The MCDE DSI link hardware which is modeled like a bridge
in DRM, connected further to the panel bridge, creating
a pipeline.
We have been using the .pre_enable(), .enable(),
.disable() and .post_disable() callbacks from the bridge
to set this up in a chained manner: first the display
controller goes online and then in successive order
each bridge in the pipeline. Inside DRM it works
like this:
drm_atomic_helper_commit_tail()
drm_atomic_helper_commit_modeset_enables()
struct drm_crtc_helper_funcs .atomic_enable()
struct drm_simple_display_pipe_funcs .enable()
MCDE display enable call
drm_atomic_bridge_chain_enable()
struct drm_bridge_funcs .pre_enable()
mcde_dsi_bridge_pre_enable()
panel_bridge_pre_enable()
struct drm_panel_funcs .prepare()
struct drm_bridge_funcs .enable()
mcde_dsi_bridge_enable()
panel_bridge_enable()
struct drm_panel_funcs .enable()
A similar sequence is executed for disabling.
Unfortunately this is not what the hardware needs: at
a certain stage in the enablement of the display
controller the DSI link needs to come up to support
video mode, else something (like a FIFO flow) locks
up the hardware and we never get picture.
Fix this by simply leaving the pre|enable and
post|disable callbacks unused, and establish two
cross-calls from the display controller to bring up
the DSI link at the right place in the display
bring-up sequence and vice versa in the shutdown
sequence.
For command mode displays, it works just fine to
also enable the display flow early. The only time
we hold it back right now is in one-shot mode,
on-demand display updates.
When combined with the previous patch and some patches
for the S6E63M0 display controller to support DSI
mode, this gives working display on the Samsung
GT-I8190 (Golden) phone. It has also been tested working
on the Samsung GT-S7710 (Skomer) phone.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: newbytee@protonmail.com
Cc: Stephan Gerhold <stephan@gerhold.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20200808223122.1492124-4-linus.walleij@linaro.org
To make sure that the MCDE is in a reasonable state during
set-up, perform a reset by power cycling the block by dropping
the on-chip regulator reference after probe. The display
subsystem (DSS) has no dedicated reset line so dropping
the EPOD regulator is the only real way of resetting it.
We introduce code to enable and disable the regulator in
the display enable/disable callbacks.
We move the generic MCDE setup such as muxing of DPI
signals and masking of interrupts to the display
handling.
When we drop the power to the whole display subsystem, not
only MCDE but also the DSI links lose their state. Therefore
we move the DSI block reset and hardware initialization
code to the mcde_dsi_bridge_pre_enable() callback so this
happens every time we start up the bridge, as we may have
lost the power.
We move the final disablement of the interrupts and
clocks to the mcde_dsi_bridge_post_disable() callback
rather than have it in the mcde_dsi_bridge_disable()
callback, as some control messages may still be sent
over the DSI host after the bridge has been shut down.
This (together with a patch for the corresponding
panel) makes the Samsung GT-S7710 successfully disable
and re-enable its display, cutting all power while
disabled and re-initializing the hardware when coming
back up.
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Cc: newbytee@protonmail.com
Cc: Stephan Gerhold <stephan@gerhold.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20200808223122.1492124-3-linus.walleij@linaro.org
Revamp the way that the flow of data to the display is
defined.
I realized that the hardware supports something like
5 different modes of flow: oneshot, command with TE IRQ,
command with BTA (bus turn around) and TE IRQ, video
with TE IRQ and video without TE IRQ instead synchronizing
to the output of the MCDE DSI formatter.
Like before the selection of the type of flow is done
from the DSI driver when we attach it to the MCDE and we
get to know what the display wants.
The new video mode synchronization method from the MCDE DSI
formatter is used on some upstream devices such as Golden.
This is the new default for video mode: stateless panels
do not as a rule generate TE IRQs.
Another semantic change is that we stop sending
a TE request before every command when sending data to
a display in command mode: this should only be explicitly
requested when using BTA, according to the vendor driver.
This has been tested and works fine with the command mode
displays I have. (All that are supported upstream.)
Reported-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Cc: Stephan Gerhold <stephan@gerhold.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20200729090915.252730-2-linus.walleij@linaro.org
Add missing include to fix build error:
hibmc_drm_drv.c:385:1: warning: data definition has no type or storage
class [enabled by default]
hibmc_drm_drv.c:385:1: error: type defaults to ‘int’ in declaration
of ‘module_init’ [-Werror=implicit-int]
hibmc_drm_drv.c:385:1: warning: parameter names (without types) in function
of ‘module_exit’ [-Werror=implicit-int]
hibmc_drm_drv.c:385:292:1: warning: parameter names (without types) in
function declaration [enabled by default]
Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1597289955-27381-1-git-send-email-tiantao6@hisilicon.com
Backmerging drm-next into drm-misc-next for nouveau and panel updates.
Resolves a conflict between ttm and nouveau, where struct ttm_mem_res got
renamed to struct ttm_resource.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
The memory used to be allocated with devres helpers and released
automatically. In rare circumstances, the memory's release could
have happened before the DRM device got released, which would have
caused memory corruption of some kind. Now we're embedding the data
structures in struct hibmc_drm_private. The whole release problem
has been resolved, because struct hibmc_drm_private is allocated
with drmm_kzalloc and always released with the DRM device.
Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1597218179-3938-3-git-send-email-tiantao6@hisilicon.com
Currently, downstream port type is only reported in debugfs. This
information should be considered important since it reflects the actual
physical connector type. Some userspace (e.g. window compositors)
may want to show this info to a user.
The 'subconnector' property is already utilized for DVI-I and TV-out for
reporting connector subtype.
The initial motivation for this feature came from i2c test [1].
It is supposed to be skipped on VGA connectors, but it cannot
detect VGA over DP and fails instead.
v2:
- Ville: utilized drm_dp_is_branch()
- Ville: implement DP 1.0 downstream type info
- Replaced create_dp_properties with add_dp_subconnector_property
- Added dp_set_subconnector_property helper
v4:
- Ville: add DP1.0 best assumption about subconnector
- Ville: assume DVI is DVI-D
- Ville: reuse Writeback enum value for Virtual subconnector
- Renamed #defines: HDMI -> HDMIA, DP -> DisplayPort
v5: rebase
v6:
- Jani Nikula: renamed a function name
- Jani Nikula: addressed the issues with documentation
[1]: https://bugs.freedesktop.org/show_bug.cgi?id=104097
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Signed-off-by: Jeevan B <jeevan.b@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1587732655-17544-1-git-send-email-jeevan.b@intel.com
There is a spelling mistake in a DRM_ERROR message. Fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
There is a spelling mistake in a DRM_ERROR message. Fix it.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
When converting to atomic the state reset was done by directly calling
the functions, and before the modeset object was fully initialized.
This means the various ->dev pointers weren't set up.
After
commit 51f644b40b
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Fri Jun 12 18:00:49 2020 +0200
drm/atomic-helper: reset vblank on crtc reset
this started to oops because now we're trying to derefence
drm_crtc->dev. Fix this up by entirely switching over to
drm_mode_config_reset, called once everything is set up.
Fixes: 51f644b40b ("drm/atomic-helper: reset vblank on crtc reset")
Reported-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Tested-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
These if statements are supposed to be true if we ended the
list_for_each_entry() loops without hitting a break statement but they
don't work.
In the first loop, we increment "i" after the "if (i == unit)" condition
so we don't necessarily know that "i" is not equal to unit at the end of
the loop.
In the second loop we exit when mode is not pointing to a valid
drm_display_mode struct so it doesn't make sense to check "mode->type".
Fixes: a278724aa2 ("drm/vmwgfx: Implement fbdev on kms v2")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
The "entry" pointer is an offset from the list head and it doesn't
point to a valid vmw_legacy_display_unit struct. Presumably the
intent was to point to the last entry.
Also the "i++" wasn't used so I have removed that as well.
Fixes: d7e1958dbe ("drm/vmwgfx: Support older hardware.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
Make use of the struct_size() helper instead of an open-coded version
in order to avoid any potential type mistakes.
This code was detected with the help of Coccinelle and, audited and
fixed manually.
