Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1626399
| From | Greg Kroah-Hartman <gregkh@linuxfoundation.org> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 4.10 01/69] cgroup, kthread: close race window where new kthreads can be migrated to non-root cgroups |
| Date | 2017-04-19 17:40 +0200 |
| Message-ID | <txZS3-59Y-37@gated-at.bofh.it> (permalink) |
| References | <txYVX-4yJ-3@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
4.10-stable review patch. If anyone has any objections, please let me know.
------------------
From: Tejun Heo <tj@kernel.org>
commit 77f88796cee819b9c4562b0b6b44691b3b7755b1 upstream.
Creation of a kthread goes through a couple interlocked stages between
the kthread itself and its creator. Once the new kthread starts
running, it initializes itself and wakes up the creator. The creator
then can further configure the kthread and then let it start doing its
job by waking it up.
In this configuration-by-creator stage, the creator is the only one
that can wake it up but the kthread is visible to userland. When
altering the kthread's attributes from userland is allowed, this is
fine; however, for cases where CPU affinity is critical,
kthread_bind() is used to first disable affinity changes from userland
and then set the affinity. This also prevents the kthread from being
migrated into non-root cgroups as that can affect the CPU affinity and
many other things.
Unfortunately, the cgroup side of protection is racy. While the
PF_NO_SETAFFINITY flag prevents further migrations, userland can win
the race before the creator sets the flag with kthread_bind() and put
the kthread in a non-root cgroup, which can lead to all sorts of
problems including incorrect CPU affinity and starvation.
This bug got triggered by userland which periodically tries to migrate
all processes in the root cpuset cgroup to a non-root one. Per-cpu
workqueue workers got caught while being created and ended up with
incorrected CPU affinity breaking concurrency management and sometimes
stalling workqueue execution.
This patch adds task->no_cgroup_migration which disallows the task to
be migrated by userland. kthreadd starts with the flag set making
every child kthread start in the root cgroup with migration
disallowed. The flag is cleared after the kthread finishes
initialization by which time PF_NO_SETAFFINITY is set if the kthread
should stay in the root cgroup.
It'd be better to wait for the initialization instead of failing but I
couldn't think of a way of implementing that without adding either a
new PF flag, or sleeping and retrying from waiting side. Even if
userland depends on changing cgroup membership of a kthread, it either
has to be synchronized with kthread_create() or periodically repeat,
so it's unlikely that this would break anything.
v2: Switch to a simpler implementation using a new task_struct bit
field suggested by Oleg.
Signed-off-by: Tejun Heo <tj@kernel.org>
Suggested-by: Oleg Nesterov <oleg@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Reported-and-debugged-by: Chris Mason <clm@fb.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
include/linux/cgroup.h | 21 +++++++++++++++++++++
include/linux/sched.h | 4 ++++
kernel/cgroup.c | 9 +++++----
kernel/kthread.c | 3 +++
4 files changed, 33 insertions(+), 4 deletions(-)
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -570,6 +570,25 @@ static inline void pr_cont_cgroup_path(s
pr_cont_kernfs_path(cgrp->kn);
}
+static inline void cgroup_init_kthreadd(void)
+{
+ /*
+ * kthreadd is inherited by all kthreads, keep it in the root so
+ * that the new kthreads are guaranteed to stay in the root until
+ * initialization is finished.
+ */
+ current->no_cgroup_migration = 1;
+}
+
+static inline void cgroup_kthread_ready(void)
+{
+ /*
+ * This kthread finished initialization. The creator should have
+ * set PF_NO_SETAFFINITY if this kthread should stay in the root.
