[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-id: <1397734130-21019-1-git-send-email-a.hajda@samsung.com>
Date: Thu, 17 Apr 2014 13:28:47 +0200
From: Andrzej Hajda <a.hajda@...sung.com>
To: dri-devel@...ts.freedesktop.org
Cc: Andrzej Hajda <a.hajda@...sung.com>,
Marek Szyprowski <m.szyprowski@...sung.com>,
Inki Dae <inki.dae@...sung.com>,
Kyungmin Park <kyungmin.park@...sung.com>,
linux-samsung-soc@...r.kernel.org (moderated list:ARM/S5P EXYNOS AR...),
Tomasz Figa <t.figa@...sung.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
David Airlie <airlied@...ux.ie>,
linux-kernel@...r.kernel.org (open list),
linux-arm-kernel@...ts.infradead.org (moderated list:ARM/S5P EXYNOS
AR...), Russell King - ARM Linux <linux@....linux.org.uk>,
Arnd Bergmann <arnd@...db.de>
Subject: [PATCH RFC 0/3] drm/exynos: refactoring drm initialization/cleanup code
Hi,
The patchset presents alternative approach to superdevice DT node
and components framework. It also refactors Exynos DRM device initialization.
The first patch uses linker sections to get rid of ifdef macros, it is not
essential for the rest of patchset but it makes code more readable.
Similar approach is used by irqchip, clks and clk_sources.
Any comments are welcome, especially regarding linker script usage in drivers.
The second patch proposes pending_components framework - lightweight alternative
for componentized devices.
Details are described in the patch description. But as it is an alternative for
component framework I would like to show differences:
- it is simpler and more straightforward,
- it requires only minimal changes to component drivers,
- only one callback to implement,
- no unwinding, component framework in case of error removes all already added
components and unbinds all already bound components,
- it is instantiated per superdevice, component framework uses global list,
the disadvantage is that we create dependency between components and
the superdevice, but in case of Exynos DRM the dependency is present already,
- no two stage component initialization: probe, bind,
- it is more lightweight,
- it clearly separates device probing order issue from registering interfaces
provided by the device.
The third patch implements pending_components framework in Exynos DRM.
Details are in patch description.
The patchset is based on exynos-drm-next branch.
Regards
Andrzej
Andrzej Hajda (3):
drm/exynos: refactor drm drivers registration code
drivers/base: provide lightweight framework for componentized devices
drm/exynos: use pending_components for components tracking
drivers/base/Kconfig | 3 +
drivers/base/Makefile | 1 +
drivers/base/pending_components.c | 93 +++++++++++
drivers/gpu/drm/exynos/Kconfig | 1 +
drivers/gpu/drm/exynos/Makefile | 2 +
drivers/gpu/drm/exynos/exynos_dp_core.c | 38 +++--
drivers/gpu/drm/exynos/exynos_drm.lds.S | 9 +
drivers/gpu/drm/exynos/exynos_drm_drv.c | 250 ++++++++++------------------
drivers/gpu/drm/exynos/exynos_drm_drv.h | 33 ++--
drivers/gpu/drm/exynos/exynos_drm_dsi.c | 43 +++--
drivers/gpu/drm/exynos/exynos_drm_fimc.c | 36 ++--
drivers/gpu/drm/exynos/exynos_drm_fimd.c | 39 +++--
drivers/gpu/drm/exynos/exynos_drm_g2d.c | 19 ++-
drivers/gpu/drm/exynos/exynos_drm_gsc.c | 32 ++--
drivers/gpu/drm/exynos/exynos_drm_ipp.c | 20 ++-
drivers/gpu/drm/exynos/exynos_drm_rotator.c | 29 +++-
drivers/gpu/drm/exynos/exynos_drm_vidi.c | 18 +-
drivers/gpu/drm/exynos/exynos_hdmi.c | 55 ++++--
drivers/gpu/drm/exynos/exynos_mixer.c | 15 +-
include/linux/pending_components.h | 30 ++++
20 files changed, 466 insertions(+), 300 deletions(-)
create mode 100644 drivers/base/pending_components.c
create mode 100644 drivers/gpu/drm/exynos/exynos_drm.lds.S
create mode 100644 include/linux/pending_components.h
--
1.8.3.2
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists