target/linux/generic/backport-6.1/836-v6.7-leds-trigger-netdev-fix-RTNL-handling-to-prevent-pot.patch

   1 From fe2b1226656afae56702d1d84c6900f6b67df297 Mon Sep 17 00:00:00 2001
   2 From: Heiner Kallweit <hkallweit1@gmail.com>
   3 Date: Fri, 1 Dec 2023 11:23:22 +0100
   4 Subject: [PATCH] leds: trigger: netdev: fix RTNL handling to prevent potential
   5  deadlock
   6
   7 When working on LED support for r8169 I got the following lockdep
   8 warning. Easiest way to prevent this scenario seems to be to take
   9 the RTNL lock before the trigger_data lock in set_device_name().
  10
  11 ======================================================
  12 WARNING: possible circular locking dependency detected
  13 6.7.0-rc2-next-20231124+ #2 Not tainted
  14 ------------------------------------------------------
  15 bash/383 is trying to acquire lock:
  16 ffff888103aa1c68 (&trigger_data->lock){+.+.}-{3:3}, at: netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
  17
  18 but task is already holding lock:
  19 ffffffff8cddf808 (rtnl_mutex){+.+.}-{3:3}, at: rtnl_lock+0x12/0x20
  20
  21 which lock already depends on the new lock.
  22
  23 the existing dependency chain (in reverse order) is:
  24
  25 -> #1 (rtnl_mutex){+.+.}-{3:3}:
  26        __mutex_lock+0x9b/0xb50
  27        mutex_lock_nested+0x16/0x20
  28        rtnl_lock+0x12/0x20
  29        set_device_name+0xa9/0x120 [ledtrig_netdev]
  30        netdev_trig_activate+0x1a1/0x230 [ledtrig_netdev]
  31        led_trigger_set+0x172/0x2c0
  32        led_trigger_write+0xf1/0x140
  33        sysfs_kf_bin_write+0x5d/0x80
  34        kernfs_fop_write_iter+0x15d/0x210
  35        vfs_write+0x1f0/0x510
  36        ksys_write+0x6c/0xf0
  37        __x64_sys_write+0x14/0x20
  38        do_syscall_64+0x3f/0xf0
  39        entry_SYSCALL_64_after_hwframe+0x6c/0x74
  40
  41 -> #0 (&trigger_data->lock){+.+.}-{3:3}:
  42        __lock_acquire+0x1459/0x25a0
  43        lock_acquire+0xc8/0x2d0
  44        __mutex_lock+0x9b/0xb50
  45        mutex_lock_nested+0x16/0x20
  46        netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
  47        call_netdevice_register_net_notifiers+0x5a/0x100
  48        register_netdevice_notifier+0x85/0x120
  49        netdev_trig_activate+0x1d4/0x230 [ledtrig_netdev]
  50        led_trigger_set+0x172/0x2c0
  51        led_trigger_write+0xf1/0x140
  52        sysfs_kf_bin_write+0x5d/0x80
  53        kernfs_fop_write_iter+0x15d/0x210
  54        vfs_write+0x1f0/0x510
  55        ksys_write+0x6c/0xf0
  56        __x64_sys_write+0x14/0x20
  57        do_syscall_64+0x3f/0xf0
  58        entry_SYSCALL_64_after_hwframe+0x6c/0x74
  59
  60 other info that might help us debug this:
  61
  62  Possible unsafe locking scenario:
  63
  64        CPU0                    CPU1
  65        ----                    ----
  66   lock(rtnl_mutex);
  67                                lock(&trigger_data->lock);
  68                                lock(rtnl_mutex);
  69   lock(&trigger_data->lock);
  70
  71  *** DEADLOCK ***
  72
  73 8 locks held by bash/383:
  74  #0: ffff888103ff33f0 (sb_writers#3){.+.+}-{0:0}, at: ksys_write+0x6c/0xf0
  75  #1: ffff888103aa1e88 (&of->mutex){+.+.}-{3:3}, at: kernfs_fop_write_iter+0x114/0x210
  76  #2: ffff8881036f1890 (kn->active#82){.+.+}-{0:0}, at: kernfs_fop_write_iter+0x11d/0x210
  77  #3: ffff888108e2c358 (&led_cdev->led_access){+.+.}-{3:3}, at: led_trigger_write+0x30/0x140
  78  #4: ffffffff8cdd9e10 (triggers_list_lock){++++}-{3:3}, at: led_trigger_write+0x75/0x140
  79  #5: ffff888108e2c270 (&led_cdev->trigger_lock){++++}-{3:3}, at: led_trigger_write+0xe3/0x140
  80  #6: ffffffff8cdde3d0 (pernet_ops_rwsem){++++}-{3:3}, at: register_netdevice_notifier+0x1c/0x120
  81  #7: ffffffff8cddf808 (rtnl_mutex){+.+.}-{3:3}, at: rtnl_lock+0x12/0x20
  82
