drm/i915: Perform an invalidate prior to executing golden renderstate

As we may have just bound the renderstate into the GGTT for execution, we
need to ensure that the GTT TLB are also flushed.

On snb-gt2, this would cause a random GPU hang at the start of a new
context (e.g. boot) and on snb-gt1, it was causing the renderstate batch
to take ~10s. It was the GPU hang that revealed the truth, as the CS
gleefully executed beyond the end of the golden renderstate batch, a good
indicator for a GTT TLB miss.

Fixes: 20fe17aa52dc ("drm/i915: Remove redundant TLB invalidate on switching contexts")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: stable@vger.kernel.org
Link: https://patchwork.freedesktop.org/patch/msgid/20170808131904.1385-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.12-rc1+

drm/i915: remove unused function declaration

This function is not part of the driver anymore.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 90f4fcd56bda ("drm/i915: Remove forced stop ring on suspend/unload")
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170804140348.24971-1-lionel.g.landwerlin@intel.com

drm/i915/selftests: Retarget igt_render_engine_reset_fallback()

The purpose of the test was to check per-engine resets would fallback to
the global reset when required, but first we actually need a test for a
basic i915_handle_error()!

Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170728112110.6464-1-chris@chris-wilson.co.uk

drm/i915: fix backlight invert for non-zero minimum brightness

When we started following the backlight minimum brightness in
6dda730e55f4 ("drm/i915: respect the VBT minimum backlight brightness")
we overlooked the brightness invert quirk. Even if we invert the
brightness, we need to take the min limit into account. We probably
missed this because the invert has only been required on gen4 for proper
operation.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101127
Fixes: 6dda730e55f4 ("drm/i915: respect the VBT minimum backlight brightness")
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170531083355.7898-1-jani.nikula@intel.com

drm/i915/shrinker: Wrap need_resched() inside preempt-disable

In order for us to successfully detect the end of a timeslice,
preemption must be disabled. Otherwise, inside the loop we may be
preempted many times without our noticing, and each time our timeslice
will be reset, invalidating need_resched()

Reported-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reported-by: Tomi Sarvela <tomi.p.sarvela@intel.com>
Fixes: 290271de34f6 ("drm/i915: Spin for struct_mutex inside shrinker")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.13-rc1+
Link: https://patchwork.freedesktop.org/patch/msgid/20170804104135.26805-1-chris@chris-wilson.co.uk
Tested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

drm/i915/perf: Initialise the dynamic sysfs attr

Use sysfs_attr_init() to dynamically initialise the
oa_config->sysfs_metric_id.attr as it has the important side-effect of
setting the lockdep key.

[    4.971513] [drm] Initialized i915 1.6.0 20170731 for 0000:00:02.0 on minor 0
[    4.973489] BUG: key ffff88026f6e7bb0 not in .data!
[    4.973506] DEBUG_LOCKS_WARN_ON(1)
[    4.973518] ------------[ cut here ]------------
[    4.973547] WARNING: CPU: 1 PID: 258 at kernel/locking/lockdep.c:3156 lockdep_init_map+0x1b2/0x1c0
[    4.973567] Modules linked in: i915(+) x86_pkg_temp_thermal intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm r8169 mei_me mii mei lpc_ich prime_numbers i2c_hid pinctrl_broxton pinctrl_intel
[    4.973645] CPU: 1 PID: 258 Comm: systemd-udevd Not tainted 4.13.0-rc3-CI-CI_DRM_2915+ #1
[    4.973664] Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./J4205-ITX, BIOS P1.10 09/29/2016
[    4.973686] task: ffff8802704c2740 task.stack: ffffc90000224000
[    4.973700] RIP: 0010:lockdep_init_map+0x1b2/0x1c0
[    4.973712] RSP: 0018:ffffc90000227a10 EFLAGS: 00010282
[    4.973726] RAX: 0000000000000016 RBX: ffff880262aac010 RCX: 0000000000000000
[    4.973741] RDX: 0000000080000001 RSI: 0000000000000001 RDI: ffffffff810ed1ab
[    4.973757] RBP: ffffc90000227a30 R08: 0000000000000001 R09: 0000000000000000
[    4.973774] R10: 0000000000000000 R11: 0000000000000000 R12: ffff88026f6e7bb0
[    4.973789] R13: 0000000000000000 R14: ffff88026f6e7b98 R15: ffffffff81a24da0
[    4.973805] FS:  00007f588d7f58c0(0000) GS:ffff88027fc80000(0000) knlGS:0000000000000000
[    4.973823] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[    4.973837] CR2: 00000082482e32a0 CR3: 0000000270531000 CR4: 00000000003406e0
[    4.973852] Call Trace:
[    4.973864]  __kernfs_create_file+0x71/0xe0
[    4.973876]  sysfs_add_file_mode_ns+0x85/0x1a0
[    4.973890]  internal_create_group+0xe5/0x2b0
[    4.973903]  sysfs_create_group+0xe/0x10
[    4.973985]  i915_perf_register+0xd9/0x220 [i915]
[    4.974044]  i915_driver_load+0xa72/0x16b0 [i915]
[    4.974124]  i915_pci_probe+0x32/0x90 [i915]

Annoyingly detected by CI, but not reported due to it occurring during boot
and disabling lockdep for later runs.

Fixes: f89823c21224 ("drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Andrzej Datczuk <andrzej.datczuk@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803223700.10329-1-chris@chris-wilson.co.uk
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

drm/i915: enable WaDisableDopClkGating for skl

This WA is required when decoupled frequencies for slice and unslice
are enabled. This disables DOP clock gating for skl.

v2: enable the WA for all gen9 platforms (not just for SKL GT4 where
    the hang issue is originally reported) to avoid rare hangs (David)
v3: as per WaDatabase, enable it only for SKL (Rodrigo)

Cc: David Weinehall <david.weinehall@linux.intel.com>
Reviewed-by: David Weinehall <david.weinehall@linux.intel.com>
Signed-off-by: Praveen Paneri <praveen.paneri@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1501781530-8186-1-git-send-email-praveen.paneri@intel.com

drm/i915: Fix PCH names for KBP and CNP.

No functional change.

KBP was based on SPT and spec wasn't clear about the full name.
There was the initial point of the "Point" confusion.

Later the split with Coffee Lake and Cannon Lake both using CNP
and also some uncertainty from the specs we had at that time
made us to propagated the mistake along.

So, let's fix this now and avoid propagating these wrong
"points".

Cc: Anusha Srivatsa <anusha.srivatsa@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Anusha Srivatsa <anusha.srivatsa@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170731185220.758-1-rodrigo.vivi@intel.com

drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface

The motivation behind this new interface is expose at runtime the
creation of new OA configs which can be used as part of the i915 perf
open interface. This will enable the kernel to learn new configs which
may be experimental, or otherwise not part of the core set currently
available through the i915 perf interface.

v2: Drop DRM_ERROR for userspace errors (Matthew)
    Add padding to userspace structure (Matthew)
    s/guid/uuid/ (Matthew)

v3: Use u32 instead of int to iterate through registers (Matthew)

v4: Lock access to dynamic config list (Lionel)

v5: by Matthew:
    Fix uninitialized error values
    Fix incorrect unwiding when opening perf stream
    Use kmalloc_array() to store register
    Use uuid_is_valid() to valid config uuids
    Declare ioctls as write only
    Check padding members are set to 0
    by Lionel:
    Return ENOENT rather than EINVAL when trying to remove non
    existing config

v6: by Chris:
    Use ref counts for OA configs
    Store UUID in drm_i915_perf_oa_config rather then using pointer
    Shuffle fields of drm_i915_perf_oa_config to avoid padding

v7: by Chris
    Rename uapi pointers fields to end with '_ptr'

v8: by Andrzej, Marek, Sebastian
    Update register whitelisting
    by Lionel
    Add more register names for documentation
    Allow configuration programming in non-paranoid mode
    Add support for value filter for a couple of registers already
    programmed in other part of the kernel

v9: Documentation fix (Lionel)
    Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej)

v10: Perform read access_ok() on register pointers (Lionel)

Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com>
Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com

drm/i915: reorder NOA register definition to follow addresses

It makes things easier to read when implementing whitelisting in the
following patches.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-6-lionel.g.landwerlin@intel.com

drm/i915/perf: disable NOA logic when not used

We already do it on Haswell and the documentation says it saves power.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-5-lionel.g.landwerlin@intel.com

drm/i915/perf: leave GDT_CHICKEN_BITS programming in configs

There will be a need for userspaces configurations to set this
register. We can apply the same model inside the kernel for test
configs.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-4-lionel.g.landwerlin@intel.com

drm/i915/perf: prune OA configs

In the following commit we'll introduce loadable userspace
configs. This change reworks how configurations are handled in the
perf driver and retains only the test configurations in kernel space.

We now store the test config in dev_priv and resolve the id only once
when opening the perf stream. The OA config is then handled through a
pointer to the structure holding the configuration details.

v2: Rework how test configs are handled (Lionel)

v3: Use u32 to hold number of register (Matthew)

v4: Removed unused dev_priv->perf.oa.current_config variable (Matthew)

v5: Lock device when accessing exclusive_stream (Lionel)

v6: Ensure OACTXCONTROL is always reprogrammed (Lionel)

v7: Switch a couple of index variable from int to u32 (Matthew)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-3-lionel.g.landwerlin@intel.com

drm/i915/perf: fix flex eu registers programming

We were reserving fewer dwords in the ring than necessary. Indeed
we're always writing all registers once, so discard the actual number
of registers given by the user and just program the whitelisted ones
once.

Fixes: 19f81df2859e ("drm/i915/perf: Add OA unit support for Gen 8+")
Reported-by: Matthew Auld <matthew.william.auld@gmail.com>
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Cc: <stable@vger.kernel.org> # v4.12+
Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-6-lionel.g.landwerlin@intel.com

drm/i915: add const to bin_attribute

Add const to bin_attribute structures as they are only passed to the
functions sysfs_{remove/create}_bin_file or
device_{remove/create}_bin_file. The corresponding arguments are of
type const, so declare the structures to be const.

Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/1501694447-14356-1-git-send-email-bhumirks@gmail.com

drm/i915: Fix out-of-bounds array access in bdw_load_gamma_lut

bdw_load_gamma_lut is writing beyond the array to the maximum value.
The intend of the function is to clamp values > 1 to 1, so write
the intended color to the max register.

This fixes the following KASAN warning:

[  197.020857] [IGT] kms_pipe_color: executing
[  197.063434] [IGT] kms_pipe_color: starting subtest ctm-0-25-pipe0
[  197.078989] ==================================================================
[  197.079127] BUG: KASAN: slab-out-of-bounds in bdw_load_gamma_lut.isra.2+0x3b9/0x570 [i915]
[  197.079188] Read of size 2 at addr ffff8800d38db150 by task kms_pipe_color/1839
[  197.079208] CPU: 2 PID: 1839 Comm: kms_pipe_color Tainted: G     U 4.13.0-rc1-patser+ #5211
[  197.079215] Hardware name: NUC5i7RYB, BIOS RYBDWi35.86A.0246.2015.0309.1355 03/09/2015
[  197.079220] Call Trace:
[  197.079230]  dump_stack+0x68/0x9e
[  197.079239]  print_address_description+0x6f/0x250
[  197.079251]  kasan_report+0x216/0x370
[  197.079374]  ? bdw_load_gamma_lut.isra.2+0x3b9/0x570 [i915]
[  197.079451]  ? gen8_write16+0x4e0/0x4e0 [i915]
[  197.079460]  __asan_report_load2_noabort+0x14/0x20
[  197.079535]  bdw_load_gamma_lut.isra.2+0x3b9/0x570 [i915]
[  197.079612]  broadwell_load_luts+0x1df/0x550 [i915]
[  197.079690]  intel_color_load_luts+0x7b/0x80 [i915]
[  197.079764]  intel_begin_crtc_commit+0x138/0x760 [i915]
[  197.079783]  drm_atomic_helper_commit_planes_on_crtc+0x1a3/0x820 [drm_kms_helper]
[  197.079859]  ? intel_pre_plane_update+0x571/0x580 [i915]
[  197.079937]  intel_update_crtc+0x238/0x330 [i915]
[  197.080016]  intel_update_crtcs+0x10f/0x210 [i915]
[  197.080092]  intel_atomic_commit_tail+0x1552/0x3340 [i915]
[  197.080101]  ? _raw_spin_unlock+0x3c/0x40
[  197.080110]  ? __queue_work+0xb40/0xbf0
[  197.080188]  ? skl_update_crtcs+0xc00/0xc00 [i915]
[  197.080195]  ? trace_hardirqs_on+0xd/0x10
[  197.080269]  ? intel_atomic_commit_ready+0x128/0x13c [i915]
[  197.080329]  ? __i915_sw_fence_complete+0x5b8/0x6d0 [i915]
[  197.080336]  ? debug_object_activate+0x39e/0x580
[  197.080397]  ? i915_sw_fence_await+0x30/0x30 [i915]
[  197.080409]  ? __might_sleep+0x15b/0x180
[  197.080483]  intel_atomic_commit+0x944/0xa70 [i915]
[  197.080490]  ? refcount_dec_and_test+0x11/0x20
[  197.080567]  ? intel_atomic_commit_tail+0x3340/0x3340 [i915]
[  197.080597]  ? drm_atomic_crtc_set_property+0x303/0x580 [drm]
[  197.080674]  ? intel_atomic_commit_tail+0x3340/0x3340 [i915]
[  197.080704]  drm_atomic_commit+0xd7/0xe0 [drm]
[  197.080722]  drm_atomic_helper_crtc_set_property+0xec/0x130 [drm_kms_helper]
[  197.080749]  drm_mode_crtc_set_obj_prop+0x7d/0xb0 [drm]
[  197.080775]  drm_mode_obj_set_property_ioctl+0x50b/0x5d0 [drm]
[  197.080783]  ? __might_fault+0x104/0x180
[  197.080809]  ? drm_mode_obj_find_prop_id+0x160/0x160 [drm]
[  197.080838]  ? drm_mode_obj_find_prop_id+0x160/0x160 [drm]
[  197.080861]  drm_ioctl_kernel+0x154/0x1a0 [drm]
[  197.080885]  drm_ioctl+0x624/0x8f0 [drm]
[  197.080910]  ? drm_mode_obj_find_prop_id+0x160/0x160 [drm]
[  197.080934]  ? drm_getunique+0x210/0x210 [drm]
[  197.080943]  ? __handle_mm_fault+0x1bd0/0x1ce0
[  197.080949]  ? lock_downgrade+0x610/0x610
[  197.080957]  ? __lru_cache_add+0x15a/0x180
[  197.080967]  do_vfs_ioctl+0xd92/0xe40
[  197.080975]  ? ioctl_preallocate+0x1b0/0x1b0
[  197.080982]  ? selinux_capable+0x20/0x20
[  197.080991]  ? __do_page_fault+0x7b7/0x9a0
[  197.080997]  ? lock_downgrade+0x5bb/0x610
[  197.081007]  ? security_file_ioctl+0x57/0x90
[  197.081016]  SyS_ioctl+0x4e/0x80
[  197.081024]  entry_SYSCALL_64_fastpath+0x18/0xad
[  197.081030] RIP: 0033:0x7f61f287a987
[  197.081035] RSP: 002b:00007fff7d44d188 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[  197.081043] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f61f287a987
[  197.081048] RDX: 00007fff7d44d1c0 RSI: 00000000c01864ba RDI: 0000000000000003
[  197.081053] RBP: 00007f61f2b3eb00 R08: 0000000000000059 R09: 0000000000000000
[  197.081058] R10: 0000002ea5c4a290 R11: 0000000000000246 R12: 00007f61f2b3eb58
[  197.081063] R13: 0000000000001010 R14: 00007f61f2b3eb58 R15: 0000000000002702

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101659
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reported-by: Martin Peres <martin.peres@linux.intel.com>
Cc: Martin Peres <martin.peres@linux.intel.com>
Fixes: 82cf435b3134 ("drm/i915: Implement color management on bdw/skl/bxt/kbl")
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Kiran S Kumar <kiran.s.kumar@intel.com>
Cc: Kausal Malladi <kausalmalladi@gmail.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: intel-gfx@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v4.7+
Link: https://patchwork.freedesktop.org/patch/msgid/20170724091431.24251-1-maarten.lankhorst@linux.intel.com
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>

drm/i915: Update DRIVER_DATE to 20170731

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Remove unused i915_err_print_instdone

Just a simple code cleanup, below commit forgot to remove a
function which it made unused:

 commit eaa14c24864ecfc60fb591f3b20747af7c67d446
 Author: Chris Wilson <chris@chris-wilson.co.uk>
 Date:   Wed Oct 19 13:52:03 2016 +0100

  drm/i915: Stop reporting error details in dmesg as well as the error-state

  As we already capture all the information from the registers into the
  error-state, also dumping that to dmesg just generates noise that upsets
  CI and users alike (and doesn't provide us with any more information).

v2: Chris Wilson dag out the relevant commit. Commit msg updated.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20170727110113.16942-1-tvrtko.ursulin@linux.intel.com

drm/i915: Include mbox details for pcode read/write failures

If we fail at punit communication, include both the mbox address and the
value we tried to write so that we can identify the invalid sequence.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170728085022.1586-1-chris@chris-wilson.co.uk
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/sdvo: Shut up state checker with hdmi cards on gen3

The hdmi bits simply don't exist, so nerf them. I think audio doesn't
work on gen3 at all, and for the limited color range we should
probably use the colorimetry sdvo paramater instead of the bit in the
port.

But fixing sdvo isn't my goal, I just want to get the backtrace out of
the way, and this takes care of that.

Still, while at it fix the missing read-out of the gen4 audio bit,
maybe that part even works ...

v2: Instead of trying to plug the damage in ->compute_config() make
sure we never set intel_sdvo->is_hdmi, which stops the bad state at
the source. Suggested by Chris Wilson. Also make sure we don't break
this by accident by putting a WARN_ON in place.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726193251.25393-1-daniel.vetter@ffwll.ch

drm/i915: Rework sdvo proxy i2c locking

lockdep complaints about a locking recursion for the i2c bus lock
because both the sdvo ddc proxy bus and the gmbus nested within use
the same locking class. It's not really a deadlock since we never nest
the other way round, but it's annoying.

Fix it by pulling the gmbus locking into the i2c lock_ops for both
i2c_adapater and making sure that the ddc_proxy_xfer function is
entirely lockless.

Re-layouting the extracted function resulted in some whitespace
cleanups, I figured we might as well keep them.

v2: Review from Chris:
- s/locked/unlocked/ since I got the naming backwards
- Use the vfuncs of the proxied adatper instead of re-rolling copies.
  That's more consistent with the other proxying we're doing.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726132647.31833-1-daniel.vetter@ffwll.ch

drm/i915: Call the unlocked version of i915_gem_object_get_pages()

When we hold for the lock for swapping out the shmem pages for the
physically contiguous pages, we have to call the unlocked version of
get_pages!

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101934
Fixes: 35d23516946e ("drm/i915: Make i915_gem_object_phys_attach() use obj->mm.lock more appropriately")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726181602.23527-2-chris@chris-wilson.co.uk
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Move i915_gem_object_phys_attach()

Prevent a forward declaration in the next patch.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726181602.23527-1-chris@chris-wilson.co.uk
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Pin the pages before acquiring struct_mutex for display

Since we don't need the struct_mutex to acquire the object's pages, call
i915_gem_object_pin_pages() before we bind the object into the GGTT.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726160038.29487-3-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Make i915_gem_object_phys_attach() use obj->mm.lock more appropriately

Actually transferring from shmemfs to the physically contiguous set of
pages should be wholly guarded by its obj->mm.lock!

v2: Remember to free the old pages.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726160038.29487-2-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Trim struct_mutex usage for kms

Reduce acquisition of struct_mutex to the critical regions that must
hold it; for KMS, we need struct_mutex currently only for the purpose of
pinning/unpinning the framebuffer's VMA into the global GTT. This allows
us to avoid taking the struct_mutex when disabling the CRTC (i.e. NULL
framebuffer objects) before a reset. (Not yet achieving the full goal of
avoiding the strut_mutex nesting, but good enough to break the first
half of the reset deadlock.)

v2: Keep pages pinning inside struct_mutex for the moment.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170726160038.29487-1-chris@chris-wilson.co.uk
[danvet: Drop another case of grabbing dev->struct_mutex around
cleanup_planes, which popped up because I had to redo the drm-next
backmerge for entirely different reasons. Acked by Chris on irc.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Handle msr read failure gracefully

When reading the i915_energy_uJ debugfs file, it tries to fetch
MSR_RAPL_POWER_UNIT, which might not be available, like in a vm
environment, causing the exception shown below.

We can easily prevent it by doing a rdmsrl_safe read instead, which will
handle the exception, allowing us to abort the debugfs file read.

This was caught by the new igt@debugfs_test@read_all_entries testcase in
the CI.

  unchecked MSR access error: RDMSR from 0x606 at rIP:0xffffffffa0078f66
    (i915_energy_uJ+0x36/0xb0 [i915])
  Call Trace:
   seq_read+0xdc/0x3a0
   full_proxy_read+0x4f/0x70
   __vfs_read+0x23/0x120
   ? putname+0x4f/0x60
   ? rcu_read_lock_sched_held+0x75/0x80
   ? entry_SYSCALL_64_fastpath+0x5/0xb1
   vfs_read+0xa0/0x150
   SyS_read+0x44/0xb0
   entry_SYSCALL_64_fastpath+0x1c/0xb1
  RIP: 0033:0x7f1f5e9f4500
  RSP: 002b:00007ffc77e65cf8 EFLAGS: 00000246 ORIG_RAX: 0000000000000000
  RAX: ffffffffffffffda RBX: ffffffff8146e003 RCX: 00007f1f5e9f4500
  RDX: 0000000000000200 RSI: 00007ffc77e65d10 RDI: 0000000000000006
  RBP: ffffc900007abf88 R08: 0000000001eaff20 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
  R13: 0000000000000006 R14: 0000000000000005 R15: 0000000001eb94db
   ? __this_cpu_preempt_check+0x13/0x20

v2:
 - Drop unsigned long long cast and improve calculation (Chris)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101901
Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/87o9s7zrx3.fsf@dilma.collabora.co.uk
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Force CPU synchronisation even if userspace requests ASYNC

The goal here was to minimise doing any thing or any check inside the
kernel that was not strictly required. For a userspace that assumes
complete control over the cache domains, the kernel is usually using
outdated information and may trigger clflushes where none were
required.

However, swapping is a situation where userspace has no knowledge of the
domain transfer, and will leave the object in the CPU cache. The kernel
must flush this out to the backing storage prior to use with the GPU. As
we use an asynchronous task tracked by an implicit fence for this, we
also need to cancel the ASYNC flag on the object so that the object will
wait for the clflush to complete before being executed. This also absolves
userspace of the responsibility imposed by commit 77ae9957897d ("drm/i915:
Enable userspace to opt-out of implicit fencing") that its needed to ensure
that the object was out of the CPU cache prior to use on the GPU.

Fixes: 77ae9957897d ("drm/i915: Enable userspace to opt-out of implicit fencing")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101571
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721145037.25105-5-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Only skip updating execobject.offset after error

I was being overly paranoid in not updating the execobject.offset after
performing the fallback copy where we set reloc.presumed_offset to -1.
The thinking was to ensure that a subsequent NORELOC execbuf would be
forced to process the invalid relocations. However this is overkill so
long as we *only* update the execobject.offset following a successful
update of the relocation value witin the batch. If we have to repeat the
execbuf due to a later interruption, then we may skip the relocations on
the second pass (honouring NORELOC) since the execobject.offset match
the actual offsets (even though reloc.presumed_offset is garbage).

Subsequent calls to execbuf with NORELOC should themselves ensure that
the reloc.presumed_offset have been corrected in case of future
migration.

Reporting back the actual execobject.offset, even when
reloc.presumed_offset is garbage, ensures that reuse of those objects
use the latest information to avoid relocations.

Fixes: 2889caa92321 ("drm/i915: Eliminate lots of iterations over the execobjects array")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101635
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721145037.25105-4-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Only mark the execobject as pinned on success

If we fail to acquire a fence (for old school fenced GPU access) then we
unwind the vma reservation, including its pin. However, we were making
the execobject as holding the pin before erring out, leading to a double
unpin:

[ 3193.991802] kernel BUG at drivers/gpu/drm/i915/i915_vma.h:287!
[ 3193.998131] invalid opcode: 0000 [#1] PREEMPT SMP
[ 3194.002816] Modules linked in: snd_hda_intel i915 vgem snd_hda_codec_analog snd_hda_codec_generic coretemp snd_hda_codec snd_hwdep snd_hda_core snd_pcm lpc_ich mei_me e1000e mei prime_numbers ptp pps_core [last unloaded: i915]
[ 3194.022841] CPU: 0 PID: 8123 Comm: kms_flip Tainted: G     U          4.13.0-rc1-CI-CI_DRM_471+ #1
[ 3194.031765] Hardware name: Dell Inc. OptiPlex 755                 /0PU052, BIOS A04 11/05/2007
[ 3194.040343] task: ffff8800785d4c40 task.stack: ffffc90001768000
[ 3194.046339] RIP: 0010:eb_release_vmas.isra.6+0x119/0x180 [i915]
[ 3194.052234] RSP: 0018:ffffc9000176ba80 EFLAGS: 00010246
[ 3194.057439] RAX: 00000000000003c0 RBX: ffff8800710fc2d8 RCX: ffff8800588e4f48
[ 3194.064546] RDX: ffffffff1fffffff RSI: 00000000ffffffff RDI: ffff8800588e00d0
[ 3194.071654] RBP: ffffc9000176bab0 R08: 0000000000000000 R09: 0000000000000000
[ 3194.078761] R10: 0000000000000040 R11: 0000000000000001 R12: ffff880060822f00
[ 3194.085867] R13: 0000000000000310 R14: 00000000000003b8 R15: ffffc9000176bbb0
[ 3194.092975] FS:  00007fd2b94aba40(0000) GS:ffff88007d200000(0000) knlGS:0000000000000000
[ 3194.101033] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 3194.106754] CR2: 00007ffbec3ff000 CR3: 0000000074e67000 CR4: 00000000000006f0
[ 3194.113861] Call Trace:
[ 3194.116321]  eb_relocate_slow+0x67/0x4e0 [i915]
[ 3194.120861]  i915_gem_do_execbuffer+0x429/0x1260 [i915]
[ 3194.126070]  ? lock_acquire+0xb5/0x210
[ 3194.129803]  ? __might_fault+0x39/0x90
[ 3194.133563]  i915_gem_execbuffer2+0x9b/0x1b0 [i915]
[ 3194.138447]  ? i915_gem_execbuffer+0x2b0/0x2b0 [i915]
[ 3194.143478]  drm_ioctl_kernel+0x64/0xb0
[ 3194.147298]  drm_ioctl+0x2cd/0x390
[ 3194.150710]  ? i915_gem_execbuffer+0x2b0/0x2b0 [i915]
[ 3194.155741]  ? finish_task_switch+0xa5/0x210
[ 3194.159993]  ? finish_task_switch+0x6a/0x210
[ 3194.164247]  do_vfs_ioctl+0x90/0x670
[ 3194.167806]  ? entry_SYSCALL_64_fastpath+0x5/0xb1
[ 3194.172492]  ? __this_cpu_preempt_check+0x13/0x20
[ 3194.177176]  ? trace_hardirqs_on_caller+0xe7/0x1c0
[ 3194.181946]  SyS_ioctl+0x3c/0x70
[ 3194.185159]  entry_SYSCALL_64_fastpath+0x1c/0xb1
[ 3194.189756] RIP: 0033:0x7fd2b76a8587
[ 3194.193314] RSP: 002b:00007fff074845b8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[ 3194.200855] RAX: ffffffffffffffda RBX: ffffffff8146da43 RCX: 00007fd2b76a8587
[ 3194.207962] RDX: 00007fff074846e0 RSI: 0000000040406469 RDI: 0000000000000003
[ 3194.215068] RBP: ffffc9000176bf88 R08: 0000000000000000 R09: 0000000000000003
[ 3194.222175] R10: 00007fd2b796bb58 R11: 0000000000000246 R12: 00007fff07484880
[ 3194.229280] R13: 0000000000000003 R14: 0000000040406469 R15: 0000000000000000
[ 3194.236386]  ? __this_cpu_preempt_check+0x13/0x20
[ 3194.241070] Code: 24 b0 00 00 00 48 85 c9 0f 84 6c ff ff ff 8b 41 20 85 c0 7e 73 83 e8 01 89 41 20 41 8b 84 24 e8 00 00 00 a8 0f 0f 85 5f ff ff ff <0f> 0b 48 83 c4 08 5b 41 5c 41 5d 41 5e 41 5f 5d f3 c3 49 8b 84
[ 3194.259943] RIP: eb_release_vmas.isra.6+0x119/0x180 [i915] RSP: ffffc9000176ba80
[ 3194.268047] ---[ end trace 1d7348c6575d8800 ]---
[ 3673.658819] softdog: Initiating panic
[ 3673.662471] Kernel panic - not syncing: Software Watchdog Timer expired
[ 3673.669066] Kernel Offset: disabled
[ 3673.672541] Rebooting in 1 seconds..

Reported-by: Tomi Sarvela <tomi.p.sarvela@intel.com>
Fixes: 2889caa92321 ("drm/i915: Eliminate lots of iterations over the execobjects array")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721145037.25105-3-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Remove assertion from raw __i915_vma_unpin()

After we detect a i915_vma pin overflow, we call __i915_vma_unpin to
cleanup. However, on an overflow the pin_count bitfield will be zero,
triggering an assertion, even though we the intention is to merely warn
and report the error back to the user (as historically the culprit has
be a leak in the display code).

Fixes: 20dfbde463c8 ("drm/i915: Wrap vma->pin_count accessors with small inline helpers")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721145037.25105-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/cnl: Fix loadgen select programming on ddi vswing sequence

The condition for setting the Loadgen Select bit of
PORT_TX_DW4 register during DDI Vswing Sequence should be
Bit rate <=6 GHz whereas the existing code checks only
Bit Rate < 6GHz. This patch fixes this condition.
While at it also remove the redundant paranthesis.

Fixes: cf54ca8bc567 ("drm/i915/cnl: Implement voltage swing sequence.")
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500329122-32662-1-git-send-email-manasi.d.navare@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/fbc: add comments to the FBC auxiliary structs

I wrote this code an year and a half ago and I couldn't exactly
remember the main differences of these two structures when reviewing a
new FBC patch. Add some comments to help explain what's the purpose of
each struct.

For the record, the original commits are:
 b183b3f14395 ("drm/i915/fbc: introduce struct intel_fbc_reg_params")
 aaf78d276ba0 ("drm/i915/fbc: introduce struct intel_fbc_state_cache")

Cc: Praveen Paneri <praveen.paneri@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170714193822.12121-1-paulo.r.zanoni@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: cleanup the CHICKEN_MISC_2 (re)definitions

* Don't define it twice.
* Define MSBs first, like the rest of i915_reg.h.
* Add CNL_ prefix to the bit that arrived in CNL.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170714175228.27019-1-paulo.r.zanoni@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/selftests: Fix kbuild error

After applying af2788925ae0 ("drm/i915: Squelch reset messages during
selftests") out of sequence, I missed fixing up a call to i915_reset().

Reported-by: kbuild test robot <kbuild-all@01.org>
Fixes: af2788925ae0 ("drm/i915: Squelch reset messages during selftests")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20170725125336.11969-1-chris@chris-wilson.co.uk
Reviewed-by: David Weinehall <david.weinehall@linux.intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Squelch reset messages during selftests

During our selftests, we try reseting the GPU tens of thousands of
times, flooding the dmesg with our reset spam drowning out any potential
warnings. Add an option to i915_reset()/i915_reset_engine() to specify a
quiet reset for selftesting.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-19-chris@chris-wilson.co.uk
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/selftest: Refactor reset locking

Extract the common barrier against rogue hangchecks from disrupting our
direct testing of resets, and in the process expand the lock to include
the per-engine reset shortcuts.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-17-chris@chris-wilson.co.uk
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Don't touch fence->error when resetting an innocent request

If the request has been completed before the reset took effect, we don't
need to mark it up as being a victim. Touching fence->error after the
fence has been signaled is detected by dma_fence_set_error() and
triggers a BUG:

[  231.743133] kernel BUG at ./include/linux/dma-fence.h:434!
[  231.743156] invalid opcode: 0000 [#1] SMP KASAN
[  231.743172] Modules linked in: i915 drm_kms_helper drm iptable_nat nf_nat_ipv4 nf_nat x86_pkg_temp_thermal iosf_mbi i2c_algo_bit cfbfillrect syscopyarea cfbimgblt sysfillrect sysimgblt fb_sys_fops cfbcopyarea fb font fbdev [last unloaded: drm]
[  231.743221] CPU: 2 PID: 20 Comm: kworker/2:0 Tainted: G     U          4.13.0-rc1+ #52
[  231.743236] Hardware name: Hewlett-Packard HP EliteBook 8460p/161C, BIOS 68SCF Ver. F.01 03/11/2011
[  231.743363] Workqueue: events_long i915_hangcheck_elapsed [i915]
[  231.743382] task: ffff8801f42e9780 task.stack: ffff8801f42f8000
[  231.743489] RIP: 0010:i915_gem_reset_engine+0x45a/0x460 [i915]
[  231.743505] RSP: 0018:ffff8801f42ff770 EFLAGS: 00010202
[  231.743521] RAX: 0000000000000007 RBX: ffff8801bf6b1880 RCX: ffffffffa02881a6
[  231.743537] RDX: dffffc0000000000 RSI: dffffc0000000000 RDI: ffff8801bf6b18c8
[  231.743551] RBP: ffff8801f42ff7c8 R08: 0000000000000001 R09: 0000000000000000
[  231.743566] R10: 0000000000000000 R11: 0000000000000000 R12: ffff8801edb02d00
[  231.743581] R13: ffff8801e19d4200 R14: 000000000000001d R15: ffff8801ce2a4000
[  231.743599] FS:  0000000000000000(0000) GS:ffff8801f5a80000(0000) knlGS:0000000000000000
[  231.743614] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  231.743629] CR2: 00007f0ebd1add10 CR3: 0000000002621000 CR4: 00000000000406e0
[  231.743643] Call Trace:
[  231.743752]  i915_gem_reset+0x6c/0x150 [i915]
[  231.743853]  i915_reset+0x175/0x210 [i915]
[  231.743958]  i915_reset_device+0x33b/0x350 [i915]
[  231.744061]  ? valleyview_pipestat_irq_handler+0xe0/0xe0 [i915]
[  231.744081]  ? trace_hardirqs_off_caller+0x70/0x110
[  231.744102]  ? _raw_spin_unlock_irqrestore+0x46/0x50
[  231.744120]  ? find_held_lock+0x119/0x150
[  231.744138]  ? mark_lock+0x6d/0x850
[  231.744241]  ? gen8_gt_irq_ack+0x1f0/0x1f0 [i915]
[  231.744262]  ? work_on_cpu_safe+0x60/0x60
[  231.744284]  ? rcu_read_lock_sched_held+0x57/0xa0
[  231.744400]  ? gen6_read32+0x2ba/0x320 [i915]
[  231.744506]  i915_handle_error+0x382/0x5f0 [i915]
[  231.744611]  ? gen6_rps_reset_ei+0x20/0x20 [i915]
[  231.744630]  ? vsnprintf+0x128/0x8e0
[  231.744649]  ? pointer+0x6b0/0x6b0
[  231.744667]  ? debug_check_no_locks_freed+0x1a0/0x1a0
[  231.744688]  ? scnprintf+0x92/0xe0
[  231.744706]  ? snprintf+0xb0/0xb0
[  231.744820]  hangcheck_declare_hang+0x15a/0x1a0 [i915]
[  231.744932]  ? engine_stuck+0x440/0x440 [i915]
[  231.744951]  ? rcu_read_lock_sched_held+0x57/0xa0
[  231.745062]  ? gen6_read32+0x2ba/0x320 [i915]
[  231.745173]  ? gen6_read16+0x320/0x320 [i915]
[  231.745284]  ? intel_engine_get_active_head+0x91/0x170 [i915]
[  231.745401]  i915_hangcheck_elapsed+0x3d8/0x400 [i915]
[  231.745424]  process_one_work+0x3e8/0xac0
[  231.745444]  ? pwq_dec_nr_in_flight+0x110/0x110
[  231.745464]  ? do_raw_spin_lock+0x8e/0x120
[  231.745484]  worker_thread+0x8d/0x720
[  231.745506]  kthread+0x19e/0x1f0
[  231.745524]  ? process_one_work+0xac0/0xac0
[  231.745541]  ? kthread_create_on_node+0xa0/0xa0
[  231.745560]  ret_from_fork+0x27/0x40
[  231.745581] Code: 8b 7d c8 e8 49 0d 02 e1 49 8b 7f 38 48 8b 75 b8 48 83 c7 10 e8 b8 89 be e1 e9 95 fc ff ff 4c 89 e7 e8 4b b9 ff ff e9 30 ff ff ff <0f> 0b 0f 1f 40 00 55 48 89 e5 41 57 41 56 41 55 41 54 49 89 fe
[  231.745767] RIP: i915_gem_reset_engine+0x45a/0x460 [i915] RSP: ffff8801f42ff770

