Re: [BUG] dev_pm_opp refcount issue on Arm Juno r0

[Viresh Kumar]

On 20-12-18, 15:27, Valentin Schneider wrote:
> Hi,
> 
> While running some hotplug torture test [1] on my Juno r0 I came across
> the follow splat:
> 
> [  716.561862] ------------[ cut here ]------------
> [  716.566451] refcount_t: underflow; use-after-free.
> [  716.571240] WARNING: CPU: 2 PID: 18 at lib/refcount.c:280 refcount_dec_not_one+0x9c/0xc0
> [  716.579246] Modules linked in:
> [  716.582269] CPU: 2 PID: 18 Comm: cpuhp/2 Not tainted 4.20.0-rc7 #39
> [  716.588469] Hardware name: ARM Juno development board (r0) (DT)
> [  716.594326] pstate: 40000005 (nZcv daif -PAN -UAO)
> [  716.599065] pc : refcount_dec_not_one+0x9c/0xc0
> [  716.603546] lr : refcount_dec_not_one+0x9c/0xc0
> [  716.608024] sp : ffff00000a063c70
> [  716.611299] x29: ffff00000a063c70 x28: 0000000000000000 
> [  716.616555] x27: 0000000000000000 x26: 0000000000000002 
> [  716.621810] x25: ffff000009169000 x24: ffff000008f8e1b0 
> [  716.627065] x23: ffff000008ce0920 x22: 00000000ffffffff 
> [  716.632319] x21: ffff000009169000 x20: ffff8009762a2664 
> [  716.637574] x19: ffff000009294a90 x18: 0000000000000400 
> [  716.642828] x17: 0000000000000000 x16: 0000000000000000 
> [  716.648082] x15: 0000000000000000 x14: 0000000000000400 
> [  716.653336] x13: 000000000000023f x12: 0000000000043705 
> [  716.658590] x11: 0000000000000108 x10: 0000000000000960 
> [  716.663844] x9 : ffff00000a063970 x8 : ffff800976943ec0 
> [  716.669098] x7 : 0000000000000000 x6 : ffff80097ff720b8 
> [  716.674353] x5 : ffff80097ff720b8 x4 : 0000000000000000 
> [  716.679607] x3 : ffff80097ff78e68 x2 : ffff80097ff720b8 
> [  716.684861] x1 : 6374e2a7925c1100 x0 : 0000000000000000 
> [  716.690115] Call trace:
> [  716.692532]  refcount_dec_not_one+0x9c/0xc0
> [  716.696669]  refcount_dec_and_mutex_lock+0x18/0x70
> [  716.701409]  _put_opp_list_kref+0x28/0x50
> [  716.705373]  _dev_pm_opp_find_and_remove_table+0x24/0x88
> [  716.710628]  _dev_pm_opp_cpumask_remove_table+0x50/0xa0
> [  716.715796]  dev_pm_opp_cpumask_remove_table+0x10/0x18
> [  716.720879]  scpi_cpufreq_exit+0x40/0x50
> [  716.724758]  cpufreq_offline+0x108/0x1e0
> [  716.728637]  cpuhp_cpufreq_offline+0xc/0x18

This probably happened due to some of recent OPP core changes and I missed
updating this platform (I updated mvebu though). The problem is completely
different from what you logs show :)

Please try the below patch.

@Sudeep: Please help review it as well.

-- 
viresh

-------------------------8<-------------------------