[Intel-gfx] [PATCH] drm/i915/bxt: Fix uninitialized variables in intel_check_sprite_plane

Jani Nikula jani.nikula at linux.intel.com
Tue Dec 8 03:20:39 PST 2015


On Mon, 07 Dec 2015, Nabendu Maiti <nabendu.bikash.maiti at intel.com> wrote:
> On 11/26/2015 11:33 PM, Nabendu Maiti wrote:
>>
>>
>> On 11/18/2015 10:56 PM, Ville Syrjälä wrote:
>>> On Wed, Nov 18, 2015 at 10:33:55PM +0530, Maiti, Nabendu Bikash wrote:
>>>>
>>>> On 11/18/2015 7:00 PM, Ville Syrjälä wrote:
>>>>> On Wed, Nov 18, 2015 at 06:48:37PM +0530, Maiti, Nabendu Bikash wrote:
>>>>>> Then just one line change of initializing the variables is better?
>>>>> Or maybe fix your compiler instead? I don't get any warning/errors 
>>>>> from
>>>>> this. What version of gcc are you using?
>>>>    I still get the warning and error if -Werror is enabled, And on
>>>> Makefile O1 optimization is enabled.
>>> And why exactly are you building with -O1?
>> I am using it for platform level debug symbol inclusion in elf which 
>> are excluded in Os/O2 . We need it.Else it will break generic kernel.
> Any comments? Please let me know if anything need to be done from my side.

Nobody is doing anything about this because it's the compiler being
silly.

We could merge your patch (provided it said this was a compiler issue)
but frankly it wouldn't take long for someone to submit a patch to drop
the unnecessary initializations again.

We could probably rearrange the code to avoid warnings (e.g. use a local
temp variable for state->visible, or use "goto out" where state->visible
is set to false) but seems like a lot of churn just to make a silly tool
happy.

Your build is bound to produce a lot of false positives across the whole
kernel build. What do you do with those?


BR,
Jani.


-- 
Jani Nikula, Intel Open Source Technology Center


More information about the Intel-gfx mailing list