Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1610806
| From | Greg Kroah-Hartman <gregkh@linuxfoundation.org> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 4.9 73/88] cgroup, net_cls: iterate the fds of only the tasks which are being migrated |
| Date | 2017-03-28 15:10 +0200 |
| Message-ID | <tpZ2O-83l-29@gated-at.bofh.it> (permalink) |
| References | <tpYJs-7Cs-3@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
4.9-stable review patch. If anyone has any objections, please let me know.
------------------
From: Tejun Heo <tj@kernel.org>
commit a05d4fd9176003e0c1f9c3d083f4dac19fd346ab upstream.
The net_cls controller controls the classid field of each socket which
is associated with the cgroup. Because the classid is per-socket
attribute, when a task migrates to another cgroup or the configured
classid of the cgroup changes, the controller needs to walk all
sockets and update the classid value, which was implemented by
3b13758f51de ("cgroups: Allow dynamically changing net_classid").
While the approach is not scalable, migrating tasks which have a lot
of fds attached to them is rare and the cost is born by the ones
initiating the operations. However, for simplicity, both the
migration and classid config change paths call update_classid() which
scans all fds of all tasks in the target css. This is an overkill for
the migration path which only needs to cover a much smaller subset of
tasks which are actually getting migrated in.
On cgroup v1, this can lead to unexpected scalability issues when one
tries to migrate a task or process into a net_cls cgroup which already
contains a lot of fds. Even if the migration traget doesn't have many
to get scanned, update_classid() ends up scanning all fds in the
target cgroup which can be extremely numerous.
Unfortunately, on cgroup v2 which doesn't use net_cls, the problem is
even worse. Before bfc2cf6f61fc ("cgroup: call subsys->*attach() only
for subsystems which are actually affected by migration"), cgroup core
would call the ->css_attach callback even for controllers which don't
see actual migration to a different css.
As net_cls is always disabled but still mounted on cgroup v2, whenever
a process is migrated on the cgroup v2 hierarchy, net_cls sees
identity migration from root to root and cgroup core used to call
->css_attach callback for those. The net_cls ->css_attach ends up
calling update_classid() on the root net_cls css to which all
processes on the system belong to as the controller isn't used. This
makes any cgroup v2 migration O(total_number_of_fds_on_the_system)
which is horrible and easily leads to noticeable stalls triggering RCU
stall warnings and so on.
The worst symptom is already fixed in upstream by bfc2cf6f61fc
("cgroup: call subsys->*attach() only for subsystems which are
actually affected by migration"); however, backporting that commit is
too invasive and we want to avoid other cases too.
This patch updates net_cls's cgrp_attach() to iterate fds of only the
processes which are actually getting migrated. This removes the
surprising migration cost which is dependent on the total number of
fds in the target cgroup. As this leaves write_classid() the only
user of update_classid(), open-code the helper into write_classid().
Reported-by: David Goode <dgoode@fb.com>
Fixes: 3b13758f51de ("cgroups: Allow dynamically changing net_classid")
Cc: Nina Schiff <ninasc@fb.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
net/core/netclassid_cgroup.c | 32 ++++++++++++++++----------------
1 file changed, 16 insertions(+), 16 deletions(-)
--- a/net/core/netclassid_cgroup.c
+++ b/net/core/netclassid_cgroup.c
@@ -69,27 +69,17 @@ static int update_classid_sock(const voi
return 0;
}
-static void update_classid(struct cgroup_subsys_state *css, void *v)
+static void cgrp_attach(struct cgroup_taskset *tset)
{
- struct css_task_iter it;
+ struct cgroup_subsys_state *css;
struct task_struct *p;
- css_task_iter_start(css, &it);
- while ((p = css_task_iter_next(&it))) {
+ cgroup_taskset_for_each(p, css, tset) {
task_lock(p);
- iterate_fd(p->files, 0, update_classid_sock, v);
+ iterate_fd(p->files, 0, update_classid_sock,
+ (void *)(unsigned long)css_cls_state(css)->classid);
task_unlock(p);
}
- css_task_iter_end(&it);
-}
-
-static void cgrp_attach(struct cgroup_taskset *tset)
-{
- struct cgroup_subsys_state *css;
-
- cgroup_taskset_first(tset, &css);
- update_classid(css,
- (void *)(unsigned long)css_cls_state(css)->classid);
}
static u64 read_classid(struct cgroup_subsys_state *css, struct cftype *cft)
@@ -101,12 +91,22 @@ static int write_classid(struct cgroup_s
u64 value)
{
struct cgroup_cls_state *cs = css_cls_state(css);
+ struct css_task_iter it;
+ struct task_struct *p;
cgroup_sk_alloc_disable();
cs->classid = (u32)value;
- update_classid(css, (void *)(unsigned long)cs->classid);
+ css_task_iter_start(css, &it);
+ while ((p = css_task_iter_next(&it))) {
+ task_lock(p);
+ iterate_fd(p->files, 0, update_classid_sock,
+ (void *)(unsigned long)cs->classid);
+ task_unlock(p);
+ }
+ css_task_iter_end(&it);
+
return 0;
}
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH 4.9 00/88] 4.9.19-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 23/88] Input: iforce - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 21/88] Input: elan_i2c - add ASUS EeeBook X205TA special touchpad fw Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 44/88] USB: usbtmc: add missing endpoint sanity check Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 79/88] intel_th: Dont leak module refcount on failure to activate Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 43/88] usb: hub: Fix crash after failure to read BOS descriptor Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 67/88] xen/acpi: upload PM state from init-domain to Xen Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 82/88] libceph: dont set weight to IN when OSD is destroyed Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 22/88] Input: i8042 - add noloop quirk for Dell Embedded Box PC 3000 Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 72/88] cpufreq: Restore policy min/max limits on CPU online Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 24/88] Input: ims-pcu - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 01/88] net/openvswitch: Set the ipv6 source tunnel key address attribute correctly Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 47/88] uwb: hwa-rc: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 50/88] iio: sw-device: Fix config group initialization Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 17/88] socket, bpf: fix sk_filter use after free in sk_clone_lock Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 41/88] USB: wusbcore: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 42/88] usb: musb: cppi41: dont check early-TX-interrupt for Isoch transfer Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 39/88] USB: lvtest: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 56/88] powerpc/64s: Fix idle wakeup potential to clobber registers Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 86/88] drm: reference count event->completion Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 85/88] nl80211: fix dumpit error path RTNL deadlocks Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 10/88] net/mlx5e: Use the proper UAPI values when offloading TC vlan actions Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 31/88] ALSA: ctxfi: Fix the incorrect check of dma_set_mask() call Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 53/88] parport: fix attempt to write duplicate procfiles Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 32/88] ALSA: hda - Adding a group of pin definition to fix headset problem Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 38/88] USB: uss720: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 28/88] Input: kbtab - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 81/88] Drivers: hv: vmbus: Dont leak memory when a channel is rescinded Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 30/88] ALSA: seq: Fix racy cell insertions during snd_seq_pool_done() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 88/88] crypto: algif_hash - avoid zero-sized array Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 55/88] ext4: lock the xattr block before checksuming it Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 61/88] hwrng: amd - Revert managed API changes Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 14:50 +0200
[PATCH 4.9 83/88] device-dax: fix pmd/pte fault fallback handling Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 45/88] USB: usbtmc: fix probe error path Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 84/88] drm/bridge: analogix dp: Fix runtime PM state on driver bind Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 62/88] hwrng: geode - Revert managed API changes Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 87/88] fbcon: Fix vc attr at deinit Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 66/88] crypto: ccp - Assign DMA commands to the channels CCP Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 48/88] mmc: ushc: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 46/88] uwb: i1480-dfu: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 80/88] Drivers: hv: vmbus: Dont leak channel ids Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:00 +0200
