openwrt/staging/blogic.git
9 years agodrm/i915: Write the SDVO reg twice on IBX
Ville Syrjälä [Tue, 5 May 2015 14:17:33 +0000 (17:17 +0300)]
drm/i915: Write the SDVO reg twice on IBX

On IBX the SDVO/HDMI register write may be masked when enabling the
port, so it may need to written twice. The HDMI code does this, but
the SDVO code does not. Add the workaround to the SDVO code as well.

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: Fix DP enhanced framing for CPT
Ville Syrjälä [Tue, 5 May 2015 14:17:31 +0000 (17:17 +0300)]
drm/i915: Fix DP enhanced framing for CPT

Currently we're always enabling enhanced framing on CPT even if the sink
doesn't support it. Fix this up by actaully looking at what the sink
tells us.

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: Clean up the CPT DP .get_hw_state() port readout
Ville Syrjälä [Tue, 5 May 2015 14:17:30 +0000 (17:17 +0300)]
drm/i915: Clean up the CPT DP .get_hw_state() port readout

Define a TRANS_DP_PIPE_TO_PORT() to make the CPT DP .get_hw_state()
pipe readout neater.

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: Clarfify the DP code platform checks
Ville Syrjälä [Tue, 5 May 2015 14:17:29 +0000 (17:17 +0300)]
drm/i915: Clarfify the DP code platform checks

intel_dp.c is a mess with all the checks for different
platform/PCH variants and ports. Try to clean it up by recognizing
the following facts:
- IVB port A, and CPT port B/C/D are always the special cases
- VLV/CHV don't have port A
- Using the same kind of logic everywhere makes things much easier to
  parse

So let's move the IVB port A and PCH port B/C/D checks to be done first,
and let the other cases fall through, and always check for these things
using the same logic.

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: Remove the double register write from intel_disable_hdmi()
Ville Syrjälä [Tue, 5 May 2015 14:17:28 +0000 (17:17 +0300)]
drm/i915: Remove the double register write from intel_disable_hdmi()

IBX can have problems with the first write to the port register getting
masked when enabling the port. We are trying to apply the workaround
also when disabling the port where it's not needed, and we also try
to apply it for CPT/PPT as well which don't need it. Just kill it.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[danvet: Resolve conflict with the remove CHV if block.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove a bogus 12bpc "toggle" from intel_disable_hdmi()
Ville Syrjälä [Tue, 5 May 2015 14:17:27 +0000 (17:17 +0300)]
drm/i915: Remove a bogus 12bpc "toggle" from intel_disable_hdmi()

The IBX 12bpc port enable toggle is only relevant when enabling
the port, not when disabling it. Also this code doesn't actually
toggle anything, and essentially just writes the port register
one extra time. Furthermore CPT/PPT don't need such workarounds
and yet we include them. Just kill it.

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/skl: Deinit/init the display at suspend/resume
Damien Lespiau [Thu, 21 May 2015 15:37:48 +0000 (16:37 +0100)]
drm/i915/skl: Deinit/init the display at suspend/resume

We need to re-init the display hardware when going out of suspend. This
includes:

  - Hooking the PCH to the reset logic
  - Restoring CDCDLK
  - Enabling the DDB power

Among those, only the CDCDLK one is a bit tricky. There's some
complexity in that:

  - DPLL0 (which is the source for CDCLK) has two VCOs, each with a set
    of supported frequencies. As eDP also uses DPLL0 for its link rate,
    once DPLL0 is on, we restrict the possible eDP link rates the chosen
    VCO.
  - CDCLK also limits the bandwidth available to push pixels.

So, as a first step, this commit restore what the BIOS set, until I can
do more testing.

In case that's of interest for the reviewer, I've unit tested the
function that derives the decimal frequency field:

  #include <stdio.h>
  #include <stdint.h>
  #include <assert.h>

  #define ARRAY_SIZE(x) (sizeof(x) / sizeof(*(x)))

  static const struct dpll_freq {
          unsigned int freq;
          unsigned int decimal;
  } freqs[] = {
          { .freq = 308570, .decimal = 0b01001100111},
          { .freq = 337500, .decimal = 0b01010100001},
          { .freq = 432000, .decimal = 0b01101011110},
          { .freq = 450000, .decimal = 0b01110000010},
          { .freq = 540000, .decimal = 0b10000110110},
          { .freq = 617140, .decimal = 0b10011010000},
          { .freq = 675000, .decimal = 0b10101000100},
  };

  static void intbits(unsigned int v)
  {
          int i;

          for(i = 10; i >= 0; i--)
                  putchar('0' + ((v >> i) & 1));
  }

  static unsigned int freq_decimal(unsigned int freq /* in kHz */)
  {
          return (freq - 1000) / 500;
  }

  static void test_freq(const struct dpll_freq *entry)
  {
          unsigned int decimal = freq_decimal(entry->freq);

          printf("freq: %d, expected: ", entry->freq);
          intbits(entry->decimal);
          printf(", got: ");
          intbits(decimal);
          putchar('\n');

          assert(decimal == entry->decimal);
  }

  int main(int argc, char **argv)
  {
          int i;

          for (i = 0; i < ARRAY_SIZE(freqs); i++)
                  test_freq(&freqs[i]);

          return 0;
  }

v2:
  - Rebase on top of -nightly
  - Use (freq - 1000) / 500 for the decimal frequency (Ville)
  - Fix setting the enable bit of HSW_NDE_RSTWRN_OPT (Ville)
  - Rename skl_display_{resume,suspend} to skl_{init,uninit}_cdclk to
    be consistent with the BXT code (Ville)
  - Store boot CDCLK in ddi_pll_init (Ville)
  - Merge dev_priv's skl_boot_cdclk into cdclk_freq
  - Use LCPLL_PLL_LOCK instead of (1 << 30) (Ville)
  - Replace various '0' by SKL_DPLL0 to be a bit more explicit that
    we're programming DPLL0
  - Busy poll the PCU before doing the frequency change. It takes about
    3/4 cycles, each separated by 10us, to get the ACK from the CPU
    (Ville)

v3:
  - Restore dev_priv->skl_boot_cdclk, leaving unification with
    dev_priv->cdclk_freq for a later patch (Daniel, Ville)

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Free RPS boosts for all laggards
Chris Wilson [Mon, 27 Apr 2015 12:41:24 +0000 (13:41 +0100)]
drm/i915: Free RPS boosts for all laggards

If the client stalls on a congested request, chosen to be 20ms old to
match throttling, allow the client a free RPS boost.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: s/rq/req/]
[danvet: s/0/NULL/ reported by 0-day build]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't downclock whilst we have clients waiting for GPU results
Chris Wilson [Mon, 27 Apr 2015 12:41:23 +0000 (13:41 +0100)]
drm/i915: Don't downclock whilst we have clients waiting for GPU results

If we have clients stalled waiting for requests, ignore the GPU if it
signals that it should downclock due to low load. This helps prevent
the automatic timeout from causing extremely long running batches from
taking even longer.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Convert RPS tracking to a intel_rps_client struct
Chris Wilson [Mon, 27 Apr 2015 12:41:22 +0000 (13:41 +0100)]
drm/i915: Convert RPS tracking to a intel_rps_client struct

Now that we have internal clients, rather than faking a whole
drm_i915_file_private just for tracking RPS boosts, create a new struct
intel_rps_client and pass it along when waiting.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: s/rq/req/]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Limit mmio flip RPS boosts
Chris Wilson [Mon, 27 Apr 2015 12:41:21 +0000 (13:41 +0100)]
drm/i915: Limit mmio flip RPS boosts

Since we will often pageflip to an active surface, we will often have to
wait for the surface to be written before issuing the flip. Also we are
likely to wait on that surface in plenty of time before the vblank.
Since we have a mechanism for boosting when a flip misses the expected
vblank, curtain the number of times we RPS boost when simply waiting for
mmioflip.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: s/rq/req/]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Limit ring synchronisation (sw sempahores) RPS boosts
Chris Wilson [Mon, 27 Apr 2015 12:41:20 +0000 (13:41 +0100)]
drm/i915: Limit ring synchronisation (sw sempahores) RPS boosts

Ring switches can occur many times per frame, and are often out of
control, causing frequent RPS boosting for no practical benefit. Treat
the sw semaphore synchronisation as a separate client and only allow it
to boost once per busy/idle cycle.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: s/rq/req/]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Inline check required for object syncing prior to execbuf
Chris Wilson [Mon, 27 Apr 2015 12:41:18 +0000 (13:41 +0100)]
drm/i915: Inline check required for object syncing prior to execbuf

This trims a little overhead from the common case of not needing to
synchronize between rings.

v2: execlists is special and likes to duplicate code.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Implement inter-engine read-read optimisations
Chris Wilson [Mon, 27 Apr 2015 12:41:17 +0000 (13:41 +0100)]
drm/i915: Implement inter-engine read-read optimisations

Currently, we only track the last request globally across all engines.
This prevents us from issuing concurrent read requests on e.g. the RCS
and BCS engines (or more likely the render and media engines). Without
semaphores, we incur costly stalls as we synchronise between rings -
greatly impacting the current performance of Broadwell versus Haswell in
certain workloads (like video decode). With the introduction of
reference counted requests, it is much easier to track the last request
per ring, as well as the last global write request so that we can
optimise inter-engine read read requests (as well as better optimise
certain CPU waits).

v2: Fix inverted readonly condition for nonblocking waits.
v3: Handle non-continguous engine array after waits
v4: Rebase, tidy, rewrite ring list debugging
v5: Use obj->active as a bitfield, it looks cool
v6: Micro-optimise, mostly involving moving code around
v7: Fix retire-requests-upto for execlists (and multiple rq->ringbuf)
v8: Rebase
v9: Refactor i915_gem_object_sync() to allow the compiler to better
optimise it.

Benchmark: igt/gem_read_read_speed
hsw:gt3e (with semaphores):
Before: Time to read-read 1024k: 275.794µs
After:  Time to read-read 1024k: 123.260µs

hsw:gt3e (w/o semaphores):
Before: Time to read-read 1024k: 230.433µs
After:  Time to read-read 1024k: 124.593µs

bdw-u (w/o semaphores):             Before          After
Time to read-read 1x1:            26.274µs       10.350µs
Time to read-read 128x128:        40.097µs       21.366µs
Time to read-read 256x256:        77.087µs       42.608µs
Time to read-read 512x512:       281.999µs      181.155µs
Time to read-read 1024x1024:    1196.141µs     1118.223µs
Time to read-read 2048x2048:    5639.072µs     5225.837µs
Time to read-read 4096x4096:   22401.662µs    21137.067µs
Time to read-read 8192x8192:   89617.735µs    85637.681µs

Testcase: igt/gem_concurrent_blit (read-read and friends)
Cc: Lionel Landwerlin <lionel.g.landwerlin@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> [v8]
[danvet: s/\<rq\>/req/g]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: s/\<rq\>/req/g
Daniel Vetter [Thu, 21 May 2015 12:21:25 +0000 (14:21 +0200)]
drm/i915: s/\<rq\>/req/g

The merged seqno->request conversion from John called request
variables req, but some (not all) of Chris' recent patches changed
those to just rq. We've had a lenghty (and inconclusive) discussion on
irc which is the more meaningful name with maybe at most a slight bias
towards req.

Given that the "don't change names without good reason to avoid
conflicts" rule applies, so lets go back to a req everywhere for
consistency. I'll sed any patches for which this will cause conflicts
before applying.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: John Harrison <John.C.Harrison@Intel.com>
[danvet: s/origina/merged/ as pointed out by Chris - the first
mass-conversion patch was from Chris, the merged one from John.]
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agodrm/i915/skl: enable WaForceContextSaveRestoreNonCoherent
Imre Deak [Tue, 19 May 2015 14:05:42 +0000 (17:05 +0300)]
drm/i915/skl: enable WaForceContextSaveRestoreNonCoherent

v2:
- set the override disable flag too on stepping F0 (mika)

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: fix WaForceContextSaveRestoreNonCoherent on steppings B0+
Imre Deak [Tue, 19 May 2015 14:05:41 +0000 (17:05 +0300)]
drm/i915/bxt: fix WaForceContextSaveRestoreNonCoherent on steppings B0+

On B0 and C0 steppings the workaround enable bit would be overriden by
default, so the overriding must be disabled.

