Merge branch 'drm-next-4.19' of git://people.freedesktop.org/~agd5f/linux into drm-next

Fixes for 4.19:
- Fix UVD 7.2 instance handling
- Fix UVD 7.2 harvesting
- GPU scheduler fix for when a process is killed
- TTM cleanups
- amdgpu CS bo_list fixes
- Powerplay fixes for polaris12 and CZ/ST
- DC fixes for link training certain HMDs
- DC fix for vega10 blank screen in certain cases

From: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180801222906.1016-1-alexander.deucher@amd.com

Merge branch 'etnaviv/next' of https://git.pengutronix.de/git/lst/linux into drm-next

From: Lucas Stach <l.stach@pengutronix.de>
"not much to de-stage this time. Changes from Philipp and Souptick to
use memset32 more and switch the fault handler to the new vm_fault_t
and two small fixes for issues that can be hit in rare corner cases
from me."

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1533563808.2809.7.camel@pengutronix.de

Merge branch 'drm-tda998x-devel' of git://git.armlinux.org.uk/~rmk/linux-arm into drm-next

From: Russell King <rmk@armlinux.org.uk>
As per the patches posted, discussed and tested by Peter Rosin, this
converts TDA998x to a bridge driver, while still allowing Armada and
TI LCDC to continue using it as they always have done.  It also gets
rid of the private .fill_modes function, and tweaks the TMDS divider
calculation to be more correct to the available information.

[airlied: fixed two conflicts]
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180802093421.GA29670@rmk-PC.armlinux.org.uk

Merge branch 'for-upstream/hdlcd' of git://linux-arm.org/linux-ld into drm-next

From: Liviu Dudau <Liviu.Dudau@arm.com>
"I managed to loose track of a few patches for HDLCD while focusing
on Mali DP and found them again when investigating an issue with
the way HDLCD behaves on teardown. They can go into drm-next for
one of the v4.19-rcX if you're not going to do another pull request
before the merge window."

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180731170831.GF17455@e110455-lin.cambridge.arm.com

drm/i2c: tda998x: correct PLL divider calculation

The serializer PLL divider is a power-of-two divider, so our calculation
which assumes that it's a numerical divider is incorrect.  Replace it
with one that results in a power-of-two divider value instead.

Tested with all supported modes with a Samsung S24C750.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: get rid of private fill_modes function

We can achieve the same effect via the get_modes() method, rather than
wrapping the fill_modes helper.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: move mode_valid() to bridge

Move the mode_valid() implementation to the bridge instead of the
connector, as we're checking the bridge's capabilities.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: register bridge outside of component helper

Register the bridge outside of the component helper as we have
drivers that wish to use the tda998x without its encoder.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: cleanup from previous changes

Cleanup the code a little from the effects of the previous changes:
- Move tda998x_destroy() to be above tda998x_create()
- Use 'dev' directly in tda998x_create() where appropriate.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: allocate tda998x_priv inside tda998x_create()

Move the tda998x_priv allocation inside tda998x_create() and simplify
the tda998x_create()'s arguments.  Pass the same to tda998x_destroy().

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: convert to bridge driver

Convert tda998x to a bridge driver with built-in encoder support for
compatibility with existing component drivers.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/etnaviv: fix crash in GPU suspend when init failed due to buffer placement

When the suballocator was unable to provide a suitable buffer for the MMUv1
linear window, we roll back the GPU initialization. As the GPU is runtime
resumed at that point we need to clear the kernel cmdbuf suballoc entry to
properly skip any attempt to manipulate the cmdbuf when the GPU gets shut
down in the runtime suspend later on.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

drm/etnaviv: change return type to vm_fault_t

Use new return type vm_fault_t for fault handler. For
now, this is just documenting that the function returns
a VM_FAULT value rather than an errno. Once all instances
are converted, vm_fault_t will become a distinct type.

