| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Since DSI PHY has been a separate platform device, it should not
depend on the resources in host to be functional. This change is
to trigger PHY operations in manager, instead of host, so that
host and PHY can be completely separated.
Signed-off-by: Hai Li <hali@codeaurora.org>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
In case of dual DSI, some registers in PHY1 have been programmed
during PLL0 clock's set_rate. The PHY1 reset called by host1 later
will silently reset those PHY1 registers. This change is to reset
and enable both PHYs before any PLL clock operation.
[Originally worked on by Hai Li <hali@codeaurora.org>. Fixed up
by Archit Taneja <architt@codeaurora.org>]
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
For some new types of DSI PHY, more settings depend on
use cases controlled by DSI manager. This change allows
DSI manager to setup PHY with a use case.
Signed-off-by: Hai Li <hali@codeaurora.org>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
The DSI host is required to configure more timings calculated
in PHY. By introducing a shared structure, this change allows
more timing information passed from PHY to host.
Signed-off-by: Hai Li <hali@codeaurora.org>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Create an init() op for dsi_phy which sets up things specific to
a given DSI PHY.
The dsi_phy driver probe expects every DSI version to get a
"dsi_phy_regulator" mmio base. This isn't the case for 8x96.
Creating an init() op will allow us to accommodate such
differences.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Add 8x96 DSI data in dsi_cfg. The downstream kernel's dsi_host driver
enables core_mmss_clk. We're seeing some branch clock warnings on
8x96 when enabling this. There doesn't seem to be any negative effect
with not enabling this clock, so use it once we figure out why we
get the warnings.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
The driver returns an error if a DSI DT node is populated, but no device
is connected to it or if the data-lane map isn't present. Ideally, such
a DSI node shouldn't be probed at all (i.e, its status should be set to
"disabled in DT"), but there isn't any harm in registering the DSI device
even if it doesn't have a bridge/panel connected to it.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
This code has been more or less picked up from the vc4 and intel
implementations of update_plane() funcs for cursor planes.
The update_plane() func is usually the drm_atomic_helper_update_plane
func that will issue an atomic commit with the plane updates. Such
commits are not intended to be done faster than the vsync rate.
The legacy cursor userspace API, on the other hand, expects the kernel
to handle cursor updates immediately.
Create a fast path in update_plane, which updates the cursor registers
and flushes the configuration. The fast path is taken when there is only
a change in the cursor's position in the crtc, or a change in the
cursor's crop co-ordinates. For anything else, we go via the slow path.
We take the slow path even when the fb changes, and when there is
currently no fb tied to the plane. This should hopefully ensure that we
always take a slow path for every new fb. This in turn should ensure that
the fb is pinned/prepared.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
In mdp5_plane_atomic_check, we get crtc_state from drm_plane_state.
Later, for cursor planes, we'll populate the update_plane() func that
takes a fast asynchronous path to implement cursor movements. There, we
would need to call a similar atomic_check func to validate the plane
state, but crtc_state would need to be derived differently.
Refactor mdp5_plane_atomic_check to mdp5_plane_atomic_check_with_state
such that the latter takes crtc_state as an argument.
This is similar to what the intel driver has done for async cursor
updates.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Register cursor drm_planes. The loop in modeset_init that inits the
planes and crtcs has to be refactored a bit. We first iterate all the
hwpipes to find the cursor planes. Then, we loop again to create
crtcs.
In msm_atomic_wait_for_commit_done, remove the check which bypasses
waiting for vsyncs if state->legacy_cursor_updates is true.
We will later create a fast path for cursor position changes in the
cursor plane's update_plane func that doesn't go via the regular
atomic commit path. For rest of cursor related updates, we will have
to wait for vsyncs, so ignore the legacy_cursor_updates flag.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
These are various changes added in preparation for cursor planes:
- Add a pipe_cursor block for 8x96 in mdp5_cfg.
- Add a new pipe CAP called MDP_PIPE_CAP_CURSOR. Use this to ensure we
assign a cursor SSPP for a drm_plane with type DRM_PLANE_TYPE_CURSOR.
- Update mdp5_ctl_blend_mask/ext_blend_mask funcs to incorporate cursor
SSPPs.
- In mdp5_ctl_blend, iterate through MAX_STAGES instead of stage_cnt,
we need to do this because we can now have empty stages in between.
- In mdp5_crtc_atomic_check, make sure that the cursor plane has the
highest zorder, and stage the cursor plane to the maximum stage #
present on the HW.
