Currently, when our driver becomes idle for i915.pc8_timeout (default:
5s) we enable PC8, so we save some power, but not everything we can.
Then, while PC8 is enabled, if we stay idle for more
autosuspend_delay_ms (default: 10s) we'll enter runtime PM and put the
graphics device in D3 state, saving even more power. The two features
are separate things with increasing levels of power savings, but if we
disable PC8 we'll never get into D3.

While from the modularity point of view it would be nice to keep these
features as separate, we have reasons to merge them:
 - We are not aware of anybody wanting a "PC8 without D3" environment.
 - If we keep both features as separate, we'll have to to test both
   PC8 and PC8+D3 code paths. We're already having a major pain to
   make QA do automated testing of just one thing, testing both paths
   will cost even more.
 - Only Haswell+ supports PC8, so if we want to add runtime PM support
   to, for example, IVB, we'll have to copy some code from the PC8
   feature to runtime PM, so merging both features as a single thing
   will make it easier for enabling runtime PM on other platforms.

This patch only does the very basic steps required to have PC8 and
runtime PM merged on a single feature: the next patches will take care
of cleaning up everything.

v2: - Rebase.
v3: - Rebase.
    - Fully remove the deprecated i915 params since Daniel doesn't
      consider them as part of the ABI.
v4: - Rebase.
    - Fix typo in the commit message.
v5: - Rebase, again.
    - Add a huge comment explaining the different forcewake usage
      (Chris, Daniel).
    - Use open-coded forcewake functions (Daniel).
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: NImre Deak <imre.deak@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When we merge PC8 and runtime PM, these new functions are going to be
called by the runtime suspend/resume functions, and their callers are
going to be removed.

v2: - Rebase

Reviewed-by: Imre Deak <imre.deak@intel.com> (v1)
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The name 'update_plane' was used both for the primary plane functions in
intel_display.c and the sprite/overlay functions in intel_sprite.c.
Rename the primary plane functions to 'update_primary_plane' to avoid
confusion.

On a similar note, intel_display.c already had a function called
intel_disable_primary_plane() that programs the hardware to disable a
pipe's primary plane.  When we hook up primary planes through the DRM
plane interface, one of the natural handler names will be
intel_primary_plane_disable(), which is very similar.  To avoid
confusion, rename the existing intel_disable_primary_plane() to
intel_disable_primary_hw_plane() to make the two names a little more
distinct.

Cc: Intel Graphics Development <intel-gfx@lists.freedesktop.org>
Signed-off-by: NMatt Roper <matthew.d.roper@intel.com>
[danvet: Fix up conflicts.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

So don't try to allocate and program it, we're only fooling ourselves.
Reported-by: N"Chang, Junxiao" <junxiao.chang@intel.com>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: NJunxiao Chang <junxiao.chang@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Apparently it is wiped out from under us, and we get some really fun
caching artifacts upon resume (it seems to be WB for all types by
default).
Reported-by: NJames Ausmus <james.ausmus@intel.com>
Signed-off-by: NBen Widawsky <ben@bwidawsk.net>
Tested-by: NJames Ausmus <james.ausmus@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76113Tested-by: NTimo Aaltonen <timo.aaltonen@canonical.com>
Cc: stable@vger.kernel.org
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The TP_printk() should never dereference any pointers, because the ring
buffer can be read at some unknown time in the future. If a device no
longer exists, it can cause a kernel oops. This also makes this
event useless when saving the ring buffer in userspaces tools such as
perf and trace-cmd.

The i915_gem_evict_vm dereferences the vm pointer which may also not
exist when the ring buffer is read sometime in the future.

Link: http://lkml.kernel.org/r/1395095198-20034-3-git-send-email-artagnon@gmail.comReported-by: NRamkumar Ramachandra <artagnon@gmail.com>
Cc: stable@vger.kernel.org # 3.13+
Fixes: bcccff84 "drm/i915: trace vm eviction instead of everything"
Signed-off-by: NSteven Rostedt <rostedt@goodmis.org>
[danvet: Try to make it actually compile]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When compiling on 32bits, I have the following warning:

drivers/gpu/drm/i915/i915_cmd_parser.c:405:4: warning: format ‘%ld’
expects argument of type ‘long int’, but argument 7 has type ‘int’
[-Wformat=]
    DRM_DEBUG_DRIVER("CMD: Command length exceeds batch length: 0x%08X
    length=%d batchlen=%ld\n",

The ptrdiff_t type has its own modifier: 't'.

Cc: Brad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Reviewed-by: NBrad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This function is only used on ILK+, so rename it accordingly.
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

No need of any here.
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This is a small follow-up fix to the series of eDP VDD back and forth
we've had recently. This is effectively a combined revert of three
commits:

commit 2c2894f698fffd8ff53e1e1d3834f9e1035b1f39
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date:   Fri Mar 7 20:05:20 2014 -0300

    drm/i915: properly disable the VDD when disabling the panel

commit b3064154
Author: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Date:   Tue Mar 4 00:42:44 2014 +0100

    drm/i915: Don't just say it, actually force edp vdd

commit dff392db
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date:   Fri Dec 6 17:32:41 2013 -0200

    drm/i915: don't touch the VDD when disabling the panel

which shows that we're pretty close back to where we started
already. The first two were basically reverting the last, but missing
the WARN. Add that back. We also OCD the intel_ prefix back to
intel_edp_panel_vdd_on() which was lost somewhere in between. The circle
closes.

For future reference, "drm/i915: don't touch the VDD when disabling the
panel" failed to take into account

commit 6cb49835
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sun May 20 17:14:50 2012 +0200

    drm/i915: enable vdd when switching off the eDP panel

and

commit 35a38556
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sun Aug 12 22:17:14 2012 +0200

    drm/i915: reorder edp disabling to fix ivb MacBook Air

Cc: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Without this the new drv_suspend/forcewake subtest I've created
doesn't result in immediately visible failures.

Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

I have been seeing this for a long time, but ignored it because it's
typically not terribly important. Recently, I really needed this info,
and it was garbage. Proof that I should have fixed it sooner. Originally
wrong from:

commit 6c7a01ec
Author: Ben Widawsky <benjamin.widawsky@intel.com>
Date:   Thu Jan 30 00:19:40 2014 -0800

    drm/i915: Capture PPGTT info on error capture

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NBen Widawsky <ben@bwidawsk.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Commit b3064154 tried to revert commit
dff392db, but wasn't complete, which
resulted in regressions on Haswell. So this commit should fix
b3064154 by undoing what it did and
providing an actual complete revert of
dff392db.

Fixes regression introduced by:
commit b3064154
Author: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Date:   Tue Mar 4 00:42:44 2014 +0100
    drm/i915: Don't just say it, actually force edp vdd

Testcase: igt/pm_pc8
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Tested-by: NPatrik Jakobsson <patrik.r.jakobsson@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We used to have per file descriptor hang stats for the
i915_get_reset_stats_ioctl() and for default context banning.

commit 0eea67eb
Author: Ben Widawsky <ben@bwidawsk.net>
Date:   Fri Dec 6 14:11:19 2013 -0800

    drm/i915: Create a per file_priv default context

made having separate hangstats in file_private redundant
as i915_hw_context already contained hangstats. So

commit c482972a
Author: Ben Widawsky <benjamin.widawsky@intel.com>
Date:   Fri Dec 6 14:11:20 2013 -0800

    drm/i915: Piggy back hangstats off of contexts

consolidated the hangstats and enabled further improvements.

commit 44e2c070
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date:   Thu Jan 30 16:01:15 2014 +0200

    drm/i915: Use i915_hw_context to set reset stats

tried to reap full benefits of consolidation but fell short
as we never 'switch' to the fake private context on gens
that don't have hw_contexts, so request->ctx remained NULL
on those.