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Roland Scheidegger <sroland@vmware.com>
This driver is tested with two panels individually with Apq8016-IFC6309 board
https://www.inforcecomputing.com/products/single-board-computers-sbc/qualcomm-snapdragon-410-inforce-6309-micro-sbc
1. 1366x768@60 auo,b101xtn01 data-mapping = "jeida-24"
2. 800x480@60 innolux,at070tn92 data-mapping = "vesa-24"
- power off sequence in proper order
- put_unaligned_be16, put_unaligned_le32 macros used
- static function for mode_valid
- len initialized
- MODE_CLOCK_HIGH handled properly
- bus_formats handled in mode_valid
- GENMASK and FIELD_PREP used
- Kconfig proper indentation
- error handling endpoint data-lanes
- check for bus_formats unsupported
- display_timings naming local variables
- help modified
- ~vsdelay dynamic value set based on the
calculation of dsi speed, output speed, blanking
- panel->connector_type removed
- dual port implemented
- devm_drm_panel_bridge_add method used instead of panel
description modified
- regulator enable and disable with proper orders and delays
as per the spec
- removed drm_connector_status
- added bus_formats
- mdelay to usleep_range
- magic number to macros for CLRSI and mux registers
description modified
- replaced u32 instead of uint32_t
- updated alphabetic order of headers
- added SPDX identifier license
Signed-off-by: Vinay Simha BN <simhavcs@gmail.com>
Reviewed-by: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Sam Ravnborg <sam.ravnborg@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1594388491-15129-2-git-send-email-simhavcs@gmail.com
VKMS needs vblank interrupts enabled to capture CRC. When vblank is
disabled, tests like kms_cursor_crc and kms_pipe_crc_basic getting stuck
waiting for a capture that will not occur until vkms wakes up. This patch
adds a helper to set composer and ensure that vblank remains enabled as
long as the CRC capture is needed.
It clears the execution of the following kms_cursor_crc subtests:
1. pipe-A-cursor-[size,alpha-opaque, NxN-(on-screen, off-screen, sliding,
random, fast-moving])] - successful when running individually.
2. pipe-A-cursor-dpms passes again
3. pipe-A-cursor-suspend also passes
The issue was initially tracked in the sequential execution of IGT
kms_cursor_crc subtests: when running the test sequence or one of its
subtests twice, the odd execs complete and the pairs get stuck in an
endless wait. In the IGT code, calling a wait_for_vblank on preparing for
CRC capture prevented the busy-wait. But the problem persisted in the
pipe-A-cursor-dpms and -suspend subtests.
Checking the history, the pipe-A-cursor-dpms subtest was successful when,
in vkms_atomic_commit_tail, instead of using the flip_done op, it used
wait_for_vblanks. Another way to prevent blocking was wait_one_vblank when
enabling crtc. However, in both cases, pipe-A-cursor-suspend persisted
blocking in the 2nd start of CRC capture, which may indicate that
something got stuck in the step of CRC setup. Indeed, wait_one_vblank in
the crc setup was able to sync things and free all kms_cursor_crc
subtests. Besides, other alternatives to force enabling vblanks or prevent
disabling them such as calling drm_crtc_put_vblank or modeset_enables
before commit_planes + offdelay = 0, also unlock all subtests executions.
Finally, due to vkms's dependence on vblank interruptions to perform
tasks, this patch uses refcount to ensure that vblanks happen when
enabling composer and while crc capture is needed.
Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
Cc: Haneen Mohammed <hamohammed.sa@gmail.com>
v2:
- extract a vkms_set_composer helper
- fix vblank refcounting for the disabling case
v3:
- make the vkms_set_composer helper static
- review the credit tags
Co-debugged-by: Sidong Yang <realwakka@gmail.com>
Signed-off-by: Sidong Yang <realwakka@gmail.com>
Signed-off-by: Melissa Wen <melissa.srw@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[danvet: add changelog back in]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20200808120900.pudwwrfz44g3rqx7@smtp.gmail.com
Reproducing bug report here:
After hibernating and resuming, DPM is not enabled. This remains the case
even if you test hibernate using the steps here:
https://www.kernel.org/doc/html/latest/power/basic-pm-debugging.html
I debugged the problem, and figured out that in the file hardwaremanager.c,
in the function, phm_enable_dynamic_state_management(), the check
'if (!hwmgr->pp_one_vf && smum_is_dpm_running(hwmgr) && !amdgpu_passthrough(adev) && adev->in_suspend)'
returns true for the hibernate case, and false for the suspend case.
This means that for the hibernate case, the AMDGPU driver doesn't enable DPM
(even though it should) and simply returns from that function.
In the suspend case, it goes ahead and enables DPM, even though it doesn't need to.
I debugged further, and found out that in the case of suspend, for the
CIK/Hawaii GPUs, smum_is_dpm_running(hwmgr) returns false, while in the case of
hibernate, smum_is_dpm_running(hwmgr) returns true.
For CIK, the ci_is_dpm_running() function calls the ci_is_smc_ram_running() function,
which is ultimately used to determine if DPM is currently enabled or not,
and this seems to provide the wrong answer.
I've changed the ci_is_dpm_running() function to instead use the same method that
some other AMD GPU chips do (e.g Fiji), which seems to read the voltage controller.
I've tested on my R9 390 and it seems to work correctly for both suspend and
hibernate use cases, and has been stable so far.
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=208839
Signed-off-by: Sandeep Raghuraman <sandy.8925@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
