drm/panfrost: Hold runtime PM reference until jobs complete

Doing a pm_runtime_put as soon as a job is submitted is wrong as it should
not happen until the job completes. It works currently because we are
relying on the autosuspend timeout to keep the h/w enabled.

Fixes: f3ba91228e8e ("drm/panfrost: Add initial panfrost driver")
Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-3-robh@kernel.org

drm/panfrost: Rework runtime PM initialization

There's a few issues with the runtime PM initialization.

The documentation states pm_runtime_set_active() should be called before
pm_runtime_enable(). The pm_runtime_put_autosuspend() could suspend the GPU
before panfrost_perfcnt_init() is called which touches the h/w. The
autosuspend delay keeps things from breaking. There's no need explicitly
power off the GPU only to wake back up with pm_runtime_get_sync(). Just
delaying pm_runtime_enable to the end of probe is sufficient.

Lets move all the runtime PM calls into the probe() function so they are
all in one place and are done after all initialization.

Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: Steven Price <steven.price@arm.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190826223317.28509-2-robh@kernel.org

drm/panfrost: Use mutex_trylock in panfrost_gem_purge

Lockdep reports a circular locking dependency with pages_lock taken in
the shrinker callback. The deadlock can't actually happen with current
users at least as a BO will never be purgeable when pages_lock is held.
To be safe, let's use mutex_trylock() instead and bail if a BO is locked
already.

Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-7-robh@kernel.org

drm/shmem: Use mutex_trylock in drm_gem_shmem_purge

Lockdep reports a circular locking dependency with pages_lock taken in
the shrinker callback. The deadlock can't actually happen with current
users at least as a BO will never be purgeable when pages_lock is held.
To be safe, let's use mutex_trylock() instead and bail if a BO is locked
already.

WARNING: possible circular locking dependency detected
5.3.0-rc1+ #100 Tainted: G             L
------------------------------------------------------
kswapd0/171 is trying to acquire lock:
000000009b9823fd (&shmem->pages_lock){+.+.}, at: drm_gem_shmem_purge+0x20/0x40

but task is already holding lock:
00000000f82369b6 (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x0/0x40

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (fs_reclaim){+.+.}:
       fs_reclaim_acquire.part.18+0x34/0x40
       fs_reclaim_acquire+0x20/0x28
       __kmalloc_node+0x6c/0x4c0
       kvmalloc_node+0x38/0xa8
       drm_gem_get_pages+0x80/0x1d0
       drm_gem_shmem_get_pages+0x58/0xa0
       drm_gem_shmem_get_pages_sgt+0x48/0xd0
       panfrost_mmu_map+0x38/0xf8 [panfrost]
       panfrost_gem_open+0xc0/0xe8 [panfrost]
       drm_gem_handle_create_tail+0xe8/0x198
       drm_gem_handle_create+0x3c/0x50
       panfrost_gem_create_with_handle+0x70/0xa0 [panfrost]
       panfrost_ioctl_create_bo+0x48/0x80 [panfrost]
       drm_ioctl_kernel+0xb8/0x110
       drm_ioctl+0x244/0x3f0
       do_vfs_ioctl+0xbc/0x910
       ksys_ioctl+0x78/0xa8
       __arm64_sys_ioctl+0x1c/0x28
       el0_svc_common.constprop.0+0x90/0x168
       el0_svc_handler+0x28/0x78
       el0_svc+0x8/0xc

-> #0 (&shmem->pages_lock){+.+.}:
       __lock_acquire+0xa2c/0x1d70
       lock_acquire+0xdc/0x228
       __mutex_lock+0x8c/0x800
       mutex_lock_nested+0x1c/0x28
       drm_gem_shmem_purge+0x20/0x40
       panfrost_gem_shrinker_scan+0xc0/0x180 [panfrost]
       do_shrink_slab+0x208/0x500
       shrink_slab+0x10c/0x2c0
       shrink_node+0x28c/0x4d8
       balance_pgdat+0x2c8/0x570
       kswapd+0x22c/0x638
       kthread+0x128/0x130
       ret_from_fork+0x10/0x18

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(fs_reclaim);
                               lock(&shmem->pages_lock);
                               lock(fs_reclaim);
  lock(&shmem->pages_lock);

 *** DEADLOCK ***

3 locks held by kswapd0/171:
 #0: 00000000f82369b6 (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x0/0x40
 #1: 00000000ceb37808 (shrinker_rwsem){++++}, at: shrink_slab+0xbc/0x2c0
 #2: 00000000f31efa81 (&pfdev->shrinker_lock){+.+.}, at: panfrost_gem_shrinker_scan+0x34/0x180 [panfrost]

Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>
Cc: Sean Paul <sean@poorly.run>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-6-robh@kernel.org

drm/shmem: Do dma_unmap_sg before purging pages

Calling dma_unmap_sg() in drm_gem_shmem_free_object() is too late if the
backing pages have already been released by the shrinker. The result is
the following abort:

Unable to handle kernel paging request at virtual address ffff8000098ed000
Mem abort info:
  ESR = 0x96000147
  Exception class = DABT (current EL), IL = 32 bits
  SET = 0, FnV = 0
  EA = 0, S1PTW = 0
Data abort info:
  ISV = 0, ISS = 0x00000147
  CM = 1, WnR = 1
swapper pgtable: 4k pages, 48-bit VAs, pgdp=0000000002f51000
[ffff8000098ed000] pgd=00000000401f8003, pud=00000000401f7003, pmd=00000000401b1003, pte=00e80000098ed712
Internal error: Oops: 96000147 [#1] SMP
Modules linked in: panfrost gpu_sched
CPU: 5 PID: 902 Comm: gnome-shell Not tainted 5.3.0-rc1+ #95
Hardware name: 96boards Rock960 (DT)
pstate: 40000005 (nZcv daif -PAN -UAO)
pc : __dma_inv_area+0x40/0x58
lr : arch_sync_dma_for_cpu+0x28/0x30
sp : ffff00001321ba30
x29: ffff00001321ba30 x28: ffff00001321bd08
x27: 0000000000000000 x26: 0000000000000009
x25: 0000ffffc1f86170 x24: 0000000000000000
x23: 0000000000000000 x22: 0000000000000000
x21: 0000000000021000 x20: ffff80003bb2d810
x19: 00000000098ed000 x18: 0000000000000000
x17: 0000000000000000 x16: ffff800023fd9480
x15: 0000000000000000 x14: 0000000000000000
x13: 0000000000000000 x12: 0000000000000000
x11: 00000000fffb9fff x10: 0000000000000000
x9 : 0000000000000000 x8 : ffff800023fd9c18
x7 : 0000000000000000 x6 : 00000000ffffffff
x5 : 0000000000000000 x4 : 0000000000021000
Purging 5693440 bytes
x3 : 000000000000003f x2 : 0000000000000040
x1 : ffff80000990e000 x0 : ffff8000098ed000
Call trace:
 __dma_inv_area+0x40/0x58
 dma_direct_sync_single_for_cpu+0x7c/0x80
 dma_direct_unmap_page+0x80/0x88
 dma_direct_unmap_sg+0x54/0x80
 drm_gem_shmem_free_object+0xfc/0x108
 panfrost_gem_free_object+0x118/0x128 [panfrost]
 drm_gem_object_free+0x18/0x90
 drm_gem_object_put_unlocked+0x58/0x80
 drm_gem_object_handle_put_unlocked+0x64/0xb0
 drm_gem_object_release_handle+0x70/0x98
 drm_gem_handle_delete+0x64/0xb0
 drm_gem_close_ioctl+0x28/0x38
 drm_ioctl_kernel+0xb8/0x110
 drm_ioctl+0x244/0x3f0
 do_vfs_ioctl+0xbc/0x910
 ksys_ioctl+0x78/0xa8
 __arm64_sys_ioctl+0x1c/0x28
 el0_svc_common.constprop.0+0x88/0x150
 el0_svc_handler+0x28/0x78
 el0_svc+0x8/0xc
 Code: 8a230000 54000060 d50b7e20 14000002 (d5087620)

Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers")
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>
Cc: Sean Paul <sean@poorly.run>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-5-robh@kernel.org

drm/panfrost: Fix possible suspend in panfrost_remove

Calls to panfrost_device_fini() access the h/w, but we already done a
pm_runtime_put_sync_autosuspend() beforehand. This only works if the
autosuspend delay is long enough. A 0ms delay will hang the system when
removing the device. Fix this by moving the pm_runtime_put_sync_suspend()
after the panfrost_device_fini() call.

Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Rob Herring <robh@kernel.org>
Reviewed-by: Steven Price <steven.price@arm.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823021216.5862-2-robh@kernel.org

drm/panfrost: Add missing check for pfdev->regulator

When modifying panfrost_devfreq_target() to support a device without a
regulator defined I missed the check on the error path. Let's add it.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: e21dd290881b ("drm/panfrost: Enable devfreq to work without regulator")
Signed-off-by: Steven Price <steven.price@arm.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20190822093218.26014-1-steven.price@arm.com

Merge tag 'drm-next-5.4-2019-08-23' of git://people.freedesktop.org/~agd5f/linux into drm-next

drm-next-5.4-2019-08-23:

amdgpu:
- Enable power features on Navi12
- Enable power features on Arcturus
- RAS updates
- Initial Renoir APU support
- Enable power featyres on Renoir
- DC gamma fixes
- DCN2 fixes
- GPU reset support for Picasso
- Misc cleanups and fixes

scheduler:
- Possible race fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823202620.3870-1-alexander.deucher@amd.com

Merge tag 'drm-misc-next-2019-08-23' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

drm-misc-next for 5.4:

UAPI Changes:

Cross-subsystem Changes:

Core Changes:
  - dma-buf: dma-fence selftests

Driver Changes:
  - kirin: Various cleanups and reworks
  - komeda: Add support for DT memory-regions
  - meson: Rely on the compatible to detect vpu features
  - omap: Implement alpha and pixel blend mode properties
  - panfrost: Implement per-fd address spaces, various fixes
  - rockchip: DSI DT binding rework
  - fbdev: Various cleanups

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <maxime.ripard@bootlin.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823083509.c7mduqdqjnxc7ubb@flea

Merge tag 'drm-hisilicon-hibmc-next-2019-08-26' of https://github.com/xin3liang/linux into drm-next

Three small cleanup and fix patches for 5.4 hisilicon hibmc driver.
I have tested and verified on taishan 2280v1/v2 machines.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: xinliang <z.liuxinliang@hisilicon.com>
Link: https://patchwork.freedesktop.org/patch/msgid/5D63A271.7080400@hisilicon.com

Merge tag 'imx-drm-next-2019-08-23' of git://git.pengutronix.de/pza/linux into drm-next

drm/imx: IPUv3 image converter fixes and improvements

Fix image converter seam handling for 1024x1024 pixel hardware
limitation at the main processing section input, improve error
handling, and slightly optimize for 1:1 conversions.
Add support for newly defined 32-bit RGB V4L2 pixel formats.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Philipp Zabel <p.zabel@pengutronix.de>
Link: https://patchwork.freedesktop.org/patch/msgid/1566573659.23587.2.camel@pengutronix.de

Merge tag 'drm-intel-next-2019-08-22' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

- More TGL enabling work (Michel, Jose, Lucas)
- Fixes on DP MST (Ville)
- More GTT and Execlists fixes and improvements (Chris)
- Code style clean-up on hdmi and dp side (Jani)
- Fix null pointer dereferrence (Xiong)
- Fix a couple of missing serialization on selftests (Chris)
- More vm locking rework (Chris)

drm-intel-next-2019-08-20:
- GuC and HuC related fixes and improvements (Daniele, Michal)
- Improve debug with more engine information and rework on debugfs files (Chris, Stuart)
- Simplify appearture address handling (Chris)
- Other fixes and cleanups around engines and execlists (Chris)
- Selftests fixes (Matt, Chris)
- Gen11 cache flush related fixes and improvements (Mika)
- More work around requests, timelines and locks to allow removal of struct_mutex (Chris)
- Add missing CML PCI ID (Anusha)
- More work on the new i915 buddy allocator (Matt)
- More headers, files and directories reorg (Daniele)
- Improvements on ggtt’s get pdp (Mika)
- Fix GPU reset (Chris)
- Fix GPIO pins on gen11 (Matt)
- Fix HW readout for crtc_clock in HDMI mode (Imre)
- Sanitize display Phy during unitit to workaround messages of HW state change during suspend (Imre)
- Be defensive when starting vma activity (Chris)
- More Tiger Lake enabling work (Michel, Daniele, Lucas)
- Relax pd_used assertion (Chris)

drm-intel-next-2019-08-13:
- More Tiger Lake enabling work (Lucas, Jose, Tomasz, Michel, Jordan, Anusha, Vandita)
- More selftest organization reworks, fixes and improvements (Lucas, Chris)
- Simplifications on GEM code like context and cleanup_early (Chris, Daniele)
- GuC and HuC related fixes and improvements (Daniele, Michal, Chris)
- Some clean up and fixes on headers, Makefile, and generated files (Lucas, Jani)
- MOCS setup clean up (Tvrtko)
- More Elkhartlake enabling work (Jose, Matt)
- Fix engine reset by clearing in flight execlists requests (Chris)
- Fix possible memory leak on intel_hdcp_auth_downstream (Wei)
- Introduce intel_gt_runtime_suspend/resume (Daniele)
- PMU improvements (Tvrtko)
- Flush extra hard after writing relocations through the GTT (Chris)
- Documentations fixes (Michal, Chris)
- Report dma_reserv allocation failure (Chris)
- Improvements around shrinker (Chris)
- More improvements around engine handling (Chris)
- Also more s/dev_priv/i915 (Chris)
- Abstract display suspend/resume operations (Rodrigo/Jani)
- Drop VM_IO from GTT mappings (Chris)
- Fix some NULL vs IS_ERR conditions (Dan)
- General improvements on error state (Chris)
- Isolate i915_getparam_iocrtl to its own file (Chris)
- Perf OA object refactor (Umesh)
- Ignore central i915->kernel_context and allocate it directly (Chris)
- More fixes and improvements around wakerefs (Chris)
- Clean-up and improvements around debugfs (Chris)
- Free the imported shmemfs file for phys objects (Chris)
- Many other fix and cleanups around engines and execlists (Chris)
- Split out uncore_mmio_debug (Daniele)
- Memory management fixes for blk and gtt (Matt)
- Introduction of buddy allocator to handle huge-pages for GTT (Matt)
- Fix ICL and TGL PG3 power domains (Anshuman)
- Extract GT IRQ to gt/ (Andi)
- Drop last_fence tracking in favor of whole vma->active (Chris)
- Make overlay to use i915_active instead of i915_active_request (Chris)
- Move misc display IRQ handling to its own function (Jose)
- Introduce new _TRANS2() macro in preparation for some coming PSR related work (Jose)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190823051435.GA23885@intel.com

drm/hisilicon/hibmc: Make CONFIG_DRM_HISI_HIBMC depend on ARM64

Hisilicon developed hibmc_drm for their arm64 based soc and did not
intend for this driver to be used on any other architecture than arm64.

Using it on amd64 leads to incorrect video modes being used, making
the screen unreadable, forcing users to manually blacklist the module
on the kernel command line to use the d-i server installer or any
graphical sessions.

Make CONFIG_DRM_HISI_HIBMC firmly depend on ARM64 to ensure it is not
built for other architectures.

Signed-off-by: Matthew Ruffell <matthew.ruffell@canonical.com>
Signed-off-by: Xinliang Liu <z.liuxinliang@hisilicon.com>

drm/hisilicon: Use dev_get_drvdata

Instead of using to_pci_dev + pci_get_drvdata,
use dev_get_drvdata to make code simpler.

Signed-off-by: Chuhong Yuan <hslester96@gmail.com>
Signed-off-by: Xinliang Liu <z.liuxinliang@hisilicon.com>

drm/hisilicon/hibmc: Using module_pci_driver.

Remove boilerplate code by using macro module_pci_driver.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Xinliang Liu <z.liuxinliang@hisilicon.com>
Signed-off-by: Xinliang Liu <z.liuxinliang@hisilicon.com>

drm/amd/display: 3.2.48

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: flicking observed while installing driver on Navi10 CF

[WHY] value of dchub_ref_clock is decided by dchubbub global timer
settings which is programmed by vbios command table disp_init.
for multi-GPU case, vbios is posted only for primary GPU. without
vbios posted for the secondary GPU, value of dchub_ref_clock is not
set properly. this value will affect dcn bandwidth calcuation and
cause underflow. user will see screen flicking during driver
installation for dual GPU case.

[HOW] dc init_hw always call vbios command table disp_init to
make sure dchubbub global timer is configured and enable.

Signed-off-by: hersen wu <hersenxs.wu@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Properly read LVTMA_PWRSEQ_CNTL

[Why]
The register LVTMA_PWRSEQ_CNTL is used to determine the power state of the
embedded display. Currently we do not actually read this register's values,
so during power down we think that this display is already off, so we skip
calling into VBIOS to actually turn it off.

[How]
 - add relevant fields to shift / mask initialization

Signed-off-by: Joshua Aberback <joshua.aberback@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: revert wait in pipelock

[why]
Previous workaround to prevent a vsync flip to be converted
to immediate flip is no longer needed, and is risky because
there are cases where it can result in infinite loop.

[how]
Remove wait loop (which is potentially infinite) before locking
pipe

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Use res_cap to acquire i2c instead of pipe count

[Why]
We should be using the ddc_num from res_caps. As the
pipe count != number of i2c resources.

[How]
Use ddc_num from res_cap instead of pipe count.

Signed-off-by: Derek Lai <Derek.Lai@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Expose OTG_V_TOTAL_MID for HW Diags

[Why]
Existing HW Features, HW Diags test requested that the
registers be exposed.

[How]
Add V_TOTAL_MID to existing DC structures.
Make sure values are passed down throughout DC
Add Register definition.
Program the additional registers
Add additional Logic for V_TOTAL_CONTROL.

Signed-off-by: Bayan Zabihiyan <bayan.zabihiyan@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix DML not calculating delivery time

[why]
Calculating DCFCLK DS time requires calculating
delivery time for luma/chroma, but this value is
not calculated in DMLv2, it was inadvertently
removed when porting DMLv2

[how]
Add the calculation back

Signed-off-by: Jun Lei <Jun.Lei@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: MST topology debugfs

DRM provides drm_dp_mst_dump_topology, which prints
useful information about MST devices

Hook this up to a debugfs file named amdgpu_mst_topology

Signed-off-by: David Francis <David.Francis@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set av_mute in hw_init for HDMI

[Description]
OS will reserve HW state in UEFI mode.
Driver init_hw reset to RGB which caused HDMI green in YCbCr mode.
read HW blank_color based on acc_mode.

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add Logging for Gamma Related information

[Why]
A recent bug showed that logging would be useful in debugging
various gamma issues.

[How]
Add logging in dc.
Fix formatting for easier graphing.

Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Enable HW rotation

[Why] HW rotation is not enabled. Calculations for cursor rotation
are wrong for the values passed to set_cursor_position.

[How] Swap Src rect and height and vertically mirror surface for
the correct surface rotation direction. Cursor position is rotated
according to angle. Offset calculations are tweaked for non-rotated
cursor hotspot and width/height.

Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add Cursor Degamma logic for DCN2

[Why]
We need to have the ability to to tell us set degamma on the cursor.

[How]
Pass a flag down to register programming that tells us if the
current surface format needs cursor degamma.

Signed-off-by: Bayan Zabihiyan <bayan.zabihiyan@amd.com>
Reviewed-by: Krunoslav Kovac <Krunoslav.Kovac@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm validation

Update bw validation to use prev and next odm pipe pointers
for populating dml inputs.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm stream release

Need to memset all odm pipes when calling dc_remove_stream_from_ctx

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dcn20 odm dpp programming

dcn20 requires special casing for odm.
This change treats odm as alternative to mpc tree on dcn20.

This is planned to be fixed in a future refactor

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Fix number of slices not being checked for dsc

[why]
num_slices_h was not being checked

[How]
Fix the typo and check num_slices_h

Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Harry Wentland <Harry.Wentland@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix odm pipe copy

ODM next and prev pipe were missing from dc_copy_state

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Nikola Cornij <Nikola.Cornij@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix trigger not generated for freesync

[Why]
In newer hardware MANUAL_FLOW_CONTROL is not a trigger bit. Due to this
front porch is fixed and in these hardware freesync does not work.

[How]
Change the programming to generate a pulse so that the event will be
triggered, front porch will be cut short and freesync will work.

Signed-off-by: Yogesh Mohan Marimuthu <yogesh.mohanmarimuthu@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix stuck test pattern on right half of display

[why]
With visual confirm enabled, displays where ODM combine is enabled
has a test pattern stuck on the right half of the display even
though the display is unblanked.

[how]
Add a condition to not show the colour ramp test pattern when the
display is unblanked.

Signed-off-by: Zi Yu Liao <ziyu.liao@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: support spdif

[Description]
port spdif fix to staging:
 spdif hardwired to afmt inst 1.
 spdif func pointer
 spdif resource allocation (reserve last audio endpoint for spdif only)

Signed-off-by: Charlene Liu <charlene.liu@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Refactoring VTEM

[Why]
Video Timing Extended Metadata packet (VTEM) is not
specific to freesync. So move it out of freesync module

[How]
- Moved VTEM from freesync module to info_packet module
- Created new structure for VTEM parameters that can be used for VRR
and FVA

Signed-off-by: Ahmad Othman <ahmad.othman@amd.com>
Reviewed-by: Chris Park <Chris.Park@amd.com>
Acked-by: Ahmad Othman <Ahmad.Othman@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: 3.2.47

Signed-off-by: Anthony Koo <Anthony.Koo@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: remove unused function

[Why]
This function is not being used, it was left in
when introducing DCN2

[How]
Remove the function

Signed-off-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Reviewed-by: Eric Bernstein <Eric.Bernstein@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: cleaned up coding error in init_hw

[why]
during a refactor a redundant code that has unknown behaviour was added.

[how]
removed old bad code

Fixes: 8a31820b1218 ("drm/amd/display: Make init_hw and init_pipes generic for seamless boot")
Signed-off-by: Martin Leung <martin.leung@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Add VM page fault handle implementation

[How] Allocate memory for default page and program memory block addr
into default page addr register.

Signed-off-by: Jaehyun Chung <jaehyun.chung@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: refactor Device ID for external chips

IEEE OUI will now be used while referring to certain vendors.
instead of normal index

Signed-off-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Zero-out dsc init regs

[why]
Before a statically allocated PPS data structure, that did
get zeroed-out at startup, had been re-used for making packed PPS
SDP. With S3 fix, using a non-initialized PPS data structure was
introduced, while wrongly assuming it'd get initialized before it's
populated. As a consequence 'vbr_enable' and perhaps some other
fields are left uninitialized when making packed PPS SDP. This can
affect 'simple_422' as well because of the way PPS SDP packing is
done (the fields are not masked first, only shifted). The behavior
will be different, depending on the content of uninitialized data.

[how]
Zero-out PPS data structure at initialization time before it's
populated

Fixes: 1a9e3d4569fc ("drm/amd/display: Set DSC before DIG front-end is connected to its back-end")
Signed-off-by: Nikola Cornij <nikola.cornij@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Implement voltage limitation stub

add new function to get the voltage at the end of
dcn_validate_bandwidth, to check against the
highest voltage we allow.

Created a stub to allow for optimizations

Signed-off-by: Joseph Gravenor <joseph.gravenor@amd.com>
Reviewed-by: Eric Yang <eric.yang2@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Sun peng Li <Sunpeng.Li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: add null checks before logging

Adding NULL checks to various parameters in log_tf, to avoid
nullptr errors

Signed-off-by: Wyatt Wood <wyatt.wood@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Acked-by: Nikola Cornij <Nikola.Cornij@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: set Hratio and VRatio in dml

Set the writeback Hratio and Vratio in dml.

Signed-off-by: Ilya Bakoulin <Ilya.Bakoulin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix dp stream enable

A previous odm change broke stream enable by always setting
n_multiply as if odm was on.

