Locking 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530
  1. The text below describes the locking rules for VFS-related methods.
  2. It is (believed to be) up-to-date. *Please*, if you change anything in
  3. prototypes or locking protocols - update this file. And update the relevant
  4. instances in the tree, don't leave that to maintainers of filesystems/devices/
  5. etc. At the very least, put the list of dubious cases in the end of this file.
  6. Don't turn it into log - maintainers of out-of-the-tree code are supposed to
  7. be able to use diff(1).
  8. Thing currently missing here: socket operations. Alexey?
  9. --------------------------- dentry_operations --------------------------
  10. prototypes:
  11. int (*d_revalidate)(struct dentry *, int);
  12. int (*d_hash) (struct dentry *, struct qstr *);
  13. int (*d_compare) (struct dentry *, struct qstr *, struct qstr *);
  14. int (*d_delete)(struct dentry *);
  15. void (*d_release)(struct dentry *);
  16. void (*d_iput)(struct dentry *, struct inode *);
  17. char *(*d_dname)((struct dentry *dentry, char *buffer, int buflen);
  18. locking rules:
  19. dcache_lock rename_lock ->d_lock may block
  20. d_revalidate: no no no yes
  21. d_hash no no no yes
  22. d_compare: no yes no no
  23. d_delete: yes no yes no
  24. d_release: no no no yes
  25. d_iput: no no no yes
  26. d_dname: no no no no
  27. --------------------------- inode_operations ---------------------------
  28. prototypes:
  29. int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
  30. struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameid
  31. ata *);
  32. int (*link) (struct dentry *,struct inode *,struct dentry *);
  33. int (*unlink) (struct inode *,struct dentry *);
  34. int (*symlink) (struct inode *,struct dentry *,const char *);
  35. int (*mkdir) (struct inode *,struct dentry *,int);
  36. int (*rmdir) (struct inode *,struct dentry *);
  37. int (*mknod) (struct inode *,struct dentry *,int,dev_t);
  38. int (*rename) (struct inode *, struct dentry *,
  39. struct inode *, struct dentry *);
  40. int (*readlink) (struct dentry *, char __user *,int);
  41. void * (*follow_link) (struct dentry *, struct nameidata *);
  42. void (*put_link) (struct dentry *, struct nameidata *, void *);
  43. void (*truncate) (struct inode *);
  44. int (*permission) (struct inode *, int, struct nameidata *);
  45. int (*check_acl)(struct inode *, int);
  46. int (*setattr) (struct dentry *, struct iattr *);
  47. int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
  48. int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
  49. ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
  50. ssize_t (*listxattr) (struct dentry *, char *, size_t);
  51. int (*removexattr) (struct dentry *, const char *);
  52. void (*truncate_range)(struct inode *, loff_t, loff_t);
  53. long (*fallocate)(struct inode *inode, int mode, loff_t offset, loff_t len);
  54. int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, u64 len);
  55. locking rules:
  56. all may block
  57. i_mutex(inode)
  58. lookup: yes
  59. create: yes
  60. link: yes (both)
  61. mknod: yes
  62. symlink: yes
  63. mkdir: yes
  64. unlink: yes (both)
  65. rmdir: yes (both) (see below)
  66. rename: yes (all) (see below)
  67. readlink: no
  68. follow_link: no
  69. put_link: no
  70. truncate: yes (see below)
  71. setattr: yes
  72. permission: no
  73. check_acl: no
  74. getattr: no
  75. setxattr: yes
  76. getxattr: no
  77. listxattr: no
  78. removexattr: yes
  79. truncate_range: yes
  80. fallocate: no
  81. fiemap: no
  82. Additionally, ->rmdir(), ->unlink() and ->rename() have ->i_mutex on
  83. victim.
  84. cross-directory ->rename() has (per-superblock) ->s_vfs_rename_sem.
  85. ->truncate() is never called directly - it's a callback, not a
  86. method. It's called by vmtruncate() - deprecated library function used by
  87. ->setattr(). Locking information above applies to that call (i.e. is
  88. inherited from ->setattr() - vmtruncate() is used when ATTR_SIZE had been
  89. passed).
  90. See Documentation/filesystems/directory-locking for more detailed discussion
  91. of the locking scheme for directory operations.
  92. --------------------------- super_operations ---------------------------
  93. prototypes:
  94. struct inode *(*alloc_inode)(struct super_block *sb);
  95. void (*destroy_inode)(struct inode *);
  96. void (*dirty_inode) (struct inode *);
  97. int (*write_inode) (struct inode *, struct writeback_control *wbc);
  98. int (*drop_inode) (struct inode *);
  99. void (*evict_inode) (struct inode *);
  100. void (*put_super) (struct super_block *);
  101. void (*write_super) (struct super_block *);
  102. int (*sync_fs)(struct super_block *sb, int wait);
  103. int (*freeze_fs) (struct super_block *);
  104. int (*unfreeze_fs) (struct super_block *);
  105. int (*statfs) (struct dentry *, struct kstatfs *);
  106. int (*remount_fs) (struct super_block *, int *, char *);
  107. void (*umount_begin) (struct super_block *);
  108. int (*show_options)(struct seq_file *, struct vfsmount *);
  109. ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
  110. ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
  111. int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
  112. locking rules:
  113. All may block [not true, see below]
  114. s_umount
  115. alloc_inode:
  116. destroy_inode:
  117. dirty_inode: (must not sleep)
  118. write_inode:
  119. drop_inode: !!!inode_lock!!!
  120. evict_inode:
  121. put_super: write
  122. write_super: read
  123. sync_fs: read
  124. freeze_fs: read
  125. unfreeze_fs: read
  126. statfs: maybe(read) (see below)
  127. remount_fs: write
  128. umount_begin: no
  129. show_options: no (namespace_sem)
  130. quota_read: no (see below)
  131. quota_write: no (see below)
  132. bdev_try_to_free_page: no (see below)
  133. ->statfs() has s_umount (shared) when called by ustat(2) (native or
  134. compat), but that's an accident of bad API; s_umount is used to pin
  135. the superblock down when we only have dev_t given us by userland to
  136. identify the superblock. Everything else (statfs(), fstatfs(), etc.)
  137. doesn't hold it when calling ->statfs() - superblock is pinned down
  138. by resolving the pathname passed to syscall.
  139. ->quota_read() and ->quota_write() functions are both guaranteed to
  140. be the only ones operating on the quota file by the quota code (via
  141. dqio_sem) (unless an admin really wants to screw up something and
  142. writes to quota files with quotas on). For other details about locking
  143. see also dquot_operations section.
  144. ->bdev_try_to_free_page is called from the ->releasepage handler of
  145. the block device inode. See there for more details.
  146. --------------------------- file_system_type ---------------------------
  147. prototypes:
  148. int (*get_sb) (struct file_system_type *, int,
  149. const char *, void *, struct vfsmount *);
  150. struct dentry *(*mount) (struct file_system_type *, int,
  151. const char *, void *);
  152. void (*kill_sb) (struct super_block *);
  153. locking rules:
  154. may block
  155. get_sb yes
  156. mount yes
  157. kill_sb yes
  158. ->get_sb() returns error or 0 with locked superblock attached to the vfsmount
  159. (exclusive on ->s_umount).
  160. ->mount() returns ERR_PTR or the root dentry.
  161. ->kill_sb() takes a write-locked superblock, does all shutdown work on it,
  162. unlocks and drops the reference.
  163. --------------------------- address_space_operations --------------------------
  164. prototypes:
  165. int (*writepage)(struct page *page, struct writeback_control *wbc);
  166. int (*readpage)(struct file *, struct page *);
  167. int (*sync_page)(struct page *);
  168. int (*writepages)(struct address_space *, struct writeback_control *);
  169. int (*set_page_dirty)(struct page *page);
  170. int (*readpages)(struct file *filp, struct address_space *mapping,
  171. struct list_head *pages, unsigned nr_pages);
  172. int (*write_begin)(struct file *, struct address_space *mapping,
  173. loff_t pos, unsigned len, unsigned flags,
  174. struct page **pagep, void **fsdata);
  175. int (*write_end)(struct file *, struct address_space *mapping,
  176. loff_t pos, unsigned len, unsigned copied,
  177. struct page *page, void *fsdata);
  178. sector_t (*bmap)(struct address_space *, sector_t);
  179. int (*invalidatepage) (struct page *, unsigned long);
  180. int (*releasepage) (struct page *, int);
  181. void (*freepage)(struct page *);
  182. int (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
  183. loff_t offset, unsigned long nr_segs);
  184. int (*get_xip_mem)(struct address_space *, pgoff_t, int, void **,
  185. unsigned long *);
  186. int (*migratepage)(struct address_space *, struct page *, struct page *);
  187. int (*launder_page)(struct page *);
  188. int (*is_partially_uptodate)(struct page *, read_descriptor_t *, unsigned long);
  189. int (*error_remove_page)(struct address_space *, struct page *);
  190. locking rules:
  191. All except set_page_dirty and freepage may block
  192. PageLocked(page) i_mutex
  193. writepage: yes, unlocks (see below)
  194. readpage: yes, unlocks
  195. sync_page: maybe
  196. writepages:
  197. set_page_dirty no
  198. readpages:
  199. write_begin: locks the page yes
  200. write_end: yes, unlocks yes
  201. bmap:
  202. invalidatepage: yes
  203. releasepage: yes
  204. freepage: yes
  205. direct_IO:
  206. get_xip_mem: maybe
  207. migratepage: yes (both)
  208. launder_page: yes
  209. is_partially_uptodate: yes
  210. error_remove_page: yes
  211. ->write_begin(), ->write_end(), ->sync_page() and ->readpage()
  212. may be called from the request handler (/dev/loop).
  213. ->readpage() unlocks the page, either synchronously or via I/O
  214. completion.
  215. ->readpages() populates the pagecache with the passed pages and starts
  216. I/O against them. They come unlocked upon I/O completion.
  217. ->writepage() is used for two purposes: for "memory cleansing" and for
  218. "sync". These are quite different operations and the behaviour may differ
  219. depending upon the mode.
  220. If writepage is called for sync (wbc->sync_mode != WBC_SYNC_NONE) then
  221. it *must* start I/O against the page, even if that would involve
  222. blocking on in-progress I/O.
  223. If writepage is called for memory cleansing (sync_mode ==
  224. WBC_SYNC_NONE) then its role is to get as much writeout underway as
  225. possible. So writepage should try to avoid blocking against
  226. currently-in-progress I/O.
  227. If the filesystem is not called for "sync" and it determines that it
  228. would need to block against in-progress I/O to be able to start new I/O
  229. against the page the filesystem should redirty the page with
  230. redirty_page_for_writepage(), then unlock the page and return zero.
  231. This may also be done to avoid internal deadlocks, but rarely.
  232. If the filesystem is called for sync then it must wait on any
  233. in-progress I/O and then start new I/O.
  234. The filesystem should unlock the page synchronously, before returning to the
  235. caller, unless ->writepage() returns special WRITEPAGE_ACTIVATE
  236. value. WRITEPAGE_ACTIVATE means that page cannot really be written out
  237. currently, and VM should stop calling ->writepage() on this page for some
  238. time. VM does this by moving page to the head of the active list, hence the
  239. name.
  240. Unless the filesystem is going to redirty_page_for_writepage(), unlock the page
  241. and return zero, writepage *must* run set_page_writeback() against the page,
  242. followed by unlocking it. Once set_page_writeback() has been run against the
  243. page, write I/O can be submitted and the write I/O completion handler must run
  244. end_page_writeback() once the I/O is complete. If no I/O is submitted, the
  245. filesystem must run end_page_writeback() against the page before returning from
  246. writepage.
  247. That is: after 2.5.12, pages which are under writeout are *not* locked. Note,
  248. if the filesystem needs the page to be locked during writeout, that is ok, too,
  249. the page is allowed to be unlocked at any point in time between the calls to
  250. set_page_writeback() and end_page_writeback().
  251. Note, failure to run either redirty_page_for_writepage() or the combination of
  252. set_page_writeback()/end_page_writeback() on a page submitted to writepage
  253. will leave the page itself marked clean but it will be tagged as dirty in the
  254. radix tree. This incoherency can lead to all sorts of hard-to-debug problems
  255. in the filesystem like having dirty inodes at umount and losing written data.
  256. ->sync_page() locking rules are not well-defined - usually it is called
  257. with lock on page, but that is not guaranteed. Considering the currently
  258. existing instances of this method ->sync_page() itself doesn't look
  259. well-defined...
  260. ->writepages() is used for periodic writeback and for syscall-initiated
  261. sync operations. The address_space should start I/O against at least
  262. *nr_to_write pages. *nr_to_write must be decremented for each page which is
  263. written. The address_space implementation may write more (or less) pages
  264. than *nr_to_write asks for, but it should try to be reasonably close. If
  265. nr_to_write is NULL, all dirty pages must be written.
  266. writepages should _only_ write pages which are present on
  267. mapping->io_pages.
  268. ->set_page_dirty() is called from various places in the kernel
  269. when the target page is marked as needing writeback. It may be called
  270. under spinlock (it cannot block) and is sometimes called with the page
  271. not locked.
  272. ->bmap() is currently used by legacy ioctl() (FIBMAP) provided by some
  273. filesystems and by the swapper. The latter will eventually go away. Please,
  274. keep it that way and don't breed new callers.
  275. ->invalidatepage() is called when the filesystem must attempt to drop
  276. some or all of the buffers from the page when it is being truncated. It
  277. returns zero on success. If ->invalidatepage is zero, the kernel uses
  278. block_invalidatepage() instead.