Ref- commit 1c8f422059ae ("mm: change return type to vm_fault_t")

Previously vm_insert_page() returns err which driver
mapped into VM_FAULT_* type. The new function
vmf_insert_page() will replace this inefficiency by
returning VM_FAULT_* type.

vmf_error() is the newly introduce inline function
in 4.17-rc6.

Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

drm/etnaviv: protect sched job submission with fence mutex

The documentation of drm_sched_job_init and drm_sched_entity_push_job has
been clarified. Both functions should be called under a shared lock, to
avoid jobs getting pushed into the scheduler queue in a different order
than their sched_fence seqnos, which will confuse checks that are looking
at the seqnos to infer information about completion order.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

drm/etnaviv: mmuv2: use memset32 to init scratch page

Replace the open-coded scratch page initialization loop with memset32

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>

drm/i2c: tda998x: move tda998x_set_config() into tda998x_create()

Move the non-DT configuration of the TDA998x into tda998x_create()
so that we do all setup in one place.

Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: split tda998x_encoder_dpms into enable/disable

This fits better with the drm_bridge callbacks for when this
driver becomes a drm_bridge.

Suggested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Peter Rosin <peda@axentia.se>
[edited by rmk to just split the tda998x_encoder_dpms() function
 and restore the double-disable protection we originally had,
 preserving original behaviour.]
Tested-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/i2c: tda998x: find the drm_device via the drm_connector

This prepares for being a drm_bridge which will not register the
encoder. That makes the connector the better choice.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/ttm: clean up non-x86 definitions on ttm_tt

All non-x86 definitions are moved to ttm_set_memory header, so remove it from
ttm_tt.c.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: Add ttm_set_pages_wc and ttm_set_pages_uc helper

These two helpers will be used on set page caching.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: fix missed conversion of set_pages_array_uc

This patch fixed the error when do not configure CONFIG_X86, otherwise, below
error will be encountered.

All errors (new ones prefixed by >>):

   drivers/gpu/drm/ttm/ttm_page_alloc_dma.c: In function 'ttm_set_pages_caching':
>> drivers/gpu/drm/ttm/ttm_page_alloc_dma.c:272:7: error: implicit declaration of function 'set_pages_array_uc'; did you mean
+'ttm_set_pages_array_uc'? [-Werror=implicit-function-declaration]
      r = set_pages_array_uc(pages, cpages);
          ^~~~~~~~~~~~~~~~~~
          ttm_set_pages_array_uc
   cc1: some warnings being treated as errors

Reported-by: kbuild test robot <lkp@intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge branch 'drm-armada-devel' of git://git.armlinux.org.uk/~rmk/linux-arm into drm-next

This set of changes migrates Armada DRM from legacy modeset to atomic
modeset.  This is everything from the "Transition Armada DRM planes to
atomic state" and "Finish Armada DRM transition to atomic modeset"
patch sets as posted on drm-devel, excluding the "Finish Armada DRM DT
support" series.

These series did not evoke any comments - if there are any, these can
be addressed via follow up patches.

Developed and tested on Dove Cubox with xf86-video-armada including the
overlay plane, and also tested with the tools in libdrm.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180730110543.GA30664@rmk-PC.armlinux.org.uk

Merge tag 'drm-msm-next-2018-07-30' of git://people.freedesktop.org/~robclark/linux into drm-next

A bit larger this time around, due to introduction of "dpu1" support
for the display controller in sdm845 and beyond.  This has been on
list and undergoing refactoring since Feb (going from ~110kloc to
~30kloc), and all my review complaints have been addressed, so I'd be
happy to see this upstream so further feature work can procede on top
of upstream.

Also includes the gpu coredump support, which should be useful for
debugging gpu crashes.  And various other misc fixes and such.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGv-8y3zguY0Mj1vh=o+vrv_bJ8AwZ96wBXYPvMeQT2XcA@mail.gmail.com

drm/amdgpu/pm: Fix potential Spectre v1

idx can be indirectly controlled by user-space, hence leading to a
potential exploitation of the Spectre variant 1 vulnerability.

