[PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[Christian König]

This is to work around problems with libva and vainfo.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/i915/i915_drv.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index bbe1a5d56480..12615641a34b 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -3039,7 +3039,7 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(I915_BATCHBUFFER, drm_noop, DRM_AUTH),
 	DRM_IOCTL_DEF_DRV(I915_IRQ_EMIT, drm_noop, DRM_AUTH),
 	DRM_IOCTL_DEF_DRV(I915_IRQ_WAIT, drm_noop, DRM_AUTH),
-	DRM_IOCTL_DEF_DRV(I915_GETPARAM, i915_getparam_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(I915_GETPARAM, i915_getparam_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_SETPARAM, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_ALLOC, drm_noop, DRM_AUTH),
 	DRM_IOCTL_DEF_DRV(I915_FREE, drm_noop, DRM_AUTH),
@@ -3052,13 +3052,13 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(I915_HWS_ADDR, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_GEM_INIT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER, i915_gem_execbuffer_ioctl, DRM_AUTH),
-	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER2_WR, i915_gem_execbuffer2_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER2_WR, i915_gem_execbuffer2_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_PIN, i915_gem_reject_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_GEM_UNPIN, i915_gem_reject_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY),
-	DRM_IOCTL_DEF_DRV(I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_SET_CACHING, i915_gem_set_caching_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_GET_CACHING, i915_gem_get_caching_ioctl, DRM_RENDER_ALLOW),
-	DRM_IOCTL_DEF_DRV(I915_GEM_THROTTLE, i915_gem_throttle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(I915_GEM_THROTTLE, i915_gem_throttle_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_ENTERVT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_GEM_LEAVEVT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
 	DRM_IOCTL_DEF_DRV(I915_GEM_CREATE, i915_gem_create_ioctl, DRM_RENDER_ALLOW),
@@ -3077,7 +3077,7 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
 	DRM_IOCTL_DEF_DRV(I915_OVERLAY_ATTRS, intel_overlay_attrs_ioctl, DRM_MASTER),
 	DRM_IOCTL_DEF_DRV(I915_SET_SPRITE_COLORKEY, intel_sprite_set_colorkey_ioctl, DRM_MASTER),
 	DRM_IOCTL_DEF_DRV(I915_GET_SPRITE_COLORKEY, drm_noop, DRM_MASTER),
-	DRM_IOCTL_DEF_DRV(I915_GEM_WAIT, i915_gem_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
+	DRM_IOCTL_DEF_DRV(I915_GEM_WAIT, i915_gem_wait_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_CREATE_EXT, i915_gem_context_create_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_DESTROY, i915_gem_context_destroy_ioctl, DRM_RENDER_ALLOW),
 	DRM_IOCTL_DEF_DRV(I915_REG_READ, i915_reg_read_ioctl, DRM_RENDER_ALLOW),
-- 
2.17.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

[PATCH 2/2] drm: revert "allow render capable master with DRM_AUTH ioctls"

[Christian König]

This reverts commit 8059add0478e29cb641936011a8fcc9ce9fd80be.

It's breaking radv and most likely old libdrm_amdgpu versions.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/drm_ioctl.c | 20 ++++----------------
 1 file changed, 4 insertions(+), 16 deletions(-)

diff --git a/drivers/gpu/drm/drm_ioctl.c b/drivers/gpu/drm/drm_ioctl.c
index d337f161909c..ce8a70875bd5 100644
--- a/drivers/gpu/drm/drm_ioctl.c
+++ b/drivers/gpu/drm/drm_ioctl.c
@@ -508,13 +508,6 @@ int drm_version(struct drm_device *dev, void *data,
 	return err;
 }
 
-static inline bool
-drm_render_driver_and_ioctl(const struct drm_device *dev, u32 flags)
-{
-	return drm_core_check_feature(dev, DRIVER_RENDER) &&
-		(flags & DRM_RENDER_ALLOW);
-}
-
 /**
  * drm_ioctl_permit - Check ioctl permissions against caller
  *
@@ -529,19 +522,14 @@ drm_render_driver_and_ioctl(const struct drm_device *dev, u32 flags)
  */
 int drm_ioctl_permit(u32 flags, struct drm_file *file_priv)
 {
-	const struct drm_device *dev = file_priv->minor->dev;
-
 	/* ROOT_ONLY is only for CAP_SYS_ADMIN */
 	if (unlikely((flags & DRM_ROOT_ONLY) && !capable(CAP_SYS_ADMIN)))
 		return -EACCES;
 
-	/* AUTH is only for master ... */
-	if (unlikely((flags & DRM_AUTH) && drm_is_primary_client(file_priv))) {
-		/* authenticated ones, or render capable on DRM_RENDER_ALLOW. */
-		if (!file_priv->authenticated &&
-		    !drm_render_driver_and_ioctl(dev, flags))
-			return -EACCES;
-	}
+	/* AUTH is only for authenticated or render client */
+	if (unlikely((flags & DRM_AUTH) && !drm_is_render_client(file_priv) &&
+		     !file_priv->authenticated))
+		return -EACCES;
 
 	/* MASTER is only for master or control clients */
 	if (unlikely((flags & DRM_MASTER) &&
-- 
2.17.1

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[Chris Wilson]

Quoting Christian König (2019-04-17 12:25:24)
> This is to work around problems with libva and vainfo.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>

Mediated HW access should not require additional DRM_AUTH, if you can
already create buffers, execbuf itself should not be any more dangerous.

Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[Daniel Vetter]

On Wed, Apr 17, 2019 at 01:25:24PM +0200, Christian König wrote:
> This is to work around problems with libva and vainfo.
> 
> Signed-off-by: Christian König <christian.koenig@amd.com>

I don't think this is a good idea since it leaves the core ioctl out, and
it leaves all other drivers out (and inconsistency in these winsys
relevant things is not great imo).

Since we seem to be stuck on the overall discussion here, on the revert:

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

Assuming we don't magically unblock ourselves suddenly ofc.
-Daniel
> ---
>  drivers/gpu/drm/i915/i915_drv.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> index bbe1a5d56480..12615641a34b 100644
> --- a/drivers/gpu/drm/i915/i915_drv.c
> +++ b/drivers/gpu/drm/i915/i915_drv.c
> @@ -3039,7 +3039,7 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
>  	DRM_IOCTL_DEF_DRV(I915_BATCHBUFFER, drm_noop, DRM_AUTH),
>  	DRM_IOCTL_DEF_DRV(I915_IRQ_EMIT, drm_noop, DRM_AUTH),
>  	DRM_IOCTL_DEF_DRV(I915_IRQ_WAIT, drm_noop, DRM_AUTH),
> -	DRM_IOCTL_DEF_DRV(I915_GETPARAM, i915_getparam_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(I915_GETPARAM, i915_getparam_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_SETPARAM, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_ALLOC, drm_noop, DRM_AUTH),
>  	DRM_IOCTL_DEF_DRV(I915_FREE, drm_noop, DRM_AUTH),
> @@ -3052,13 +3052,13 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
>  	DRM_IOCTL_DEF_DRV(I915_HWS_ADDR, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_INIT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER, i915_gem_execbuffer_ioctl, DRM_AUTH),
> -	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER2_WR, i915_gem_execbuffer2_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(I915_GEM_EXECBUFFER2_WR, i915_gem_execbuffer2_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_PIN, i915_gem_reject_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_UNPIN, i915_gem_reject_pin_ioctl, DRM_AUTH|DRM_ROOT_ONLY),
> -	DRM_IOCTL_DEF_DRV(I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(I915_GEM_BUSY, i915_gem_busy_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_SET_CACHING, i915_gem_set_caching_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_GET_CACHING, i915_gem_get_caching_ioctl, DRM_RENDER_ALLOW),
> -	DRM_IOCTL_DEF_DRV(I915_GEM_THROTTLE, i915_gem_throttle_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(I915_GEM_THROTTLE, i915_gem_throttle_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_ENTERVT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_LEAVEVT, drm_noop, DRM_AUTH|DRM_MASTER|DRM_ROOT_ONLY),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_CREATE, i915_gem_create_ioctl, DRM_RENDER_ALLOW),
> @@ -3077,7 +3077,7 @@ static const struct drm_ioctl_desc i915_ioctls[] = {
>  	DRM_IOCTL_DEF_DRV(I915_OVERLAY_ATTRS, intel_overlay_attrs_ioctl, DRM_MASTER),
>  	DRM_IOCTL_DEF_DRV(I915_SET_SPRITE_COLORKEY, intel_sprite_set_colorkey_ioctl, DRM_MASTER),
>  	DRM_IOCTL_DEF_DRV(I915_GET_SPRITE_COLORKEY, drm_noop, DRM_MASTER),
> -	DRM_IOCTL_DEF_DRV(I915_GEM_WAIT, i915_gem_wait_ioctl, DRM_AUTH|DRM_RENDER_ALLOW),
> +	DRM_IOCTL_DEF_DRV(I915_GEM_WAIT, i915_gem_wait_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_CREATE_EXT, i915_gem_context_create_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_GEM_CONTEXT_DESTROY, i915_gem_context_destroy_ioctl, DRM_RENDER_ALLOW),
>  	DRM_IOCTL_DEF_DRV(I915_REG_READ, i915_reg_read_ioctl, DRM_RENDER_ALLOW),
> -- 
> 2.17.1
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[Emil Velikov]

On 2019/04/17, Christian König wrote:
> This is to work around problems with libva and vainfo.
> 
This part reverts a commit from 2013. Something I would wager that will cause
multiple problems across the board.

If we look at the intel libva driver in particular, as-is this will cause all
get param queries on the render node to fail. Something that used to work
since render node support was introduced.

Analogously this will break the Intel Mesa driver when using the render node
and issuing the ioctls.

HTH
Emil




_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/2] drm/i915: remove DRM_AUTH from IOCTLs which also have DRM_RENDER_ALLOW

[Emil Velikov]

On 2019/04/17, Emil Velikov wrote:
> On 2019/04/17, Christian König wrote:
> > This is to work around problems with libva and vainfo.
> > 
> This part reverts a commit from 2013. Something I would wager that will cause
> multiple problems across the board.
> 
> If we look at the intel libva driver in particular, as-is this will cause all
> get param queries on the render node to fail. Something that used to work
> since render node support was introduced.
> 
> Analogously this will break the Intel Mesa driver when using the render node
> and issuing the ioctls.
> 
Silly moment of mine - I've misread which piece gets reverted.
Pardon for the noise.

Personally I would use something like the following, which helps the reader
understand what's happening.


At the moment the following ioctls are exposed by both primary and render
nodes. Yet there's no code-flow difference based on the originating node type.
Thus in reality the exact same code is executed without the "authentication".

So let's drop the DRM_AUTH annotation.

In practise this fixes cases like (but not limited to) libva where the primary
node is opened, yet authentication is lacking prior to issuing the ioctl.

I915_GETPARAM
I915_GEM_EXECBUFFER2_WR
I915_GEM_BUSY
I915_GEM_THROTTLE
I915_GEM_WAIT

HTH
Emil


_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel