drm/i915/bxt: Add MST support when do DPLL calculation

Add the missing INTEL_OUTPUT_DP_MST case in bxt_get_dpll()
to correctly initialize the crtc_state and port plls when
link training a DP MST monitor on BXT/APL devices.

Fixes: a277ca7dc01d ("drm/i915: Split bxt_ddi_pll_select()")
Bugs: https://bugs.freedesktop.org/show_bug.cgi?id=99572
Reviewed-by: Cooper Chiou <cooper.chiou@intel.com>
Reviewed-by: Gary C Wang <gary.c.wang@intel.com>
Reviewed-by: Ciobanu, Nathan D <nathan.d.ciobanu@intel.com>
Reviewed-by: Herbert, Marc <marc.herbert@intel.com>
Reviewed-by: Bride, Jim <jim.bride@intel.com>
Reviewed-by: Navare, Manasi D <manasi.d.navare@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: <stable@vger.kernel.org> # v4.9+
Signed-off-by: Lee, Shawn C <shawn.c.lee@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1486096329-6255-1-git-send-email-shawn.c.lee@intel.com

drm/i915: remove 512GB allocation warning

Now that we have selftests in place exercising truly huge allocations
we will start to hit the 512GB warning, so now seems like a good time to
remove this user-triggerable WARN.

Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1486047300-13198-1-git-send-email-matthew.auld@intel.com
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: Recreate internal objects with single page segments if dmar fails

If we fail to dma-map the object, the most common cause is lack of space
inside the SW-IOTLB due to fragmentation. If we recreate the_sg_table
using segments of PAGE_SIZE (and single page allocations), we may succeed
in remapping the scatterlist.

First became a significant problem for the mock selftests after commit
5584f1b1d73e ("drm/i915: fix i915 running as dom0 under Xen") increased
the max_order.

Fixes: 920cf4194954 ("drm/i915: Introduce an internal allocator for disposable private objects")
Fixes: 5584f1b1d73e ("drm/i915: fix i915 running as dom0 under Xen")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170202132721.12711-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.10-rc1+

drm/i915: don't warn about Skylake CPU - KabyPoint PCH combo

Apparently there are machines out there with Skylake CPU and KabyPoint
PCH. Judging from our driver code, there doesn't seem to be any code
paths that would do anything different between SunrisePoint and
KabyPoint PCHs, so it would seem okay to accept the combo without
warnings.

Fixes: 22dea0be50b2 ("drm/i915: Introduce Kabypoint PCH for Kabylake H/DT.")
References: https://lists.freedesktop.org/archives/intel-gfx/2017-February/118611.html
Reported-by: Rainer Koenig <Rainer.Koenig@ts.fujitsu.com>
Cc: Rainer Koenig <Rainer.Koenig@ts.fujitsu.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: <stable@vger.kernel.org> # v4.8+
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485956769-26015-1-git-send-email-jani.nikula@intel.com

drm/i915: Build DRM range manager selftests for CI

Build the struct drm_mm selftests so that we can trivially run them
within our CI.

"Enable debug, become developer." - Joonas Lahtinen

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170125101102.9010-3-chris@chris-wilson.co.uk

drm/i915: fix i915 running as dom0 under Xen

Commit 920cf4194954ec ("drm/i915: Introduce an internal allocator for
disposable private objects") introduced a regression for the kernel
running as Xen dom0: when switching to graphics mode a GPU HANG
occurred.

Reason seems to be a missing adaption similar to that done in
commit 7453c549f5f648 ("swiotlb: Export swiotlb_max_segment to users")
to i915_gem_object_get_pages_internal().

So limit the maximum page order to be used according to the maximum
swiotlb segment size instead to the complete swiotlb size.

Fixes: 920cf4194954 ("drm/i915: Introduce an internal allocator for disposable private objects")
Signed-off-by: Juergen Gross <jgross@suse.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170202094711.939-1-jgross@suse.com
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.10-rc1+
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: Fix POWER_DOMAIN_AUDIO refcounting.

If the crtc was brought up with audio before the driver loads,
then crtc_disable will remove a refcount to audio that doesn't exist
before.

Fortunately we already set power domains on readout, so we can just add
the power domain handling to get_crtc_power_domains, which will update
the power domains correctly in all cases.

This was found when testing module reload on CI with the crtc enabled,
which resulted in the following warn after module reload + modeset:

[   24.197041] ------------[ cut here ]------------
[   24.197075] WARNING: CPU: 0 PID: 99 at drivers/gpu/drm/i915/intel_runtime_pm.c:1790 intel_display_power_put+0x134/0x140 [i915]
[   24.197076] Use count on domain AUDIO is already zero
[   24.197098] CPU: 0 PID: 99 Comm: kworker/u8:2 Not tainted 4.9.0-CI-Trybot_393+ #1
[   24.197099] Hardware name:                  /NUC6i5SYB, BIOS SYSKLi35.86A.0042.2016.0409.1246 04/09/2016
[   24.197102] Workqueue: events_unbound async_run_entry_fn
[   24.197105]  ffffc900003c7688 ffffffff81435b35 ffffc900003c76d8 0000000000000000
[   24.197107]  ffffc900003c76c8 ffffffff8107e4d6 000006fe5dc36f28 ffff88025dc30054
[   24.197109]  ffff88025dc36f28 ffff88025dc30000 ffff88025dc30000 0000000000000015
[   24.197110] Call Trace:
[   24.197113]  [<ffffffff81435b35>] dump_stack+0x67/0x92
[   24.197116]  [<ffffffff8107e4d6>] __warn+0xc6/0xe0
[   24.197118]  [<ffffffff8107e53a>] warn_slowpath_fmt+0x4a/0x50
[   24.197149]  [<ffffffffa039b4b4>] intel_display_power_put+0x134/0x140 [i915]
[   24.197187]  [<ffffffffa04217dd>] intel_disable_ddi+0x4d/0x80 [i915]
[   24.197223]  [<ffffffffa03f388f>] intel_encoders_disable.isra.74+0x7f/0x90 [i915]
[   24.197257]  [<ffffffffa03f6c05>] haswell_crtc_disable+0x55/0x170 [i915]
[   24.197292]  [<ffffffffa03fec88>] intel_atomic_commit_tail+0x108/0xfd0 [i915]
[   24.197295]  [<ffffffff810d47c6>] ? __lock_is_held+0x66/0x90
[   24.197330]  [<ffffffffa03fff79>] intel_atomic_commit+0x429/0x560 [i915]
[   24.197332]  [<ffffffff81570186>] ?drm_atomic_add_affected_connectors+0x56/0xf0
[   24.197334]  [<ffffffff8156f726>] drm_atomic_commit+0x46/0x50
[   24.197336]  [<ffffffff81553f87>] restore_fbdev_mode+0x147/0x270
[   24.197337]  [<ffffffff81555bee>] drm_fb_helper_restore_fbdev_mode_unlocked+0x2e/0x70
[   24.197339]  [<ffffffff81555aa8>] drm_fb_helper_set_par+0x28/0x50
[   24.197374]  [<ffffffffa041c7d3>] intel_fbdev_set_par+0x13/0x70 [i915]
[   24.197376]  [<ffffffff8149e07a>] fbcon_init+0x57a/0x600
[   24.197379]  [<ffffffff81514b71>] visual_init+0xd1/0x130
[   24.197381]  [<ffffffff8151603c>] do_bind_con_driver+0x1bc/0x3a0
[   24.197384]  [<ffffffff81516521>] do_take_over_console+0x111/0x180
[   24.197386]  [<ffffffff8149e152>] do_fbcon_takeover+0x52/0xb0
[   24.197387]  [<ffffffff814a12c3>] fbcon_event_notify+0x723/0x850
[   24.197390]  [<ffffffff810a4830>] ?__blocking_notifier_call_chain+0x30/0x70
[   24.197392]  [<ffffffff810a44a4>] notifier_call_chain+0x34/0xa0
[   24.197394]  [<ffffffff810a4848>] __blocking_notifier_call_chain+0x48/0x70
[   24.197397]  [<ffffffff810a4881>] blocking_notifier_call_chain+0x11/0x20
[   24.197398]  [<ffffffff814a4556>] fb_notifier_call_chain+0x16/0x20
[   24.197400]  [<ffffffff814a678c>] register_framebuffer+0x24c/0x330
[   24.197402]  [<ffffffff815558d9>] drm_fb_helper_initial_config+0x219/0x3c0
[   24.197436]  [<ffffffffa041d373>] intel_fbdev_initial_config+0x13/0x30 [i915]
[   24.197438]  [<ffffffff810a5d44>] async_run_entry_fn+0x34/0x140
[   24.197440]  [<ffffffff8109c26c>] process_one_work+0x1ec/0x6b0
[   24.197442]  [<ffffffff8109c1e6>] ? process_one_work+0x166/0x6b0
[   24.197445]  [<ffffffff8109c779>] worker_thread+0x49/0x490
[   24.197447]  [<ffffffff8109c730>] ? process_one_work+0x6b0/0x6b0
[   24.197448]  [<ffffffff810a2a9b>] kthread+0xeb/0x110
[   24.197451]  [<ffffffff810a29b0>] ? kthread_park+0x60/0x60
[   24.197453]  [<ffffffff818241a7>] ret_from_fork+0x27/0x40
[   24.197476] ---[ end trace bda64b683b8e8162 ]---

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1481812185-19098-3-git-send-email-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Disable all crtcs during driver unload, v2.

We may keep the crtc's enabled when userspace unsets all framebuffers but
keeps the crtc active. This exposes a WARN in fbc_global disable, and
a lot of bugs in our hardware readout code. Solve this by disabling
all crtc's for now.

Changes since v1:
- Use lock_all_ctx instead of lock_all.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1481812185-19098-4-git-send-email-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/execlists: Add interrupt-pending check to intel_execlists_idle()

Primarily this serves as a sanity check that the bit has been cleared
before we suspend (and hasn't reappeared after resume).

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Tvrtko Ursulin <tursulin@ursulin.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170201131222.11882-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>

drm/i915/execlists: Skip resetting RING_CONTEXT_STATUS_PTR

As we now flag when the GPU signals a context-switch and do not read the
status register before we see that signal, we do not have to ensure that
it is cleared upon reset (and can leave it to the GPU to reset it from
the power context).

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Tvrtko Ursulin <tursulin@ursulin.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170201125338.12932-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>

drm/i915: Add MIPI_IO WA and program DSI regulators

Enable MIPI IO WA for BXT DSI as per bspec and
program the DSI regulators.

v2: Moved IO enable to pre-enable as per Mika's
review comments. Also reused the existing register
definition for BXT_P_CR_GT_DISP_PWRON.

v3: Added Programming the DSI regulators as per disable/enable
sequences.

v4: Restricting regulator changes to BXT as suggested by
Jani/Mika

v5: Removed redundant read/modify for regulator register as
per Jani's comment. Maintain enable/disable symmetry as per spec.

Signed-off-by: Uma Shankar <uma.shankar@intel.com>
Signed-off-by: Vidya Srinivas <vidya.srinivas@intel.com>
Acked-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485353603-11260-1-git-send-email-vidya.srinivas@intel.com

drm/i915: Flush untouched framebuffers before display on !llc

On a non-llc system, the objects are created with .cache_level =
CACHE_NONE and so the transition to uncached for scanout is a no-op.
However, if the object was never written to, it will still be in the CPU
domain (having been zeroed out by shmemfs). Those cachelines need to be
flushed prior to display.

Reported-and-tested-by: Vito Caputo
Fixes: a6a7cc4b7db6 ("drm/i915: Always flush the dirty CPU cache when pinning the scanout")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.10-rc1+
Link: http://patchwork.freedesktop.org/patch/msgid/20170109111932.6342-1-chris@chris-wilson.co.uk
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Merge remote-tracking branch 'airlied/drm-next' into drm-intel-next-queued

Chris Wilson wants the new fence tracepoint added in

commit 8c96c678011eeb1676da18f203e90dea7e0d69d2
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Jan 24 11:57:58 2017 +0000

    dma/fence: Export enable-signaling tracepoint for emission by drivers

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

Merge branch 'exynos-drm-next' of git://git./linux/kernel/git/daeinki/drm-exynos into drm-next

adding runtime PM support to MIC driver, and including some
cleanups - especially using atomic helper functions
instead of specific ones - and fixups.

* 'exynos-drm-next' of git://git.kernel.org/pub/scm/linux/kernel/git/daeinki/drm-exynos:
  drm/exynos: g2d: prevent integer overflow in
  drm/exynos: fix a timeout loop
  drm/exynos: use atomic helper commit
  drm/exynos: remove unnecessary codes
  drm/exynos: mic: Add runtime PM support
  drm/exynos: Stop using drm_framebuffer_unregister_private
  drm/exynos: mic: Fix parse_dt function
  drm/exynos: mic: Add mode_set callback function

Merge tag 'sunxi-drm-for-4.11' of https://git./linux/kernel/git/mripard/linux into drm-next

Allwinner DRM changes for 4.11

Just one minor fix.

* tag 'sunxi-drm-for-4.11' of https://git.kernel.org/pub/scm/linux/kernel/git/mripard/linux:
  drm/sun4i: Fix a return value in case of error

Merge branch 'drm-vmwgfx-next' of git://people.freedesktop.org/~syeh/repos_linux into drm-next

This is to address what we've discussed, moving some of the minor changes
into a drm-next request.

* 'drm-vmwgfx-next' of git://people.freedesktop.org/~syeh/repos_linux:
  drm/vmwgfx: Use kmemdup instead of kmalloc and memcpy
  drm/vmwgfx: Fix depth input into drm_mode_legacy_fb_format
  drm/vmwgfx: Fix a potential integer overflow
  drm/vmwgfx: Clear an uninitialized struct member
  drm/vmwgfx: Annotate ignored return values
  drm/vmwgfx: Clear uninitialized fields of a parameter

Merge branch 'drm-next-4.11' of git://people.freedesktop.org/~agd5f/linux into drm-next

This is the main feature pull for radeon and amdgpu for 4.11.  Highlights:
- Power and clockgating improvements
- Preliminary SR-IOV support
- ttm buffer priority support
- ttm eviction fixes
- Removal of the ttm lru callbacks
- Remove SI DPM quirks due to MC firmware issues
- Handle VFCT with multiple vbioses
- Powerplay improvements
- Lots of driver cleanups

* 'drm-next-4.11' of git://people.freedesktop.org/~agd5f/linux: (120 commits)
  drm/amdgpu: fix amdgpu_bo_va_mapping flags
  drm/amdgpu: access stolen VRAM directly on CZ (v2)
  drm/amdgpu: access stolen VRAM directly on KV/KB (v2)
  drm/amdgpu: fix kernel panic when dpm disabled on Kv.
  drm/amdgpu: fix dpm bug on Kv.
  drm/amd/powerplay: fix regresstion issue can't set manual dpm mode.
  drm/amdgpu: handle vfct with multiple vbios images
  drm/radeon: handle vfct with multiple vbios images
  drm/amdgpu: move misc si headers into amdgpu
  drm/amdgpu: remove unused header si_reg.h
  drm/radeon: drop pitcairn dpm quirks
  drm/amdgpu: drop pitcairn dpm quirks
  drm: radeon: radeon_ttm: Handle return NULL error from ioremap_nocache
  drm/amd/amdgpu/amdgpu_ttm: Handle return NULL error from ioremap_nocache
  drm/amdgpu: add new virtual display ID
  drm/amd/amdgpu: remove the uncessary parameter for ib scheduler
  drm/amdgpu: Bring bo creation in line with radeon driver (v2)
  drm/amd/powerplay: fix misspelling in header guard
  drm/ttm: revert "add optional LRU removal callback v2"
  drm/ttm: revert "implement LRU add callbacks v2"
  ...

Merge tag 'drm-misc-next-2017-01-30' of git://anongit.freedesktop.org/git/drm-misc into drm-next

Another round of -misc stuff:
- Noralf debugfs cleanup cleanup (not yet everything, some more driver
  patches awaiting acks).
- More doc work.
- edid/infoframe fixes from Ville.
- misc 1-patch fixes all over, as usual

Noralf needs this for his tinydrm pull request.

* tag 'drm-misc-next-2017-01-30' of git://anongit.freedesktop.org/git/drm-misc: (48 commits)
  drm/vc4: Remove vc4_debugfs_cleanup()
  dma/fence: Export enable-signaling tracepoint for emission by drivers
  drm/tilcdc: Remove tilcdc_debugfs_cleanup()
  drm/tegra: Remove tegra_debugfs_cleanup()
  drm/sti: Remove drm_debugfs_remove_files() calls
  drm/radeon: Remove drm_debugfs_remove_files() call
  drm/omap: Remove omap_debugfs_cleanup()
  drm/hdlcd: Remove hdlcd_debugfs_cleanup()
  drm/etnaviv: Remove etnaviv_debugfs_cleanup()
  drm/etnaviv: allow build with COMPILE_TEST
  drm/amd/amdgpu: Remove drm_debugfs_remove_files() call
  drm/prime: Clarify DMA-BUF/GEM Object lifetime
  drm/ttm: Make sure BOs being swapped out are cacheable
  drm/atomic: Remove drm_atomic_debugfs_cleanup()
  drm: drm_minor_register(): Clean up debugfs on failure
  drm: debugfs: Remove all files automatically on cleanup
  drm/fourcc: add vivante tiled layout format modifiers
  drm/edid: Set YQ bits in the AVI infoframe according to CEA-861-F
  drm/edid: Set AVI infoframe Q even when QS=0
  drm/edid: Introduce drm_hdmi_avi_infoframe_quant_range()
  ...

Merge tag 'drm/atmel-hlcdc/for-4.11' of git://git./linux/kernel/git/bbrezillon/linux into drm-next

Contains a single patch to create the fbdev at driver's registration
time instead of waiting for the connector status change.

* tag 'drm/atmel-hlcdc/for-4.11' of git://git.kernel.org/pub/scm/linux/kernel/git/bbrezillon/linux:
  drm/atmel-hlcdc: Rework the fbdev creation logic

Merge tag 'zxdrm-4.11' of git://git./linux/kernel/git/shawnguo/linux into drm-next

ZTE DRM driver updates for 4.11:
 - Add missing selection of VIDEOMODE_HELPERS in Kconfig, since ZTE DRM
   driver uses drm_display_mode_to_videomode().
 - Enable HDMI audio support through SPDIF interface based on generic
   hdmi-audio-codec driver.
 - Enable VOU VL (Video Layer) to support overlay plane with scaling
   function.
 - Refine zx_vou driver a bit and then add TV Encoder output device
   support.

[airlied: fixup plane format change]

* tag 'zxdrm-4.11' of git://git.kernel.org/pub/scm/linux/kernel/git/shawnguo/linux:
  drm: zte: add tvenc driver support
  dt: add bindings for ZTE tvenc device
  drm: zte: add function to configure vou_ctrl dividers
  drm: zte: move struct vou_inf into zx_vou driver
  drm: zte: add interlace mode support
  drm: zte: add overlay plane support
  drm: zte: add .atomic_disable hook to disable graphic layer
  drm: zte: make zx_plane accessible from zx_vou driver
  drm: zte: support hdmi audio through spdif
  drm: zte: select VIDEOMODE_HELPERS in Kconfig

Merge branch 'drm-intel-next' of git://anongit.freedesktop.org/git/drm-intel into drm-next

Updated pull request after I pulled first time :)

* 'drm-intel-next' of git://anongit.freedesktop.org/git/drm-intel:
  drm/i915: Pevent copying uninitialised garbage into vma->ggtt_view

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

Hope I'm not too late before the cutoff for the v4.11 with these patches.
Mostly an asorted set of fixes that we have discovered while playing with
the code and preparing for the next set of features.

* 'for-upstream/mali-dp' of git://linux-arm.org/linux-ld:
  drm: mali-dp: fix stride setting for multi-plane formats
  drm: mali-dp: Add plane offset to the plane's physical start address register
  drm: mali-dp: Check for sufficient address space
  drm: mali-dp: Check hw version matches device-tree
  drm: mali-dp: Rename malidp_input_format to malidp_pixel_format
  drm: mali-dp: fix Lx_CONTROL register fields clobber
  drm: mali-dp: Fix transposed horizontal/vertical flip
  drm: mali-dp: Fix destination size handling when rotating
  drm: mali-dp: Don't force source size == crtc size
  drm: mali-dp: Check more use cases in the plane's ->atomic_check()
  drm: malidp: Remove event_list member from struct malidp_drm
  drm/arm/malidp: Fix possible dereference of NULL

drm/i915: Sanity check the computed size and base of stolen memory

Just do a quick check that the stolen memory address range doesn't
overflow our chosen integer type.

v2: Add add_overflows() to utils with the promise that gcc7 can do this
better than C and then maybe it will have a proper definition in core.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170130134721.5159-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Treat stolen memory as DMA addresses

The conversion of stolen to use phys_addr_t (from essentially u32)
sparked an interesting discussion. We treat stolen memory as only
accessible from the GPU (the DMA device) - an attempt to use it from the
CPU will generate a MCE on gen6 onwards, although it is in theory a
physical address that can be dereferenced from the CPU as demonstrated
by earlier generations. As such, using phys_addr_t has the wrong
connotations and as we pass the address into the DMA device via
dma_addr_t (through the scatterlists used to program the GTT entries),
we should treat it as dma_addr_t throughout.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170127165531.28135-1-chris@chris-wilson.co.uk
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>

drm/i915: Be defensive when cleaning up i915_gem_internal pages

If we abort the i915_gem_internal get_pages, we mark the failing sg as
the last. However, that means we iterate upto and including the failing
sg element and results in us trying to free the unallocated sg_page().

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170131104630.3074-1-chris@chris-wilson.co.uk
Reviewed-by: Matthew Auld <matthew.auld@intel.com>

drm/i915: clean up unused vgpu_read/write

Having converted the force_wake_get/_put routines for a vGPU to be no-op,
we can use the common mmio accessors and remove our specialised routines
that simply skipped the calls to control force_wake.

Signed-off-by: Weinan Li <weinan.z.li@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1485408228-12932-1-git-send-email-weinan.z.li@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: noop forcewake get/put when vgpu activated

For a virtualized GPU, the host maintains the forcewake state on the real
device. As we don't control forcewake ourselves, we can simply set
force_wake_get() and force_wake_put() to be no-ops. By setting the vfuncs,
we adjust both the manual control of forcewake and around the mmio
accessors (making our vgpu specific mmio routines redundant and to be
removed in the next patch).

Signed-off-by: Weinan Li <weinan.z.li@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1485408013-12780-1-git-send-email-weinan.z.li@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

drm/i915: Record more information about the hanging contexts

Include extra information such as the user_handle and hw_id so that
userspace can identify which of their contexts hung, useful if they are
performing self-diagnositics.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170129092433.10483-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>

drm/i915: Add early BXT sdv to the list of preproduction machines

Extend intel_detect_preproduction_hw() to include BXT A and B steppings.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Zanoni, Paulo R" <paulo.r.zanoni@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170130104458.2653-3-chris@chris-wilson.co.uk
Acked-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Mark the kernel as tainted if we fail the preproduction check

Preproduction sdv are not supported beyond the release of production
hardware, and continued use is ill-advised. Mark the kernel as tainted
to reinforce the error.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Zanoni, Paulo R" <paulo.r.zanoni@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Acked-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170130104458.2653-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915: Make intel_detect_preproduction_hw easier to extend

As we add new generations, we should keep detecting new pre-production
system development platforms that were temporarily enabled to facilitate
initial development and now superseded by production systems. To make
it easier to add more platforms, split the if into a series of logical
operations.

v2: s/sdv/pre/ - not all system development vehicles are for
preproduction usage.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Zanoni, Paulo R" <paulo.r.zanoni@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170130104458.2653-1-chris@chris-wilson.co.uk
Acked-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/exynos: g2d: prevent integer overflow in

The size computations done in the ioctl function use an integer.
If userspace submits a request with req->cmd_nr or req->cmd_buf_nr
set to INT_MAX, the integer computations overflow later, leading
to potential (kernel) memory corruption.

Prevent this issue by enforcing a limit on the number of submitted
commands, so that we have enough headroom later for the size
computations.

Note that this change has no impact on the currently available
users in userspace, like e.g. libdrm/exynos.

While at it, also make a comment about the size computation more
detailed.

Signed-off-by: Joonyoung Shim <jy0922.shim@samsung.com>
Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: fix a timeout loop

We were trying to print an error message if we timed out here, but the
loop actually ends with "tries" set to UINT_MAX and not zero.  Fix this
by changing from tries-- to --tries.

A for loop would actually be the most natural way to do this.  My fix
means we only loop 99 times instead of 100 but that's probably ok.

Fixes: a696394c5224 ('drm/exynos: mixer: simplify loop in vp_win_reset()')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: use atomic helper commit

This patch replaces specific atomic commit function
with atomic helper commit one.

For this, it removes existing atomic commit function
and relevant code specific to Exynos DRM and makes
atomic helper commit to be used instead.

Below are changes for the use of atomic helper commit:
- add atomic_commit_tail callback specific to Exynos DRM
  . default implemention of atomic helper doesn't mesh well
    with runtime PM so the device driver which supports runtime
    PM should call drm_atomic_helper_commit_modeset_enables function
    prior to drm_atomic_helper_commit_planes function call.
    atomic_commit_tail callback implements this call ordering.
- allow plane commit only in case that CRTC device is enabled.
  . for this, it calls atomic_helper_commit_planes function
    with DRM_PLANE_COMMIT_ACTIVE_ONLY flag in atomic_commit_tail callback.

Signed-off-by: Inki Dae <inki.dae@samsung.com>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.com>

drm/exynos: remove unnecessary codes

This patch removes exynos_drm_crtc_cancel_page_flip call
when drm is closed because at that time, events will be released
by drm_events_release function.

Changelog v1:
- remove exynos_drm_crtc_cancel_page_flip function also because
  this funtion isn't used anymore.

Signed-off-by: Inki Dae <inki.dae@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>

drm/exynos: mic: Add runtime PM support

This patch adds runtime support calls to notify device core when MIC
device is really in use. Runtime PM is implemented by enabling and
disabling clocks like in other Exynos DRM subdrivers. Adding runtime
PM support is needed to let power domain with this device to be turned
off when display is not used.

Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: Stop using drm_framebuffer_unregister_private

This is the deprecated function for when you embedded the framebuffer
somewhere else (which breaks refcounting). But exynos is using
drm_framebuffer_remove and a free-standing fb, so this is rendundant.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: mic: Fix parse_dt function

The OF graph is not necessary because the panel is a child of
dsi. therefore, the parse_dt function of dsi does not need to
check the remote_node connected to the panel. and the whole
parse_dt function should be refactored later.

Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/exynos: mic: Add mode_set callback function

Before applying the patch, used the of_get_videomode function to
parse the display-timings in the panel which is the child driver
of dsi in the devicetree. this is wrong. So removed the
of_get_videomode and fixed to get videomode struct through
mode_set callback function.

Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com>
Reviewed-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Inki Dae <inki.dae@samsung.com>

drm/i915: use variadic macros and arrays to choose port/pipe based registers

This allows the use of more than 3 ports/pipes/whatever without tricks,
even if the register offsets are not evenly spaced.

There's the risk of out of bounds access if we're not careful; currently
that would "just" lead to the wrong register offset being used. It might
be possible to add build bug ons for build time constant indexing.

We already have ports defined up to E, not sure if we might have bugs
related to them and the current _PORT3() macro.

   text    data     bss     dec     hex filename
1239868   46199    4096 1290163  13afb3 drivers/gpu/drm/i915/i915.ko
1238828   46199    4096 1289123  13aba3 drivers/gpu/drm/i915/i915.ko

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Tvrtko Ursulin <tursulin@ursulin.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485532626-20923-1-git-send-email-jani.nikula@intel.com

drm/i915: Create context desc template when context is created

Move the invariant parts of context desc setup from execlist init
to context creation. This is advantageous when we need to
create different templates based on the context parametrization,
ie. for svm capable contexts.

v2: s/create/default, remove engine->ctx_desc_template

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485522189-31984-1-git-send-email-mika.kuoppala@intel.com

drm/atmel-hlcdc: Rework the fbdev creation logic

Now that we wait for DRM panels to be available before registering the
DRM device (returning -EPROBE_DEFER if the panel has not been probed
yet), we no longer need to put the fbdev creation code in
->output_poll_changed().

This removes the 10 secs delay between DRM dev registration and fbdev
creation (polling period = 10 seconds).

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Reported-by: Alex Vazquez <avazquez.dev@gmail.com>

drm/vc4: Remove vc4_debugfs_cleanup()

drm_debugfs_cleanup() now removes all minor->debugfs_list entries
automatically, so the drm_driver.debugfs_cleanup callback is not
needed.

Cc: eric@anholt.net
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170126225621.12314-17-noralf@tronnes.org

drm/i915/glk: Program pipe gamma and degamma tables

The gamma tables in Geminilake were changed. There is no split-gamma
mode. Instead, there is a dedicated degamma table that is enabled
whenever pipe CSC is enabled.

The dedicated gamma table has 16 bit precision but doesn't support
separate channels. Since that doesn't match the per-channel format of
the degamma LUT property, for now only a linear table is loaded and the
property ignored.

v2: Remove empty line. (Ville)
    Reuse broadwell code. (Ville)

v3: Don't write PIPE_CSC_MODE. (Ville)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170127090230.20302-1-ander.conselvan.de.oliveira@intel.com

drm/i915: Split broadwell_load_luts() into smaller functions

Split the logic for progamming each LUT out of broadwell_load_luts(), so
we can reuse part of it for geminilake.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485429865-10687-4-git-send-email-ander.conselvan.de.oliveira@intel.com

drm/i915/glk: Plane color correction register changes

In Geminilake, the bits for enabling pipe csc, pipe gamma and plane
gamma moved to a new register. So update the plane update functions
to set the right bits.

Pipe CSC is kept disabled though, since enabling that also enables the
dedicated degamma table, and that is not properly programmed yet,
leading to a black screen.

v2: Use plane_id. (Ville)
    Remove unnecessary variable. (Ville)
    Keep registers in offset order. (Ville)
    Don't set plane gamma disable twice. (Ander)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Ander Conselvan De Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485429865-10687-3-git-send-email-ander.conselvan.de.oliveira@intel.com

drm/i915: Disable plane gamma in SKL+ sprite planes

The plane gamma tables are never programmed, so just disable it, like it
is done for the primary plane.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485429865-10687-2-git-send-email-ander.conselvan.de.oliveira@intel.com

drm/i915/glk: Turn on workarounds that apply to Geminilake too

Apply workarounds to Geminilake, and annotate those that are applied
unconditionally when they apply to GLK based on the workaround database.

v2: Fix commit message typos. (David)
v3: Rebase.
Cc: David Weinehall <david.weinehall@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: David Weinehall <david.weinehall@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1485422218-9102-1-git-send-email-ander.conselvan.de.oliveira@intel.com

drm: zte: add tvenc driver support

It adds the TV Encoder driver to support video output in PAL and NTSC
format.  The driver uses syscon/regmap interface to configure register
bit sitting in SYSCTRL module for DAC power control.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

dt: add bindings for ZTE tvenc device

It adds bindings doc for ZTE VOU TV Encoder device.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Acked-by: Rob Herring <robh@kernel.org>

drm: zte: add function to configure vou_ctrl dividers

The clock control module (CRM) cannot always provide desired frequency
for all VOU output devices.  That's why VOU integrates a few dividers
to further divide the clocks from CRM.  Let's add an interface for
configuring these dividers.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: move struct vou_inf into zx_vou driver

Although data in struct vou_inf is defined per output device, it doesn't
belong to the device itself but VOU control module.  All these data can
just be defined in VOU driver, and output device driver only needs to
invoke VOU driver function with device ID to enable/disable specific
output device.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: add interlace mode support

It adds interlace mode support in VOU TIMING_CTRL and channel control
block, so that VOU driver gets ready to support output device in
interlace mode like TV Encoder.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: add overlay plane support

It enables VOU VL (Video Layer) to support overlay plane with scaling
function.  VL0 has some quirks on scaling support.  We choose to skip it
and only adds VL1 and VL2 into DRM core for now.

