• Tejun Heo's avatar
    cgroup/pids: remove spurious suspicious RCU usage warning · 228514bf
    Tejun Heo authored
    commit 1d18c274 upstream.
    
    pids_can_fork() is special in that the css association is guaranteed
    to be stable throughout the function and thus doesn't need RCU
    protection around task_css access.  When determining the css to charge
    the pid, task_css_check() is used to override the RCU sanity check.
    
    While adding a warning message on fork rejection from pids limit,
    135b8b37 ("cgroup: Add pids controller event when fork fails
    because of pid limit") incorrectly added a task_css access which is
    neither RCU protected or explicitly annotated.  This triggers the
    following suspicious RCU usage warning when RCU debugging is enabled.
    
      cgroup: fork rejected by pids controller in
    
      ===============================
      [ ERR: suspicious RCU usage.  ]
      4.10.0-work+ #1 Not tainted
      -------------------------------
      ./include/linux/cgroup.h:435 suspicious rcu_dereference_check() usage!
    
      other info that might help us debug this:
    
      rcu_scheduler_active = 2, debug_locks = 0
      1 lock held by bash/1748:
       #0:  (&cgroup_threadgroup_rwsem){+++++.}, at: [<ffffffff81052c96>] _do_fork+0xe6/0x6e0
    
      stack backtrace:
      CPU: 3 PID: 1748 Comm: bash Not tainted 4.10.0-work+ #1
      Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.3-1.fc25 04/01/2014
      Call Trace:
       dump_stack+0x68/0x93
       lockdep_rcu_suspicious+0xd7/0x110
       pids_can_fork+0x1c7/0x1d0
       cgroup_can_fork+0x67/0xc0
       copy_process.part.58+0x1709/0x1e90
       _do_fork+0xe6/0x6e0
       SyS_clone+0x19/0x20
       do_syscall_64+0x5c/0x140
       entry_SYSCALL64_slow_path+0x25/0x25
      RIP: 0033:0x7f7853fab93a
      RSP: 002b:00007ffc12d05c90 EFLAGS: 00000246 ORIG_RAX: 0000000000000038
      RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7853fab93a
      RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000001200011
      RBP: 00007ffc12d05cc0 R08: 0000000000000000 R09: 00007f78548db700
      R10: 00007f78548db9d0 R11: 0000000000000246 R12: 00000000000006d4
      R13: 0000000000000001 R14: 0000000000000000 R15: 000055e3ebe2c04d
      /asdf
    
    There's no reason to dereference task_css again here when the
    associated css is already available.  Fix it by replacing the
    task_cgroup() call with css->cgroup.
    Signed-off-by: 's avatarTejun Heo <tj@kernel.org>
    Reported-by: 's avatarMike Galbraith <efault@gmx.de>
    Fixes: 135b8b37 ("cgroup: Add pids controller event when fork fails because of pid limit")
    Cc: Kenny Yu <kennyyu@fb.com>
    Signed-off-by: 's avatarTejun Heo <tj@kernel.org>
    Signed-off-by: 's avatarGreg Kroah-Hartman <gregkh@linuxfoundation.org>
    228514bf