[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <20241023192229.64333482554de0135397cb23@linux-foundation.org>
Date: Wed, 23 Oct 2024 19:22:29 -0700
From: Andrew Morton <akpm@...ux-foundation.org>
To: Matthew Brost <matthew.brost@...el.com>
Cc: <intel-xe@...ts.freedesktop.org>, <dri-devel@...ts.freedesktop.org>,
<linux-mm@...ck.org>, <linux-kernel@...r.kernel.org>,
<simona.vetter@...ll.ch>, <felix.kuehling@....com>, <apopple@...dia.com>
Subject: Re: [PATCH v3 1/1] mm/migrate: Add migrate_device_pfns
On Thu, 24 Oct 2024 01:50:34 +0000 Matthew Brost <matthew.brost@...el.com> wrote:
> > On Wed, 23 Oct 2024 16:39:44 -0700 Matthew Brost <matthew.brost@...el.com> wrote:
> >
> > > Implement migrate_device_pfns to prepare an array of PFNs for migration.
> > > Handles non-contiguous ranges of device pages that require migration.
> >
> > OK, that's "what". We're more interested in "why".
> >
>
> Sure can add. The 'why' is:
>
> A non-contiguous allocation of device pages can occur if a device is
> under memory pressure within a single driver allocation of device
> memory. Additionally, a driver allocation of memory can also be evicted
> under memory pressure. Therefore, an interface for migrating a set of
> non-contiguous device pages is required.
OK, thanks. But when merging a new interface such as this we like to
see the code which will actually use the interface. Along with reasons
to believe that the calling code will actually be merged, so we don't
end up with a new interface which has no callers.
Apart from that, I suspect that it makes more sense to merge this via
the DRM tree, alongside the code which uses it.
Powered by blists - more mailing lists