include <linux/elf.h> |
60347f67 | 86 | #include <linux/pid_namespace.h> |
22d917d8 | 87 | #include <linux/user_namespace.h> |
5ad4e53b | 88 | #include <linux/fs_struct.h> |
5a0e3ad6 | 89 | #include <linux/slab.h> |
4eb5aaa3 | 90 | #include <linux/sched/autogroup.h> |
6e84f315 | 91 | #include <linux/sched/mm.h> |
f7ccbae4 | 92 | #include <linux/sched/coredump.h> |
b17b0153 | 93 | #include <linux/sched/debug.h> |
3905f9ad | 94 | #include <linux/sched/stat.h> |
48f6a7a5 | 95 | #include <linux/posix-timers.h> |
04a8682a | 96 | #include <linux/time_namespace.h> |
e79f15a4 | 97 | #include <linux/resctrl.h> |
c2f273eb | 98 | #include <linux/cn_proc.h> |
d21077fb | 99 | #include <linux/ksm.h> |
267c068e | 100 | #include <uapi/linux/lsm.h> |
43d2b113 | 101 | #include <trace/events/oom.h> |
1da177e4 | 102 | #include "internal.h" |
faf60af1 | 103 | #include "fd.h" |
1da177e4 | 104 | |
ac7f1061 AD |
105 | #include "../../lib/kstrtox.h" |
106 | ||
0f2fe20f EB |
107 | /* NOTE: |
108 | * Implementing inode permission operations in /proc is almost | |
109 | * certainly an error. Permission checks need to happen during | |
110 | * each system call not at open time. The reason is that most of | |
111 | * what we wish to check for permissions in /proc varies at runtime. | |
112 | * | |
113 | * The classic example of a problem is opening file descriptors | |
114 | * in /proc for a task before it execs a suid executable. | |
115 | */ | |
116 | ||
efb1a57d AD |
117 | static u8 nlink_tid __ro_after_init; |
118 | static u8 nlink_tgid __ro_after_init; | |
1270dd8d | 119 | |
1da177e4 | 120 | struct pid_entry { |
cedbccab | 121 | const char *name; |
623f594e | 122 | unsigned int len; |
d161a13f | 123 | umode_t mode; |
c5ef1c42 | 124 | const struct inode_operations *iop; |
00977a59 | 125 | const struct file_operations *fop; |
20cdc894 | 126 | union proc_op op; |
1da177e4 LT |
127 | }; |
128 | ||
61a28784 | 129 | #define NOD(NAME, MODE, IOP, FOP, OP) { \ |
20cdc894 | 130 | .name = (NAME), \ |
c5141e6d | 131 | .len = sizeof(NAME) - 1, \ |
20cdc894 EB |
132 | .mode = MODE, \ |
133 | .iop = IOP, \ | |
134 | .fop = FOP, \ | |
135 | .op = OP, \ | |
136 | } | |
137 | ||
631f9c18 AD |
138 | #define DIR(NAME, MODE, iops, fops) \ |
139 | NOD(NAME, (S_IFDIR|(MODE)), &iops, &fops, {} ) | |
140 | #define LNK(NAME, get_link) \ | |
61a28784 | 141 | NOD(NAME, (S_IFLNK|S_IRWXUGO), \ |
20cdc894 | 142 | &proc_pid_link_inode_operations, NULL, \ |
631f9c18 AD |
143 | { .proc_get_link = get_link } ) |
144 | #define REG(NAME, MODE, fops) \ | |
145 | NOD(NAME, (S_IFREG|(MODE)), NULL, &fops, {}) | |
631f9c18 | 146 | #define ONE(NAME, MODE, show) \ |
6d9c939d | 147 | NOD(NAME, (S_IFREG|(MODE)), \ |
be614086 | 148 | NULL, &proc_single_file_operations, \ |
631f9c18 | 149 | { .proc_show = show } ) |
267c068e | 150 | #define ATTR(LSMID, NAME, MODE) \ |
6d9c939d CS |
151 | NOD(NAME, (S_IFREG|(MODE)), \ |
152 | NULL, &proc_pid_attr_operations, \ | |
267c068e | 153 | { .lsmid = LSMID }) |
1da177e4 | 154 | |
aed54175 VN |
155 | /* |
156 | * Count the number of hardlinks for the pid_entry table, excluding the . | |
157 | * and .. links. | |
158 | */ | |
1270dd8d | 159 | static unsigned int __init pid_entry_nlink(const struct pid_entry *entries, |
aed54175 VN |
160 | unsigned int n) |
161 | { | |
162 | unsigned int i; | |
163 | unsigned int count; | |
164 | ||
1270dd8d | 165 | count = 2; |
aed54175 VN |
166 | for (i = 0; i < n; ++i) { |
167 | if (S_ISDIR(entries[i].mode)) | |
168 | ++count; | |
169 | } | |
170 | ||
171 | return count; | |
172 | } | |
173 | ||
f7ad3c6b | 174 | static int get_task_root(struct task_struct *task, struct path *root) |
1da177e4 | 175 | { |
7c2c7d99 HD |
176 | int result = -ENOENT; |
177 | ||
0494f6ec | 178 | task_lock(task); |
f7ad3c6b MS |
179 | if (task->fs) { |
180 | get_fs_root(task->fs, root); | |
7c2c7d99 HD |
181 | result = 0; |
182 | } | |
0494f6ec | 183 | task_unlock(task); |
7c2c7d99 | 184 | return result; |
0494f6ec MS |
185 | } |
186 | ||
7773fbc5 | 187 | static int proc_cwd_link(struct dentry *dentry, struct path *path) |
0494f6ec | 188 | { |
2b0143b5 | 189 | struct task_struct *task = get_proc_task(d_inode(dentry)); |
0494f6ec | 190 | int result = -ENOENT; |
99f89551 EB |
191 | |
192 | if (task) { | |
f7ad3c6b MS |
193 | task_lock(task); |
194 | if (task->fs) { | |
195 | get_fs_pwd(task->fs, path); | |
196 | result = 0; | |
197 | } | |
198 | task_unlock(task); | |
99f89551 EB |
199 | put_task_struct(task); |
200 | } | |
1da177e4 LT |
201 | return result; |
202 | } | |
203 | ||
7773fbc5 | 204 | static int proc_root_link(struct dentry *dentry, struct path *path) |
1da177e4 | 205 | { |
2b0143b5 | 206 | struct task_struct *task = get_proc_task(d_inode(dentry)); |
1da177e4 | 207 | int result = -ENOENT; |
99f89551 EB |
208 | |
209 | if (task) { | |
f7ad3c6b | 210 | result = get_task_root(task, path); |
99f89551 EB |
211 | put_task_struct(task); |
212 | } | |
1da177e4 LT |
213 | return result; |
214 | } | |
215 | ||
d26d0cd9 LT |
216 | /* |
217 | * If the user used setproctitle(), we just get the string from | |
218 | * user space at arg_start, and limit it to a maximum of one page. | |
219 | */ | |
220 | static ssize_t get_mm_proctitle(struct mm_struct *mm, char __user *buf, | |
221 | size_t count, unsigned long pos, | |
222 | unsigned long arg_start) | |
223 | { | |
224 | char *page; | |
225 | int ret, got; | |
226 | ||
227 | if (pos >= PAGE_SIZE) | |
228 | return 0; | |
229 | ||
230 | page = (char *)__get_free_page(GFP_KERNEL); | |
231 | if (!page) | |
232 | return -ENOMEM; | |
233 | ||
234 | ret = 0; | |
235 | got = access_remote_vm(mm, arg_start, page, PAGE_SIZE, FOLL_ANON); | |
236 | if (got > 0) { | |
237 | int len = strnlen(page, got); | |
238 | ||
239 | /* Include the NUL character if it was found */ | |
240 | if (len < got) | |
241 | len++; | |
242 | ||
243 | if (len > pos) { | |
244 | len -= pos; | |
245 | if (len > count) | |
246 | len = count; | |
247 | len -= copy_to_user(buf, page+pos, len); | |
248 | if (!len) | |
249 | len = -EFAULT; | |
250 | ret = len; | |
251 | } | |
252 | } | |
253 | free_page((unsigned long)page); | |
254 | return ret; | |
255 | } | |
256 | ||
e4b4e441 | 257 | static ssize_t get_mm_cmdline(struct mm_struct *mm, char __user *buf, |
5ab82718 | 258 | size_t count, loff_t *ppos) |
1da177e4 | 259 | { |
c2c0bb44 | 260 | unsigned long arg_start, arg_end, env_start, env_end; |
5ab82718 | 261 | unsigned long pos, len; |
d26d0cd9 | 262 | char *page, c; |
c2c0bb44 | 263 | |
c2c0bb44 | 264 | /* Check if process spawned far enough to have cmdline. */ |
e4b4e441 LT |
265 | if (!mm->env_end) |
266 | return 0; | |
c2c0bb44 | 267 | |
88aa7cc6 | 268 | spin_lock(&mm->arg_lock); |
c2c0bb44 AD |
269 | arg_start = mm->arg_start; |
270 | arg_end = mm->arg_end; | |
271 | env_start = mm->env_start; | |
272 | env_end = mm->env_end; | |
88aa7cc6 | 273 | spin_unlock(&mm->arg_lock); |
c2c0bb44 | 274 | |
5ab82718 LT |
275 | if (arg_start >= arg_end) |
276 | return 0; | |
6a6cbe75 | 277 | |
2ca66ff7 | 278 | /* |
d26d0cd9 LT |
279 | * We allow setproctitle() to overwrite the argument |
280 | * strings, and overflow past the original end. But | |
281 | * only when it overflows into the environment area. | |
2ca66ff7 | 282 | */ |
d26d0cd9 | 283 | if (env_start != arg_end || env_end < env_start) |
5ab82718 | 284 | env_start = env_end = arg_end; |
d26d0cd9 | 285 | len = env_end - arg_start; |
f5b65348 | 286 | |
5ab82718 | 287 | /* We're not going to care if "*ppos" has high bits set */ |
d26d0cd9 LT |
288 | pos = *ppos; |
289 | if (pos >= len) | |
5ab82718 | 290 | return 0; |
d26d0cd9 LT |
291 | if (count > len - pos) |
292 | count = len - pos; | |
293 | if (!count) | |
294 | return 0; | |
295 | ||
296 | /* | |
297 | * Magical special case: if the argv[] end byte is not | |
298 | * zero, the user has overwritten it with setproctitle(3). | |
299 | * | |
300 | * Possible future enhancement: do this only once when | |
301 | * pos is 0, and set a flag in the 'struct file'. | |
302 | */ | |
303 | if (access_remote_vm(mm, arg_end-1, &c, 1, FOLL_ANON) == 1 && c) | |
304 | return get_mm_proctitle(mm, buf, count, pos, arg_start); | |
a0a07b87 | 305 | |
d26d0cd9 LT |
306 | /* |
307 | * For the non-setproctitle() case we limit things strictly | |
308 | * to the [arg_start, arg_end[ range. | |
309 | */ | |
310 | pos += arg_start; | |
3d712546 | 311 | if (pos < arg_start || pos >= arg_end) |
5ab82718 | 312 | return 0; |
3d712546 LT |
313 | if (count > arg_end - pos) |
314 | count = arg_end - pos; | |
5ab82718 LT |
315 | |
316 | page = (char *)__get_free_page(GFP_KERNEL); | |
317 | if (!page) | |
318 | return -ENOMEM; | |
319 | ||
320 | len = 0; | |
321 | while (count) { | |
322 | int got; | |
323 | size_t size = min_t(size_t, PAGE_SIZE, count); | |
324 | ||
3d712546 LT |
325 | got = access_remote_vm(mm, pos, page, size, FOLL_ANON); |
326 | if (got <= 0) | |
5ab82718 | 327 | break; |
3d712546 | 328 | got -= copy_to_user(buf, page, got); |
5ab82718 LT |
329 | if (unlikely(!got)) { |
330 | if (!len) | |
331 | len = -EFAULT; | |
332 | break; | |
c2c0bb44 | 333 | } |
5ab82718 LT |
334 | pos += got; |
335 | buf += got; | |
336 | len += got; | |
337 | count -= got; | |
c2c0bb44 AD |
338 | } |
339 | ||
c2c0bb44 | 340 | free_page((unsigned long)page); |
5ab82718 | 341 | return len; |
1da177e4 LT |
342 | } |
343 | ||
e4b4e441 LT |
344 | static ssize_t get_task_cmdline(struct task_struct *tsk, char __user *buf, |
345 | size_t count, loff_t *pos) | |
346 | { | |
347 | struct mm_struct *mm; | |
348 | ssize_t ret; | |
349 | ||
350 | mm = get_task_mm(tsk); | |
351 | if (!mm) | |
352 | return 0; | |
353 | ||
354 | ret = get_mm_cmdline(mm, buf, count, pos); | |
c2c0bb44 | 355 | mmput(mm); |
e4b4e441 LT |
356 | return ret; |
357 | } | |
358 | ||
359 | static ssize_t proc_pid_cmdline_read(struct file *file, char __user *buf, | |
360 | size_t count, loff_t *pos) | |
361 | { | |
362 | struct task_struct *tsk; | |
363 | ssize_t ret; | |
364 | ||
365 | BUG_ON(*pos < 0); | |
366 | ||
367 | tsk = get_proc_task(file_inode(file)); | |
368 | if (!tsk) | |
369 | return -ESRCH; | |
370 | ret = get_task_cmdline(tsk, buf, count, pos); | |
371 | put_task_struct(tsk); | |
372 | if (ret > 0) | |
373 | *pos += ret; | |
374 | return ret; | |
1da177e4 LT |
375 | } |
376 | ||
c2c0bb44 AD |
377 | static const struct file_operations proc_pid_cmdline_ops = { |
378 | .read = proc_pid_cmdline_read, | |
379 | .llseek = generic_file_llseek, | |
380 | }; | |
381 | ||
1da177e4 LT |
382 | #ifdef CONFIG_KALLSYMS |
383 | /* | |
384 | * Provides a wchan file via kallsyms in a proper one-value-per-file format. | |
385 | * Returns the resolved symbol. If that fails, simply return the address. | |
386 | */ | |
edfcd606 AD |
387 | static int proc_pid_wchan(struct seq_file *m, struct pid_namespace *ns, |
388 | struct pid *pid, struct task_struct *task) | |
1da177e4 | 389 | { |
ffb45122 | 390 | unsigned long wchan; |
54354c6a | 391 | char symname[KSYM_NAME_LEN]; |
1da177e4 | 392 | |
54354c6a KC |
393 | if (!ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS)) |
394 | goto print0; | |
1da177e4 | 395 | |
54354c6a KC |
396 | wchan = get_wchan(task); |
397 | if (wchan && !lookup_symbol_name(wchan, symname)) { | |
398 | seq_puts(m, symname); | |
399 | return 0; | |
400 | } | |
25ce3191 | 401 | |
54354c6a KC |
402 | print0: |
403 | seq_putc(m, '0'); | |
25ce3191 | 404 | return 0; |
1da177e4 LT |
405 | } |
406 | #endif /* CONFIG_KALLSYMS */ | |
407 | ||
a9712bc1 AV |
408 | static int lock_trace(struct task_struct *task) |
409 | { | |
f7cfd871 | 410 | int err = down_read_killable(&task->signal->exec_update_lock); |
a9712bc1 AV |
411 | if (err) |
412 | return err; | |
caaee623 | 413 | if (!ptrace_may_access(task, PTRACE_MODE_ATTACH_FSCREDS)) { |
f7cfd871 | 414 | up_read(&task->signal->exec_update_lock); |
a9712bc1 AV |
415 | return -EPERM; |
416 | } | |
417 | return 0; | |
418 | } | |
419 | ||
420 | static void unlock_trace(struct task_struct *task) | |
421 | { | |
f7cfd871 | 422 | up_read(&task->signal->exec_update_lock); |
a9712bc1 AV |
423 | } |
424 | ||
2ec220e2 KC |
425 | #ifdef CONFIG_STACKTRACE |
426 | ||
427 | #define MAX_STACK_TRACE_DEPTH 64 | |
428 | ||
429 | static int proc_pid_stack(struct seq_file *m, struct pid_namespace *ns, | |
430 | struct pid *pid, struct task_struct *task) | |
431 | { | |
2ec220e2 | 432 | unsigned long *entries; |
a9712bc1 | 433 | int err; |
2ec220e2 | 434 | |
f8a00cef JH |
435 | /* |
436 | * The ability to racily run the kernel stack unwinder on a running task | |
437 | * and then observe the unwinder output is scary; while it is useful for | |
438 | * debugging kernel issues, it can also allow an attacker to leak kernel | |
439 | * stack contents. | |
440 | * Doing this in a manner that is at least safe from races would require | |
441 | * some work to ensure that the remote task can not be scheduled; and | |
442 | * even then, this would still expose the unwinder as local attack | |
443 | * surface. | |
444 | * Therefore, this interface is restricted to root. | |
445 | */ | |
446 | if (!file_ns_capable(m->file, &init_user_ns, CAP_SYS_ADMIN)) | |
447 | return -EACCES; | |
448 | ||
6da2ec56 KC |
449 | entries = kmalloc_array(MAX_STACK_TRACE_DEPTH, sizeof(*entries), |
450 | GFP_KERNEL); | |
2ec220e2 KC |
451 | if (!entries) |
452 | return -ENOMEM; | |
453 | ||
a9712bc1 AV |
454 | err = lock_trace(task); |
455 | if (!err) { | |
e988e5ec | 456 | unsigned int i, nr_entries; |
5d008fb4 | 457 | |
e988e5ec TG |
458 | nr_entries = stack_trace_save_tsk(task, entries, |
459 | MAX_STACK_TRACE_DEPTH, 0); | |
a9712bc1 | 460 | |
e988e5ec | 461 | for (i = 0; i < nr_entries; i++) { |
8f5abe84 | 462 | seq_printf(m, "[<0>] %pB\n", (void *)entries[i]); |
a9712bc1 | 463 | } |
e988e5ec | 464 | |
a9712bc1 | 465 | unlock_trace(task); |
2ec220e2 KC |
466 | } |
467 | kfree(entries); | |
468 | ||
a9712bc1 | 469 | return err; |
2ec220e2 KC |
470 | } |
471 | #endif | |
472 | ||
5968cece | 473 | #ifdef CONFIG_SCHED_INFO |
1da177e4 LT |
474 | /* |
475 | * Provides /proc/PID/schedstat | |
476 | */ | |
f6e826ca AD |
477 | static int proc_pid_schedstat(struct seq_file *m, struct pid_namespace *ns, |
478 | struct pid *pid, struct task_struct *task) | |
1da177e4 | 479 | { |
5968cece | 480 | if (unlikely(!sched_info_on())) |
08b55775 | 481 | seq_puts(m, "0 0 0\n"); |
5968cece NR |
482 | else |
483 | seq_printf(m, "%llu %llu %lu\n", | |
25ce3191 JP |
484 | (unsigned long long)task->se.sum_exec_runtime, |
485 | (unsigned long long)task->sched_info.run_delay, | |
486 | task->sched_info.pcount); | |
487 | ||
488 | return 0; | |
1da177e4 LT |
489 | } |
490 | #endif | |
491 | ||
9745512c AV |
492 | #ifdef CONFIG_LATENCYTOP |
493 | static int lstats_show_proc(struct seq_file *m, void *v) | |
494 | { | |
495 | int i; | |
13d77c37 HS |
496 | struct inode *inode = m->private; |
497 | struct task_struct *task = get_proc_task(inode); | |
9745512c | 498 | |
13d77c37 HS |
499 | if (!task) |
500 | return -ESRCH; | |
501 | seq_puts(m, "Latency Top version : v0.1\n"); | |
f6d2f584 | 502 | for (i = 0; i < LT_SAVECOUNT; i++) { |
34e49d4f JP |
503 | struct latency_record *lr = &task->latency_record[i]; |
504 | if (lr->backtrace[0]) { | |
9745512c | 505 | int q; |
34e49d4f JP |
506 | seq_printf(m, "%i %li %li", |
507 | lr->count, lr->time, lr->max); | |
9745512c | 508 | for (q = 0; q < LT_BACKTRACEDEPTH; q++) { |
34e49d4f | 509 | unsigned long bt = lr->backtrace[q]; |
accddc41 | 510 | |
34e49d4f | 511 | if (!bt) |
9745512c | 512 | break; |
34e49d4f | 513 | seq_printf(m, " %ps", (void *)bt); |
9745512c | 514 | } |
9d6de12f | 515 | seq_putc(m, '\n'); |
9745512c AV |
516 | } |
517 | ||
518 | } | |
13d77c37 | 519 | put_task_struct(task); |
9745512c AV |
520 | return 0; |
521 | } | |
522 | ||
523 | static int lstats_open(struct inode *inode, struct file *file) | |
524 | { | |
13d77c37 | 525 | return single_open(file, lstats_show_proc, inode); |
d6643d12 HS |
526 | } |
527 | ||
9745512c AV |
528 | static ssize_t lstats_write(struct file *file, const char __user *buf, |
529 | size_t count, loff_t *offs) | |
530 | { | |
496ad9aa | 531 | struct task_struct *task = get_proc_task(file_inode(file)); |
9745512c | 532 | |
13d77c37 HS |
533 | if (!task) |
534 | return -ESRCH; | |
e02c9b0d | 535 | clear_tsk_latency_tracing(task); |
13d77c37 | 536 | put_task_struct(task); |
9745512c AV |
537 | |
538 | return count; | |
539 | } | |
540 | ||
541 | static const struct file_operations proc_lstats_operations = { | |
542 | .open = lstats_open, | |
543 | .read = seq_read, | |
544 | .write = lstats_write, | |
545 | .llseek = seq_lseek, | |
13d77c37 | 546 | .release = single_release, |
9745512c AV |
547 | }; |
548 | ||
549 | #endif | |
550 | ||
6ba51e37 AD |
551 | static int proc_oom_score(struct seq_file *m, struct pid_namespace *ns, |
552 | struct pid *pid, struct task_struct *task) | |
1da177e4 | 553 | { |
ca79b0c2 | 554 | unsigned long totalpages = totalram_pages() + total_swap_pages; |
b95c35e7 | 555 | unsigned long points = 0; |
9066e5cf YS |
556 | long badness; |
557 | ||
558 | badness = oom_badness(task, totalpages); | |
559 | /* | |
560 | * Special case OOM_SCORE_ADJ_MIN for all others scale the | |
561 | * badness value into [0, 2000] range which we have been | |
562 | * exporting for a long time so userspace might depend on it. | |
563 | */ | |
564 | if (badness != LONG_MIN) | |
565 | points = (1000 + badness * 1000 / (long)totalpages) * 2 / 3; | |
1da177e4 | 566 | |
25ce3191 JP |
567 | seq_printf(m, "%lu\n", points); |
568 | ||
569 | return 0; | |
1da177e4 LT |
570 | } |
571 | ||
d85f50d5 | 572 | struct limit_names { |
cedbccab AD |
573 | const char *name; |
574 | const char *unit; | |
d85f50d5 NH |
575 | }; |
576 | ||
577 | static const struct limit_names lnames[RLIM_NLIMITS] = { | |
cff4edb5 | 578 | [RLIMIT_CPU] = {"Max cpu time", "seconds"}, |
d85f50d5 NH |
579 | [RLIMIT_FSIZE] = {"Max file size", "bytes"}, |
580 | [RLIMIT_DATA] = {"Max data size", "bytes"}, | |
581 | [RLIMIT_STACK] = {"Max stack size", "bytes"}, | |
582 | [RLIMIT_CORE] = {"Max core file size", "bytes"}, | |
583 | [RLIMIT_RSS] = {"Max resident set", "bytes"}, | |
584 | [RLIMIT_NPROC] = {"Max processes", "processes"}, | |
585 | [RLIMIT_NOFILE] = {"Max open files", "files"}, | |