The WA was added in
commit 83a24979c40ebbf0fa0cd14df16f74142f373cd3
Author: Nick Hoath <nicholas.hoath@intel.com>
Date:   Fri Apr 10 13:12:26 2015 +0100

    drm/i915/bxt: Add WaForceContextSaveRestoreNonCoherent

Spotted-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Force clean compilation with -Werror
Chris Wilson [Wed, 20 May 2015 13:12:47 +0000 (14:12 +0100)]
drm/i915: Force clean compilation with -Werror

Our driver compiles clean (nowadays thanks to 0day) but for me, at least,
it would be beneficial if the compiler threw an error rather than a
warning when it found a piece of suspect code. (I use this to
compile-check patch series and want to break on the first compiler error
in order to fix the patch.)

v2: Kick off a new "Debugging" submenu for i915.ko

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Add "DRM i915" to the menu name as requested by Chris.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Kill the dev variable in intel_suspend_complete()
Damien Lespiau [Wed, 20 May 2015 13:45:16 +0000 (14:45 +0100)]
drm/i915: Kill the dev variable in intel_suspend_complete()

The macros we use there are the magic ones that can take either dev or
dev_priv. We'd like to move as much as possible towards dev_priv though.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Add a space after ', ' and don't capitalize mid-sentence
Damien Lespiau [Wed, 20 May 2015 13:45:15 +0000 (14:45 +0100)]
drm/i915: Add a space after ', ' and don't capitalize mid-sentence

Couldn't let it go!

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: Also add bxt_resume_prepare() to the S3/S4 path
Damien Lespiau [Wed, 20 May 2015 13:45:14 +0000 (14:45 +0100)]
drm/i915/bxt: Also add bxt_resume_prepare() to the S3/S4 path

Currently bxt_resume_prepare() is only used in the runtime-resume path.
Add it to the full S3/S4 path as well.

v2: Rebase on top of the vlv_resume_prepare() shuffling around

Cc: Imre Deak <imre.deak@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com> (v1)
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Swapping 90 and 270 to be compliant with Xrandr
Sonika Jindal [Wed, 20 May 2015 08:10:48 +0000 (13:40 +0530)]
drm/i915/skl: Swapping 90 and 270 to be compliant with Xrandr

Since DRM_ROTATE is counter clockwise (which is compliant with Xrandr),
and HW rotation is clockwise, swapping 90/270 to work as expected from
userspace.

v2: Rebased

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
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: Update comment in clear_intel_crtc_state()
Ander Conselvan de Oliveira [Wed, 20 May 2015 06:03:27 +0000 (09:03 +0300)]
drm/i915: Update comment in clear_intel_crtc_state()

Explain why a few fields of the new pipe_config have their values
preserved, while the others are zeroed.

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/skl: add F0 stepping ID
Imre Deak [Tue, 19 May 2015 12:05:00 +0000 (15:05 +0300)]
drm/i915/skl: add F0 stepping ID

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: limit WaDisableMaskBasedCammingInRCC to stepping A
Imre Deak [Tue, 19 May 2015 12:04:59 +0000 (15:04 +0300)]
drm/i915/bxt: limit WaDisableMaskBasedCammingInRCC to stepping A

Also make the WA comment consistent with the rest, where the stepping
info is not shown.

Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Tighten the exposure ARGB/ABGR 8888 formats
Damien Lespiau [Tue, 19 May 2015 11:29:16 +0000 (12:29 +0100)]
drm/i915: Tighten the exposure ARGB/ABGR 8888 formats

ARGB8888 is used for cursors on all platforms so we need to allow it
everywhere.

ABGR8888 is currently only honoured:
  - on VLV/CHV in sprite planes
  - on SKL+ for primary and sprite planes
so only allow it for those platforms.

Note that we only support ARGB8888/ABGR8888 on the primary plane for
SKL/BXT because we have in line of sight the pipe bottom color on those
platforms and because the primary plane programming on VLV/CHV doesn't
anything different for those formats today.

v2: Fix the logic to forbid the creation ABGR2101010 fbs (Ville)
v3: Still allow the creation of ARGB8888 fbs now that cursor planes use
    real fb objects (found by PRTS).

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix 'suspedn' typo
Damien Lespiau [Mon, 18 May 2015 18:53:48 +0000 (19:53 +0100)]
drm/i915: Fix 'suspedn' typo

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: Move around lane stagger calculation
Vandana Kannan [Wed, 13 May 2015 06:50:35 +0000 (12:20 +0530)]
drm/i915/bxt: Move around lane stagger calculation

Making lane stagger calculation common for HDMI and DP

v2: Imre's comments addressed
- Remove lane stagger from bxt_clk_div and make it a local variable in
ddi_pll_select

Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: Port PLL programming BUN
Vandana Kannan [Wed, 13 May 2015 06:48:52 +0000 (12:18 +0530)]
drm/i915/bxt: Port PLL programming BUN

BUN 1: prop_coeff, int_coeff, tdctargetcnt programming updated and tied to
VCO frequencies. Program i_lockthresh in PORT_PLL_9.

VCO calculated based on the formula:
Desired Output = Port bit rate in MHz (DisplayPort HBR2 is 5400 MHz)
Fast Clock = Desired Output / 2
VCO = Fast Clock * P1 * P2

Prop_coeff, int_coeff, and tdctargetcnt modified according to above
calculation.

BUN 2: Port PLLs require additional programming at certain frequencies -
DCO amplitude in PORT_PLL_10

Review comments from Siva which were addressed in the initial version of the
patch.
- Change PORT_PLL_LOCK_THRESHOLD to PORT_PLL_LOCK_THRESHOLD_MASK
- Calculate for HDMI
- Correct values for vco = 5.4
- return in case of invalid vco range

v2: Imre's review comments addressed
- change dcoampovr_en to dcoampovr_en_h
- change PORT_PLL_DCO_AMP_OVR_EN to PORT_PLL_DCO_AMP_OVR_EN_H
- Correct lane stagger value for 324MHz
- Make coef common for HDMI and DP
- remove superfluous comments

v3: Imre's comments addressed
- Remove Prop_coeff, int_coeff, tdctargetcnt, dcoampovr_en, gain_ctl,
dcoampovr_en_h from bxt_clk_div and make them local variables.

Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
Reviewed-by: Sivakumar Thulasimani <sivakumar.thulasimani@intel.com> [v1]
Cc: Sivakumar Thulasimani <sivakumar.thulasimani@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: add HAS_DP_MST feature test macro
Jani Nikula [Mon, 18 May 2015 14:10:01 +0000 (17:10 +0300)]
drm/i915: add HAS_DP_MST feature test macro

Be in line with other features that we have.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/dp: make link rate printing prettier
Jani Nikula [Mon, 18 May 2015 13:01:45 +0000 (16:01 +0300)]
drm/i915/dp: make link rate printing prettier

Turn

    [drm:intel_dp_print_rates] source rates: 162000,270000,540000,
    [drm:intel_dp_print_rates] sink rates: 162000,270000,
    [drm:intel_dp_print_rates] common rates: 162000,270000,

into

    [drm:intel_dp_print_rates] source rates: 162000, 270000, 540000
    [drm:intel_dp_print_rates] sink rates: 162000, 270000
    [drm:intel_dp_print_rates] common rates: 162000, 270000

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bxt: Update the Broxton PCI ids
Damien Lespiau [Fri, 15 May 2015 18:43:56 +0000 (19:43 +0100)]
drm/i915/bxt: Update the Broxton PCI ids

Cc: Imre Deak <imre.deak@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't expose ARGB1555 on gen2/3
Damien Lespiau [Fri, 15 May 2015 18:06:03 +0000 (19:06 +0100)]
drm/i915: Don't expose ARGB1555 on gen2/3

Signed-off-by: Damien Lespiau <damien.lespiau@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 ARBG/ABGR 2101010 on platform not supporting those formats
Damien Lespiau [Fri, 15 May 2015 18:06:01 +0000 (19:06 +0100)]
drm/i915: Remove ARBG/ABGR 2101010 on platform not supporting those formats

We just have have VLV and CHV sprites programming the hardware
differently for the ABGR2101010 so keep them working.

Signed-off-by: Damien Lespiau <damien.lespiau@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 the COMMON_PRIMARY_FORMATS defines
Damien Lespiau [Fri, 15 May 2015 18:06:00 +0000 (19:06 +0100)]
drm/i915: Remove the COMMON_PRIMARY_FORMATS defines

That define makes it hard to figure out what is the actual list of
formats at a glance. Expand it then.

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use uninterruptible mutex_lock for userptr bo creation
Chris Wilson [Fri, 15 May 2015 10:42:21 +0000 (11:42 +0100)]
drm/i915: Use uninterruptible mutex_lock for userptr bo creation

Mika encountered one pathological scenario under X where acquiring all
the mm locks (required to insert a mmu notifier) was very slow, so slow
that by the time we tried to lock the struct_mutex with the usual call
to i915_mutex_lock_interruptible(), X's signal timer had fired causing
us to restart the ioctl (and so looped indefinitely).

While I suspect this is the result of another bug (something leaking mm
perhaps?) we can forgo the error checking and interuptible nature of the
lock here so we only have to pay the expense once and get on with it.
This does expose the userptr creation routine to a driver livelock
though by not being interruptible.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
[danvet: Init ret to avoid issues reported by PRTS.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Preserve ddi_pll_sel when allocating new pipe_config
Ander Conselvan de Oliveira [Fri, 15 May 2015 08:51:50 +0000 (11:51 +0300)]
drm/i915: Preserve ddi_pll_sel when allocating new pipe_config

When the modeset code is reached with a CRTC that only needs a flip, the
code that assigns PLLs is skipped. But since there is still a state swap
for that CRTC, the current PLL assignment needs to be preserved. I
missed the ddi_pll_sel field in the following commit, which causes
warnings in DDI platforms.

commit 4978cc93d9ac240b435ce60431aef24239b4c270
Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Date:   Tue Apr 21 17:13:21 2015 +0300

    drm/i915: Preserve shared DPLL information in new pipe_config

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90410
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Dump some DPLL fields in pipe config debug
Tvrtko Ursulin [Thu, 14 May 2015 12:38:31 +0000 (13:38 +0100)]
drm/i915: Dump some DPLL fields in pipe config debug

v2: Split strings to 80 char, add ddi_pll_sel and fixed typo. (Damien Lespiau)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't overwrite (e)DP PLL selection on SKL
Ander Conselvan de Oliveira [Fri, 15 May 2015 10:34:29 +0000 (13:34 +0300)]
drm/i915: Don't overwrite (e)DP PLL selection on SKL

In the following commit, the place where the contents of dpll_hw_state
in crtc_state where zeroed was changed. Prior to that commit, it
happened when the new state was allocated, but now that happens just
before the call the .crtc_compute_clock() hook. The DP code for SKL,
however, sets up the (private) PLL in the encoder compute config
function that has already run by the time that memset() is reached,
causing the previous value to be lost.

This patch fixes the issue by moving the memset() down the call chain,
so that it is only called if the values in dpll_hw_state are going to be
updated.

commit 4978cc93d9ac240b435ce60431aef24239b4c270
Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Date:   Tue Apr 21 17:13:21 2015 +0300

    drm/i915: Preserve shared DPLL information in new pipe_config

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90462
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Reported-and-tested-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bios: be more explicit about discarding iomem address space
Jani Nikula [Wed, 13 May 2015 12:34:05 +0000 (15:34 +0300)]
drm/i915/bios: be more explicit about discarding iomem address space

Add one explicit discard of __iomem address space qualifier in
validate_vbt(), and respect it otherwise. This adds clarity in the code,
and reduces the sparse warnings from the module to just one.

Quoting Daniel, "The vbt really is plain old memory. Except that it's
reserved in the e820 table as something special and hence treated as io
range by the kernel. But it is memory, hence casting away the __iomem is
imo the right approach."

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Output scaler related pipe config debug in a single line
Tvrtko Ursulin [Wed, 13 May 2015 15:51:08 +0000 (16:51 +0100)]
drm/i915: Output scaler related pipe config debug in a single line

Just so it is grouped logically in line with other data and makes a
rather verbose output a bit shorter.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chandra Konduru <chandra.konduru@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Chandra Konduru <chandra.konduru@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bios: abstract finding VBT in BIOS to a separate function
Jani Nikula [Wed, 13 May 2015 12:34:04 +0000 (15:34 +0300)]
drm/i915/bios: abstract finding VBT in BIOS to a separate function

Improve clarity. No functional changes.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/bios: remove a redundant NULL pointer check
Jani Nikula [Wed, 13 May 2015 12:34:03 +0000 (15:34 +0300)]
drm/i915/bios: remove a redundant NULL pointer check

We never pass a non-NULL vbt to validate_vbt, and we can safely expect
the callers to not change.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Remove unnecessary local variables in skl_plane_ctl*()
Damien Lespiau [Fri, 15 May 2015 14:07:02 +0000 (15:07 +0100)]
drm/i915/skl: Remove unnecessary local variables in skl_plane_ctl*()

Ville noticed in another patch we we didn't need them at all, so remove
them. It's worth saying that it makes no difference to code generated as
gcc is clever enough to optimize it out.

v2: Remove 'break' after 'return' in switches (Ville)

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/dsi: add support for DSI PLL N1 divisor values
Jani Nikula [Wed, 13 May 2015 07:35:25 +0000 (10:35 +0300)]
drm/i915/dsi: add support for DSI PLL N1 divisor values

Currently DSI PLL N1 is hardcoded off. Make it possible to use it
later. This should have no functional changes for now.

v2: s/ffz(~(n))/ffs(n) - 1/ (Ville)

Signed-off-by: Jani Nikula <jani.nikula@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: Documentation for CSR firmware
Animesh Manna [Wed, 13 May 2015 16:43:29 +0000 (22:13 +0530)]
drm/i915/skl: Documentation for CSR firmware

Added docbook info regarding context save and restore (CSR)
firmware support added from gen9 onwards to drive newly added
DMC (Display microcontroller) in display engine.

v1: Initial version as RFC.

v2: Used "DOC:" tag for csr description based on review comment from Daniel.

Signed-off-by: Animesh Manna <animesh.manna@intel.com>
Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't read dpcd for disconnected ports
Mika Kuoppala [Fri, 15 May 2015 10:09:21 +0000 (13:09 +0300)]
drm/i915: Don't read dpcd for disconnected ports

Reading from disconnected ports will spit out timeout error
on the dmesg. Skip the attempted read if the port is not
connected and avoid confusing users/testcases about
expected timeouts.

This new dpcd debugfs entry was introduced by commit aa7471d228eb
("drm/i915: add i915 specific connector debugfs file for DPCD")

v2 by Jani: move the check at the top, out of the loop.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90060
Tested-by: yex.tian@intel.com
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Bob Paauwe <bob.j.paauwe@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/edid: fix a debug message
Dan Carpenter [Tue, 12 May 2015 18:07:37 +0000 (21:07 +0300)]
drm/edid: fix a debug message

There is an extra semi-colon on the if statement so the debug output
always says "Failed to write EDID checksum" even when it didn't fail.

Fixes: 559be30cb74d ('drm/i915: Implement the intel_dp_autotest_edid function for DP EDID complaince tests')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make the sprite formats const
Damien Lespiau [Tue, 12 May 2015 15:13:22 +0000 (16:13 +0100)]
drm/i915: Make the sprite formats const

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Don't use the intel_ prefix for gen-specific data
Damien Lespiau [Tue, 12 May 2015 15:13:18 +0000 (16:13 +0100)]
drm/i915: Don't use the intel_ prefix for gen-specific data

We now prefix our functions/enums/data with the first platform it has
been introduced. Do that for the primary plane formats.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: s/gen2/i8xx/ and s/gen4/i965/ ...]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Support the advertized index format
Damien Lespiau [Tue, 12 May 2015 15:13:17 +0000 (16:13 +0100)]
drm/i915/skl: Support the advertized index format

We advertize C8 in the primary plane formats didn't have the
corresponding code to set PLANE_CTL accordingly.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Replace BUG() by MISSING_CASE() in skl_plane_ctl_format()
Damien Lespiau [Tue, 12 May 2015 15:13:16 +0000 (16:13 +0100)]
drm/i915/skl: Replace BUG() by MISSING_CASE() in skl_plane_ctl_format()

Let's be consistent with the others skl_plane_ctl_*() functions and use
a MISSING_CASE(). Not only that, but it's a rude to BUG() the whole
machine here.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/skl: Rename a local variable to fit in 80 chars
Damien Lespiau [Tue, 12 May 2015 15:13:15 +0000 (16:13 +0100)]
drm/i915/skl: Rename a local variable to fit in 80 chars

No reason to not follow the 80 chars rule, renaming the local variable
makes it easy.

Cc: Chandra Konduru <chandra.konduru@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: Add a new line before return
Damien Lespiau [Tue, 12 May 2015 15:13:14 +0000 (16:13 +0100)]
drm/i915/skl: Add a new line before return

We usually use a new line before those kind of return statements. Also
the various skl_plane_ctl*() functions weren't consistent.

Cc: Chandra Konduru <chandra.konduru@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: Leave a new line between variable declarations and code
Damien Lespiau [Tue, 12 May 2015 15:13:13 +0000 (16:13 +0100)]
drm/i915/skl: Leave a new line between variable declarations and code

Cc: Chandra Konduru <chandra.konduru@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: clean up dsi pll calculation
Jani Nikula [Tue, 12 May 2015 12:23:09 +0000 (15:23 +0300)]
drm/i915: clean up dsi pll calculation

Improve readability. No functional changes.

v2: use more rational types (Ville)

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: don't register invalid gmbus pins for skl
Jani Nikula [Wed, 6 May 2015 12:33:44 +0000 (15:33 +0300)]
drm/i915: don't register invalid gmbus pins for skl

Do not expose invalid gmbus pins as i2c devices to userspace.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: don't register invalid gmbus pins for bdw
Jani Nikula [Wed, 6 May 2015 12:33:43 +0000 (15:33 +0300)]
drm/i915: don't register invalid gmbus pins for bdw

Do not expose invalid gmbus pins as i2c devices to userspace.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: constify validate_vbt in VBT parsing
Jani Nikula [Tue, 12 May 2015 12:41:32 +0000 (15:41 +0300)]
drm/i915: constify validate_vbt in VBT parsing

Make input and output of validate_vbt const, and fix the fallout. We
shouldn't modify the VBT, so make the compiler help us here.

v2: use pointer arithmetics on void* to simplify (Ville)

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: constify find_section in VBT parsing
Jani Nikula [Wed, 15 Apr 2015 12:18:28 +0000 (15:18 +0300)]
drm/i915: constify find_section in VBT parsing

Make input and output of find_section const, and fix the fallout. We
shouldn't modify the VBT, so make the compiler help us here.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Use HOTPLUG_INT_STATUS_G4X on VLV/CHV
Ville Syrjälä [Mon, 11 May 2015 17:49:10 +0000 (20:49 +0300)]
drm/i915: Use HOTPLUG_INT_STATUS_G4X on VLV/CHV

Use HOTPLUG_INT_STATUS_G4X instead of HOTPLUG_INT_STATUS_I915 on VLV/CHV
so that we don't confuse the AUX status bits with SDVO status bits.

