Re: [PATCH v2] rust: place generated init_module() function in .init.text

[Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>]

On Tue, Feb 6, 2024 at 3:50 PM Thomas Bertschinger
<tahbertschinger@gmail.com> wrote:
>
> I can send a v3 that fixes the wording of the comment; it's worth being
> correct.

(Perhaps you didn't mean this, but just to be clear...)

No, safety sections are not incorrect just because they are not as
relaxed as they could be. They aren't "normal" comments -- they don't
document what the function does or could do.

In many cases, relaxing the preconditions is what you want, of course,
because that makes the function easier to call (and ideally make them
safe); but in some cases you may not want that, e.g. because you don't
want callers to start to rely on something you may change later even
if you happen to support it now.

Because of this, the absence of a safety section can actually be what
makes it incorrect, unlike "normal" comments where removing them is a
easy way to avoid mistakes in the comment :)

Now, we may nevertheless want to relax the preconditions here or
document (somewhere that is not the safety section) what the function
could offer, but my point is that it is perfectly OK to have a
function with restricted preconditions.

And this is one of those cases where it can make sense: unless we
expect to have a use case in-tree for this, I don't see the need to
let others call this.

Cheers,
Miguel