When an encoder is shared on several connectors there is only
one hotplug line, thus this line needs to be shared among these
connectors.
If HPD detect only works reliably on a subset of those connectors,
we want to poll the others. Thus we need to make sure that storm
detection doesn't mess up the settings for those connectors.
Therefore we store the settings in the intel_connector struct and
restore them from there.
If nothing is set but the encoder has a hpd_pin set we assume this
connector is hotplug capable.
On init/reset we make sure the polled state of the connectors
is (re)set to the default value, the HPD interrupts are marked
enabled.
Signed-off-by: NEgbert Eich <eich@suse.de>
Reviewed-by: NJani Nikula <jani.nikula@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

It returns true if we've requested to turn the power well on and it's
really on. It also returns true for all the previous gens.

For now there's just one caller, but I'm going to add more.
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Clock computations and handling are highly encoder specific, both in
the optimal clock selection and also in which clocks to use and when
sharing of clocks is possible.

So the best place to do this is somewhere in the encoders, with a
generic fallback for those encoders without special needs. To facility
this, add a pipe_config->clocks_set boolean.

This patch here is only prep work, it simply sets the computed clock
values in pipe_config->dpll, and uses that data in the hw clock
setting functions.

Haswell code isn't touched, simply because Haswell clocks work much
different and need their own infrastructure (with probably a
Haswell-specific config->ddi_clock substruct).
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Jesse Barnes noticed in his review of my DP cleanup series that
intel_edp_target_clock is now unused. Checking related code I've
noticed that also intel_edp_link_config is long unused.

Kill them both.

Wrt leaky eDP functions used in the common crtc code, the only thing
still left is intel_encoder_is_pch_edp. That one is just due to the
massive confusion between eDP vs. DP and port A vs. port D. Crtc code
should at most concern itself with the later, never with the former.

But that's material for another patch series.

Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We need it in the fdi m_n computation, which nicely kills almost
all ugly special cases in there.

It looks like we also need this to handle 12bpc hdmi correctly.

Eventually it might be better to switch things around and put the
target clock into adjusted_mode->clock and create a new pipe_config
parameter for the port link clock.

v2: Add a massive comment in the code to explain this mess.

v3: s/dp_target_clock/pixel_target_clock in anticipation of the hdmi
use-case.
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We need a flag to designate dp encoders and the dp link m_n parameters
in the pipe config for that. And now that the pipe bpp computations
have been moved up and stored in the pipe config, too, we can do this
without losing our sanity.

v2: Rebased on top of Takashi Iwai's fix to (again) fix the target
clock handling for eDP. Luckily the new code is sane enough and just
does the right thing!

v3: Move ->has_dp_encoder to this patch (Jesse).
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

There's a rather decent confusion going on around transcoder m_n
values. So let's clarify:
- All dp encoders need this, either on the pch transcoder if it's a
  pch port, or on the cpu transcoder/pipe if it's a cpu port.
- fdi links need to have the right m_n values for the fdi link set in
  the cpu transcoder.

To handle the pch vs transcoder stuff a bit better, extract transcoder
set_m_n helpers. To make them simpler, set intel_crtc->cpu_transcoder
als in ironlake_crtc_mode_set, so that gen5+ (where the cpu m_n
registers are all at the same offset) can use it.

Haswell modeset is decently confused about dp vs. edp vs. fdi. dp vs.
edp works exactly the same as dp (since there's no pch dp any more),
so use that as a check. And only set up the fdi m_n values if we
really have a pch encoder present (which means we have a VGA encoder).

On ilk+ we've called ironlake_set_m_n both for cpu_edp and for pch
encoders. Now that dp_set_m_n handles all dp links (thanks to the
pch encoder check), we can ditch the cpu_edp stuff from the
fdi_set_m_n function.

Since the dp_m_n values are not readily available, we need to
carefully coax the edp values out of the encoder. Hence we can't (yet)
kill this superflous complexity.

v2: Rebase on top of the ivb fdi B/C check patch - we need to properly
clear intel_crtc->fdi_lane, otherwise those checks will misfire.

v3: Rebased on top of a s/IS_HASWELL/HAS_DDI/ patch from Paulo Zanoni.

v4: Drop the addition of has_dp_encoder, it's in the wrong patch (Jesse).
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

No constant alpha yet though, that needs a new ioctl and/or property to
get/set.

v2: use drm_plane_format_cpp (Ville)
    fix up vlv_disable_plane, remove IVB bits (Ville)
    remove error path rework (Ville)
    fix component order confusion (Ville)
    clean up platform init (Ville)
    use compute_offset_xtiled (Ville)
v3: fix up more format confusion (Ville)
    update to new page offset function (Ville)
v4: remove incorrect formats from framebuffer_init (Ville)
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

As Thomas Gleixner spotted, it's rather horrible racy:
- We can miss almost a full tick, so need to compensate by 1 jiffy.
- We need to re-check the condition when having timed-out, since a
  the last check could have been before the timeout expired. E.g. when
  we've been preempted or a long irq happened.

Cc: Thomas Gleixner <tglx@linutronix.de>
Reported-by: NJack Winter <jbh@alchemy.lu>
Cc: Jack Winter <jbh@alchemy.lu>
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Since

commit bcf9dcc1
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Sun Jul 15 09:42:38 2012 +0100

    drm/i915: Workaround hang with BSD and forcewake on SandyBridge

and

commit 0cc2764c
Author: Ben Widawsky <ben@bwidawsk.net>
Date:   Sat Sep 1 22:59:48 2012 -0700

    drm/i915: use cpu_relax() in wait_for_atomic

these two macros are essentially the same, so unify them. We keep the
_us version since it's a nice documentation for smaller timeouts.

v2: Fixup time unit conversion, _wait_for takes ms (Ville).

Cc: Jack Winter <jbh@alchemy.lu>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The old code only handled either 6bpc or 8bpc. Since it's easy to do,
reorganize the code to be a bit more generic so that it can also handle
10bpc and 12bpc. Note that we still start with 8bpc, so there's no
functional change.

Also, since we no don't need to compute the 6BPC flag in the mode_valid
callback, we can consolidate things a bit. That requires though that
the link bw computation is moved up in the compute_config callback.
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We want to compute this earlier. To avoid a big complicated patch,
this patch here just does the big search&replace and still calls the
old functions at the same places.
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Now that we have a useful struct for this, let's use it. Some neat
pointer-chasing required, but it's all there already.

v2: Rebased on top of the added Haswell limited color range support.
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This is used way too often in the enable/disable paths. And will
be even more useful in the future.

Note that correct semantics of this change highly depend upon
correct updating of intel_crtc->config: Like with all other
modeset state, we need to call ->disable with the old config,
but ->mode_set and ->enable with the new config.

v2: Do not yet use the flag in the ->disable callbacks - atm we don't
yet have support for the information stored in the pipe_config in the
hw state readout code, so this will be wrong at boot-up/resume.

v3: Rebased on top of the hdmi/dp ddi encoder merging.

v4: Fixup stupid rebase error which lead to a NULL vfunc deref.

v5: On haswell the VGA port is on the PCH!

v6: s/IS_HASWELL/HAS_DDI/, spotted by Paulo Zanoni. Also add a missing
parameter name in a function declaration.

v7: Don't forget to git add ...
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Used by SDVO (and hopefully, eventually HDMI, if we ever get around
to fixing up the low dotclock CEA modes ...).

This required adding a new encoder->mode_set callback to be able to
pass around the intel_crtc_config.
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Only used by the lvds encoder. Note that we shouldn't do the same
simple conversion with the FORCE_6BPC flag, since that's much better
handled by moving all the pipe_bpc computation around.

This requires that we pass the pipe config around to encoders, so
that they can set special attributes and set constraints. To do so
introduce a new ->compute_config encoder callback, which is called in
stead of the drm crtc helper's ->mode_fixup.

To avoid massive churn all over the codebase we don't want to convert
all existing ->mode_fixup functions. Instead I've opted to convert
them on an as-needed basis (mostly to cut down on rebase conflicts and
to have more freedom to experiment around while developing the
patches).
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Currently only containing the requested and the adjusted mode. And
only crtc callbacks are converted somewhat to it, encoders will be
done on a as-needed basis (simply too much churn in one patch
otherwise).

Future patches will add tons more useful stuff to this struct,
starting with the very simple.

v2: Store the pipe_config in the intel_crtc, so that the ->mode-set,
->enable and also ->disable have easy access to it.

v3: Store the pipe config in the right crtc ...

v4: Rebased.

v5: Fixup an OOPS when trying to kfree an ERR_PTR.

v6: Used drm_moode_copy and some other small cleanups as suggested
by Ville Syrjälä.

v7: drm_mode_copy preserves the mode id of the destination, so no need
to clear it again (Ville).

v8: Break a long line spotted by Paulo.
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

To clean up hotplug support we add a new enum to intel_encoder:
enum hpd_pin. It allows the encoder to request a hpd line but leave
the details which IRQ is responsible on which chipset generation
to i915_irq.c.
This way requesting hotplug support will become really simple on
the encoder/connector level.
Signed-off-by: NEgbert Eich <eich@suse.de>
Acked-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

To be used to restore sprite state on resume.

v2: move sprite tracking bits up so we don't track modified sprite state
v3: use src_x/y in sprite suspend/resume code (Ville)
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We are really talking about the transcoder function here and the disable
version uses trancoder in its name already, so let's try to be
consistent.
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

We trim the fb to fit the CRTC by computing the offset of that CRTC to
its nearest tile_row origin. This allows us to use framebuffers that are
larger than the CRTC limits without additional work.

However, we failed to compute the offset for a linear framebuffer
correctly as we treated its x-advance in whole tiles (instead of the
linear increment expected), leaving the CRTC misaligned with its
contents.

Fixes regression from commit c2c75131
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Thu Jul 5 12:17:30 2012 +0200

    drm/i915: adjust framebuffer base address on gen4+

v2: Adjust relative x-coordinate after linear alignment (vsyrjala)
v3: Repaint with pokadots (vsyrjala)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=61152Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Some (but not all) of the HDMI registers can be used to control sDVO,
so those registers have two names. IMHO, when we're talking about
HDMI, we really should call the HDMI control register "hdmi_reg"
instead of "sdvox_reg", otherwise we'll just confuse people reading
our code (we now have platforms with HDMI but without SDVO). So now
"struct intel_hdmi" has a member called "hdmi_reg" instead of
"sdvox_reg".

Also, don't worry: "struct intel_sdvo" still has a member called
"sdvo_reg".

v2: Rebase (v1 was sent in May 2012).
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

This way we can remove some duplicated code and avoid more mistakes
and regressions with these registers in the future.
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

GPU reset will drop all flips that are still in the ring. So after the
reset, call update_plane() for all CRTCs to make sure the primary
planes are scanning out from the correct buffer.

Also finish all pending flips. That means user space will get its
page flip events and won't get stuck waiting for them.

v2: Explicitly finish page flips instead of relying on FLIP_DONE
    interrupt being generated by the base address update.
v3: Make two loops over crtcs to avoid deadlocks with the crtc mutex
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
[danvet: Fixup long line complaint from checkpatch.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Similarly to:

  commit 6a0d1df3d3a0d2370541164eb0595fe35dcd6de3
  Author: Damien Lespiau <damien.lespiau@intel.com>
  Date:   Tue Dec 11 15:18:28 2012 +0000

      drm/i915: Preserve the FDI line reversal override bit on CPT

DDI port support lane reversal to easy the PCB layouting work. Let's
preserve the bit configured by the BIOS (until we find how to correctly
retrieve the information from the VBT, but this does sound more fragile
then just relying on the BIOS that has, hopefully, been validated
already.
Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

If a GPU reset occurs while a page flip has been submitted to the ring,
the flip will never complete once the ring has been reset.

The GPU reset can be detected by sampling the reset_counter before the
flip is submitted, and then while waiting for the flip, the sampled
counter is compared with the current reset_counter value.
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NDamien Lespiau <damien.lespiau@intel.com>
[danvet: Move the reset_counter assignment to an earlier place in
common code as discussed on the mailing list.]
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60140Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Now that the driver is in control of whether it needs to disable
everything at take-over or not, we can rip this all out.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

When adding the fb idle detection to mark-inactive, it was forgotten
that userspace can drive the processing of retire-requests. We assumed
that it would be principally driven by the retire requests worker,
running once every second whilst active and so we would get the deferred
timer for free. Instead we spend too many CPU cycles reclocking the LVDS
preventing real work from being done.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reported-and-tested-by: NAlexander Lam <lambchop468@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=58843
Cc: stable@vger.kernel.org
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Our suspend code touches a lot of registers all over the place, so we
need to enable the power well before suspending.
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: NJani Nikula <jani.nikula@intel.com>
[danvet: Fixup compilation by stealing the header decl from the
dynamic power wells patch.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The current code was wrong in many different ways, so this is a full
rewrite. We don't have "different power wells for different parts of
the GPU", we have a single power well, but we have multiple registers
that can be used to request enabling/disabling the power well. So
let's be a good citizen and only use the register we're suppose to
use, except when we're loading the driver, where we clear the request
made by the BIOS.

If any of the registers is requesting the power well to be enabled, it
will be enabled. If none of the registers is requesting the power well
to be enabled, it will be disabled.

For now we're just forcing the power well to be enabled, but in the
next commits we'll change this.

V2:
  - Remove debug messages that could be misleading due to possible
    race conditions with KVMr, Debug and BIOS.
  - Don't wait on disabling: after a conversaion with a hardware
    engineer we discovered that the "restriction" on bit 31 is just
    for the "enable" case, and we don't even need to wait on the
    "disable" case.
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

ELD info should be updated dynamically according to hot plug event.
For haswell chip, clear/set the eld valid bit and output enable bit
from callback intel_disable/eanble_ddi().
Reviewed-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
Signed-off-by: NWang Xingchao <xingchao.wang@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The AVI infoframe is able to inform the display whether the source is
sending full or limited range RGB data.

As per CEA-861 [1] we must first check whether the display reports the
quantization range as selectable, and if so we can set the approriate
bits in the AVI inforframe.

[1] CEA-861-E - 6.4 Format of Version 2 AVI InfoFrame

v2: Give the Q bits better names, add spec chapter information
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Add a new "Automatic" mode to the "Broadcast RGB" range property.
When selected the driver automagically selects between full range and
limited range output.

Based on CEA-861 [1] guidelines, limited range output is selected if the
mode is a CEA mode, except 640x480. Otherwise full range output is used.
Additionally DVI monitors should most likely default to full range
always.

As per DP1.2a [2] DisplayPort should always use full range for 18bpp, and
otherwise will follow CEA-861 rules.

NOTE: The default value for the property will now be "Automatic"
so some people may be affected in case they're relying on the
current full range default.

[1] CEA-861-E - 5.1 Default Encoding Parameters
[2] VESA DisplayPort Ver.1.2a - 5.1.1.1 Video Colorimetry

v2: Use has_hdmi_sink to check if a HDMI monitor is present
v3: Add information about relevant spec chapters
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

The RGB color range select bit on the DP/SDVO/HDMI registers
disappeared when PCH was introduced, and instead a new PIPECONF bit
was added that performs the same function.

Add a new INTEL_MODE_LIMITED_COLOR_RANGE private mode flag, and set
it in the encoder mode_fixup if limited color range is requested.
Set the the PIPECONF bit 13 based on the flag.

Experimentation showed that simply toggling the bit while the pipe is
active doesn't work. We need to restart the pipe, which luckily already
happens.

The DP/SDVO/HDMI bit 8 is marked MBZ in the docs, so avoid setting it,
although it doesn't seem to do any harm in practice.