586 | [RLIMIT_MEMLOCK] = {"Max locked memory", "bytes"}, | |
587 | [RLIMIT_AS] = {"Max address space", "bytes"}, | |
588 | [RLIMIT_LOCKS] = {"Max file locks", "locks"}, | |
589 | [RLIMIT_SIGPENDING] = {"Max pending signals", "signals"}, | |
590 | [RLIMIT_MSGQUEUE] = {"Max msgqueue size", "bytes"}, | |
591 | [RLIMIT_NICE] = {"Max nice priority", NULL}, | |
592 | [RLIMIT_RTPRIO] = {"Max realtime priority", NULL}, | |
8808117c | 593 | [RLIMIT_RTTIME] = {"Max realtime timeout", "us"}, |
d85f50d5 NH |
594 | }; |
595 | ||
596 | /* Display limits for a process */ | |
1c963eb1 AD |
597 | static int proc_pid_limits(struct seq_file *m, struct pid_namespace *ns, |
598 | struct pid *pid, struct task_struct *task) | |
d85f50d5 NH |
599 | { |
600 | unsigned int i; | |
d85f50d5 | 601 | unsigned long flags; |
d85f50d5 NH |
602 | |
603 | struct rlimit rlim[RLIM_NLIMITS]; | |
604 | ||
a6bebbc8 | 605 | if (!lock_task_sighand(task, &flags)) |
d85f50d5 | 606 | return 0; |
d85f50d5 NH |
607 | memcpy(rlim, task->signal->rlim, sizeof(struct rlimit) * RLIM_NLIMITS); |
608 | unlock_task_sighand(task, &flags); | |
d85f50d5 NH |
609 | |
610 | /* | |
611 | * print the file header | |
612 | */ | |
afe922c2 AD |
613 | seq_puts(m, "Limit " |
614 | "Soft Limit " | |
615 | "Hard Limit " | |
616 | "Units \n"); | |
d85f50d5 NH |
617 | |
618 | for (i = 0; i < RLIM_NLIMITS; i++) { | |
619 | if (rlim[i].rlim_cur == RLIM_INFINITY) | |
1c963eb1 | 620 | seq_printf(m, "%-25s %-20s ", |
25ce3191 | 621 | lnames[i].name, "unlimited"); |
d85f50d5 | 622 | else |
1c963eb1 | 623 | seq_printf(m, "%-25s %-20lu ", |
25ce3191 | 624 | lnames[i].name, rlim[i].rlim_cur); |
d85f50d5 NH |
625 | |
626 | if (rlim[i].rlim_max == RLIM_INFINITY) | |
1c963eb1 | 627 | seq_printf(m, "%-20s ", "unlimited"); |
d85f50d5 | 628 | else |
1c963eb1 | 629 | seq_printf(m, "%-20lu ", rlim[i].rlim_max); |
d85f50d5 NH |
630 | |
631 | if (lnames[i].unit) | |
1c963eb1 | 632 | seq_printf(m, "%-10s\n", lnames[i].unit); |
d85f50d5 | 633 | else |
1c963eb1 | 634 | seq_putc(m, '\n'); |
d85f50d5 NH |
635 | } |
636 | ||
1c963eb1 | 637 | return 0; |
d85f50d5 NH |
638 | } |
639 | ||
ebcb6734 | 640 | #ifdef CONFIG_HAVE_ARCH_TRACEHOOK |
09d93bd6 AD |
641 | static int proc_pid_syscall(struct seq_file *m, struct pid_namespace *ns, |
642 | struct pid *pid, struct task_struct *task) | |
ebcb6734 | 643 | { |
631b7aba SRRH |
644 | struct syscall_info info; |
645 | u64 *args = &info.data.args[0]; | |
25ce3191 JP |
646 | int res; |
647 | ||
648 | res = lock_trace(task); | |
a9712bc1 AV |
649 | if (res) |
650 | return res; | |
ebcb6734 | 651 | |
631b7aba | 652 | if (task_current_syscall(task, &info)) |
09d93bd6 | 653 | seq_puts(m, "running\n"); |
631b7aba SRRH |
654 | else if (info.data.nr < 0) |
655 | seq_printf(m, "%d 0x%llx 0x%llx\n", | |
656 | info.data.nr, info.sp, info.data.instruction_pointer); | |
a9712bc1 | 657 | else |
09d93bd6 | 658 | seq_printf(m, |
631b7aba SRRH |
659 | "%d 0x%llx 0x%llx 0x%llx 0x%llx 0x%llx 0x%llx 0x%llx 0x%llx\n", |
660 | info.data.nr, | |
ebcb6734 | 661 | args[0], args[1], args[2], args[3], args[4], args[5], |
631b7aba | 662 | info.sp, info.data.instruction_pointer); |
a9712bc1 | 663 | unlock_trace(task); |
25ce3191 JP |
664 | |
665 | return 0; | |
ebcb6734 RM |
666 | } |
667 | #endif /* CONFIG_HAVE_ARCH_TRACEHOOK */ | |
668 | ||
1da177e4 LT |
669 | /************************************************************************/ |
670 | /* Here the fs part begins */ | |
671 | /************************************************************************/ | |
672 | ||
673 | /* permission checks */ | |
51a18734 | 674 | static bool proc_fd_access_allowed(struct inode *inode) |
1da177e4 | 675 | { |
778c1144 | 676 | struct task_struct *task; |
51a18734 | 677 | bool allowed = false; |
df26c40e EB |
678 | /* Allow access to a task's file descriptors if it is us or we |
679 | * may use ptrace attach to the process and find out that | |
680 | * information. | |
778c1144 EB |
681 | */ |
682 | task = get_proc_task(inode); | |
df26c40e | 683 | if (task) { |
caaee623 | 684 | allowed = ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS); |
778c1144 | 685 | put_task_struct(task); |
df26c40e | 686 | } |
778c1144 | 687 | return allowed; |
1da177e4 LT |
688 | } |
689 | ||
c1632a0f | 690 | int proc_setattr(struct mnt_idmap *idmap, struct dentry *dentry, |
549c7297 | 691 | struct iattr *attr) |
6d76fa58 LT |
692 | { |
693 | int error; | |
2b0143b5 | 694 | struct inode *inode = d_inode(dentry); |
6d76fa58 LT |
695 | |
696 | if (attr->ia_valid & ATTR_MODE) | |
697 | return -EPERM; | |
698 | ||
c1632a0f | 699 | error = setattr_prepare(&nop_mnt_idmap, dentry, attr); |
1025774c CH |
700 | if (error) |
701 | return error; | |
702 | ||
c1632a0f | 703 | setattr_copy(&nop_mnt_idmap, inode, attr); |
1025774c | 704 | return 0; |
6d76fa58 LT |
705 | } |
706 | ||
0499680a VK |
707 | /* |
708 | * May current process learn task's sched/cmdline info (for hide_pid_min=1) | |
709 | * or euid/egid (for hide_pid_min=2)? | |
710 | */ | |
fa10fed3 | 711 | static bool has_pid_permissions(struct proc_fs_info *fs_info, |
0499680a | 712 | struct task_struct *task, |
e61bb8b3 | 713 | enum proc_hidepid hide_pid_min) |
0499680a | 714 | { |
24a71ce5 AG |
715 | /* |
716 | * If 'hidpid' mount option is set force a ptrace check, | |
717 | * we indicate that we are using a filesystem syscall | |
718 | * by passing PTRACE_MODE_READ_FSCREDS | |
719 | */ | |
720 | if (fs_info->hide_pid == HIDEPID_NOT_PTRACEABLE) | |
721 | return ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS); | |
722 | ||
fa10fed3 | 723 | if (fs_info->hide_pid < hide_pid_min) |
0499680a | 724 | return true; |
fa10fed3 | 725 | if (in_group_p(fs_info->pid_gid)) |
0499680a | 726 | return true; |
caaee623 | 727 | return ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS); |
0499680a VK |
728 | } |
729 | ||
730 | ||
4609e1f1 | 731 | static int proc_pid_permission(struct mnt_idmap *idmap, |
549c7297 | 732 | struct inode *inode, int mask) |
0499680a | 733 | { |
fa10fed3 | 734 | struct proc_fs_info *fs_info = proc_sb_info(inode->i_sb); |
0499680a VK |
735 | struct task_struct *task; |
736 | bool has_perms; | |
737 | ||
738 | task = get_proc_task(inode); | |
a2ef990a XF |
739 | if (!task) |
740 | return -ESRCH; | |
fa10fed3 | 741 | has_perms = has_pid_permissions(fs_info, task, HIDEPID_NO_ACCESS); |
0499680a VK |
742 | put_task_struct(task); |
743 | ||
744 | if (!has_perms) { | |
fa10fed3 | 745 | if (fs_info->hide_pid == HIDEPID_INVISIBLE) { |
0499680a VK |
746 | /* |
747 | * Let's make getdents(), stat(), and open() | |
748 | * consistent with each other. If a process | |
749 | * may not stat() a file, it shouldn't be seen | |
750 | * in procfs at all. | |
751 | */ | |
752 | return -ENOENT; | |
753 | } | |
754 | ||
755 | return -EPERM; | |
756 | } | |
4609e1f1 | 757 | return generic_permission(&nop_mnt_idmap, inode, mask); |
0499680a VK |
758 | } |
759 | ||
760 | ||
761 | ||
c5ef1c42 | 762 | static const struct inode_operations proc_def_inode_operations = { |
6d76fa58 LT |
763 | .setattr = proc_setattr, |
764 | }; | |
765 | ||
be614086 EB |
766 | static int proc_single_show(struct seq_file *m, void *v) |
767 | { | |
768 | struct inode *inode = m->private; | |
9d78edea | 769 | struct pid_namespace *ns = proc_pid_ns(inode->i_sb); |
76f668be | 770 | struct pid *pid = proc_pid(inode); |
be614086 EB |
771 | struct task_struct *task; |
772 | int ret; | |
773 | ||
be614086 EB |
774 | task = get_pid_task(pid, PIDTYPE_PID); |
775 | if (!task) | |
776 | return -ESRCH; | |
777 | ||
778 | ret = PROC_I(inode)->op.proc_show(m, ns, pid, task); | |
779 | ||
780 | put_task_struct(task); | |
781 | return ret; | |
782 | } | |
783 | ||
784 | static int proc_single_open(struct inode *inode, struct file *filp) | |
785 | { | |
c6a34058 | 786 | return single_open(filp, proc_single_show, inode); |
be614086 EB |
787 | } |
788 | ||
789 | static const struct file_operations proc_single_file_operations = { | |
790 | .open = proc_single_open, | |
791 | .read = seq_read, | |
792 | .llseek = seq_lseek, | |
793 | .release = single_release, | |
794 | }; | |
795 | ||
5381e169 ON |
796 | |
797 | struct mm_struct *proc_mem_open(struct inode *inode, unsigned int mode) | |
1da177e4 | 798 | { |
5381e169 ON |
799 | struct task_struct *task = get_proc_task(inode); |
800 | struct mm_struct *mm = ERR_PTR(-ESRCH); | |
e268337d | 801 | |
5381e169 | 802 | if (task) { |
caaee623 | 803 | mm = mm_access(task, mode | PTRACE_MODE_FSCREDS); |
5381e169 | 804 | put_task_struct(task); |
e268337d | 805 | |
5381e169 ON |
806 | if (!IS_ERR_OR_NULL(mm)) { |
807 | /* ensure this mm_struct can't be freed */ | |
f1f10076 | 808 | mmgrab(mm); |
5381e169 ON |
809 | /* but do not pin its memory */ |
810 | mmput(mm); | |
811 | } | |
812 | } | |
813 | ||
814 | return mm; | |
815 | } | |
816 | ||
817 | static int __mem_open(struct inode *inode, struct file *file, unsigned int mode) | |
818 | { | |
819 | struct mm_struct *mm = proc_mem_open(inode, mode); | |
e268337d LT |
820 | |
821 | if (IS_ERR(mm)) | |
822 | return PTR_ERR(mm); | |
823 | ||
e268337d | 824 | file->private_data = mm; |
1da177e4 LT |
825 | return 0; |
826 | } | |
827 | ||
b409e578 CW |
828 | static int mem_open(struct inode *inode, struct file *file) |
829 | { | |
bc452b4b DH |
830 | int ret = __mem_open(inode, file, PTRACE_MODE_ATTACH); |
831 | ||
832 | /* OK to pass negative loff_t, we can catch out-of-range */ | |
833 | file->f_mode |= FMODE_UNSIGNED_OFFSET; | |
834 | ||
835 | return ret; | |
b409e578 CW |
836 | } |
837 | ||
572d34b9 ON |
838 | static ssize_t mem_rw(struct file *file, char __user *buf, |
839 | size_t count, loff_t *ppos, int write) | |
1da177e4 | 840 | { |
e268337d | 841 | struct mm_struct *mm = file->private_data; |
572d34b9 ON |
842 | unsigned long addr = *ppos; |
843 | ssize_t copied; | |
1da177e4 | 844 | char *page; |
272ddc8b | 845 | unsigned int flags; |
1da177e4 | 846 | |
e268337d LT |
847 | if (!mm) |
848 | return 0; | |
99f89551 | 849 | |
0ee931c4 | 850 | page = (char *)__get_free_page(GFP_KERNEL); |
30cd8903 | 851 | if (!page) |
e268337d | 852 | return -ENOMEM; |
1da177e4 | 853 | |
f7ca54f4 | 854 | copied = 0; |
388f7934 | 855 | if (!mmget_not_zero(mm)) |
6d08f2c7 ON |
856 | goto free; |
857 | ||
f511c0b1 | 858 | flags = FOLL_FORCE | (write ? FOLL_WRITE : 0); |
6347e8d5 | 859 | |
1da177e4 | 860 | while (count > 0) { |
d238692b | 861 | size_t this_len = min_t(size_t, count, PAGE_SIZE); |
1da177e4 | 862 | |
572d34b9 | 863 | if (write && copy_from_user(page, buf, this_len)) { |
1da177e4 LT |
864 | copied = -EFAULT; |
865 | break; | |
866 | } | |
572d34b9 | 867 | |
6347e8d5 | 868 | this_len = access_remote_vm(mm, addr, page, this_len, flags); |
572d34b9 | 869 | if (!this_len) { |
1da177e4 LT |
870 | if (!copied) |
871 | copied = -EIO; | |
872 | break; | |
873 | } | |
572d34b9 ON |
874 | |
875 | if (!write && copy_to_user(buf, page, this_len)) { | |
876 | copied = -EFAULT; | |
877 | break; | |
878 | } | |
879 | ||
880 | buf += this_len; | |
881 | addr += this_len; | |
882 | copied += this_len; | |
883 | count -= this_len; | |
1da177e4 | 884 | } |
572d34b9 | 885 | *ppos = addr; |
30cd8903 | 886 | |
6d08f2c7 ON |
887 | mmput(mm); |
888 | free: | |
30cd8903 | 889 | free_page((unsigned long) page); |
1da177e4 LT |
890 | return copied; |
891 | } | |
1da177e4 | 892 | |
572d34b9 ON |
893 | static ssize_t mem_read(struct file *file, char __user *buf, |
894 | size_t count, loff_t *ppos) | |
895 | { | |
896 | return mem_rw(file, buf, count, ppos, 0); | |
897 | } | |
898 | ||
899 | static ssize_t mem_write(struct file *file, const char __user *buf, | |
900 | size_t count, loff_t *ppos) | |
901 | { | |
902 | return mem_rw(file, (char __user*)buf, count, ppos, 1); | |
903 | } | |
904 | ||
85863e47 | 905 | loff_t mem_lseek(struct file *file, loff_t offset, int orig) |
1da177e4 LT |
906 | { |
907 | switch (orig) { | |
908 | case 0: | |
909 | file->f_pos = offset; | |
910 | break; | |
911 | case 1: | |
912 | file->f_pos += offset; | |
913 | break; | |
914 | default: | |
915 | return -EINVAL; | |
916 | } | |
917 | force_successful_syscall_return(); | |
918 | return file->f_pos; | |
919 | } | |
920 | ||
e268337d LT |
921 | static int mem_release(struct inode *inode, struct file *file) |
922 | { | |
923 | struct mm_struct *mm = file->private_data; | |
71879d3c | 924 | if (mm) |
6d08f2c7 | 925 | mmdrop(mm); |
e268337d LT |
926 | return 0; |
927 | } | |
928 | ||
00977a59 | 929 | static const struct file_operations proc_mem_operations = { |
1da177e4 LT |
930 | .llseek = mem_lseek, |
931 | .read = mem_read, | |
932 | .write = mem_write, | |
933 | .open = mem_open, | |
e268337d | 934 | .release = mem_release, |
1da177e4 LT |
935 | }; |
936 | ||
b409e578 CW |
937 | static int environ_open(struct inode *inode, struct file *file) |
938 | { | |
939 | return __mem_open(inode, file, PTRACE_MODE_READ); | |
940 | } | |
941 | ||
315e28c8 JP |
942 | static ssize_t environ_read(struct file *file, char __user *buf, |
943 | size_t count, loff_t *ppos) | |
944 | { | |
315e28c8 JP |
945 | char *page; |
946 | unsigned long src = *ppos; | |
b409e578 CW |
947 | int ret = 0; |
948 | struct mm_struct *mm = file->private_data; | |
a3b609ef | 949 | unsigned long env_start, env_end; |
315e28c8 | 950 | |
8148a73c MK |
951 | /* Ensure the process spawned far enough to have an environment. */ |
952 | if (!mm || !mm->env_end) | |
b409e578 | 953 | return 0; |
315e28c8 | 954 | |
0ee931c4 | 955 | page = (char *)__get_free_page(GFP_KERNEL); |
315e28c8 | 956 | if (!page) |
b409e578 | 957 | return -ENOMEM; |
315e28c8 | 958 | |
d6f64b89 | 959 | ret = 0; |
388f7934 | 960 | if (!mmget_not_zero(mm)) |
b409e578 | 961 | goto free; |
a3b609ef | 962 | |
88aa7cc6 | 963 | spin_lock(&mm->arg_lock); |
a3b609ef MG |
964 | env_start = mm->env_start; |
965 | env_end = mm->env_end; | |
88aa7cc6 | 966 | spin_unlock(&mm->arg_lock); |
a3b609ef | 967 | |
315e28c8 | 968 | while (count > 0) { |
e8905ec2 DH |
969 | size_t this_len, max_len; |
970 | int retval; | |
315e28c8 | 971 | |
a3b609ef | 972 | if (src >= (env_end - env_start)) |
315e28c8 JP |
973 | break; |
974 | ||
a3b609ef | 975 | this_len = env_end - (env_start + src); |
e8905ec2 DH |
976 | |
977 | max_len = min_t(size_t, PAGE_SIZE, count); | |
978 | this_len = min(max_len, this_len); | |
315e28c8 | 979 | |
7f7ccc2c | 980 | retval = access_remote_vm(mm, (env_start + src), page, this_len, FOLL_ANON); |
315e28c8 JP |
981 | |
982 | if (retval <= 0) { | |
983 | ret = retval; | |
984 | break; | |
985 | } | |
986 | ||
987 | if (copy_to_user(buf, page, retval)) { | |
988 | ret = -EFAULT; | |
989 | break; | |
990 | } | |
991 | ||
992 | ret += retval; | |
993 | src += retval; | |
994 | buf += retval; | |
995 | count -= retval; | |
996 | } | |
997 | *ppos = src; | |
315e28c8 | 998 | mmput(mm); |
b409e578 CW |
999 | |
1000 | free: | |
315e28c8 | 1001 | free_page((unsigned long) page); |
315e28c8 JP |
1002 | return ret; |
1003 | } | |
1004 | ||
1005 | static const struct file_operations proc_environ_operations = { | |
b409e578 | 1006 | .open = environ_open, |
315e28c8 | 1007 | .read = environ_read, |
87df8424 | 1008 | .llseek = generic_file_llseek, |
b409e578 | 1009 | .release = mem_release, |
315e28c8 JP |
1010 | }; |
1011 | ||
c5317167 AV |
1012 | static int auxv_open(struct inode *inode, struct file *file) |
1013 | { | |
1014 | return __mem_open(inode, file, PTRACE_MODE_READ_FSCREDS); | |
1015 | } | |
1016 | ||
1017 | static ssize_t auxv_read(struct file *file, char __user *buf, | |
1018 | size_t count, loff_t *ppos) | |
1019 | { | |
1020 | struct mm_struct *mm = file->private_data; | |
1021 | unsigned int nwords = 0; | |
06b2849d LY |
1022 | |
1023 | if (!mm) | |
1024 | return 0; | |
c5317167 AV |
1025 | do { |
1026 | nwords += 2; | |
1027 | } while (mm->saved_auxv[nwords - 2] != 0); /* AT_NULL */ | |
1028 | return simple_read_from_buffer(buf, count, ppos, mm->saved_auxv, | |
1029 | nwords * sizeof(mm->saved_auxv[0])); | |
1030 | } | |
1031 | ||
1032 | static const struct file_operations proc_auxv_operations = { | |
1033 | .open = auxv_open, | |
1034 | .read = auxv_read, | |
1035 | .llseek = generic_file_llseek, | |
1036 | .release = mem_release, | |
1037 | }; | |
1038 | ||
fa0cbbf1 DR |
1039 | static ssize_t oom_adj_read(struct file *file, char __user *buf, size_t count, |
1040 | loff_t *ppos) | |
1041 | { | |
496ad9aa | 1042 | struct task_struct *task = get_proc_task(file_inode(file)); |
fa0cbbf1 DR |
1043 | char buffer[PROC_NUMBUF]; |
1044 | int oom_adj = OOM_ADJUST_MIN; | |
1045 | size_t len; | |
fa0cbbf1 DR |
1046 | |
1047 | if (!task) | |
1048 | return -ESRCH; | |
f913da59 MH |
1049 | if (task->signal->oom_score_adj == OOM_SCORE_ADJ_MAX) |
1050 | oom_adj = OOM_ADJUST_MAX; | |
1051 | else | |
1052 | oom_adj = (task->signal->oom_score_adj * -OOM_DISABLE) / | |
1053 | OOM_SCORE_ADJ_MAX; | |
fa0cbbf1 | 1054 | put_task_struct(task); |
66606567 CH |
1055 | if (oom_adj > OOM_ADJUST_MAX) |
1056 | oom_adj = OOM_ADJUST_MAX; | |
fa0cbbf1 DR |
1057 | len = snprintf(buffer, sizeof(buffer), "%d\n", oom_adj); |
1058 | return simple_read_from_buffer(buf, count, ppos, buffer, len); | |
1059 | } | |
1060 | ||
1d5f0acb MH |
1061 | static int __set_oom_adj(struct file *file, int oom_adj, bool legacy) |
1062 | { | |
44a70ade | 1063 | struct mm_struct *mm = NULL; |
1d5f0acb MH |
1064 | struct task_struct *task; |
1065 | int err = 0; | |
1066 | ||
1067 | task = get_proc_task(file_inode(file)); | |
1068 | if (!task) | |
1069 | return -ESRCH; | |
1070 | ||
1071 | mutex_lock(&oom_adj_mutex); | |
1072 | if (legacy) { | |
1073 | if (oom_adj < task->signal->oom_score_adj && | |
1074 | !capable(CAP_SYS_RESOURCE)) { | |
1075 | err = -EACCES; | |
1076 | goto err_unlock; | |
1077 | } | |
1078 | /* | |
1079 | * /proc/pid/oom_adj is provided for legacy purposes, ask users to use | |
1080 | * /proc/pid/oom_score_adj instead. | |
1081 | */ | |
1082 | pr_warn_once("%s (%d): /proc/%d/oom_adj is deprecated, please use /proc/%d/oom_score_adj instead.\n", | |
1083 | current->comm, task_pid_nr(current), task_pid_nr(task), | |
1084 | task_pid_nr(task)); | |
1085 | } else { | |
1086 | if ((short)oom_adj < task->signal->oom_score_adj_min && | |
1087 | !capable(CAP_SYS_RESOURCE)) { | |
1088 | err = -EACCES; | |
1089 | goto err_unlock; | |
1090 | } | |
1091 | } | |
1092 | ||
44a70ade MH |
1093 | /* |
1094 | * Make sure we will check other processes sharing the mm if this is | |
1095 | * not vfrok which wants its own oom_score_adj. | |
1096 | * pin the mm so it doesn't go away and get reused after task_unlock | |
1097 | */ | |
1098 | if (!task->vfork_done) { | |
1099 | struct task_struct *p = find_lock_task_mm(task); | |
1100 | ||
1101 | if (p) { | |
67197a4f | 1102 | if (test_bit(MMF_MULTIPROCESS, &p->mm->flags)) { |
44a70ade | 1103 | mm = p->mm; |
f1f10076 | 1104 | mmgrab(mm); |
44a70ade MH |
1105 | } |
1106 | task_unlock(p); | |
1107 | } | |
1108 | } | |
1109 | ||
1d5f0acb MH |
1110 | task->signal->oom_score_adj = oom_adj; |
1111 | if (!legacy && has_capability_noaudit(current, CAP_SYS_RESOURCE)) | |
1112 | task->signal->oom_score_adj_min = (short)oom_adj; | |
1113 | trace_oom_score_adj_update(task); | |
44a70ade MH |
1114 | |
1115 | if (mm) { | |
1116 | struct task_struct *p; | |
1117 | ||
1118 | rcu_read_lock(); | |
1119 | for_each_process(p) { | |
1120 | if (same_thread_group(task, p)) | |
1121 | continue; | |
1122 | ||
1123 | /* do not touch kernel threads or the global init */ | |
1124 | if (p->flags & PF_KTHREAD || is_global_init(p)) | |
1125 | continue; | |
1126 | ||
1127 | task_lock(p); | |
1128 | if (!p->vfork_done && process_shares_mm(p, mm)) { | |
44a70ade MH |
1129 | p->signal->oom_score_adj = oom_adj; |
1130 | if (!legacy && has_capability_noaudit(current, CAP_SYS_RESOURCE)) | |
1131 | p->signal->oom_score_adj_min = (short)oom_adj; | |
1132 | } | |
1133 | task_unlock(p); | |
1134 | } | |
1135 | rcu_read_unlock(); | |
1136 | mmdrop(mm); | |
1137 | } | |
1d5f0acb MH |
1138 | err_unlock: |
1139 | mutex_unlock(&oom_adj_mutex); | |
1140 | put_task_struct(task); | |
1141 | return err; | |
1142 | } | |
f913da59 | 1143 | |
b72bdfa7 DR |
1144 | /* |
1145 | * /proc/pid/oom_adj exists solely for backwards compatibility with previous | |
1146 | * kernels. The effective policy is defined by oom_score_adj, which has a | |
1147 | * different scale: oom_adj grew exponentially and oom_score_adj grows linearly. | |
1148 | * Values written to oom_adj are simply mapped linearly to oom_score_adj. | |
1149 | * Processes that become oom disabled via oom_adj will still be oom disabled | |
1150 | * with this implementation. | |
1151 | * | |
1152 | * oom_adj cannot be removed since existing userspace binaries use it. | |
1153 | */ | |
fa0cbbf1 DR |
1154 | static ssize_t oom_adj_write(struct file *file, const char __user *buf, |
1155 | size_t count, loff_t *ppos) | |
1156 | { | |
860a2e7f | 1157 | char buffer[PROC_NUMBUF] = {}; |
fa0cbbf1 | 1158 | int oom_adj; |
fa0cbbf1 DR |
1159 | int err; |
1160 | ||
fa0cbbf1 DR |
1161 | if (count > sizeof(buffer) - 1) |
1162 | count = sizeof(buffer) - 1; | |
1163 | if (copy_from_user(buffer, buf, count)) { | |
1164 | err = -EFAULT; | |
1165 | goto out; | |
1166 | } | |
1167 | ||
1168 | err = kstrtoint(strstrip(buffer), 0, &oom_adj); | |
1169 | if (err) | |
1170 | goto out; | |
1171 | if ((oom_adj < OOM_ADJUST_MIN || oom_adj > OOM_ADJUST_MAX) && | |
1172 | oom_adj != OOM_DISABLE) { | |
1173 | err = -EINVAL; | |
1174 | goto out; | |
1175 | } | |
1176 | ||
fa0cbbf1 DR |
1177 | /* |
1178 | * Scale /proc/pid/oom_score_adj appropriately ensuring that a maximum | |
1179 | * value is always attainable. | |
1180 | */ | |
1181 | if (oom_adj == OOM_ADJUST_MAX) | |
1182 | oom_adj = OOM_SCORE_ADJ_MAX; | |
1183 | else | |
1184 | oom_adj = (oom_adj * OOM_SCORE_ADJ_MAX) / -OOM_DISABLE; | |
1185 | ||
1d5f0acb | 1186 | err = __set_oom_adj(file, oom_adj, true); |
fa0cbbf1 DR |
1187 | out: |
1188 | return err < 0 ? err : count; | |
1189 | } | |
1190 | ||
1191 | static const struct file_operations proc_oom_adj_operations = { | |
1192 | .read = oom_adj_read, | |
1193 | .write = oom_adj_write, | |
1194 | .llseek = generic_file_llseek, | |
1195 | }; | |
1196 | ||
a63d83f4 DR |
1197 | static ssize_t oom_score_adj_read(struct file *file, char __user *buf, |
1198 | size_t count, loff_t *ppos) | |
1199 | { | |
496ad9aa | 1200 | struct task_struct *task = get_proc_task(file_inode(file)); |
a63d83f4 | 1201 | char buffer[PROC_NUMBUF]; |
a9c58b90 | 1202 | short oom_score_adj = OOM_SCORE_ADJ_MIN; |
a63d83f4 DR |
1203 | size_t len; |
1204 | ||
1205 | if (!task) | |
1206 | return -ESRCH; | |
f913da59 | 1207 | oom_score_adj = task->signal->oom_score_adj; |
a63d83f4 | 1208 | put_task_struct(task); |
a9c58b90 | 1209 | len = snprintf(buffer, sizeof(buffer), "%hd\n", oom_score_adj); |
a63d83f4 DR |
1210 | return simple_read_from_buffer(buf, count, ppos, buffer, len); |
1211 | } | |
1212 | ||
1213 | static ssize_t oom_score_adj_write(struct file *file, const char __user *buf, | |
1214 | size_t count, loff_t *ppos) | |
1215 | { | |
860a2e7f | 1216 | char buffer[PROC_NUMBUF] = {}; |
0a8cb8e3 | 1217 | int oom_score_adj; |
a63d83f4 DR |
1218 | int err; |
1219 | ||
a63d83f4 DR |
1220 | if (count > sizeof(buffer) - 1) |
1221 | count = sizeof(buffer) - 1; | |
723548bf DR |
1222 | if (copy_from_user(buffer, buf, count)) { |
1223 | err = -EFAULT; | |
1224 | goto out; | |
1225 | } | |
a63d83f4 | 1226 | |
0a8cb8e3 | 1227 | err = kstrtoint(strstrip(buffer), 0, &oom_score_adj); |
a63d83f4 | 1228 | if (err) |
723548bf | 1229 | goto out; |
a63d83f4 | 1230 | if (oom_score_adj < OOM_SCORE_ADJ_MIN || |
723548bf DR |
1231 | oom_score_adj > OOM_SCORE_ADJ_MAX) { |
1232 | err = -EINVAL; | |
1233 | goto out; | |
1234 | } | |
a63d83f4 | 1235 | |
1d5f0acb | 1236 | err = __set_oom_adj(file, oom_score_adj, false); |
723548bf DR |
1237 | out: |
1238 | return err < 0 ? err : count; | |
a63d83f4 DR |
1239 | } |
1240 | ||
1241 | static const struct file_operations proc_oom_score_adj_operations = { | |
1242 | .read = oom_score_adj_read, | |
1243 | .write = oom_score_adj_write, | |
6038f373 | 1244 | .llseek = default_llseek, |
a63d83f4 DR |
1245 | }; |
1246 | ||
4b7d248b | 1247 | #ifdef CONFIG_AUDIT |
b4eb4f7f | 1248 | #define TMPBUFLEN 11 |
1da177e4 LT |
1249 | static ssize_t proc_loginuid_read(struct file * file, char __user * buf, |
1250 | size_t count, loff_t *ppos) | |
1251 | { | |
496ad9aa | 1252 | struct inode * inode = file_inode(file); |
99f89551 | 1253 | struct task_struct *task = get_proc_task(inode); |
1da177e4 LT |
1254 | ssize_t length; |
1255 | char tmpbuf[TMPBUFLEN]; | |
1256 | ||
99f89551 EB |
1257 | if (!task) |
1258 | return -ESRCH; | |
1da177e4 | 1259 | length = scnprintf(tmpbuf, TMPBUFLEN, "%u", |
e1760bd5 EB |
1260 | from_kuid(file->f_cred->user_ns, |
1261 | audit_get_loginuid(task))); | |
99f89551 | 1262 | put_task_struct(task); |
1da177e4 LT |
1263 | return simple_read_from_buffer(buf, count, ppos, tmpbuf, length); |
1264 | } | |
1265 | ||
1266 | static ssize_t proc_loginuid_write(struct file * file, const char __user * buf, | |
1267 | size_t count, loff_t *ppos) | |
1268 | { | |
496ad9aa | 1269 | struct inode * inode = file_inode(file); |
1da177e4 | 1270 | uid_t loginuid; |
e1760bd5 | 1271 | kuid_t kloginuid; |
774636e1 | 1272 | int rv; |
1da177e4 | 1273 | |
4ea33a97 JA |
1274 | /* Don't let kthreads write their own loginuid */ |
1275 | if (current->flags & PF_KTHREAD) | |
1276 | return -EPERM; | |
1277 | ||
7dc52157 PM |
1278 | rcu_read_lock(); |
1279 | if (current != pid_task(proc_pid(inode), PIDTYPE_PID)) { | |
1280 | rcu_read_unlock(); | |
1da177e4 | 1281 | return -EPERM; |
7dc52157 PM |
1282 | } |
1283 | rcu_read_unlock(); | |
1da177e4 | 1284 | |
1da177e4 LT |
1285 | if (*ppos != 0) { |
1286 | /* No partial writes. */ | |
1287 | return -EINVAL; | |
1288 | } | |
1da177e4 | 1289 | |
774636e1 AD |
1290 | rv = kstrtou32_from_user(buf, count, 10, &loginuid); |
1291 | if (rv < 0) | |
1292 | return rv; | |
81407c84 EP |
1293 | |
1294 | /* is userspace tring to explicitly UNSET the loginuid? */ | |
1295 | if (loginuid == AUDIT_UID_UNSET) { | |
1296 | kloginuid = INVALID_UID; | |
1297 | } else { | |
1298 | kloginuid = make_kuid(file->f_cred->user_ns, loginuid); | |
774636e1 AD |
1299 | if (!uid_valid(kloginuid)) |
1300 | return -EINVAL; | |
e1760bd5 EB |
1301 | } |
1302 | ||
774636e1 AD |
1303 | rv = audit_set_loginuid(kloginuid); |
1304 | if (rv < 0) | |
1305 | return rv; | |
1306 | return count; | |
1da177e4 LT |
1307 | } |
1308 | ||
00977a59 | 1309 | static const struct file_operations proc_loginuid_operations = { |
1da177e4 LT |
1310 | .read = proc_loginuid_read, |
1311 | .write = proc_loginuid_write, | |
87df8424 | 1312 | .llseek = generic_file_llseek, |
1da177e4 | 1313 | }; |
1e0bd755 EP |
1314 | |
1315 | static ssize_t proc_sessionid_read(struct file * file, char __user * buf, | |
1316 | size_t count, loff_t *ppos) | |
1317 | { | |
496ad9aa | 1318 | struct inode * inode = file_inode(file); |
1e0bd755 EP |
1319 | struct task_struct *task = get_proc_task(inode); |
1320 | ssize_t length; | |
1321 | char tmpbuf[TMPBUFLEN]; | |
1322 | ||
1323 | if (!task) | |
1324 | return -ESRCH; | |
1325 | length = scnprintf(tmpbuf, TMPBUFLEN, "%u", | |
1326 | audit_get_sessionid(task)); | |
1327 | put_task_struct(task); | |
1328 | return simple_read_from_buffer(buf, count, ppos, tmpbuf, length); | |
1329 | } | |
1330 | ||
1331 | static const struct file_operations proc_sessionid_operations = { | |
1332 | .read = proc_sessionid_read, | |
87df8424 | 1333 | .llseek = generic_file_llseek, |
1e0bd755 | 1334 | }; |
1da177e4 LT |
1335 | #endif |
1336 | ||
f4f154fd AM |
1337 | #ifdef CONFIG_FAULT_INJECTION |
1338 | static ssize_t proc_fault_inject_read(struct file * file, char __user * buf, | |
1339 | size_t count, loff_t *ppos) | |
1340 | { | |
496ad9aa | 1341 | struct task_struct *task = get_proc_task(file_inode(file)); |
f4f154fd AM |
1342 | char buffer[PROC_NUMBUF]; |
1343 | size_t len; | |
1344 | int make_it_fail; | |
f4f154fd AM |
1345 | |
1346 | if (!task) | |
1347 | return -ESRCH; | |
1348 | make_it_fail = task->make_it_fail; | |
1349 | put_task_struct(task); | |
1350 | ||
1351 | len = snprintf(buffer, sizeof(buffer), "%i\n", make_it_fail); | |
0c28f287 AM |
1352 | |
1353 | return simple_read_from_buffer(buf, count, ppos, buffer, len); | |
f4f154fd AM |
1354 | } |
1355 | ||
1356 | static ssize_t proc_fault_inject_write(struct file * file, | |
1357 | const char __user * buf, size_t count, loff_t *ppos) | |
1358 | { | |
1359 | struct task_struct *task; | |
860a2e7f | 1360 | char buffer[PROC_NUMBUF] = {}; |
f4f154fd | 1361 | int make_it_fail; |
774636e1 | 1362 | int rv; |
f4f154fd AM |
1363 | |
1364 | if (!capable(CAP_SYS_RESOURCE)) | |
1365 | return -EPERM; | |
860a2e7f | 1366 | |
f4f154fd AM |
1367 | if (count > sizeof(buffer) - 1) |
1368 | count = sizeof(buffer) - 1; | |
1369 | if (copy_from_user(buffer, buf, count)) | |
1370 | return -EFAULT; | |
774636e1 AD |
1371 | rv = kstrtoint(strstrip(buffer), 0, &make_it_fail); |
1372 | if (rv < 0) | |
1373 | return rv; | |
16caed31 DJ |
1374 | if (make_it_fail < 0 || make_it_fail > 1) |
1375 | return -EINVAL; | |
1376 | ||
496ad9aa | 1377 | task = get_proc_task(file_inode(file)); |
f4f154fd AM |
1378 | if (!task) |
1379 | return -ESRCH; | |
1380 | task->make_it_fail = make_it_fail; | |
1381 | put_task_struct(task); | |
cba8aafe VL |
1382 | |
1383 | return count; | |
f4f154fd AM |
1384 | } |
1385 | ||
00977a59 | 1386 | static const struct file_operations proc_fault_inject_operations = { |
f4f154fd AM |
1387 | .read = proc_fault_inject_read, |
1388 | .write = proc_fault_inject_write, | |
87df8424 | 1389 | .llseek = generic_file_llseek, |
f4f154fd | 1390 | }; |
e41d5818 DV |
1391 | |
1392 | static ssize_t proc_fail_nth_write(struct file *file, const char __user *buf, | |
1393 | size_t count, loff_t *ppos) | |
1394 | { | |
1395 | struct task_struct *task; | |
9049f2f6 AM |
1396 | int err; |
1397 | unsigned int n; | |
e41d5818 | 1398 | |
1203c8e6 AM |
1399 | err = kstrtouint_from_user(buf, count, 0, &n); |
1400 | if (err) | |
1401 | return err; | |
1402 | ||
e41d5818 DV |
1403 | task = get_proc_task(file_inode(file)); |
1404 | if (!task) | |
1405 | return -ESRCH; | |
9f7118b2 | 1406 | task->fail_nth = n; |
e41d5818 | 1407 | put_task_struct(task); |
1203c8e6 | 1408 | |
e41d5818 DV |
1409 | return count; |
1410 | } | |
1411 | ||
1412 | static ssize_t proc_fail_nth_read(struct file *file, char __user *buf, | |
1413 | size_t count, loff_t *ppos) | |
1414 | { | |
1415 | struct task_struct *task; | |
bfc74093 AM |
1416 | char numbuf[PROC_NUMBUF]; |
1417 | ssize_t len; | |
e41d5818 DV |
1418 | |
1419 | task = get_proc_task(file_inode(file)); | |
1420 | if (!task) | |
1421 | return -ESRCH; | |
9f7118b2 | 1422 | len = snprintf(numbuf, sizeof(numbuf), "%u\n", task->fail_nth); |
1203c8e6 | 1423 | put_task_struct(task); |
a44937fe | 1424 | return simple_read_from_buffer(buf, count, ppos, numbuf, len); |
e41d5818 DV |
1425 | } |
1426 | ||
1427 | static const struct file_operations proc_fail_nth_operations = { | |
1428 | .read = proc_fail_nth_read, | |
1429 | .write = proc_fail_nth_write, | |
1430 | }; | |
f4f154fd AM |
1431 | #endif |
1432 | ||
9745512c | 1433 | |
43ae34cb IM |
1434 | #ifdef CONFIG_SCHED_DEBUG |
1435 | /* | |
1436 | * Print out various scheduling related per-task fields: | |
1437 | */ | |
1438 | static int sched_show(struct seq_file *m, void *v) | |
1439 | { | |
1440 | struct inode *inode = m->private; | |
9d78edea | 1441 | struct pid_namespace *ns = proc_pid_ns(inode->i_sb); |
43ae34cb IM |
1442 | struct task_struct *p; |
1443 | ||
43ae34cb IM |
1444 | p = get_proc_task(inode); |
1445 | if (!p) | |
1446 | return -ESRCH; | |
74dc3384 | 1447 | proc_sched_show_task(p, ns, m); |
43ae34cb IM |
1448 | |
1449 | put_task_struct(p); | |
1450 | ||
1451 | return 0; | |
1452 | } | |
1453 | ||
1454 | static ssize_t | |
1455 | sched_write(struct file *file, const char __user *buf, | |
1456 | size_t count, loff_t *offset) | |
1457 | { | |
496ad9aa | 1458 | struct inode *inode = file_inode(file); |
43ae34cb IM |
1459 | struct task_struct *p; |
1460 | ||
43ae34cb IM |
1461 | p = get_proc_task(inode); |
1462 | if (!p) | |
1463 | return -ESRCH; | |
1464 | proc_sched_set_task(p); | |
1465 | ||
1466 | put_task_struct(p); | |
1467 | ||
1468 | return count; | |
1469 | } | |
1470 | ||
1471 | static int sched_open(struct inode *inode, struct file *filp) | |
1472 | { | |
c6a34058 | 1473 | return single_open(filp, sched_show, inode); |
43ae34cb IM |
1474 | } |
1475 | ||
1476 | static const struct file_operations proc_pid_sched_operations = { | |
1477 | .open = sched_open, | |
1478 | .read = seq_read, | |
1479 | .write = sched_write, | |
1480 | .llseek = seq_lseek, | |
5ea473a1 | 1481 | .release = single_release, |
43ae34cb IM |
1482 | }; |
1483 | ||
1484 | #endif | |
1485 | ||
5091faa4 MG |
1486 | #ifdef CONFIG_SCHED_AUTOGROUP |
1487 | /* | |
1488 | * Print out autogroup related information: | |
1489 | */ | |
1490 | static int sched_autogroup_show(struct seq_file *m, void *v) | |
1491 | { | |
1492 | struct inode *inode = m->private; | |
1493 | struct task_struct *p; | |
1494 | ||
1495 | p = get_proc_task(inode); | |
1496 | if (!p) | |
1497 | return -ESRCH; | |
1498 | proc_sched_autogroup_show_task(p, m); | |
1499 | ||
1500 | put_task_struct(p); | |
1501 | ||
1502 | return 0; | |
1503 | } | |
1504 | ||
1505 | static ssize_t | |
1506 | sched_autogroup_write(struct file *file, const char __user *buf, | |
1507 | size_t count, loff_t *offset) | |
1508 | { | |
496ad9aa | 1509 | struct inode *inode = file_inode(file); |
5091faa4 | 1510 | struct task_struct *p; |
860a2e7f | 1511 | char buffer[PROC_NUMBUF] = {}; |
0a8cb8e3 | 1512 | int nice; |
5091faa4 MG |
1513 | int err; |
1514 | ||
5091faa4 MG |
1515 | if (count > sizeof(buffer) - 1) |
1516 | count = sizeof(buffer) - 1; | |
1517 | if (copy_from_user(buffer, buf, count)) | |
1518 | return -EFAULT; | |
1519 | ||
0a8cb8e3 AD |
1520 | err = kstrtoint(strstrip(buffer), 0, &nice); |
1521 | if (err < 0) | |
1522 | return err; | |
5091faa4 MG |
1523 | |
1524 | p = get_proc_task(inode); | |
1525 | if (!p) | |
1526 | return -ESRCH; | |
1527 | ||
2e5b5b3a | 1528 | err = proc_sched_autogroup_set_nice(p, nice); |
5091faa4 MG |
1529 | if (err) |
1530 | count = err; | |
1531 | ||
1532 | put_task_struct(p); | |
1533 | ||
1534 | return count; | |
1535 | } | |
1536 | ||
1537 | static int sched_autogroup_open(struct inode *inode, struct file *filp) | |
1538 | { | |
1539 | int ret; | |
1540 | ||
1541 | ret = single_open(filp, sched_autogroup_show, NULL); | |
1542 | if (!ret) { | |
1543 | struct seq_file *m = filp->private_data; | |
1544 | ||
1545 | m->private = inode; | |
1546 | } | |
1547 | return ret; | |
1548 | } | |
1549 | ||
1550 | static const struct file_operations proc_pid_sched_autogroup_operations = { | |
1551 | .open = sched_autogroup_open, | |
1552 | .read = seq_read, | |
1553 | .write = sched_autogroup_write, | |
1554 | .llseek = seq_lseek, | |
1555 | .release = single_release, | |
1556 | }; | |
1557 | ||
1558 | #endif /* CONFIG_SCHED_AUTOGROUP */ | |
1559 | ||
04a8682a AV |
1560 | #ifdef CONFIG_TIME_NS |
1561 | static int timens_offsets_show(struct seq_file *m, void *v) | |
1562 | { | |
1563 | struct task_struct *p; | |
1564 | ||
1565 | p = get_proc_task(file_inode(m->file)); | |
1566 | if (!p) | |
1567 | return -ESRCH; | |
1568 | proc_timens_show_offsets(p, m); | |
1569 | ||
1570 | put_task_struct(p); | |
1571 | ||
1572 | return 0; | |
1573 | } | |
1574 | ||
1575 | static ssize_t timens_offsets_write(struct file *file, const char __user *buf, | |
1576 | size_t count, loff_t *ppos) | |
1577 | { | |
1578 | struct inode *inode = file_inode(file); | |
1579 | struct proc_timens_offset offsets[2]; | |
1580 | char *kbuf = NULL, *pos, *next_line; | |
1581 | struct task_struct *p; | |
1582 | int ret, noffsets; | |
1583 | ||
1584 | /* Only allow < page size writes at the beginning of the file */ | |
1585 | if ((*ppos != 0) || (count >= PAGE_SIZE)) | |
1586 | return -EINVAL; | |
1587 | ||
1588 | /* Slurp in the user data */ | |
1589 | kbuf = memdup_user_nul(buf, count); | |
1590 | if (IS_ERR(kbuf)) | |
1591 | return PTR_ERR(kbuf); | |
1592 | ||
1593 | /* Parse the user data */ | |
1594 | ret = -EINVAL; | |
1595 | noffsets = 0; | |
1596 | for (pos = kbuf; pos; pos = next_line) { | |
1597 | struct proc_timens_offset *off = &offsets[noffsets]; | |
94d440d6 | 1598 | char clock[10]; |
04a8682a AV |
1599 | int err; |
1600 | ||
1601 | /* Find the end of line and ensure we don't look past it */ | |
1602 | next_line = strchr(pos, '\n'); | |
1603 | if (next_line) { | |
1604 | *next_line = '\0'; | |
1605 | next_line++; | |
1606 | if (*next_line == '\0') | |
1607 | next_line = NULL; | |
1608 | } | |
1609 | ||
94d440d6 | 1610 | err = sscanf(pos, "%9s %lld %lu", clock, |
04a8682a AV |
1611 | &off->val.tv_sec, &off->val.tv_nsec); |
1612 | if (err != 3 || off->val.tv_nsec >= NSEC_PER_SEC) | |
1613 | goto out; | |
94d440d6 AV |
1614 | |
1615 | clock[sizeof(clock) - 1] = 0; | |
1616 | if (strcmp(clock, "monotonic") == 0 || | |
1617 | strcmp(clock, __stringify(CLOCK_MONOTONIC)) == 0) | |
1618 | off->clockid = CLOCK_MONOTONIC; | |
1619 | else if (strcmp(clock, "boottime") == 0 || | |
1620 | strcmp(clock, __stringify(CLOCK_BOOTTIME)) == 0) | |
1621 | off->clockid = CLOCK_BOOTTIME; | |
1622 | else | |
1623 | goto out; | |
1624 | ||
04a8682a AV |
1625 | noffsets++; |
1626 | if (noffsets == ARRAY_SIZE(offsets)) { | |
1627 | if (next_line) | |
1628 | count = next_line - kbuf; | |
1629 | break; | |
1630 | } | |
1631 | } | |
1632 | ||
1633 | ret = -ESRCH; | |
1634 | p = get_proc_task(inode); | |
1635 | if (!p) | |
1636 | goto out; | |
1637 | ret = proc_timens_set_offset(file, p, offsets, noffsets); | |
1638 | put_task_struct(p); | |
1639 | if (ret) | |
1640 | goto out; | |
1641 | ||
1642 | ret = count; | |
1643 | out: | |
1644 | kfree(kbuf); | |
1645 | return ret; | |
1646 | } | |
1647 | ||
1648 | static int timens_offsets_open(struct inode *inode, struct file *filp) | |
1649 | { | |
1650 | return single_open(filp, timens_offsets_show, inode); | |
1651 | } | |
1652 | ||
1653 | static const struct file_operations proc_timens_offsets_operations = { | |
1654 | .open = timens_offsets_open, | |
1655 | .read = seq_read, | |
1656 | .write = timens_offsets_write, | |
1657 | .llseek = seq_lseek, | |
1658 | .release = single_release, | |
1659 | }; | |
1660 | #endif /* CONFIG_TIME_NS */ | |
1661 | ||
4614a696 | 1662 | static ssize_t comm_write(struct file *file, const char __user *buf, |
1663 | size_t count, loff_t *offset) | |
1664 | { | |
496ad9aa | 1665 | struct inode *inode = file_inode(file); |
4614a696 | 1666 | struct task_struct *p; |
860a2e7f | 1667 | char buffer[TASK_COMM_LEN] = {}; |
830e0fc9 | 1668 | const size_t maxlen = sizeof(buffer) - 1; |
4614a696 | 1669 | |
830e0fc9 | 1670 | if (copy_from_user(buffer, buf, count > maxlen ? maxlen : count)) |
4614a696 | 1671 | return -EFAULT; |
1672 | ||
1673 | p = get_proc_task(inode); | |
1674 | if (!p) | |
1675 | return -ESRCH; | |
1676 | ||
c2f273eb | 1677 | if (same_thread_group(current, p)) { |
4614a696 | 1678 | set_task_comm(p, buffer); |
c2f273eb OK |
1679 | proc_comm_connector(p); |
1680 | } | |
4614a696 | 1681 | else |
1682 | count = -EINVAL; | |
1683 | ||
1684 | put_task_struct(p); | |
1685 | ||
1686 | return count; | |
1687 | } | |
1688 | ||
1689 | static int comm_show(struct seq_file *m, void *v) | |
1690 | { | |
1691 | struct inode *inode = m->private; | |
1692 | struct task_struct *p; | |
1693 | ||
1694 | p = get_proc_task(inode); | |
1695 | if (!p) | |
1696 | return -ESRCH; | |
1697 | ||
88b72b31 TH |
1698 | proc_task_name(m, p, false); |
1699 | seq_putc(m, '\n'); | |
4614a696 | 1700 | |
1701 | put_task_struct(p); | |
1702 | ||
1703 | return 0; | |
1704 | } | |
1705 | ||
1706 | static int comm_open(struct inode *inode, struct file *filp) | |
1707 | { | |
c6a34058 | 1708 | return single_open(filp, comm_show, inode); |
4614a696 | 1709 | } |
1710 | ||
1711 | static const struct file_operations proc_pid_set_comm_operations = { | |
1712 | .open = comm_open, | |
1713 | .read = seq_read, | |
1714 | .write = comm_write, | |
1715 | .llseek = seq_lseek, | |
1716 | .release = single_release, | |
1717 | }; | |
1718 | ||
7773fbc5 | 1719 | static int proc_exe_link(struct dentry *dentry, struct path *exe_path) |
925d1c40 MH |
1720 | { |
1721 | struct task_struct *task; | |
925d1c40 MH |
1722 | struct file *exe_file; |
1723 | ||
2b0143b5 | 1724 | task = get_proc_task(d_inode(dentry)); |
925d1c40 MH |
1725 | if (!task) |
1726 | return -ENOENT; | |
cd81a917 | 1727 | exe_file = get_task_exe_file(task); |
925d1c40 | 1728 | put_task_struct(task); |
925d1c40 MH |
1729 | if (exe_file) { |
1730 | *exe_path = exe_file->f_path; | |
1731 | path_get(&exe_file->f_path); | |
1732 | fput(exe_file); | |
1733 | return 0; | |
1734 | } else | |
1735 | return -ENOENT; | |
1736 | } | |
1737 | ||
6b255391 | 1738 | static const char *proc_pid_get_link(struct dentry *dentry, |
fceef393 AV |
1739 | struct inode *inode, |
1740 | struct delayed_call *done) | |
1da177e4 | 1741 | { |
408ef013 | 1742 | struct path path; |
1da177e4 LT |
1743 | int error = -EACCES; |
1744 | ||
6b255391 AV |
1745 | if (!dentry) |
1746 | return ERR_PTR(-ECHILD); | |
1747 | ||
778c1144 EB |
1748 | /* Are we allowed to snoop on the tasks file descriptors? */ |
1749 | if (!proc_fd_access_allowed(inode)) | |
1da177e4 | 1750 | goto out; |
1da177e4 | 1751 | |
408ef013 CH |
1752 | error = PROC_I(inode)->op.proc_get_link(dentry, &path); |
1753 | if (error) | |
1754 | goto out; | |
1755 | ||
1bc82070 | 1756 | error = nd_jump_link(&path); |
1da177e4 | 1757 | out: |
008b150a | 1758 | return ERR_PTR(error); |
1da177e4 LT |
1759 | } |
1760 | ||
5b09c9fe | 1761 | static int do_proc_readlink(const struct path *path, char __user *buffer, int buflen) |
1da177e4 | 1762 | { |
3a72917c | 1763 | char *tmp = kmalloc(PATH_MAX, GFP_KERNEL); |
3dcd25f3 | 1764 | char *pathname; |
1da177e4 LT |
1765 | int len; |
1766 | ||
1767 | if (!tmp) | |
1768 | return -ENOMEM; | |
0c28f287 | 1769 | |
3a72917c | 1770 | pathname = d_path(path, tmp, PATH_MAX); |
3dcd25f3 JB |
1771 | len = PTR_ERR(pathname); |
1772 | if (IS_ERR(pathname)) | |
1da177e4 | 1773 | goto out; |
3a72917c | 1774 | len = tmp + PATH_MAX - 1 - pathname; |
1da177e4 LT |
1775 | |
1776 | if (len > buflen) | |
1777 | len = buflen; | |
3dcd25f3 | 1778 | if (copy_to_user(buffer, pathname, len)) |
1da177e4 LT |
1779 | len = -EFAULT; |
1780 | out: | |
3a72917c | 1781 | kfree(tmp); |
1da177e4 LT |
1782 | return len; |
1783 | } | |
1784 | ||
1785 | static int proc_pid_readlink(struct dentry * dentry, char __user * buffer, int buflen) | |
1786 | { | |
1787 | int error = -EACCES; | |
2b0143b5 | 1788 | struct inode *inode = d_inode(dentry); |
3dcd25f3 | 1789 | struct path path; |
1da177e4 | 1790 | |
778c1144 EB |
1791 | /* Are we allowed to snoop on the tasks file descriptors? */ |
1792 | if (!proc_fd_access_allowed(inode)) | |
1da177e4 | 1793 | goto out; |
1da177e4 | 1794 | |
7773fbc5 | 1795 | error = PROC_I(inode)->op.proc_get_link(dentry, &path); |
1da177e4 LT |
1796 | if (error) |
1797 | goto out; | |
1798 | ||
3dcd25f3 JB |
1799 | error = do_proc_readlink(&path, buffer, buflen); |
1800 | path_put(&path); | |
1da177e4 | 1801 | out: |
1da177e4 LT |
1802 | return error; |
1803 | } | |
1804 | ||
faf60af1 | 1805 | const struct inode_operations proc_pid_link_inode_operations = { |
1da177e4 | 1806 | .readlink = proc_pid_readlink, |
6b255391 | 1807 | .get_link = proc_pid_get_link, |
6d76fa58 | 1808 | .setattr = proc_setattr, |
1da177e4 LT |
1809 | }; |
1810 | ||
28a6d671 EB |
1811 | |
1812 | /* building an inode */ | |
1813 | ||
c6eb50d2 | 1814 | void task_dump_owner(struct task_struct *task, umode_t mode, |
68eb94f1 EB |
1815 | kuid_t *ruid, kgid_t *rgid) |
1816 | { | |
1817 | /* Depending on the state of dumpable compute who should own a | |
1818 | * proc file for a task. | |
1819 | */ | |
1820 | const struct cred *cred; | |
1821 | kuid_t uid; | |
1822 | kgid_t gid; | |
1823 | ||
2e0ad552 AD |
1824 | if (unlikely(task->flags & PF_KTHREAD)) { |
1825 | *ruid = GLOBAL_ROOT_UID; | |
1826 | *rgid = GLOBAL_ROOT_GID; | |
1827 | return; | |
1828 | } | |
1829 | ||
68eb94f1 EB |
1830 | /* Default to the tasks effective ownership */ |
1831 | rcu_read_lock(); | |
1832 | cred = __task_cred(task); | |
1833 | uid = cred->euid; | |
1834 | gid = cred->egid; | |
1835 | rcu_read_unlock(); | |
1836 | ||
1837 | /* | |
1838 | * Before the /proc/pid/status file was created the only way to read | |
1839 | * the effective uid of a /process was to stat /proc/pid. Reading | |
1840 | * /proc/pid/status is slow enough that procps and other packages | |
1841 | * kept stating /proc/pid. To keep the rules in /proc simple I have | |
1842 | * made this apply to all per process world readable and executable | |
1843 | * directories. | |
1844 | */ | |
1845 | if (mode != (S_IFDIR|S_IRUGO|S_IXUGO)) { | |
1846 | struct mm_struct *mm; | |
1847 | task_lock(task); | |
1848 | mm = task->mm; | |
1849 | /* Make non-dumpable tasks owned by some root */ | |
1850 | if (mm) { | |
1851 | if (get_dumpable(mm) != SUID_DUMP_USER) { | |
1852 | struct user_namespace *user_ns = mm->user_ns; | |
1853 | ||
1854 | uid = make_kuid(user_ns, 0); | |
1855 | if (!uid_valid(uid)) | |
1856 | uid = GLOBAL_ROOT_UID; | |
1857 | ||
1858 | gid = make_kgid(user_ns, 0); | |
1859 | if (!gid_valid(gid)) | |
1860 | gid = GLOBAL_ROOT_GID; | |
1861 | } | |
1862 | } else { | |
1863 | uid = GLOBAL_ROOT_UID; | |
1864 | gid = GLOBAL_ROOT_GID; | |
1865 | } | |
1866 | task_unlock(task); | |
1867 | } | |
1868 | *ruid = uid; | |
1869 | *rgid = gid; | |
1870 | } | |
1871 | ||
7bc3e6e5 EB |
1872 | void proc_pid_evict_inode(struct proc_inode *ei) |
1873 | { | |
1874 | struct pid *pid = ei->pid; | |
1875 | ||
1876 | if (S_ISDIR(ei->vfs_inode.i_mode)) { | |
63f818f4 | 1877 | spin_lock(&pid->lock); |
7bc3e6e5 | 1878 | hlist_del_init_rcu(&ei->sibling_inodes); |
63f818f4 | 1879 | spin_unlock(&pid->lock); |
7bc3e6e5 EB |
1880 | } |
1881 | ||
1882 | put_pid(pid); | |
1883 | } | |
1884 | ||
d919a1e7 | 1885 | struct inode *proc_pid_make_inode(struct super_block *sb, |
db978da8 | 1886 | struct task_struct *task, umode_t mode) |
28a6d671 EB |
1887 | { |
1888 | struct inode * inode; | |
1889 | struct proc_inode *ei; | |
7bc3e6e5 | 1890 | struct pid *pid; |
1da177e4 | 1891 | |
28a6d671 | 1892 | /* We need a new inode */ |
1da177e4 | 1893 | |
28a6d671 EB |
1894 | inode = new_inode(sb); |
1895 | if (!inode) | |
1896 | goto out; | |
1897 | ||
1898 | /* Common stuff */ | |
1899 | ei = PROC_I(inode); | |
db978da8 | 1900 | inode->i_mode = mode; |
85fe4025 | 1901 | inode->i_ino = get_next_ino(); |
200d9421 | 1902 | simple_inode_init_ts(inode); |
28a6d671 EB |
1903 | inode->i_op = &proc_def_inode_operations; |
1904 | ||
1905 | /* | |
1906 | * grab the reference to task. | |
1907 | */ | |
7bc3e6e5 EB |
1908 | pid = get_task_pid(task, PIDTYPE_PID); |
1909 | if (!pid) | |
28a6d671 EB |
1910 | goto out_unlock; |
1911 | ||
7bc3e6e5 EB |
1912 | /* Let the pid remember us for quick removal */ |
1913 | ei->pid = pid; | |
7bc3e6e5 | 1914 | |
68eb94f1 | 1915 | task_dump_owner(task, 0, &inode->i_uid, &inode->i_gid); |
28a6d671 EB |
1916 | security_task_to_inode(task, inode); |
1917 | ||
1da177e4 | 1918 | out: |
28a6d671 EB |
1919 | return inode; |
1920 | ||
1921 | out_unlock: | |
1922 | iput(inode); | |
1923 | return NULL; | |
1da177e4 LT |
1924 | } |
1925 | ||
d919a1e7 ZC |
1926 | /* |
1927 | * Generating an inode and adding it into @pid->inodes, so that task will | |
1928 | * invalidate inode's dentry before being released. | |
1929 | * | |
1930 | * This helper is used for creating dir-type entries under '/proc' and | |
1931 | * '/proc/<tgid>/task'. Other entries(eg. fd, stat) under '/proc/<tgid>' | |
1932 | * can be released by invalidating '/proc/<tgid>' dentry. | |
1933 | * In theory, dentries under '/proc/<tgid>/task' can also be released by | |
1934 | * invalidating '/proc/<tgid>' dentry, we reserve it to handle single | |
1935 | * thread exiting situation: Any one of threads should invalidate its | |
1936 | * '/proc/<tgid>/task/<pid>' dentry before released. | |
1937 | */ | |
1938 | static struct inode *proc_pid_make_base_inode(struct super_block *sb, | |
1939 | struct task_struct *task, umode_t mode) | |
1940 | { | |
1941 | struct inode *inode; | |
1942 | struct proc_inode *ei; | |
1943 | struct pid *pid; | |
1944 | ||
1945 | inode = proc_pid_make_inode(sb, task, mode); | |
1946 | if (!inode) | |
1947 | return NULL; | |
1948 | ||
1949 | /* Let proc_flush_pid find this directory inode */ | |
1950 | ei = PROC_I(inode); | |
1951 | pid = ei->pid; | |
1952 | spin_lock(&pid->lock); | |
1953 | hlist_add_head_rcu(&ei->sibling_inodes, &pid->inodes); | |
1954 | spin_unlock(&pid->lock); | |
1955 | ||
1956 | return inode; | |
1957 | } | |
1958 | ||
b74d24f7 | 1959 | int pid_getattr(struct mnt_idmap *idmap, const struct path *path, |
549c7297 | 1960 | struct kstat *stat, u32 request_mask, unsigned int query_flags) |
1da177e4 | 1961 | { |
a528d35e | 1962 | struct inode *inode = d_inode(path->dentry); |
fa10fed3 | 1963 | struct proc_fs_info *fs_info = proc_sb_info(inode->i_sb); |
28a6d671 | 1964 | struct task_struct *task; |
c69e8d9c | 1965 | |
0d72b928 | 1966 | generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat); |
1da177e4 | 1967 | |
dcb0f222 EB |
1968 | stat->uid = GLOBAL_ROOT_UID; |
1969 | stat->gid = GLOBAL_ROOT_GID; | |
94116929 | 1970 | rcu_read_lock(); |
28a6d671 EB |
1971 | task = pid_task(proc_pid(inode), PIDTYPE_PID); |
1972 | if (task) { | |
fa10fed3 | 1973 | if (!has_pid_permissions(fs_info, task, HIDEPID_INVISIBLE)) { |
0499680a VK |
1974 | rcu_read_unlock(); |
1975 | /* | |
1976 | * This doesn't prevent learning whether PID exists, | |
1977 | * it only makes getattr() consistent with readdir(). | |
1978 | */ | |
1979 | return -ENOENT; | |
1980 | } | |
68eb94f1 | 1981 | task_dump_owner(task, inode->i_mode, &stat->uid, &stat->gid); |
1da177e4 | 1982 | } |
28a6d671 | 1983 | rcu_read_unlock(); |
d6e71144 | 1984 | return 0; |
1da177e4 LT |
1985 | } |
1986 | ||
1da177e4 LT |
1987 | /* dentry stuff */ |
1988 | ||
1989 | /* | |
1bbc5513 AV |
1990 | * Set <pid>/... inode ownership (can change due to setuid(), etc.) |
1991 | */ | |
1992 | void pid_update_inode(struct task_struct *task, struct inode *inode) | |
1993 | { | |
1994 | task_dump_owner(task, inode->i_mode, &inode->i_uid, &inode->i_gid); | |
1995 | ||
1996 | inode->i_mode &= ~(S_ISUID | S_ISGID); | |
1997 | security_task_to_inode(task, inode); | |
1998 | } | |
1999 | ||
2000 | /* | |
1da177e4 LT |
2001 | * Rewrite the inode's ownerships here because the owning task may have |
2002 | * performed a setuid(), etc. | |
99f89551 | 2003 | * |
1da177e4 | 2004 | */ |
1bbc5513 | 2005 | static int pid_revalidate(struct dentry *dentry, unsigned int flags) |
1da177e4 | 2006 | { |
34286d66 NP |
2007 | struct inode *inode; |
2008 | struct task_struct *task; | |
da4d6b9c | 2009 | int ret = 0; |
c69e8d9c | 2010 | |
da4d6b9c SB |
2011 | rcu_read_lock(); |
2012 | inode = d_inode_rcu(dentry); | |
2013 | if (!inode) | |
2014 | goto out; | |
2015 | task = pid_task(proc_pid(inode), PIDTYPE_PID); | |
34286d66 | 2016 | |
99f89551 | 2017 | if (task) { |
1bbc5513 | 2018 | pid_update_inode(task, inode); |
da4d6b9c | 2019 | ret = 1; |
1da177e4 | 2020 | } |
da4d6b9c SB |
2021 | out: |
2022 | rcu_read_unlock(); | |
2023 | return ret; | |
1da177e4 LT |
2024 | } |
2025 | ||
d855a4b7 ON |
2026 | static inline bool proc_inode_is_dead(struct inode *inode) |
2027 | { | |
2028 | return !proc_pid(inode)->tasks[PIDTYPE_PID].first; | |
2029 | } | |
2030 | ||
1dd704b6 DH |
2031 | int pid_delete_dentry(const struct dentry *dentry) |
2032 | { | |
2033 | /* Is the task we represent dead? | |
2034 | * If so, then don't put the dentry on the lru list, | |
2035 | * kill it immediately. | |
2036 | */ | |
2b0143b5 | 2037 | return proc_inode_is_dead(d_inode(dentry)); |
1dd704b6 DH |
2038 | } |
2039 | ||
6b4e306a | 2040 | const struct dentry_operations pid_dentry_operations = |
28a6d671 EB |
2041 | { |
2042 | .d_revalidate = pid_revalidate, | |
2043 | .d_delete = pid_delete_dentry, | |
2044 | }; | |
2045 | ||
2046 | /* Lookups */ | |
2047 | ||
1c0d04c9 EB |
2048 | /* |
2049 | * Fill a directory entry. | |
2050 | * | |
2051 | * If possible create the dcache entry and derive our inode number and | |
2052 | * file type from dcache entry. | |
2053 | * | |
2054 | * Since all of the proc inode numbers are dynamically generated, the inode | |
d2928e85 | 2055 | * numbers do not exist until the inode is cache. This means creating |
1c0d04c9 EB |
2056 | * the dcache entry in readdir is necessary to keep the inode numbers |
2057 | * reported by readdir in sync with the inode numbers reported | |
2058 | * by stat. | |
2059 | */ | |
f0c3b509 | 2060 | bool proc_fill_cache(struct file *file, struct dir_context *ctx, |
a4ef3895 | 2061 | const char *name, unsigned int len, |
c5141e6d | 2062 | instantiate_t instantiate, struct task_struct *task, const void *ptr) |
61a28784 | 2063 | { |
f0c3b509 | 2064 | struct dentry *child, *dir = file->f_path.dentry; |
1df98b8b | 2065 | struct qstr qname = QSTR_INIT(name, len); |
61a28784 | 2066 | struct inode *inode; |
0168b9e3 AV |
2067 | unsigned type = DT_UNKNOWN; |
2068 | ino_t ino = 1; | |
61a28784 | 2069 | |
1df98b8b | 2070 | child = d_hash_and_lookup(dir, &qname); |
61a28784 | 2071 | if (!child) { |
3781764b AV |
2072 | DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wq); |
2073 | child = d_alloc_parallel(dir, &qname, &wq); | |
2074 | if (IS_ERR(child)) | |
1df98b8b | 2075 | goto end_instantiate; |
3781764b | 2076 | if (d_in_lookup(child)) { |
0168b9e3 AV |
2077 | struct dentry *res; |
2078 | res = instantiate(child, task, ptr); | |
3781764b | 2079 | d_lookup_done(child); |
0168b9e3 AV |
2080 | if (unlikely(res)) { |
2081 | dput(child); | |
2082 | child = res; | |
d85b399b AV |
2083 | if (IS_ERR(child)) |
2084 | goto end_instantiate; | |
3781764b | 2085 | } |
61a28784 EB |
2086 | } |
2087 | } | |
2b0143b5 | 2088 | inode = d_inode(child); |
147ce699 AV |
2089 | ino = inode->i_ino; |
2090 | type = inode->i_mode >> 12; | |
61a28784 | 2091 | dput(child); |
d85b399b | 2092 | end_instantiate: |
f0c3b509 | 2093 | return dir_emit(ctx, name, len, ino, type); |
61a28784 EB |
2094 | } |
2095 | ||
640708a2 PE |
2096 | /* |
2097 | * dname_to_vma_addr - maps a dentry name into two unsigned longs | |
2098 | * which represent vma start and end addresses. | |
2099 | */ | |
2100 | static int dname_to_vma_addr(struct dentry *dentry, | |
