Re: [PATCH 3/5] drm: Add and export function drm_gem_cma_mmap_noncoherent

[Paul Cercueil]

Hi Christoph,

Le mar. 3 nov. 2020 à 18:50, Christoph Hellwig <hch@xxxxxxxxxxxxx> a 
écrit :
> On Mon, Nov 02, 2020 at 10:06:49PM +0000, Paul Cercueil wrote:
> >  This function can be used by drivers that need to mmap dumb buffers
> >  created with non-coherent backing memory.
> >
> >  Signed-off-by: Paul Cercueil <paul@xxxxxxxxxxxxxxx>
> >  ---
> >   drivers/gpu/drm/drm_gem_cma_helper.c | 39 
> > ++++++++++++++++++++++++++++
> >   include/drm/drm_gem_cma_helper.h     |  2 ++
> >   2 files changed, 41 insertions(+)
> >
> >  diff --git a/drivers/gpu/drm/drm_gem_cma_helper.c 
> > b/drivers/gpu/drm/drm_gem_cma_helper.c
> >  index 3bdd67795e20..4ed63f4896bd 100644
> >  --- a/drivers/gpu/drm/drm_gem_cma_helper.c
> >  +++ b/drivers/gpu/drm/drm_gem_cma_helper.c
> >  @@ -387,6 +387,45 @@ int drm_gem_cma_mmap(struct file *filp, struct 
> > vm_area_struct *vma)
> >   }
> >   EXPORT_SYMBOL_GPL(drm_gem_cma_mmap);
> >
> >  +/**
> >  + * drm_gem_cma_mmap_noncoherent - memory-map a CMA GEM object with
> >  + *     non-coherent cache attribute
> >  + * @filp: file object
> >  + * @vma: VMA for the area to be mapped
> >  + *
> >  + * Just like drm_gem_cma_mmap, but for a GEM object backed by 
> > non-coherent
> >  + * memory.
> >  + *
> >  + * Returns:
> >  + * 0 on success or a negative error code on failure.
> >  + */
> >  +int drm_gem_cma_mmap_noncoherent(struct file *filp, struct 
> > vm_area_struct *vma)
> >  +{
> >  +	struct drm_gem_cma_object *cma_obj;
> >  +	int ret;
> >  +
> >  +	ret = drm_gem_mmap(filp, vma);
> >  +	if (ret)
> >  +		return ret;
> >  +
> >  +	cma_obj = to_drm_gem_cma_obj(vma->vm_private_data);
> >  +
> >  +	/*
> >  +	 * Clear the VM_PFNMAP flag that was set by drm_gem_mmap(), and 
> > set the
> >  +	 * vm_pgoff (used as a fake buffer offset by DRM) to 0 as we want 
> > to map
> >  +	 * the whole buffer.
> >  +	 */
> >  +	vma->vm_flags &= ~VM_PFNMAP;
> >  +	vma->vm_pgoff = 0;
> >  +	vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
> >  +
> >  +	return remap_pfn_range(vma, vma->vm_start,
> >  +			       cma_obj->paddr >> PAGE_SHIFT,
> >  +			       vma->vm_end - vma->vm_start,
> >  +			       vma->vm_page_prot);
>
> Per patch 1 cma_obj->paddr is the dma address, while remap_pfn_range
> expects a physical address.  This does not work.

Ok, what would be the correct way to mmap_noncoherent?

-Paul