GitHub/moto-9609/android_kernel_motorola_exynos9610.git
9 years agodrm/i915: Do not use ggtt_view with (aliasing) PPGTT
Joonas Lahtinen [Mon, 16 Mar 2015 12:11:13 +0000 (14:11 +0200)]
drm/i915: Do not use ggtt_view with (aliasing) PPGTT

GGTT views are only applicable when dealing with GGTT. Change the code to
reject ggtt_view where it should not be used and require it when it should
be.

v2:
- Dropped _ppgtt_ infixes, allow both types to be passed
- Disregard other but normal views when no view is specified
- More checks that valid parameters are passed
- More readable error checking

v3:
- Prefer WARN_ONCE over BUG_ON when there is code path for failure

Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
[danvet: Drop unecessary forward decl from earlier patch iterations.]
[danvet: Remove unused variable spotted by Tvrtko.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix sink crc connector iteration
Rodrigo Vivi [Fri, 13 Mar 2015 23:13:59 +0000 (16:13 -0700)]
drm/i915: Fix sink crc connector iteration

Regressed by this commit:

commit 3455454e18ca3f92c565700539e744c620d8276b
Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Date:   Tue Mar 3 15:21:56 2015 +0200

    drm/i915: Add a for_each_intel_connector macro

Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agoMerge tag 'drm-intel-fixes-2015-03-19' into drm-intel-next
Daniel Vetter [Fri, 20 Mar 2015 10:43:59 +0000 (11:43 +0100)]
Merge tag 'drm-intel-fixes-2015-03-19' into drm-intel-next

Backmerge because of numerous and interleaving conflicts and git
rerere getting confused a bit too often.

Conflicts:
drivers/gpu/drm/i915/intel_display.c

All conflicts are because of -next patches backported to -fixes, so
just go with the code in -next.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/i915: Make sure the primary plane is enabled before reading out the fb state
Damien Lespiau [Thu, 5 Feb 2015 19:35:13 +0000 (19:35 +0000)]
drm/i915: Make sure the primary plane is enabled before reading out the fb state

We don't want to end up in a state where we track that the pipe has its
primary plane enabled when primary plane registers are programmed with
values that look possible but the plane actually disabled.

Refuse to read out the fb state when the primary plane isn't enabled.

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Suggested-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reported-by: Andrey Skvortsov <andrej.skvortzov@gmail.com>
Reported-by: Steven Rostedt <rostedt@goodmis.org>
Reference: http://mid.gmane.org/20150203191507.GA2374@crion86
Tested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agodrm/i915: Update DRIVER_DATE to 20150313
Daniel Vetter [Fri, 13 Mar 2015 21:15:40 +0000 (22:15 +0100)]
drm/i915: Update DRIVER_DATE to 20150313

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix vmap_batch page iterator overrun
Mika Kuoppala [Fri, 13 Mar 2015 13:21:53 +0000 (15:21 +0200)]
drm/i915: Fix vmap_batch page iterator overrun

vmap_batch() calculates amount of needed pages for the mapping
we are going to create. And it uses this page count as an
argument for the for_each_sg_pages() macro. The macro takes the number
of sg list entities as an argument, not the page count. So we ended
up iterating through all the pages on the mapped object, corrupting
memory past the smaller pages[] array.

Fix this by bailing out when we have enough pages.

This regression has been introduced in

commit 17cabf571e50677d980e9ab2a43c5f11213003ae
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Wed Jan 14 11:20:57 2015 +0000

    drm/i915: Trim the command parser allocations

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Export total subslice and EU counts
Jeff McGee [Mon, 9 Mar 2015 23:06:54 +0000 (16:06 -0700)]
drm/i915: Export total subslice and EU counts

Setup new I915_GETPARAM ioctl entries for subslice total and
EU total. Userspace drivers need these values when constructing
GPGPU commands. This kernel query method is intended to replace
the PCI ID-based tables that userspace drivers currently maintain.
The kernel driver can employ fuse register reads as needed to
ensure the most accurate determination of GT config attributes.
This first became important with Cherryview in which the config
could differ between devices with the same PCI ID.

The kernel detection of these values is device-specific and not
included in this patch. Because zero is not a valid value for any of
these parameters, a value of zero is interpreted as unknown for the
device. Userspace drivers should continue to maintain ID-based tables
for older devices not supported by the new query method.

v2: Increment our I915_GETPARAM indices to fit after REVISION
    which was merged ahead of us.

For: VIZ-4636
Signed-off-by: Jeff McGee <jeff.mcgee@intel.com>
Tested-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Acked-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: redefine WARN_ON_ONCE to include the condition
Jani Nikula [Thu, 12 Mar 2015 11:01:12 +0000 (13:01 +0200)]
drm/i915: redefine WARN_ON_ONCE to include the condition

Same as

commit c883ef1b1c998d2d66866772fd0fc34afa45641e
Author: Mika Kuoppala <miku@iki.fi>
Date:   Tue Oct 28 17:32:30 2014 +0200

    drm/i915: Redefine WARN_ON to include the condition

but for WARN_ON_ONCE. Since the kernel WARN_ON_ONCE actually picks up
*our* version of WARN_ON, we end up with messages like

[  838.285319] WARN_ON(!__warned)

which are not that helpful.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Implement WaDisableHBR2
Damien Lespiau [Wed, 11 Feb 2015 17:43:24 +0000 (17:43 +0000)]
drm/i915/skl: Implement WaDisableHBR2

v2: Use the recently introduced INTEL_REVID() and SKL_REVID defines
    (Nick Hoath)

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Tested-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89554
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove the preliminary_hw_support shackles from CHV
Ville Syrjälä [Wed, 11 Mar 2015 20:52:28 +0000 (22:52 +0200)]
drm/i915: Remove the preliminary_hw_support shackles from CHV

CHV should be in a good enough shape now, so let's drop the
.is_preliminary flag.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Read CHV_PLL_DW8 from the correct offset
Ville Syrjälä [Wed, 11 Mar 2015 20:52:08 +0000 (22:52 +0200)]
drm/i915: Read CHV_PLL_DW8 from the correct offset

We accidentally pass 'pipe' instead of 'port' to CHV_PLL_DW8() and
with PIPE_C we end up at register offset 0x8320 which isn't the
0x8020 we wanted. Fix it.

The problem was fortunately caught by the sanity check in vlv_dpio_read():
WARNING: CPU: 1 PID: 238 at ../drivers/gpu/drm/i915/intel_sideband.c:200 vlv_dpio_read+0x77/0x80 [i915]()
DPIO read pipe C reg 0x8320 == 0xffffffff

The problem got introduced with this commit:
 commit 71af07f91f12bbab96335e202c82525d31680960
 Author: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
 Date:   Thu Mar 5 19:33:08 2015 +0530

    drm/i915: Update prop, int co-eff and gain threshold for CHV

Cc: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Todd Previte <tprevite@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Rewrite IVB FDI bifurcation conflict checks
Ville Syrjälä [Wed, 11 Mar 2015 16:52:31 +0000 (18:52 +0200)]
drm/i915: Rewrite IVB FDI bifurcation conflict checks

Ignore the current state of the pipe and just check crtc_state->enable
and the number of FDI lanes required. This means we don't accidentally
mistake the FDI lanes as being available of one of the pipes just
happens to be disabled at the time of the check. Also we no longer
consider pipe C to require FDI lanes when it's driving the eDP
transcoder.

We also take the opportunity to make the code a bit nicer looking by
hiding the ugly bits in the new pipe_required_fdi_lanes() function.

Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Rewrite some some of the FDI lane checks
Ville Syrjälä [Wed, 11 Mar 2015 16:52:30 +0000 (18:52 +0200)]
drm/i915: Rewrite some some of the FDI lane checks

The logic in the FDI lane checks is very hard for my poor brain to
grasp. Rewrite it in a more straightforward way.

Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Enable the RPS interrupts programming
Akash Goel [Fri, 6 Mar 2015 05:37:24 +0000 (11:07 +0530)]
drm/i915/skl: Enable the RPS interrupts programming

Enable the RPS interrupts programming(enable/disable/reset) for GEN9,
as missing changes to enable the RPS support on GEN9 have been added.

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Enabling processing of Turbo interrupts
Akash Goel [Fri, 6 Mar 2015 05:37:23 +0000 (11:07 +0530)]
drm/i915/skl: Enabling processing of Turbo interrupts

Earlier Turbo interrupts were not being processed for SKL,
as something was amiss in turbo programming for SKL.
Now missing changes have been added, so enabling the Turbo
interrupt processing for SKL.

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the i915_frequency_info debugfs function
Akash Goel [Fri, 6 Mar 2015 05:37:21 +0000 (11:07 +0530)]
drm/i915/skl: Updated the i915_frequency_info debugfs function

Added support for SKL in the i915_frequency_info debugfs function

v2:
- corrected the handling of reqf (Damien)
- Reorderd the platform check for cagf (Ville)

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Simplify the way BC bifurcation state consistency is kept
Ander Conselvan de Oliveira [Wed, 11 Mar 2015 11:35:43 +0000 (13:35 +0200)]
drm/i915: Simplify the way BC bifurcation state consistency is kept

Remove the global modeset resource function that would disable the
bifurcation bit, and instead enable/disable it when enabling the pch
transcoder. The mode set consistency check should prevent us from
disabling the bit if pipe C is enabled so the change should be safe.

Note that this doens't affect the logic that prevents the bit being
set while a pipe is active, since the patch retains the behavior of
only chaging the bit if necessary. Because of the checks during mode
set, the first change would necessarily happen with both pipes B and
C disabled, and any subsequent write would be skipped.

v2: Only change the bit during pch trancoder enable. (Ville)

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the act_freq_mhz_show sysfs function
Akash Goel [Fri, 6 Mar 2015 05:37:22 +0000 (11:07 +0530)]
drm/i915/skl: Updated the act_freq_mhz_show sysfs function

Added support for SKL in the act_freq_mhz_show sysfs function

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the gen9_enable_rps function
Akash Goel [Fri, 6 Mar 2015 05:37:20 +0000 (11:07 +0530)]
drm/i915/skl: Updated the gen9_enable_rps function

On SKL, GT frequency is programmed in units of 16.66 MHZ units compared
to 50 MHZ for older platforms. Also the time value specified for Up/Down EI &
Up/Down thresholds are expressed in units of 1.33 us, compared to 1.28
us for older platforms. So updated the gen9_enable_rps function as per that.

v2: Updated to use new macro GT_INTERVAL_FROM_US

v3: Removed the initial setup of certain registers, from gen9_enable_rps,
    which gets overridden later from gen6_set_rps (Damien)

v4: Removed the enabling of rps interrupts, from gen9_enable_rps.
    To be done from intel_gen6_powersave_work only, as done for other
    platforms also.

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the gen6_rps_limits function
Akash Goel [Fri, 6 Mar 2015 05:37:19 +0000 (11:07 +0530)]
drm/i915/skl: Updated the gen6_rps_limits function

RP Interrupt Up/Down Frequency Limits register (A014) definition
has changed for SKL. Updated the gen6_rps_limits function as per that

v2: Renamed the function to intel_rps_limits (Chris)

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Restructured the gen6_set_rps_thresholds function
Akash Goel [Fri, 6 Mar 2015 05:37:18 +0000 (11:07 +0530)]
drm/i915/skl: Restructured the gen6_set_rps_thresholds function

Prior to SKL, the time period programmed in Up/Down EI & Up/Down
threshold registers was in units of 1.28 micro seconds. But for
SKL, the units have changed (1.333 micro seconds).
Have generalized the implementation of gen6_set_rps_thresholds function,
by removing the hard coding done in it as per 1.28 micro seconds.

v2: Renamed the local variables & removed superfluous comments (Chris)

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the gen6_set_rps function
Akash Goel [Fri, 6 Mar 2015 05:37:17 +0000 (11:07 +0530)]
drm/i915/skl: Updated the gen6_set_rps function

On SKL, the frequency is programmed differently in RPNSWREQ (A008)
register (from bits 23 to 31, compared to bits 24 to 31). So updated
the gen6_set_rps function, as per this change.

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated the gen6_init_rps_frequencies function
Akash Goel [Fri, 6 Mar 2015 05:37:16 +0000 (11:07 +0530)]
drm/i915/skl: Updated the gen6_init_rps_frequencies function

On SKL the frequency is specified in units of 16.66 MHZ, barring the
RP_STATE_CAP(0x5998) register, which still reports frequency in units
of 50 MHZ. So an extra conversion is required in gen6_init_rps_frequencies
function for SKL, to store the frequency values as per the actual hardware unit.

v2: Corrected the conversion from 50 to 16.66 MHZ (Ville)

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Updated intel_gpu_freq() and intel_freq_opcode()
Akash Goel [Fri, 6 Mar 2015 05:37:15 +0000 (11:07 +0530)]
drm/i915/skl: Updated intel_gpu_freq() and intel_freq_opcode()

On SKL, frequency is specified in units of 16.66 MHZ.
Updated the intel_gpu_freq() and intel_freq_opecode() functions
to do the conversion appropriately.

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Added new macros
Akash Goel [Fri, 6 Mar 2015 05:37:14 +0000 (11:07 +0530)]
drm/i915/skl: Added new macros

For SKL, register definition for RPNSWREQ (A008), RPSTAT1(A01C)
have changed slightly. Also on SKL, frequency is specified in
units of 16.66 MHZ, compared to 50 MHZ for most of the earlier
platforms and the time values are expressed in units of 1.33 us,
compared to 1.28 us for earlier platforms.
Added new macros for the aforementioned changes.

v2: Renamed the GT_FREQ_FROM_PERIOD macro to GT_INTERVAL_FROM_US (Damien)

v3: Removed the implicit use of dev_priv in GT_INTERVAL_FROM_US macro (Chris)

Signed-off-by: Akash Goel <akash.goel@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: remove indirection in the PCI ID macros
Jani Nikula [Tue, 3 Feb 2015 12:34:05 +0000 (14:34 +0200)]
drm/i915: remove indirection in the PCI ID macros

Spell all the PCI IDs out to be able to quickly grep for the IDs. No
functional changes.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: Add GT1/2 to comments to not loose that distinction.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use FW_WM() macro for older gmch platforms too
Ville Syrjälä [Tue, 10 Mar 2015 15:02:21 +0000 (17:02 +0200)]
drm/i915: Use FW_WM() macro for older gmch platforms too

Use the FW_WM() macro from the VLV wm code to polish up the wm
code for older gmch platforms.

Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add polish to VLV WM shift+mask operations
Ville Syrjälä [Tue, 10 Mar 2015 14:16:28 +0000 (16:16 +0200)]
drm/i915: Add polish to VLV WM shift+mask operations

Wrap the FW register value shift+mask operations into a macro to hide
the ugliness a bit. Also might avoid bugs due to typos.

Also rename all the primary/sprite plane low order bit masks to have the
_VLV suffix, so that we can use the FW_WM_VLV() macro instead of the
FW_WM() macro for them in a consistent manner. Cursor and all the high
order bits are left to use the FW_WM() macro as there's no real
confusion with them.

Cc: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use plane->state->fb instead of plane->fb in intel_plane_restore()
Ville Syrjälä [Tue, 10 Mar 2015 11:15:23 +0000 (13:15 +0200)]
drm/i915: Use plane->state->fb instead of plane->fb in intel_plane_restore()

plane->fb is not as reliable as plane->state->fb so let's convert
intel_plane_restore() over the the new way of thinking as well.

Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Reduce clutter by using the local plane pointer
Ville Syrjälä [Tue, 10 Mar 2015 11:15:22 +0000 (13:15 +0200)]
drm/i915: Reduce clutter by using the local plane pointer

No need to go dig throguh intel_crtc->base.cursor when we already have
the same thing as 'plane' local variable.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove debug prints from primary plane update funcs
Ville Syrjälä [Tue, 10 Mar 2015 11:15:21 +0000 (13:15 +0200)]
drm/i915: Remove debug prints from primary plane update funcs

These are now called from the plane commit hooks, so they really need to
be fast or else we risk atomic update failures. So kill the debug prints
which are slowing things down massively.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add ULL postfix to VGT_MAGIC constant
Daniel Vetter [Wed, 11 Mar 2015 07:27:25 +0000 (08:27 +0100)]
drm/i915: Add ULL postfix to VGT_MAGIC constant

Without this Dave's 32bit rhel compiler is annoyed. Don't ask me about
the exact rules for this stuff though, but this should be safe.