This fixes the check for odm by making sure opp count is >1
rather than not 0.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: load iram for abm 2.3

[Why]
ABM 2.3 firmware expects information in iRAM that differs from previous
versions of ABM, so a mechanism is required to provide it with that
information.

[How]
Extend the existing iRAM definition to include parameters added by
ABM 2.3, and load it if DMCU is running ABM 2.3.

Signed-off-by: Josip Pavic <Josip.Pavic@amd.com>
Reviewed-by: Anthony Koo <Anthony.Koo@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix MPO HUBP underflow with Scatter Gather

[why]
With Scatter Gather enabled, HUBP underflows during MPO enabled video
playback. hubp_init has a register write that fixes this problem, but
the register is cleared when HUBP gets power gated.

[how]
Make a call to hubp_init during enable_plane, so that the fix can
be applied after HUBP powers back on again.

Signed-off-by: Zi Yu Liao <ziyu.liao@amd.com>
Reviewed-by: Tony Cheng <Tony.Cheng@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: fix audio endpoint not getting disabled issue

[Why]
Disable_audio_stream gets enum option as a paramenter which will decide
if we free acquired resources or not. However checks for the option is
guarded by the other condition which check if audio stream is getting
diabled more than once. With both conditions combined, if we attempt to
disable audio stream twice in a row, first with keep and second with
free as an option, we will never free any resources, which will make
system think there is audio endpoint connected even after we plug out
the device

[How]
Get rid of option as parameter to disable_audio_stream and move the part
of the code that free acquired resources to outside where to keep or to
free resources is actually determined

Signed-off-by: Su Sung Chung <Su.Chung@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: Delete dead code in command_table_helper

[Why]
dig_encoder_sel_to_atom will always return zero on any ASIC version
past DCE80 since programming of the FE selection is handled by
driver, but the translation code was left in the function, making
it look like a coding error.

[How]
Remove code that has no effect, and replace with a comment describing
why it returns zero.

Signed-off-by: Julian Parkin <julian.parkin@amd.com>
Reviewed-by: Wenjing Liu <Wenjing.Liu@amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: re structure odm to allow 4 to 1 support

Currently odm is handled using top_bottom pipe by special casing
the differing opps to differentiate from mpc combine.

Since top/bottom pipe list was made to track mpc muxing this creates
difficulties in adding a 4 pipe odm case support.

Rather than continue using mpc combine list, this change reworks odm
to use it's own linked list to keep track of odm combine pipes. This
also opens up options for using mpo with odm, if a practical use case
is ever found.

Signed-off-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Reviewed-by: Charlene Liu <Charlene.Liu@amd.com>
Acked-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: set adev->num_vmhubs for gmc6,7,8

So that we properly handle them on older asics.

Fixes: 3ff985485b29 ("drm/amdgpu: Export function to flush TLB of specific vm hub")
Tested-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Xiaojie Yuan <xiaojie.yuan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/display: add flag for multi-display mclk switching

Add a dcfeaturemask flag for mclk switching.  Disable by default;
enable once the feature has seen more testing.

Set amdgpu.dcfeaturemask=2 on the kernel command line in grub
to enable this.

Acked-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/display: update bw_calcs to take pipe sync into account (v3)

Properly set all_displays_in_sync so that when the data is
propagated to powerplay, it's set properly and we can enable
mclk switching when all monitors are in sync.

v2: fix logic, clean up
v3: check for blending chains, simplify logic

Acked-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/powerplay/vega10: enable mclk switching if monitors are synced

If DC has synced the displays, we can enable mclk switching to
save power.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/powerplay/smu7: enable mclk switching if monitors are synced

If DC has synced the displays, we can enable mclk switching to
save power.

Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/powerplay: Fix Vega20 power reading again

For the 40.46 SMU release, they changed CurrSocketPower to
AverageSocketPower, but this was changed back in 40.47 so just check if
it's 40.46 and make the appropriate change

Tested with 40.45, 40.46 and 40.47 successfully

Signed-off-by: Kent Russell <kent.russell@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

amd/amdkfd: add Arcturus vf DID support

Add the virtual function PCI device id.

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

drm/amd/powerpaly: fix navi series custom peak level value error

fix other navi asic set peak performance level error.
because the navi10_ppt.c will handle navi12 14 asic,
it will use navi10 peak value to set other asic, it is not correct.

after patch:
only navi10 use custom peak value, other asic will used default value.

Signed-off-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: correct ras error count type

Use unsigned long type for the same ras count variable.
This will avoid overflow on 64 bit system.

Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

omapdrm: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Sebastian Reichel <sebastian.reichel@collabora.com>
Cc: Jyri Sarha <jsarha@ti.com>
Cc: Tony Lindgren <tony@atomide.com>
Cc: zhong jiang <zhongjiang@huawei.com>
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190613115749.GC26335@kroah.com
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190704023557.4551-1-huangfq.daxian@gmail.com

drm/omap: Add 'alpha' and 'pixel blend mode' plane properties

Add the following properties for planes:
* alpha
* pixel blend mode. Only "Pre-multiplied" and "Coverage" are supported

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190711135219.23402-1-jjhiblot@ti.com

Merge branch 'linux-5.4' of git://github.com/skeggsb/linux into drm-next

This is mostly just the stuff I missed last round.  Various cleanup
patches + fixes, improvements to display colour management, and some
code to avoid loading when power cables aren't properly attached.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Ben Skeggs <skeggsb@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CACAvsv7hqj9_VHq+YiGL8Z8XsU2vPbqbNPC=LeN1Rb0XxMQypQ@mail.gmail.com

drm/nouveau/volt: Fix for some cards having 0 maximum voltage

Some, mostly Fermi, vbioses appear to have zero max voltage. That causes Nouveau to not parse voltage entries, thus users not being able to set higher clocks.

When changing this value Nvidia driver still appeared to ignore it, and I wasn't able to find out why, thus the code is ignoring the value if it is zero.

CC: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Mark Menzynski <mmenzyns@redhat.com>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dispnv50: Fix runtime PM ref tracking for non-blocking modesets

This is something that got noticed a while ago back when I was fixing a
large number of runtime PM related issues in nouveau, but never got
fixed:

https://patchwork.freedesktop.org/series/46815/#rev7

It's not safe to iterate the entire list of CRTCs in
nv50_disp_atomic_commit(), as we could be doing a non-blocking modeset
on one CRTC in parallel with one or more other CRTCs. Likewise, this
means it's also not safe to do so in order to track runtime PM state.
While this code is certainly wrong, so far the only issues I've seen
this cause in the wild is the occasional PM ref unbalance after an
atomic check failure + module reloading (since the PCI device will
outlive nouveau in such scenarios).

So, do this far more elegantly: grab a runtime PM ref across the modeset
and commit tail, then grab/put references for each CRTC enable/disable.
This also ends up being much simpler then the previous broken solution
we had.

Finally, since we've removed all it's users: get rid of
nouveau_drm->have_disp_power_ref.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dispnv04: Remove runtime PM

Originally when trying to fix the issue of runtime PM references with
non-blocking CRTCs on nv50, I ended up stumbling on this code when
trying to remove nouveau_drm->have_disp_power_ref, and attempted to fix
it to remove the dependency on have_disp_power_ref. However, Ilia Mirkin
pointed out that this code is actually completely useless, as pre-nv50
never had runtime PM support in the first place! Go figure.

So, since it's useless just get rid of it. Note that since the only
thing nouveau_crtc_set_config() was doing was grabbing a runtime PM ref,
calling drm_crtc_helper_set_config() then dropping the ref; we can just
remove the function entirely and just call drm_crtc_helper_set_config()
directly.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gpio: check function 76 in the power check as well

Added GPIO is "Power Alert". It's uncertain if this
GPIO is set on GPU initialization or only if a change is detected by the
GPU at runtime.

This GPIO can be found on Tesla and sometimes on Fermi GPUs.

Untested, wrote according to documentation.

Signed-off-by: Mark Menzynski <mmenzyns@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gpio: check the gpio function 16 in the power check as well

Added GPIO is "Thermal and External Power Detect". It's uncertain if this
GPIO is set on GPU initialization or only if a change is detected by the
GPU at runtime.

This GPIO can be found in Rankine and Curie and rarely on Tesla GPUs
VBIOS.

Untested, wrote according to documentation.

Signed-off-by: Mark Menzynski <mmenzyns@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gpio: fail if gpu external power is missing

Currently, nouveau doesn't check if GPU is missing power. This
patch makes nouveau fail when this happens on latest GPUs.

It checks GPIO function 121 (External Power Emergency), which
should detect power problems on GPU initialization.

This can be disabled with nouveau.config=NvPowerChecks=1

Tested on TU104, GP106 and GF100.

v3:
*  Add config override for disabling power checks

Signed-off-by: Mark Menzynski <mmenzyns@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/gpio: sort gpios by values

One gpio was in wrong place, moved it for better readability.

Signed-off-by: Mark Menzynski <mmenzyns@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/therm: don't attempt fan control where PMU is already managing it

There's already a condition in place which attempts to detect this, but
since we've begun to require a PMU subdev even on boards where we don't
load a custom FW, it's become inaccurate.

This will prevent unnecessarily running a periodic fan update thread on
GP100 and newer, where we don't yet override the default PMU FW.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/therm: skip probing for devices not specified in thermal tables

Saves some time during driver load, as described by the relevant section[1]
of the DCB 4.x specification.