This issue was detected with the help of Smatch:

drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c:408 amdgpu_set_pp_force_state()
warn: potential spectre issue 'data.states'

Fix this by sanitizing idx before using it to index data.states

Notice that given that speculation windows are large, the policy is
to kill the speculation on the first load and not worry if it can be
completed with a dependent load/store [1].

[1] https://marc.info/?l=linux-kernel&m=152449131114778&w=2

Cc: stable@vger.kernel.org
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add missing void parameter to dc_create_transfer_func

Add a missing void parameter to function dc_create_transfer_func, fixes
sparse warning:

warning: non-ANSI function declaration of function 'dc_create_transfer_func'

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: Replace ttm_bo_unref with ttm_bo_put

The function ttm_bo_put releases a reference to a TTM buffer object. The
function's name is more aligned to the Linux kernel convention of naming
ref-counting function _get and _put.

A call to ttm_bo_unref takes the address of the TTM BO object's pointer and
clears the pointer's value to NULL. This is not necessary in most cases and
sometimes even worked around by the calling code. A call to ttm_bo_put only
releases the reference without clearing the pointer.

The current behaviour of cleaning the pointer is kept in the calling code,
but should be removed if not required in a later patch.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: Replace ttm_bo_reference with ttm_bo_get

The function ttm_bo_get acquires a reference on a TTM buffer object. The
function's name is more aligned to the Linux kernel convention of naming
ref-counting function _get and _put.

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Replace ttm_bo_unref with ttm_bo_put

The function ttm_bo_put releases a reference to a TTM buffer object. The
function's name is more aligned to the Linux kernel convention of naming
ref-counting function _get and _put.

A call to ttm_bo_unref takes the address of the TTM BO object's pointer and
clears the pointer's value to NULL. This is not necessary in most cases and
sometimes even worked around by the calling code. A call to ttm_bo_put only
releases the reference without clearing the pointer.

The current behaviour of cleaning the pointer is kept in the calling code,
but should be removed if not required in a later patch.

v2:
 * set prefix to drm/amdgpu

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Replace ttm_bo_reference with ttm_bo_get

The function ttm_bo_get acquires a reference on a TTM buffer object. The
function's name is more aligned to the Linux kernel convention of naming
ref-counting function _get and _put.

v2:
 * changed prefix to drm/amdgpu

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/scheduler: stop setting rq to NULL

We removed the redundancy of having an extra scheduler field, so we
can't set the rq to NULL any more or otherwise won't know which
scheduler to use for the cleanup.

Just remove the entity from the scheduling list instead.

Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=107367
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/scheduler: only kill entity if last user is killed v2

Note which task is using the entity and only kill it if the last user of
the entity is killed. This should prevent problems when entities are leaked to
child processes.

v2: add missing kernel doc

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Acked-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: create an empty bo_list if no handle is provided

Instead of having extra handling just create an empty bo_list when no
handle is provided.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: allocate the bo_list array after the list

This avoids multiple allocations for the head and the array.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add bo_list iterators

Add helpers to iterate over all entries in a bo_list.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: nuke amdgpu_bo_list_free

The RCU grace period is harmless and avoiding it is not worth the effort
of doubling the implementation.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: always recreate bo_list

The bo_list handle is allocated by OP_CREATE, so in OP_UPDATE here we just
re-create the bo_list object and replace the handle. This way we don't
need locking to protect the bo_list because it's always re-created when
changed.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: move bo_list defines to amdgpu_bo_list.h

Further demangle amdgpu.h

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new amdgpu_vm_bo_trace_cs() function v2

This allows us to trace all VM ranges which should be valid inside a CS.

v2: dump mappings without BO as well

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-and-tested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com> (v1)
Reviewed-by: Huang Rui <ray.huang@amd.com> (v1)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: return error if both BOs and bo_list handle is given

Return -EINVAL when both the BOs as well as a list handle is provided in
the IOCTL.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix total size calculation

long might only be 32bit in size and we can easily use more than 4GB
here.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Acked-by: Huang Rui <ray.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/sched: remove unneeded -Iinclude/drm compiler flag

I refactored the include directives under include/drm/ some time ago.
This flag is unneeded.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Acked-by: Nayan Deshmukh <nayan26deshmukh@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add proper error handling to amdgpu_bo_list_get

Otherwise we silently don't use a BO list when the handle is invalid.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Chunming Zhou <david1.zhou@amd.com>
Reviewed-by: Huang Rui <ray.huang@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix a reversed condition

This test was reversed so it would end up leading to vddnb value
can't be read via hwmon on APU.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/pp: Convert voltage unit in mV*4 to mV on CZ/ST

the voltage showed in debugfs and hwmon should be in mV

Reviewed-by: Evan Quan <evan.quan@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/amd/pp: Delete unused temp variables

Only delete the dead temp variables in Polaris.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/pp/Polaris12: Fix a chunk of registers missed to program

DIDTConfig_Polaris12[] table missed a big chunk of data.

Pointed by aidan.fabius <aidan.fabius@coreavi.com>

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

drm/arm/hdlcd: Reject atomic commits that disable only the plane

The HDLCD engine needs an active plane while the CRTC is active, as
it will start scanning out data from HDLCD_REG_FB_BASE once it gets
enabled. Make sure that the only available plane doesn't get disabled
while the CRTC remains active, as this will scanout invalid data.

Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm: arm: hdlcd: Use drm_atomic_helper_shutdown() to disable planes on removal

The plane cleanup handler currently calls drm_plane_helper_disable(),
which is a legacy helper function. Replace it with a call to
drm_atomic_helper_shutdown() at removal time. The plane .destroy()
handler now consisting only of a call to drm_plane_cleanup(), replace it
with direct calls to that function.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm: arm: hdlcd: Don't destroy plane manually in hdlcd_setup_crtc()

The top-level error handler calls drm_mode_config_cleanup() which will
destroy all planes. There's no need to destroy them manually in lower
error handlers.

Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/arm/hdlcd: Use drm_fb_cma_fbdev_init/fini()

Use drm_fb_cma_fbdev_init() and drm_fb_cma_fbdev_fini() which relies on
the fact that drm_device holds a pointer to the drm_fb_helper structure.
This means that the driver doesn't have to keep track of that.
Also use the drm_fb_helper functions directly.

Cc: Liviu Dudau <liviu.dudau@arm.com>
Cc: Brian Starkey <brian.starkey@arm.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Liviu Dudau <liviu.dudau@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

drm/arm/hdlcd: Use drm_mode_config_helper_suspend/resume()

Replace driver's code with the generic helpers that do the same thing
including the NULL check.

Cc: Liviu Dudau <liviu.dudau@arm.com>
Cc: Brian Starkey <brian.starkey@arm.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>

Merge branch 'drm-udl-next' of git://people.freedesktop.org/~airlied/linux into drm-next

A set of cleanups and fixes for Mikulas for using udl on arm boards.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAPM=9twQNgrmfe0=Okq1NTgWHRQXy+AzeDy8A0p_-y856p4vtA@mail.gmail.com

udl-kms: dont spam the syslog with debug messages

The udl kms driver writes messages to the syslog whenever some application
opens or closes /dev/fb0 and whenever the user switches between the
Xserver and the console.

This patch changes the priority of these messages to debug.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: use spin_lock_irq instead of spin_lock_irqsave

spin_lock_irqsave and spin_unlock_irqrestore is inteded to be called from
a context where it is unknown if interrupts are enabled or disabled (such
as interrupt handlers). From a process context, we should call
spin_lock_irq and spin_unlock_irq, that avoids the costly pushf and popf
instructions.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: avoid prefetch

Modern processors can detect linear memory accesses and prefetch data
automatically, so there's no need to use prefetch.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: avoid division

Division is slow, so it shouldn't be done by the pixel generating code.
The driver supports only 2 or 4 bytes per pixel, so we can replace
division with a shift.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: fix crash due to uninitialized memory

We must use kzalloc when allocating the fb_deferred_io structure.
Otherwise, the field first_io is undefined and it causes a crash.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: handle allocation failure

Allocations larger than PAGE_ALLOC_COSTLY_ORDER are unreliable and they
may fail anytime. This patch fixes the udl kms driver so that when a large
alloactions fails, it tries to do multiple smaller allocations.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>

udl-kms: change down_interruptible to down

If we leave urbs around, it causes not only leak, but also memory
corruption. This patch fixes the function udl_free_urb_list, so that it
always waits for all urbs that are in progress.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>

dt-bindings: msm/disp: Add bindings for Snapdragon 845 DPU

Adds bindings for Snapdragon 845 display processing unit

Changes in v2:
- Use SoC specific compatibles for mdss and dpu (Rob Herring)
- Use assigned-clocks to set initial clock frequency (Rob Herring)

Changes in v3 (all suggested by Rob Herring):
- Rename mdss_phys to mdss
- Correct description for clocks/assigned-clocks
- Rename mdp_phys to mdp
- Rename vbif_phys to vbif
- Remove redundant interrupt-parent from mdss_mdp
- Fully specify 'ranges' and use relative reg address in mdss_mdp

Cc: Rob Herring <robh+dt@kernel.org>
Signed-off-by: Jeykumar Sankaran <jsanka@codeaurora.org>
Signed-off-by: Rajesh Yadav <ryadav@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

dt-bindings: msm/dsi: Add mdp transfer time to msm dsi binding

Adds mdp transfer time to msm dsi binding

Changes in v3:
- Added Rob's R-b

Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jeykumar Sankaran <jsanka@codeaurora.org>
Signed-off-by: Rajesh Yadav <ryadav@codeaurora.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/disp/dpu: Mark a handful of functions as static

Mark a number of static functions that are only unsed in the file
that defines them and remove the prototypes from the headers where
needed.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/disp/dpu: Remove unused functions from dpu_formats.c

Remove dpu_format_get_block_size, dpu_format_get_framebuffer_size,
dpu_set_scaler_v2 and dpu_copy_formats they are unused and unneeded.

Signed-off-by: Jordan Crouse <jcrouse@codeauorora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/disp/dpu: Remove dpu_kms_utils

None of the functions in dpu_kms_utils.c seem to be used so
remove them all.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/disp/dpu: Remove unused code from drm_crtc.c

Remove a chunk of unused code from drm_crtc.c, namely
dpu_crtc_res_add, dpu_crtc_res_get, dpu_crtc_res_put
and associated static functions.

Also zap dpu_crtc_event_queue(), helper functions
and members.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/disp/dpu: Remove unused code from drm_encoder.c

Remove dpu_encoder_check_mode and dpu_encoder_helper_hw_release
frmo drm_encoder.c as they appear to be unused.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: Replace PTR_RET with PTR_ERR_OR_ZERO

PTR_RET is deprecated, use PTR_ERR_OR_ZERO instead.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: mark PM functions as __maybe_unused

The suspend/resume functions are not referenced when power
management is disabled:

drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c:1288:12: error: 'dpu_runtime_resume' defined but not used [-Werror=unused-function]
drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c:1261:12: error: 'dpu_runtime_suspend' defined but not used [-Werror=unused-function]

This marks them as __maybe_unused to let the compiler
drop the functions without complaining.

Fixes: 591225291ca2 ("drm/msm: Add SDM845 DPU support")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/dpu: fix mismatch in function argument.

Fix the sparse error. the dpu_rm_init declaration is not consistent
with the implement.

