Re: [BUG v6.0-rc2] lockdep splat on ct_kernel_enter()

[Frederic Weisbecker]

Le Mon, Aug 22, 2022 at 04:44:04PM -0400, Steven Rostedt a écrit :
> My tests are failing because of this splat:
> 
> [   16.073659] ------------[ cut here ]------------
> [   16.074407] bus: 'platform': add driver acpi-ged
> [   16.074424] DEBUG_LOCKS_WARN_ON(lockdep_hardirqs_enabled())
> [   16.074424] WARNING: CPU: 0 PID: 0 at kernel/locking/lockdep.c:5506 check_flags+0x114/0x1d0
> [   16.083020] bus: 'acpi': add driver ac
> [   16.074424] Modules linked in:
> [   16.074424] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 6.0.0-rc2-test+ #18
> [   16.096856] bus: 'acpi': add driver button
> [   16.074424] Hardware name: MSI MS-7823/CSM-H87M-G43 (MS-7823), BIOS V1.6 02/22/2014
> [   16.074424] RIP: 0010:check_flags+0x114/0x1d0
> [   16.103642] bus: 'acpi': __driver_probe_device: matched device PNP0C0C:00 with driver button
> [   16.074424] Code: ff ff e8 ef 0a aa ff 85 c0 74 21 44 8b 0d bc 56 c6 00 45 85 c9 75 15 48 c7 c6 c7 e8 37 b9 48 c7 c7 a0 96 36 b9 e8 54 c9 f9 ff <0f> 0b 48 c7 c7 b0 fc 37 b9 e8 2e 20 fa ff eb 99 85 c0 0f 85 38 ff
> [   16.074424] RSP: 0000:ffffffffb9603d28 EFLAGS: 00010082
> [   16.114488] bus: 'acpi': really_probe: probing driver button with device PNP0C0C:00
> [   16.074424] 
> [   16.074424] RAX: 0000000000000000 RBX: ffffffffb96d6c20 RCX: 0000000000000000
> [   16.074424] RDX: 0000000000000000 RSI: ffffffffb81171da RDI: 0000000000000001
> [   16.126499] device: 'input0': device_add
> [   16.074424] RBP: ffffffffb9603d28 R08: 0000000000000000 R09: 0000000000000001
> [   16.074424] R10: 0000000000000000 R11: 0000000000000030 R12: ffffffffb961aa80
> [   16.074424] R13: ffffffffb961b3f0 R14: 0000000000000000 R15: ffffffffb9758878
> [   16.153763] PM: Adding info for No Bus:input0
> [   16.074424] FS:  0000000000000000(0000) GS:ffff961b55c00000(0000) knlGS:0000000000000000
> [   16.074424] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   16.166543] input: Power Button as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0C:00/input/input0
> [   16.074424] CR2: ffff961b5edff000 CR3: 00000000b1612001 CR4: 00000000001706f0
> [   16.074424] Call Trace:
> [   16.074424]  <TASK>
> [   16.175153] device: 'wakeup15': device_add
> [   16.074424]  lock_is_held_type+0x6f/0x130
> [   16.186284]  rcu_read_lock_sched_held+0x4a/0x90
> [   16.186284]  trace_rcu_dyntick+0x3a/0xe0
> [   16.186284]  ct_kernel_enter.constprop.0+0x66/0xa0
> [   16.186284]  ct_idle_exit+0xd/0x30
> [   16.186284]  cpuidle_enter_state+0x28a/0x310

I'm surprised because ct_cpuidle_exit() calls lockdep_hardirqs_off().
And in fact we shouldn't even need to disable IRQS from
ct_idle_exit() anymore because cpuidle_enter_state() already does it.

So I must be missing something...