[1] https://nvidia.github.io/open-gpu-doc/DCB/DCB-4.x-Specification.html#_i2c_device_table

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100-: attach pixel blend mode property to planes

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100-: attach alpha property to planes

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100-: add support for plane zpos property

Has a nice side-effect that we only update HW for this when it changes now,
rather than every time we do a page flip.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: attach immutable zpos property to planes

Defaulting to the fixed layout enforced in HW by EVO, and that we
currently use by default on NVD.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: create primary plane before overlay planes

zpos normalisation uses plane id to determine ordering for duplicate zpos
values, and we likely want to keep primary plane on the bottom here.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: use __drm_atomic_helper_plane_reset()

We have some of this open-coded already, use the helper to prevent problems
when adding (for example) support for the alpha property.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100-: implement csc + enable modern colour managment properties

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100-: use premultiplied alpha blending between planes

This is apparently the assumed default behaviour when blend properties
are absent.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: enable modern color management properties

For GF119:GV100, we can enable DEGAMMA/CTM/GAMMA. For earlier GPUs, as
there is no CTM, having both degamma and gamma is a bit pointless. Later
GPUs currently lack an implementation.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gf119-: add ctm property support

This adds support on GF119:GV100 (exclusive) for CTM (aka CSC).

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>

drm/nouveau/kms/nv50-: remove overlay alpha formats

The overlay logic can only do colorkey-based selection, not
alpha-blending.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: fix parsing of mmu fault data

Pascal was particularly incorrect, as the register changed to be more in the
same format as the MMU fault buffers are.

Shouldn't have impacted much more than confusing MMU fault log messages.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gf1xx: convert to using nvkm_fault_data

Would like to be able to reuse gf100_fifo_intr_fault() for (some of) the
later chipsets too, as it's identical.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/init: handle INIT_RESET_END devinit opcode

Signal that the reset sequence has completed.

This opcode signals that the software reset sequence has completed.
Ordinarily, no actual operations are performed by the opcode.
However it allows for possible software work arounds by devinit
engines in software agents other than the VBIOS, such as the resman,
FCODE, and EFI driver.

Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/init: handle INIT_RESET_BEGUN devinit opcode

Signal that the reset sequence has begun.

This opcode signals that the software reset sequence has begun.
Ordinarily, no actual operations are performed by the opcode.
However it allows for possible software work arounds by devinit
engines in software agents other than the VBIOS, such as the resman,
FCODE, and EFI driver.

Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios: downgrade absence of tmds table to info from an error

Absence of a TMDS Info Table is common on Optimus setups where the NVIDIA
gpu is not connected directly to any outputs.

Reporting an error in this scenario is too harsh. Accordingly, change the
error message to an info message.

By default the error message also causes a boot flicker for these sytems.

Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: Disable atomic support on a per-device basis

We now have per-device driver_features, so let's use that
to disable atomic only for pre-nv50.

Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: Lyude Paul <lyude@redhat.com>
Cc: nouveau@lists.freedesktop.org
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: add fp16 scanout support

Older hardware seems to want 0..1024 values, while new hardware takes
0..1 values. We set the gain to 1024 for the earlier display classes.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/nv50-: disable input lut harder

Under some circumstances, it could be left enabled when it shouldn't be.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/tu102-: disable input lut when input is already FP16

On Turing, an input LUT is required to transform inputs in fixed-point
formats to FP16 for the internal display pipe.  We provide an identity
mapping whenever a window is enabled for this reason.

HW has error checks to ensure when the input is already FP16, that the
input LUT is also disabled.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/kms/gv100: allow windows to use PACKED8BPP formats

Required for upcoming FP16 scanout support.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: fix nvif/device.h is included more than once

remove duplicate inclusion of nvif/device.h

Issue identified by includecheck

Signed-off-by: Hariprasad Kelam <hariprasad.kelam@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/dispnv04: subdev/bios.h is included more than once

remove duplicate inclusion of subdev/bios.h

Issue identified by includecheck

Signed-off-by: Hariprasad Kelam <hariprasad.kelam@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/mmu: use struct_size() helper

Make use of the struct_size() helper instead of an open-coded version
in order to avoid any potential type mistakes, in particular in the
context in which this code is being used.

So, replace the following form:

sizeof(*kind) + sizeof(*kind->data) * mmu->kind_nr;

with:

struct_size(kind, data, mmu->kind_nr)

This code was detected with the help of Coccinelle.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: drop use of drmp.h

Drop use of the deprecated drmP.h file from drm/nouveau.

Build tested using allyesconfig and allmodconfig.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: nouveau@lists.freedesktop.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: drop drmP.h from all header files

Drop include of the deprecated drmP.h from all nouveau heder files.
This allows us to remove drmP.h from all .c files without any
side-effects in a follow-up commit.

Build tested using allyeyconfig and allmodconfig

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: nouveau@lists.freedesktop.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: drop drmP.h from nouveau_drv.h

Drop the deprecated drmP.h header from nouveau_drv.h.
Fix fallout in other parts of the driver.

Build tested using allmodconfig and allyesconfig.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: nouveau@lists.freedesktop.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: drop use of DRM_UDELAY

The DRM_UDELAY is a simple wrapper for udealy() and to be consistent
call udelay() direct like in may other places.
This avoids the need to pull in drm_os_linux.h when we later
drop drmP.h uses in nouveau.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Ben Skeggs <bskeggs@redhat.com>
Cc: nouveau@lists.freedesktop.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>