Avoid pointless log spam as below while handling AUX interrupts:
[drm:intel_hpd_irq_handler] hotplug event received, stat 0x00000040, dig 0x00000000
[drm:intel_hpd_irq_handler] hotplug event received, stat 0x00000040, dig 0x00000000
[drm:intel_hpd_irq_handler] hotplug event received, stat 0x00000040, dig 0x00000000
[drm:intel_hpd_irq_handler] hotplug event received, stat 0x00000040, dig 0x00000000
[drm:intel_hpd_irq_handler] hotplug event received, stat 0x00000040, dig 0x00000000
[drm:intel_dp_aux_ch] dp_aux_ch timeout status 0x71450064

Note that there's no functional issue, it's just that the sdvo bits
overlap with the dp aux bits. Hence every time we receive an aux
interrupt we also think there's an sdvo hpd interrupt, but due to lack
of any sdvo encoders nothing ever happens because of that.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Add Ville's explanation why nothing functional really
changes.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove excess inline keywords
Ville Syrjälä [Mon, 11 May 2015 17:49:09 +0000 (20:49 +0300)]
drm/i915: Remove excess inline keywords

Remove some inline keywords. One of the functions has clearly outgrown
it anyway, so let's just leave it to the compiler.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/gtt: Fix the boundary check for vm area
Mika Kuoppala [Tue, 12 May 2015 07:35:08 +0000 (10:35 +0300)]
drm/i915/gtt: Fix the boundary check for vm area

The check for start + length >= total_vm_size is
wrong since start + length can be exactly the size of
the vm.

Fix the check to allow allocation to boundary.

Fixes a regression in commit 4dd738e9cd79
("drm/i915: Fix 32b overflow check in gen8_ppgtt_alloc_page_directories")

Testcase: igt/gem_evict_everything/swapping-interruptible
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90399
Tested-by: Lu Hua <huax.lu@intel.com>
Cc: Chris Wilson <chris@chris.wilson.co.uk>
Cc: Dave Gordon <david.s.gordon@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Make scaler_id check in check_crtc_state work for all gens
Chandra Konduru [Mon, 11 May 2015 21:35:47 +0000 (14:35 -0700)]
drm/i915: Make scaler_id check in check_crtc_state work for all gens

During check_crtc_state, scaler_id mispatch is being reported for HSW.
This is applicable for skl+ and not for HSW. It is introduced by
commit id:
    commit a1b2278e4dfcd2dbea85e319ebf73a6b7b2f180b
    Author: Chandra Konduru <chandra.konduru@intel.com>
    Date:   Tue Apr 7 15:28:45 2015 -0700

        drm/i915: skylake panel fitting using shared scalers

This patch will make sure that we leave scaler_id as 0 for platforms
before skl and set for skl+ only. This way scaler_id check during
check_crtc_state will pass for both prior to skl and skl+ platforms.

v2:
-Leave scaler_id as 0 for gen < 9 (Daniel)

Signed-off-by: Chandra Konduru <chandra.konduru@intel.com>
References: http://lists.freedesktop.org/archives/intel-gfx/2015-May/065741.html
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix race on unreferencing the wrong mmio-flip-request
Chris Wilson [Mon, 27 Apr 2015 12:41:16 +0000 (13:41 +0100)]
drm/i915: Fix race on unreferencing the wrong mmio-flip-request

As we perform the mmio-flip without any locking and then try to acquire
the struct_mutex prior to dereferencing the request, it is possible for
userspace to queue a new pageflip before the worker can finish clearing
the old state - and then it will clear the new flip request. The result
is that the new flip could be completed before the GPU has finished
rendering.

The bugs stems from removing the seqno checking in
commit 536f5b5e86b225dab94c7ff8061ae482b6077387
Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Date:   Thu Nov 6 11:03:40 2014 +0200

    drm/i915: Make mmio flip wait for seqno in the work function

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Ensure cache flushes prior to doing CS flips
Chris Wilson [Mon, 27 Apr 2015 12:41:15 +0000 (13:41 +0100)]
drm/i915: Ensure cache flushes prior to doing CS flips

Synchronising to an object active on the same ring is a no-op, for the
benefit of execbuffer scheduler. However, for CS flips this means that
we can forgo checking whether the last write request of the object is
actually queued and more importantly whether the cache flush for the
write was emitted.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove domain flubbing from i915_gem_object_finish_gpu()
Chris Wilson [Mon, 27 Apr 2015 12:41:14 +0000 (13:41 +0100)]
drm/i915: Remove domain flubbing from i915_gem_object_finish_gpu()

We no longer interpolate domains in the same manner, and even if we did,
we should trust setting either of the other write domains would trigger
an invalidation rather than force it. Remove the tweaking of the
read_domains since it serves no purpose and use
i915_gem_object_wait_rendering() directly.

Note that this goes back to

commit a8198eea156df47e0e843ac5c7d4c8774e121c42
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Wed Apr 13 22:04:09 2011 +0100

    drm/i915: Introduce i915_gem_object_finish_gpu()

and gpu domain tracking died in

commit cc889e0f6ce6a63c62db17d702ecfed86d58083f
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Wed Jun 13 20:45:19 2012 +0200

    drm/i915: disable flushing_list/gpu_write_list

which is more than 1 year older.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Add notes with information dug out of git history.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Warn when execlists changes context without IRQs
Peter Antoine [Mon, 11 May 2015 15:03:27 +0000 (16:03 +0100)]
drm/i915: Warn when execlists changes context without IRQs

If an batch ends while the IRQs are not turned on the notification can
go missing and the GPU can hang. So generate a warning in this case.

Signed-off-by: Peter Antoine <peter.antoine@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: call intel_tile_height with correct parameter
Chandra Konduru [Sat, 9 May 2015 03:22:46 +0000 (20:22 -0700)]
drm/i915: call intel_tile_height with correct parameter

In skylake update plane functions, intel_tile_height() is called with
bits_per_pixel instead of pixel_format. Correcting it.

Signed-off-by: Chandra Konduru <chandra.konduru@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
[danvet: Fixup alignment.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove duplicated intel_tile_height declaration
Tvrtko Ursulin [Fri, 8 May 2015 12:02:36 +0000 (13:02 +0100)]
drm/i915: Remove duplicated intel_tile_height declaration

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/chv: Set min freq to efficient frequency on chv
Deepak S [Sat, 9 May 2015 12:45:46 +0000 (18:15 +0530)]
drm/i915/chv: Set min freq to efficient frequency on chv

After feedback from the hardware team, now we set the GPU min/idel freq to RPe.
Punit is expecting us to operate GPU between Rpe & Rp0. If we drop the
frequency to RPn, punit is failing to change the vgg input voltage to
minimum :(

Since Punit validates the rps range [RPe, RP0]. This patch
removes unused cherryview_rps_min_freq function.

v2: Change commit message

v3: set min_freq before idle_freq (chris)

v4: Squash 'Remove unused rps min function' patch

Signed-off-by: Deepak S <deepak.s@linux.intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915/chv: Extend set idle rps wa to chv
Deepak S [Sat, 9 May 2015 12:41:54 +0000 (18:11 +0530)]
drm/i915/chv: Extend set idle rps wa to chv

It is observed on BSW that requesting a new frequency from Punit
does nothing when the GPU is in rc6, and if we let GPU enter rc6
with a high frequency, Vnn remains slightly higher than at minimum
frequency. Extending vlv_set_rps_idle() workaround on CHV/BSW.

v2: Update commit msg (Ville)

suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Deepak S <deepak.s@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/vlv: Remove wait for for punit to updates freq.
Deepak S [Sat, 9 May 2015 12:34:44 +0000 (18:04 +0530)]
drm/i915/vlv: Remove wait for for punit to updates freq.

When GPU is idle on VLV, Request freq to punit should be good enough to
get the voltage back to VNN. Also, make sure gfx clock force applies
before requesting the freq fot vlv.

v2: Do forcewake before setting idle frequency (ville)
    Update function comments to match the code (Deepak)

v3: Fix get/put across idle frequency Request. (Ville)

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=75244
suggested-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Deepak S <deepak.s@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: Be optimistic about future display engines having 7 WM levels
Damien Lespiau [Sat, 9 May 2015 01:05:55 +0000 (02:05 +0100)]
drm/i915: Be optimistic about future display engines having 7 WM levels

As we're doing throughout the code, being optimistic that platform n + 1
will mostly reuse the same things as platform n allows us to minimize
the enabling work needed.

This time, it's about the number of WM levels.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Always keep crtc_state->active in sync with enable
Maarten Lankhorst [Mon, 11 May 2015 08:45:15 +0000 (10:45 +0200)]
drm/i915: Always keep crtc_state->active in sync with enable

With the recent modeset internal rework, we wind up setting crtc_state->enable
to false, but leave crtc_state->active as true following a
drmModeSetCrtc(fb=0), which is incorrect.  This mismatch gets caught by
drm_atomic_crtc_check() and causes subsequent atomic operations (such as plane
updates while the CRTC is disabled) to fail.

Bisect points to

        commit dad9a7d6d96630182fb52aae7c3856e9e7285e13
        Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
        Date:   Tue Apr 21 17:13:19 2015 +0300

            drm/i915: Use atomic helpers for computing changed flags

as the commit that actually triggers the regression.

The difference compared to (which this patch reverts)

commit 90d469067d0808ddbd9be2c97a4a8e14037b5e46
Author: Matt Roper <matthew.d.roper@intel.com>
Date:   Thu May 7 14:31:28 2015 -0700

    drm/i915: Set crtc_state->active to false when CRTC is disabled (v2)

is that we know keep state->active/enable in sync for all legacy
modeset paths, as it should be.

Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Reported-and-Tested-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
[danvet: Directly squash in the revert and augment the commit
message.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Revert "drm/i915: Set crtc_state->active to false when CRTC is disabled (v2)"

This reverts commit 90d469067d0808ddbd9be2c97a4a8e14037b5e46.

9 years agodrm/i915: Adding dbuf support for skl nv12 format.
Chandra Konduru [Mon, 27 Apr 2015 22:47:37 +0000 (15:47 -0700)]
drm/i915: Adding dbuf support for skl nv12 format.

Skylake nv12 format requires dbuf (aka. ddb) calculations
and programming for each of y and uv sub-planes. Made minor
changes to reuse current dbuf calculations and programming
for uv plane. i.e., with this change, existing computation
is used for either packed format or uv portion of nv12
depending on incoming format. Added new code for dbuf
computation and programming for y plane.

This patch is a pre-requisite for adding NV12 format support.
Actual nv12 support is coming in later patches.

Signed-off-by: Chandra Konduru <chandra.konduru@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Fix typo in intel_runtime_pm.c
Masanari Iida [Sat, 9 May 2015 16:00:23 +0000 (01:00 +0900)]
drm/i915: Fix typo in intel_runtime_pm.c

This patch fix spelling typo in intel_runtime_pm.c

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
9 years agodrm/i915: Remove unused variable from i915_gem_mmap_gtt
Daniel Vetter [Fri, 8 May 2015 15:46:40 +0000 (17:46 +0200)]
drm/i915: Remove unused variable from i915_gem_mmap_gtt

Lost in

commit c5ad54cf7dd8922bd1cee2d5871aebf73dc9638e
Author: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Date:   Wed May 6 14:36:09 2015 +0300

    drm/i915: Use partial view in mmap fault handler

Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
9 years agoBackmerge v4.1-rc4 into into drm-next
Dave Airlie [Wed, 20 May 2015 06:23:53 +0000 (16:23 +1000)]
Backmerge v4.1-rc4 into into drm-next

We picked up a silent conflict in amdkfd with drm-fixes and drm-next,
backmerge v4.1-rc5 and fix the conflicts

Signed-off-by: Dave Airlie <airlied@redhat.com>
Conflicts:
drivers/gpu/drm/drm_irq.c

9 years agoMerge tag 'drm-amdkfd-next-2015-05-19' of git://people.freedesktop.org/~gabbayo/linux...
Dave Airlie [Tue, 19 May 2015 23:20:48 +0000 (09:20 +1000)]
Merge tag 'drm-amdkfd-next-2015-05-19' of git://people.freedesktop.org/~gabbayo/linux into drm-next

- Add the interrupts & events modules, including new IOCTLs to create and wait
  on events. The HSA RT open source stack is mainly using events to know when
  a dispatched work has been completed. In addition, this module is
  a pre-requisite for the next module I'm going to upstream - debugger support

  This module also handles H/W exceptions, such as memory exception received
  through the IOMMUv2 H/W and Bad Opcode exception receieved from the GPU.

- Adding a new kernel module parameter to let the user decide whether he wants
  to receive a SIGTERM when a memory exception occurs inside the GPU kernel and
  the HSA application doesn't wait on an appropriate event, or if he just want
  to receive notification about this event in dmesg. The default is the latter.

