cgroups v2 allows the cpuset controller to be enabled/disabled on
demand. On Fedora 32, cpuset is disabled by default. To enable it,
a user needs to:
# cd /sys/fs/cgroup/
# echo +cpuset > cgroup.subtree_control
Existing cgroups will expose the cpuset interface (e.g., cpuset.cpus
file). By default, cpuset.cpus has no CPU assigned, which means that
existing tasks will move to a cpuset without cpus.
With that in mind, look what happens if a SCHED_DEADLINE task exists
on any cgroup (user.slice by default on Fedora):
----- %< -----
# chrt -d --sched-period 1000000000 --sched-runtime 100000000 0 sleep 100 &
# cd /sys/fs/cgroup/
# echo '+cpuset' > cgroup.subtree_control
[ 65.384041] BUG: unable to handle page fault for address: ffffffffb720f7e0
[ 65.384551] #PF: supervisor read access in kernel mode
[ 65.384923] #PF: error_code(0x0000) - not-present page
[ 65.385298] PGD 61a15067 P4D 61a15067 PUD 61a16063 PMD 800fffff9ddff062
[ 65.385781] Oops: 0000 [#1] SMP PTI
[ 65.386042] CPU: 0 PID: 799 Comm: sh Not tainted 5.10.0-rc3 #1
[ 65.386461] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.13.0-2.fc32 04/01/2014
[ 65.387077] RIP: 0010:dl_task_can_attach+0x40/0x250
[ 65.387429] Code: 54 55 53 48 83 ec 18 48 89 3c 24 bf ff ff ff ff e8 05 a2 52 00
4c 63 f0 48 c7 c5 00 9e 02 00 4a 8b 04 f5 00 09 47 b6 48 89 ea
<4c> 8b a4 10 e0 09 00 00 49 8d 44 24 40 48 89 c7 48 89 44 24
08 e8
[ 65.388768] RSP: 0018:ffffaee8c056fcd8 EFLAGS: 00010283
[ 65.389148] RAX: ffffffffb71e5000 RBX: ffffaee8c056fdd0 RCX: 0000000000000040
[ 65.389661] RDX: 0000000000029e00 RSI: ffff9db202534e48 RDI: ffffffffb6d3a3e0
[ 65.390174] RBP: 0000000000029e00 R08: 0000000000000000 R09: 0000000000000004
[ 65.390686] R10: 0000000000000001 R11: 00000000ffa6fbff R12: ffffaee8c056fbf0
[ 65.391196] R13: ffff9db2024e1400 R14: 0000000000000004 R15: ffff9db20ebb31e0
[ 65.391710] FS: 00007f6df41b1740(0000) GS:ffff9db377c00000(0000) knlGS:0000000000000000
[ 65.392289] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 65.392705] CR2: ffffffffb720f7e0 CR3: 000000010680a003 CR4: 0000000000370ef0
[ 65.393220] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 65.393732] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 65.394244] Call Trace:
[ 65.394437] cpuset_can_attach+0x8b/0x110
[ 65.394732] cgroup_migrate_execute+0x70/0x430
[ 65.395057] cgroup_update_dfl_csses+0x222/0x230
[ 65.395392] cgroup_subtree_control_write+0x2c6/0x3c0
[ 65.395759] kernfs_fop_write+0xce/0x1b0
[ 65.396048] vfs_write+0xc2/0x230
[ 65.396291] ksys_write+0x4f/0xc0
[ 65.396533] do_syscall_64+0x33/0x40
[ 65.396797] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 65.397166] RIP: 0033:0x7f6df42a6537
[ 65.397428] Code: 0d 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f
1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05
<48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 48 89 54 24 18 48 89
74 24
[ 65.398766] RSP: 002b:00007ffee4128018 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
[ 65.399838] RAX: ffffffffffffffda RBX: 0000000000000008 RCX: 00007f6df42a6537
[ 65.400923] RDX: 0000000000000008 RSI: 000055b3f7e549e0 RDI: 0000000000000001
[ 65.402003] RBP: 000055b3f7e549e0 R08: 000000000000000a R09: 0000000000000007
[ 65.403082] R10: 0000000000000004 R11: 0000000000000246 R12: 0000000000000008
[ 65.404156] R13: 00007f6df4378500 R14: 0000000000000008 R15: 00007f6df4378700
[ 65.405218] Modules linked in: <lots of modules>
[ 65.414172] CR2: ffffffffb720f7e0
[ 65.415117] ---[ end trace 2dbff1a688549e65 ]---
----- >% -----
That happens because on dl_task_can_attach():
dest_cpu = cpumask_any_and(cpu_active_mask, cs_cpus_allowed);
returns a non active cpu.
Initially, I thought about returning an error and blocking the
operation. However, that is indeed not needed. The cpuset without
CPUs assigned will be a non-root cpuset, hence its cpu mask will
be the same as the root one. So, the bandwidth was already accounted,
and the task can proceed.
Signed-off-by: Daniel Bristot de Oliveira <bristot@xxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Juri Lelli <juri.lelli@xxxxxxxxxx>
Cc: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
Cc: Dietmar Eggemann <dietmar.eggemann@xxxxxxx>
Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
Cc: Ben Segall <bsegall@xxxxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxx>
Cc: Daniel Bristot de Oliveira <bristot@xxxxxxxxxx>
Cc: Li Zefan <lizefan@xxxxxxxxxx>
Cc: Tejun Heo <tj@xxxxxxxxxx>
Cc: Johannes Weiner <hannes@xxxxxxxxxxx>
Cc: Valentin Schneider <valentin.schneider@xxxxxxx>
Cc: linux-kernel@xxxxxxxxxxxxxxx
Cc: cgroups@xxxxxxxxxxxxxxx
---
kernel/sched/deadline.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 943aa32cc1bc..788a391657a5 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -2871,6 +2871,13 @@ int dl_task_can_attach(struct task_struct *p,
bool overflow;
int ret;
+ /*
+ * The cpuset has no cpus assigned, so the thread will not
+ * change its affinity.
+ */
+ if (cpumask_empty(cs_cpus_allowed))
+ return 0;
+
/*
* The task is not moving to another root domain, so it is
* already accounted.
--
2.29.2
