linux-kernel - Re: [PATCH] drm/kms/atomic: Improved the func drm_atomic_set_crtc_for

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] [thread-next>] [day] [month] [year] [list]

Message-ID: <4fd92ffb-2518-1e9a-881e-5f4b0e4d4922@linux.intel.com>
Date:   Mon, 30 Jul 2018 10:00:08 +0200
From:   Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>
To:     Satendra Singh Thakur <satendra.t@...sung.com>,
        Gustavo Padovan <gustavo@...ovan.org>,
        Sean Paul <seanpaul@...omium.org>,
        David Airlie <airlied@...ux.ie>,
        dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Cc:     vineet.j@...sung.com, hemanshu.s@...sung.com, sst2005@...il.com
Subject: Re: [PATCH] drm/kms/atomic: Improved the func
 drm_atomic_set_crtc_for_connector

Op 30-07-18 om 08:14 schreef Satendra Singh Thakur:
> a. Used drm_atomic_get_crtc_state instead of drm_atomic_get_new_crtc_state.
>    Anyway new_state and state are same, this should make no difference.
>    This change will make the code of this func and the func
>    drm_atomic_set_crtc_for_plane similar.
> b. Added error checking after this func call
> c. Currently conn_state->crtc is getting assigned a value at two places
>    We can just reduce this assignment to one
>
> Signed-off-by: Satendra Singh Thakur <satendra.t@...sung.com>
> ---
>  drivers/gpu/drm/drm_atomic.c | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/gpu/drm/drm_atomic.c b/drivers/gpu/drm/drm_atomic.c
> index 895741e..907fb2d 100644
> --- a/drivers/gpu/drm/drm_atomic.c
> +++ b/drivers/gpu/drm/drm_atomic.c
> @@ -1552,16 +1552,16 @@ drm_atomic_set_crtc_for_connector(struct drm_connector_state *conn_state,
>  		return 0;
>  
>  	if (conn_state->crtc) {
> -		crtc_state = drm_atomic_get_new_crtc_state(conn_state->state,
> +		crtc_state = drm_atomic_get_crtc_state(conn_state->state,
>  							   conn_state->crtc);
> +		if (IS_ERR(crtc_state))
> +			return PTR_ERR(crtc_state);
The error checking is not required when using drm_atomic_get_new_crtc_state,
because the end of drm_atomic_get_connector_state() will fetch it for us.

The other place where conn_state->crtc is assigned is in drm_atomic_set_crtc_for_connector(),
which also makes sure we have the crtc_state part of the commit.

It would be a bug if we would have a conn_state without conn_state->crtc's state.
Did you hit a crash here?
>  		crtc_state->connector_mask &=
>  			~(1 << drm_connector_index(conn_state->connector));
>  
>  		drm_connector_put(conn_state->connector);
> -		conn_state->crtc = NULL;
>  	}
> -
>  	if (crtc) {
>  		crtc_state = drm_atomic_get_crtc_state(conn_state->state, crtc);
>  		if (IS_ERR(crtc_state))
> @@ -1571,7 +1571,6 @@ drm_atomic_set_crtc_for_connector(struct drm_connector_state *conn_state,
>  			1 << drm_connector_index(conn_state->connector);
>  
>  		drm_connector_get(conn_state->connector);
> -		conn_state->crtc = crtc;
>  
>  		DRM_DEBUG_ATOMIC("Link connector state %p to [CRTC:%d:%s]\n",
>  				 conn_state, crtc->base.id, crtc->name);
> @@ -1579,7 +1578,7 @@ drm_atomic_set_crtc_for_connector(struct drm_connector_state *conn_state,
>  		DRM_DEBUG_ATOMIC("Link connector state %p to [NOCRTC]\n",
>  				 conn_state);
>  	}
> -
> +	conn_state->crtc = crtc;
>  	return 0;
>  }
>  EXPORT_SYMBOL(drm_atomic_set_crtc_for_connector);

~Maarten