2101 | unsigned long *start, unsigned long *end) | |
2102 | { | |
ac7f1061 AD |
2103 | const char *str = dentry->d_name.name; |
2104 | unsigned long long sval, eval; | |
2105 | unsigned int len; | |
2106 | ||
35318db5 AD |
2107 | if (str[0] == '0' && str[1] != '-') |
2108 | return -EINVAL; | |
ac7f1061 AD |
2109 | len = _parse_integer(str, 16, &sval); |
2110 | if (len & KSTRTOX_OVERFLOW) | |
2111 | return -EINVAL; | |
2112 | if (sval != (unsigned long)sval) | |
2113 | return -EINVAL; | |
2114 | str += len; | |
2115 | ||
2116 | if (*str != '-') | |
640708a2 | 2117 | return -EINVAL; |
ac7f1061 AD |
2118 | str++; |
2119 | ||
35318db5 AD |
2120 | if (str[0] == '0' && str[1]) |
2121 | return -EINVAL; | |
ac7f1061 AD |
2122 | len = _parse_integer(str, 16, &eval); |
2123 | if (len & KSTRTOX_OVERFLOW) | |
2124 | return -EINVAL; | |
2125 | if (eval != (unsigned long)eval) | |
2126 | return -EINVAL; | |
2127 | str += len; | |
2128 | ||
2129 | if (*str != '\0') | |
2130 | return -EINVAL; | |
2131 | ||
2132 | *start = sval; | |
2133 | *end = eval; | |
640708a2 PE |
2134 | |
2135 | return 0; | |
2136 | } | |
2137 | ||
0b728e19 | 2138 | static int map_files_d_revalidate(struct dentry *dentry, unsigned int flags) |
640708a2 PE |
2139 | { |
2140 | unsigned long vm_start, vm_end; | |
2141 | bool exact_vma_exists = false; | |
2142 | struct mm_struct *mm = NULL; | |
2143 | struct task_struct *task; | |
640708a2 PE |
2144 | struct inode *inode; |
2145 | int status = 0; | |
2146 | ||
0b728e19 | 2147 | if (flags & LOOKUP_RCU) |
640708a2 PE |
2148 | return -ECHILD; |
2149 | ||
2b0143b5 | 2150 | inode = d_inode(dentry); |
640708a2 PE |
2151 | task = get_proc_task(inode); |
2152 | if (!task) | |
2153 | goto out_notask; | |
2154 | ||
caaee623 | 2155 | mm = mm_access(task, PTRACE_MODE_READ_FSCREDS); |
2344bec7 | 2156 | if (IS_ERR_OR_NULL(mm)) |
640708a2 PE |
2157 | goto out; |
2158 | ||
2159 | if (!dname_to_vma_addr(dentry, &vm_start, &vm_end)) { | |
d8ed45c5 | 2160 | status = mmap_read_lock_killable(mm); |
cd9e2bb8 KK |
2161 | if (!status) { |
2162 | exact_vma_exists = !!find_exact_vma(mm, vm_start, | |
2163 | vm_end); | |
d8ed45c5 | 2164 | mmap_read_unlock(mm); |
cd9e2bb8 | 2165 | } |
640708a2 PE |
2166 | } |
2167 | ||
2168 | mmput(mm); | |
2169 | ||
2170 | if (exact_vma_exists) { | |
68eb94f1 EB |
2171 | task_dump_owner(task, 0, &inode->i_uid, &inode->i_gid); |
2172 | ||
640708a2 PE |
2173 | security_task_to_inode(task, inode); |
2174 | status = 1; | |
2175 | } | |
2176 | ||
2177 | out: | |
2178 | put_task_struct(task); | |
2179 | ||
2180 | out_notask: | |
640708a2 PE |
2181 | return status; |
2182 | } | |
2183 | ||
2184 | static const struct dentry_operations tid_map_files_dentry_operations = { | |
2185 | .d_revalidate = map_files_d_revalidate, | |
2186 | .d_delete = pid_delete_dentry, | |
2187 | }; | |
2188 | ||
6b255391 | 2189 | static int map_files_get_link(struct dentry *dentry, struct path *path) |
640708a2 PE |
2190 | { |
2191 | unsigned long vm_start, vm_end; | |
2192 | struct vm_area_struct *vma; | |
2193 | struct task_struct *task; | |
2194 | struct mm_struct *mm; | |
2195 | int rc; | |
2196 | ||
2197 | rc = -ENOENT; | |
2b0143b5 | 2198 | task = get_proc_task(d_inode(dentry)); |
640708a2 PE |
2199 | if (!task) |
2200 | goto out; | |
2201 | ||
2202 | mm = get_task_mm(task); | |
2203 | put_task_struct(task); | |
2204 | if (!mm) | |
2205 | goto out; | |
2206 | ||
2207 | rc = dname_to_vma_addr(dentry, &vm_start, &vm_end); | |
2208 | if (rc) | |
2209 | goto out_mmput; | |
2210 | ||
d8ed45c5 | 2211 | rc = mmap_read_lock_killable(mm); |
cd9e2bb8 KK |
2212 | if (rc) |
2213 | goto out_mmput; | |
2214 | ||
70335abb | 2215 | rc = -ENOENT; |
640708a2 PE |
2216 | vma = find_exact_vma(mm, vm_start, vm_end); |
2217 | if (vma && vma->vm_file) { | |
08582d67 | 2218 | *path = *file_user_path(vma->vm_file); |
640708a2 PE |
2219 | path_get(path); |
2220 | rc = 0; | |
2221 | } | |
d8ed45c5 | 2222 | mmap_read_unlock(mm); |
640708a2 PE |
2223 | |
2224 | out_mmput: | |
2225 | mmput(mm); | |
2226 | out: | |
2227 | return rc; | |
2228 | } | |
2229 | ||
2230 | struct map_files_info { | |
20d28cde AD |
2231 | unsigned long start; |
2232 | unsigned long end; | |
7b540d06 | 2233 | fmode_t mode; |
640708a2 PE |
2234 | }; |
2235 | ||
bdb4d100 | 2236 | /* |
12886f8a AR |
2237 | * Only allow CAP_SYS_ADMIN and CAP_CHECKPOINT_RESTORE to follow the links, due |
2238 | * to concerns about how the symlinks may be used to bypass permissions on | |
2239 | * ancestor directories in the path to the file in question. | |
bdb4d100 CO |
2240 | */ |
2241 | static const char * | |
6b255391 | 2242 | proc_map_files_get_link(struct dentry *dentry, |
fceef393 AV |
2243 | struct inode *inode, |
2244 | struct delayed_call *done) | |
bdb4d100 | 2245 | { |
12886f8a | 2246 | if (!checkpoint_restore_ns_capable(&init_user_ns)) |
bdb4d100 CO |
2247 | return ERR_PTR(-EPERM); |
2248 | ||
fceef393 | 2249 | return proc_pid_get_link(dentry, inode, done); |
bdb4d100 CO |
2250 | } |
2251 | ||
2252 | /* | |
6b255391 | 2253 | * Identical to proc_pid_link_inode_operations except for get_link() |
bdb4d100 CO |
2254 | */ |
2255 | static const struct inode_operations proc_map_files_link_inode_operations = { | |
2256 | .readlink = proc_pid_readlink, | |
6b255391 | 2257 | .get_link = proc_map_files_get_link, |
bdb4d100 CO |
2258 | .setattr = proc_setattr, |
2259 | }; | |
2260 | ||
0168b9e3 AV |
2261 | static struct dentry * |
2262 | proc_map_files_instantiate(struct dentry *dentry, | |
640708a2 PE |
2263 | struct task_struct *task, const void *ptr) |
2264 | { | |
7b540d06 | 2265 | fmode_t mode = (fmode_t)(unsigned long)ptr; |
640708a2 PE |
2266 | struct proc_inode *ei; |
2267 | struct inode *inode; | |
2268 | ||
0168b9e3 | 2269 | inode = proc_pid_make_inode(dentry->d_sb, task, S_IFLNK | |
db978da8 AG |
2270 | ((mode & FMODE_READ ) ? S_IRUSR : 0) | |
2271 | ((mode & FMODE_WRITE) ? S_IWUSR : 0)); | |
640708a2 | 2272 | if (!inode) |
0168b9e3 | 2273 | return ERR_PTR(-ENOENT); |
640708a2 PE |
2274 | |
2275 | ei = PROC_I(inode); | |
6b255391 | 2276 | ei->op.proc_get_link = map_files_get_link; |
640708a2 | 2277 | |
bdb4d100 | 2278 | inode->i_op = &proc_map_files_link_inode_operations; |
640708a2 | 2279 | inode->i_size = 64; |
640708a2 PE |
2280 | |
2281 | d_set_d_op(dentry, &tid_map_files_dentry_operations); | |
0168b9e3 | 2282 | return d_splice_alias(inode, dentry); |
640708a2 PE |
2283 | } |
2284 | ||
2285 | static struct dentry *proc_map_files_lookup(struct inode *dir, | |
00cd8dd3 | 2286 | struct dentry *dentry, unsigned int flags) |
640708a2 PE |
2287 | { |
2288 | unsigned long vm_start, vm_end; | |
2289 | struct vm_area_struct *vma; | |
2290 | struct task_struct *task; | |
0168b9e3 | 2291 | struct dentry *result; |
640708a2 PE |
2292 | struct mm_struct *mm; |
2293 | ||
0168b9e3 | 2294 | result = ERR_PTR(-ENOENT); |
640708a2 PE |
2295 | task = get_proc_task(dir); |
2296 | if (!task) | |
2297 | goto out; | |
2298 | ||
0168b9e3 | 2299 | result = ERR_PTR(-EACCES); |
caaee623 | 2300 | if (!ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS)) |
640708a2 PE |
2301 | goto out_put_task; |
2302 | ||
0168b9e3 | 2303 | result = ERR_PTR(-ENOENT); |
640708a2 | 2304 | if (dname_to_vma_addr(dentry, &vm_start, &vm_end)) |
eb94cd96 | 2305 | goto out_put_task; |
640708a2 PE |
2306 | |
2307 | mm = get_task_mm(task); | |
2308 | if (!mm) | |
eb94cd96 | 2309 | goto out_put_task; |
640708a2 | 2310 | |
cd9e2bb8 | 2311 | result = ERR_PTR(-EINTR); |
d8ed45c5 | 2312 | if (mmap_read_lock_killable(mm)) |
cd9e2bb8 KK |
2313 | goto out_put_mm; |
2314 | ||
2315 | result = ERR_PTR(-ENOENT); | |
640708a2 PE |
2316 | vma = find_exact_vma(mm, vm_start, vm_end); |
2317 | if (!vma) | |
2318 | goto out_no_vma; | |
2319 | ||
05f56484 | 2320 | if (vma->vm_file) |
0168b9e3 | 2321 | result = proc_map_files_instantiate(dentry, task, |
05f56484 | 2322 | (void *)(unsigned long)vma->vm_file->f_mode); |
640708a2 PE |
2323 | |
2324 | out_no_vma: | |
d8ed45c5 | 2325 | mmap_read_unlock(mm); |
cd9e2bb8 | 2326 | out_put_mm: |
640708a2 | 2327 | mmput(mm); |
640708a2 PE |
2328 | out_put_task: |
2329 | put_task_struct(task); | |
2330 | out: | |
0168b9e3 | 2331 | return result; |
640708a2 PE |
2332 | } |
2333 | ||
2334 | static const struct inode_operations proc_map_files_inode_operations = { | |
2335 | .lookup = proc_map_files_lookup, | |
2336 | .permission = proc_fd_permission, | |
2337 | .setattr = proc_setattr, | |
2338 | }; | |
2339 | ||
2340 | static int | |
f0c3b509 | 2341 | proc_map_files_readdir(struct file *file, struct dir_context *ctx) |
640708a2 | 2342 | { |
640708a2 PE |
2343 | struct vm_area_struct *vma; |
2344 | struct task_struct *task; | |
2345 | struct mm_struct *mm; | |
f0c3b509 | 2346 | unsigned long nr_files, pos, i; |
94f8f3b0 | 2347 | GENRADIX(struct map_files_info) fa; |
f0c3b509 | 2348 | struct map_files_info *p; |
640708a2 | 2349 | int ret; |
5f14b924 | 2350 | struct vma_iterator vmi; |
640708a2 | 2351 | |
94f8f3b0 KO |
2352 | genradix_init(&fa); |
2353 | ||
640708a2 | 2354 | ret = -ENOENT; |
f0c3b509 | 2355 | task = get_proc_task(file_inode(file)); |
640708a2 PE |
2356 | if (!task) |
2357 | goto out; | |
2358 | ||
2359 | ret = -EACCES; | |
caaee623 | 2360 | if (!ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS)) |
640708a2 PE |
2361 | goto out_put_task; |
2362 | ||
2363 | ret = 0; | |
f0c3b509 AV |
2364 | if (!dir_emit_dots(file, ctx)) |
2365 | goto out_put_task; | |
640708a2 | 2366 | |
f0c3b509 AV |
2367 | mm = get_task_mm(task); |
2368 | if (!mm) | |
2369 | goto out_put_task; | |
cd9e2bb8 | 2370 | |
89154dd5 | 2371 | ret = mmap_read_lock_killable(mm); |
cd9e2bb8 KK |
2372 | if (ret) { |
2373 | mmput(mm); | |
2374 | goto out_put_task; | |
2375 | } | |
640708a2 | 2376 | |
f0c3b509 | 2377 | nr_files = 0; |
640708a2 | 2378 | |
f0c3b509 AV |
2379 | /* |
2380 | * We need two passes here: | |
2381 | * | |
c1e8d7c6 ML |
2382 | * 1) Collect vmas of mapped files with mmap_lock taken |
2383 | * 2) Release mmap_lock and instantiate entries | |
f0c3b509 AV |
2384 | * |
2385 | * otherwise we get lockdep complained, since filldir() | |
c1e8d7c6 | 2386 | * routine might require mmap_lock taken in might_fault(). |
f0c3b509 | 2387 | */ |
640708a2 | 2388 | |
5f14b924 LH |
2389 | pos = 2; |
2390 | vma_iter_init(&vmi, mm, 0); | |
2391 | for_each_vma(vmi, vma) { | |
94f8f3b0 KO |
2392 | if (!vma->vm_file) |
2393 | continue; | |
2394 | if (++pos <= ctx->pos) | |
2395 | continue; | |
f0c3b509 | 2396 | |
94f8f3b0 KO |
2397 | p = genradix_ptr_alloc(&fa, nr_files++, GFP_KERNEL); |
2398 | if (!p) { | |
f0c3b509 | 2399 | ret = -ENOMEM; |
89154dd5 | 2400 | mmap_read_unlock(mm); |
f0c3b509 AV |
2401 | mmput(mm); |
2402 | goto out_put_task; | |
640708a2 | 2403 | } |
f0c3b509 | 2404 | |
94f8f3b0 KO |
2405 | p->start = vma->vm_start; |
2406 | p->end = vma->vm_end; | |
2407 | p->mode = vma->vm_file->f_mode; | |
640708a2 | 2408 | } |
89154dd5 | 2409 | mmap_read_unlock(mm); |
fe079a5e | 2410 | mmput(mm); |
f0c3b509 AV |
2411 | |
2412 | for (i = 0; i < nr_files; i++) { | |
20d28cde AD |
2413 | char buf[4 * sizeof(long) + 2]; /* max: %lx-%lx\0 */ |
2414 | unsigned int len; | |
2415 | ||
94f8f3b0 | 2416 | p = genradix_ptr(&fa, i); |
20d28cde | 2417 | len = snprintf(buf, sizeof(buf), "%lx-%lx", p->start, p->end); |
f0c3b509 | 2418 | if (!proc_fill_cache(file, ctx, |
20d28cde | 2419 | buf, len, |
f0c3b509 AV |
2420 | proc_map_files_instantiate, |
2421 | task, | |
2422 | (void *)(unsigned long)p->mode)) | |
2423 | break; | |
2424 | ctx->pos++; | |
640708a2 PE |
2425 | } |
2426 | ||
640708a2 PE |
2427 | out_put_task: |
2428 | put_task_struct(task); | |
2429 | out: | |
94f8f3b0 | 2430 | genradix_free(&fa); |
640708a2 PE |
2431 | return ret; |
2432 | } | |
2433 | ||
2434 | static const struct file_operations proc_map_files_operations = { | |
2435 | .read = generic_read_dir, | |
f50752ea AV |
2436 | .iterate_shared = proc_map_files_readdir, |
2437 | .llseek = generic_file_llseek, | |
640708a2 PE |
2438 | }; |
2439 | ||
b18b6a9c | 2440 | #if defined(CONFIG_CHECKPOINT_RESTORE) && defined(CONFIG_POSIX_TIMERS) |
48f6a7a5 PE |
2441 | struct timers_private { |
2442 | struct pid *pid; | |
2443 | struct task_struct *task; | |
2444 | struct sighand_struct *sighand; | |
57b8015e | 2445 | struct pid_namespace *ns; |
48f6a7a5 PE |
2446 | unsigned long flags; |
2447 | }; | |
2448 | ||
2449 | static void *timers_start(struct seq_file *m, loff_t *pos) | |
2450 | { | |
2451 | struct timers_private *tp = m->private; | |
2452 | ||
2453 | tp->task = get_pid_task(tp->pid, PIDTYPE_PID); | |
2454 | if (!tp->task) | |
2455 | return ERR_PTR(-ESRCH); | |
2456 | ||
2457 | tp->sighand = lock_task_sighand(tp->task, &tp->flags); | |
2458 | if (!tp->sighand) | |
2459 | return ERR_PTR(-ESRCH); | |
2460 | ||
2461 | return seq_list_start(&tp->task->signal->posix_timers, *pos); | |
2462 | } | |
2463 | ||
2464 | static void *timers_next(struct seq_file *m, void *v, loff_t *pos) | |
2465 | { | |
2466 | struct timers_private *tp = m->private; | |
2467 | return seq_list_next(v, &tp->task->signal->posix_timers, pos); | |
2468 | } | |
2469 | ||
2470 | static void timers_stop(struct seq_file *m, void *v) | |
2471 | { | |
2472 | struct timers_private *tp = m->private; | |
2473 | ||
2474 | if (tp->sighand) { | |
2475 | unlock_task_sighand(tp->task, &tp->flags); | |
2476 | tp->sighand = NULL; | |
2477 | } | |
2478 | ||
2479 | if (tp->task) { | |
2480 | put_task_struct(tp->task); | |
2481 | tp->task = NULL; | |
2482 | } | |
2483 | } | |
2484 | ||
2485 | static int show_timer(struct seq_file *m, void *v) | |
2486 | { | |
2487 | struct k_itimer *timer; | |
57b8015e PE |
2488 | struct timers_private *tp = m->private; |
2489 | int notify; | |
cedbccab | 2490 | static const char * const nstr[] = { |
57b8015e PE |
2491 | [SIGEV_SIGNAL] = "signal", |
2492 | [SIGEV_NONE] = "none", | |
2493 | [SIGEV_THREAD] = "thread", | |
2494 | }; | |
48f6a7a5 PE |
2495 | |
2496 | timer = list_entry((struct list_head *)v, struct k_itimer, list); | |
57b8015e PE |
2497 | notify = timer->it_sigev_notify; |
2498 | ||
48f6a7a5 | 2499 | seq_printf(m, "ID: %d\n", timer->it_id); |
ba3edf1f | 2500 | seq_printf(m, "signal: %d/%px\n", |
25ce3191 JP |
2501 | timer->sigq->info.si_signo, |
2502 | timer->sigq->info.si_value.sival_ptr); | |
57b8015e | 2503 | seq_printf(m, "notify: %s/%s.%d\n", |
25ce3191 JP |
2504 | nstr[notify & ~SIGEV_THREAD_ID], |
2505 | (notify & SIGEV_THREAD_ID) ? "tid" : "pid", | |
2506 | pid_nr_ns(timer->it_pid, tp->ns)); | |
15ef0298 | 2507 | seq_printf(m, "ClockID: %d\n", timer->it_clock); |
48f6a7a5 PE |
2508 | |
2509 | return 0; | |
2510 | } | |
2511 | ||
2512 | static const struct seq_operations proc_timers_seq_ops = { | |
2513 | .start = timers_start, | |
2514 | .next = timers_next, | |
2515 | .stop = timers_stop, | |
2516 | .show = show_timer, | |
2517 | }; | |
2518 | ||
2519 | static int proc_timers_open(struct inode *inode, struct file *file) | |
2520 | { | |
2521 | struct timers_private *tp; | |
2522 | ||
2523 | tp = __seq_open_private(file, &proc_timers_seq_ops, | |
2524 | sizeof(struct timers_private)); | |
2525 | if (!tp) | |
2526 | return -ENOMEM; | |
2527 | ||
2528 | tp->pid = proc_pid(inode); | |
9d78edea | 2529 | tp->ns = proc_pid_ns(inode->i_sb); |
48f6a7a5 PE |
2530 | return 0; |
2531 | } | |
2532 | ||
2533 | static const struct file_operations proc_timers_operations = { | |
2534 | .open = proc_timers_open, | |
2535 | .read = seq_read, | |
2536 | .llseek = seq_lseek, | |
2537 | .release = seq_release_private, | |
2538 | }; | |
b5946bea | 2539 | #endif |
640708a2 | 2540 | |
5de23d43 JS |
2541 | static ssize_t timerslack_ns_write(struct file *file, const char __user *buf, |
2542 | size_t count, loff_t *offset) | |
2543 | { | |
2544 | struct inode *inode = file_inode(file); | |
2545 | struct task_struct *p; | |
2546 | u64 slack_ns; | |
2547 | int err; | |
2548 | ||
2549 | err = kstrtoull_from_user(buf, count, 10, &slack_ns); | |
2550 | if (err < 0) | |
2551 | return err; | |
2552 | ||
2553 | p = get_proc_task(inode); | |
2554 | if (!p) | |
2555 | return -ESRCH; | |
2556 | ||
4b2bd5fe | 2557 | if (p != current) { |
8da0b4f6 BG |
2558 | rcu_read_lock(); |
2559 | if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) { | |
2560 | rcu_read_unlock(); | |
4b2bd5fe JS |
2561 | count = -EPERM; |
2562 | goto out; | |
2563 | } | |
8da0b4f6 | 2564 | rcu_read_unlock(); |
5de23d43 | 2565 | |
4b2bd5fe JS |
2566 | err = security_task_setscheduler(p); |
2567 | if (err) { | |
2568 | count = err; | |
2569 | goto out; | |
2570 | } | |
904763e1 JS |
2571 | } |
2572 | ||
7abbaf94 JS |
2573 | task_lock(p); |
2574 | if (slack_ns == 0) | |
2575 | p->timer_slack_ns = p->default_timer_slack_ns; | |
2576 | else | |
2577 | p->timer_slack_ns = slack_ns; | |
2578 | task_unlock(p); | |
2579 | ||
2580 | out: | |
5de23d43 JS |
2581 | put_task_struct(p); |
2582 | ||
2583 | return count; | |
2584 | } | |
2585 | ||
2586 | static int timerslack_ns_show(struct seq_file *m, void *v) | |
2587 | { | |
2588 | struct inode *inode = m->private; | |
2589 | struct task_struct *p; | |
7abbaf94 | 2590 | int err = 0; |
5de23d43 JS |
2591 | |
2592 | p = get_proc_task(inode); | |
2593 | if (!p) | |
2594 | return -ESRCH; | |
2595 | ||
4b2bd5fe | 2596 | if (p != current) { |
8da0b4f6 BG |
2597 | rcu_read_lock(); |
2598 | if (!ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE)) { | |
2599 | rcu_read_unlock(); | |
4b2bd5fe JS |
2600 | err = -EPERM; |
2601 | goto out; | |
2602 | } | |
8da0b4f6 BG |
2603 | rcu_read_unlock(); |
2604 | ||
4b2bd5fe JS |
2605 | err = security_task_getscheduler(p); |
2606 | if (err) | |
2607 | goto out; | |
2608 | } | |
904763e1 | 2609 | |
7abbaf94 JS |
2610 | task_lock(p); |
2611 | seq_printf(m, "%llu\n", p->timer_slack_ns); | |
2612 | task_unlock(p); | |
2613 | ||
2614 | out: | |
5de23d43 JS |
2615 | put_task_struct(p); |
2616 | ||
2617 | return err; | |
2618 | } | |
2619 | ||
2620 | static int timerslack_ns_open(struct inode *inode, struct file *filp) | |
2621 | { | |
2622 | return single_open(filp, timerslack_ns_show, inode); | |
2623 | } | |
2624 | ||
2625 | static const struct file_operations proc_pid_set_timerslack_ns_operations = { | |
2626 | .open = timerslack_ns_open, | |
2627 | .read = seq_read, | |
