Re: [PATCH v2] mm: Introduce new function vm_insert_kmem_page

[Souptick Joarder <jrdr.linux@gmail.com>]

On Fri, Oct 5, 2018 at 11:39 PM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> On Fri, Oct 5, 2018 at 2:11 PM Souptick Joarder <jrdr.linux@gmail.com> wrote:
> >
> > On Fri, Oct 5, 2018 at 4:19 PM Miguel Ojeda
> > <miguel.ojeda.sandonis@gmail.com> wrote:
> > >
> > >   1. Introduce the vmf_* API
> > >   2. Change all PF-users users to that (leaving all non-PF ones
> > > untouched!) -- if this is too big, you can split this patch into
> > > several patches, one per subsystem, etc.
> >
> > We are done with step 2. All the PF-users are converted to use
> > vmf_insert_page. ( Ref - linux-next-20181005)
>
> They are not supposed to be "steps". You did it with 70+ commits (!!)
> over the course of several months. Why a tree wasn't created, stuff
> developed there, and when done, submitted it for review?

Because we already have a plan for entire vm_fault_t migration and
the * instruction * was to send one patch per driver.
>
> > >
> > > Otherwise, if you want to pursue Matthew's idea:
> > >
> > >   4. Introduce the vm_insert_range (possibly leveraging
> > > vm_insert_page, or not; you have to see what is best).
> > >   5. Replace those callers that can take advantage of vm_insert_range
> > >   6. Remove vm_insert_page and replace callers with vm_insert_range
> > > (only if it is not worth to keep vm_insert_range, again justifying it
> > > *on its own merits*)
> >
> > Step 4 to 6, going to do it.  It is part of plan now :-)
> >
>
> Fine, but you haven't answered to the other parts of my email: you
> don't explain why you choose one alternative over the others, you
> simply keep changing the approach.

We are going in circles here. That you want to convert vm_insert_page
to vmf_insert_page for the PF case is fine and understood. However,
you don't *need* to introduce a new name for the remaining non-PF
cases if the function is going to be the exact same thing as before.
You say "The final goal is to remove vm_insert_page", but you haven't
justified *why* you need to remove that name.

I think I have given that answer. If we don't remove vm_insert_page,
future #PF caller will have option to use it. But those should be
restricted. How are we going to restrict vm_insert_page in one half
of kernel when other half is still using it  ?? Is there any way ? ( I don't
know)