On 8/29/19 4:27 PM, Marcelo Tosatti wrote:
> On Thu, Aug 29, 2019 at 04:10:27PM +0100, Joao Martins wrote:
>> When cpus != maxcpus cpuidle-haltpoll will fail to register all vcpus
>> past the online ones and thus fail to register the idle driver.
>> This is because cpuidle_add_sysfs() will return with -ENODEV as a
>> consequence from get_cpu_device() return no device for a non-existing
>> CPU.
>>
>> Instead switch to cpuidle_register_driver() and manually register each
>> of the present cpus through cpuhp_setup_state() callback and future
>> ones that get onlined. This mimmics similar logic that intel_idle does.
>>
>> Fixes: fa86ee90eb11 ("add cpuidle-haltpoll driver")
>> Signed-off-by: Joao Martins <joao.m.martins@xxxxxxxxxx>
>> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>
>> ---
>> v2:
>> * move cpus_read_unlock() right after unregistering all cpuidle_devices;
>> (Marcello Tosatti)
>> * redundant usage of cpuidle_unregister() when only
>> cpuidle_unregister_driver() suffices; (Marcelo Tosatti)
>> * cpuhp_setup_state() returns a state (> 0) on success with CPUHP_AP_ONLINE_DYN
>> thus we set @ret to 0
[ ... ]
>
> Reviewed-by: Marcelo Tosatti <mtosatti@xxxxxxxxxx>
>
Thanks!
Meanwhile upon re-reading cpuhp_setup_state() I found out the teardown/offlining
and haltpoll_uninit() could be a bit simplified. So I sent out a new version[0],
but didn't add your Rb because there's was some very slight functional changes.
[0] https://lore.kernel.org/kvm/20190902104031.9296-1-joao.m.martins@xxxxxxxxxx/
Joao