- Create drm_crtc_funcs that doesn't try to implement cursors using the
older LM cursor HW.
- Pass drm_plane_type in mdp5_plane_init instead of a bool telling
whether plane is primary or not.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
In MDP5 Layer Mixer HW, the blender output is only the blended color
components (i.e R, G and B, or COLOR0/1/2 in MDP5 HW terminology). This
is fed to the BG input of the next blender. We also need to provide an
alpha (COLOR3) value for the BG input at the next stage.
This is configured via using the REG_MDP5_LM_BLEND_COLOR_OUT register.
For each stage, we can propagate either the BG or FG alpha to the next
stage.
The approach taken by the driver is to propagate FG alpha, if the plane
staged on that blender has an alpha. If it doesn't, we try to propagate
the base layer's alpha.
This is borrowed from downstream MDP5 kernel driver. Without this, we
don't see any cursor plane content.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
The MDP5 plane's atomic_check ops doesn't perform clipping tests.
This didn't hurt us much in the past, but clipping becomes important
with cursor planes.
Use drm_plane_helper_check_state, the way rockchip/intel/mtk drivers
already do. Use these drivers as reference.
Clipping requires knowledge of the crtc width and height. This requires
us to call drm_atomic_helper_check_modeset before
drm_atomic_helper_check_planes in the driver's atomic_check op, because
check_modetest will populate the mode for the crtc, needed to populate
the clip rectangle.
We update the plane_enabled(state) local helper to use state->visible,
since state->visible and 'state->fb && state->crtc' represent the same
thing.
One issue with the existing code is that we don't have a way to disable
the plane when it's completely clipped out. Until there isn't an update
on the crtc (which would de-stage the plane), we would still see the
plane in its last 'visible' configuration.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Use SSPP_NONE in mdp5_plane_pipe() if there is now hwpipe allocated for
the drm_plane. Returning '0' means we are returning VIG0 pipe.
Also, use the mdp5_pipe enum to pass around the stage array. Initialize
the stage to SSPP_NONE by default.
We do the above because 1) Cursor plane has to be staged at the topmost
blender of the LM, which can result in empty stages in between 2) In
the future, when we support multiple LMs per CRTC. We could have stages
which don't have any pipe assigned to them.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
We currently create CRTCs equaling to the # of Layer Mixer blocks we
have on the MDP5 HW. This number is generally more than the # of encoders
(INTFs) we have in the MDSS HW. The number of encoders connected to
displays on the platform (as described by DT) would be even lesser.
Create only N drm_crtcs, where N is the number of drm_encoders
successfully registered. To do this, we call modeset_init_intf() before
we init the drm_crtcs and drm_planes.
Because of this change, setting encoder->possible_crtcs needs to be moved
from construct_encoder() to a later point when we know how many CRTCs we
have.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Count can't be non-zero. Changing to uint will also prevent future
warnings.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
For the DSI interfaces, the mdp5_kms core creates 2 encoders for video
and command modes.
Create only a single encoder per interface. When creating the encoder, set
the interface type to MDP5_INTF_MODE_NONE. It's the bridge (DSI/HDMI/eDP)
driver's responsibility to set a different interface type. It can use the
the kms func op set_encoder_mode to change the mode of operation, which
in turn would configure the interface type for the INTF.
In mdp5_cmd_encoder.c, we remove the redundant code, and make the commmand
mode funcs as helpers that are used in mdp5_encoder.c
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Rename the mdp5_encoder_* ops for active displays to
mdp5_vid_encoder_* ops.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
The mdp5 kms driver currently sets up multiple encoders per interface
(INTF), one for each kind of mode of operation it supports.
We create 2 drm_encoders for DSI, one for Video Mode and the other
for Command Mode operation. The reason behind this approach could have
been that we aren't aware of the DSI device's mode of operation when
we create the encoders.
This makes things a bit complicated, since these encoders have to
be further attached to the same DSI bridge. The easier way out is
to create a single encoder, and make the DSI driver set its mode
of operation when we know what the DSI device's mode flags are.
Start with providing a way to set the mdp5_intf_mode using a kms
func that sets the encoder's mode of operation. When constructing
a DSI encoder, we set the mode of operation to Video Mode as
default. When the DSI device is attached to the host, we probe the
DSI mode flags and set the corresponding mode of operation.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
We currently create 2 encoders for DSI interfaces, one for command
mode and other for video mode operation. This isn't needed as we
can't really use both the encoders at the same time. It also makes
connecting bridges harder.
Switch to creating a single encoder. For now, we assume that the
encoder is configured only in video mode. Later, the same encoder
would be usable in both modes.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
The commit "drm: bridge: Link encoder and bridge in core code" updated
the drm_bridge_attach() API to also include the drm_encoder pointer
the bridge attaches to.
The func msm_dsi_manager_bridge_init() now relies on the drm_encoder
pointer stored in msm_dsi->encoders to pass the encoder to the bridge
API.
msm_dsi->encoders is unfortunately set after this function is called,
resulting in us passing a NULL pointer to drm_brigde_attach. This
results in an error and the DSI driver probe fails.
Move the initialization of msm_dsi->encoders[] a bit up. Also, don't
try to set the encoder's bridge. That's now managed by the bridge
API.
Cc: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Define the block in advance so that the generated mdp5.xml.h doesn't
break build.
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Signed-off-by: Rob Clark <robdclark@gmail.com>
|
|
Suggested by Rob Herring. We still support the old names for
compatibility with downstream android dt files.
Cc: Rob Herring <robh@kernel.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Rob Herring <robh@kernel.org>
|
|
This was never documented or used in upstream dtb. It is used by
downstream bindings from android device kernels. But the quirks are
a property of the gpu revision, and as such are redundant to be listed
separately in dt. Instead, move the quirks to the device table.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
|
|
The original way we determined the gpu version was based on downstream
bindings from android kernel. A cleaner way is to get the version from
the compatible string.
Note that no upstream dtb uses these bindings. But the code still
supports falling back to the legacy bindings (with a warning), so that
we are still compatible with the gpu dt node from android device
kernels.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Rob Herring <robh@kernel.org>
|
|
The plan is to use the OPP bindings. For now, remove the documentation
for qcom,gpu-pwrlevels, and make the driver fall back to a safe low
clock if the node is not present.
Note that no upstream dtb use this node. For now we keep compatibility
with this node to avoid breaking compatibility with downstream android
dt files.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Acked-by: Rob Herring <robh@kernel.org>
|
|
etnaviv_gem.h header gets included twice. Remove duplicate.
Signed-off-by: Liviu Dudau <liviu@dudau.co.uk>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
|
|
Now that commandstreams are handled through the cmdbuf suballocator
the workaround to make the IOVA games work is not needed anymore.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
There are 3 big benefits to suballocating a single big DMA buffer
for command submission:
1. Avoid hammering CMA. The old way of allocating and freeing a DMA
buffer for each submission was hitting some of the real slow
pathes in CMA, as this allocator was not designed for a concurrent
small buffers load.
2. Less TLB flushes on IOMMUv2. If a new command buffer is mapped into
the GPU address space the MMU TLBs need to be flushed. By having
one big buffer statically mapped to the GPU, a lot of those flushes
can be avoided.
3. No funky workarounds for GC3000. The FE TLB flush on GC3000 isn't
reliable. To work around that we tried to lay out the cmdbufs in
the GPU address space in a way to avoid this issue. This hasn't
always worked if the address space is crowded. A single statically
mapped buffer avoids the erratum completely.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Don't allow IOMMUv2 to peek directly into the cmdbuf, but get the
needed PA through a dedicated function.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Don't call the IOMMU directly, but go through the new cmdbuf abstraction.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
This will get more complex with the following changes, so move it
into its own place.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
This ensures that the GPU isn't able to write into already freed
objects, as doing this in the IOVA reaper isn't enough, as the
gem_free_object path will also cause unmaps to happen.
On MMUv2 this also ensures that stale entries, which may have
been prefetched into the TLB will be purged.
The flush is low overhead, as it gets batched up with the next
user command buffer, so this isn't incuring an overhead for
each buffer map/unmap.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
|
|
Declare etnaviv_iommu_ops structure as const as it is only used when
the reference of one of its field is stored in the ops field of a
iommu_domain structure. This ops field is of type const, so
etnaviv_iommu_ops structures having similar properties can be declared
const too.
Done using Coccinelle.
Before and after size details of .o file remains the same after
cross compiling for arm architecture.
lst: Trimmed commit message, apply the same change to iommu_v2.
Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
|
|
The size computations done in the ioctl function use an integer.
If userspace submits a request with req->cmd_nr or req->cmd_buf_nr
set to INT_MAX, the integer computations overflow later, leading
to potential (kernel) memory corruption.
Prevent this issue by enforcing a limit on the number of submitted
commands, so that we have enough headroom later for the size
computations.
Note that this change has no impact on the currently available
users in userspace, like e.g. libdrm/exynos.
While at it, also make a comment about the size computation more
detailed.
Signed-off-by: Joonyoung Shim <jy0922.shim@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
We were trying to print an error message if we timed out here, but the
loop actually ends with "tries" set to UINT_MAX and not zero. Fix this
by changing from tries-- to --tries.
A for loop would actually be the most natural way to do this. My fix
means we only loop 99 times instead of 100 but that's probably ok.
Fixes: a696394c5224 ('drm/exynos: mixer: simplify loop in vp_win_reset()')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
This patch replaces specific atomic commit function
with atomic helper commit one.
For this, it removes existing atomic commit function
and relevant code specific to Exynos DRM and makes
atomic helper commit to be used instead.
Below are changes for the use of atomic helper commit:
- add atomic_commit_tail callback specific to Exynos DRM
. default implemention of atomic helper doesn't mesh well
with runtime PM so the device driver which supports runtime
PM should call drm_atomic_helper_commit_modeset_enables function
prior to drm_atomic_helper_commit_planes function call.
atomic_commit_tail callback implements this call ordering.
- allow plane commit only in case that CRTC device is enabled.
. for this, it calls atomic_helper_commit_planes function
with DRM_PLANE_COMMIT_ACTIVE_ONLY flag in atomic_commit_tail callback.
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.com>
|
|
This patch removes exynos_drm_crtc_cancel_page_flip call
when drm is closed because at that time, events will be released
by drm_events_release function.
Changelog v1:
- remove exynos_drm_crtc_cancel_page_flip function also because
this funtion isn't used anymore.
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
|
|
This patch adds runtime support calls to notify device core when MIC
device is really in use. Runtime PM is implemented by enabling and
disabling clocks like in other Exynos DRM subdrivers. Adding runtime
PM support is needed to let power domain with this device to be turned
off when display is not used.
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
This is the deprecated function for when you embedded the framebuffer
somewhere else (which breaks refcounting). But exynos is using
drm_framebuffer_remove and a free-standing fb, so this is rendundant.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
The OF graph is not necessary because the panel is a child of
dsi. therefore, the parse_dt function of dsi does not need to
check the remote_node connected to the panel. and the whole
parse_dt function should be refactored later.
Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
Before applying the patch, used the of_get_videomode function to
parse the display-timings in the panel which is the child driver
of dsi in the devicetree. this is wrong. So removed the
of_get_videomode and fixed to get videomode struct through
mode_set callback function.
Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
|
|
Set up the PULSE_EATER register (0x0010C) in etnaviv_gpu_hw_init. This
ports three mostly undocumented model/revision-specific register
overrides from the Vivante kernel driver.
This is relevant as at least the "disable internal DFS" for revisions >
0x5420 has shown to have a huge impact on shader performance (sped up
memory read performance by 7.5x and write performance by 1.5x) on an
affected GPU.
Signed-off-by: Wladimir J. van der Laan <laanwj@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
|
|
- Add PS.INST_ADDR (0x01028) and VS.INST_ADDR (0x0086C): GC3000 loads
shader code from these addresses if ICACHE is used.
- Add new NFE vertex stream addresses (0x14600).
- Add PE Multple Render Target pipe addresses (0x14800).
- Add TS Multiple Render Target pipe addresses (0x017C0, 0x17E0).
Signed-off-by: Wladimir J. van der Laan <laanwj@gmail.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
|
|
Now that we wait for DRM panels to be available before registering the
DRM device (returning -EPROBE_DEFER if the panel has not been probed
yet), we no longer need to put the fbdev creation code in
->output_poll_changed().
This removes the 10 secs delay between DRM dev registration and fbdev
creation (polling period = 10 seconds).
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Reported-by: Alex Vazquez <avazquez.dev@gmail.com>
|
|
drm_debugfs_cleanup() now removes all minor->debugfs_list entries
automatically, so the drm_driver.debugfs_cleanup callback is not
needed.
Cc: eric@anholt.net
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170126225621.12314-17-noralf@tronnes.org
|
|
It adds the TV Encoder driver to support video output in PAL and NTSC
format. The driver uses syscon/regmap interface to configure register
bit sitting in SYSCTRL module for DAC power control.
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
|
Hai Li
Archit Taneja
Rob Clark
Liviu Dudau
Lucas Stach
Bhumika Goyal
Joonyoung Shim
Dan Carpenter
Inki Dae
Marek Szyprowski
Daniel Vetter
Hoegeun Kwon
Wladimir J. van der Laan
Boris Brezillon
Noralf Trønnes
Shawn Guo