Re: [PATCH v3 1/2] x86/IRQ: make max number of guests for a shared IRQ configurable

[Jan Beulich <jbeulich@xxxxxxxx>]

On 07.12.2020 12:28, Igor Druzhinin wrote:
> On 07/12/2020 09:43, Jan Beulich wrote:
>> On 06.12.2020 18:43, Igor Druzhinin wrote:
>>> @@ -1633,11 +1640,12 @@ int pirq_guest_bind(struct vcpu *v, struct pirq 
>>> *pirq, int will_share)
>>>          goto retry;
>>>      }
>>>  
>>> -    if ( action->nr_guests == IRQ_MAX_GUESTS )
>>> +    if ( action->nr_guests == irq_max_guests )
>>>      {
>>> -        printk(XENLOG_G_INFO "Cannot bind IRQ%d to dom%d. "
>>> -               "Already at max share.\n",
>>> -               pirq->pirq, v->domain->domain_id);
>>> +        printk(XENLOG_G_INFO
>>> +               "Cannot bind IRQ%d to dom%pd: already at max share %u ",
> 
> I noticed it just now but could you also remove stray "dom" left in this line 
> while commiting.

Oh, sure.

>>> +               pirq->pirq, v->domain, irq_max_guests);
>>> +        printk("(increase with irq-max-guests= option)\n");
>>
>> Now two separate printk()s are definitely worse. Then putting the
>> part of the format string inside the parentheses on a separate line
>> would still be better (and perhaps a sensible compromise with the
>> grep-ability desire).
> 
> Now I'm confused because you asked me not to split the format string between 
> the lines which
> wouldn't be possible without splitting printk's. I didn't really want to drop 
> anything
> informative.

"Not splitting" really was meant in the sense of the words: No
splitting at all. Even less so across multiple printk()-s. But
since the line would get really long, I can live with the
outlined compromise.

Jan