Re: [PATCH] mm-move-mremap-from-file_operations-to-vm_operations_struct-fix

[Oleg Nesterov <oleg@redhat.com>]

On 07/20, Benjamin LaHaise wrote:
>
> On Mon, Jul 20, 2015 at 07:33:11PM +0200, Oleg Nesterov wrote:
> >
> > And if we accept the fact this memory is locked and if we properly account
> > it, then may be we can just kill aio_migratepage(), aio_private_file(), and
> > change aio_setup_ring() to simply use install_special_mapping(). This will
> > greatly simplify the code. But let me remind that I know nothing about aio,
> > so please don't take my thoughts seriously.
>
> No, you can't get rid of that code.  The page migration is required when
> CPUs/memory is offlined and data needs to be moved to another node.

Of course, if we remove aio_migratepage() then aio can't be moved,

> Similarly, support for mremap() is also required for container migration /
> restoration.

This is not the problem. And one of the reasons to move ->mremap() into
vm_operations_struct was that install_special_mapping() can use it.

> Given how small the amount of memory aio can pin

I agree, but why should we worry about migration then? let this memory be
unmovable, don't use GFP_RECLAIMABLE/MOVABLE, etc.

But again, again, please ignore. This all is off-topic and my understanding
is very limited.

> it is unlikely that the accounting of
> how much aio has pinned will make any real difference in the big picture.

Agreed, but this can help to remove the system-wide aio-max-nr. Again,
unpriviliged user can steal aio.

Oleg.