[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250930182920.5604ca49@fedora>
Date: Tue, 30 Sep 2025 18:29:20 +0200
From: Boris Brezillon <boris.brezillon@...labora.com>
To: Loïc Molinari <loic.molinari@...labora.com>
Cc: Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>, Maxime Ripard
<mripard@...nel.org>, Thomas Zimmermann <tzimmermann@...e.de>, David Airlie
<airlied@...il.com>, Simona Vetter <simona@...ll.ch>, Jani Nikula
<jani.nikula@...ux.intel.com>, Joonas Lahtinen
<joonas.lahtinen@...ux.intel.com>, Rodrigo Vivi <rodrigo.vivi@...el.com>,
Tvrtko Ursulin <tursulin@...ulin.net>, Rob Herring <robh@...nel.org>,
Steven Price <steven.price@....com>, Liviu Dudau <liviu.dudau@....com>,
Melissa Wen <mwen@...lia.com>, Maíra Canal
<mcanal@...lia.com>, Hugh Dickins <hughd@...gle.com>, Baolin Wang
<baolin.wang@...ux.alibaba.com>, Andrew Morton <akpm@...ux-foundation.org>,
Al Viro <viro@...iv.linux.org.uk>, Mikołaj Wasiak
<mikolaj.wasiak@...el.com>, Christian Brauner <brauner@...nel.org>, Nitin
Gote <nitin.r.gote@...el.com>, Andi Shyti <andi.shyti@...ux.intel.com>,
linux-kernel@...r.kernel.org, dri-devel@...ts.freedesktop.org,
intel-gfx@...ts.freedesktop.org, linux-mm@...ck.org, kernel@...labora.com
Subject: Re: [PATCH 2/8] drm/gem: Introduce drm_gem_get_unmapped_area() fop
On Tue, 30 Sep 2025 18:09:37 +0200
Loïc Molinari <loic.molinari@...labora.com> wrote:
> On 30/09/2025 12:30, Boris Brezillon wrote:
> > On Mon, 29 Sep 2025 22:03:10 +0200
> >
> > Loïc Molinari <loic.molinari@...labora.com> wrote:
> >> +unsigned long drm_gem_get_unmapped_area(struct file *filp, unsigned long uaddr,
> >> + unsigned long len, unsigned long pgoff,
> >> + unsigned long flags)
> >> +{
> >> +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >> + struct drm_gem_object *obj;
> >> + unsigned long ret;
> >> +
> >> + obj = drm_gem_object_lookup_from_offset(filp, pgoff, len >> PAGE_SHIFT);
> >> + if (IS_ERR(obj))
> >> + return mm_get_unmapped_area(current->mm, filp, uaddr, len, 0,
> >> + flags);
> >> +
> >> + ret = shmem_get_unmapped_area(obj->filp, uaddr, len, 0, flags);
> >> +
> >> + drm_gem_object_put(obj);
> >> +
> >> + return ret;
> >> +#else
> >> + return mm_get_unmapped_area(current->mm, filp, uaddr, len, 0, flags);
> >
> > Looks like the above code covers the non-THP case too, do we really need
> > to specialize for !CONFIG_TRANSPARENT_HUGEPAGE here?
>
> It does cover the !CONFIG_TRANSPARENT_HUGEPAGE case
> (shmem_get_unmapped_area() would just call and return the
> mm_get_unmapped_area() address) but the idea here is to avoid the GEM
> object lookup cost by calling mm_get_unmapped_area() directly.
I'd expect the extra GEM lookup to be negligible compared to the overall
mmap() operation to be honest, but I guess if we really want to avoid
the overhead, we could still write it without this ifdef.
if (!IS_ENABLED(TRANSPARENT_HUGEPAGE))
return mm_get_unmapped_area(current->mm, filp, uaddr,
len, 0, flags);
...
My main concern is that shmem_get_unmapped_area() evolves with more
!TRANSPARENT_HUGEPAGE cases, and by calling mm_get_unmapped_area()
directly, we miss the opportunity to get optimizations for these cases,
just like we missed them by not forwarding the ->get_unmapped_area()
requests to the shmem layer so far.
>
> >> +#endif
> >> +}
> >> +EXPORT_SYMBOL(drm_gem_get_unmapped_area);
>
> Loïc
Powered by blists - more mailing lists