Re: [PATCH] drm/etnaviv: switch to PFN mappings

From: Daniel Vetter <daniel@ffwll.ch>
To: Lucas Stach <l.stach@pengutronix.de>
Cc: kernel@pengutronix.de, patchwork-lst@pengutronix.de,
	etnaviv@lists.freedesktop.org, dri-devel@lists.freedesktop.org,
	Russell King <linux+etnaviv@armlinux.org.uk>
Subject: Re: [PATCH] drm/etnaviv: switch to PFN mappings
Date: Wed, 16 Nov 2022 10:23:39 +0100	[thread overview]
Message-ID: <Y3SsG0eW1HFZIOb/@phenom.ffwll.local> (raw)
In-Reply-To: <a105ba6089f77f1c3b8310b06815cb733eb7841c.camel@pengutronix.de>

On Wed, Nov 16, 2022 at 10:17:02AM +0100, Lucas Stach wrote:
> Am Freitag, dem 09.09.2022 um 11:20 +0200 schrieb Lucas Stach:
> > There is no reason to use page based mappings, as the established
> > mappings are special driver mappings anyways and should not be
> > handled like normal pages.
> > 
> > Be consistent with what other drivers do and use raw PFN based
> > mappings.
> 
> Anyone up for reviewing this? Basically it moves etnaviv closer to what
> all the other DRM drivers are doing.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Note that since you've had VM_IO before it was already VM_SPECIAL, so
outside of etnaviv (for stuff like get_user_pages/userptr) there's not
really a difference. But it's nice to be a tad more consistent, since page
mapping + VM_IO really doesn't make that much sense :-)
-Daniel

> 
> Regards,
> Lucas
> 
> > 
> > Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
> > ---
> >  drivers/gpu/drm/etnaviv/etnaviv_gem.c | 11 ++++++-----
> >  1 file changed, 6 insertions(+), 5 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem.c b/drivers/gpu/drm/etnaviv/etnaviv_gem.c
> > index d45bf0368339..68e4446a94ad 100644
> > --- a/drivers/gpu/drm/etnaviv/etnaviv_gem.c
> > +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem.c
> > @@ -130,7 +130,7 @@ static int etnaviv_gem_mmap_obj(struct etnaviv_gem_object *etnaviv_obj,
> >  {
> >  	pgprot_t vm_page_prot;
> >  
> > -	vma->vm_flags |= VM_IO | VM_MIXEDMAP | VM_DONTEXPAND | VM_DONTDUMP;
> > +	vma->vm_flags |= VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP;
> >  
> >  	vm_page_prot = vm_get_page_prot(vma->vm_flags);
> >  
> > @@ -165,7 +165,8 @@ static vm_fault_t etnaviv_gem_fault(struct vm_fault *vmf)
> >  	struct vm_area_struct *vma = vmf->vma;
> >  	struct drm_gem_object *obj = vma->vm_private_data;
> >  	struct etnaviv_gem_object *etnaviv_obj = to_etnaviv_bo(obj);
> > -	struct page **pages, *page;
> > +	struct page **pages;
> > +	unsigned long pfn;
> >  	pgoff_t pgoff;
> >  	int err;
> >  
> > @@ -189,12 +190,12 @@ static vm_fault_t etnaviv_gem_fault(struct vm_fault *vmf)
> >  	/* We don't use vmf->pgoff since that has the fake offset: */
> >  	pgoff = (vmf->address - vma->vm_start) >> PAGE_SHIFT;
> >  
> > -	page = pages[pgoff];
> > +	pfn = page_to_pfn(pages[pgoff]);
> >  
> >  	VERB("Inserting %p pfn %lx, pa %lx", (void *)vmf->address,
> > -	     page_to_pfn(page), page_to_pfn(page) << PAGE_SHIFT);
> > +	     pfn, pfn << PAGE_SHIFT);
> >  
> > -	return vmf_insert_page(vma, vmf->address, page);
> > +	return vmf_insert_pfn(vma, vmf->address, pfn);
> >  }
> >  
> >  int etnaviv_gem_mmap_offset(struct drm_gem_object *obj, u64 *offset)
> 
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch