Re: [PATCH 1/2] xen/x86: io_apic: Introduce a command line option to skip timer check

[Jan Beulich <jbeulich@xxxxxxxx>]

On 15.09.2023 15:18, Julien Grall wrote:
> On 07/09/2023 15:09, Jan Beulich wrote:
>> On 18.08.2023 15:44, Julien Grall wrote:
>>> --- a/docs/misc/xen-command-line.pandoc
>>> +++ b/docs/misc/xen-command-line.pandoc
>>> @@ -1896,6 +1896,13 @@ This option is ignored in **pv-shim** mode.
>>>   ### nr_irqs (x86)
>>>   > `= <integer>`
>>>   
>>> +### no_timer_works (x86)
>>> +> `=<boolean>`
>>> +
>>> +> Default: `true`
>>> +
>>> +Disables the code which tests for broken timer IRQ sources.
>>
>> In description and code it's "check", but here it's "works". Likely
>> just a typo. But I'd prefer if we didn't introduce any new "no*"
>> options which then can be negated to "no-no*". Make it "timer-check"
>> (also avoiding the underscore, no matter that Linux uses it), or
>> alternatively make it a truly positive option, e.g. "timer-irq-works".
> 
> I don't mind too much about using - over _ but it is never clear why you 
> strongly push for it (and whether the others agrees).

Informal feedback suggested that various people agree and no-one strongly
disagrees to the argument of underscore really only being an auxiliary
separator character, when no better one can be used, and it also being
two keypresses to type on most keyboards, when dash is just one.

> Is this documented 
> somewhere? If not, can you do it so everyone can apply it consistently? 
> (At least I would not remember to ask for it because I am happy with the _).

As to documenting - it's not clear to me where such documentation ought
to go. In a way this is coding style, so it could be ./CODING_STYLE, but
then my experience with proposing changes there has been at best mixed.

Jan