Hi Linus,
This pull request adds support for pidfd polling and the pidfd_open()
syscall to support retrieving pidfds for processes created without
CLONE_PIDFD.
The following changes since commit a188339ca5a396acc588e5851ed7e19f66b0ebd9:
Linux 5.2-rc1 (2019-05-19 15:47:09 -0700)
are available in the Git repository at:
g...@gitolite.kernel.org:pub/scm/linux/kernel/git/brauner/linux
tags/pidfd-updates-v5.3
for you to fetch changes up to 172bb24a4f480c180bee646f6616f714ac4bcab2:
tests: add pidfd_open() tests (2019-06-28 12:17:55 +0200)
/* Summary */
With this tag come two main features.
First, it adds polling support for pidfds. This allows process managers to
know when a (non-parent) process dies in a race-free way. The nofitication
mechanism used follows the same logic that is currently used when the
parent of a task is notified of a child's death.
With this patchset it is possible to put pidfds in an {e}poll loop and get
reliable notifications for process (i.e. thread-group) exit.
The second feature compliments the first one by making it possible to
retrieve pollable pidfds for processes that were not created using
CLONE_PIDFD.
A lot of processes get created with traditional PID-based calls such as
fork() or clone() (without CLONE_PIDFD). For these processes a caller can
currently not create a pollable pidfd. This is a problem for Android's low
memory killer (LMK) and service managers such as systemd.
Both patchsets are accompanied by selftests.
/* Testing */
All patches are based on v5.2-rc1 and have been sitting in linux-next since
then and have not caused any failures or warnings.
/* Conflicts with v5.2 */
A test-merge of my tree into pristine v5.2 revealed conflicts in the
following two files:
- tools/testing/selftests/pidfd/Makefile
- tools/testing/selftests/pidfd/pidfd_test.c
Fwiw, both conflicts should be trivially resolveable by just accepting all
changes introduced by this tag but I am also happy to provide a
fixed-up/rebased tree.
Note, following Al's changes in
6fd2fe494b17 (" copy_process(): don't use ksys_close() on cleanups")
the function pidfd_create() that pidfd_open() relied on got removed. I
fixed this up *without* rebasing in-tree to not break linux-next and
minimize merge conflicts with other branches. I hope that was ok to do
(The original branch does still exist at [2] for comparison.).
/* Conflicts with other trees */
Based on linux-next, I am only aware of a single trivial conflict with
Andrew's tree where a change to switch struct pid to use refcount_t
has included a new header.
/* Syscall number 434 */
pidfd_open() uses syscall number 434. I'm not aware of any other syscall
targeted for 5.3 that has chosen the same number.
The syscall required no arch-specific massaging and has hence been added to
all architectures at the same time.
/* Adoption in userspace */
Good news is that the work done so far and the work done in this branch for
pidfd_open() and polling support do already see some adoption:
- Android is in the process of backporting this work to all their LTS
kernels (cf. [3]).
- Service managers make use of pidfd_send_signal but will need to wait
until we enable waiting on pidfds for full adoption.
- And projects I maintain make use of both pidfd_send_signal and
CLONE_PIDFD (cf. [4]) and will use polling support and pidfd_open() too.
/* New signing subkey */
So that there are no suprises, please note that I signed-off the tag with a
new signing key: 0x91C61BC06578DCA2
It's an ed25519 signing subkey that I moved to a Nitrokey and that's
available from the new, safer keys.openpgp.org keyserver.
Given the SKS churn I figured it might be a good idea to use a new,
dedicated kernel-only subkey instead of my general signing subkey.
Please consider pulling these changes from the signed pidfd-updates-v5.3 tag.
Thanks!
Christian
[1]: 6fd2fe494b17bf2dec37b610d23a43a72b16923a
[2]:
https://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux.git/log/?h=pidfd_open
[3]:
https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.9+backport%22
https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.14+backport%22
https://android-review.googlesource.com/q/topic:%22pidfd+polling+support+4.19+backport%22
[4]:
https://github.com/lxc/lxc/blob/aab6e3eb73c343231cdde775db938994fc6f2803/src/lxc/start.c#L1753
----------------------------------------------------------------
pidfd-updates-v5.3
----------------------------------------------------------------
Christian Brauner (3):
pid: add pidfd_open()
arch: wire-up pidfd_open()
tests: add pidfd_open() tests
Joel Fernandes (Google) (2):
pidfd: add polling support
pidfd: add polling selftests
arch/alpha/kernel/syscalls/syscall.tbl | 1 +
arch/arm/tools/syscall.tbl | 1 +
arch/arm64/include/asm/unistd.h | 2 +-
arch/arm64/include/asm/unistd32.h | 2 +
arch/ia64/kernel/syscalls/syscall.tbl | 1 +
arch/m68k/kernel/syscalls/syscall.tbl | 1 +
arch/microblaze/kernel/syscalls/syscall.tbl | 1 +
arch/mips/kernel/syscalls/syscall_n32.tbl | 1 +
arch/mips/kernel/syscalls/syscall_n64.tbl | 1 +
arch/mips/kernel/syscalls/syscall_o32.tbl | 1 +
arch/parisc/kernel/syscalls/syscall.tbl | 1 +
arch/powerpc/kernel/syscalls/syscall.tbl | 1 +
arch/s390/kernel/syscalls/syscall.tbl | 1 +
arch/sh/kernel/syscalls/syscall.tbl | 1 +
arch/sparc/kernel/syscalls/syscall.tbl | 1 +
arch/x86/entry/syscalls/syscall_32.tbl | 1 +
arch/x86/entry/syscalls/syscall_64.tbl | 1 +
arch/xtensa/kernel/syscalls/syscall.tbl | 1 +
include/linux/pid.h | 3 +
include/linux/syscalls.h | 1 +
include/uapi/asm-generic/unistd.h | 4 +-
kernel/fork.c | 26 +++
kernel/pid.c | 71 +++++++
kernel/signal.c | 11 ++
tools/testing/selftests/pidfd/.gitignore | 1 +
tools/testing/selftests/pidfd/Makefile | 5 +-
tools/testing/selftests/pidfd/pidfd.h | 57 ++++++
tools/testing/selftests/pidfd/pidfd_open_test.c | 169 ++++++++++++++++
tools/testing/selftests/pidfd/pidfd_test.c | 252 ++++++++++++++++++++----
29 files changed, 576 insertions(+), 44 deletions(-)
create mode 100644 tools/testing/selftests/pidfd/pidfd.h
create mode 100644 tools/testing/selftests/pidfd/pidfd_open_test.c