2628 | .write = timerslack_ns_write, | |
2629 | .llseek = seq_lseek, | |
2630 | .release = single_release, | |
2631 | }; | |
2632 | ||
0168b9e3 AV |
2633 | static struct dentry *proc_pident_instantiate(struct dentry *dentry, |
2634 | struct task_struct *task, const void *ptr) | |
444ceed8 | 2635 | { |
c5141e6d | 2636 | const struct pid_entry *p = ptr; |
444ceed8 EB |
2637 | struct inode *inode; |
2638 | struct proc_inode *ei; | |
444ceed8 | 2639 | |
0168b9e3 | 2640 | inode = proc_pid_make_inode(dentry->d_sb, task, p->mode); |
444ceed8 | 2641 | if (!inode) |
0168b9e3 | 2642 | return ERR_PTR(-ENOENT); |
444ceed8 EB |
2643 | |
2644 | ei = PROC_I(inode); | |
444ceed8 | 2645 | if (S_ISDIR(inode->i_mode)) |
bfe86848 | 2646 | set_nlink(inode, 2); /* Use getattr to fix if necessary */ |
444ceed8 EB |
2647 | if (p->iop) |
2648 | inode->i_op = p->iop; | |
2649 | if (p->fop) | |
2650 | inode->i_fop = p->fop; | |
2651 | ei->op = p->op; | |
1bbc5513 | 2652 | pid_update_inode(task, inode); |
fb045adb | 2653 | d_set_d_op(dentry, &pid_dentry_operations); |
0168b9e3 | 2654 | return d_splice_alias(inode, dentry); |
444ceed8 EB |
2655 | } |
2656 | ||
1da177e4 LT |
2657 | static struct dentry *proc_pident_lookup(struct inode *dir, |
2658 | struct dentry *dentry, | |
d5a572a4 AD |
2659 | const struct pid_entry *p, |
2660 | const struct pid_entry *end) | |
1da177e4 | 2661 | { |
99f89551 | 2662 | struct task_struct *task = get_proc_task(dir); |
0168b9e3 | 2663 | struct dentry *res = ERR_PTR(-ENOENT); |
1da177e4 | 2664 | |
99f89551 EB |
2665 | if (!task) |
2666 | goto out_no_task; | |
1da177e4 | 2667 | |
20cdc894 EB |
2668 | /* |
2669 | * Yes, it does not scale. And it should not. Don't add | |
2670 | * new entries into /proc/<tgid>/ without very good reasons. | |
2671 | */ | |
d5a572a4 | 2672 | for (; p < end; p++) { |
1da177e4 LT |
2673 | if (p->len != dentry->d_name.len) |
2674 | continue; | |
26b95137 AD |
2675 | if (!memcmp(dentry->d_name.name, p->name, p->len)) { |
2676 | res = proc_pident_instantiate(dentry, task, p); | |
1da177e4 | 2677 | break; |
26b95137 | 2678 | } |
1da177e4 | 2679 | } |
99f89551 EB |
2680 | put_task_struct(task); |
2681 | out_no_task: | |
0168b9e3 | 2682 | return res; |
1da177e4 LT |
2683 | } |
2684 | ||
f0c3b509 | 2685 | static int proc_pident_readdir(struct file *file, struct dir_context *ctx, |
c5141e6d | 2686 | const struct pid_entry *ents, unsigned int nents) |
28a6d671 | 2687 | { |
f0c3b509 AV |
2688 | struct task_struct *task = get_proc_task(file_inode(file)); |
2689 | const struct pid_entry *p; | |
28a6d671 | 2690 | |
28a6d671 | 2691 | if (!task) |
f0c3b509 | 2692 | return -ENOENT; |
28a6d671 | 2693 | |
f0c3b509 AV |
2694 | if (!dir_emit_dots(file, ctx)) |
2695 | goto out; | |
2696 | ||
2697 | if (ctx->pos >= nents + 2) | |
2698 | goto out; | |
28a6d671 | 2699 | |
bac5f5d5 | 2700 | for (p = ents + (ctx->pos - 2); p < ents + nents; p++) { |
f0c3b509 AV |
2701 | if (!proc_fill_cache(file, ctx, p->name, p->len, |
2702 | proc_pident_instantiate, task, p)) | |
2703 | break; | |
2704 | ctx->pos++; | |
2705 | } | |
28a6d671 | 2706 | out: |
61a28784 | 2707 | put_task_struct(task); |
f0c3b509 | 2708 | return 0; |
1da177e4 LT |
2709 | } |
2710 | ||
28a6d671 | 2711 | #ifdef CONFIG_SECURITY |
591a22c1 KC |
2712 | static int proc_pid_attr_open(struct inode *inode, struct file *file) |
2713 | { | |
94f0b2d4 LT |
2714 | file->private_data = NULL; |
2715 | __mem_open(inode, file, PTRACE_MODE_READ_FSCREDS); | |
2716 | return 0; | |
591a22c1 KC |
2717 | } |
2718 | ||
28a6d671 EB |
2719 | static ssize_t proc_pid_attr_read(struct file * file, char __user * buf, |
2720 | size_t count, loff_t *ppos) | |
2721 | { | |
496ad9aa | 2722 | struct inode * inode = file_inode(file); |
04ff9708 | 2723 | char *p = NULL; |
28a6d671 EB |
2724 | ssize_t length; |
2725 | struct task_struct *task = get_proc_task(inode); | |
2726 | ||
28a6d671 | 2727 | if (!task) |
04ff9708 | 2728 | return -ESRCH; |
28a6d671 | 2729 | |
267c068e | 2730 | length = security_getprocattr(task, PROC_I(inode)->op.lsmid, |
c8e477c6 | 2731 | file->f_path.dentry->d_name.name, |
04ff9708 | 2732 | &p); |
28a6d671 | 2733 | put_task_struct(task); |
04ff9708 AV |
2734 | if (length > 0) |
2735 | length = simple_read_from_buffer(buf, count, ppos, p, length); | |
2736 | kfree(p); | |
28a6d671 | 2737 | return length; |
1da177e4 LT |
2738 | } |
2739 | ||
28a6d671 EB |
2740 | static ssize_t proc_pid_attr_write(struct file * file, const char __user * buf, |
2741 | size_t count, loff_t *ppos) | |
2742 | { | |
496ad9aa | 2743 | struct inode * inode = file_inode(file); |
41089b6d | 2744 | struct task_struct *task; |
bb646cdb | 2745 | void *page; |
41089b6d | 2746 | int rv; |
b21507e2 | 2747 | |
bfb819ea | 2748 | /* A task may only write when it was the opener. */ |
591a22c1 | 2749 | if (file->private_data != current->mm) |
bfb819ea KC |
2750 | return -EPERM; |
2751 | ||
41089b6d AD |
2752 | rcu_read_lock(); |
2753 | task = pid_task(proc_pid(inode), PIDTYPE_PID); | |
2754 | if (!task) { | |
2755 | rcu_read_unlock(); | |
2756 | return -ESRCH; | |
2757 | } | |
b21507e2 | 2758 | /* A task may only write its own attributes. */ |
41089b6d AD |
2759 | if (current != task) { |
2760 | rcu_read_unlock(); | |
2761 | return -EACCES; | |
2762 | } | |
35a196be PM |
2763 | /* Prevent changes to overridden credentials. */ |
2764 | if (current_cred() != current_real_cred()) { | |
2765 | rcu_read_unlock(); | |
2766 | return -EBUSY; | |
2767 | } | |
41089b6d | 2768 | rcu_read_unlock(); |
b21507e2 | 2769 | |
28a6d671 EB |
2770 | if (count > PAGE_SIZE) |
2771 | count = PAGE_SIZE; | |
2772 | ||
2773 | /* No partial writes. */ | |
28a6d671 | 2774 | if (*ppos != 0) |
41089b6d | 2775 | return -EINVAL; |
28a6d671 | 2776 | |
bb646cdb AV |
2777 | page = memdup_user(buf, count); |
2778 | if (IS_ERR(page)) { | |
41089b6d | 2779 | rv = PTR_ERR(page); |
28a6d671 | 2780 | goto out; |
bb646cdb | 2781 | } |
28a6d671 | 2782 | |
107db7c7 | 2783 | /* Guard against adverse ptrace interaction */ |
41089b6d AD |
2784 | rv = mutex_lock_interruptible(¤t->signal->cred_guard_mutex); |
2785 | if (rv < 0) | |
107db7c7 DH |
2786 | goto out_free; |
2787 | ||
267c068e | 2788 | rv = security_setprocattr(PROC_I(inode)->op.lsmid, |
6d9c939d CS |
2789 | file->f_path.dentry->d_name.name, page, |
2790 | count); | |
b21507e2 | 2791 | mutex_unlock(¤t->signal->cred_guard_mutex); |
28a6d671 | 2792 | out_free: |
bb646cdb | 2793 | kfree(page); |
28a6d671 | 2794 | out: |
41089b6d | 2795 | return rv; |
28a6d671 EB |
2796 | } |
2797 | ||
00977a59 | 2798 | static const struct file_operations proc_pid_attr_operations = { |
591a22c1 | 2799 | .open = proc_pid_attr_open, |
28a6d671 EB |
2800 | .read = proc_pid_attr_read, |
2801 | .write = proc_pid_attr_write, | |
87df8424 | 2802 | .llseek = generic_file_llseek, |
591a22c1 | 2803 | .release = mem_release, |
28a6d671 EB |
2804 | }; |
2805 | ||
6d9c939d CS |
2806 | #define LSM_DIR_OPS(LSM) \ |
2807 | static int proc_##LSM##_attr_dir_iterate(struct file *filp, \ | |
2808 | struct dir_context *ctx) \ | |
2809 | { \ | |
2810 | return proc_pident_readdir(filp, ctx, \ | |
2811 | LSM##_attr_dir_stuff, \ | |
2812 | ARRAY_SIZE(LSM##_attr_dir_stuff)); \ | |
2813 | } \ | |
2814 | \ | |
2815 | static const struct file_operations proc_##LSM##_attr_dir_ops = { \ | |
2816 | .read = generic_read_dir, \ | |
0a2c2baa | 2817 | .iterate_shared = proc_##LSM##_attr_dir_iterate, \ |
6d9c939d CS |
2818 | .llseek = default_llseek, \ |
2819 | }; \ | |
2820 | \ | |
2821 | static struct dentry *proc_##LSM##_attr_dir_lookup(struct inode *dir, \ | |
2822 | struct dentry *dentry, unsigned int flags) \ | |
2823 | { \ | |
2824 | return proc_pident_lookup(dir, dentry, \ | |
2825 | LSM##_attr_dir_stuff, \ | |
d5a572a4 | 2826 | LSM##_attr_dir_stuff + ARRAY_SIZE(LSM##_attr_dir_stuff)); \ |
6d9c939d CS |
2827 | } \ |
2828 | \ | |
2829 | static const struct inode_operations proc_##LSM##_attr_dir_inode_ops = { \ | |
2830 | .lookup = proc_##LSM##_attr_dir_lookup, \ | |
2831 | .getattr = pid_getattr, \ | |
2832 | .setattr = proc_setattr, \ | |
2833 | } | |
2834 | ||
2835 | #ifdef CONFIG_SECURITY_SMACK | |
2836 | static const struct pid_entry smack_attr_dir_stuff[] = { | |
267c068e | 2837 | ATTR(LSM_ID_SMACK, "current", 0666), |
6d9c939d CS |
2838 | }; |
2839 | LSM_DIR_OPS(smack); | |
2840 | #endif | |
2841 | ||
6413f852 JJ |
2842 | #ifdef CONFIG_SECURITY_APPARMOR |
2843 | static const struct pid_entry apparmor_attr_dir_stuff[] = { | |
267c068e CS |
2844 | ATTR(LSM_ID_APPARMOR, "current", 0666), |
2845 | ATTR(LSM_ID_APPARMOR, "prev", 0444), | |
2846 | ATTR(LSM_ID_APPARMOR, "exec", 0666), | |
6413f852 JJ |
2847 | }; |
2848 | LSM_DIR_OPS(apparmor); | |
2849 | #endif | |
2850 | ||
c5141e6d | 2851 | static const struct pid_entry attr_dir_stuff[] = { |
267c068e CS |
2852 | ATTR(LSM_ID_UNDEF, "current", 0666), |
2853 | ATTR(LSM_ID_UNDEF, "prev", 0444), | |
2854 | ATTR(LSM_ID_UNDEF, "exec", 0666), | |
2855 | ATTR(LSM_ID_UNDEF, "fscreate", 0666), | |
2856 | ATTR(LSM_ID_UNDEF, "keycreate", 0666), | |
2857 | ATTR(LSM_ID_UNDEF, "sockcreate", 0666), | |
6d9c939d CS |
2858 | #ifdef CONFIG_SECURITY_SMACK |
2859 | DIR("smack", 0555, | |
2860 | proc_smack_attr_dir_inode_ops, proc_smack_attr_dir_ops), | |
2861 | #endif | |
6413f852 JJ |
2862 | #ifdef CONFIG_SECURITY_APPARMOR |
2863 | DIR("apparmor", 0555, | |
2864 | proc_apparmor_attr_dir_inode_ops, proc_apparmor_attr_dir_ops), | |
2865 | #endif | |
28a6d671 EB |
2866 | }; |
2867 | ||
f0c3b509 | 2868 | static int proc_attr_dir_readdir(struct file *file, struct dir_context *ctx) |
28a6d671 | 2869 | { |
f0c3b509 AV |
2870 | return proc_pident_readdir(file, ctx, |
2871 | attr_dir_stuff, ARRAY_SIZE(attr_dir_stuff)); | |
28a6d671 EB |
2872 | } |
2873 | ||
00977a59 | 2874 | static const struct file_operations proc_attr_dir_operations = { |
1da177e4 | 2875 | .read = generic_read_dir, |
f50752ea AV |
2876 | .iterate_shared = proc_attr_dir_readdir, |
2877 | .llseek = generic_file_llseek, | |
1da177e4 LT |
2878 | }; |
2879 | ||
72d9dcfc | 2880 | static struct dentry *proc_attr_dir_lookup(struct inode *dir, |
00cd8dd3 | 2881 | struct dentry *dentry, unsigned int flags) |
28a6d671 | 2882 | { |
7bcd6b0e | 2883 | return proc_pident_lookup(dir, dentry, |
d5a572a4 AD |
2884 | attr_dir_stuff, |
2885 | attr_dir_stuff + ARRAY_SIZE(attr_dir_stuff)); | |
28a6d671 EB |
2886 | } |
2887 | ||
c5ef1c42 | 2888 | static const struct inode_operations proc_attr_dir_inode_operations = { |
72d9dcfc | 2889 | .lookup = proc_attr_dir_lookup, |
99f89551 | 2890 | .getattr = pid_getattr, |
6d76fa58 | 2891 | .setattr = proc_setattr, |
1da177e4 LT |
2892 | }; |
2893 | ||
28a6d671 EB |
2894 | #endif |
2895 | ||
698ba7b5 | 2896 | #ifdef CONFIG_ELF_CORE |
3cb4a0bb KH |
2897 | static ssize_t proc_coredump_filter_read(struct file *file, char __user *buf, |
2898 | size_t count, loff_t *ppos) | |
2899 | { | |
496ad9aa | 2900 | struct task_struct *task = get_proc_task(file_inode(file)); |
3cb4a0bb KH |
2901 | struct mm_struct *mm; |
2902 | char buffer[PROC_NUMBUF]; | |
2903 | size_t len; | |
2904 | int ret; | |
2905 | ||
2906 | if (!task) | |
2907 | return -ESRCH; | |
2908 | ||
2909 | ret = 0; | |
2910 | mm = get_task_mm(task); | |
2911 | if (mm) { | |
2912 | len = snprintf(buffer, sizeof(buffer), "%08lx\n", | |
2913 | ((mm->flags & MMF_DUMP_FILTER_MASK) >> | |
2914 | MMF_DUMP_FILTER_SHIFT)); | |
2915 | mmput(mm); | |
2916 | ret = simple_read_from_buffer(buf, count, ppos, buffer, len); | |
2917 | } | |
2918 | ||
2919 | put_task_struct(task); | |
2920 | ||
2921 | return ret; | |
2922 | } | |
2923 | ||
2924 | static ssize_t proc_coredump_filter_write(struct file *file, | |
2925 | const char __user *buf, | |
2926 | size_t count, | |
2927 | loff_t *ppos) | |
2928 | { | |
2929 | struct task_struct *task; | |
2930 | struct mm_struct *mm; | |
3cb4a0bb KH |
2931 | unsigned int val; |
2932 | int ret; | |
2933 | int i; | |
2934 | unsigned long mask; | |
2935 | ||
774636e1 AD |
2936 | ret = kstrtouint_from_user(buf, count, 0, &val); |
2937 | if (ret < 0) | |
2938 | return ret; | |
3cb4a0bb KH |
2939 | |
2940 | ret = -ESRCH; | |
496ad9aa | 2941 | task = get_proc_task(file_inode(file)); |
3cb4a0bb KH |
2942 | if (!task) |
2943 | goto out_no_task; | |
2944 | ||
3cb4a0bb KH |
2945 | mm = get_task_mm(task); |
2946 | if (!mm) | |
2947 | goto out_no_mm; | |
41a0c249 | 2948 | ret = 0; |
3cb4a0bb KH |
2949 | |
2950 | for (i = 0, mask = 1; i < MMF_DUMP_FILTER_BITS; i++, mask <<= 1) { | |
2951 | if (val & mask) | |
2952 | set_bit(i + MMF_DUMP_FILTER_SHIFT, &mm->flags); | |
2953 | else | |
2954 | clear_bit(i + MMF_DUMP_FILTER_SHIFT, &mm->flags); | |
2955 | } | |
2956 | ||
2957 | mmput(mm); | |
2958 | out_no_mm: | |
2959 | put_task_struct(task); | |
2960 | out_no_task: | |
774636e1 AD |
2961 | if (ret < 0) |
2962 | return ret; | |
2963 | return count; | |
3cb4a0bb KH |
2964 | } |
2965 | ||
2966 | static const struct file_operations proc_coredump_filter_operations = { | |
2967 | .read = proc_coredump_filter_read, | |
2968 | .write = proc_coredump_filter_write, | |
87df8424 | 2969 | .llseek = generic_file_llseek, |
3cb4a0bb KH |
2970 | }; |
2971 | #endif | |
2972 | ||
aba76fdb | 2973 | #ifdef CONFIG_TASK_IO_ACCOUNTING |
19aadc98 | 2974 | static int do_io_accounting(struct task_struct *task, struct seq_file *m, int whole) |
297c5d92 | 2975 | { |
23202220 | 2976 | struct task_io_accounting acct; |
293eb1e7 | 2977 | int result; |
5995477a | 2978 | |
f7cfd871 | 2979 | result = down_read_killable(&task->signal->exec_update_lock); |
293eb1e7 VK |
2980 | if (result) |
2981 | return result; | |
2982 | ||
caaee623 | 2983 | if (!ptrace_may_access(task, PTRACE_MODE_READ_FSCREDS)) { |
293eb1e7 VK |
2984 | result = -EACCES; |
2985 | goto out_unlock; | |
2986 | } | |
1d1221f3 | 2987 | |
1df4bd83 | 2988 | if (whole) { |
23202220 ON |
2989 | struct signal_struct *sig = task->signal; |
2990 | struct task_struct *t; | |
1df4bd83 ON |
2991 | unsigned int seq = 1; |
2992 | unsigned long flags; | |
2993 | ||
2994 | rcu_read_lock(); | |
2995 | do { | |
2996 | seq++; /* 2 on the 1st/lockless path, otherwise odd */ | |
2997 | flags = read_seqbegin_or_lock_irqsave(&sig->stats_lock, &seq); | |
5995477a | 2998 | |
1df4bd83 ON |
2999 | acct = sig->ioac; |
3000 | __for_each_thread(sig, t) | |
3001 | task_io_accounting_add(&acct, &t->ioac); | |
5995477a | 3002 | |
1df4bd83 ON |
3003 | } while (need_seqretry(&sig->stats_lock, seq)); |
3004 | done_seqretry_irqrestore(&sig->stats_lock, seq, flags); | |
3005 | rcu_read_unlock(); | |
23202220 ON |
3006 | } else { |
3007 | acct = task->ioac; | |
297c5d92 | 3008 | } |
23202220 | 3009 | |
25ce3191 JP |
3010 | seq_printf(m, |
3011 | "rchar: %llu\n" | |
3012 | "wchar: %llu\n" | |
3013 | "syscr: %llu\n" | |
3014 | "syscw: %llu\n" | |
3015 | "read_bytes: %llu\n" | |
3016 | "write_bytes: %llu\n" | |
3017 | "cancelled_write_bytes: %llu\n", | |
3018 | (unsigned long long)acct.rchar, | |
3019 | (unsigned long long)acct.wchar, | |
3020 | (unsigned long long)acct.syscr, | |
3021 | (unsigned long long)acct.syscw, | |
3022 | (unsigned long long)acct.read_bytes, | |
3023 | (unsigned long long)acct.write_bytes, | |
3024 | (unsigned long long)acct.cancelled_write_bytes); | |
3025 | result = 0; | |
3026 | ||
293eb1e7 | 3027 | out_unlock: |
f7cfd871 | 3028 | up_read(&task->signal->exec_update_lock); |
293eb1e7 | 3029 | return result; |
297c5d92 AR |
3030 | } |
3031 | ||
19aadc98 AD |
3032 | static int proc_tid_io_accounting(struct seq_file *m, struct pid_namespace *ns, |
3033 | struct pid *pid, struct task_struct *task) | |
297c5d92 | 3034 | { |
19aadc98 | 3035 | return do_io_accounting(task, m, 0); |
aba76fdb | 3036 | } |
297c5d92 | 3037 | |
19aadc98 AD |
3038 | static int proc_tgid_io_accounting(struct seq_file *m, struct pid_namespace *ns, |
3039 | struct pid *pid, struct task_struct *task) | |
297c5d92 | 3040 | { |
19aadc98 | 3041 | return do_io_accounting(task, m, 1); |
297c5d92 AR |
3042 | } |
3043 | #endif /* CONFIG_TASK_IO_ACCOUNTING */ | |
aba76fdb | 3044 | |
22d917d8 EB |
3045 | #ifdef CONFIG_USER_NS |
3046 | static int proc_id_map_open(struct inode *inode, struct file *file, | |
ccf94f1b | 3047 | const struct seq_operations *seq_ops) |
22d917d8 EB |
3048 | { |
3049 | struct user_namespace *ns = NULL; | |
3050 | struct task_struct *task; | |
3051 | struct seq_file *seq; | |
3052 | int ret = -EINVAL; | |
3053 | ||
3054 | task = get_proc_task(inode); | |
3055 | if (task) { | |
3056 | rcu_read_lock(); | |
3057 | ns = get_user_ns(task_cred_xxx(task, user_ns)); | |
3058 | rcu_read_unlock(); | |
3059 | put_task_struct(task); | |
3060 | } | |
3061 | if (!ns) | |
3062 | goto err; | |
3063 | ||
3064 | ret = seq_open(file, seq_ops); | |
3065 | if (ret) | |
3066 | goto err_put_ns; | |
3067 | ||
3068 | seq = file->private_data; | |
3069 | seq->private = ns; | |
3070 | ||
3071 | return 0; | |
3072 | err_put_ns: | |
3073 | put_user_ns(ns); | |
3074 | err: | |
3075 | return ret; | |
3076 | } | |
3077 | ||
3078 | static int proc_id_map_release(struct inode *inode, struct file *file) | |
3079 | { | |
3080 | struct seq_file *seq = file->private_data; | |
3081 | struct user_namespace *ns = seq->private; | |
3082 | put_user_ns(ns); | |
3083 | return seq_release(inode, file); | |
3084 | } | |
3085 | ||
3086 | static int proc_uid_map_open(struct inode *inode, struct file *file) | |
3087 | { | |
3088 | return proc_id_map_open(inode, file, &proc_uid_seq_operations); | |
3089 | } | |
3090 | ||
3091 | static int proc_gid_map_open(struct inode *inode, struct file *file) | |
3092 | { | |
3093 | return proc_id_map_open(inode, file, &proc_gid_seq_operations); | |
3094 | } | |
3095 | ||
f76d207a EB |
3096 | static int proc_projid_map_open(struct inode *inode, struct file *file) |
3097 | { | |
3098 | return proc_id_map_open(inode, file, &proc_projid_seq_operations); | |
3099 | } | |
3100 | ||
22d917d8 EB |
3101 | static const struct file_operations proc_uid_map_operations = { |
3102 | .open = proc_uid_map_open, | |
3103 | .write = proc_uid_map_write, | |
3104 | .read = seq_read, | |
3105 | .llseek = seq_lseek, | |
3106 | .release = proc_id_map_release, | |
3107 | }; | |
3108 | ||
3109 | static const struct file_operations proc_gid_map_operations = { | |