Signed-off-by: zhong jiang <zhongjiang@huawei.com>
[robclark un-typo'd subject line]
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: avoid deprecated do_gettimeofday

All users of do_gettimeofday() have been removed, but this one recently
crept in, along with an incorrect printing of the microseconds portion.

This converts it to using ktime_get_real_timespec64() as a direct
replacement, and adds the leading zeroes. I considered using monotonic
times (ktime_get()) instead, but as this timestamp appears to only
be used for humans rather than compared with other timestamps, the
real time domain is probably good enough.

Fixes: e43b045e2c82 ("drm/msm/gpu: Capture the state of the GPU")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: dsi: Handle dual-channel for 6G as well

This fixes up a collision between introducing dual-channel support and
the dsi refactors. This patch applies the same dual-channel
considerations and pclk calculations to both v2 and 6G, with a bit of
abstracting for good measure.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: dpu: Use clock-names instead of assigned-clock-names

In these cases, we want to enumerate _all_ clocks, not just the ones
that are assigned a rate.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm: dpu: Use 'vsync' instead of 'vsync_clk' in cmdmode encoder

Should work with the legacy handling in of, but we shouldn't rely on
that.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: Add the buffer objects from the submit to the crash dump

For hangs, dump copy out the contents of the buffer objects attached to the
guilty submission and print them in the crash dump report.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/adreno: Add a5xx specific registers for the GPU state

HLSQ, SP and TP registers are only accessible from a special
aperture and to make matters worse the aperture is blocked from
the CPU on targets that can support secure rendering. Luckily the
GPU hardware has its own purpose built register dumper that can
access the registers from the aperture. Add a5xx specific code
to program the crashdumper and retrieve the wayward registers
and dump them for the crash state.

Also, remove a block of registers the regular CPU accessible
list that aren't useful for debug which helps reduce the size
of the crash state file by a goodly amount.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/adreno: Add ringbuffer data to the GPU state

Add the contents of each ringbuffer to the GPU state and dump the
data in the crash file encoded with ascii85. To save space only
the used portions of the ringbuffer are dumped.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/adreno: Convert the show/crash file format

Convert the format of the 'show' debugfs file and the crash
dump to a  format resembling YAML. This should be easier to
parse and be more flexible for future changes and expansions.

v2: Use a standard .rst for the msm crashdump documentation

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: Capture the GPU state on a GPU hang

Capture the GPU state on a GPU hang and store it for later playback
via the devcoredump facility. Only one crash state is stored at a
time on the assumption that the first hang is usually the most
interesting. The existing crash state can be cleared after capturing
it and then a new one will be captured on the next hang.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: Rearrange the code that collects the task during a hang

Do a bit of cleanup to prepare for upcoming changes to pass the
hanging task comm and cmdline to the crash dump function.

v2: Use GFP_ATOMIC while holding the rcu lock per Chris Wilson

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: Convert the GPU show function to use the GPU state

Convert the existing GPU show function to use the GPU state to
dump the information rather than reading it directly from the hardware.
This will require an additional step to capture the state before
dumping it for the existing nodes but it will greatly facilitate reusing
the same code for dumping a previously captured state from a GPU hang.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/msm/gpu: Capture the state of the GPU

Add the infrastructure to capture the current state of the GPU and
store it in memory so that it can be dumped later.

For now grab the same basic ringbuffer information and registers
that are provided by the debugfs 'gpu' node but obviously this should
be extended to capture a much larger set of GPU information.

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm: Add puts callback for the coredump printer

Add a puts function for the coredump printer to bypass printf()
for constant strings for a speed boost. Reorganize the
coredump printf callback to share as much code as possible.

v2: Try to reuse code between print and puts as suggested by
    Chris Wilson

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm: Add a -puts() function for the seq_file printer

Add a puts() function to use seq_puts() to help speed up
up print time for constant strings.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm: Add drm_puts() to complement drm_printf()

Add drm_puts() for a much faster path to print constant strings
into a drm_printer object with memcpy and friends. This can
have seconds off of really large outputs such as GPU dumps.

