On 07/18/2017 03:51 PM, Tejun Heo wrote:
> Hello, Waiman.
>
> On Tue, Jul 18, 2017 at 03:32:16PM -0400, Waiman Long wrote:
>> It was found that when a cgroup2 filesystem was mounted, control
>> files other than the base cgroup.* ones were not shown in the root
>> directory. They were shown only after some controllers were activated
>> in the root's cgroup.subtree_control file.
>>
>> This was caused by a lack of the kernfs_activate() call which was fixed
>> by this patch.
>>
>> Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
>> ---
>> kernel/cgroup/cgroup.c | 4 +++-
>> 1 file changed, 3 insertions(+), 1 deletion(-)
>>
>> diff --git a/kernel/cgroup/cgroup.c b/kernel/cgroup/cgroup.c
>> index 89a23c6..fb1893b 100644
>> --- a/kernel/cgroup/cgroup.c
>> +++ b/kernel/cgroup/cgroup.c
>> @@ -2024,8 +2024,10 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type,
>>
>> dentry = cgroup_do_mount(&cgroup2_fs_type, flags, &cgrp_dfl_root,
>> CGROUP2_SUPER_MAGIC, ns);
>> - if (!IS_ERR(dentry))
>> + if (!IS_ERR(dentry)) {
>> apply_cgroup_root_flags(root_flags);
>> + kernfs_activate(cgrp_dfl_root.cgrp.kn);
>> + }
> Heh, that's tricky. I'm not quite sure where the unactivated files
> are being added tho because that'd be where we should be activating.
> I *think* that they are already activated as part of
> cgroup_add_cftypes() but I am obviously qmissing something. I'll try
> to repro the issue and find where we're skipping the activation call.
>
> Thanks!
>
>From my own debugging, the controller files (e.g. the debug controller)
were indirectly populated by the rebind_subsystems() call.
[ 1.628103] css_populate_dir: init subsystem debug
[ 1.628944] ------------[ cut here ]------------
[ 1.629796] WARNING: CPU: 0 PID: 1 at kernel/cgroup/cgroup.c:1584
css_populate_dir+0x141/0x180
[ 1.631401] Modules linked in:
[ 1.632052] CPU: 0 PID: 1 Comm: systemd Tainted: G W
4.13.0-rc1-cgroup2+ #13
[ 1.633642] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
[ 1.634627] task: ffff91d84e1f0000 task.stack: ffffb40b8189c000
[ 1.635630] RIP: 0010:css_populate_dir+0x141/0x180
[ 1.636548] RSP: 0018:ffffb40b8189fb88 EFLAGS: 00010246
[ 1.637457] RAX: 0000000000000026 RBX: ffff91d84fcb7ec0 RCX:
ffffffffaec60ae8
[ 1.638538] RDX: 0000000000000000 RSI: 0000000000000082 RDI:
0000000000000246
[ 1.639689] RBP: ffffb40b8189fbb0 R08: 0000000000000000 R09:
0000000000000325
[ 1.640889] R10: 00000000ffffffff R11: 0000000000000324 R12:
ffffffffaecec3f8
[ 1.642071] R13: ffff91d84fcb7ec0 R14: ffffffffaf1f6770 R15:
ffffffffaeceea60
[ 1.643263] FS: 00007fc8c9624940(0000) GS:ffff91dad7200000(0000)
knlGS:0000000000000000
[ 1.651469] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 1.653184] CR2: 00007fc8c9630000 CR3: 0000000190bff000 CR4:
00000000001406f0
[ 1.654401] Call Trace:
[ 1.654975] cgroup_apply_control_enable+0x103/0x340
[ 1.655906] ? css_next_descendant_pre+0x35/0x40
[ 1.656834] ? cgroup_propagate_control+0x101/0x150
[ 1.657719] cgroup_apply_control+0x1a/0x30
[ 1.658521] rebind_subsystems+0x18a/0x3b0
[ 1.659272] cgroup_setup_root+0x18f/0x380
[ 1.660086] cgroup1_mount+0x2c5/0x490
[ 1.660842] cgroup_mount+0x9d/0x390
[ 1.661586] mount_fs+0x39/0x150
[ 1.662238] vfs_kern_mount+0x67/0x130
[ 1.663027] do_mount+0x1e2/0xc90
[ 1.663718] ? kmem_cache_alloc_trace+0x14b/0x1b0
[ 1.664548] SyS_mount+0x83/0xd0
[ 1.665254] entry_SYSCALL_64_fastpath+0x1a/0xa5
For the default cgroup2 root, kernfs_activate() was only called at the
beginning in cgroup_init() with only the base cgroup files added. No
more call after that until I touched the cgroup.subtree_control file.
Cheers,
Longman
--
To unsubscribe from this list: send the line "unsubscribe cgroups" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