At first glance this looks to be related to commit c64992e035d7
("drm/i915: Look for active requests earlier in the reset path"), but it
could easily happen before as well. On the other hand, we no longer
logged victims due to the active_request being dropped earlier.

v2: Be trickier to unwind the incomplete request as we cannot rely on
request retirement for the lockless per-engine reset.
v3: Reprobe the active request at the time of the reset.

Reported-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fixes: c64992e035d7 ("drm/i915: Look for active requests earlier in the reset path")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-15-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> #v1
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Enforce that CS packets are qword aligned

We require the caller to ensure that the packets they wish to emit into
the CS ring are qword aligned (i.e. have an even number of dwords).
Double check this.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721161101.1618-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/glk: set HDMI 2.0 identifier

This patch sets the is_hdmi2_src identifier in drm connector
for GLK platform. GLK contains a native HDMI 2.0 controller.
This identifier will help the EDID handling functions to save
lot of work which is specific to HDMI 2.0 sources.

V3: Added this patch
V4: Rebase
V4: Rebase
V5: Added r-b from Ander
V6: Rebase
V7: Rebase
V8: Rebase
V9: Added r-b from Ville
V9: Added r-b from Imre

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500650709-14447-7-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: set colorspace for YCBCR420 outputs

When output colorspace is YCBCR420, we have to load the
corresponding colorspace in AVI infoframe. This patch fills
the colorspace of AVI infoframe as per the output mode.

V2: Rebase
V3: Rebase
V4: Rebase
V5: Added r-b from Ander
V6: Checking RGB/YCBCR420 output only (Ville)
V7: Add colorspace info in driver(not drm layer) (Ville)
V8: Rebase
V9: Added r-b from Ville
V10: Added r-b from Imre

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500650709-14447-6-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: prepare csc unit for YCBCR420 output

To support ycbcr output, we need a pipe CSC block to do
RGB->YCBCR conversion.

Current Intel platforms have only one pipe CSC unit, so
we can either do color correction using it, or we can perform
RGB->YCBCR conversion.

This function adds a csc handler, which uses recommended bspec
values to perform RGB->YCBCR conversion (target color space BT709)

V2: Rebase
V3: Rebase
V4: Rebase
V5: Addressed review comments from Ander
    - Remove extra line added in the patch
    - Add the spec details in the commit message
    - Combine two if(cond) while calling intel_crtc_compute_config
V6: Handle YCBCR420 outputs only (Ville)
V7: Addressed review comments from Ville:
    - Add description about target colorspace
    - Remove the comments from CSC function
    - DRM_DEBUG->DEBUG_KMS for atomic failure due to CSC unit busy
    - Remove unnecessary debug message about YCBCR420 possibe
V8: Addressed review comments from Ville:
    - Remove extra comment, not required.
    - Do not add extra variable for CTM, reuse pipe_config
    Added r-b from Ville
V9: Remove extra whitespace (Imre)
V10: Added r-b from Imre

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500650709-14447-5-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: prepare pipe for YCBCR420 output

To get HDMI YCBCR420 output, the PIPEMISC register should be
programmed to:
- Generate YCBCR output (bit 11)
- In case of YCBCR420 outputs, it should be programmed in full
  blend mode to use the scaler in 5x3 ratio (bits 26 and 27)

This patch:
- Adds definition of these bits.
- Programs PIPEMISC for YCBCR420 outputs.
- Adds readouts to compare HW and SW states.

V2: rebase
V3: rebase
V4: rebase
V5: added r-b from Ander
V6: Handle only YCBCR420 outputs (ville)
V7: rebase
V8: Addressed review comments from Ville
    - Add readouts for state->ycbcr420 and 420 pixel_clock.
    - Handle warning due to mismatch in clock for ycbcr420 clock.
    - Rename PIPEMISC macros to match the Bspec.
    - Add a debug print stating if YCBCR 4:2:0 output enabled.
    Added r-b from Ville
V9: Addressed review comments from Imre:
    - Add 420 mode clock adjustment in intel_hdmi_mode_valid to
      prevent 420_only modes getting rejected for high clock.
    - Add port clock adjustment for ycbcr420 modes in ddi_get_clock
    - Rename macros as per Ville's suggestion.
    - Remove unnecessary wl changes.
V10: Added r-b from Imre
V11: Fixed faulty dotclock handling, and addressed missing comment
     from previous set of review comments (Imre)
V12: Fixed dotclock for 12bpc too, removed 420 check for GEN < 10

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500904172-31717-1-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: prepare scaler for YCBCR420 modeset

To get a YCBCR420 output from intel platforms, we need one
scaler to scale down YCBCR444 samples to YCBCR420 samples.

This patch:
- Does scaler allocation for HDMI ycbcr420 outputs.
- Programs PIPE_MISC register for ycbcr420 output.

V2: rebase
V3: rebase
V4: rebase
V5: addressed review comments from Ander:
    - No need to check both scaler_user && hdmi_output.
      Check for scaler_user is enough.
V6: rebase
V7: Do not create a new scaler user, use existing pipe scaler user.
V8: rebase
V9: Addressed review comments from Ville:
    - Remove leftover comment for HDMI scaler user.
    - Remove unnecessary blank line.
    - Make scaler alocation failure a DEBUG log instead of ERROR.
    Added r-b from Ville
V10: Update commit message as per latest code (Imre)
     Added r-b from Imre

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Ander Conselvan De Oliveira <conselvan2@gmail.com>
Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500650709-14447-3-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: add config function for YCBCR420 outputs

This patch checks encoder level support for YCBCR420 outputs.
The logic goes as simple as this:
If the input mode is YCBCR420-only mode: prepare HDMI for
YCBCR420 output, else continue with RGB output mode.

It checks if the mode is YCBCR420 and source can support this
output then it marks the ycbcr_420 output indicator into crtc
state, for further staging in driver.

V2: Split the patch into two, kept helper functions in DRM layer.
V3: Changed the compute_config function based on new DRM API.
V4: Rebase
V5: Rebase
V6: Check and handle YCBCR420-only modes, discard the property
    based approach (Ville)
V7: Addressed review comments from Ville
    - add else case in 12BPC check.
    - extract ycbcr420 state inside hdmi_12bpc_possible function.
V8: Addressed review comments from Ville
    - Remove extra blank lines.
    - Remove "HDMI" from the description of ycbcr420 state variable.
    - Remove local variable, use crtc_state->ycbcr420 instead.
    Added r-b from Ville.
V9: Rebase
V10: Added r-b from Imre

Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1500650709-14447-2-git-send-email-shashank.sharma@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Gather all the power well->domain mappings to one place

Shuffle the power well->domain mapping macros around so they are at one
place in old->new GEN order.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-19-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Move hsw_power_well_enable() next to the rest of HSW helpers

Move the helper next to the rest of HSW specific code.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-18-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/gen9+: Unify the HSW/BDW and GEN9+ power well helpers

After the previous refactorings the HSW/BDW and GEN9+ power well helpers
are practically identical, so use the HSW power well helpers for GEN9+
too. This means using the HSW power well ops instead of the SKL one and
setting the irq_pipe_mask, has_vga and has_fuses attributes as needed.

v2:
- Rebased on v2 of patch 15.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-7-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw+: Add has_fuses power well attribute

The pattern of a power well backing a set of fuses whose initialization
we need to wait for during power well enabling is common to all GEN9+
platforms. Adding support for this to the HSW power well enable helper
allows us to use the HSW/BDW power well code for GEN9+ as well in a
follow-up patch.

v2:
- Use an enum for power gates instead of raw numbers. (Ville)

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-6-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw, bdw: Wait for the power well disabled state

Similarly to GEN9+ waiting for the power well disabled state is a safer
option and also provides diagnostic info if the disabling didn't succeed
or the power well was forced on by an external requester. While at it
also use the existing GEN9+ helper to wait for the enabled state.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-15-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw, bdw: Add irq_pipe_mask, has_vga power well attributes

The pattern of a power well backing a set of pipe IRQ or VGA
functionality applies to all HSW+ platforms. Using power well attributes
instead of platform checks to decide whether to init/reset pipe IRQs and
VGA correspondingly is cleaner and it allows us to unify the HSW/BDW and
GEN9+ power well code in follow-up patches.

Also use u8 for pipe_mask in related helpers to match the type in the
power well struct.

v2:
- Use u8 instead of u32 for irq_pipe_mask. (Ville)

v3:
- Use u8 for pipe_mask in related helpers too for clarity.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170712155413.29839-1-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw+: Unify the hsw/bdw and gen9+ power well req/state macros

Although on HSW/BDW there is only a single display global power well,
it's programmed the same way as other GEN9+ power wells. This also
means we can get at the HSW/BDW request and status flags the same way
it's done on GEN9+ by assigning the corresponding HSW/BDW power well ID.
This ID was assigned in a recent patch, so we can now switch to using
the same macros everywhere on HSW+.

Updating the HSW power well control register with RMW is not strictly
necessary, but this will allow us to use the same code for GEN9+.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-13-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw, bdw: Split power well set to enable/disable helpers

We can reduce the code indentation by splitting the set helper to
separate enable/disable helpers. This also allows us to unify the
HSW/BDW and GEN9+ power well ops in follow-up patches, which introduces
some differences between the enable and disable helpers.

While at it also remove the redundant enable/disable debug messages,
the same info is printed already elsewhere.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-12-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw, bdw: Remove redundant state check during power well toggling

Similarly to the GEN9 power well toggling, saving an occasional extra
MMIO write is not worth the code complexity, let's simplify things.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-11-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/gen9+: Remove redundant state check during power well toggling

Atm we enable/disable a power well only if it wasn't already
enabled/disabled respectively. The only reason for this I can think of
is to save the extra MMIO writes. Since the HW state matches the power
well's usage counter most of the time the overhead due to these MMIOs is
insignificant. Let's simplify the code by making the writes
unconditional.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-10-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/gen9+: Remove redundant power well state assert during enabling

We check already for power wells that are unexpectedly on (or forced on)
during power well disabling. Those checks also account for other
power well requesters like KVMR or DEBUG. As such this check is
redundant, let's remove it to simplify things.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-9-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/bxt, glk: Give a proper name to the power well struct phy field

Follow-up patches will add new fields to the i915_power_well struct that
are specific to the hsw_power_well_ops helpers. Prepare for this by
changing the generic 'data' field to a union of platform specific
structs.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-8-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Check for duplicated power well IDs

Check that all the power well IDs are unique on the given platform.

v2:
- Fix using BIT_ULL() instead of BIT() for 64 bit mask.
v3:
- Move the check to a separate function. (Ville)

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-4-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/hsw, bdw: Add an ID for the global display power well

Add an ID for the HSW/BDW global display power well for consistency. The
ID is selected so that it can be used to get at the HW request and
status flags with the corresponding GEN9+ macros. Unifying the HSW/BDW
and GEN9+ versions of these macros and the power well ops using them
will be done in follow-up patches.

v2:
- Rebased on v2 of patch 2.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-3-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/gen2: Add an ID for the display pipes power well

Make the I830 power well ID assignment explicit for consistency.

v2:
- s/GEN2/I830/ in the comment, since other GEN2s don't have the power
  well. (Ville)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-2-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Assign everywhere the always-on power well ID

Power well IDs are used for lookup so they must be unique. To ensure
this assign the always-on power well ID everywhere where it's missing.
This didn't cause a problem so far, since we didn't need to look up
power wells that happened to share their IDs.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-4-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Unify power well ID enums

Atm, the power well IDs are defined in separate platform specific enums,
which isn't ideal for the following reasons:
- the IDs are used by helpers like lookup_power_well() in a platform
  independent way
- the always-on power well is used by multiple platforms and so needs
  now separate IDs, although these IDs refer to the same thing

To make things more consistent use a single enum instead of the two
separate ones, listing the IDs per platform (or set of very similar
platforms like all GEN9/10). Replace the separate always-on power
well IDs with a single ID.

While at it also add a note clarifying the distinction between regular
power wells that follow a common programming pattern and custom ones
that are programmed in some other way. The IDs for regular power wells
need to stay fixed, since they also define the request and state HW flag
positions in their corresponding power well control register(s).

v2:
- Add comment about id to req,status bit mapping to the enum. (Rodrigo)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170711204236.5618-1-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/chv: Add unique power well ID for the pipe A power well

The power well IDs are used for lookup, so they must be unique on a
given platform; ensure this on CHV. This didn't cause an actual problem
since we didn't need to look up power wells which happened to share an
ID.

Mark this new power well as custom, since its programming pattern
doesn't follow that of the rest of VLV/CHV power wells.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/1499352040-8819-2-git-send-email-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Simplify scaler init during CRTC HW readout

The crtc state starts out being bzero'd, so no need to clear
scaler_users. Also intel_crtc_init_scalers() knows already which
platforms have scalers, so no need for the platform check here.
Similarly intel_crtc_init_scalers() will init scaler_id as required,
so no need to do it here separately.

Cc: Chandra Konduru <chandra.konduru@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170719225057.20131-2-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Fix scaler init during CRTC HW state readout

The scaler allocation code depends on a non-zero default value for the
crtc scaler_id, so make sure we initialize the scaler state accordingly
even if the crtc is off. This fixes at least an initial YUV420 modeset
(added in a follow-up patchset by Shashank) when booting with the screen
off: after the initial HW readout and modeset which enables the scaler a
subsequent modeset will disable the scaler which isn't properly
allocated. This results in a funky HW state where the pipe scaler HW
registers can't be modified and the normally black screen is grey and
shifted to the right or jitters.

The problem was revealed by Shashank's YUV420 patchset and first
reported by Ville.

v2:
- In the stable tag also include versions which need backporting (Jani)

Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Chandra Konduru <chandra.konduru@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: <stable@vger.kernel.org> # 4.2.x
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: a1b2278e4dfc ("drm/i915: skylake panel fitting using shared scalers")
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Mahesh Kumar <mahesh1.kumar@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170720112820.26816-1-imre.deak@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915/selftests: Exercise independence of per-engine resets

If all goes well, resetting one engine should not affect the operation of
any others. So to test this, we setup a continuous stream of requests
onto to each of the "innocent" engines whilst constantly resetting our
target engine.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-16-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Disable per-engine reset for Broxton

Triggering a GPU reset for one engine affects another, notably
corrupting the context status buffer (CSB) effectively losing track of
inflight requests.

Adding a few printks:
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index ad41836fa5e5..a969456bc0fa 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -1953,6 +1953,7 @@ int i915_reset_engine(struct intel_engine_cs *engine)
                goto out;
        }

+       pr_err("Resetting %s\n", engine->name);
        ret = intel_gpu_reset(engine->i915, intel_engine_flag(engine));
        if (ret) {
                /* If we fail here, we expect to fallback to a global reset */
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 716e5c9ea222..a72bc35d0870 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -355,6 +355,7 @@ static void execlists_submit_ports(struct intel_engine_cs *engine)
                                execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_IN);
                        port_set(&port[n], port_pack(rq, count));
                        desc = execlists_update_context(rq);
+                       pr_err("%s: in (rq=%x) ctx=%d\n", engine->name, rq->global_seqno, upper_32_bits(desc));
                        GEM_DEBUG_EXEC(port[n].context_id = upper_32_bits(desc));
                } else {
                        GEM_BUG_ON(!n);
@@ -594,9 +595,23 @@ static void intel_lrc_irq_handler(unsigned long data)
                        if (!(status & GEN8_CTX_STATUS_COMPLETED_MASK))
                                continue;

+                       pr_err("%s: out CSB (%x head=%d, tail=%d), ctx=%d, rq=%d\n",
+                                       engine->name,
+                                       readl(csb_mmio),
+                                       head, tail,
+                                       readl(buf+2*head+1),
+                                       port->context_id);
+
                        /* Check the context/desc id for this event matches */
-                       GEM_DEBUG_BUG_ON(readl(buf + 2 * head + 1) !=
-                                        port->context_id);
+                       if (readl(buf + 2 * head + 1) != port->context_id) {
+                               pr_err("%s: BUG CSB (%x head=%d, tail=%d), ctx=%d, rq=%d\n",
+                                               engine->name,
+                                               readl(csb_mmio),
+                                               head, tail,
+                                               readl(buf+2*head+1),
+                                               port->context_id);
+                               BUG();
+                       }

                        rq = port_unpack(port, &count);
                        GEM_BUG_ON(count == 0);

Results in:

[ 6423.006602] Resetting rcs0
[ 6423.009080] rcs0: in (rq=fffffe70) ctx=1
[ 6423.009216] rcs0: in (rq=fffffe6f) ctx=3
[ 6423.009542] rcs0: out CSB (2 head=1, tail=2), ctx=3, rq=3
[ 6423.009619] Resetting bcs0
[ 6423.009980] rcs0: BUG CSB (0 head=1, tail=2), ctx=0, rq=3

Note that this bug may be affect all machines and not just Broxton,
Broxton is just the first machine on which I have confirmed this bug.