Reported-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/fourcc: 64 #defines need ULL postfix
Daniel Vetter [Wed, 11 Mar 2015 07:26:23 +0000 (08:26 +0100)]
drm/fourcc: 64 #defines need ULL postfix

I have no idea about the exact rules, but this angered Dave's 32bit
rhel gcc.

Reported-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/i915: Don't assume primary & cursor are always on for wm calculation (v4)
Matt Roper [Mon, 9 Mar 2015 18:06:02 +0000 (11:06 -0700)]
drm/i915: Don't assume primary & cursor are always on for wm calculation (v4)

Current ILK-style watermark code assumes the primary plane and cursor
plane are always enabled.  This assumption, along with the combination
of two independent commits that got merged at the same time, results in
a NULL dereference.  The offending commits are:

        commit fd2d61341bf39d1054256c07d6eddd624ebc4241
        Author: Matt Roper <matthew.d.roper@intel.com>
        Date:   Fri Feb 27 10:12:01 2015 -0800

            drm/i915: Use plane->state->fb in watermark code (v2)

and

        commit 0fda65680e92545caea5be7805a7f0a617fb6c20
        Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
        Date:   Fri Feb 27 15:12:35 2015 +0000

            drm/i915/skl: Update watermarks for Y tiling

The first commit causes us to use the FB from plane->state->fb rather
than the legacy plane->fb, which is updated a bit later in the process.

The second commit includes a change that now triggers watermark
reprogramming on primary plane enable/disable where we didn't have one
before (which wasn't really correct, but we had been getting lucky
because we always calculated as if the primary plane was on).

Together, these two commits cause the watermark calculation to
(properly) see plane->state->fb = NULL when we're in the process of
disabling the primary plane.  However the existing watermark code
assumes there's always a primary fb and tries to dereference it to find
out pixel format / bpp information.

The fix is to make ILK-style watermark calculation actually check the
true status of primary & cursor planes and adjust our watermark logic
accordingly.

v2: Update unchecked uses of state->fb for other platforms (pnv, skl,
    etc.).  Note that this is just a temporary fix.  Ultimately the
    useful information is going to be computed at check time and stored
    right in the state structures so that we don't have to figure this
    all out while we're supposed to be programming the watermarks.
    (caught by Tvrtko)

v3: Fix a couple copy/paste mistakes in SKL code. (Tvrtko)

v4: Only add FB checks for ILK/SKL codepaths.  Older platforms still use
    intel_crtc_active() and will shortcircuit out of watermark
    calculations before ever trying to dereference the primary plane's
    framebuffer.

Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reported-by: Michael Leuchtenburg <michael@slashhome.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89388
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Move drm_framebuffer_unreference out of struct_mutex for flips
Chris Wilson [Mon, 16 Feb 2015 14:31:49 +0000 (14:31 +0000)]
drm/i915: Move drm_framebuffer_unreference out of struct_mutex for flips

intel_user_framebuffer_destroy() requires the struct_mutex for its
object bookkeeping, so this means that all calls to
drm_framebuffer_unreference must not hold that lock.

Regression from commit ab8d66752a9c28cd6c94fa173feacdfc1554aa03
Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Date:   Mon Feb 2 15:44:15 2015 +0000

    drm/i915: Track old framebuffer instead of object

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89166
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
[danvet: Clarify commit message slightly.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Disable DDR DVFS on CHV
Ville Syrjälä [Thu, 5 Mar 2015 19:19:52 +0000 (21:19 +0200)]
drm/i915: Disable DDR DVFS on CHV

DDR DVFS introduces massive memory latencies which can't be handled by
the PND deadline stuff. Instead the watermarks will need to be
programmed to compensate for the latency and the deadlines will need to
be programmed to tight fixed values. That means DDR DVFS can only be
enabled if the display FIFOs are large enough, and that pretty much
means we have to manually repartition them to suit the needs of the
moment.

That's a lot of change, so in the meantime let's just disable DDR DVFS
to get the display(s) to be stable.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Enable the maxfifo PM5 mode when appropriate on CHV
Ville Syrjälä [Thu, 5 Mar 2015 19:19:51 +0000 (21:19 +0200)]
drm/i915: Enable the maxfifo PM5 mode when appropriate on CHV

CHV has a new knob in Punit to select between some memory power savings
modes PM2 and PM5. We can allow the deeper PM5 when maxfifo mode is
enabled, so let's do so in the hopes for moar power savings.

v2: Put the thing into a separate function to avoid churn later
v3: Don't break VLV

Reviewed-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Arun R Murthy <arun.r.murthy@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Program PFI credits for VLV
Vidya Srinivas [Thu, 5 Mar 2015 19:19:50 +0000 (21:19 +0200)]
drm/i915: Program PFI credits for VLV

PFI credit programming is required when CD clock (related to data flow from
display pipeline to end display) is greater than CZ clock (related to data
flow from memory to display plane). This programming should be done when all
planes are OFF to avoid intermittent hangs while accessing memory even from
different Gfx units (not just display).

If cdclk/czclk >=1, PFI credits could be set as any number. To get better
performance, larger PFI credit can be assigned to PND. Otherwise if
cdclk/czclk<1, the default PFI credit of 8 should be set.

v2:
    - Change log to lower log level instead of DRM_ERROR
    - Change function name to valleyview_program_pfi_credits
    - Move program PFI credits to modeset_init instead of intel_set_mode
    - Change magic numbers to logical constants

[vsyrjala v3:
 - only program in response to cdclk update
 - program the credits also when cdclk<czclk
 - add CHV bits
 v4:
 - Change CHV cdclk<czclk credits to 12 (Vijay)]

Signed-off-by: Vidya Srinivas <vidya.srinivas@intel.com>
Signed-off-by: Gajanan Bhat <gajanan.bhat@intel.com>
Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Rewrite VLV/CHV watermark code
Ville Syrjälä [Thu, 5 Mar 2015 19:19:49 +0000 (21:19 +0200)]
drm/i915: Rewrite VLV/CHV watermark code

Assuming the PND deadline mechanism works reasonably we should do
memory requests as early as possible so that PND has schedule the
requests more intelligently. Currently we're still calculating
the watermarks as if VLV/CHV are identical to g4x, which isn't
the case.

The current code also seems to calculate insufficient watermarks
and hence we're seeing some underruns, especially on high resolution
displays.

To fix it just rip out the current code and replace is with something
that tries to utilize PND as efficiently as possible.

We now calculate the WM watermark to trigger when the FIFO still has
256us worth of data. 256us is the maximum deadline value supoorted by
PND, so issuing memory requests earlier would mean we probably couldn't
utilize the full FIFO as PND would attempt to return the data at
least in at least 256us. We also clamp the watermark to at least 8
cachelines as that's the magic watermark that enabling trickle feed
would also impose. I'm assuming it matches some burst size.

In theory we could just enable trickle feed and ignore the WM values,
except trickle feed doesn't work with max fifo mode anyway, so we'd
still need to calculate the SR watermarks. It seems cleaner to just
disable trickle feed and calculate all watermarks the same way. Also
trickle feed wouldn't account for the 256us max deadline value, thoguh
that may be a moot point in non-max fifo mode sicne the FIFOs are fairly
small.

On VLV max fifo mode can be used with either primary or sprite planes.
So the code now also checks all the planes (apart from the cursor)
when calculating the SR plane watermark.

We don't have to worry about the WM1 watermarks since we're using the
PND deadline scheme which means the hardware ignores WM1 values.

v2: Use plane->state->fb instead of plane->fb

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make sure we invalidate frontbuffer on fbcon.
Rodrigo Vivi [Tue, 10 Mar 2015 00:57:07 +0000 (17:57 -0700)]
drm/i915: Make sure we invalidate frontbuffer on fbcon.

There are some cases like suspend/resume or dpms off/on sequences
that can flush frontbuffer bits. In these cases features that relies
on frontbuffer tracking can start working and user can stop getting
screen updates on fbcon having impression the system is frozen.

So, let's make sure we also invalidate frontbuffer on fbdev blank.

v2: Daniel was right, backtrace didn't show other path than this blank
one so let's make sure frontbuffer bits gets invalidate here instead of
on random write operations that doesn't garantee we track all frontbuffer
writes.

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[danvet: Exchange code comments for one that complains about the
locking, like in set_par.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Update prop, int co-eff and gain threshold for CHV
Vijay Purushothaman [Thu, 5 Mar 2015 14:03:08 +0000 (19:33 +0530)]
drm/i915: Update prop, int co-eff and gain threshold for CHV

This patch implements latest PHY changes in Gain, prop and int co-efficients
based on the vco freq.

v2: Split the original changes into multiple smaller patches based on
review by Ville

v3: Addressed Ville's review comments. Fixed the error introduced in v2.
Clear the old bits before we modify those bits as part of RMW.

v4: TDC target cnt is 10 bits and not 8 bits (Ville)

Signed-off-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Initialize CHV digital lock detect threshold
Vijay Purushothaman [Thu, 5 Mar 2015 14:02:06 +0000 (19:32 +0530)]
drm/i915: Initialize CHV digital lock detect threshold

Initialize lock detect threshold and select coarse threshold for the
case where M2 fraction division is disabled.

v2: Split the changes into multiple smaller patches (Ville)
v3: Clear out the old bits before we modify those bits as RMW (Ville)
v4: Reset coarse threshold when M2 fraction is enabled (Ville)

Signed-off-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Disable M2 frac division for integer case
Vijay Purushothaman [Thu, 5 Mar 2015 14:00:57 +0000 (19:30 +0530)]
drm/i915: Disable M2 frac division for integer case

v2 : Handle M2 frac division for both M2 frac and int cases

v3 : Addressed Ville's review comments. Cleared the old bits for RMW

v4 : Fix feedfwd gain (Ville)

Signed-off-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Reviewed-by: Ville Syrjala <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Spelling s/auxilliary/auxiliary/
Geert Uytterhoeven [Mon, 9 Mar 2015 20:21:08 +0000 (21:21 +0100)]
drm/i915: Spelling s/auxilliary/auxiliary/

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use crtc->state->active in ilk/skl watermark calculations (v3)
Matt Roper [Mon, 9 Mar 2015 17:19:24 +0000 (10:19 -0700)]
drm/i915: Use crtc->state->active in ilk/skl watermark calculations (v3)

Existing watermark code calls intel_crtc_active() to determine whether a CRTC
is active for the purpose of watermark calculations (and bails out early if it
determines the CRTC is not active).  However intel_crtc_active() only returns
true if crtc->primary->fb is non-NULL, which isn't appropriate in the modern
age of universal planes and atomic modeset since userspace can now disable the
primary plane, but leave the CRTC (and other planes) running.

Note that commit

        commit 0fda65680e92545caea5be7805a7f0a617fb6c20
        Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
        Date:   Fri Feb 27 15:12:35 2015 +0000

            drm/i915/skl: Update watermarks for Y tiling

adds a test for primary plane enable/disable to trigger a watermark update
(previously we ignored updates to primary planes, which wasn't really correct,
but we got lucky since we always pretended the primary plane was on).  Tvrtko's
patch tries to update watermarks when we re-enable the primary plane, but that
watermark computation gets aborted early because intel_crtc_active() returns
false due to the disabled primary plane.

Switch the ILK and SKL watermark code over to use crtc->state->active rather
than calling intel_crtc_active() so that we'll properly compute watermarks when
re-enabling the primary plane.

Note that this commit doesn't touch callsites in the watermark code for
older platforms since there were concerns that doing so would lead to
other types of breakage.

Also note that all of the watermark calculation at the moment takes place after
new crtc/plane states are swapped into the DRM objects.  This will change in
the future, so we'll be working with in-flight state objects, but for the time
being, crtc->state is what we want to operate on.

v2: Don't drop primary->fb check from intel_crtc_active(), but rather replace
    ILK/SKL callsites with direct tests of crtc->state->active.  There is
    concern that messing with intel_crtc_active() will lead to other breakage for
    old hardware platforms.  (Ville)

v3: Use intel_crtc->active for now rather than crtc->state->active since
    we don't have CRTC states properly hooked up and initialized yet.
    We'll defer the switch to crtc->state->active until the atomic CRTC
    state work is farther along. (Ville)

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Update intel_crtc_active() to use state values (v2)
Matt Roper [Mon, 9 Mar 2015 17:19:23 +0000 (10:19 -0700)]
drm/i915: Update intel_crtc_active() to use state values (v2)

With the switch to atomic plumbing for planes, some of our commit-time
work (e.g., watermarks) is done after the new atomic state is swapped
into the relevant DRM object, but before the DRM core has a chance to
update its legacy state values.  Switch intel_crtc_active() to look at
the state objects rather than legacy fields to ensure we operate on the
proper values.

Note that we're continuing to use intel_crtc->active here for the time
being since crtc->state isn't really hooked up yet.  Once CRTC states
are wired up properly, we'll want to switch this over to use
crtc->state->active instead.

v2: Switch back to intel_crtc->active for now; when Ander's work on CRTC
    states is ready, we can flip this over to use crtc->state->active
    instead. (Ville)

Cc: Ander Conselvan De Oliveira <conselvan2@gmail.com>
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Exit early from psr_status if PSR is not supported by the device
Damien Lespiau [Mon, 9 Mar 2015 14:17:58 +0000 (14:17 +0000)]
drm/i915: Exit early from psr_status if PSR is not supported by the device

Static analysis was complaining that a path existed where we could use
stat[] uninitialized. Fix this by simplifying the logic to exit early if
PSR isn't supported.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix chv cdclk support
Ville Syrjälä [Mon, 2 Mar 2015 18:07:17 +0000 (20:07 +0200)]
drm/i915: Fix chv cdclk support

The specs seem to be full of misinformation wrt. the Punit register
0x36. Some versions still show the old VLV bit layout, some the new
layout, and all of them seem to tell us nonsense about the cdclk
value encoding.

Testing on actual hardware has shown that we simply need to program
the desired CCK divider into the Punit register using the new layout of
the bits. Doing that, the status bit change to indicate the same value,
and the CCK 0x6b register also changes accordingly to indicate that CCK
is now using the new divider.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Reviewed-by: Yogesh Mohan Marimuthu <yogesh.mohan.marimuthu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Allow pixel clock up to 95% of cdclk on CHV
Ville Syrjälä [Mon, 2 Mar 2015 18:07:16 +0000 (20:07 +0200)]
drm/i915: Allow pixel clock up to 95% of cdclk on CHV

Supposedly CHV can sustain a pixel clock of up to 95% of
cdclk, as opposed to the 90% limit that was used old older
platforms. Update the cdclk selection code to allow for this.

This will allow eg. HDMI 4k modes with their 297MHz pixel clock
while still respecting the 320 MHz cdclk limit on CHV.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Reviewed-by: Yogesh Mohan Marimuthu <yogesh.mohan.marimuthu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: port A fuse straps don't work on early SKL steppings
Jesse Barnes [Fri, 6 Mar 2015 23:53:32 +0000 (15:53 -0800)]
drm/i915/skl: port A fuse straps don't work on early SKL steppings

So try to enumerate eDP unconditionally in those cases.

Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: Add wa tag Damien dug out.]
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Restore the DDI translation tables when enabling PW1
Damien Lespiau [Fri, 6 Mar 2015 18:50:53 +0000 (18:50 +0000)]
drm/i915/skl: Restore the DDI translation tables when enabling PW1

I was dumping the DDI translation tables to make sure my patch updating
the HDMI entry was doing the right thing when I noticed that the table
was showing reset values after DPMS.

And indeed, the DDI translation registers are in power well 1 on SKL,
and so we're losing their values when shutting down eDP.

Calling intel_prepare_ddi() on PW1 enabling re-programs the table.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove unused condition in hsw_power_well_post_enable()
Damien Lespiau [Fri, 6 Mar 2015 18:50:52 +0000 (18:50 +0000)]
drm/i915: Remove unused condition in hsw_power_well_post_enable()

We don't use this function on gen9, no need for that test here.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Restore pipe interrupt registers after power well enabling
Damien Lespiau [Fri, 6 Mar 2015 18:50:51 +0000 (18:50 +0000)]
drm/i915/skl: Restore pipe interrupt registers after power well enabling

The pipe interrupt registers are in the actual pipe power well, so we
need to restore them when re-enable the corresponding power well.

I've also copied what we do on HSW/BDW for VGA, even if the we haven't
enabled unclaimed registers just yet.

v2: Don't run skl_power_well_post_enable() if the power well is already
    enabled (Paulo)

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Mirror what we do on HSW for the power well enable log message
Damien Lespiau [Fri, 6 Mar 2015 18:50:50 +0000 (18:50 +0000)]
drm/i915/skl: Mirror what we do on HSW for the power well enable log message

Just to be more consistent with what we do on HSW.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Introduce enable_requested and is_enabled in the power well code
Damien Lespiau [Fri, 6 Mar 2015 18:50:49 +0000 (18:50 +0000)]
drm/i915/skl: Introduce enable_requested and is_enabled in the power well code

Just like what we do for HSW/BDW, having those variables makes it a bit
easier to parse the code.

Suggested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Make gen8_irq_power_well_post_enable() take a pipe mask
Damien Lespiau [Fri, 6 Mar 2015 18:50:48 +0000 (18:50 +0000)]
drm/i915/skl: Make gen8_irq_power_well_post_enable() take a pipe mask

While we only need to restore pipe B/C interrupt registers on BDW when
enabling the power well, skylake a bit more flexible and we'll also need
to restore the pipe A registers as it has its own power well that can be
toggled.

Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/chv: Add CHV HW status to SSEU status
Jeff McGee [Fri, 27 Feb 2015 18:22:32 +0000 (10:22 -0800)]
drm/i915/chv: Add CHV HW status to SSEU status

Collect the currently enabled counts of slice, subslice, and
execution units using the power gate control ack message
registers specific to Cherryview.

Slice/subslice/EU info and hardware status can now be
determined for CHV, so allow the debugfs SSEU status dump
to proceed for CHV devices.

Signed-off-by: Jeff McGee <jeff.mcgee@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/chv: Determine CHV slice/subslice/EU info
Jeff McGee [Fri, 27 Feb 2015 20:12:28 +0000 (12:12 -0800)]
drm/i915/chv: Determine CHV slice/subslice/EU info

Total EU was already being detected on CHV, so we just add the
additional info parameters. The detection method is changed to
be more robust in the case of subslice fusing - we don't want
to trust the EU fuse bits corresponding to subslices which are
fused-off.

v2: Fixed subslice disable bitmasks and removed unnecessary ?
    operation (Ville)

Signed-off-by: Jeff McGee <jeff.mcgee@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make sure PND deadline mode is enabled on VLV/CHV
Ville Syrjälä [Thu, 5 Mar 2015 19:19:48 +0000 (21:19 +0200)]
drm/i915: Make sure PND deadline mode is enabled on VLV/CHV

Poke at the CBR1_VLV register during init_clock_gating to make sure the
PND deadline scheme is used.

The hardware has two modes of operation wrt. watermarks:

1) PND deadline mode:
 - memory request deadline is calculated from actual FIFO level * DDL
 - WM1 watermark values are unused (AFAIK)
 - WM watermark level defines when to start fetching data from memory
   (assuming trickle feed is not used)

2) backup mode
 - deadline is based on FIFO status, DDL is unused
 - FIFO split into three regions with WM and WM1 watermarks, each
   part specifying a different FIFO status

We want to use the PND deadline mode, so let's make sure the chicken
bit is in the correct position on init.

Also take the opportunity to refactor the shared code between VLV and
CHV to a shared function.

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Read out display FIFO size on VLV/CHV
Ville Syrjälä [Thu, 5 Mar 2015 19:19:47 +0000 (21:19 +0200)]
drm/i915: Read out display FIFO size on VLV/CHV

VLV/CHV have similar DSPARB registers as older platforms, just more of
them due to more planes. Add a bit of code to read out the current FIFO
split from the registers. Will be useful later when we improve the WM
calculations.

v2: Add display_mmio_offset to DSPARB

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Pass plane to vlv_compute_drain_latency()
Ville Syrjälä [Thu, 5 Mar 2015 19:19:46 +0000 (21:19 +0200)]
drm/i915: Pass plane to vlv_compute_drain_latency()

Now that we have drm_planes for the cursor and primary we can move the
pixel_size handling into vlv_compute_drain_latency() and just pass the
appropriate plane to it.

v2: Check plane->state->fb instead of plane->fb

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> (v1)
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Resolve conflict with Matt's s/plane->fb/plane->state->fb/
patch.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Reorganize VLV DDL setup
Ville Syrjälä [Thu, 5 Mar 2015 19:19:45 +0000 (21:19 +0200)]
drm/i915: Reorganize VLV DDL setup

Introduce struct vlv_wm_values to house VLV watermark/drain latency
values. We start by using it when computing the drain latency values.

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Hide VLV DDL precision handling
Ville Syrjälä [Thu, 5 Mar 2015 19:19:44 +0000 (21:19 +0200)]
drm/i915: Hide VLV DDL precision handling

Move the DDL precision handling into vlv_compute_drain_latency() so the
callers don't have to duplicate the same code to deal with it.

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Simplify VLV drain latency computation
Ville Syrjälä [Thu, 5 Mar 2015 19:19:43 +0000 (21:19 +0200)]
drm/i915: Simplify VLV drain latency computation

The current drain lantency computation relies on hardcoded limits to
determine when the to use the low vs. high precision multiplier.
Rewrite the code to use a more straightforward approach.

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Kill DRAIN_LATENCY_PRECISION_* defines
Ville Syrjälä [Thu, 5 Mar 2015 19:19:42 +0000 (21:19 +0200)]
drm/i915: Kill DRAIN_LATENCY_PRECISION_* defines

Kill the silly DRAIN_LATENCY_PRECISION_* defines and just use the raw
number instead.

v2: Move the sprite 32/16 -> 16/8 preision multiplier
    change to another patch (Jesse)

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Reduce CHV DDL multiplier to 16/8
Ville Syrjälä [Thu, 5 Mar 2015 19:19:41 +0000 (21:19 +0200)]
drm/i915: Reduce CHV DDL multiplier to 16/8

Apparently we must yet halve the DDL drain latency from what we're
using currently. This little nugget is not in any spec, but came
down through the grapevine.

This makes the displays a bit more stable. Not quite fully stable but at
least they don't fall over immediately on driver load.

v2: Update high_precision in valleyview_update_sprite_wm() too (Jesse)

Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Disable the mmio.debug WARN after it fires
Chris Wilson [Thu, 18 Dec 2014 10:55:50 +0000 (10:55 +0000)]
drm/i915: Disable the mmio.debug WARN after it fires

If we have a single unclaimed register, we will have lots. A WARN for
each one makes the machine unusable and does not aid debugging. Convert
the i915.mmio_debug option to a counter for how many WARNs to fire
before shutting up. Even when i915.mmio_debug was disabled it would
continue to shout an *ERROR* for every interrupt, without any
information at all for debugging.

The massive verbiage was added in
commit 5978118c39c2f72fd8b39ef9c086723542384809
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date:   Wed Jul 16 17:49:29 2014 -0300

    drm/i915: reorganize the unclaimed register detection code

v2: Automatically enable invalid mmio reporting for the *next* invalid
access if mmio_debug is disabled by default. This should give us clearer
debug information without polluting the logs too much.
v3: Compile fixes, rebase.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
[danvet: Update modparam text per the thread.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: fix simple_return.cocci warnings
kbuild test robot [Thu, 5 Mar 2015 14:03:08 +0000 (22:03 +0800)]
drm/i915: fix simple_return.cocci warnings

drivers/gpu/drm/i915/intel_ringbuffer.c:435:1-4: WARNING: end returns can be simpified

 Simplify a trivial if-return sequence.  Possibly combine with a
 preceding function call.
Generated by: scripts/coccinelle/misc/simple_return.cocci

CC: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Do both mt and gen6 style forcewake reset on ivb probe
Mika Kuoppala [Fri, 27 Feb 2015 16:11:09 +0000 (18:11 +0200)]
drm/i915: Do both mt and gen6 style forcewake reset on ivb probe

commit 05a2fb157e44 ("drm/i915: Consolidate forcewake code")
failed to take into account that we have used to reset both
the gen6 style and the multithreaded style forcewake registers.
This is due to fact that ivb can use either, depending on how the
bios has set up the machine.

Mimic the old semantics before we have determined the correct variety
and reset both before the ecobus probe.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Huang Ying <ying.huang@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove ironlake rc6 support
John Harrison [Thu, 5 Mar 2015 14:03:03 +0000 (14:03 +0000)]
drm/i915: Remove ironlake rc6 support

Apparently, this has never worked reliably and is currently disabled. Also, the
gains are not particularly impressive. Thus rather than try to keep unused code
from decaying and having to update it for other driver changes, it was decided
to simply remove it.

For: VIZ-5115
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't clobber plane state on internal disables
Matt Roper [Wed, 4 Mar 2015 18:49:04 +0000 (10:49 -0800)]
drm/i915: Don't clobber plane state on internal disables

We need to disable all sprite planes when disabling the CRTC.  We had
been using the top-level atomic 'disable' entrypoint to accomplish this,
which was wrong.  Not only can this lead to various locking issues, it
also modifies the actual plane state, making it impossible to restore
the plane properly later.  For example, a DPMS off followed by a DPMS on
will result in any sprite planes in use not being restored properly.

The proper solution here is to call directly into our 'commit plane'
hook with a copy of the plane's current state that has 'visible' set to
false.  Committing this dummy state will turn off the plane, but will
not touch the actual plane->state pointer, allowing us to properly
restore the plane state later.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Setup all page directories for gen8
Mika Kuoppala [Wed, 4 Mar 2015 12:55:17 +0000 (14:55 +0200)]
drm/i915: Setup all page directories for gen8

If the requested size is less than what the full range
of pdps can address, we end up setting pdps for only the
requested area.

The logical context however needs all pdp entries to be valid.
Prior to commit 06fda602dbca ("drm/i915: Create page table allocators")
we have been writing pdp entries with dma address of zero instead
of valid pdps. This is supposedly bad even if those pdps are not
addressed.

As commit 06fda602dbca ("drm/i915: Create page table allocators")
introduced more dynamic structure for pdps, we ended up oopsing
when we populated the lrc context. Analyzing this oops revealed
the fact that we have not been writing valid pdps with bsw, as
it is doing the ppgtt init with 2GB limit in some cases.

We should do the right thing and setup the non addressable part
pdps/pde/pte to scratch page through the minimal structure by
having just pdp with pde entries pointing to same page with
pte entries pointing to scratch page.

But instead of going through that trouble, setup all the pdps
through individual pd pages and pt entries, even for non
addressable parts. And let the clear range point them to scratch
page. This way we populate the lrc with valid pdps and wait
for dynamic page allocation work to land, and do the heavy lifting
for truncating page table tree according to usage.

The regression of oopsing in init was introduced by
commit 06fda602dbca ("drm/i915: Create page table allocators")

v2: Clear the range for the unused part also (Ville)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89350
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Valtteri Rantala <valtteri.rantala@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Program PLL for edp1.4 intermediate frequencies
Sonika Jindal [Sat, 21 Feb 2015 05:42:13 +0000 (11:12 +0530)]
drm/i915/skl: Program PLL for edp1.4 intermediate frequencies

v2: Making the link_clock half in switch inline with the DPLL_CTRL1_* macros
(Ville)

Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Add support for edp 1.4 intermediate frequencies
Sonika Jindal [Thu, 5 Mar 2015 04:32:30 +0000 (10:02 +0530)]
drm/i915/skl: Add support for edp 1.4 intermediate frequencies

eDp 1.4 supports custom frequencies.
Skylake supports following intermediate frequencies : 3.24 GHz, 2.16 GHz and
4.32 GHz along with usual LBR, HBR and HBR2 frequencies.
Read sink supported frequencies and get common frequencies from sink and
source and use these for link training.

v2: Rebased, removed calculation of min_clock since for edp it is taken as
max_clock (as per comment).
v3: Keeping single array for link rates (Satheesh)
v4: Setting LINK_BW_SET to 0 when setting LINK_RATE_SET (Satheesh)
v5: Some minor nits (Ville)
v6: Keeping separate arrays for source and sink rates (Ville)
v7: Remove redundant setting of DP_LINK_BW_SET to 0 (Ville)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Read sink supported rates from edp panel
Sonika Jindal [Thu, 5 Mar 2015 04:33:58 +0000 (10:03 +0530)]
drm/i915/skl: Read sink supported rates from edp panel

v2: Using DP_SUPPORTED_LINK_RATES macro for supported_rates array (Satheesh).
v3: Reading dpcd's supported link rates tables based upon edp version in the
same patch.
v4: Move version check under is_edp (Satheesh)
v5: Using le16 for rates, some naming, and removing nested if block (Ville)
v6: Correctly using DP_MAX_SUPPORTED_RATES and removing DP_SUPPORTED_LINK_RATES
(Ville)
v7: Incorrectly removed DP_SUPPORTED_LINK_RATES in v6, re-adding it
v8: Checking return value of intel_dp_dpcd_read_wake() (Ville)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: add frontbuffer tracking to FBC
Paulo Zanoni [Fri, 13 Feb 2015 19:23:46 +0000 (17:23 -0200)]
drm/i915: add frontbuffer tracking to FBC

Kill the blt/render tracking we currently have and use the frontbuffer
tracking infrastructure.

Don't enable things by default yet.

v2: (Rodrigo) Fix small conflict on rebase and typo at subject.
v3: (Paulo) Rebase on RENDER_CS change.
v4: (Paulo) Rebase.
v5: (Paulo) Simplify: flushes don't have origin (Daniel).
            Also rebase due to patch order changes.

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fixing mutex deadlock window at eDP DRRS
Ramalingam C [Tue, 3 Mar 2015 06:41:46 +0000 (12:11 +0530)]
drm/i915: Fixing mutex deadlock window at eDP DRRS

In invalidate and flush functions of eDP DRRS, if deferred downclock
work starts execution at a time window between acquiring the drrs
mutex and cancellation of the deferred work
(intel_edp_drrs_downclock_work), then deferred work will find
drrs mutex locked and wait for the same.

Meanwhile the function that acquired mutex drrs invalidate/flush will
wait for the completion of the deferred work before releasing the mutex.
Thats a deadlock.

To avoid such deadlock scenario, this change cancels the deferred work
before acquiring the mutex at invalidate and flush functions.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add debugfs entry for DRRS
Vandana Kannan [Tue, 3 Mar 2015 15:23:10 +0000 (20:53 +0530)]
drm/i915: Add debugfs entry for DRRS

Adding a debugfs entry to determine if DRRS is supported or not

V2: [By Ram]: Following details about the active crtc will be filled
in seq-file of the debugfs
1. Encoder output type
2. DRRS Support on this CRTC
3. DRRS current state
4. Current Vrefresh
Format is as follows:
CRTC 1:  Output: eDP, DRRS Supported: Yes (Seamless), DRRS_State: DRRS_HIGH_RR, Vrefresh: 60
CRTC 2:  Output: HDMI, DRRS Supported : No, VBT DRRS_type: Seamless
CRTC 1:  Output: eDP, DRRS Supported: Yes (Seamless), DRRS_State: DRRS_LOW_RR, Vrefresh: 40
CRTC 2:  Output: HDMI, DRRS Supported : No, VBT DRRS_type: Seamless

V3: [By Ram]: Readability is improved.
Another error case is covered [Daniel]

V4: [By Ram]: Current status of the Idleness DRRS along with
the Front buffer bits are added to the debugfs. [Rodrigo]

V5: [By Ram]: Rephrased to make it easy to understand.
And format is modified. [Rodrigo]

V6: [By Ram]: Modeset mutex are acquired for each crtc along with
renaming the Idleness detection states  [Daniel]

Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
[danvet: dump full busy_frontbuffer_bits and remove the dubios
computed logical state of DRRS - debugfs is about what is fact,
developers should reach their own conclusion when debugging issues.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add I915_PARAM_REVISION
Neil Roberts [Wed, 4 Mar 2015 14:41:16 +0000 (14:41 +0000)]
drm/i915: Add I915_PARAM_REVISION

Adds a parameter which can be used with DRM_I915_GETPARAM to query the
GPU revision. The intention is to use this in Mesa to implement the
WaDisableSIMD16On3SrcInstr workaround on Skylake but only for
revision 2.

Signed-off-by: Neil Roberts <neil@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Improve staged config logging
Ander Conselvan de Oliveira [Tue, 3 Mar 2015 13:21:57 +0000 (15:21 +0200)]
drm/i915: Improve staged config logging

When logging that full mode switch is necessary, log which connector,
encoder or crtc has caused it, so it is easier to figure out what is
goind on by just looking at the log.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add a for_each_intel_connector macro
Ander Conselvan de Oliveira [Tue, 3 Mar 2015 13:21:56 +0000 (15:21 +0200)]
drm/i915: Add a for_each_intel_connector macro

We have similar macros for crtcs and encoders, and the pattern happens
often enough to justify the macro.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Set crtc backpointer when duplicating crtc state
Ander Conselvan de Oliveira [Tue, 3 Mar 2015 13:21:55 +0000 (15:21 +0200)]
drm/i915: Set crtc backpointer when duplicating crtc state

In the path were there is no state to duplicate, the allocated crtc
state wouldn't have the crtc backpointer initialized.

Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Reduce CHV DPLL min vco frequency to 4.8 GHz
Ville Syrjälä [Thu, 26 Feb 2015 19:01:52 +0000 (21:01 +0200)]
drm/i915: Reduce CHV DPLL min vco frequency to 4.8 GHz

The current minimum vco frequency leaves us with a gap in our supported
frequencies at 233-243 MHz. Your typical 2560x1440@60 display wants a
pixel clock of 241.5 MHz, which is just withing that gap. Reduce the
allowed vco min frequency to 4.8GHz to reduce the gap to 233-240 MHz,
and thus allow such displays to work.

4.8 GHz is actually the documented (at least in some docs) limit of the
PLL, and we just picked 4.86 GHz originally because that was the lowest
value produced by the PLL spreadsheet, which obviously didn't consider
2560x1440 displays.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Vijay Purushothaman <vijay.a.purushothaman@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: also do frontbuffer tracking on pwrites
Paulo Zanoni [Fri, 13 Feb 2015 19:23:45 +0000 (17:23 -0200)]
drm/i915: also do frontbuffer tracking on pwrites

We need this for FBC, and possibly for PSR too.

v2: Don't only flush: invalidate too (Daniel).

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: pass which operation triggered the frontbuffer tracking
Paulo Zanoni [Fri, 13 Feb 2015 19:23:44 +0000 (17:23 -0200)]
drm/i915: pass which operation triggered the frontbuffer tracking

We want to port FBC to the frontbuffer tracking infrastructure, but
for that we need to know what caused the object invalidation so
we can react accordingly: CPU mmaps need manual, GTT mmaps and
flips don't need handling and ring rendering needs nukes.

v2: - s/ORIGIN_RENDER/ORIGIN_CS/ (Daniel, Rodrigo)
    - Fix copy/pasted wrong documentation
    - Rebase
v3: - Rebase
v4: - Don't pass the operation to flushes (Daniel).

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agoRevert "drm/i915: Switch planes from transitional helpers to full atomic helpers"
Daniel Vetter [Mon, 2 Mar 2015 15:35:20 +0000 (16:35 +0100)]
Revert "drm/i915: Switch planes from transitional helpers to full atomic helpers"

This reverts commit 3f678c96abb43a977d2ea41aefccdc49e8a3e896.

We've been a bit too optimistic with this one here :(

The trouble is that internally we're still using these plane
update/disable hooks. Which was totally ok pre-atomic since the drm
core did all the book-keeping updating and these just mostly updated
hw state. But with atomic there's lots more going on, and it causes
heaps of trouble with the load detect code.

This one specifically cause a deadlock since both the load detect code
and the nested plane atomic helper functions tried to grab the same
locks. It only blows up because of the evil tricks though we play with
the implicit ww acquire context.

Applying this revert unearths the NULL deref on already freed
framebuffer objects reported as a regression in 4.0 by various people.

Fixing this will be fairly invasive, hence revert even for the
4.1-next queue.

Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul Bolle <pebolle@tiscali.nl>
Acked-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/i915/skl: Only use the 800mV+2bB HDMI translation entry
Damien Lespiau [Mon, 2 Mar 2015 16:19:36 +0000 (16:19 +0000)]
drm/i915/skl: Only use the 800mV+2bB HDMI translation entry

This translation entry was updated after electrical validation by the hw
team. The other entries are removed from existence as they aren't
validated and because the sole use of a certain type of level shifter
for SKL products is anticipated.

v2: Remove all the other entries and force the use of the 800mv+2dB
    config (Sonika)

Suggested-by: Sonika Jindal <sonika.jindal@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make for_each_sprite() take dev_priv as argument
Damien Lespiau [Sat, 28 Feb 2015 14:54:09 +0000 (14:54 +0000)]
drm/i915: Make for_each_sprite() take dev_priv as argument

Implicit usage of local variables in macros isn't exactly the greatest
thing in the world, especially when that variable is the drm device and
we want to move towards a broader use of the i915 device structure.

Let's make for_each_sprite() take dev_priv as its first argument then.

Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Chris Wilson <chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make for_each_plane() take dev_priv as argument
Damien Lespiau [Sat, 28 Feb 2015 14:54:08 +0000 (14:54 +0000)]
drm/i915: Make for_each_plane() take dev_priv as argument

Implicit usage of local variables in macros isn't exactly the greatest
thing in the world, especially when that variable is the drm device and
we want to move towards a broader use of the i915 device structure.

Let's make for_each_plane() take dev_priv as its first argument then.

Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Chris Wilson <chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove irq-related FIXME in reset code
Daniel Vetter [Mon, 23 Feb 2015 11:03:28 +0000 (12:03 +0100)]
drm/i915: Remove irq-related FIXME in reset code

With the two-step reset counter increments which braket the actual
reset code and the subsequent wake-up we're guaranteeing that all the
lockless waiters _will_ be woken up. And since we unconditionally bail
out of waits with -EAGAIN (or -EIO) in that case there is not risk of
lost interrupt enabling bits when the lockless wait code races against
a gpu reset.

Let's remove this FIXME as resolved then.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix trivial typos in comments and warning message
Yannick Guerrini [Sat, 28 Feb 2015 16:20:41 +0000 (17:20 +0100)]
drm/i915: Fix trivial typos in comments and warning message

Change 'mutliple' to 'multiple'
Change 'mutlipler' to 'multiplier'
Change 'Haswel' to 'Haswell'

Signed-off-by: Yannick Guerrini <yguerrini@tomshardware.fr>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use plane->state->fb in watermark code (v2)
Matt Roper [Fri, 27 Feb 2015 18:12:01 +0000 (10:12 -0800)]
drm/i915: Use plane->state->fb in watermark code (v2)

plane->fb is a legacy pointer that not always be up-to-date (or updated
early enough).  Make sure the watermark code uses plane->state->fb so
that we're always doing our calculations based on the correct
framebuffers.

This patch was generated by Coccinelle with the following semantic
patch:

        @@
        struct drm_plane *P;
        @@
        - P->fb
        + P->state->fb

v2: Rebase

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Kill intel_crtc->cursor_{width, height} (v2)
Matt Roper [Fri, 27 Feb 2015 18:12:00 +0000 (10:12 -0800)]
drm/i915: Kill intel_crtc->cursor_{width, height} (v2)

The cursor size fields in intel_crtc just duplicate the data from
cursor->state.crtc_{w,h} so we don't need them any more.  Worse, their
use in the watermark code actually introduces a subtle bug since they
don't get updated to mirror the state values until the plane commit
stage, which is *after* we've already used them to calculate new
watermark values.  This happens because we had to move watermark updates
slightly earlier (outside vblank evasion) in commit

        commit 32b7eeec4d1e861230b09d437e95d76c86ff4a68
        Author: Matt Roper <matthew.d.roper@intel.com>
        Date:   Wed Dec 24 07:59:06 2014 -0800

            drm/i915: Refactor work that can sleep out of commit (v7)

Dropping the intel_crtc fields and just using the state values (which
are properly updated by the time watermark updates happen) should solve
the problem.

Aside from the actual removal of the struct fields (which are formatted
in a way that I couldn't figure out how to match in Coccinelle), the
rest of this patch was generated via the following semantic patch:

        // Drop assignment
        @@
        struct intel_crtc *C;
        struct drm_plane_state S;
        @@
        (
        - C->cursor_width = S.crtc_w;
        |
        - C->cursor_height = S.crtc_h;
        )

        // Replace usage
        @@
        struct intel_crtc *C;
        expression E;
        @@
        (
        - C->cursor_width
        + C->base.cursor->state->crtc_w
        |
        - C->cursor_height
        + C->base.cursor->state->crtc_h
        |
        - to_intel_crtc(E)->cursor_width
        + E->cursor->state->crtc_w
        |
        - to_intel_crtc(E)->cursor_height
        + E->cursor->state->crtc_h
        )

v2: Rebase

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Joe Konno <joe.konno@linux.intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89346
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Ensure plane->state->fb stays in sync with plane->fb
Xi Ruoyao [Thu, 12 Mar 2015 12:16:32 +0000 (20:16 +0800)]
drm/i915: Ensure plane->state->fb stays in sync with plane->fb

plane->state->fb and plane->fb should always reference the same FB so
that atomic and legacy codepaths have the same view of display state.
However, there are some places in kernel code that directly set
plane->fb and neglect to update plane->state->fb. If we never do a
successful update through the atomic pipeline, the RmFB cleanup code
will look at the plane->state->fb pointer, which has never actually
been set to a legitimate value, and try to clean it up, leading to
BUG's.

