Re: [Buildroot] [RFC v3 1/5] package/qt6: Add qmake functionality

From: Thomas Petazzoni via buildroot <buildroot@buildroot.org>
To: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Cc: buildroot@buildroot.org,
	Andrea Ricchi <andrea.ricchi@amarulasolutions.com>,
	Roy Kollen Svendsen <roykollensvendsen@gmail.com>,
	Angelo Compagnucci <angelo.compagnucci@gmail.com>,
	Jesse Van Gavere <jesseevg@gmail.com>,
	linux-amarula@amarulasolutions.com,
	Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>,
	Angelo Compagnucci <angelo@amarulasolutions.com>
Subject: Re: [Buildroot] [RFC v3 1/5] package/qt6: Add qmake functionality
Date: Fri, 15 Aug 2025 17:28:51 +0200	[thread overview]
Message-ID: <20250815172851.77009124@windsurf> (raw)
In-Reply-To: <20250814153259.6822-2-dario.binacchi@amarulasolutions.com>

Hello Dario,

Thanks for bringing this further. This really seems like a copy/paste
of the Qt5 logic, with even some parts that are dead code (see below).
How much have you validated that everything added by this patch is
actually *needed*, as opposed to just blindly copy/pasted from the Qt5
logic?

Also, in the commit log:

> flags/cross-compiler the biggest difference is that where previously
> qt.conf was used to fixup per-paths builds, this is now always
> created with changed contents so qmake looks at the correct target
> directories.

I'm a bit confused by what this means. Is this about
QT6_INSTALL_QT_CONF being called not only by QT6_QT_CONF_FIXUP in the
BR2_PER_PACKAGE_DIRECTORIES=y case but also being called by:

QT6BASE_POST_INSTALL_STAGING_HOOKS += QT6_INSTALL_QT_CONF

 ?

If so, why does this has to be different?

On Thu, 14 Aug 2025 17:32:55 +0200
Dario Binacchi <dario.binacchi@amarulasolutions.com> wrote:

> diff --git a/package/qt6/qt6base/qt6base.mk b/package/qt6/qt6base/qt6base.mk
> index 51a4f1b5be7f..697d1a96a920 100644
> --- a/package/qt6/qt6base/qt6base.mk
> +++ b/package/qt6/qt6base/qt6base.mk
> @@ -55,6 +55,9 @@ QT6BASE_CONF_OPTS = \
>  	-DFEATURE_system_zlib=ON \
>  	-DFEATURE_system_libb2=ON
>  
> +QT6BASE_CFLAGS = $(TARGET_CFLAGS)
> +QT6BASE_CXXFLAGS = $(TARGET_CXXFLAGS)

These variables are not "changed" anywhere, so you could directly use
TARGET_CFLAGS/TARGET_CXXFLAGS. Unless the goal is to keep the logic as
similar as qt5 as possible.

> +
>  # x86 optimization options. While we have a BR2_X86_CPU_HAS_AVX512, it
>  # is not clear yet how it maps to all the avx512* options of Qt, so we
>  # for now keeps them disabled.
> @@ -431,5 +434,20 @@ define QT6BASE_RM_USR_MKSPECS
>  endef
>  QT6BASE_TARGET_FINALIZE_HOOKS += QT6BASE_RM_USR_MKSPECS
>  
> +define QT6BASE_MAKE_BR_SPEC
> +	mkdir -p $(@D)/mkspecs/devices/linux-buildroot-g++/
> +	sed 's/@EGLFS_DEVICE@/$(QT6BASE_EGLFS_DEVICE)/g' \
> +		$(QT6BASE_PKGDIR)/qmake.conf.in > \
> +		$(@D)/mkspecs/devices/linux-buildroot-g++/qmake.conf

$(QT6BASE_EGLFS_DEVICE) doesn't seem to be defined anywhere, so I have
some doubts that this is doing something useful.

This also means the @EGLFS_DEVICE@ in qmake.conf.in is not relevant.

> +	$(INSTALL) -m 0644 -D $(QT6BASE_PKGDIR)/qplatformdefs.h \
> +		$(@D)/mkspecs/devices/linux-buildroot-g++/qplatformdefs.h
> +endef
> +
> +QT6BASE_CONF_OPTS += -DQT_QMAKE_DEVICE_OPTIONS=CROSS_COMPILE="$(TARGET_CROSS)";BR_COMPILER_CFLAGS="$(QT6BASE_CFLAGS)";BR_COMPILER_CXXFLAGS="$(QT6BASE_CXX_FLAGS)"

What is this doing? Does it actually work? I'm asking because that
doesn't seem to exist in the Qt5 case.

Thanks!

Thomas
-- 
Thomas Petazzoni, co-owner and CEO, Bootlin
Embedded Linux and Kernel engineering and training
https://bootlin.com
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot