linux/fs
Linus Torvalds 3a34b13a88 pipe: make pipe writes always wake up readers
Since commit 1b6b26ae70 ("pipe: fix and clarify pipe write wakeup
logic") we have sanitized the pipe write logic, and would only try to
wake up readers if they needed it.

In particular, if the pipe already had data in it before the write,
there was no point in trying to wake up a reader, since any existing
readers must have been aware of the pre-existing data already.  Doing
extraneous wakeups will only cause potential thundering herd problems.

However, it turns out that some Android libraries have misused the EPOLL
interface, and expected "edge triggered" be to "any new write will
trigger it".  Even if there was no edge in sight.

Quoting Sandeep Patil:
 "The commit 1b6b26ae70 ('pipe: fix and clarify pipe write wakeup
  logic') changed pipe write logic to wakeup readers only if the pipe
  was empty at the time of write. However, there are libraries that
  relied upon the older behavior for notification scheme similar to
  what's described in [1]

  One such library 'realm-core'[2] is used by numerous Android
  applications. The library uses a similar notification mechanism as GNU
  Make but it never drains the pipe until it is full. When Android moved
  to v5.10 kernel, all applications using this library stopped working.

  The library has since been fixed[3] but it will be a while before all
  applications incorporate the updated library"

Our regression rule for the kernel is that if applications break from
new behavior, it's a regression, even if it was because the application
did something patently wrong.  Also note the original report [4] by
Michal Kerrisk about a test for this epoll behavior - but at that point
we didn't know of any actual broken use case.

So add the extraneous wakeup, to approximate the old behavior.

[ I say "approximate", because the exact old behavior was to do a wakeup
  not for each write(), but for each pipe buffer chunk that was filled
  in. The behavior introduced by this change is not that - this is just
  "every write will cause a wakeup, whether necessary or not", which
  seems to be sufficient for the broken library use. ]

It's worth noting that this adds the extraneous wakeup only for the
write side, while the read side still considers the "edge" to be purely
about reading enough from the pipe to allow further writes.

See commit f467a6a664 ("pipe: fix and clarify pipe read wakeup logic")
for the pipe read case, which remains that "only wake up if the pipe was
full, and we read something from it".