Add a quick helper function to synchronize plane->state->fb with
plane->fb and call it everywhere the driver tries to manually set
plane->fb outside of the atomic pipeline. In this function, use
drm_atomic_set_fb_for_plane instead of writing plane->state->fb
directly to keep the reference count right.

This is modified from Matt Roper's patch to drm-intel-nightly with
commit id

commit afd65eb4cc0578a9c07d621acdb8a570e2782bf7
Author: Matt Roper <matthew.d.roper@intel.com>
Date:   Tue Feb 3 13:10:04 2015 -0800

    drm/i915: Ensure plane->state->fb stays in sync with plane->fb

However this bug exists in mainline kernel too, so I created this to fix
it in mainline kernel.

A minor change is to use drm_atomic_set_fb_for_plane instead of update
reference count manually.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88909
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=93711
Signed-off-by: Xi Ruoyao <xry111@outlook.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
[Jani: included the patch notes in the commit message]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
9 years agoLinux 4.0-rc4
Linus Torvalds [Mon, 16 Mar 2015 00:38:20 +0000 (17:38 -0700)]
Linux 4.0-rc4

9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Sun, 15 Mar 2015 22:20:09 +0000 (15:20 -0700)]
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull drm fix from Dave Airlie:
 "An oops snuck in in an -rc3 patch, this fixes it"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  [PATCH] drm/mm: Fix support 4 GiB and larger ranges

9 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 15 Mar 2015 22:07:08 +0000 (15:07 -0700)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clock framework fixes from Michael Turquette:
 "The clk fixes for 4.0-rc4 comprise three themes.

  First are the usual driver fixes for new regressions since v3.19.

  Second are fixes to the common clock divider type caused by recent
  changes to how we round clock rates.  This affects many clock drivers
  that use this common code.

  Finally there are fixes for drivers that improperly compared struct
  clk pointers (drivers must not deref these pointers).  While some of
  these drivers have done this for a long time, this did not cause a
  problem until we started generating unique struct clk pointers for
  every consumer.  A new function, clk_is_match was introduced to get
  these drivers working again and they are fixed up to no longer deref
  the pointers themselves"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  ASoC: kirkwood: fix struct clk pointer comparing
  ASoC: fsl_spdif: fix struct clk pointer comparing
  ARM: imx: fix struct clk pointer comparing
  clk: introduce clk_is_match
  clk: don't export static symbol
  clk: divider: fix calculation of initial best divider when rounding to closest
  clk: divider: fix selection of divider when rounding to closest
  clk: divider: fix calculation of maximal parent rate for a given divider
  clk: divider: return real rate instead of divider value
  clk: qcom: fix platform_no_drv_owner.cocci warnings
  clk: qcom: fix platform_no_drv_owner.cocci warnings
  clk: qcom: Add PLL4 vote clock
  clk: qcom: lcc-msm8960: Fix PLL rate detection
  clk: qcom: Fix slimbus n and m val offsets
  clk: ti: Fix FAPLL parent enable bit handling

9 years ago[PATCH] drm/mm: Fix support 4 GiB and larger ranges
Krzysztof Kolasa [Sun, 15 Mar 2015 19:22:36 +0000 (20:22 +0100)]
[PATCH] drm/mm: Fix support 4 GiB and larger ranges

bad argument if(tmp)... in check_free_hole

fix oops: kernel BUG at drivers/gpu/drm/drm_mm.c:305!

[airlied: excellent, this was my task for today].

Signed-off-by: Krzysztof Kolasa <kkolasa@winsoft.pl>
Reviewed-by: Chris wilson <chris@chris-wilson.co.uk>
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agoMerge tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm...
Linus Torvalds [Sun, 15 Mar 2015 17:49:38 +0000 (10:49 -0700)]
Merge tag 'fixes-for-linus' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Arnd Bergmann:
 "This is a rather unpleasantly large set of bug fixes for arm-soc, Most
  of them because of cross-tree dependencies for Exynos where we should
  have figured out the right path to merge things before the merge
  window, and then the maintainer being unable to sort things out in
  time during a business trip.

  The other changes contained here are the usual collection:

  MAINTAINERS file updates
   - Gregory Clement is now a co-maintainer for the legacy Marvell EBU
     platforms
   - A MAINTAINERS entry for the Freescale Vybrid platform that was
     added last year
   - Matt Porter no longer works as a maintainer on Broadcom SoCs

  Build-time issues
   - A compile-time error for at91
   - Several minor DT fixes on at91, imx, exynos, socfpga, and omap
   - The new digicolor platform was not correctly enabled at all

  Configuration issues
   - Two defconfig fix for regressions using USB on versatile express
     and on OMAP3
   - Enabling all 8 CPUs on Allwinner/SUNxi
   - Enabling the new STiH410 platform to be usable

  Bug fixes in platform code
   - A missing barrier for socfpga
   - Fixing LPDDR1 self-refresh mode on at91
   - Fixing RTC interrupt numbers on Exynos3250
   - Fixing a cache-coherency issues in CPU power-down on Exynos5
   - Multiple small OMAP power management fixes"

* tag 'fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (69 commits)
  MAINTAINERS: Add myself as co-maintainer to the legacy support of the mvebu SoCs
  ARM: at91: pm_slowclock: fix the compilation error
  ARM: at91/dt: fix USB high-speed clock to select UTMI
  ARM: at91/dt: fix at91 udc compatible strings
  ARM: at91/dt: declare matrix node as a syscon device
  ARM: vexpress: update CONFIG_USB_ISP1760 option
  ARM: digicolor: add the machine directory to Makefile
  ARM: STi: Add STiH410 SoC support
  MAINTAINERS: add Freescale Vybrid SoC
  MAINTAINERS: Remove self as ARM mach-bcm co-maintainer
  ARM: imx6sl-evk: set swbst_reg as vbus's parent reg
  ARM: imx6qdl-sabresd: set swbst_reg as vbus's parent reg
  ARM: at91/dt: at91sam9261: fix clocks and clock-names in udc definition
  ARM: OMAP2+: Fix wl12xx on dm3730-evm with mainline u-boot
  ARM: OMAP: enable TWL4030_USB in omap2plus_defconfig
  ARM: dts: dra7x-evm: avoid possible contention while muxing on CAN lines
  ARM: dts: dra7x-evm: Don't use dcan1_rx.gpio1_15 in DCAN pinctrl
  ARM: dts: am43xx: fix SLEWCTRL_FAST pinctrl binding
  ARM: dts: am33xx: fix SLEWCTRL_FAST pinctrl binding
  ARM: dts: OMAP5: fix polling intervals for thermal zones
  ...

9 years agoMerge tag 'irqchip-fixes-4.0' of git://git.infradead.org/users/jcooper/linux
Linus Torvalds [Sun, 15 Mar 2015 17:41:30 +0000 (10:41 -0700)]
Merge tag 'irqchip-fixes-4.0' of git://git.infradead.org/users/jcooper/linux

Pull irqchip fixes from Jason Cooper:
 "armada-370-xp:
   - Chained per-cpu interrupts

  gic{,-v3,v3-its}"
   - Various fixes for safer operation"

* tag 'irqchip-fixes-4.0' of git://git.infradead.org/users/jcooper/linux:
  irqchip: gicv3-its: Support safe initialization
  irqchip: gicv3-its: Define macros for GITS_CTLR fields
  irqchip: gicv3-its: Add limitation to page order
  irqchip: gicv3-its: Use 64KB page as default granule
  irqchip: gicv3-its: Zero itt before handling to hardware
  irqchip: gic-v3: Fix out of bounds access to cpu_logical_map
  irqchip: gic: Fix unsafe locking reported by lockdep
  irqchip: gicv3-its: Fix unsafe locking reported by lockdep
  irqchip: gicv3-its: Iterate over PCI aliases to generate ITS configuration
  irqchip: gicv3-its: Allocate enough memory for the full range of DeviceID
  irqchip: gicv3-its: Fix ITS CPU init
  irqchip: armada-370-xp: Fix chained per-cpu interrupts