TODO:
- the PIPECONF bit too seems to have disappeared from HSW. Need a
  volunteer to test if it's just a documentation issue or if it's really
  gone. If the bit is gone and no easy replacement is found, then I suppose
  we may need to use the pipe CSC unit to perform the range compression.

v2: Use mode private_flags instead of intel_encoder virtual functions
v3: Moved the intel_dp color_range handling after bpc check to help
    later patches

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=46800Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Note: This patch also adds a little helper intel_crtc_restore_mode for
the common case where we do a full modeset but with the same
parameters, e.g. to undo bios damage or update a property.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NRodrigo Vivi <rodrigo.vivi@gmail.com>
[danvet: Added note.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

If you unplug the hdmi connector slowly enough, the hotplug interrupt
fires but then the kernel code tries to read the EDID and succeeds
(because the connector is still half connected, the HPD pin is shorter
than the others, and DDC works). Since EDID succeeds it thinks the
monitor is still connected.

To prevent that, read the live HPD status in the hotplug handler before
trying to read the EDID.

v2: Rename the function to ibx_ (Chris Wilson)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=55372Signed-off-by: NDamien Lespiau <damien.lespiau@intel.com>
Reviewed-by: NJani Nikula <jani.nikula@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

For GMCH platforms we set up the hpd irq registers in the irq
postinstall hook. But since we only enable the irq sources we actually
need in PORT_HOTPLUG_EN/STATUS, taking dev_priv->hotplug_supported_mask
into account, no hpd interrupt sources is enabled since

commit 52d7eced
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sat Dec 1 21:03:22 2012 +0100

    drm/i915: reorder setup sequence to have irqs for output setup

Wrongly set-up interrupts also lead to broken hw-based load-detection
on at least GM45, resulting in ghost VGA/TV-out outputs.

To fix this, delay the hotplug register setup until after all outputs
are set up, by moving it into a new dev_priv->display.hpd_irq_callback.
We might also move the PCH_SPLIT platforms to such a setup eventually.

Another funny part is that we need to delay the fbdev initial config
probing until after the hpd regs are setup, for otherwise it'll detect
ghost outputs. But we can only enable the hpd interrupt handling
itself (and the output polling) _after_ that initial scan, due to
massive locking brain-damage in the fbdev setup code. Add a big
comment to explain this cute little dragon lair.

v2: Encapsulate all the fbdev handling by wrapping the move call into
intel_fbdev_initial_config in intel_fb.c. Requested by Chris Wilson.

v3: Applied bikeshed from Jesse Barnes.

v4: Imre Deak noticed that we also need to call intel_hpd_init after
the drm_irqinstall calls in the gpu reset and resume paths - otherwise
hotplug will be broken. Also improve the comment a bit about why
hpd_init needs to be called before we set up the initial fbdev config.

Bugzilla: Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54943Reported-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> (v3)
Reviewed-by: NImre Deak <imre.deak@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

More specifically, the LPT FDI RX only supports 8bpc and a maximum of
2 lanes, so anything above that won't work and should be rejected.
Signed-off-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Before queuing the flip but crucially after attaching the unpin-work to
the crtc, we continue to setup the unpin-work. However, should the
hardware fire early, we see the connected unpin-work and queue the task.
The task then promptly runs and unpins the fb before we finish taking
the required references or even pinning it... Havoc.

To close the race, we use the flip-pending atomic to indicate when the
flip is finally setup and enqueued. So during the flip-done processing,
we can check more accurately whether the flip was expected.

v2: Add the appropriate mb() to ensure that the writes to the page-flip
worker are complete prior to marking it active and emitting the MI_FLIP.
On the read side, the mb should be enforced by the spinlocks.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: stable@vger.kernel.org
[danvet: Review the barriers a bit, we need a write barrier both
before and after updating ->pending. Similarly we need a read barrier
in the interrupt handler both before and after reading ->pending. With
well-ordered irqs only one barrier in each place should be required,
but since this patch explicitly sets out to combat spurious interrupts
with is staged activation of the unpin work we need to go full-bore on
the barriers, too. Discussed with Chris Wilson on irc and changes
acked by him.]
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>

Just a prep patch to make this a property of intel_lvds. Makes more
sense, removes clutter from intel_display.c and eventually I want to
move all the encoder special cases wrt clock handling to encoders
anyway.

v2: Add an intel_ prefixe to is_dual_link_lvds since it's non-static
now.
Reviewed-by: NPaulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>