Hi,
> > + vma->vm_flags |= VM_IO | VM_PFNMAP | VM_DONTEXPAND |
> > VM_DONTDUMP;
> > + vma->vm_page_prot =
> > pgprot_writecombine(vm_get_page_prot(vma->vm_flags));
> > + vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);
> > + }
>
> Totall
On Fri, Sep 6, 2019 at 2:13 PM Gerd Hoffmann wrote:
>
> Hi,
>
> > I think if we do an mmap callback, it should replace all the mmap handling
> > (except the drm_gem_object_get) that drm_gem_mmap_obj does. So maybe
> > something like the below:
>
> [ snip ]
>
> > Since I remember quite a few disc
Hi,
> I think if we do an mmap callback, it should replace all the mmap handling
> (except the drm_gem_object_get) that drm_gem_mmap_obj does. So maybe
> something like the below:
[ snip ]
> Since I remember quite a few discussions where the default vma flag
> wrangling we're doing is seriousl
On Thu, Aug 08, 2019 at 03:44:05PM +0200, Gerd Hoffmann wrote:
> drm_gem_object_funcs->vm_ops alone can't handle
> everything mmap() needs. Add a new callback for it.
>
> Signed-off-by: Gerd Hoffmann
> ---
> include/drm/drm_gem.h | 9 +
> drivers/gpu/drm/drm_gem.c | 6 ++
> 2 fi
drm_gem_object_funcs->vm_ops alone can't handle
everything mmap() needs. Add a new callback for it.
Signed-off-by: Gerd Hoffmann
---
include/drm/drm_gem.h | 9 +
drivers/gpu/drm/drm_gem.c | 6 ++
2 files changed, 15 insertions(+)
diff --git a/include/drm/drm_gem.h b/include/drm
