Re: [PATCH 5/5] x86: don't build unused entry code when !PV32

[Jan Beulich <jbeulich@xxxxxxxx>]

On 04.01.2021 16:53, Roger Pau Monné wrote:
> On Mon, Jan 04, 2021 at 02:56:12PM +0100, Jan Beulich wrote:
>> On 28.12.2020 16:30, Roger Pau Monné wrote:
>>> I would like to have Andrew's opinion on this one (as you and him tend
>>> to modify more asm code than myself). There are quite a lot of
>>> addition to the assembly code, and IMO it makes the code more complex
>>> which I think we should try to avoid, as assembly is already hard
>>> enough.
>>
>> Well, while I can see your point (and I indeed asked myself the same
>> question when making this change), this merely follows the route
>> started with the addition on CONFIG_PV conditionals. If we think that
>> prior step didn't set a good precedent, we ought to undo it.
>> Otherwise I see no good argument against doing the same kind of
>> transformation a 2nd time (and further ones, if need be down the
>> road).
> 
> I think we need to apply some common sense and reach consensus about
> where it's fine to make code conditional at build time as to not make
> the existing code much harder to read and reason about. This is mostly
> a subjective decision, so I understand your concern.
> 
> I still think I would like Andrew opinion on this one, as said you and
> him are the ones mostly doing assembly coding. I find it already hard
> to follow myself without the conditionals.

Oh, sure - my prior response in no way meant to be an objection to
your request for Andrew's take on this.

Jan