Function zx_plane_atomic_disable() gets moved around with no changes to
save a forward declaration.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: add .atomic_disable hook to disable graphic layer

There are a few hardware bits for each graphic layer to control main/aux
channel and clock selection, as well as the layer enabling.  These bits
sit outside the layer block itself, but in VOU control glue block.  We
currently set these bits up at CRTC initialization for once, and do not
support disabling the layer.

This patch creates a pair of functions zx_vou_layer_enable[disable] to
be invoked from plane hooks .atomic_update and .atomic_disable to set up
and tear down the layer.  This is generic for both graphic and video
layers, so it will make the overlay plane support to be added later much
easier.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: make zx_plane accessible from zx_vou driver

Move struct zx_plane from zx_plane.c to zx_plane.h, so that it can be
accessed from zx_vou driver, and we can save the use of struct
zx_layer_data completely.  More importantly, those additional data used
by VOU controller to enable/disable graphic and video layers can later
be added and accessed much more easily from zx_vou driver.

While at it, we make two changes to zx_plane_init() interface:

 - Encode struct device pointer in zx_plane, so that we do not need to
   pass it as a parameter.
 - Change return of zx_plane_init() from struct drm_plane pointer to
   error code, since we can get the pointer from zx_plane in zx_vou
   driver now.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>

drm: zte: support hdmi audio through spdif

It enables HDMI audio support through SPDIF interface based on generic
hdmi-audio-codec driver.  The HDMI hardware supports more audio
interfaces than SPDIF, like I2S, which may be added later.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

drm: zte: select VIDEOMODE_HELPERS in Kconfig

ZTE DRM driver uses drm_display_mode_to_videomode() in function
zx_crtc_enable().  Select VIDEOMODE_HELPERS in Kconfig to fix the
following link error.

  LD      vmlinux.o
  MODPOST vmlinux.o
drivers/built-in.o: In function `zx_crtc_enable':
:(.text+0xbdeb8): undefined reference to `drm_display_mode_to_videomode'

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>

drm/i915: Support explicit fencing for execbuf

Now that the user can opt-out of implicit fencing, we need to give them
back control over the fencing. We employ sync_file to wrap our
drm_i915_gem_request and provide an fd that userspace can merge with
other sync_file fds and pass back to the kernel to wait upon before
future execution.

Testcase: igt/gem_exec_fence
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Acked-by: Chad Versace <chadversary@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20170127094008.27489-2-chris@chris-wilson.co.uk

drm/i915: Enable userspace to opt-out of implicit fencing

Userspace is faced with a dilemma. The kernel requires implicit fencing
to manage resource usage (we always must wait for the GPU to finish
before releasing its PTE) and for third parties. However, userspace may
wish to avoid this serialisation if it is either using explicit fencing
between parties and wants more fine-grained access to buffers (e.g. it
may partition the buffer between uses and track fences on ranges rather
than the implicit fences tracking the whole object). It follows that
userspace needs a mechanism to avoid the kernel's serialisation on its
implicit fences before execbuf execution.

The next question is whether this is an object, execbuf or context flag.
Hybrid users (such as using explicit EGL_ANDROID_native_sync fencing on
shared winsys buffers, but implicit fencing on internal surfaces)
require a per-object level flag. Given that this flag need to be only
set once for the lifetime of the object, this reduces the convenience of
having an execbuf or context level flag (and avoids having multiple
pieces of uABI controlling the same feature).

Incorrect use of this flag will result in rendering corruption and GPU
hangs - but will not result in use-after-free or similar resource
tracking issues.

Serious caveat: write ordering is not strictly correct after setting
this flag on a render target on multiple engines. This affects all
subsequent GEM operations (execbuf, set-domain, pread) and shared
dma-buf operations. A fix is possible - but costly (both in terms of
further ABI changes and runtime overhead).

Testcase: igt/gem_exec_async
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Acked-by: Chad Versace <chadversary@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20170127094008.27489-1-chris@chris-wilson.co.uk

dma/fence: Export enable-signaling tracepoint for emission by drivers

Currently this tracepoint is solely used by dma_fence_enable_sw_signaling,
however I have a need to manually perform the hw enabling of the
signaling and would like to emit this tracepoint for completeness.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170124115758.31353-1-chris@chris-wilson.co.uk

drm/amdgpu: fix amdgpu_bo_va_mapping flags

They are 64bit not 32 for a while now.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: access stolen VRAM directly on CZ (v2)

We don't need to use the PCI BAR on APUs. This allows us to access
the full VRAM directly without being limited by the BAR size.

v2: squash in 64bit shift fix

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: access stolen VRAM directly on KV/KB (v2)

We don't need to use the PCI BAR on APUs. This allows us to access
the full VRAM directly without being limited by the BAR size.

v2: squash in 64bit shift fix

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix kernel panic when dpm disabled on Kv.

Return early if it's disabled.

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

drm/amdgpu: fix dpm bug on Kv.

1. current_ps/request_ps not update.
2. compare crrent_ps and request_ps, if same,
   don't re-set power state.

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

drm/amd/powerplay: fix regresstion issue can't set manual dpm mode.

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

drm/amdgpu: handle vfct with multiple vbios images

The vfct table can contain multiple vbios images if the
platform contains multiple GPUs. Noticed by netkas on
phoronix forums.  This patch fixes those platforms.

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

drm/radeon: handle vfct with multiple vbios images

The vfct table can contain multiple vbios images if the
platform contains multiple GPUs. Noticed by netkas on
phoronix forums.  This patch fixes those platforms.

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

drm/amdgpu: move misc si headers into amdgpu

Move these to the amdgpu directory to match what we
do for other asics.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: remove unused header si_reg.h

All of these are available elsewhere.

