Re: [PATCH net-next 00/12] Fix rtnl_mutex deadlock with DPAA2 and SFP modules

[Ioana Ciornei]

On Tue, Nov 29, 2022 at 04:12:09PM +0200, Vladimir Oltean wrote:
> This patch set deliberately targets net-next and lacks Fixes: tags due
> to caution on my part.
> 
> While testing some SFP modules on the Solidrun Honeycomb LX2K platform,
> I noticed that rebooting causes a deadlock:
> 
> ============================================
> WARNING: possible recursive locking detected
> 6.1.0-rc5-07010-ga9b9500ffaac-dirty #656 Not tainted
> --------------------------------------------
> systemd-shutdow/1 is trying to acquire lock:
> ffffa62db6cf42f0 (rtnl_mutex){+.+.}-{4:4}, at: rtnl_lock+0x1c/0x30
> 
> but task is already holding lock:
> ffffa62db6cf42f0 (rtnl_mutex){+.+.}-{4:4}, at: rtnl_lock+0x1c/0x30
> 
> other info that might help us debug this:
>  Possible unsafe locking scenario:
> 
>        CPU0
>        ----
>   lock(rtnl_mutex);
>   lock(rtnl_mutex);
> 
>  *** DEADLOCK ***
> 
>  May be due to missing lock nesting notation
> 
> 6 locks held by systemd-shutdow/1:
>  #0: ffffa62db6863c70 (system_transition_mutex){+.+.}-{4:4}, at: __do_sys_reboot+0xd4/0x260
>  #1: ffff2f2b0176f100 (&dev->mutex){....}-{4:4}, at: device_shutdown+0xf4/0x260
>  #2: ffff2f2b017be900 (&dev->mutex){....}-{4:4}, at: device_shutdown+0x104/0x260
>  #3: ffff2f2b017680f0 (&dev->mutex){....}-{4:4}, at: device_release_driver_internal+0x40/0x260
>  #4: ffff2f2b0e1608f0 (&dev->mutex){....}-{4:4}, at: device_release_driver_internal+0x40/0x260
>  #5: ffffa62db6cf42f0 (rtnl_mutex){+.+.}-{4:4}, at: rtnl_lock+0x1c/0x30
> 
> stack backtrace:
> CPU: 6 PID: 1 Comm: systemd-shutdow Not tainted 6.1.0-rc5-07010-ga9b9500ffaac-dirty #656
> Hardware name: SolidRun LX2160A Honeycomb (DT)
> Call trace:
>  lock_acquire+0x68/0x84
>  __mutex_lock+0x98/0x460
>  mutex_lock_nested+0x2c/0x40
>  rtnl_lock+0x1c/0x30
>  sfp_bus_del_upstream+0x1c/0xac
>  phylink_destroy+0x1c/0x50
>  dpaa2_mac_disconnect+0x28/0x70
>  dpaa2_eth_remove+0x1dc/0x1f0
>  fsl_mc_driver_remove+0x24/0x60
>  device_remove+0x70/0x80
>  device_release_driver_internal+0x1f0/0x260
>  device_links_unbind_consumers+0xe0/0x110
>  device_release_driver_internal+0x138/0x260
>  device_release_driver+0x18/0x24
>  bus_remove_device+0x12c/0x13c
>  device_del+0x16c/0x424
>  fsl_mc_device_remove+0x28/0x40
>  __fsl_mc_device_remove+0x10/0x20
>  device_for_each_child+0x5c/0xac
>  dprc_remove+0x94/0xb4
>  fsl_mc_driver_remove+0x24/0x60
>  device_remove+0x70/0x80
>  device_release_driver_internal+0x1f0/0x260
>  device_release_driver+0x18/0x24
>  bus_remove_device+0x12c/0x13c
>  device_del+0x16c/0x424
>  fsl_mc_bus_remove+0x8c/0x10c
>  fsl_mc_bus_shutdown+0x10/0x20
>  platform_shutdown+0x24/0x3c
>  device_shutdown+0x15c/0x260
>  kernel_restart+0x40/0xa4
>  __do_sys_reboot+0x1e4/0x260
>  __arm64_sys_reboot+0x24/0x30
> 
> But fixing this appears to be not so simple. The patch set represents my
> attempt to address it.
> 
> In short, the problem is that dpaa2_mac_connect() and dpaa2_mac_disconnect()
> call 2 phylink functions in a row, one takes rtnl_lock() itself -
> phylink_create(), and one which requires rtnl_lock() to be held by the
> caller - phylink_fwnode_phy_connect(). The existing approach in the
> drivers is too simple. We take rtnl_lock() when calling dpaa2_mac_connect(),
> which is what results in the deadlock.
> 
> Fixing just that creates another problem. The drivers make use of
> rtnl_lock() for serializing with other code paths too. I think I've
> found all those code paths, and established other mechanisms for
> serializing with them.
> 
> Vladimir Oltean (12):
>   net: dpaa2-eth: don't use -ENOTSUPP error code
>   net: dpaa2: replace dpaa2_mac_is_type_fixed() with
>     dpaa2_mac_is_type_phy()
>   net: dpaa2-mac: absorb phylink_start() call into dpaa2_mac_start()
>   net: dpaa2-mac: remove defensive check in dpaa2_mac_disconnect()
>   net: dpaa2-eth: assign priv->mac after dpaa2_mac_connect() call
>   net: dpaa2-switch: assign port_priv->mac after dpaa2_mac_connect()
>     call
>   net: dpaa2: publish MAC stringset to ethtool -S even if MAC is missing
>   net: dpaa2-switch replace direct MAC access with
>     dpaa2_switch_port_has_mac()
>   net: dpaa2-eth: connect to MAC before requesting the "endpoint
>     changed" IRQ
>   net: dpaa2-eth: serialize changes to priv->mac with a mutex
>   net: dpaa2-switch: serialize changes to priv->mac with a mutex
>   net: dpaa2-mac: move rtnl_lock() only around
>     phylink_{,dis}connect_phy()
> 
>  .../freescale/dpaa2/mac-phy-support.rst       |  9 +-
>  .../net/ethernet/freescale/dpaa2/dpaa2-eth.c  | 87 ++++++++++++-------
>  .../net/ethernet/freescale/dpaa2/dpaa2-eth.h  | 11 +--
>  .../ethernet/freescale/dpaa2/dpaa2-ethtool.c  | 70 ++++++++++-----
>  .../net/ethernet/freescale/dpaa2/dpaa2-mac.c  | 16 +++-
>  .../net/ethernet/freescale/dpaa2/dpaa2-mac.h  | 10 ++-
>  .../freescale/dpaa2/dpaa2-switch-ethtool.c    | 45 ++++++----
>  .../ethernet/freescale/dpaa2/dpaa2-switch.c   | 59 +++++++++----
>  .../ethernet/freescale/dpaa2/dpaa2-switch.h   |  9 +-
>  9 files changed, 212 insertions(+), 104 deletions(-)

For the entire patch set:

Reviewed-by: Ioana Ciornei <ioana.ciornei@xxxxxxx>
Tested-by: Ioana Ciornei <ioana.ciornei@xxxxxxx>