+ */
+ current->no_cgroup_migration = 0;
+}
+
#else /* !CONFIG_CGROUPS */
struct cgroup_subsys_state;
@@ -590,6 +609,8 @@ static inline void cgroup_free(struct ta
static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
+static inline void cgroup_init_kthreadd(void) {}
+static inline void cgroup_kthread_ready(void) {}
static inline bool task_under_cgroup_hierarchy(struct task_struct *task,
struct cgroup *ancestor)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1620,6 +1620,10 @@ struct task_struct {
#ifdef CONFIG_COMPAT_BRK
unsigned brk_randomized:1;
#endif
+#ifdef CONFIG_CGROUPS
+ /* disallow userland-initiated cgroup migration */
+ unsigned no_cgroup_migration:1;
+#endif
unsigned long atomic_flags; /* Flags needing atomic access. */
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -2920,11 +2920,12 @@ static ssize_t __cgroup_procs_write(stru
tsk = tsk->group_leader;
/*
- * Workqueue threads may acquire PF_NO_SETAFFINITY and become
- * trapped in a cpuset, or RT worker may be born in a cgroup
- * with no rt_runtime allocated. Just say no.
+ * kthreads may acquire PF_NO_SETAFFINITY during initialization.
+ * If userland migrates such a kthread to a non-root cgroup, it can
+ * become trapped in a cpuset, or RT kthread may be born in a
+ * cgroup with no rt_runtime allocated. Just say no.
*/
- if (tsk == kthreadd_task || (tsk->flags & PF_NO_SETAFFINITY)) {
+ if (tsk->no_cgroup_migration || (tsk->flags & PF_NO_SETAFFINITY)) {
ret = -EINVAL;
goto out_unlock_rcu;
}
--- a/kernel/kthread.c
+++ b/kernel/kthread.c
@@ -18,6 +18,7 @@
#include <linux/freezer.h>
#include <linux/ptrace.h>
#include <linux/uaccess.h>
+#include <linux/cgroup.h>
#include <trace/events/sched.h>
static DEFINE_SPINLOCK(kthread_create_lock);
@@ -223,6 +224,7 @@ static int kthread(void *_create)
ret = -EINTR;
if (!test_bit(KTHREAD_SHOULD_STOP, &self->flags)) {
+ cgroup_kthread_ready();
__kthread_parkme(self);
ret = threadfn(data);
}
@@ -536,6 +538,7 @@ int kthreadd(void *unused)
set_mems_allowed(node_states[N_MEMORY]);
current->flags |= PF_NOFREEZE;
+ cgroup_init_kthreadd();
for (;;) {
set_current_state(TASK_INTERRUPTIBLE);
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH 4.10 00/69] 4.10.12-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 18/69] CIFS: store results of cifs_reopen_file to avoid infinite wait Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 20/69] perf annotate s390: Fix perf annotate error -95 (4.10 regression) Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 48/69] cpufreq: Bring CPUs up even if cpufreq_online() failed Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 35/69] scsi: sr: Sanity check returned mode data Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 15/69] drm/etnaviv: fix missing unlock on error in etnaviv_gpu_submit() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 08/69] zsmalloc: expand class bit Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 11/69] drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 26/69] x86/vdso: Ensure vdso32_enabled gets set to valid values only Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:40 +0200
[PATCH 4.10 42/69] make skb_copy_datagram_msg() et.al. preserve ->msg_iter on error Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 67/69] mm: Tighten x86 /dev/mem with zeroing reads Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 39/69] target: Avoid mappedlun symlink creation during lun shutdown Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 52/69] char: lack of bool string made CONFIG_DEVPORT always on Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 34/69] iscsi-target: Drop work-around for legacy GlobalSAN initiator Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 43/69] libnvdimm: fix blk free space accounting Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 40/69] xen, fbfront: fix connecting to backend Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 69/69] virtio-console: avoid DMA from stack Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 65/69] ACPI / EC: Use busy polling mode when GPE is not enabled Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 66/69] rtc: tegra: Implement clock handling Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 41/69] [iov_iter] new privimitive: iov_iter_revert() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 37/69] scsi: qla2xxx: Add fix to read correct register value for ISP82xx. Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 16:50 +0200
[PATCH 4.10 56/69] crypto: algif_aead - Fix bogus request dereference in completion function Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 54/69] zram: do not use copy_page with non-page aligned address Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 58/69] crypto: ahash - Fix EINPROGRESS notification callback Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 61/69] [media] dvb-usb-v2: avoid use-after-free Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 36/69] scsi: sd: Consider max_xfer_blocks if opt_xfer_blocks is unusable Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 59/69] crypto: lrw - Fix use-after-free on EINPROGRESS Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 38/69] scsi: sd: Fix capacity calculation with 32-bit sector_t Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 57/69] crypto: xts - Fix use-after-free on EINPROGRESS Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 64/69] x86/xen: Fix APIC id mismatch warning on Intel Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 62/69] ASoC: Intel: select DW_DMAC_CORE since its mandatory Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:20 +0200
[PATCH 4.10 55/69] ftrace: Fix function pid filter on instances Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 06/69] thp: fix MADV_DONTNEED vs. MADV_FREE race Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 23/69] x86/signals: Fix lower/upper bound reporting in compat siginfo Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 45/69] libnvdimm: band aid btt vs clear poison locking Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 07/69] thp: fix MADV_DONTNEED vs clear soft dirty race Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 50/69] ftrace: Fix removing of second function probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 49/69] irqchip/irq-imx-gpcv2: Fix spinlock initialization Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 51/69] drm/i915/gvt: set the correct default value of CTX STATUS PTR Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 46/69] can: ifi: use correct register to read rx status Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 31/69] efi/libstub: Skip GOP with PIXEL_BLT_ONLY format Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 22/69] x86/efi: Dont try to reserve runtime regions Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 25/69] x86, pmem: fix broken __copy_user_nocache cache-bypass assumptions Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 28/69] acpi, nfit, libnvdimm: fix interleave set cookie calculation (64-bit comparison) Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 27/69] x86/vdso: Plug race between mapping and ELF header setup Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 53/69] Revert "MIPS: Lantiq: Fix cascaded IRQ setup" Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 32/69] efi/fb: Avoid reconfiguration of BAR that covers the framebuffer Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 09/69] orangefs: free superblock when mount fails Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 03/69] tcmu: Fix possible overwrite of t_data_sgs last iov[] Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 04/69] tcmu: Fix wrongly calculating of the base_command_size Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 29/69] ACPI / scan: Set the visited flag for all enumerated devices Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 47/69] pwm: rockchip: State of PWM clock should synchronize with PWM enabled state Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 05/69] tcmu: Skip Data-Out blocks before gathering Data-In buffer for BIDI case Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 60/69] parisc: Fix get_user() for 64-bit value on 32-bit kernel Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 44/69] libnvdimm: fix reconfig_mutex, mmap_sem, and jbd2_handle lockdep splat Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:30 +0200
[PATCH 4.10 19/69] Input: xpad - add support for Razer Wildcat gamepad Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 17/69] CIFS: reconnect thread reschedule itself Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 13/69] drm/nouveau/kms/nv50: fix double dma_fence_put() when destroying plane state Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 02/69] audit: make sure we dont let the retry queue grow without bounds Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 01/69] cgroup, kthread: close race window where new kthreads can be migrated to non-root cgroups Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 10/69] drm/nouveau/mpeg: mthd returns true on success now Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
[PATCH 4.10 12/69] drm/nouveau/kms/nv50: fix setting of HeadSetRasterVertBlankDmi method Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-19 17:40 +0200
Re: [PATCH 4.10 00/69] 4.10.12-stable review Shuah Khan <shuahkh@osg.samsung.com> - 2017-04-19 22:40 +0200
Re: [PATCH 4.10 00/69] 4.10.12-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-20 08:40 +0200
Re: [PATCH 4.10 00/69] 4.10.12-stable review Guenter Roeck <linux@roeck-us.net> - 2017-04-20 01:30 +0200
Re: [PATCH 4.10 00/69] 4.10.12-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-04-20 08:40 +0200
csiph-web