linux-kernel - Re: [PATCH v2.1 1/2] drm/msm: remove resv fields from msm_gem

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: <20190514181229.GR17077@art_vandelay>
Date:   Tue, 14 May 2019 14:12:29 -0400
From:   Sean Paul <sean@...rly.run>
To:     Brian Masney <masneyb@...tation.org>
Cc:     robdclark@...il.com, sean@...rly.run, bjorn.andersson@...aro.org,
        dri-devel@...ts.freedesktop.org, linux-arm-msm@...r.kernel.org,
        freedreno@...ts.freedesktop.org, airlied@...ux.ie, daniel@...ll.ch,
        linux-kernel@...r.kernel.org, linus.walleij@...aro.org,
        jonathan@...ek.ca, robh@...nel.org
Subject: Re: [PATCH v2.1 1/2] drm/msm: remove resv fields from msm_gem_object
 struct

On Mon, May 13, 2019 at 07:41:04PM -0400, Brian Masney wrote:
> The msm_gem_object structure contains resv and _resv fields that are
> no longer needed since the reservation object is now stored on
> drm_gem_object. msm_atomic_prepare_fb() and msm_atomic_prepare_fb()
> both referenced the wrong reservation object, and would lead to an
> attempt to dereference a NULL pointer. Correct those two cases to
> point to the correct reservation object.
> 
> Signed-off-by: Brian Masney <masneyb@...tation.org>

Thanks Brian for your patches. I've applied them to drm-misc-next-fixes for
inclusion in 5.2

Sean

> Reviewed-by: Bjorn Andersson <bjorn.andersson@...aro.org>
> Tested-by: Bjorn Andersson <bjorn.andersson@...aro.org>
> Fixes: dd55cf6929e6 ("drm: msm: Switch to use drm_gem_object reservation_object")
> ---
> This is a split out version of this patch from where I am working to get
> the display working on the Nexus 5:
> 
> https://lore.kernel.org/lkml/20190509020352.14282-2-masneyb@onstation.org/
> 
> Bjorn asks:
>     This resolves a NULL-pointer dereference about to show up in v5.2-rc1,
>     so please pick this up for -rc.
> 
> In this version, I dropped the change to msm_gem_new_impl() that
> mistakenly removed 'msm_obj->base.resv = resv;'. I did not put a v3 on
> this patch since I wanted to keep that version number for the larger work
> to get the display working on the Nexus 5 so I went with 2.1 here. :)
> 
>  drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c | 4 +---
>  drivers/gpu/drm/msm/msm_atomic.c          | 4 +---
>  drivers/gpu/drm/msm/msm_gem.h             | 4 ----
>  3 files changed, 2 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> index 4fca24b8702c..f3d009a3dc63 100644
> --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_plane.c
> @@ -781,7 +781,6 @@ static int dpu_plane_prepare_fb(struct drm_plane *plane,
>  	struct dpu_plane_state *pstate = to_dpu_plane_state(new_state);
>  	struct dpu_hw_fmt_layout layout;
>  	struct drm_gem_object *obj;
> -	struct msm_gem_object *msm_obj;
>  	struct dma_fence *fence;
>  	struct dpu_kms *kms = _dpu_plane_get_kms(&pdpu->base);
>  	int ret;
> @@ -800,8 +799,7 @@ static int dpu_plane_prepare_fb(struct drm_plane *plane,
>  	 *       implicit fence and fb prepare by hand here.
>  	 */
>  	obj = msm_framebuffer_bo(new_state->fb, 0);
> -	msm_obj = to_msm_bo(obj);
> -	fence = reservation_object_get_excl_rcu(msm_obj->resv);
> +	fence = reservation_object_get_excl_rcu(obj->resv);
>  	if (fence)
>  		drm_atomic_set_fence_for_plane(new_state, fence);
>  
> diff --git a/drivers/gpu/drm/msm/msm_atomic.c b/drivers/gpu/drm/msm/msm_atomic.c
> index f5b1256e32b6..131c23a267ee 100644
> --- a/drivers/gpu/drm/msm/msm_atomic.c
> +++ b/drivers/gpu/drm/msm/msm_atomic.c
> @@ -49,15 +49,13 @@ int msm_atomic_prepare_fb(struct drm_plane *plane,
>  	struct msm_drm_private *priv = plane->dev->dev_private;
>  	struct msm_kms *kms = priv->kms;
>  	struct drm_gem_object *obj;
> -	struct msm_gem_object *msm_obj;
>  	struct dma_fence *fence;
>  
>  	if (!new_state->fb)
>  		return 0;
>  
>  	obj = msm_framebuffer_bo(new_state->fb, 0);
> -	msm_obj = to_msm_bo(obj);
> -	fence = reservation_object_get_excl_rcu(msm_obj->resv);
> +	fence = reservation_object_get_excl_rcu(obj->resv);
>  
>  	drm_atomic_set_fence_for_plane(new_state, fence);
>  
> diff --git a/drivers/gpu/drm/msm/msm_gem.h b/drivers/gpu/drm/msm/msm_gem.h
> index c5ac781dffee..812d1b1369a5 100644
> --- a/drivers/gpu/drm/msm/msm_gem.h
> +++ b/drivers/gpu/drm/msm/msm_gem.h
> @@ -86,10 +86,6 @@ struct msm_gem_object {
>  
>  	struct llist_node freed;
>  
> -	/* normally (resv == &_resv) except for imported bo's */
> -	struct reservation_object *resv;
> -	struct reservation_object _resv;
> -
>  	/* For physically contiguous buffers.  Used when we don't have
>  	 * an IOMMU.  Also used for stolen/splashscreen buffer.
>  	 */
> -- 
> 2.20.1
> 

-- 
Sean Paul, Software Engineer, Google / Chromium OS