Link: https://lore.kernel.org/lkml/CAHk-=wjeG0q1vgzu4iJhW5juPkTsjTYmiqiMUYAebWW+0bam6w@mail.gmail.com/ [1]
Link: https://github.com/realm/realm-core [2]
Link: https://github.com/realm/realm-core/issues/4666 [3]
Link: https://lore.kernel.org/lkml/CAKgNAkjMBGeAwF=2MKK758BhxvW58wYTgYKB2V-gY1PwXxrH+Q@mail.gmail.com/ [4]
Link: https://lore.kernel.org/lkml/20210729222635.2937453-1-sspatil@android.com/
Reported-by: Sandeep Patil <sspatil@android.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-30 15:42:34 -07:00
..
9p 9p for 5.13-rc1 2021-05-07 11:18:52 -07:00
adfs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
affs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
afs afs: Remove redundant assignment to ret 2021-07-21 15:11:22 +01:00
autofs autofs: should_expire() argument is guaranteed to be positive 2021-03-24 14:14:27 -04:00
befs fs/befs: Delete obsolete TODO file 2021-03-30 16:54:49 -07:00
bfs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
btrfs for-5.14-rc3-tag 2021-07-30 10:50:09 -07:00
cachefiles fscache, cachefiles: Add alternate API to use kiocb for read/write to cache 2021-04-23 10:14:32 +01:00
ceph ceph: don't WARN if we're still opening a session to an MDS 2021-07-20 17:57:33 +02:00
cifs cifs: fix fallocate when trying to allocate a hole. 2021-07-22 21:24:22 -05:00
coda coda: fix reference counting in coda_file_mmap error path 2021-04-23 14:42:39 -07:00
configfs configfs: fix the read and write iterators 2021-07-13 20:56:24 +02:00
cramfs cramfs: use %pD instead of messing with file_dentry()->d_name 2021-01-05 23:02:47 -05:00
crypto fscrypt: fix derivation of SipHash keys on big endian CPUs 2021-06-05 00:52:52 -07:00
debugfs Linux 5.13-rc6 2021-06-14 09:07:45 +02:00
devpts
dlm fs: dlm: invalid buffer access in lookup error 2021-06-11 12:44:47 -05:00
ecryptfs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
efivarfs efivars: convert to fileattr 2021-04-12 15:04:29 +02:00
efs
erofs erofs: clean up file headers & footers 2021-06-08 00:41:24 +08:00
exfat Description for this pull request: 2021-07-06 11:06:04 -07:00
exportfs exportfs: Add a function to return the raw output from fh_to_dentry() 2020-12-09 09:39:38 -05:00
ext2 fs/ext2: Avoid page_address on pages returned by ext2_get_page 2021-07-16 12:36:51 +02:00
ext4 Ext4 regression and bug fixes for v5.14-rc1 2021-07-09 09:57:27 -07:00
f2fs f2fs: drop dirty node pages when cp is in error status 2021-07-06 22:05:06 -07:00
fat mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
freevxfs
fscache fscache, cachefiles: Add alternate API to use kiocb for read/write to cache 2021-04-23 10:14:32 +01:00
fuse fuse update for 5.14 2021-07-06 11:17:41 -07:00
gfs2 Various minor gfs2 cleanups and fixes 2021-06-29 20:23:08 -07:00
hfs hfs: add lock nesting notation to hfs_find_init 2021-07-15 10:13:49 -07:00
hfsplus hfsplus: report create_date to kstat.btime 2021-07-01 11:06:06 -07:00
hostfs Merge branch 'work.misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-05-02 09:14:01 -07:00
hpfs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
hugetlbfs hugetlbfs: fix mount mode command line processing 2021-07-23 17:43:28 -07:00
iomap iomap: Don't create iomap_page objects in iomap_page_mkwrite_actor 2021-07-15 09:58:06 -07:00
isofs isofs: remove redundant continue statement 2021-06-17 17:11:42 +02:00
jbd2 ext4: inline jbd2_journal_[un]register_shrinker() 2021-07-08 08:37:31 -04:00
jffs2 This pull request contains changes for JFFS2, UBI and UBIFS 2021-05-04 18:08:40 -07:00
jfs JFS fixes for 5.14 2021-07-02 14:25:17 -07:00
kernfs Driver core changes for 5.14-rc1 2021-07-05 13:51:41 -07:00
lockd lockd: Update the NLMv4 SHARE results encoder to use struct xdr_stream 2021-07-06 20:14:44 -04:00
minix mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