If the drm_printer object supports a custom puts function then
use that otherwise fall back to the slower legacy printf call.

v2: Add documentation for drm_puts() per Daniel Vetter

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
[robclark fix minor htmldocs warning]
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm: drm_printer: Add printer for devcoredump

Add a drm printer suitable for use with the read callback for
devcoredump or other suitable buffer based output format that
isn't otherwise covered by seq_file.

v2: Add improved documentation per Daniel Vetter

Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

include: Move ascii85 functions from i915 to linux/ascii85.h

The i915 DRM driver very cleverly used ascii85 encoding for their
GPU state file. Move the encode functions to a general header file to
support other drivers that might be interested in the same
functionality.

v4: Make the return value const char * as suggested by Chris Wilson
v3: Fix error_puts -> err_puts pointed out by the 01.org bot
v2: Update API to be cleaner for the caller as suggested by Chris Wilson

Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>

drm/armada: remove obsolete fb unreferencing kfifo and workqueue

Remove the obsolete fb unreferencing system that is no longer used
since we've transitioned to atomic modeset.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: remove unnecessary armada_plane structure

We no longer require a private armada_plane structure, so eliminate
it, and use the drm_plane structure directly.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: remove unnecessary armada_ovl_plane structure

We no longer need a private plane structure, so get rid of it.  Use the
drm_plane structure directly.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: update primary framebuffer parameters on mode change

The framebuffer base address and toggling mode needs to be updated
when the interlaced flag for mode changes is updated.  Arrange to
reprogram these parameters when only the mode has changed.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: update planes after the dumb frame is complete

Write out the plane updates after the dumb frame has completed, but
just before the blank period.  This allows all the plane updates to
be performed in a flicker-free non-tearing manner.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: switch overlay plane to atomic modeset

Switch the overlay plane away from the transitional helpers and legacy
methods, and use atomic helpers instead to implement the legacy
set_plane ioctl methods.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: switch primary plane to atomic modeset

Switch the primary plane away from the transitional helpers, and
use the atomic helpers instead to implement the legacy set_plane
ioctl call for this plane.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: switch legacy modeset to atomic modeset

Switch the legacy set_config() method to use the atomic modeset
helper, which allows us to get rid of the legacy dpms, prepare,
commit, mode_set, mode_set_base and disable helper methods.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: enable atomic modeset support

Enable atomic modeset helpers, and internal DRM use of atomic modeset
with armada-drm.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: implement atomic_enable()/atomic_disable() methods

Implement the atomic_enable()/atomic_disable() methods used by the
atomic modeset helpers.  atomic_disable() will need some transitional
code during conversion to ensure proper ordering is maintained.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: unhook dpms state from armada_drm_crtc_update()

Explicitly pass in the desired enable/disable state into
armada_drm_crtc_update() rather than having it use the DPMS state
stored in our crtc structure.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: push responsibility for clock management to backend

Push responsibility for managing the clock during DPMS down into the
variant backend, rather than the CRTC layer having knowledge of its
state.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: handle atomic modeset crtc events

Prepare handling for atomic modeset CRTC events.  Currently, using the
transition helpers, CRTC events do not exist, but once we switch to
proper atomic modeset, they have to be handled.

We queue an event for the next vblank in two places:
- armada_drm_crtc_atomic_flush() provided we aren't doing an
  atomic modeset.
- armada_drm_crtc_commit() if we are committing a modeset.

This ensures that the event is sent at the correct time (after all
updates have been written to the hardware and after the following
vblank.)

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

drm/armada: clean up SPU_ADV_REG

Rather than writing all bits of SPU_ADV_REG on modeset, only write
what we need to change, and initialise the register in the variant
initialisation.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>