Re: [PATCH] x86/IRQ: bump max number of guests for a shared IRQ to 31

[Jan Beulich <jbeulich@xxxxxxxx>]

On 02.12.2020 17:34, Igor Druzhinin wrote:
> On 02/12/2020 15:21, Jan Beulich wrote:
>> On 02.12.2020 15:53, Igor Druzhinin wrote:
>>> On 02/12/2020 09:25, Jan Beulich wrote:
>>>> Instead I'm wondering whether this wouldn't better be a Kconfig
>>>> setting (or even command line controllable). There don't look to be
>>>> any restrictions on the precise value chosen (i.e. 2**n-1 like is
>>>> the case for old and new values here, for whatever reason), so a
>>>> simple permitted range of like 4...64 would seem fine to specify.
>>>> Whether the default then would want to be 8 (close to the current
>>>> 7) or higher (around the actually observed maximum) is a different
>>>> question.
>>>
>>> I'm in favor of a command line argument here - it would be much less trouble
>>> if a higher limit was suddenly necessary in the field. The default IMO
>>> should definitely be higher than 8 - I'd stick with number 32 which to me
>>> should cover our real world scenarios and apply some headroom for the 
>>> future.
>>
>> Well, I'm concerned of the extra memory overhead. Every IRQ,
>> sharable or not, will get the extra slots allocated with the
>> current scheme. Perhaps a prereq change then would be to only
>> allocate multi-guest arrays for sharable IRQs, effectively
>> shrinking the overhead in particular for all MSI ones?
> 
> That's one way to improve overall system scalability but in that area
> there is certainly much bigger fish to fry elsewhere. With 32 elements in the
> array we get 200 bytes of overhead per structure, with 16 it's just 72 extra
> bytes which in the unattainable worst case scenario of every single vector 
> taken
> in 512 CPU machine would only account for several MB of overhead.

I'm generally unhappy with this way of thinking, as this is what has
been leading to unnecessary growth of all sorts of software and its
needs of resources. Yes, there surely are larger gains to be had
elsewhere, but that's imo still no excuse to grow memory allocations
"blindly" despite it being clear that in a fair share of cases a
fair part of the allocated memory won't be used. This said, ...

> I'd start with dynamic array allocation first and setting the limit to 16 that
> should be enough for now. And then if that default value needs to be raised
> we can consider further improvements.

... I'm puzzled by this plan of yours, because unless I'm
misunderstanding dynamic array allocation is what I've been asking
for, effectively. Now that we have xmalloc_flex_struct(), this
should even be relatively straightforward, i.e. in particular with
no need to open code complex expressions.

Jan