  279. ->releasepage() is called when the kernel is about to try to drop the
  280. buffers from the page in preparation for freeing it. It returns zero to
  281. indicate that the buffers are (or may be) freeable. If ->releasepage is zero,
  282. the kernel assumes that the fs has no private interest in the buffers.
  283. ->freepage() is called when the kernel is done dropping the page
  284. from the page cache.
  285. ->launder_page() may be called prior to releasing a page if
  286. it is still found to be dirty. It returns zero if the page was successfully
  287. cleaned, or an error value if not. Note that in order to prevent the page
  288. getting mapped back in and redirtied, it needs to be kept locked
  289. across the entire operation.
  290. ----------------------- file_lock_operations ------------------------------
  291. prototypes:
  292. void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
  293. void (*fl_release_private)(struct file_lock *);
  294. locking rules:
  295. file_lock_lock may block
  296. fl_copy_lock: yes no
  297. fl_release_private: maybe no
  298. ----------------------- lock_manager_operations ---------------------------
  299. prototypes:
  300. int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
  301. void (*fl_notify)(struct file_lock *); /* unblock callback */
  302. int (*fl_grant)(struct file_lock *, struct file_lock *, int);
  303. void (*fl_release_private)(struct file_lock *);
  304. void (*fl_break)(struct file_lock *); /* break_lease callback */
  305. int (*fl_change)(struct file_lock **, int);
  306. locking rules:
  307. file_lock_lock may block
  308. fl_compare_owner: yes no
  309. fl_notify: yes no
  310. fl_grant: no no
  311. fl_release_private: maybe no
  312. fl_break: yes no
  313. fl_change yes no
  314. --------------------------- buffer_head -----------------------------------
  315. prototypes:
  316. void (*b_end_io)(struct buffer_head *bh, int uptodate);
  317. locking rules:
  318. called from interrupts. In other words, extreme care is needed here.
  319. bh is locked, but that's all warranties we have here. Currently only RAID1,
  320. highmem, fs/buffer.c, and fs/ntfs/aops.c are providing these. Block devices
  321. call this method upon the IO completion.
  322. --------------------------- block_device_operations -----------------------
  323. prototypes:
  324. int (*open) (struct block_device *, fmode_t);
  325. int (*release) (struct gendisk *, fmode_t);
  326. int (*ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
  327. int (*compat_ioctl) (struct block_device *, fmode_t, unsigned, unsigned long);
  328. int (*direct_access) (struct block_device *, sector_t, void **, unsigned long *);
  329. int (*media_changed) (struct gendisk *);
  330. void (*unlock_native_capacity) (struct gendisk *);
  331. int (*revalidate_disk) (struct gendisk *);
  332. int (*getgeo)(struct block_device *, struct hd_geometry *);
  333. void (*swap_slot_free_notify) (struct block_device *, unsigned long);
  334. locking rules:
  335. bd_mutex
  336. open: yes
  337. release: yes
  338. ioctl: no
  339. compat_ioctl: no
  340. direct_access: no
  341. media_changed: no
  342. unlock_native_capacity: no
  343. revalidate_disk: no
  344. getgeo: no
  345. swap_slot_free_notify: no (see below)
  346. media_changed, unlock_native_capacity and revalidate_disk are called only from
  347. check_disk_change().
  348. swap_slot_free_notify is called with swap_lock and sometimes the page lock
  349. held.
  350. --------------------------- file_operations -------------------------------
  351. prototypes:
  352. loff_t (*llseek) (struct file *, loff_t, int);
  353. ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
  354. ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
  355. ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
  356. ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
  357. int (*readdir) (struct file *, void *, filldir_t);
  358. unsigned int (*poll) (struct file *, struct poll_table_struct *);
  359. long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
  360. long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
  361. int (*mmap) (struct file *, struct vm_area_struct *);
  362. int (*open) (struct inode *, struct file *);
  363. int (*flush) (struct file *);
  364. int (*release) (struct inode *, struct file *);
  365. int (*fsync) (struct file *, int datasync);
  366. int (*aio_fsync) (struct kiocb *, int datasync);
  367. int (*fasync) (int, struct file *, int);
  368. int (*lock) (struct file *, int, struct file_lock *);
  369. ssize_t (*readv) (struct file *, const struct iovec *, unsigned long,
  370. loff_t *);
  371. ssize_t (*writev) (struct file *, const struct iovec *, unsigned long,
  372. loff_t *);
  373. ssize_t (*sendfile) (struct file *, loff_t *, size_t, read_actor_t,
  374. void __user *);
  375. ssize_t (*sendpage) (struct file *, struct page *, int, size_t,
  376. loff_t *, int);
  377. unsigned long (*get_unmapped_area)(struct file *, unsigned long,
  378. unsigned long, unsigned long, unsigned long);
  379. int (*check_flags)(int);
  380. int (*flock) (struct file *, int, struct file_lock *);
  381. ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *,
  382. size_t, unsigned int);
  383. ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *,
  384. size_t, unsigned int);
  385. int (*setlease)(struct file *, long, struct file_lock **);
  386. };
  387. locking rules:
  388. All may block except for ->setlease.
  389. No VFS locks held on entry except for ->fsync and ->setlease.
  390. ->fsync() has i_mutex on inode.
  391. ->setlease has the file_list_lock held and must not sleep.
  392. ->llseek() locking has moved from llseek to the individual llseek
  393. implementations. If your fs is not using generic_file_llseek, you
  394. need to acquire and release the appropriate locks in your ->llseek().
  395. For many filesystems, it is probably safe to acquire the inode
  396. mutex or just to use i_size_read() instead.
  397. Note: this does not protect the file->f_pos against concurrent modifications
  398. since this is something the userspace has to take care about.
  399. ->fasync() is responsible for maintaining the FASYNC bit in filp->f_flags.
  400. Most instances call fasync_helper(), which does that maintenance, so it's
  401. not normally something one needs to worry about. Return values > 0 will be
  402. mapped to zero in the VFS layer.
  403. ->readdir() and ->ioctl() on directories must be changed. Ideally we would
  404. move ->readdir() to inode_operations and use a separate method for directory
  405. ->ioctl() or kill the latter completely. One of the problems is that for
  406. anything that resembles union-mount we won't have a struct file for all
  407. components. And there are other reasons why the current interface is a mess...
  408. ->read on directories probably must go away - we should just enforce -EISDIR
  409. in sys_read() and friends.
  410. --------------------------- dquot_operations -------------------------------
  411. prototypes:
  412. int (*write_dquot) (struct dquot *);
  413. int (*acquire_dquot) (struct dquot *);
  414. int (*release_dquot) (struct dquot *);
  415. int (*mark_dirty) (struct dquot *);
  416. int (*write_info) (struct super_block *, int);
  417. These operations are intended to be more or less wrapping functions that ensure
  418. a proper locking wrt the filesystem and call the generic quota operations.
  419. What filesystem should expect from the generic quota functions:
  420. FS recursion Held locks when called
  421. write_dquot: yes dqonoff_sem or dqptr_sem
  422. acquire_dquot: yes dqonoff_sem or dqptr_sem
  423. release_dquot: yes dqonoff_sem or dqptr_sem
  424. mark_dirty: no -
  425. write_info: yes dqonoff_sem
  426. FS recursion means calling ->quota_read() and ->quota_write() from superblock
  427. operations.
  428. More details about quota locking can be found in fs/dquot.c.
  429. --------------------------- vm_operations_struct -----------------------------
  430. prototypes:
  431. void (*open)(struct vm_area_struct*);
  432. void (*close)(struct vm_area_struct*);
  433. int (*fault)(struct vm_area_struct*, struct vm_fault *);
  434. int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
  435. int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);
  436. locking rules:
  437. mmap_sem PageLocked(page)
  438. open: yes
  439. close: yes
  440. fault: yes can return with page locked
  441. page_mkwrite: yes can return with page locked
  442. access: yes
  443. ->fault() is called when a previously not present pte is about
  444. to be faulted in. The filesystem must find and return the page associated
  445. with the passed in "pgoff" in the vm_fault structure. If it is possible that
  446. the page may be truncated and/or invalidated, then the filesystem must lock
  447. the page, then ensure it is not already truncated (the page lock will block
  448. subsequent truncate), and then return with VM_FAULT_LOCKED, and the page
  449. locked. The VM will unlock the page.
  450. ->page_mkwrite() is called when a previously read-only pte is
  451. about to become writeable. The filesystem again must ensure that there are
  452. no truncate/invalidate races, and then return with the page locked. If
  453. the page has been truncated, the filesystem should not look up a new page
  454. like the ->fault() handler, but simply return with VM_FAULT_NOPAGE, which
  455. will cause the VM to retry the fault.
  456. ->access() is called when get_user_pages() fails in
  457. acces_process_vm(), typically used to debug a process through
  458. /proc/pid/mem or ptrace. This function is needed only for
  459. VM_IO | VM_PFNMAP VMAs.
  460. ================================================================================
  461. Dubious stuff
  462. (if you break something or notice that it is broken and do not fix it yourself
  463. - at least put it here)