]> asedeno.scripts.mit.edu Git - linux.git/commit
stop_machine: Avoid potential race behaviour
authorMark Rutland <mark.rutland@arm.com>
Mon, 7 Oct 2019 10:45:36 +0000 (11:45 +0100)
committerThomas Gleixner <tglx@linutronix.de>
Thu, 17 Oct 2019 10:47:12 +0000 (12:47 +0200)
commitb1fc5833357524d5d342737913dbe32ff3557bc5
tree0e8dd11256f35518e7767cf1c5f49d544e962a6b
parent4f5cafb5cb8471e54afdc9054d973535614f7675
stop_machine: Avoid potential race behaviour

Both multi_cpu_stop() and set_state() access multi_stop_data::state
racily using plain accesses. These are subject to compiler
transformations which could break the intended behaviour of the code,
and this situation is detected by KCSAN on both arm64 and x86 (splats
below).

Improve matters by using READ_ONCE() and WRITE_ONCE() to ensure that the
compiler cannot elide, replay, or tear loads and stores.

In multi_cpu_stop() the two loads of multi_stop_data::state are expected to
be a consistent value, so snapshot the value into a temporary variable to
ensure this.

The state transitions are serialized by atomic manipulation of
multi_stop_data::num_threads, and other fields in multi_stop_data are not
modified while subject to concurrent reads.

KCSAN splat on arm64:

| BUG: KCSAN: data-race in multi_cpu_stop+0xa8/0x198 and set_state+0x80/0xb0
|
| write to 0xffff00001003bd00 of 4 bytes by task 24 on cpu 3:
|  set_state+0x80/0xb0
|  multi_cpu_stop+0x16c/0x198
|  cpu_stopper_thread+0x170/0x298
|  smpboot_thread_fn+0x40c/0x560
|  kthread+0x1a8/0x1b0
|  ret_from_fork+0x10/0x18
|
| read to 0xffff00001003bd00 of 4 bytes by task 14 on cpu 1:
|  multi_cpu_stop+0xa8/0x198
|  cpu_stopper_thread+0x170/0x298
|  smpboot_thread_fn+0x40c/0x560
|  kthread+0x1a8/0x1b0
|  ret_from_fork+0x10/0x18
|
| Reported by Kernel Concurrency Sanitizer on:
| CPU: 1 PID: 14 Comm: migration/1 Not tainted 5.3.0-00007-g67ab35a199f4-dirty #3
| Hardware name: linux,dummy-virt (DT)

KCSAN splat on x86:

| write to 0xffffb0bac0013e18 of 4 bytes by task 19 on cpu 2:
|  set_state kernel/stop_machine.c:170 [inline]
|  ack_state kernel/stop_machine.c:177 [inline]
|  multi_cpu_stop+0x1a4/0x220 kernel/stop_machine.c:227
|  cpu_stopper_thread+0x19e/0x280 kernel/stop_machine.c:516
|  smpboot_thread_fn+0x1a8/0x300 kernel/smpboot.c:165
|  kthread+0x1b5/0x200 kernel/kthread.c:255
|  ret_from_fork+0x35/0x40 arch/x86/entry/entry_64.S:352
|
| read to 0xffffb0bac0013e18 of 4 bytes by task 44 on cpu 7:
|  multi_cpu_stop+0xb4/0x220 kernel/stop_machine.c:213
|  cpu_stopper_thread+0x19e/0x280 kernel/stop_machine.c:516
|  smpboot_thread_fn+0x1a8/0x300 kernel/smpboot.c:165
|  kthread+0x1b5/0x200 kernel/kthread.c:255
|  ret_from_fork+0x35/0x40 arch/x86/entry/entry_64.S:352
|
| Reported by Kernel Concurrency Sanitizer on:
| CPU: 7 PID: 44 Comm: migration/7 Not tainted 5.3.0+ #1
| Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Marco Elver <elver@google.com>
Link: https://lkml.kernel.org/r/20191007104536.27276-1-mark.rutland@arm.com
kernel/stop_machine.c