[Buildroot] [PATCH v2] qt5base: don't enable KMS if it doesn't have udev

[Thomas Petazzoni <thomas.petazzoni@free-electrons.com>]

Hello,

On Fri,  8 Jul 2016 10:17:32 +0900, Akihiko Odaki wrote:
> Signed-off-by: Akihiko Odaki <akihiko.odaki.4i@stu.hosei.ac.jp>

Thanks for this new iteration. When you send a patch that fixes an
autobuilder failure, please include something like this in the commit
log:

Fixes:

    http://autobuild.buildroot.org/results/....

(which should point to the autobuilder failure being fixed by this
patch).

However, I have some comments below.

> diff --git a/package/qt5/qt5base/qt5base.mk b/package/qt5/qt5base/qt5base.mk
> index bf541b0..60b4e94 100644
> --- a/package/qt5/qt5base/qt5base.mk
> +++ b/package/qt5/qt5base/qt5base.mk
> @@ -30,8 +30,11 @@ QT5BASE_CONFIGURE_OPTS += \
>  
>  # Uses libgbm from mesa3d
>  ifeq ($(BR2_PACKAGE_MESA3D_OPENGL_EGL),y)
> -QT5BASE_CONFIGURE_OPTS += -kms -gbm
> +QT5BASE_CONFIGURE_OPTS += -gbm
>  QT5BASE_DEPENDENCIES += mesa3d
> +ifeq ($(BR2_PACKAGE_HAS_UDEV),y)
> +QT5BASE_CONFIGURE_OPTS += -kms
> +endif

Does it really make sense to enable GBM without KMS? The only place
where I see GBM actually being used in Qt's configure script is:

if [ "$CFG_EGLFS" = "yes" ] && [ "$CFG_KMS" = "yes" ] && [ "$CFG_GBM" = "yes" ]; then
    QT_CONFIG="$QT_CONFIG eglfs_gbm"
    CFG_EGLFS_GBM="yes"

Which seems to indicate that *both* KMS and GBM are needed in order to
enable the eglfs_gbm module. Can you check this? If that's the case,
then a better patch would be:

ifeq ($(BR2_PACKAGE_MESA3D_OPENGL_EGL)$(BR2_PACKAGE_HAS_UDEV),yy)
QT5BASE_CONFIGURE_OPTS += -kms -gbm
else
...
endif

Thanks,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com