3110 | .open = proc_gid_map_open, | |
3111 | .write = proc_gid_map_write, | |
3112 | .read = seq_read, | |
3113 | .llseek = seq_lseek, | |
3114 | .release = proc_id_map_release, | |
3115 | }; | |
f76d207a EB |
3116 | |
3117 | static const struct file_operations proc_projid_map_operations = { | |
3118 | .open = proc_projid_map_open, | |
3119 | .write = proc_projid_map_write, | |
3120 | .read = seq_read, | |
3121 | .llseek = seq_lseek, | |
3122 | .release = proc_id_map_release, | |
3123 | }; | |
9cc46516 EB |
3124 | |
3125 | static int proc_setgroups_open(struct inode *inode, struct file *file) | |
3126 | { | |
3127 | struct user_namespace *ns = NULL; | |
3128 | struct task_struct *task; | |
3129 | int ret; | |
3130 | ||
3131 | ret = -ESRCH; | |
3132 | task = get_proc_task(inode); | |
3133 | if (task) { | |
3134 | rcu_read_lock(); | |
3135 | ns = get_user_ns(task_cred_xxx(task, user_ns)); | |
3136 | rcu_read_unlock(); | |
3137 | put_task_struct(task); | |
3138 | } | |
3139 | if (!ns) | |
3140 | goto err; | |
3141 | ||
3142 | if (file->f_mode & FMODE_WRITE) { | |
3143 | ret = -EACCES; | |
3144 | if (!ns_capable(ns, CAP_SYS_ADMIN)) | |
3145 | goto err_put_ns; | |
3146 | } | |
3147 | ||
3148 | ret = single_open(file, &proc_setgroups_show, ns); | |
3149 | if (ret) | |
3150 | goto err_put_ns; | |
3151 | ||
3152 | return 0; | |
3153 | err_put_ns: | |
3154 | put_user_ns(ns); | |
3155 | err: | |
3156 | return ret; | |
3157 | } | |
3158 | ||
3159 | static int proc_setgroups_release(struct inode *inode, struct file *file) | |
3160 | { | |
3161 | struct seq_file *seq = file->private_data; | |
3162 | struct user_namespace *ns = seq->private; | |
3163 | int ret = single_release(inode, file); | |
3164 | put_user_ns(ns); | |
3165 | return ret; | |
3166 | } | |
3167 | ||
3168 | static const struct file_operations proc_setgroups_operations = { | |
3169 | .open = proc_setgroups_open, | |
3170 | .write = proc_setgroups_write, | |
3171 | .read = seq_read, | |
3172 | .llseek = seq_lseek, | |
3173 | .release = proc_setgroups_release, | |
3174 | }; | |
22d917d8 EB |
3175 | #endif /* CONFIG_USER_NS */ |
3176 | ||
47830723 KC |
3177 | static int proc_pid_personality(struct seq_file *m, struct pid_namespace *ns, |
3178 | struct pid *pid, struct task_struct *task) | |
3179 | { | |
a9712bc1 AV |
3180 | int err = lock_trace(task); |
3181 | if (!err) { | |
3182 | seq_printf(m, "%08x\n", task->personality); | |
3183 | unlock_trace(task); | |
3184 | } | |
3185 | return err; | |
47830723 KC |
3186 | } |
3187 | ||
7c23b330 JP |
3188 | #ifdef CONFIG_LIVEPATCH |
3189 | static int proc_pid_patch_state(struct seq_file *m, struct pid_namespace *ns, | |
3190 | struct pid *pid, struct task_struct *task) | |
3191 | { | |
3192 | seq_printf(m, "%d\n", task->patch_state); | |
3193 | return 0; | |
3194 | } | |
3195 | #endif /* CONFIG_LIVEPATCH */ | |
3196 | ||
76093853 | 3197 | #ifdef CONFIG_KSM |
3198 | static int proc_pid_ksm_merging_pages(struct seq_file *m, struct pid_namespace *ns, | |
3199 | struct pid *pid, struct task_struct *task) | |
3200 | { | |
3201 | struct mm_struct *mm; | |
3202 | ||
3203 | mm = get_task_mm(task); | |
3204 | if (mm) { | |
3205 | seq_printf(m, "%lu\n", mm->ksm_merging_pages); | |
3206 | mmput(mm); | |
3207 | } | |
3208 | ||
3209 | return 0; | |
3210 | } | |
cb4df4ca | 3211 | static int proc_pid_ksm_stat(struct seq_file *m, struct pid_namespace *ns, |
3212 | struct pid *pid, struct task_struct *task) | |
3213 | { | |
3214 | struct mm_struct *mm; | |
3215 | ||
3216 | mm = get_task_mm(task); | |
3217 | if (mm) { | |
3218 | seq_printf(m, "ksm_rmap_items %lu\n", mm->ksm_rmap_items); | |
6080d19f | 3219 | seq_printf(m, "ksm_zero_pages %lu\n", mm->ksm_zero_pages); |
d21077fb SR |
3220 | seq_printf(m, "ksm_merging_pages %lu\n", mm->ksm_merging_pages); |
3221 | seq_printf(m, "ksm_process_profit %ld\n", ksm_process_profit(mm)); | |
cb4df4ca | 3222 | mmput(mm); |
3223 | } | |
3224 | ||
3225 | return 0; | |
3226 | } | |
76093853 | 3227 | #endif /* CONFIG_KSM */ |
3228 | ||
c8d12627 AP |
3229 | #ifdef CONFIG_STACKLEAK_METRICS |
3230 | static int proc_stack_depth(struct seq_file *m, struct pid_namespace *ns, | |
3231 | struct pid *pid, struct task_struct *task) | |
3232 | { | |
3233 | unsigned long prev_depth = THREAD_SIZE - | |
3234 | (task->prev_lowest_stack & (THREAD_SIZE - 1)); | |
3235 | unsigned long depth = THREAD_SIZE - | |
3236 | (task->lowest_stack & (THREAD_SIZE - 1)); | |
3237 | ||
3238 | seq_printf(m, "previous stack depth: %lu\nstack depth: %lu\n", | |
3239 | prev_depth, depth); | |
3240 | return 0; | |
3241 | } | |
3242 | #endif /* CONFIG_STACKLEAK_METRICS */ | |
3243 | ||
28a6d671 EB |
3244 | /* |
3245 | * Thread groups | |
3246 | */ | |
00977a59 | 3247 | static const struct file_operations proc_task_operations; |
c5ef1c42 | 3248 | static const struct inode_operations proc_task_inode_operations; |
20cdc894 | 3249 | |
c5141e6d | 3250 | static const struct pid_entry tgid_base_stuff[] = { |
631f9c18 AD |
3251 | DIR("task", S_IRUGO|S_IXUGO, proc_task_inode_operations, proc_task_operations), |
3252 | DIR("fd", S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations), | |
640708a2 | 3253 | DIR("map_files", S_IRUSR|S_IXUSR, proc_map_files_inode_operations, proc_map_files_operations), |
7bc3fa01 | 3254 | DIR("fdinfo", S_IRUGO|S_IXUGO, proc_fdinfo_inode_operations, proc_fdinfo_operations), |
6b4e306a | 3255 | DIR("ns", S_IRUSR|S_IXUGO, proc_ns_dir_inode_operations, proc_ns_dir_operations), |
b2211a36 | 3256 | #ifdef CONFIG_NET |
631f9c18 | 3257 | DIR("net", S_IRUGO|S_IXUGO, proc_net_inode_operations, proc_net_operations), |
b2211a36 | 3258 | #endif |
631f9c18 | 3259 | REG("environ", S_IRUSR, proc_environ_operations), |
c5317167 | 3260 | REG("auxv", S_IRUSR, proc_auxv_operations), |
631f9c18 | 3261 | ONE("status", S_IRUGO, proc_pid_status), |
35a35046 | 3262 | ONE("personality", S_IRUSR, proc_pid_personality), |
1c963eb1 | 3263 | ONE("limits", S_IRUGO, proc_pid_limits), |
43ae34cb | 3264 | #ifdef CONFIG_SCHED_DEBUG |
631f9c18 | 3265 | REG("sched", S_IRUGO|S_IWUSR, proc_pid_sched_operations), |
5091faa4 MG |
3266 | #endif |
3267 | #ifdef CONFIG_SCHED_AUTOGROUP | |
3268 | REG("autogroup", S_IRUGO|S_IWUSR, proc_pid_sched_autogroup_operations), | |
04a8682a AV |
3269 | #endif |
3270 | #ifdef CONFIG_TIME_NS | |
3271 | REG("timens_offsets", S_IRUGO|S_IWUSR, proc_timens_offsets_operations), | |
ebcb6734 | 3272 | #endif |
4614a696 | 3273 | REG("comm", S_IRUGO|S_IWUSR, proc_pid_set_comm_operations), |
ebcb6734 | 3274 | #ifdef CONFIG_HAVE_ARCH_TRACEHOOK |
09d93bd6 | 3275 | ONE("syscall", S_IRUSR, proc_pid_syscall), |
43ae34cb | 3276 | #endif |
c2c0bb44 | 3277 | REG("cmdline", S_IRUGO, proc_pid_cmdline_ops), |
631f9c18 AD |
3278 | ONE("stat", S_IRUGO, proc_tgid_stat), |
3279 | ONE("statm", S_IRUGO, proc_pid_statm), | |
b7643757 | 3280 | REG("maps", S_IRUGO, proc_pid_maps_operations), |
28a6d671 | 3281 | #ifdef CONFIG_NUMA |
b7643757 | 3282 | REG("numa_maps", S_IRUGO, proc_pid_numa_maps_operations), |
28a6d671 | 3283 | #endif |
631f9c18 AD |
3284 | REG("mem", S_IRUSR|S_IWUSR, proc_mem_operations), |
3285 | LNK("cwd", proc_cwd_link), | |
3286 | LNK("root", proc_root_link), | |
3287 | LNK("exe", proc_exe_link), | |
3288 | REG("mounts", S_IRUGO, proc_mounts_operations), | |
3289 | REG("mountinfo", S_IRUGO, proc_mountinfo_operations), | |
3290 | REG("mountstats", S_IRUSR, proc_mountstats_operations), | |
1e883281 | 3291 | #ifdef CONFIG_PROC_PAGE_MONITOR |
631f9c18 | 3292 | REG("clear_refs", S_IWUSR, proc_clear_refs_operations), |
b7643757 | 3293 | REG("smaps", S_IRUGO, proc_pid_smaps_operations), |
493b0e9d | 3294 | REG("smaps_rollup", S_IRUGO, proc_pid_smaps_rollup_operations), |
32ed74a4 | 3295 | REG("pagemap", S_IRUSR, proc_pagemap_operations), |
28a6d671 EB |
3296 | #endif |
3297 | #ifdef CONFIG_SECURITY | |
631f9c18 | 3298 | DIR("attr", S_IRUGO|S_IXUGO, proc_attr_dir_inode_operations, proc_attr_dir_operations), |
28a6d671 EB |
3299 | #endif |
3300 | #ifdef CONFIG_KALLSYMS | |
edfcd606 | 3301 | ONE("wchan", S_IRUGO, proc_pid_wchan), |
28a6d671 | 3302 | #endif |
2ec220e2 | 3303 | #ifdef CONFIG_STACKTRACE |
35a35046 | 3304 | ONE("stack", S_IRUSR, proc_pid_stack), |
28a6d671 | 3305 | #endif |
5968cece | 3306 | #ifdef CONFIG_SCHED_INFO |
f6e826ca | 3307 | ONE("schedstat", S_IRUGO, proc_pid_schedstat), |
28a6d671 | 3308 | #endif |
9745512c | 3309 | #ifdef CONFIG_LATENCYTOP |
631f9c18 | 3310 | REG("latency", S_IRUGO, proc_lstats_operations), |
9745512c | 3311 | #endif |
8793d854 | 3312 | #ifdef CONFIG_PROC_PID_CPUSET |
52de4779 | 3313 | ONE("cpuset", S_IRUGO, proc_cpuset_show), |
a424316c PM |
3314 | #endif |
3315 | #ifdef CONFIG_CGROUPS | |
006f4ac4 | 3316 | ONE("cgroup", S_IRUGO, proc_cgroup_show), |
e79f15a4 CY |
3317 | #endif |
3318 | #ifdef CONFIG_PROC_CPU_RESCTRL | |
3319 | ONE("cpu_resctrl_groups", S_IRUGO, proc_resctrl_show), | |
28a6d671 | 3320 | #endif |
6ba51e37 | 3321 | ONE("oom_score", S_IRUGO, proc_oom_score), |
fa0cbbf1 | 3322 | REG("oom_adj", S_IRUGO|S_IWUSR, proc_oom_adj_operations), |
a63d83f4 | 3323 | REG("oom_score_adj", S_IRUGO|S_IWUSR, proc_oom_score_adj_operations), |
4b7d248b | 3324 | #ifdef CONFIG_AUDIT |
631f9c18 AD |
3325 | REG("loginuid", S_IWUSR|S_IRUGO, proc_loginuid_operations), |
3326 | REG("sessionid", S_IRUGO, proc_sessionid_operations), | |
28a6d671 | 3327 | #endif |
f4f154fd | 3328 | #ifdef CONFIG_FAULT_INJECTION |
631f9c18 | 3329 | REG("make-it-fail", S_IRUGO|S_IWUSR, proc_fault_inject_operations), |
168c42bc | 3330 | REG("fail-nth", 0644, proc_fail_nth_operations), |
f4f154fd | 3331 | #endif |
698ba7b5 | 3332 | #ifdef CONFIG_ELF_CORE |
631f9c18 | 3333 | REG("coredump_filter", S_IRUGO|S_IWUSR, proc_coredump_filter_operations), |
3cb4a0bb | 3334 | #endif |
aba76fdb | 3335 | #ifdef CONFIG_TASK_IO_ACCOUNTING |
19aadc98 | 3336 | ONE("io", S_IRUSR, proc_tgid_io_accounting), |
aba76fdb | 3337 | #endif |
22d917d8 EB |
3338 | #ifdef CONFIG_USER_NS |
3339 | REG("uid_map", S_IRUGO|S_IWUSR, proc_uid_map_operations), | |
3340 | REG("gid_map", S_IRUGO|S_IWUSR, proc_gid_map_operations), | |
f76d207a | 3341 | REG("projid_map", S_IRUGO|S_IWUSR, proc_projid_map_operations), |
9cc46516 | 3342 | REG("setgroups", S_IRUGO|S_IWUSR, proc_setgroups_operations), |
22d917d8 | 3343 | #endif |
b18b6a9c | 3344 | #if defined(CONFIG_CHECKPOINT_RESTORE) && defined(CONFIG_POSIX_TIMERS) |
48f6a7a5 PE |
3345 | REG("timers", S_IRUGO, proc_timers_operations), |
3346 | #endif | |
5de23d43 | 3347 | REG("timerslack_ns", S_IRUGO|S_IWUGO, proc_pid_set_timerslack_ns_operations), |
7c23b330 JP |
3348 | #ifdef CONFIG_LIVEPATCH |
3349 | ONE("patch_state", S_IRUSR, proc_pid_patch_state), | |
3350 | #endif | |
c8d12627 AP |
3351 | #ifdef CONFIG_STACKLEAK_METRICS |
3352 | ONE("stack_depth", S_IRUGO, proc_stack_depth), | |
3353 | #endif | |
68bc30bb AL |
3354 | #ifdef CONFIG_PROC_PID_ARCH_STATUS |
3355 | ONE("arch_status", S_IRUGO, proc_pid_arch_status), | |
3356 | #endif | |
0d8315dd YZ |
3357 | #ifdef CONFIG_SECCOMP_CACHE_DEBUG |
3358 | ONE("seccomp_cache", S_IRUSR, proc_pid_seccomp_cache), | |
3359 | #endif | |
76093853 | 3360 | #ifdef CONFIG_KSM |
3361 | ONE("ksm_merging_pages", S_IRUSR, proc_pid_ksm_merging_pages), | |
cb4df4ca | 3362 | ONE("ksm_stat", S_IRUSR, proc_pid_ksm_stat), |
76093853 | 3363 | #endif |
28a6d671 | 3364 | }; |
1da177e4 | 3365 | |
f0c3b509 | 3366 | static int proc_tgid_base_readdir(struct file *file, struct dir_context *ctx) |
1da177e4 | 3367 | { |
f0c3b509 AV |
3368 | return proc_pident_readdir(file, ctx, |
3369 | tgid_base_stuff, ARRAY_SIZE(tgid_base_stuff)); | |
1da177e4 LT |
3370 | } |
3371 | ||
00977a59 | 3372 | static const struct file_operations proc_tgid_base_operations = { |
1da177e4 | 3373 | .read = generic_read_dir, |
f50752ea AV |
3374 | .iterate_shared = proc_tgid_base_readdir, |
3375 | .llseek = generic_file_llseek, | |
1da177e4 LT |
3376 | }; |
3377 | ||
3eb39f47 CB |
3378 | struct pid *tgid_pidfd_to_pid(const struct file *file) |
3379 | { | |
30d158b1 | 3380 | if (file->f_op != &proc_tgid_base_operations) |
3eb39f47 CB |
3381 | return ERR_PTR(-EBADF); |
3382 | ||
3383 | return proc_pid(file_inode(file)); | |
3384 | } | |
3385 | ||
00cd8dd3 AV |
3386 | static struct dentry *proc_tgid_base_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags) |
3387 | { | |
7bcd6b0e | 3388 | return proc_pident_lookup(dir, dentry, |
d5a572a4 AD |
3389 | tgid_base_stuff, |
3390 | tgid_base_stuff + ARRAY_SIZE(tgid_base_stuff)); | |
1da177e4 LT |
3391 | } |
3392 | ||
c5ef1c42 | 3393 | static const struct inode_operations proc_tgid_base_inode_operations = { |
28a6d671 | 3394 | .lookup = proc_tgid_base_lookup, |
99f89551 | 3395 | .getattr = pid_getattr, |
6d76fa58 | 3396 | .setattr = proc_setattr, |
0499680a | 3397 | .permission = proc_pid_permission, |
1da177e4 | 3398 | }; |
1da177e4 | 3399 | |
0895e91d | 3400 | /** |
7bc3e6e5 EB |
3401 | * proc_flush_pid - Remove dcache entries for @pid from the /proc dcache. |
3402 | * @pid: pid that should be flushed. | |
0895e91d | 3403 | * |
7bc3e6e5 EB |
3404 | * This function walks a list of inodes (that belong to any proc |
3405 | * filesystem) that are attached to the pid and flushes them from | |
3406 | * the dentry cache. | |
0895e91d RD |
3407 | * |
3408 | * It is safe and reasonable to cache /proc entries for a task until | |
3409 | * that task exits. After that they just clog up the dcache with | |
3410 | * useless entries, possibly causing useful dcache entries to be | |
7bc3e6e5 EB |
3411 | * flushed instead. This routine is provided to flush those useless |
3412 | * dcache entries when a process is reaped. | |
0895e91d RD |
3413 | * |
3414 | * NOTE: This routine is just an optimization so it does not guarantee | |
7bc3e6e5 EB |
3415 | * that no dcache entries will exist after a process is reaped |
3416 | * it just makes it very unlikely that any will persist. | |
60347f67 PE |
3417 | */ |
3418 | ||
7bc3e6e5 | 3419 | void proc_flush_pid(struct pid *pid) |
60347f67 | 3420 | { |
63f818f4 | 3421 | proc_invalidate_siblings_dcache(&pid->inodes, &pid->lock); |
60347f67 PE |
3422 | } |
3423 | ||
0168b9e3 | 3424 | static struct dentry *proc_pid_instantiate(struct dentry * dentry, |
c52a47ac | 3425 | struct task_struct *task, const void *ptr) |
444ceed8 | 3426 | { |
444ceed8 EB |
3427 | struct inode *inode; |
3428 | ||
d919a1e7 ZC |
3429 | inode = proc_pid_make_base_inode(dentry->d_sb, task, |
3430 | S_IFDIR | S_IRUGO | S_IXUGO); | |
444ceed8 | 3431 | if (!inode) |
0168b9e3 | 3432 | return ERR_PTR(-ENOENT); |
444ceed8 | 3433 | |
444ceed8 EB |
3434 | inode->i_op = &proc_tgid_base_inode_operations; |
3435 | inode->i_fop = &proc_tgid_base_operations; | |
3436 | inode->i_flags|=S_IMMUTABLE; | |
aed54175 | 3437 | |
1270dd8d | 3438 | set_nlink(inode, nlink_tgid); |
1bbc5513 | 3439 | pid_update_inode(task, inode); |
444ceed8 | 3440 | |
fb045adb | 3441 | d_set_d_op(dentry, &pid_dentry_operations); |
0168b9e3 | 3442 | return d_splice_alias(inode, dentry); |
444ceed8 EB |
3443 | } |
3444 | ||
867aaccf | 3445 | struct dentry *proc_pid_lookup(struct dentry *dentry, unsigned int flags) |
1da177e4 LT |
3446 | { |
3447 | struct task_struct *task; | |
1da177e4 | 3448 | unsigned tgid; |
fa10fed3 | 3449 | struct proc_fs_info *fs_info; |
b488893a | 3450 | struct pid_namespace *ns; |
0168b9e3 | 3451 | struct dentry *result = ERR_PTR(-ENOENT); |
1da177e4 | 3452 | |
dbcdb504 | 3453 | tgid = name_to_int(&dentry->d_name); |
1da177e4 LT |
3454 | if (tgid == ~0U) |
3455 | goto out; | |
3456 | ||
fa10fed3 AG |
3457 | fs_info = proc_sb_info(dentry->d_sb); |
3458 | ns = fs_info->pid_ns; | |
de758734 | 3459 | rcu_read_lock(); |
b488893a | 3460 | task = find_task_by_pid_ns(tgid, ns); |
1da177e4 LT |
3461 | if (task) |
3462 | get_task_struct(task); | |
de758734 | 3463 | rcu_read_unlock(); |
1da177e4 LT |
3464 | if (!task) |
3465 | goto out; | |
3466 | ||
24a71ce5 AG |
3467 | /* Limit procfs to only ptraceable tasks */ |
3468 | if (fs_info->hide_pid == HIDEPID_NOT_PTRACEABLE) { | |
3469 | if (!has_pid_permissions(fs_info, task, HIDEPID_NO_ACCESS)) | |
3470 | goto out_put_task; | |
3471 | } | |
3472 | ||
0168b9e3 | 3473 | result = proc_pid_instantiate(dentry, task, NULL); |
24a71ce5 | 3474 | out_put_task: |
1da177e4 | 3475 | put_task_struct(task); |
1da177e4 | 3476 | out: |
0168b9e3 | 3477 | return result; |
1da177e4 LT |
3478 | } |
3479 | ||
1da177e4 | 3480 | /* |
0804ef4b | 3481 | * Find the first task with tgid >= tgid |
0bc58a91 | 3482 | * |
1da177e4 | 3483 | */ |
19fd4bb2 EB |
3484 | struct tgid_iter { |
3485 | unsigned int tgid; | |
0804ef4b | 3486 | struct task_struct *task; |
19fd4bb2 EB |
3487 | }; |
3488 | static struct tgid_iter next_tgid(struct pid_namespace *ns, struct tgid_iter iter) | |
3489 | { | |
0804ef4b | 3490 | struct pid *pid; |
1da177e4 | 3491 | |
19fd4bb2 EB |
3492 | if (iter.task) |
3493 | put_task_struct(iter.task); | |
454cc105 | 3494 | rcu_read_lock(); |
0804ef4b | 3495 | retry: |
19fd4bb2 EB |
3496 | iter.task = NULL; |
3497 | pid = find_ge_pid(iter.tgid, ns); | |
0804ef4b | 3498 | if (pid) { |
19fd4bb2 | 3499 | iter.tgid = pid_nr_ns(pid, ns); |
3147d8aa EB |
3500 | iter.task = pid_task(pid, PIDTYPE_TGID); |
3501 | if (!iter.task) { | |
19fd4bb2 | 3502 | iter.tgid += 1; |
0804ef4b | 3503 | goto retry; |
19fd4bb2 EB |
3504 | } |
3505 | get_task_struct(iter.task); | |
0bc58a91 | 3506 | } |
454cc105 | 3507 | rcu_read_unlock(); |
19fd4bb2 | 3508 | return iter; |
1da177e4 LT |
3509 | } |
3510 | ||
0097875b | 3511 | #define TGID_OFFSET (FIRST_PROCESS_ENTRY + 2) |
0804ef4b | 3512 | |
1da177e4 | 3513 | /* for the /proc/ directory itself, after non-process stuff has been done */ |
f0c3b509 | 3514 | int proc_pid_readdir(struct file *file, struct dir_context *ctx) |
1da177e4 | 3515 | { |
19fd4bb2 | 3516 | struct tgid_iter iter; |
fa10fed3 | 3517 | struct proc_fs_info *fs_info = proc_sb_info(file_inode(file)->i_sb); |
9d78edea | 3518 | struct pid_namespace *ns = proc_pid_ns(file_inode(file)->i_sb); |
f0c3b509 | 3519 | loff_t pos = ctx->pos; |
1da177e4 | 3520 | |
021ada7d | 3521 | if (pos >= PID_MAX_LIMIT + TGID_OFFSET) |
f0c3b509 | 3522 | return 0; |
1da177e4 | 3523 | |
0097875b | 3524 | if (pos == TGID_OFFSET - 2) { |
fa10fed3 | 3525 | struct inode *inode = d_inode(fs_info->proc_self); |
db963164 | 3526 | if (!dir_emit(ctx, "self", 4, inode->i_ino, DT_LNK)) |
f0c3b509 | 3527 | return 0; |
0097875b EB |
3528 | ctx->pos = pos = pos + 1; |
3529 | } | |
3530 | if (pos == TGID_OFFSET - 1) { | |
fa10fed3 | 3531 | struct inode *inode = d_inode(fs_info->proc_thread_self); |
0097875b EB |
3532 | if (!dir_emit(ctx, "thread-self", 11, inode->i_ino, DT_LNK)) |
3533 | return 0; | |
3534 | ctx->pos = pos = pos + 1; | |
021ada7d | 3535 | } |
0097875b | 3536 | iter.tgid = pos - TGID_OFFSET; |
19fd4bb2 | 3537 | iter.task = NULL; |
19fd4bb2 EB |
3538 | for (iter = next_tgid(ns, iter); |
3539 | iter.task; | |
3540 | iter.tgid += 1, iter = next_tgid(ns, iter)) { | |
e3912ac3 | 3541 | char name[10 + 1]; |
a4ef3895 | 3542 | unsigned int len; |
3ba4bcee ED |
3543 | |
3544 | cond_resched(); | |
fa10fed3 | 3545 | if (!has_pid_permissions(fs_info, iter.task, HIDEPID_INVISIBLE)) |
f0c3b509 | 3546 | continue; |
0499680a | 3547 | |
e3912ac3 | 3548 | len = snprintf(name, sizeof(name), "%u", iter.tgid); |
f0c3b509 AV |
3549 | ctx->pos = iter.tgid + TGID_OFFSET; |
3550 | if (!proc_fill_cache(file, ctx, name, len, | |
3551 | proc_pid_instantiate, iter.task, NULL)) { | |
19fd4bb2 | 3552 | put_task_struct(iter.task); |
f0c3b509 | 3553 | return 0; |
1da177e4 | 3554 | } |
0bc58a91 | 3555 | } |
f0c3b509 | 3556 | ctx->pos = PID_MAX_LIMIT + TGID_OFFSET; |
0bc58a91 EB |
3557 | return 0; |
3558 | } | |
1da177e4 | 3559 | |
1b3044e3 JD |
3560 | /* |
3561 | * proc_tid_comm_permission is a special permission function exclusively | |
3562 | * used for the node /proc/<pid>/task/<tid>/comm. | |
3563 | * It bypasses generic permission checks in the case where a task of the same | |
3564 | * task group attempts to access the node. | |
3565 | * The rationale behind this is that glibc and bionic access this node for | |
3566 | * cross thread naming (pthread_set/getname_np(!self)). However, if | |
3567 | * PR_SET_DUMPABLE gets set to 0 this node among others becomes uid=0 gid=0, | |
3568 | * which locks out the cross thread naming implementation. | |
3569 | * This function makes sure that the node is always accessible for members of | |
3570 | * same thread group. | |
3571 | */ | |
4609e1f1 | 3572 | static int proc_tid_comm_permission(struct mnt_idmap *idmap, |
549c7297 | 3573 | struct inode *inode, int mask) |
1b3044e3 JD |
3574 | { |
3575 | bool is_same_tgroup; | |
3576 | struct task_struct *task; | |
3577 | ||
3578 | task = get_proc_task(inode); | |
3579 | if (!task) | |
3580 | return -ESRCH; | |
3581 | is_same_tgroup = same_thread_group(current, task); | |
3582 | put_task_struct(task); | |
3583 | ||
3584 | if (likely(is_same_tgroup && !(mask & MAY_EXEC))) { | |
3585 | /* This file (/proc/<pid>/task/<tid>/comm) can always be | |
3586 | * read or written by the members of the corresponding | |
3587 | * thread group. | |
3588 | */ | |
3589 | return 0; | |
3590 | } | |
3591 | ||
4609e1f1 | 3592 | return generic_permission(&nop_mnt_idmap, inode, mask); |
1b3044e3 JD |
3593 | } |
3594 | ||
3595 | static const struct inode_operations proc_tid_comm_inode_operations = { | |
ccf61486 AS |
3596 | .setattr = proc_setattr, |
3597 | .permission = proc_tid_comm_permission, | |
1b3044e3 JD |
3598 | }; |
3599 | ||
28a6d671 EB |
3600 | /* |
3601 | * Tasks | |
3602 | */ | |
c5141e6d | 3603 | static const struct pid_entry tid_base_stuff[] = { |
631f9c18 | 3604 | DIR("fd", S_IRUSR|S_IXUSR, proc_fd_inode_operations, proc_fd_operations), |
7bc3fa01 | 3605 | DIR("fdinfo", S_IRUGO|S_IXUGO, proc_fdinfo_inode_operations, proc_fdinfo_operations), |
6b4e306a | 3606 | DIR("ns", S_IRUSR|S_IXUGO, proc_ns_dir_inode_operations, proc_ns_dir_operations), |
6ba8ed79 EB |
3607 | #ifdef CONFIG_NET |
3608 | DIR("net", S_IRUGO|S_IXUGO, proc_net_inode_operations, proc_net_operations), | |
3609 | #endif | |
631f9c18 | 3610 | REG("environ", S_IRUSR, proc_environ_operations), |
c5317167 | 3611 | REG("auxv", S_IRUSR, proc_auxv_operations), |
631f9c18 | 3612 | ONE("status", S_IRUGO, proc_pid_status), |
35a35046 | 3613 | ONE("personality", S_IRUSR, proc_pid_personality), |
1c963eb1 | 3614 | ONE("limits", S_IRUGO, proc_pid_limits), |
43ae34cb | 3615 | #ifdef CONFIG_SCHED_DEBUG |
631f9c18 | 3616 | REG("sched", S_IRUGO|S_IWUSR, proc_pid_sched_operations), |
ebcb6734 | 3617 | #endif |
1b3044e3 JD |
3618 | NOD("comm", S_IFREG|S_IRUGO|S_IWUSR, |
3619 | &proc_tid_comm_inode_operations, | |
3620 | &proc_pid_set_comm_operations, {}), | |
ebcb6734 | 3621 | #ifdef CONFIG_HAVE_ARCH_TRACEHOOK |
09d93bd6 | 3622 | ONE("syscall", S_IRUSR, proc_pid_syscall), |
43ae34cb | 3623 | #endif |
c2c0bb44 | 3624 | REG("cmdline", S_IRUGO, proc_pid_cmdline_ops), |
631f9c18 AD |
3625 | ONE("stat", S_IRUGO, proc_tid_stat), |
3626 | ONE("statm", S_IRUGO, proc_pid_statm), | |
871305bb | 3627 | REG("maps", S_IRUGO, proc_pid_maps_operations), |
2e13ba54 | 3628 | #ifdef CONFIG_PROC_CHILDREN |
81841161 CG |
3629 | REG("children", S_IRUGO, proc_tid_children_operations), |
3630 | #endif | |
28a6d671 | 3631 | #ifdef CONFIG_NUMA |
871305bb | 3632 | REG("numa_maps", S_IRUGO, proc_pid_numa_maps_operations), |
28a6d671 | 3633 | #endif |
631f9c18 AD |
3634 | REG("mem", S_IRUSR|S_IWUSR, proc_mem_operations), |
3635 | LNK("cwd", proc_cwd_link), | |
3636 | LNK("root", proc_root_link), | |
3637 | LNK("exe", proc_exe_link), | |
3638 | REG("mounts", S_IRUGO, proc_mounts_operations), | |
3639 | REG("mountinfo", S_IRUGO, proc_mountinfo_operations), | |
1e883281 | 3640 | #ifdef CONFIG_PROC_PAGE_MONITOR |
631f9c18 | 3641 | REG("clear_refs", S_IWUSR, proc_clear_refs_operations), |
871305bb | 3642 | REG("smaps", S_IRUGO, proc_pid_smaps_operations), |
493b0e9d | 3643 | REG("smaps_rollup", S_IRUGO, proc_pid_smaps_rollup_operations), |
32ed74a4 | 3644 | REG("pagemap", S_IRUSR, proc_pagemap_operations), |
28a6d671 EB |
3645 | #endif |
3646 | #ifdef CONFIG_SECURITY | |
631f9c18 | 3647 | DIR("attr", S_IRUGO|S_IXUGO, proc_attr_dir_inode_operations, proc_attr_dir_operations), |
28a6d671 EB |
3648 | #endif |
3649 | #ifdef CONFIG_KALLSYMS | |
edfcd606 | 3650 | ONE("wchan", S_IRUGO, proc_pid_wchan), |
28a6d671 | 3651 | #endif |
2ec220e2 | 3652 | #ifdef CONFIG_STACKTRACE |
35a35046 | 3653 | ONE("stack", S_IRUSR, proc_pid_stack), |
28a6d671 | 3654 | #endif |
5968cece | 3655 | #ifdef CONFIG_SCHED_INFO |
f6e826ca | 3656 | ONE("schedstat", S_IRUGO, proc_pid_schedstat), |
28a6d671 | 3657 | #endif |
9745512c | 3658 | #ifdef CONFIG_LATENCYTOP |
631f9c18 | 3659 | REG("latency", S_IRUGO, proc_lstats_operations), |
9745512c | 3660 | #endif |
8793d854 | 3661 | #ifdef CONFIG_PROC_PID_CPUSET |
52de4779 | 3662 | ONE("cpuset", S_IRUGO, proc_cpuset_show), |
a424316c PM |
3663 | #endif |
3664 | #ifdef CONFIG_CGROUPS | |
006f4ac4 | 3665 | ONE("cgroup", S_IRUGO, proc_cgroup_show), |
e79f15a4 CY |
3666 | #endif |
3667 | #ifdef CONFIG_PROC_CPU_RESCTRL | |
3668 | ONE("cpu_resctrl_groups", S_IRUGO, proc_resctrl_show), | |
28a6d671 | 3669 | #endif |
6ba51e37 | 3670 | ONE("oom_score", S_IRUGO, proc_oom_score), |
fa0cbbf1 | 3671 | REG("oom_adj", S_IRUGO|S_IWUSR, proc_oom_adj_operations), |
a63d83f4 | 3672 | REG("oom_score_adj", S_IRUGO|S_IWUSR, proc_oom_score_adj_operations), |
4b7d248b | 3673 | #ifdef CONFIG_AUDIT |
631f9c18 | 3674 | REG("loginuid", S_IWUSR|S_IRUGO, proc_loginuid_operations), |
26ec3c64 | 3675 | REG("sessionid", S_IRUGO, proc_sessionid_operations), |
28a6d671 | 3676 | #endif |
f4f154fd | 3677 | #ifdef CONFIG_FAULT_INJECTION |
631f9c18 | 3678 | REG("make-it-fail", S_IRUGO|S_IWUSR, proc_fault_inject_operations), |
1203c8e6 | 3679 | REG("fail-nth", 0644, proc_fail_nth_operations), |
f4f154fd | 3680 | #endif |
297c5d92 | 3681 | #ifdef CONFIG_TASK_IO_ACCOUNTING |
19aadc98 | 3682 | ONE("io", S_IRUSR, proc_tid_io_accounting), |
297c5d92 | 3683 | #endif |
22d917d8 EB |
3684 | #ifdef CONFIG_USER_NS |
3685 | REG("uid_map", S_IRUGO|S_IWUSR, proc_uid_map_operations), | |
3686 | REG("gid_map", S_IRUGO|S_IWUSR, proc_gid_map_operations), | |
f76d207a | 3687 | REG("projid_map", S_IRUGO|S_IWUSR, proc_projid_map_operations), |
9cc46516 | 3688 | REG("setgroups", S_IRUGO|S_IWUSR, proc_setgroups_operations), |
22d917d8 | 3689 | #endif |
7c23b330 JP |
3690 | #ifdef CONFIG_LIVEPATCH |
3691 | ONE("patch_state", S_IRUSR, proc_pid_patch_state), | |
3692 | #endif | |
68bc30bb AL |
3693 | #ifdef CONFIG_PROC_PID_ARCH_STATUS |
3694 | ONE("arch_status", S_IRUGO, proc_pid_arch_status), | |
3695 | #endif | |
0d8315dd YZ |
3696 | #ifdef CONFIG_SECCOMP_CACHE_DEBUG |
3697 | ONE("seccomp_cache", S_IRUSR, proc_pid_seccomp_cache), | |
3698 | #endif | |
76093853 | 3699 | #ifdef CONFIG_KSM |
3700 | ONE("ksm_merging_pages", S_IRUSR, proc_pid_ksm_merging_pages), | |
cb4df4ca | 3701 | ONE("ksm_stat", S_IRUSR, proc_pid_ksm_stat), |
76093853 | 3702 | #endif |
28a6d671 EB |
3703 | }; |
3704 | ||
f0c3b509 | 3705 | static int proc_tid_base_readdir(struct file *file, struct dir_context *ctx) |
28a6d671 | 3706 | { |
f0c3b509 AV |
3707 | return proc_pident_readdir(file, ctx, |
3708 | tid_base_stuff, ARRAY_SIZE(tid_base_stuff)); | |
28a6d671 EB |
3709 | } |
3710 | ||
00cd8dd3 AV |
3711 | static struct dentry *proc_tid_base_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags) |
3712 | { | |
7bcd6b0e | 3713 | return proc_pident_lookup(dir, dentry, |
d5a572a4 AD |
3714 | tid_base_stuff, |
3715 | tid_base_stuff + ARRAY_SIZE(tid_base_stuff)); | |
28a6d671 EB |
3716 | } |
3717 | ||
00977a59 | 3718 | static const struct file_operations proc_tid_base_operations = { |
28a6d671 | 3719 | .read = generic_read_dir, |
f50752ea AV |
3720 | .iterate_shared = proc_tid_base_readdir, |
3721 | .llseek = generic_file_llseek, | |
28a6d671 EB |
3722 | }; |
3723 | ||
c5ef1c42 | 3724 | static const struct inode_operations proc_tid_base_inode_operations = { |
28a6d671 EB |
3725 | .lookup = proc_tid_base_lookup, |
3726 | .getattr = pid_getattr, | |
3727 | .setattr = proc_setattr, | |
3728 | }; | |
3729 | ||
0168b9e3 AV |
3730 | static struct dentry *proc_task_instantiate(struct dentry *dentry, |
3731 | struct task_struct *task, const void *ptr) | |
444ceed8 | 3732 | { |
444ceed8 | 3733 | struct inode *inode; |
d919a1e7 ZC |
3734 | inode = proc_pid_make_base_inode(dentry->d_sb, task, |
3735 | S_IFDIR | S_IRUGO | S_IXUGO); | |
444ceed8 | 3736 | if (!inode) |
0168b9e3 | 3737 | return ERR_PTR(-ENOENT); |
1bbc5513 | 3738 | |
444ceed8 EB |
3739 | inode->i_op = &proc_tid_base_inode_operations; |
3740 | inode->i_fop = &proc_tid_base_operations; | |
1bbc5513 | 3741 | inode->i_flags |= S_IMMUTABLE; |
aed54175 | 3742 | |
1270dd8d | 3743 | set_nlink(inode, nlink_tid); |
1bbc5513 | 3744 | pid_update_inode(task, inode); |
444ceed8 | 3745 | |
fb045adb | 3746 | d_set_d_op(dentry, &pid_dentry_operations); |
0168b9e3 | 3747 | return d_splice_alias(inode, dentry); |
444ceed8 EB |
3748 | } |
3749 | ||
00cd8dd3 | 3750 | static struct dentry *proc_task_lookup(struct inode *dir, struct dentry * dentry, unsigned int flags) |
28a6d671 | 3751 | { |
28a6d671 EB |
3752 | struct task_struct *task; |
3753 | struct task_struct *leader = get_proc_task(dir); | |
28a6d671 | 3754 | unsigned tid; |
fa10fed3 | 3755 | struct proc_fs_info *fs_info; |
b488893a | 3756 | struct pid_namespace *ns; |
0168b9e3 | 3757 | struct dentry *result = ERR_PTR(-ENOENT); |
28a6d671 EB |
3758 | |
3759 | if (!leader) | |
3760 | goto out_no_task; | |
3761 | ||
dbcdb504 | 3762 | tid = name_to_int(&dentry->d_name); |
28a6d671 EB |
3763 | if (tid == ~0U) |
3764 | goto out; | |
3765 | ||
fa10fed3 AG |
3766 | fs_info = proc_sb_info(dentry->d_sb); |
3767 | ns = fs_info->pid_ns; | |
28a6d671 | 3768 | rcu_read_lock(); |
b488893a | 3769 | task = find_task_by_pid_ns(tid, ns); |
28a6d671 EB |
3770 | if (task) |
3771 | get_task_struct(task); | |
3772 | rcu_read_unlock(); | |
3773 | if (!task) | |
3774 | goto out; | |
bac0abd6 | 3775 | if (!same_thread_group(leader, task)) |
28a6d671 EB |
3776 | goto out_drop_task; |
3777 | ||
0168b9e3 | 3778 | result = proc_task_instantiate(dentry, task, NULL); |
28a6d671 EB |
3779 | out_drop_task: |
3780 | put_task_struct(task); | |
3781 | out: | |
3782 | put_task_struct(leader); | |
3783 | out_no_task: | |
0168b9e3 | 3784 | return result; |
28a6d671 EB |
3785 | } |
3786 | ||
0bc58a91 EB |
3787 | /* |
3788 | * Find the first tid of a thread group to return to user space. | |
3789 | * | |
3790 | * Usually this is just the thread group leader, but if the users | |
3791 | * buffer was too small or there was a seek into the middle of the | |
3792 | * directory we have more work todo. | |
3793 | * | |
3794 | * In the case of a short read we start with find_task_by_pid. | |
3795 | * | |
3796 | * In the case of a seek we start with the leader and walk nr | |
3797 | * threads past it. | |
3798 | */ | |
9f6e963f ON |
3799 | static struct task_struct *first_tid(struct pid *pid, int tid, loff_t f_pos, |
3800 | struct pid_namespace *ns) | |
0bc58a91 | 3801 | { |
d855a4b7 | 3802 | struct task_struct *pos, *task; |
9f6e963f ON |
3803 | unsigned long nr = f_pos; |
3804 | ||
3805 | if (nr != f_pos) /* 32bit overflow? */ | |
3806 | return NULL; | |
1da177e4 | 3807 | |
cc288738 | 3808 | rcu_read_lock(); |
d855a4b7 ON |
3809 | task = pid_task(pid, PIDTYPE_PID); |
3810 | if (!task) | |
3811 | goto fail; | |
3812 | ||
3813 | /* Attempt to start with the tid of a thread */ | |
9f6e963f | 3814 | if (tid && nr) { |
b488893a | 3815 | pos = find_task_by_pid_ns(tid, ns); |
d855a4b7 | 3816 | if (pos && same_thread_group(pos, task)) |
a872ff0c | 3817 | goto found; |
0bc58a91 | 3818 | } |
1da177e4 | 3819 | |
0bc58a91 | 3820 | /* If nr exceeds the number of threads there is nothing todo */ |
9f6e963f | 3821 | if (nr >= get_nr_threads(task)) |
c986c14a | 3822 | goto fail; |
1da177e4 | 3823 | |
a872ff0c ON |
3824 | /* If we haven't found our starting place yet start |
3825 | * with the leader and walk nr threads forward. | |
0bc58a91 | 3826 | */ |
dce8f8ed | 3827 | for_each_thread(task, pos) { |
9f6e963f | 3828 | if (!nr--) |
c986c14a | 3829 | goto found; |
63993102 | 3830 | } |
c986c14a ON |
3831 | fail: |
3832 | pos = NULL; | |
3833 | goto out; | |
a872ff0c ON |
3834 | found: |
3835 | get_task_struct(pos); | |
3836 | out: | |
cc288738 | 3837 | rcu_read_unlock(); |
0bc58a91 EB |
3838 | return pos; |
3839 | } | |
3840 | ||
3841 | /* | |
3842 | * Find the next thread in the thread list. | |
3843 | * Return NULL if there is an error or no next thread. | |
3844 | * | |
3845 | * The reference to the input task_struct is released. | |
3846 | */ | |
3847 | static struct task_struct *next_tid(struct task_struct *start) | |
3848 | { | |
c1df7fb8 | 3849 | struct task_struct *pos = NULL; |
cc288738 | 3850 | rcu_read_lock(); |
c1df7fb8 | 3851 | if (pid_alive(start)) { |
33a98138 ON |
3852 | pos = __next_thread(start); |
3853 | if (pos) | |
c1df7fb8 ON |
3854 | get_task_struct(pos); |
3855 | } | |
cc288738 | 3856 | rcu_read_unlock(); |
0bc58a91 EB |
3857 | put_task_struct(start); |
3858 | return pos; | |
1da177e4 LT |
3859 | } |
3860 | ||
3861 | /* for the /proc/TGID/task/ directories */ | |
f0c3b509 | 3862 | static int proc_task_readdir(struct file *file, struct dir_context *ctx) |
1da177e4 | 3863 | { |
d855a4b7 ON |
3864 | struct inode *inode = file_inode(file); |
3865 | struct task_struct *task; | |
b488893a | 3866 | struct pid_namespace *ns; |
f0c3b509 | 3867 | int tid; |
1da177e4 | 3868 | |
d855a4b7 | 3869 | if (proc_inode_is_dead(inode)) |
f0c3b509 | 3870 | return -ENOENT; |
1da177e4 | 3871 | |
f0c3b509 | 3872 | if (!dir_emit_dots(file, ctx)) |
d855a4b7 | 3873 | return 0; |
1da177e4 | 3874 | |
0bc58a91 EB |
3875 | /* f_version caches the tgid value that the last readdir call couldn't |
3876 | * return. lseek aka telldir automagically resets f_version to 0. | |
3877 | */ | |
9d78edea | 3878 | ns = proc_pid_ns(inode->i_sb); |
f0c3b509 AV |
3879 | tid = (int)file->f_version; |
3880 | file->f_version = 0; | |
d855a4b7 | 3881 | for (task = first_tid(proc_pid(inode), tid, ctx->pos - 2, ns); |
0bc58a91 | 3882 | task; |
f0c3b509 | 3883 | task = next_tid(task), ctx->pos++) { |
e3912ac3 | 3884 | char name[10 + 1]; |
a4ef3895 | 3885 | unsigned int len; |
0658a096 | 3886 | |
b488893a | 3887 | tid = task_pid_nr_ns(task, ns); |
0658a096 FW |
3888 | if (!tid) |
3889 | continue; /* The task has just exited. */ | |
e3912ac3 | 3890 | len = snprintf(name, sizeof(name), "%u", tid); |
f0c3b509 AV |
3891 | if (!proc_fill_cache(file, ctx, name, len, |
3892 | proc_task_instantiate, task, NULL)) { | |
0bc58a91 EB |
3893 | /* returning this tgid failed, save it as the first |
3894 | * pid for the next readir call */ | |
f0c3b509 | 3895 | file->f_version = (u64)tid; |
0bc58a91 | 3896 | put_task_struct(task); |
1da177e4 | 3897 | break; |
0bc58a91 | 3898 | } |
1da177e4 | 3899 | } |
d855a4b7 | 3900 | |
f0c3b509 | 3901 | return 0; |
1da177e4 | 3902 | } |
6e66b52b | 3903 | |
b74d24f7 | 3904 | static int proc_task_getattr(struct mnt_idmap *idmap, |
549c7297 | 3905 | const struct path *path, struct kstat *stat, |
a528d35e | 3906 | u32 request_mask, unsigned int query_flags) |
6e66b52b | 3907 | { |
a528d35e | 3908 | struct inode *inode = d_inode(path->dentry); |
99f89551 | 3909 | struct task_struct *p = get_proc_task(inode); |
0d72b928 | 3910 | generic_fillattr(&nop_mnt_idmap, request_mask, inode, stat); |
6e66b52b | 3911 | |
99f89551 | 3912 | if (p) { |
99f89551 | 3913 | stat->nlink += get_nr_threads(p); |
99f89551 | 3914 | put_task_struct(p); |
6e66b52b EB |
3915 | } |
3916 | ||
3917 | return 0; | |
3918 | } | |
28a6d671 | 3919 | |
c5ef1c42 | 3920 | static const struct inode_operations proc_task_inode_operations = { |
28a6d671 EB |
3921 | .lookup = proc_task_lookup, |
3922 | .getattr = proc_task_getattr, | |
3923 | .setattr = proc_setattr, | |
0499680a | 3924 | .permission = proc_pid_permission, |
28a6d671 EB |
3925 | }; |
3926 | ||
00977a59 | 3927 | static const struct file_operations proc_task_operations = { |
28a6d671 | 3928 | .read = generic_read_dir, |
f50752ea AV |
3929 | .iterate_shared = proc_task_readdir, |
3930 | .llseek = generic_file_llseek, | |
28a6d671 | 3931 | }; |
1270dd8d AD |
3932 | |
3933 | void __init set_proc_pid_nlink(void) | |
3934 | { | |
3935 | nlink_tid = pid_entry_nlink(tid_base_stuff, ARRAY_SIZE(tid_base_stuff)); | |
3936 | nlink_tgid = pid_entry_nlink(tgid_base_stuff, ARRAY_SIZE(tgid_base_stuff)); | |
3937 | } |