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/radeon: drop pitcairn dpm quirks

No longer necessary with the new 58 mc ucode.

bug:
https://bugs.freedesktop.org/show_bug.cgi?id=76490

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

drm/amdgpu: drop pitcairn dpm quirks

No longer necessary with the new 58 mc ucode.

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

drm: radeon: radeon_ttm: Handle return NULL error from ioremap_nocache

Here, If ioremap_nocache will fail. It will return NULL.
Kernel can run into a NULL-pointer dereference.
This error check will avoid NULL pointer dereference.

Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu/amdgpu_ttm: Handle return NULL error from ioremap_nocache

Here, If ioremap_nocache will fail. It will return NULL.
Kernel can run into a NULL-pointer dereference.
This error check will avoid NULL pointer dereference.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Arvind Yadav <arvind.yadav.cs@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: add new virtual display ID

In the case of pass-through, amdgpu.ko may be included into a image
with the hard code ID therefore loading driver with specified virtual
display ID will lose efficacy when the BDF of GPU modifies.So add the
new ID string "all" for it as same as vf case what does.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Yintian Tao <yttao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: remove the uncessary parameter for ib scheduler

Signed-off-by: Junwei Zhang <Jerry.Zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Bring bo creation in line with radeon driver (v2)

Add the bo creation changes that have been done to the radeon driver in
recent times, e.g. disable GTT WC on 32 bit because it is broken there,
and also disable it generally (and print a warning message) when
CONFIG_X86_PAT is not set.

v2: agd: fix warning in defined(CONFIG_X86) && !defined(CONFIG_X86_PAT)
    case

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Nils Holland <nholland@tisys.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/powerplay: fix misspelling in header guard

In smu7_clockpowergating.h, the #ifndef statement which prevents
multiple inclusions of the header file uses _SMU7_CLOCK_POWER_GATING_H_
but the following #define statement uses _SMU7_CLOCK__POWER_GATING_H_.

Signed-off-by: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: revert "add optional LRU removal callback v2"

Without the custom LRU management the callback is not used any more.

agd: fix trivial warning

Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: revert "implement LRU add callbacks v2"

The additional housekeeping had too much CPU overhead,
let's use the BO priorities instead.

agd: also revert hibmc changes

Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-and-Tested-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: double the priority of kernel allocations

Give kernel allocations a higher priority cause it is often
more work to swap them back in.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: user BO priority instead of self coding it (v2)

Keeping groups of BOs on the LRU is to time consuming on command submission.

Instead use the newly added BO priority to give a certain eviction order.

v2: agd: trivial warning fix

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: add BO priorities for the LRUs

This way the driver can specify a priority for a BO which has the effect that
a BO is only evicted when all other BOs with a lower priority are evicted
first.

Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: remove allow_errors parameter from ttm_bo_force_list_clean

Not allowing errors here is completely pointless and actually dangerous
cause trying to continue on an error can cause an endless loop.

Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger.He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu/vi: fix mailbox irq mistake

For virt, freed mailbox irq should be handled in hw fini, not hw
init. Correct it.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Reviewed-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: fix reboot failure issue for virtualization

Reboot process will call HW fini functions of IP blocks. For virt,
need to send event three before hw fini and send event four after
hw fini.

Signed-off-by: Xiangliang Yu <Xiangliang.Yu@amd.com>
Reviewed-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amd/amdgpu: Add PCI info to gca_config debugfs

So we can determine which device the entry is before connecting
a display.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:insert switch buffer only for VM submit

for non-VM submit which is from kernel side, no need
to switch buffer at all.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:Preamble is forbid to be ignored in SRIOV

SR-IOV requires the preamble.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:set cond_exec polling value to 1 in ring_init

no need to set it per ib_schedule(), hw won't override
this polling address.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:in cntx_ctrl we need insert meta-init for CE/DE(V2)

to support SRIOV preemption.

v2:
fix emit_frame_size

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:introduce new flag to identify VM domain for cntx_ctrl

To determine whether the context uses GPUVM or not.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:implement CE/DE meta-init routines

those package need to insert into ring buffer for
SRIOV case. they are used to let CP do preemption.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:implement ring_write_multiple

Write multiple dwords to the ring.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu:add META_DATA struct for CSA/SRIOV v2

META-DATA is used in GFX cmd submit, we have two
format suit for META-DATA-init, one is legacy and another
is for chained-ib preempt, which is used in vulkan
UMD.

v2: drop use CP version number to judge if chain-ib
supports or not, we wait for it mature

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refine ci uvd dpm code.

Fix up the powergating logic.

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Ack-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: simplify allocation of scratch regs

The scratch regs are sequential so there's no need to keep
them in an array, we can just return the index of the first
free register + the base register. Also change the array
of bools for keeping track of the free regs to a bitfield.

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Nils Wallménius <nils.wallmenius@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Refactor flip into prepare submit and submit. (v3)

Make pflip atomic friendly. Split the fuinction into
whatever can fail part and the actual flip submit part.
Call the pre-submit function before atomic states
are swapped so in case of error we can fail the
IOCTL.

v2:
Update due to  target_vblank code change.
Fix identetation.
Change return type for amdgpu_crtc_submit_flip to void
v3: agd: fix formatting

Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: Switch error code when bo pin fails. (v3)

Switching to -ENOMEM resolves a hang in case the driver
uses this function in atomic code path.

v2: Update commit message.
v3:
Remove forcing -EINVAL in case of amdgpu_bo_pin fail
since there is no apparent reason for not just passing
through the the return value.

Signed-off-by: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>