netfs netfs: fix test for whether we can skip read when writing beyond EOF 2021-06-21 21:24:07 +01:00
nfs NFS client updates for Linux 5.14 2021-07-09 09:43:57 -07:00
nfs_common nfs_common: fix doc warning 2021-07-06 20:14:41 -04:00
nfsd block-5.14-2021-07-08 2021-07-09 12:05:33 -07:00
nilfs2 Merge branch 'akpm' (patches from Andrew) 2021-07-02 12:08:10 -07:00
nls
notify fanotify: fix copy_event_to_user() fid error clean up 2021-06-14 12:16:37 +02:00
ntfs Merge branch 'work.iov_iter' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-07-03 11:30:04 -07:00
ocfs2 ocfs2: issue zeroout to EOF blocks 2021-07-30 10:14:39 -07:00
omfs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
openpromfs openpromfs: don't do unlock_new_inode() until the new inode is set up 2021-03-12 22:15:22 -05:00
orangefs orangefs: fix orangefs df output. 2021-06-28 08:40:08 -04:00
overlayfs overlayfs update for 5.13 2021-04-30 15:17:08 -07:00
proc Merge branch 'work.namei' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-07-03 11:41:14 -07:00
pstore for-5.14/drivers-2021-06-29 2021-06-30 12:21:16 -07:00
qnx4
qnx6
quota quota: remove unnecessary oom message 2021-06-22 10:40:52 +02:00
ramfs fs: move ramfs_aops to libfs 2021-06-29 10:53:48 -07:00
reiserfs reiserfs: check directory items on read from disk 2021-07-16 12:36:51 +02:00
romfs
squashfs squashfs: add option to panic on errors 2021-06-29 10:53:46 -07:00
sysfs sysfs: Support zapping of binary attr mmaps 2021-01-12 14:26:31 +01:00
sysv mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
tracefs tracing: Fix various typos in comments 2021-03-23 14:08:18 -04:00
ubifs ubifs: Set/Clear I_LINKABLE under i_lock for whiteout inode 2021-06-22 09:21:39 +02:00
udf \n 2021-07-01 12:06:39 -07:00
ufs mm: require ->set_page_dirty to be explicitly wired up 2021-06-29 10:53:48 -07:00
unicode .gitignore: prefix local generated files with a slash 2021-05-02 00:43:35 +09:00
vboxsf vboxsf: Add support for the atomic_open directory-inode op 2021-06-23 14:36:52 +02:00
verity fsverity: relax build time dependency on CRYPTO_SHA256 2021-04-22 17:31:32 +10:00
xfs Fixes for 5.14-rc: 2021-07-18 11:27:25 -07:00
zonefs zonefs: remove redundant null bio check 2021-07-16 13:45:18 +09:00
aio.c Revert "mremap: don't allow MREMAP_DONTUNMAP on special_mappings and aio" 2021-04-30 11:20:39 -07:00
anon_inodes.c fs: anon_inodes: rephrase to appropriate kernel-doc 2021-01-15 12:17:25 -05:00
attr.c ima: handle idmapped mounts 2021-01-24 14:27:20 +01:00
bad_inode.c fs: make helpers idmap mount aware 2021-01-24 14:27:20 +01:00
binfmt_aout.c binfmt: remove in-tree usage of MAP_EXECUTABLE 2021-06-29 10:53:50 -07:00
binfmt_elf_fdpic.c Merge branch 'akpm' (patches from Andrew) 2021-06-29 17:29:11 -07:00
binfmt_elf.c Merge branch 'akpm' (patches from Andrew) 2021-06-29 17:29:11 -07:00
binfmt_flat.c binfmt: remove in-tree usage of MAP_EXECUTABLE 2021-06-29 10:53:50 -07:00
binfmt_misc.c binfmt_misc: fix possible deadlock in bm_register_write 2021-03-13 11:27:30 -08:00
binfmt_script.c
block_dev.c block-5.14-2021-07-30 2021-07-30 11:08:12 -07:00
buffer.c mm/writeback: move __set_page_dirty() to core mm 2021-06-29 10:53:48 -07:00
char_dev.c
compat_binfmt_elf.c get rid of COMPAT_ELF_EXEC_PAGESIZE 2021-01-06 08:42:51 -05:00
coredump.c Merge branch 'work.namei' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-07-03 11:41:14 -07:00
d_path.c getcwd(2): clean up error handling 2021-05-18 20:15:58 -04:00
dax.c dax: fix ENOMEM handling in grab_mapping_entry() 2021-06-29 10:53:47 -07:00
dcache.c useful constants: struct qstr for ".." 2021-04-15 22:36:45 -04:00
direct-io.c fs: direct-io: fix missing sdio->boundary 2021-04-09 14:54:23 -07:00
drop_caches.c
eventfd.c
eventpoll.c fs/epoll: restore waking from ep_done_scan() 2021-05-06 19:24:13 -07:00
exec.c Merge branch 'akpm' (patches from Andrew) 2021-07-02 12:08:10 -07:00
fcntl.c fcntl: Fix unreachable code in do_fcntl() 2021-07-12 11:09:13 -05:00
fhandle.c switch file_open_root() to struct path 2021-04-07 13:56:43 -04:00
file_table.c
file.c Merge branch 'work.file' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-05-03 11:05:28 -07:00
filesystems.c
fs_context.c fs: add vfs_parse_fs_param_source() helper 2021-07-14 09:19:06 -07:00
fs_parser.c vfs: fs_parser: clean up kernel-doc warnings 2021-04-30 11:20:35 -07:00
fs_pin.c
fs_struct.c
fs_types.c
fs-writeback.c writeback, cgroup: do not reparent dax inodes 2021-07-23 17:43:28 -07:00
fsopen.c
init.c init: handle idmapped mounts 2021-01-24 14:27:19 +01:00
inode.c mm: remove nrexceptional from inode: remove BUG_ON 2021-05-05 11:27:20 -07:00
internal.h cgroup1: fix leaked context root causing sporadic NULL deref in LTP 2021-07-21 06:39:20 -10:00
io_uring.c io_uring: fix poll requests leaking second poll entries 2021-07-28 07:24:57 -06:00
io-wq.c io_uring: explicitly catch any illegal async queue attempt 2021-07-23 16:44:51 -06:00
io-wq.h io_uring: move creds from io-wq work to io_kiocb 2021-06-18 09:22:02 -06:00
ioctl.c vfs: add fileattr ops 2021-04-12 15:04:23 +02:00
Kconfig mm: hugetlb: introduce CONFIG_HUGETLB_PAGE_FREE_VMEMMAP_DEFAULT_ON 2021-06-30 20:47:26 -07:00
Kconfig.binfmt binfmt: remove support for em86 (alpha only) 2021-07-25 22:33:03 -07:00
kernel_read_file.c switch file_open_root() to struct path 2021-04-07 13:56:43 -04:00
libfs.c fs: remove noop_set_page_dirty() 2021-06-29 10:53:48 -07:00
locks.c Additional fixes and clean-ups for NFSD since tags/nfsd-5.13, 2021-05-05 13:44:19 -07:00
Makefile binfmt: remove support for em86 (alpha only) 2021-07-25 22:33:03 -07:00
mbcache.c
mount.h mount: make {lock,unlock}_mount_hash() static 2021-01-24 14:29:34 +01:00
mpage.c block: rename BIO_MAX_PAGES to BIO_MAX_VECS 2021-03-11 07:47:48 -07:00
namei.c Merge branch 'work.namei' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-07-03 11:41:14 -07:00
namespace.c mount: Support "nosymfollow" in new mount api 2021-06-01 12:09:27 +02:00
no-block.c
nsfs.c
open.c Merge branch 'work.namei' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs 2021-07-03 11:41:14 -07:00
pipe.c pipe: make pipe writes always wake up readers 2021-07-30 15:42:34 -07:00
pnode.c
pnode.h mount: fix mounting of detached mounts onto targets that reside on shared mounts 2021-03-08 15:18:43 +01:00
posix_acl.c fs: make helpers idmap mount aware 2021-01-24 14:27:20 +01:00
proc_namespace.c fs: introduce MOUNT_ATTR_IDMAP 2021-01-24 14:43:45 +01:00
read_write.c teach sendfile(2) to handle send-to-pipe directly 2021-01-25 23:29:36 -05:00
readdir.c readdir: make sure to verify directory entry for legacy interfaces too 2021-04-17 11:39:49 -07:00
remap_range.c ioctl: handle idmapped mounts 2021-01-24 14:27:19 +01:00
select.c kernel, fs: Introduce and use set_restart_fn() and arch_set_restart_data() 2021-03-16 22:13:10 +01:00
seq_file.c seq_file: disallow extremely large seq buffer allocations 2021-07-19 17:18:48 -07:00
signalfd.c signalfd: Remove SIL_PERF_EVENT fields from signalfd_siginfo 2021-05-18 16:20:54 -05:00
splice.c for-5.12/block-2021-02-17 2021-02-21 11:02:48 -08:00
stack.c
stat.c fs: fix reporting supported extra file attributes for statx() 2021-04-17 23:03:50 -04:00
statfs.c s390,alpha: switch to 64-bit ino_t 2021-02-13 17:17:53 +01:00
super.c block: move bd_mutex to struct gendisk 2021-06-01 07:44:32 -06:00
sync.c
timerfd.c
userfaultfd.c userfaultfd: do not untag user pointers 2021-07-23 17:43:28 -07:00
utimes.c utimes: handle idmapped mounts 2021-01-24 14:27:18 +01:00
xattr.c xattr: fix kernel-doc for mnt_userns and vfs xattr helpers 2021-03-23 11:20:26 +01:00