  83 stack backtrace:
  84 CPU: 0 PID: 383 Comm: bash Not tainted 6.7.0-rc2-next-20231124+ #2
  85 Hardware name: Default string Default string/Default string, BIOS ADLN.M6.SODIMM.ZB.CY.015 08/08/2023
  86 Call Trace:
  87  <TASK>
  88  dump_stack_lvl+0x5c/0xd0
  89  dump_stack+0x10/0x20
  90  print_circular_bug+0x2dd/0x410
  91  check_noncircular+0x131/0x150
  92  __lock_acquire+0x1459/0x25a0
  93  lock_acquire+0xc8/0x2d0
  94  ? netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
  95  __mutex_lock+0x9b/0xb50
  96  ? netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
  97  ? __this_cpu_preempt_check+0x13/0x20
  98  ? netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
  99  ? __cancel_work_timer+0x11c/0x1b0
 100  ? __mutex_lock+0x123/0xb50
 101  mutex_lock_nested+0x16/0x20
 102  ? mutex_lock_nested+0x16/0x20
 103  netdev_trig_notify+0xec/0x190 [ledtrig_netdev]
 104  call_netdevice_register_net_notifiers+0x5a/0x100
 105  register_netdevice_notifier+0x85/0x120
 106  netdev_trig_activate+0x1d4/0x230 [ledtrig_netdev]
 107  led_trigger_set+0x172/0x2c0
 108  ? preempt_count_add+0x49/0xc0
 109  led_trigger_write+0xf1/0x140
 110  sysfs_kf_bin_write+0x5d/0x80
 111  kernfs_fop_write_iter+0x15d/0x210
 112  vfs_write+0x1f0/0x510
 113  ksys_write+0x6c/0xf0
 114  __x64_sys_write+0x14/0x20
 115  do_syscall_64+0x3f/0xf0
 116  entry_SYSCALL_64_after_hwframe+0x6c/0x74
 117 RIP: 0033:0x7f269055d034
 118 Code: c7 00 16 00 00 00 b8 ff ff ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 f3 0f 1e fa 80 3d 35 c3 0d 00 00 74 13 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 48 83 ec 28 48 89 54 24 18 48
 119 RSP: 002b:00007ffddb7ef748 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
 120 RAX: ffffffffffffffda RBX: 0000000000000007 RCX: 00007f269055d034
 121 RDX: 0000000000000007 RSI: 000055bf5f4af3c0 RDI: 0000000000000001
 122 RBP: 000055bf5f4af3c0 R08: 0000000000000073 R09: 0000000000000001
 123 R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000007
 124 R13: 00007f26906325c0 R14: 00007f269062ff20 R15: 0000000000000000
 125  </TASK>
 126
 127 Fixes: d5e01266e7f5 ("leds: trigger: netdev: add additional specific link speed mode")
 128 Cc: stable@vger.kernel.org
 129 Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
 130 Reviewed-by: Andrew Lunn <andrew@lunn.ch>
 131 Acked-by: Lee Jones <lee@kernel.org>
 132 Link: https://lore.kernel.org/r/fb5c8294-2a10-4bf5-8f10-3d2b77d2757e@gmail.com
 133 Signed-off-by: Jakub Kicinski <kuba@kernel.org>
 134 ---
 135  drivers/leds/trigger/ledtrig-netdev.c | 11 +++++++----
 136  1 file changed, 7 insertions(+), 4 deletions(-)
 137
 138 --- a/drivers/leds/trigger/ledtrig-netdev.c
 139 +++ b/drivers/leds/trigger/ledtrig-netdev.c
 140 @@ -235,6 +235,11 @@ static int set_device_name(struct led_ne
 141  {
 142         cancel_delayed_work_sync(&trigger_data->work);
 143
 144 +       /*
 145 +        * Take RTNL lock before trigger_data lock to prevent potential
 146 +        * deadlock with netdev notifier registration.
 147 +        */
 148 +       rtnl_lock();
 149         mutex_lock(&trigger_data->lock);
 150
 151         if (trigger_data->net_dev) {
 152 @@ -254,16 +259,14 @@ static int set_device_name(struct led_ne
 153         trigger_data->carrier_link_up = false;
 154         trigger_data->link_speed = SPEED_UNKNOWN;
 155         trigger_data->duplex = DUPLEX_UNKNOWN;
 156 -       if (trigger_data->net_dev != NULL) {
 157 -               rtnl_lock();
 158 +       if (trigger_data->net_dev)
 159                 get_device_state(trigger_data);
 160 -               rtnl_unlock();
 161 -       }
 162
 163         trigger_data->last_activity = 0;
 164
 165         set_baseline_state(trigger_data);
 166         mutex_unlock(&trigger_data->lock);
 167 +       rtnl_unlock();
 168
 169         return 0;
 170  }