Fixes: 142bc7d99bcf ("drm/i915: Modify error handler for per engine hang recovery")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Acked-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-13-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Emit a user level message when resetting the GPU (or engine)

Although a banned context will be told to -EIO off if they try to submit
more requests, we have a discrepancy between whole device resets and
per-engine resets where we report the GPU reset but not the engine
resets. This leaves a bit of mystery as to why the context was banned,
and also reduces awareness overall of when a GPU (engine) reset occurs
with its possible side-effects.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-13-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Make i915_gem_context_mark_guilty() safe for unlocked updates

Since we make call i915_gem_context_mark_guilty() concurrently when
resetting different engines in parallel, we need to make sure that our
updates are safe for the unlocked access.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-12-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Clear engine irq posted following a reset

When the GPU is reset, we want to discard all pending notifications as
either we have manually completed them, or they are no longer
applicable. Make sure we do reset the engine->irq_posted prior to
re-enabling the engine (e.g. the interrupt tasklets) in
i915_gem_reset_finish_engine().

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-11-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Assert that machine is wedged for nop_submit_request

We should only ever do nop_submit_request when the machine is wedged, so
assert it is so.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-10-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Wake up waiters after setting the WEDGED bit

After setting the WEDGED bit, make sure that we do wake up waiters as
they may not be waiting for a request completion yet, just for its
execution.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-9-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Move idle checks before intel_engine_init_global_seqno()

intel_engine_init_globa_seqno() may be called from an uncontrolled
set-wedged path where we have given up waiting for broken hw and declare
it defunct. Along that path, any sanity checks that the hw is idle
before we adjust its state will expectedly fail, so we simply cannot.
Instead of asserting inside init_global_seqno, we move them to the
normal caller reset_all_global_seqno() as it handles runtime seqno
wraparound.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-8-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Clear execlist port[] before updating seqno on wedging

When we wedge the device, we clear out the in-flight requests and
advance the breadcrumb to indicate they are complete. However, the
breadcrumb advance includes an assert that the engine is idle, so that
advancement needs to be the last step to ensure we pass our own sanity
checks.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-7-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Check the execlist queue for pending requests before declaring idle

Including a check against the execlist queue before calling the engine
idle and passing hangcheck.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-6-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Check execlist/ring status during hangcheck

Before we declare an engine as idle, check if there are any pending
execlist context-switches and if the ring itself reports as idle.
Otherwise, we may be left in a situation where we miss a crucial
execlist event (or something more sinister) yet the requests complete.
Since the seqno write happens, we believe the engine to be truly idle.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-5-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Flush the execlist ports if idle

When doing a GPU reset, the CSB register will be trashed and we will
lose any context-switch notifications that happened since the tasklet
was disabled. If we find that all requests on this engine were
completed, we want to make sure that the ELSP tracker is similarly empty
so that we do not feed back in the completed requests upon recovering
from the reset.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-4-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Serialize per-engine resets against new requests

We rely on disabling the execlists (by stopping the tasklet) to prevent
new requests from submitting to the engine ELSP before we are ready.
However, we re-enable the engine before we call init_hw which gives
userspace the opportunity to subit a new request which is then
overwritten by init_hw -- but not before the HW may have started
executing. The subsequent out-of-order CSB is detected by our sanity
checks in intel_lrc_irq_handler().

Fixes: a1ef70e14453 ("drm/i915: Add support for per engine reset recovery")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Michel Thierry <michel.thierry@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-3-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Reset context image on engines after triggering the reset

We try to fixup the context image after the reset to ensure that there
are no more pending writes from the hw that may conflict and to fixup
any that were in flight.

Fixes: a1ef70e14453 ("drm/i915: Add support for per engine reset recovery")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-2-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/i915: Report execlists irq bit in debugfs

As part of the knowing whether there is outstanding data in the CSB,
also check whether there is an outstanding IRQ notification.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170721123238.16428-1-chris@chris-wilson.co.uk
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Merge airlied/drm-next into drm-intel-next-queued

Resync with upstream to avoid git getting too badly confused. Also, we
have a conflict with the drm_vblank_cleanup removal, which cannot be
resolved by simply taking our side. Bake that in properly.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm: linux-next: build failure after merge of the drm-misc tree

Hi all,

After merging the drm-misc tree, today's linux-next build (x86_64
allmodconfig) failed like this:

drivers/staging/vboxvideo/vbox_drv.c:235:2: error: unknown field 'set_busid' specified in initializer
  .set_busid = drm_pci_set_busid,
  ^
drivers/staging/vboxvideo/vbox_drv.c:235:15: error: 'drm_pci_set_busid' undeclared here (not in a function)
  .set_busid = drm_pci_set_busid,
               ^
drivers/staging/vboxvideo/vbox_drv.c: In function 'vbox_init':
drivers/staging/vboxvideo/vbox_drv.c:273:9: error: implicit declaration of function 'drm_pci_init' [-Werror=implicit-function-declaration]
  return drm_pci_init(&driver, &vbox_pci_driver);
         ^
drivers/staging/vboxvideo/vbox_drv.c: In function 'vbox_exit':
drivers/staging/vboxvideo/vbox_drv.c:278:2: error: implicit declaration of function 'drm_pci_exit' [-Werror=implicit-function-declaration]
  drm_pci_exit(&driver, &vbox_pci_driver);
  ^

Caused by commits

  5c484cee7ef9 ("drm: Remove drm_driver->set_busid hook")
  10631d724def ("drm/pci: Deprecate drm_pci_init/exit completely")

interacting with commit

  dd55d44f4084 ("staging: vboxvideo: Add vboxvideo to drivers/staging")

from the staging.current tree.

I have applied the following merge fix patch - please check that it
is correct.

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Wed, 19 Jul 2017 11:41:01 +1000
Subject: [PATCH] drm: fixes for staging due to API changes in the drm core

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Backmerge tag 'v4.13-rc2' into drm-next

Linux 4.13-rc2

This is required for drm-misc fixing.

Merge tag 'drm-misc-next-2017-07-26' of git://anongit.freedesktop.org/git/drm-misc into drm-next

drm-misc-next-2017-07-18:
Core Changes:
- A couple fixes to only opening crc when needed (Maarten)
- Change atomic helper swap_state to be interruptible (Maarten)
- fb_helper: Support waiting for an output before setting up (Daniel)
- Allow drivers supporting runtime_pm to use helper_commit_tail (Maxime)

Driver Changes:
- misc: Use %pOF to print device node names (Rob)
- Miscellaneous fixes

drm-misc-next-2017-07-18:
UAPI Changes:
- Fail commits which request an event without including a crtc (Andrey)

Core Changes:
- Add YCBCR 4:2:0 support (Shashank)
- s/drm_atomic_replace_property_blob/drm_property_replace_blob/ (Peter)
- Add proper base class for private objs instead of using void* (Ville)
- Remove pending_read/write_domains from drm_gem_object (Chris)
- Add async plane update support (ie: cursor) to atomic helpers (Gustavo)
- Add old state to .enable and rename to .atomic_enable (Laurent)
- Add drm_atomic_helper_wait_for_flip_done() (Boris)
- Remove drm_driver->set_busid hook (Daniel)
- Migrate vblank documentation into the source files (Daniel)
- Add fb_helper->lock instead of abusing modeset lock (Thierry/Daniel)

Driver Changes:
- stm: Add STM32 DSI controller driver (Phillipe)
- amdgpu: Numerous small/misc fixes
- bridge: Add Synopsys Designware MIPI DSI host bridge driver (Phillipe)
- tinydrm: Add support for Pervasive Displays RePaper displays (Noralf)
- misc: Replace for_each_[obj]_in_state to prep for removal (Maarten)
- misc: Use .atomic_disable for atomic drivers (Laurent)
- vgem: Pin pages when mapped/exported (Chris)
- dw_hdmi: Add support for Rockchip RK3399 (Mark)
- atmel-hlcdc: Add 8-bit color look-up table format (Peter)
- vc4: Send vblank event when disabling a crtc (Boris)
- vc4: Use atomic helpers for fence waits (Eric)
- misc: drop drm_vblank_cleanup cargo-cult (Daniel)

Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Boris Brezillon <boris.brezillon@free-electrons.com>
Cc: Eric Anholt <eric@anholt.net>
Cc: Peter Rosin <peda@axentia.se>
Cc: Mark Yao <mark.yao@rock-chips.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Andrey Grodzovsky <Andrey.Grodzovsky@amd.com>
Cc: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Cc: Gustavo Padovan <gustavo.padovan@collabora.com>
Cc: Thierry Reding <treding@nvidia.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Peter Rosin <peda@axentia.se>
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Philippe CORNU <philippe.cornu@st.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
* tag 'drm-misc-next-2017-07-26' of git://anongit.freedesktop.org/git/drm-misc: (171 commits)
  drm/hisilicon: fix build error without fbdev emulation
  drm/atomic: implement drm_atomic_helper_commit_tail for runtime_pm users
  drm: Improve kerneldoc for drm_modeset_lock
  drm/hisilicon: Remove custom FB helper deferred setup
  drm/exynos: Remove custom FB helper deferred setup
  drm/fb-helper: Support deferred setup
  dma-fence: Don't BUG_ON when not absolutely needed
  drm: Convert to using %pOF instead of full_name
  drm/syncobj: Fix kerneldoc
  drm/atomic: Allow drm_atomic_helper_swap_state to fail
  drm/atomic: Add __must_check to drm_atomic_helper_swap_state.
  drm/vc4: Handle drm_atomic_helper_swap_state failure
  drm/tilcdc: Handle drm_atomic_helper_swap_state failure
  drm/tegra: Handle drm_atomic_helper_swap_state failure
  drm/msm: Handle drm_atomic_helper_swap_state failure
  drm/mediatek: Handle drm_atomic_helper_swap_state failure
  drm/i915: Handle drm_atomic_helper_swap_state failure
  drm/atmel-hlcdc: Handle drm_atomic_helper_swap_state failure
  drm/nouveau: Handle drm_atomic_helper_swap_state failure
  drm/atomic: Change drm_atomic_helper_swap_state to return an error.
  ...

drm/hisilicon: fix build error without fbdev emulation

We cannot reference priv->fbdev outside of the #ifdef:

drivers/net/virtio_net.c:1881:12: error: 'virtnet_restore_up' defined but not used [-Werror=unused-function]
 static int virtnet_restore_up(struct virtio_device *vdev)
drivers/net/virtio_net.c:1859:13: error: 'virtnet_freeze_down' defined but not used [-Werror=unused-function]
 static void virtnet_freeze_down(struct virtio_device *vdev)

As the #ifdef is a bit annoying here, this removes it entirely
and uses an IS_ENABLED() check in it place where needed.

Fixes: b4dd9f1ffaba ("drm/hisilicon: Remove custom FB helper deferred setup")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
[danvet: One step further, also remove the IS_ENABLED checks, core
no-ops out the fb helper functions that the cma helpers use. Discussed
with Arnd on dri-devel.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170725180555.3699056-1-arnd@arndb.de
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/atomic: implement drm_atomic_helper_commit_tail for runtime_pm users

The current drm_atomic_helper_commit_tail helper works only if the CRTC is
accessible, and documents an alternative implementation that is supposed to
be used if that happens.

That implementation is then duplicated by some drivers. Instead of
documenting it, let's implement an helper that all the relevant users can
use directly.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/a8f92dc70048bab746e94dadd1c23200626aff60.1500555652.git-series.maxime.ripard@free-electrons.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm: Improve kerneldoc for drm_modeset_lock

Explain better when the drm_modeset_acquire_ctx parameter can
be skipped for drm_modeset_lock() call.

Signed-off-by: Liviu Dudau <Liviu.Dudau@arm.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170720160748.12856-1-Liviu.Dudau@arm.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/hisilicon: Remove custom FB helper deferred setup

The FB helper core now supports deferred setup, so the driver's custom
implementation can be removed.

v2: Dont' resurrect drm_vblank_cleanup.

Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
Cc: Rongrong Zou <zourongrong@gmail.com>
Cc: Xinwei Kong <kong.kongxinwei@hisilicon.com>
Cc: Chen Feng <puck.chen@hisilicon.com>
Signed-off-by: Thierry Reding <treding@nvidia.com> (v1)
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170706130023.28417-5-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/exynos: Remove custom FB helper deferred setup

The FB helper core now supports deferred setup, so the driver's custom
implementation can be removed.

v2: Drop NULL check, drm_fb_helper_hotplug_event handles that already.

Cc: Inki Dae <inki.dae@samsung.com>
Cc: Joonyoung Shim <jy0922.shim@samsung.com>
Cc: Seung-Woo Kim <sw0312.kim@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Thierry Reding <treding@nvidia.com> (v1)
Reviewed-by: Inki Dae <inki.dae@samsung.com>
Tested-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170706130023.28417-4-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/fb-helper: Support deferred setup

FB helper code falls back to a 1024x768 mode if no outputs are connected
or don't report back any modes upon initialization. This can be annoying
because outputs that are added to FB helper later on can't be used with
FB helper if they don't support a matching mode.

The fallback is in place because VGA connectors can happen to report an
unknown connection status even when they are in fact connected.

Some drivers have custom solutions in place to defer FB helper setup
until at least one output is connected. But the logic behind these
solutions is always the same and there is nothing driver-specific about
it, so a better alterative is to fix the FB helper core and add support
for all drivers automatically.

This patch adds support for deferred FB helper setup. It checks all the
connectors for their connection status, and if all of them report to be
disconnected marks the FB helper as needing deferred setup. Whet setup
is deferred, the FB helper core will automatically retry setup after a
hotplug event, and it will keep trying until it succeeds.

v2: Rebase onto my entirely reworked fbdev helper locking. One big
difference is that this version again drops&reacquires the fbdev lock
(which is now fb_helper->lock, but before this patch series it was
mode_config->mutex), because register_framebuffer must be able to
recurse back into fbdev helper code for the initial screen setup.

v3: __drm_fb_helper_initial_config must hold fb_helper->lock upon
return, I've fumbled that in the deferred setup case (Liviu).

v4: I was blind, redo this all. __drm_fb_helper_initial_config
shouldn't need to reacquire fb_helper->lock, that just confuses
callers. I myself got confused by kernel_fb_helper_lock and somehow
thought it's the same as fb_helper->lock. Tsk.

Also simplify the logic a bit (we don't need two functions to probe
connectors), we can stick much closer to the existing code. And update
some comments I've spotted that are outdated.

v5: Don't pass -EAGAIN to drivers, it's just an internal error code
(Liviu).

v6: Add _and_unlock suffix to clarify locking (Maarten)

Cc: Liviu Dudau <liviu@dudau.co.uk>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Thierry Reding <treding@nvidia.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Tested-by: John Stultz <john.stultz@linaro.org>
Signed-off-by: Thierry Reding <treding@nvidia.com> (v1)
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@intel.com>
Reviewed-by: Liviu Dudau <liviu@dudau.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170706130023.28417-3-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

dma-fence: Don't BUG_ON when not absolutely needed

It makes debugging a massive pain.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Reviewed-by: Lucas Stach <l.stach@pengutronix.de>
Acked-by: Sumit Semwal <sumit.semwal@linaro.org>
Acked-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170720125107.26693-1-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm: Convert to using %pOF instead of full_name

Now that we have a custom printf format specifier, convert users of
full_name to use %pOF instead. This is preparation to remove storing
of the full path string for each node.

Signed-off-by: Rob Herring <robh@kernel.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: Inki Dae <inki.dae@samsung.com>
Cc: Joonyoung Shim <jy0922.shim@samsung.com>
Cc: Seung-Woo Kim <sw0312.kim@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: Kukjin Kim <kgene@kernel.org>
Cc: Krzysztof Kozlowski <krzk@kernel.org>
Cc: Javier Martinez Canillas <javier@osg.samsung.com>
Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
Cc: Rongrong Zou <zourongrong@gmail.com>
Cc: Xinwei Kong <kong.kongxinwei@hisilicon.com>
Cc: Chen Feng <puck.chen@hisilicon.com>
Cc: CK Hu <ck.hu@mediatek.com>
Cc: Philipp Zabel <p.zabel@pengutronix.de>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Neil Armstrong <narmstrong@baylibre.com>
Cc: Carlo Caione <carlo@caione.org>
Cc: Kevin Hilman <khilman@baylibre.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Mark Yao <mark.yao@rock-chips.com>
Cc: Heiko Stuebner <heiko@sntech.de>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Chen-Yu Tsai <wens@csie.org>
Cc: Jyri Sarha <jsarha@ti.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-samsung-soc@vger.kernel.org
Cc: linux-mediatek@lists.infradead.org
Cc: linux-amlogic@lists.infradead.org
Cc: linux-renesas-soc@vger.kernel.org
Cc: linux-rockchip@lists.infradead.org
Partially-Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Maxime Ripard <maxime.ripard@free-electrons.com>
[seanpaul changed subject prefix and fixed conflict in stm/ltdc.c]
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

Merge airlied/drm-next into drm-misc-next

I need this to be able to apply the deferred fbdev setup patches, I
need the relevant prep work that landed through the drm-intel tree.

Also squash in conflict fixup from Laurent Pinchart.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/syncobj: Fix kerneldoc

make htmldocs helps with catching these.

Cc: Dave Airlie <airlied@gmail.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20170718074113.5554-1-daniel.vetter@ffwll.ch
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/atomic: Allow drm_atomic_helper_swap_state to fail

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-13-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/atomic: Add __must_check to drm_atomic_helper_swap_state.

Now that all drivers check the return value, convert swap_state to
__must_check. This is done separately to force build warnings if we
missed a driver.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-12-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/vc4: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

VC4 has its own nonblocking modeset tracking through the vc4->async_modeset
semaphore, so it doesn't need to stall in swap_state. Pass stall = false
and BUG_ON when it returns an error. This should never happen for !stall.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Eric Anholt <eric@anholt.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-11-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/tilcdc: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Jyri Sarha <jsarha@ti.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-10-maarten.lankhorst@linux.intel.com
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/tegra: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Thierry Reding <thierry.reding@gmail.com>
Cc: Jonathan Hunter <jonathanh@nvidia.com>
Cc: linux-tegra@vger.kernel.org
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-9-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/msm: Handle drm_atomic_helper_swap_state failure

drm_atomic_helper_swap_state() will be changed to interruptible waiting
in the next few commits, so all drivers have to be changed to handling
failure.

MSM has its own busy tracking, which means the swap_state call can be
done with stall = false, in which case it should never return an error.
Handle failure with BUG_ON for this reason.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: linux-arm-msm@vger.kernel.org
Cc: freedreno@lists.freedesktop.org
Link: http://patchwork.freedesktop.org/patch/msgid/20170711143314.2148-8-maarten.lankhorst@linux.intel.com
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Tested-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>