On 14.11.2025 10:27, Oleksii Kurochko wrote:
> On 11/6/25 3:05 PM, Jan Beulich wrote:
>> On 20.10.2025 17:57, Oleksii Kurochko wrote:
>>> +void vmid_flush_hart(void)
>>> +{
>>> + struct vmid_data *data = &this_cpu(vmid_data);
>>> +
>>> + if ( !data->used )
>>> + return;
>>> +
>>> + if ( likely(++data->generation != 0) )
>>> + return;
>>> +
>>> + /*
>>> + * VMID generations are 64 bit. Overflow of generations never happens.
>>> + * For safety, we simply disable ASIDs, so correctness is established;
>>> it
>>> + * only runs a bit slower.
>>> + */
>>> + printk("%s: VMID generation overrun. Disabling VMIDs.\n", __func__);
>> Is logging of the function name of any value here?
>
> Agree, there is no any sense for the logging of the function name.
>
>> Also, despite the x86
>> original havinbg it like this - generally no full stops please if log
>> messages. "VMID generation overrun; disabling VMIDs\n" would do.
>
> Sure, I will drop it and will try to not add it in such cases. But could you
> please remind (if I asked that before) me what is the reason why full stop
> shouldn't be presented in such cases?
First: Consistency across the code base. Second: Meaningless characters
needlessly consume serial line bandwidth.
Jan