Fix this by 'switching' to fake context so that when
request is submitted to ring, proper context gets assigned
to it.

Testcase: igt/drv_hangman
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76055Signed-off-by: NMika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This is a regression introduced in

commit 0294ae7b
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Mar 13 12:00:29 2014 +0000

    drm/i915: Consolidate forcewake resetting to a single function

The reordered setup sequence ended up calling del_timer_sync before
the timer was set up correctly, resulting in endless hilarity when
loading the driver.

Compared to Ben's patch (which moved around the setup_timer call to
sanitize_early) this moves the sanitize_early call around in the
driver load call. This way we avoid calling setup_timer again in the
resume code (where we also call sanitize_early).

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Tested-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76242Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The functionality remains largerly the same. The main difference is that
i2c-over-aux defer timeouts are increased to be safe for all use cases
instead of depending on DP device type and properties.
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Do a slight rearrangement of the switch to prep for follow-up.
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Functionality remains largely the same as before.

Note that the retry loops and native reply handling all moved into the
core drm helper functions now.
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
[danvet: Fix up the stray ; Rodrigo spotted in his review and add a
note to the commit message to answer Rodrigo's question in his review.]
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This is prep work for conversion to generic drm i2c-over-aux helpers
where we won't have the function to do this at.
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Introduce _edp_panel_vdd_on() that returns true if the call enabled vdd,
and a matching disable is needed. Keep edp_panel_vdd_on() as a helper
for when it is expected the vdd is off.
Signed-off-by: NJani Nikula <jani.nikula@intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

DRM drivers share a common address_space across all character-devices of a
single DRM device. This allows simple buffer eviction and mapping-control.
However, DRM core currently waits for the first ->open() on any char-dev
to mark the underlying inode as backing inode of the device. This delayed
initialization causes ugly conditions all over the place:
  if (dev->dev_mapping)
    do_sth();

To avoid delayed initialization and to stop reusing the inode of the
char-dev, we allocate an anonymous inode for each DRM device and reset
filp->f_mapping to it on ->open().
Signed-off-by: NDavid Herrmann <dh.herrmann@gmail.com>

We have two paths that try to reset the forcewake registers back to
known good values, with slightly different semantics and levels of
paranoia. Combine the two by passing a parameter to either restore the
forcewake status or to clear our bookkeeping, and raise the paranoia
level to max.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

It's been in there since forever, and no one cared. Doesn't put a too
good light onto our bug handling and QA efforts really ...

References: https://bugs.freedesktop.org/attachment.cgi?id=90970Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Now that we regularly defer the forcewake dance to a timer func, it is
likely to fire after we disable the device during suspend. This
generates an oops as we detect inconsistency in the hardware state. So
before suspend, we want to complete the outstanding dance and generally
sanitize the registers before handing back to the BIOS.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When we change the cache_level for an object we need to make sure
we don't put differing types of snoopable memory too close to each
other on non-LLC machines.

Currently i915_gem_object_set_cache_level() will stop looking when
it finds just one vma that has such a conflict. Drop the bogus break
statement to make sure it will unbind all vmas which need to be moved
around to avoid the conflict.

I suppose this is a theoretical issue as currently we don't enable
ppgtt on non-LLC machines, so each object can only have one vma.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We will call ppgtt_bind_vma() with flags != 0, so the WARN_ON(flags)
is bogus. Kill it.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

While wandering in the spec, I noticed that BDW removes those 2 bits
from INSTPM. I couldn't find any direct way to invalidate the TLB (ie
without the ring working already). Maybe someone will be more lucky.

At least, we now know we may be a problem.
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Reviewed-by: NBen Widawsky <ben@bwidawsk.net>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

I have the occasional absent cursor on i845 and I want to know why.
This should help by revealing the last known cursor state.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We don't need to hold struct_mutex all through intel_pipe_set_base(),
just need to hold it while pinning/unpinning the buffers.

So reduce the struct_mutext usage in intel_pipe_set_base() just like we
did for the sprite code in:
 commit 82284b6b
 Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
 Date:   Tue Oct 1 18:02:12 2013 +0300

    drm/i915: Reduce the time we hold struct mutex in sprite update_plane code

The FBC and PSR locking is still entirely fubar. That stuff was
previouly done while holding struct_mutex, so leave it there for now.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Based on Bspec the command parser must be stopped prior to
issuing sync flush. This should be done by the caller of
intel_ring_setup_status_page. Patch adds a warning if it is
not done.

v2: rebased based on new patch (wait for ring to become idle)
Signed-off-by: NNaresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

make sure we wait for rings to become idle once they are
disabled. In case of timeout print an error message
Signed-off-by: NNaresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
[danvet: Frob patch as suggested by Chris.]
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Rings should be idle before issuing sync_flush
(in intel_ring_setup_status_page). This patch moves the ring
disabling before doing the HW status page setup.
Signed-off-by: NNaresh Kumar Kachhi <naresh.kumar.kachhi@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Our code allows have a PPGTT that is smaller than the maximum size for
GEN6-GEN7. Though I don't think this actually ever occurs, the code may
as well work properly and more importantly look correct by using the
variable size instead of the HW max.
Signed-off-by: NBen Widawsky <ben@bwidawsk.net>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

I'm not clear if the hardware is still subject to the same prefetching
issues that made us use a scratch page in the first place. In either
case, we're using garbage with the current code (we will end up using
offset 0).

This may be the cause of our current gem_cpu_reloc regression with
PPGTT. I cannot test it at the moment.
Signed-off-by: NBen Widawsky <ben@bwidawsk.net>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Introduced in
commit e0e33f8ff6f0b6d286afc314802be4993341bd47
Author: Imre Deak <imre.deak@intel.com>
Date:   Tue Mar 4 19:23:07 2014 +0200

The impact was luckily minimal, due to the extra check we do against a
software pipestat IRQ mask.

Caught by Fengguang's 0-day tester.

Cc: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: NImre Deak <imre.deak@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

BSpec is a bit unclear whether HDMI+HDMI cloning should work on g4x.
Tests on real hardware say that it does. Since g4x can't send
infoframes to more than one HDMI port anyway, we don't lose anything
by allow it.

For PCH platforms BSpec explicitly forbids HDMI+HDMI cloning.

Whether HDMI+HDMI cloning might also work on VLV is a bit unclear, but
since we'd at least lose the capability of sending infoframes to more
than one cloned HDMI port, it doesn't seem like a good idea to allow it.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=73850Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

HDMI+VGA cloning should be supported on all platforms. The only real
obstacle is the 1.5x clock adjustment for 12bpc HDMI, but that is now
taken care of, so we can allow HDMI+VGA cloning.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=73850Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When cloning HDMI with other output types, we can't use 12bpc since the
clocks for the other encoder types would be off. So have
intel_hdmi_compute_config() check if there are other encoders besides
HDMI being fed from the same pipe, and if so, pick 8bpc insted if 12bpc.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Currently we allow encoders to indicate whether they can be part of a
cloned set with just one flag. That's not flexible enough to describe
the actual hardware capabilities. Instead make it a bitmask of encoder
types with which the current encoder can be cloned.

For now we set the bitmask to allow DVO+DVO and DVO+VGA, which should
match what the old boolean flag allowed. We will add some more cloning
options in the future.

Note that this patch also removes the encoder.possible_clones setting
from encoder setup code - we compute this dynamically.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
[danvet: Add Ville's explanation why removing the encoder
possible_clones is save.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When adding new gunk, _always_ think of a good place. Start/end
usually just means that this didn't happen, and on top of that results
in needless conflicts with other patches doing the same.

Introduced in

commit 62d5d69b
Author: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Date:   Tue Feb 25 17:11:28 2014 +0200

    drm/i915: Add suspend count to error state

Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>