[Mesa-dev] [PATCH 6/8] freedreno: drop unnecessary -Wno-packed-bitfield-compat

Rob Clark robdclark at gmail.com
Thu Feb 25 01:37:59 UTC 2016


On Wed, Feb 24, 2016 at 8:21 PM, Emil Velikov <emil.l.velikov at gmail.com> wrote:
> On 24 February 2016 at 19:11, Rob Clark <robdclark at gmail.com> wrote:
>> hmm, we did need this at one point in time for the packed structs used
>> for shader instruction assembly.. maybe newer gcc doesn't warn about
>> this (since it mostly seems related to a bug in older gcc versions)?
>> I guess since (afaiu) android provides it's own compiler per android
>> release, I suppose we don't have to worry about someone building w/
>> any random old gcc version, so I don't see any issue dropping this for
>> the android build.
>>
>> On Wed, Feb 24, 2016 at 1:56 PM, Rob Herring <robh at kernel.org> wrote:
>>> Enabling this warning doesn't generate any warnings with gcc, but is an
>>> unknown option for clang, so drop it.
>>>
>>> Cc: Rob Clark <robdclark at gmail.com>
>>> Signed-off-by: Rob Herring <robh at kernel.org>
>>
>> Acked-by: Rob Clark <robdclark at gmail.com>
>>
> It would be nice to keep things aligned between both builds. Rob H/Rob
> C any objections if we drop it from the autotools one ?
> Alternatively can we please keep it, commented out, with the above
> note as comment in the code.

Without remembering which gcc version I was using at the time that I
added that, I am less confident about removing it from it from
autotools build.  I guess if android change was a comment-out rather
than a delete, at least w/ git-blame you could more easily trace it
back to a commit msg..

(I wonder if there is a clever way to, perhaps via patchwork, go from
a commit-id back to mailing list discussion?)

BR,
-R


More information about the mesa-dev mailing list