Re: [edk2-devel] [PATCH v8 00/37] Enable LoongArch virtual machine in edk2

["Laszlo Ersek" <lersek@redhat.com>]

On 2/2/24 04:36, Ni, Ray wrote:
> Chao,
> 
> When I review the detailed changes, I found a generic issue:
> 
>   Some changes to INF files are not necessary. They are just reordering
> the source files or libs within the same section.

I'm myself a bit torn about this. I generally like to keep entries in
[LibraryClasses] etc lexicographically sorted. However, it indeed
becomes confusing when a lib class is added or removed *together* with
sorting the entire list afresh. I think such changes are worth splitting
in to: first, sort the entries (a no-op patch), then implement the change.

> 
>   I hope that you could avoid these unnecessary changes.

I wouldn't necessarily call them unnecessary. I think sorting / cleanups
are worthwhile, but they should be well separated from the other types
of changes.

>   Another point is please make sure do not add unnecessary lib
> dependencies for x86 (e.g.: SafeIntLib). That might cause
>   some platform build breaks due to not including the SafeIntLib in DSC.

Can you clarify?

SafeIntLib is generally a good idea to *use* in C code, and when one
uses it, it must be listed in [LibraryClasses]. I agree that, when we
extend an INF file with a new dependency, we should grep edk2 (at least)
for platform DSCs that refer to that INF, and see if we have to resolve
the lib class.

Of course, when the C code doesn't actually consume SafeIntLib APIs,
then listing SafeIntLib in the INF file is just a bug.

Laszlo



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#115048): https://edk2.groups.io/g/devel/message/115048
Mute This Topic: https://groups.io/mt/103971622/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-