linux-kernel - Re: [PATCH] drm: atomic helper: fix W=1 warnings

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: <20191004130835.GX1208@intel.com>
Date:   Fri, 4 Oct 2019 16:08:35 +0300
From:   Ville Syrjälä <ville.syrjala@...ux.intel.com>
To:     Benjamin GAIGNARD <benjamin.gaignard@...com>
Cc:     Benjamin Gaignard <benjamin.gaignard@...aro.org>,
        David Airlie <airlied@...ux.ie>,
        Daniel Vetter <daniel@...ll.ch>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        ML dri-devel <dri-devel@...ts.freedesktop.org>
Subject: Re: [PATCH] drm: atomic helper: fix W=1 warnings

On Fri, Oct 04, 2019 at 12:36:56PM +0000, Benjamin GAIGNARD wrote:
> 
> On 10/4/19 2:27 PM, Ville Syrjälä wrote:
> > On Fri, Oct 04, 2019 at 12:48:02PM +0200, Benjamin Gaignard wrote:
> >> Le jeu. 3 oct. 2019 à 17:46, Ville Syrjälä
> >> <ville.syrjala@...ux.intel.com> a écrit :
> >>> On Thu, Oct 03, 2019 at 05:37:15PM +0200, Benjamin Gaignard wrote:
> >>>> Le jeu. 3 oct. 2019 à 17:05, Ville Syrjälä
> >>>> <ville.syrjala@...ux.intel.com> a écrit :
> >>>>> On Thu, Oct 03, 2019 at 04:46:54PM +0200, Benjamin Gaignard wrote:
> >>>>>> Le jeu. 3 oct. 2019 à 16:27, Ville Syrjälä
> >>>>>> <ville.syrjala@...ux.intel.com> a écrit :
> >>>>>>> On Mon, Sep 09, 2019 at 03:52:05PM +0200, Benjamin Gaignard wrote:
> >>>>>>>> Fix warnings with W=1.
> >>>>>>>> Few for_each macro set variables that are never used later.
> >>>>>>>> Prevent warning by marking these variables as __maybe_unused.
> >>>>>>>>
> >>>>>>>> Signed-off-by: Benjamin Gaignard <benjamin.gaignard@...com>
> >>>>>>>> ---
> >>>>>>>>   drivers/gpu/drm/drm_atomic_helper.c | 36 ++++++++++++++++++------------------
> >>>>>>>>   1 file changed, 18 insertions(+), 18 deletions(-)
> >>>>>>>>
> >>>>>>>> diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c
> >>>>>>>> index aa16ea17ff9b..b69d17b0b9bd 100644
> >>>>>>>> --- a/drivers/gpu/drm/drm_atomic_helper.c
> >>>>>>>> +++ b/drivers/gpu/drm/drm_atomic_helper.c
> >>>>>>>> @@ -262,7 +262,7 @@ steal_encoder(struct drm_atomic_state *state,
> >>>>>>>>              struct drm_encoder *encoder)
> >>>>>>>>   {
> >>>>>>>>        struct drm_crtc_state *crtc_state;
> >>>>>>>> -     struct drm_connector *connector;
> >>>>>>>> +     struct drm_connector __maybe_unused *connector;
> >>>>>>> Rather ugly. IMO would be nicer if we could hide something inside
> >>>>>>> the iterator macros to suppress the warning.
> >>>>>> Ok but how ?
> >>>>>> connector is assigned in the macros but not used later and we can't
> >>>>>> set "__maybe_unused"
> >>>>>> in the macro.
> >>>>>> Does another keyword exist for that ?
> >>>>> Stick a (void)(connector) into the macro?
> >>>> That could work but it will look strange inside the macro.
> >>>>
> >>>>> Another (arguably cleaner) idea would be to remove the connector/crtc/plane
> >>>>> argument from the iterators entirely since it's redundant, and instead just
> >>>>> extract it from the appropriate new/old state as needed.
> >>>>>
> >>>>> We could then also add a for_each_connector_in_state()/etc. which omit
> >>>>> s the state arguments and just has the connector argument, for cases where
> >>>>> you don't care about the states when iterating.
> >>>> That may lead to get a macro for each possible combination of used variables.
> >>> We already have new/old/oldnew, so would "just" add one more.
> >> Not just one, it will be one each new/old/oldnew macro to be able to distinguish
> >> when connector is used or not.
> > What I'm suggesting is this:
> > for_each_connector_in_state(state, connector, i)
> > for_each_old_connector_in_state(state, old_conn_state, i)
> > for_each_new_connector_in_state(state, new_conn_state, i)
> > for_each_oldnew_connector_in_state(state, old_conn_state, new_conn_state, i)
> >
> > So only four in total for each object type, instead of the current
> > three.
> 
> You are missing these cases: old and connector, new and connector,
> 
> old and new and connector are needed together.

No, that's redundant. You can always get the connector from
old/new_conn_state->connector if you need it.

-- 
Ville Syrjälä
Intel