[PATCH 4.9 08/88] net: vrf: Reset rt6i_idev in local dst after put Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 71/88] arm64: kaslr: Fix up the kernel image alignment Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 06/88] net: unix: properly re-increment inflight counter of GC discarded candidates Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 63/88] clk: sunxi-ng: sun6i: Fix enable bit offset for hdmi-ddc module clock Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 59/88] mmc: sdhci: Do not disable interrupts while waiting for clock Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 51/88] iio: hid-sensor-trigger: Change get poll value function order to avoid sensor properties losing after resume from S3 Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 52/88] iio: magnetometer: ak8974: remove incorrect __exit markups Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 05/88] openvswitch: Add missing case OVS_TUNNEL_KEY_ATTR_PAD Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 78/88] jbd2: dont leak memory if setting up journal fails Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 58/88] mmc: sdhci-of-arasan: fix incorrect timeout clock Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 54/88] ext4: mark inode dirty after converting inline directory Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 73/88] cgroup, net_cls: iterate the fds of only the tasks which are being migrated Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 74/88] blk-mq: dont complete un-started request in timeout handler Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 40/88] USB: idmouse: fix NULL-deref at probe Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 64/88] clk: sunxi-ng: mp: Adjust parent rate for pre-dividers Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 77/88] auxdisplay: img-ascii-lcd: add missing sentinel entry in img_ascii_lcd_matches Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 76/88] drm/amdgpu: reinstate oland workaround for sclk Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 09/88] net/mlx5: Add missing entries for set/query rate limit commands Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 57/88] mmc: sdhci-of-at91: Support external regulators Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 69/88] Revert "ARM: at91/dt: sama5d2: Use new compatible for ohci node" Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 65/88] mwifiex: pcie: dont leak DMA buffers when removing Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 68/88] iommu/vt-d: Fix NULL pointer dereference in device_to_iommu Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 37/88] usb-core: Add LINEAR_FRAME_INTR_BINTERVAL USB quirk Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 70/88] ARM: at91: pm: cpu_idle: switch DDR to power-down mode Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 49/88] iio: adc: ti_am335x_adc: fix fifo overrun recovery Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
[PATCH 4.9 75/88] libceph: force GFP_NOIO for socket allocations Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:10 +0200
Re: [PATCH 4.9 75/88] libceph: force GFP_NOIO for socket allocations Michal Hocko <mhocko@kernel.org> - 2017-03-29 10:20 +0200
[PATCH 4.9 14/88] ipv6: make sure to initialize sockc.tsflags before first use Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 16/88] ipv4: provide stronger user input validation in nl_fib_input() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 02/88] net: bcmgenet: Do not suspend PHY if Wake-on-LAN is enabled Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 25/88] Input: hanwang - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 15/88] net: bcmgenet: remove bcmgenet_internal_phy_setup() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 04/88] amd-xgbe: Fix jumbo MTU processing on newer hardware Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 36/88] usb: gadget: f_uvc: Fix SuperSpeed companion descriptors wBytesPerInterval Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 29/88] Input: sur40 - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 03/88] net: properly release sk_frag.page Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 26/88] Input: yealink - validate number of endpoints before using them Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 12/88] net/mlx5e: Count GSO packets correctly Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 13/88] net/mlx5e: Count LRO packets correctly Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 33/88] USB: serial: option: add Quectel UC15, UC20, EC21, and EC25 modems Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 20/88] Input: ALPS - fix trackstick button handling on V8 devices Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 18/88] tcp: initialize icsk_ack.lrcvtime at session start time Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 19/88] Input: ALPS - fix V8+ protocol handling (73 03 28) Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
[PATCH 4.9 11/88] net/mlx5: Increase number of max QPs in default profile Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-28 15:20 +0200
Re: [PATCH 4.9 00/88] 4.9.19-stable review Shuah Khan <shuahkh@osg.samsung.com> - 2017-03-28 21:50 +0200
Re: [PATCH 4.9 00/88] 4.9.19-stable review Guenter Roeck <linux@roeck-us.net> - 2017-03-29 06:50 +0200
Re: [PATCH 4.9 00/88] 4.9.19-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2017-03-29 07:50 +0200
csiph-web