linux-kernel - Re: [PATCH 1/8] drm: add mmap() to drm_gem_object

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20190917132701.GW3958@phenom.ffwll.local>
Date:   Tue, 17 Sep 2019 15:27:01 +0200
From:   Daniel Vetter <daniel@...ll.ch>
To:     Gerd Hoffmann <kraxel@...hat.com>
Cc:     dri-devel@...ts.freedesktop.org,
        Thomas Zimmermann <tzimmermann@...e.de>,
        Daniel Vetter <daniel.vetter@...ll.ch>,
        Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
        Maxime Ripard <maxime.ripard@...tlin.com>,
        Sean Paul <sean@...rly.run>, David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel@...ll.ch>,
        open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/8] drm: add mmap() to drm_gem_object_funcs

On Fri, Sep 13, 2019 at 02:29:01PM +0200, Gerd Hoffmann wrote:
> drm_gem_object_funcs->vm_ops alone can't handle everything which needs
> to be done for mmap(), tweaking vm_flags for example.  So add a new
> mmap() callback to drm_gem_object_funcs where this code can go to.
> 
> Note that the vm_ops field is not used in case the mmap callback is
> presnt, it is expected that the callback sets vma->vm_ops instead.
> 
> drm_gem_mmap_obj() will use the new callback for object specific mmap
> setup.  With this in place the need for driver-speific fops->mmap
> callbacks goes away, drm_gem_mmap can be hooked instead.
> 
> drm_gem_prime_mmap() will use the new callback too to just mmap gem
> objects directly instead of jumping though loops to make
> drm_gem_object_lookup() and fops->mmap work.
> 
> Signed-off-by: Gerd Hoffmann <kraxel@...hat.com>
> ---
>  include/drm/drm_gem.h       | 14 ++++++++++++++
>  drivers/gpu/drm/drm_gem.c   | 27 ++++++++++++++++++---------
>  drivers/gpu/drm/drm_prime.c |  9 +++++++++
>  3 files changed, 41 insertions(+), 9 deletions(-)
> 
> diff --git a/include/drm/drm_gem.h b/include/drm/drm_gem.h
> index 6aaba14f5972..e71f75a2ab57 100644
> --- a/include/drm/drm_gem.h
> +++ b/include/drm/drm_gem.h
> @@ -150,6 +150,20 @@ struct drm_gem_object_funcs {
>  	 */
>  	void (*vunmap)(struct drm_gem_object *obj, void *vaddr);
>  
> +	/**
> +	 * @mmap:
> +	 *
> +	 * Handle mmap() of the gem object, setup vma accordingly.
> +	 *
> +	 * This callback is optional.
> +	 *
> +	 * The callback is used by by both drm_gem_mmap_obj() and
> +	 * drm_gem_prime_mmap().  When @mmap is present @vm_ops is not
> +	 * used, the @mmap callback must set vma->vm_ops instead.
> +	 *
> +	 */
> +	int (*mmap)(struct drm_gem_object *obj, struct vm_area_struct *vma);
> +
>  	/**
>  	 * @vm_ops:
>  	 *
> diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c
> index 6854f5867d51..56f42e0f2584 100644
> --- a/drivers/gpu/drm/drm_gem.c
> +++ b/drivers/gpu/drm/drm_gem.c
> @@ -1099,22 +1099,31 @@ int drm_gem_mmap_obj(struct drm_gem_object *obj, unsigned long obj_size,
>  		     struct vm_area_struct *vma)
>  {
>  	struct drm_device *dev = obj->dev;
> +	int ret;
>  
>  	/* Check for valid size. */
>  	if (obj_size < vma->vm_end - vma->vm_start)
>  		return -EINVAL;
>  
> -	if (obj->funcs && obj->funcs->vm_ops)
> -		vma->vm_ops = obj->funcs->vm_ops;
> -	else if (dev->driver->gem_vm_ops)
> -		vma->vm_ops = dev->driver->gem_vm_ops;
> -	else
> -		return -EINVAL;
> +	if (obj->funcs && obj->funcs->mmap) {
> +		ret = obj->funcs->mmap(obj, vma);
> +		if (ret)
> +			return ret;
> +		WARN_ON(!(vma->vm_flags & VM_DONTEXPAND));

Yeah I think checking for VM_DONTEXPAND is even better than checking for
just VM_SPECIAL.

Reviewed-by: Daniel Vetter <daniel.vetter@...ll.ch>
> +	} else {
> +		if (obj->funcs && obj->funcs->vm_ops)
> +			vma->vm_ops = obj->funcs->vm_ops;
> +		else if (dev->driver->gem_vm_ops)
> +			vma->vm_ops = dev->driver->gem_vm_ops;
> +		else
> +			return -EINVAL;
> +
> +		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);
> +	}
>  
> -	vma->vm_flags |= VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP;
>  	vma->vm_private_data = obj;
> -	vma->vm_page_prot = pgprot_writecombine(vm_get_page_prot(vma->vm_flags));
> -	vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);
>  
>  	/* Take a ref for this mapping of the object, so that the fault
>  	 * handler can dereference the mmap offset's pointer to the object.
> diff --git a/drivers/gpu/drm/drm_prime.c b/drivers/gpu/drm/drm_prime.c
> index 0a2316e0e812..0814211b0f3f 100644
> --- a/drivers/gpu/drm/drm_prime.c
> +++ b/drivers/gpu/drm/drm_prime.c
> @@ -713,6 +713,15 @@ int drm_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma)
>  	struct file *fil;
>  	int ret;
>  
> +	if (obj->funcs && obj->funcs->mmap) {
> +		ret = obj->funcs->mmap(obj, vma);
> +		if (ret)
> +			return ret;
> +		vma->vm_private_data = obj;
> +		drm_gem_object_get(obj);
> +		return 0;
> +	}
> +
>  	priv = kzalloc(sizeof(*priv), GFP_KERNEL);
>  	fil = kzalloc(sizeof(*fil), GFP_KERNEL);
>  	if (!priv || !fil) {
> -- 
> 2.18.1
> 

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