Re: [PATCH] remoteproc: Use unbounded/high priority workqueue for recovery work

[Mukesh Ojha <quic_mojha@quicinc.com>]

On 3/12/2022 2:31 AM, Bjorn Andersson wrote:
> On Wed 19 Jan 13:30 CST 2022, Mukesh Ojha wrote:
>
>> There could be a scenario where there is too much load(n number
>> of tasks which is affined) on a core on which rproc recovery
>> is queued. Due to which, it takes number of seconds to complete
>> the recovery.
>>
>> If we make this queue unbounded and move it to high priority worker
>> pool then this work can be attempted to finished in less time.
> I unfortunately find this reasoning for adding WQ_HIGHPRI rather
> speculative. Please describe a concrete case that warrants the new
> work queue to be high priority.
>
> What is "number of seconds", what is "less time" and why is it more
> important to recover some remote processor than whatever else the system
> is busy doing?

Meanwhile, I will try to check if making it unbound only helps us in our 
low latency use cases.
So, does it make sense to make it  Unbound | freezable ?

-Mukesh

> Thanks,
> Bjorn
>
>> Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
>> ---
>>   drivers/remoteproc/remoteproc_core.c | 14 ++++++++++++--
>>   1 file changed, 12 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/remoteproc/remoteproc_core.c b/drivers/remoteproc/remoteproc_core.c
>> index 69f51ac..efb6316 100644
>> --- a/drivers/remoteproc/remoteproc_core.c
>> +++ b/drivers/remoteproc/remoteproc_core.c
>> @@ -59,6 +59,7 @@ static int rproc_release_carveout(struct rproc *rproc,
>>   
>>   /* Unique indices for remoteproc devices */
>>   static DEFINE_IDA(rproc_dev_index);
>> +static struct workqueue_struct *rproc_recovery_wq;
>>   
>>   static const char * const rproc_crash_names[] = {
>>   	[RPROC_MMUFAULT]	= "mmufault",
>> @@ -2752,8 +2753,10 @@ void rproc_report_crash(struct rproc *rproc, enum rproc_crash_type type)
>>   	dev_err(&rproc->dev, "crash detected in %s: type %s\n",
>>   		rproc->name, rproc_crash_to_string(type));
>>   
>> -	/* Have a worker handle the error; ensure system is not suspended */
>> -	queue_work(system_freezable_wq, &rproc->crash_handler);
>> +	if (rproc_recovery_wq)
>> +		queue_work(rproc_recovery_wq, &rproc->crash_handler);
>> +	else
>> +		queue_work(system_freezable_wq, &rproc->crash_handler);
>>   }
>>   EXPORT_SYMBOL(rproc_report_crash);
>>   
>> @@ -2802,6 +2805,11 @@ static void __exit rproc_exit_panic(void)
>>   
>>   static int __init remoteproc_init(void)
>>   {
>> +	rproc_recovery_wq = alloc_workqueue("rproc_recovery_wq", WQ_UNBOUND |
>> +				WQ_HIGHPRI | WQ_FREEZABLE, 0);
>> +	if (!rproc_recovery_wq)
>> +		pr_err("remoteproc: creation of rproc_recovery_wq failed\n");
>> +
>>   	rproc_init_sysfs();
>>   	rproc_init_debugfs();
>>   	rproc_init_cdev();
>> @@ -2818,6 +2826,8 @@ static void __exit remoteproc_exit(void)
>>   	rproc_exit_panic();
>>   	rproc_exit_debugfs();
>>   	rproc_exit_sysfs();
>> +	if (rproc_recovery_wq)
>> +		destroy_workqueue(rproc_recovery_wq);
>>   }
>>   module_exit(remoteproc_exit);
>>   
>> -- 
>> 2.7.4
>>