Re: [PATCH] sparc/pci: Make pci_poke_lock a raw_spinlock_t.

[Waiman Long <llong@redhat.com>]

On 11/25/24 4:25 PM, Guenter Roeck wrote:
> On 11/25/24 12:54, Waiman Long wrote:
>>
>> On 11/25/24 3:23 PM, Guenter Roeck wrote:
>>> On 11/25/24 12:06, Guenter Roeck wrote:
>>>> On 11/25/24 11:33, Waiman Long wrote:
>>>> [ ... ]
>>>>>> Fixing that finally gives me a clean run. Nevertheless, that 
>>>>>> makes me wonder:
>>>>>> Should I just disable CONFIG_PROVE_RAW_LOCK_NESTING for sparc 
>>>>>> runtime tests ?
>>>>>
>>>>> If no one is tryng to ever enable PREEMPT_RT on SPARC, I suppose 
>>>>> you could disable CONFIG_PROVE_RAW_LOCK_NESTING to avoid the trouble.
>>>>>
>>>>
>>>> SGTM. I'll do that unless someone gives me a good reason to keep it 
>>>> enabled.
>>>>
>>>
>>> Actually it can not be disabled with a configuration flag. It is
>>> automatically enabled. I'll have to disable PROVE_LOCKING to disable 
>>> it.
>>>
>>> config PROVE_RAW_LOCK_NESTING
>>>         bool                    <---- no longer user configurable
>>>         depends on PROVE_LOCKING
>>>         default y
>>>         help
>>>          Enable the raw_spinlock vs. spinlock nesting checks which 
>>> ensure
>>>          that the lock nesting rules for PREEMPT_RT enabled kernels are
>>>          not violated.
>>>
>>> I don't really like that, and I don't understand the logic behind it,
>>> but it is what it is.
>>>
>>> FWIW, the description of commit 560af5dc839 is misleading. It says 
>>> "Enable
>>> PROVE_RAW_LOCK_NESTING _by default_" (emphasis mine). That is not 
>>> what the
>>> commit does. It force-enables PROVE_RAW_LOCK_NESTING if 
>>> PROVE_LOCKING is
>>> enabled. It is all or nothing.
>>>
>> I think we can relax it by
>>
>> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
>> index 5d9eca035d47..bfdbd3fa2d29 100644
>> --- a/lib/Kconfig.debug
>> +++ b/lib/Kconfig.debug
>> @@ -1399,7 +1399,7 @@ config PROVE_LOCKING
>>   config PROVE_RAW_LOCK_NESTING
>>          bool
>>          depends on PROVE_LOCKING
>> -       default y
>> +       default y if ARCH_SUPPORTS_RT
>>          help
>>           Enable the raw_spinlock vs. spinlock nesting checks which 
>> ensure
>>           that the lock nesting rules for PREEMPT_RT enabled kernels are
>>
>> Sebastian, what do you think?
>>
>
>     depends on PROVE_LOCKING && ARCH_SUPPORTS_RT
>
> seems to make more sense to me.

That will work too, but that will enforce that arches with no 
ARCH_SUPPORTS_RT will not be able to enable PROVE_RAW_LOCK_NESTING even 
if people want to try it out.

Cheers,
Longman