- Additional improvements for SDMA code

- Update my email address in Maintainers file.

* tag 'drm-amdkfd-next-2015-05-19' of git://people.freedesktop.org/~gabbayo/linux:
  drm/amdkfd: change driver version to 0.7.2
  drm/amdkfd: Implement events IOCTLs
  drm/amdkfd: Add module parameter of send_sigterm
  drm/amdkfd: Add bad opcode exception handling
  drm/amdkfd: Add memory exception handling
  drm/amdkfd: Add the events module
  drm/amdkfd: add events IOCTL set definitions
  drm/amdkfd: Add interrupt handling module
  drm/radeon: Add init interrupt kfd->kgd interface
  MAINTAINERS: update amdkfd Oded's email address
  drm/amdkfd: make the sdma vm init to be asic specific
  drm/amdkfd: Use new struct for asic specific ops
  drm/amdkfd: reformat some debug prints
  drm/amdkfd: Remove unessary void pointer cast

9 years agoMerge tag 'topic/drm-misc-2015-05-19' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Tue, 19 May 2015 23:19:58 +0000 (09:19 +1000)]
Merge tag 'topic/drm-misc-2015-05-19' of git://anongit.freedesktop.org/drm-intel into drm-next

Scattering of random drm core patches. Bunch of atomic prep work too, but
the final bits for blob properties, atomic modesets and lifting the
experimental tag on the atomic ioctl are still blocked on Daniel Stone
finalizing and testing the weston support for it. I hope that we can get
it all ready for 4.2 though.

* tag 'topic/drm-misc-2015-05-19' of git://anongit.freedesktop.org/drm-intel: (22 commits)
  drm/atomic: Allow drivers to subclass drm_atomic_state, v3
  drm/atomic: remove duplicated assignment of old_plane_state
  drm/dp: Fix comment in DP helper
  drm/atomic: add drm_atomic_get_existing_*_state helpers
  drm/core: get rid of -Iinclude/drm
  drm/i915: get rid of -Iinclude/drm
  drm/atomic-helpers: Export drm_atomic_helper_update_legacy_modeset_state
  drm/atomic-helpers: Update vblank timestamping constants
  drm/sysfs: remove unnecessary connector type checks
  drm/sysfs: split DVI-I and TV-out attributes
  drm/sysfs: make optional attribute groups per connector type
  drm/sysfs: add a helper for extracting connector type from kobject
  drm/edid: Add CEA modes before inferred modes
  drm/prime: Allow internal imports without import_sg_table
  drm: Add reference counting to blob properties
  drm: Introduce blob_lock
  drm: Introduce helper for replacing blob properties
  drm: Don't leak path blob property when updating
  drm/atomic: Don't open-code CRTC state destroy
  drm/edid: Add DMT modes with ID > 0x50
  ...

9 years agodrm/amdkfd: change driver version to 0.7.2
Oded Gabbay [Tue, 21 Apr 2015 12:09:48 +0000 (15:09 +0300)]
drm/amdkfd: change driver version to 0.7.2

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Implement events IOCTLs
Andrew Lewycky [Tue, 9 Sep 2014 12:22:05 +0000 (15:22 +0300)]
drm/amdkfd: Implement events IOCTLs

Signed-off-by: Andrew Lewycky <Andrew.Lewycky@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add module parameter of send_sigterm
Oded Gabbay [Wed, 24 Dec 2014 11:30:52 +0000 (13:30 +0200)]
drm/amdkfd: Add module parameter of send_sigterm

This patch adds a new kernel module parameter to amdkfd,
called send_sigterm.

This parameter specifies whether amdkfd should send the
SIGTERM signal to an HSA process, when the following conditions
occur:

1. The GPU triggers an exception regarding a kernel that was
   issued by this process.

2. The HSA process isn't waiting on an event that handles
   this exception.

The default behavior is not to send a SIGTERM and suffice
with a dmesg error print.

Reviewed-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add bad opcode exception handling
Alexey Skidanov [Tue, 25 Nov 2014 08:34:31 +0000 (10:34 +0200)]
drm/amdkfd: Add bad opcode exception handling

Signed-off-by: Alexey Skidanov <alexey.skidanov@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add memory exception handling
Alexey Skidanov [Tue, 14 Apr 2015 15:05:49 +0000 (18:05 +0300)]
drm/amdkfd: Add memory exception handling

This patch adds Peripheral Page Request (PPR) failure processing
and reporting.

Bad address or pointer to a system memory block with inappropriate
read/write permission cause such PPR failure during a user queue
processing. PPR request handling is done by IOMMU driver notifying
AMDKFD module on PPR failure.

The process triggering a PPR failure will be notified by
appropriate event or SIGTERM signal will be sent to it.

v3:
- Change all bool fields in struct kfd_memory_exception_failure to
  uint32_t

Signed-off-by: Alexey Skidanov <alexey.skidanov@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add the events module
Andrew Lewycky [Sun, 10 May 2015 09:15:46 +0000 (12:15 +0300)]
drm/amdkfd: Add the events module

This patch adds the events module (kfd_events.c) and the interrupt
handle module for Kaveri (cik_event_interrupt.c).

The patch updates the interrupt_is_wanted(), so that it now calls the
interrupt isr function specific for the device that received the
interrupt. That function(implemented in cik_event_interrupt.c)
returns whether this interrupt is of interest to us or not.

The patch also updates the interrupt_wq(), so that it now calls the
device's specific wq function, which checks the interrupt source
and tries to signal relevant events.

v2:

Increase limit of signal events to 4096 per process
Remove bitfields from struct cik_ih_ring_entry
Rename radeon_kfd_event_mmap to kfd_event_mmap
Add debug prints to allocate_free_slot and allocate_signal_page
Make allocate_event_notification_slot return a correct value
Add warning prints to create_signal_event
Remove error print from IOCTL path
Reformatted debug prints in kfd_event_mmap
Map correct size (as received from mmap) in kfd_event_mmap

v3:

Reduce limit of signal events back to 256 per process
Fix allocation of kernel memory for signal events

Signed-off-by: Andrew Lewycky <Andrew.Lewycky@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: add events IOCTL set definitions
Andrew Lewycky [Sun, 7 Dec 2014 15:05:11 +0000 (17:05 +0200)]
drm/amdkfd: add events IOCTL set definitions

- AMDKFD_IOC_CREATE_EVENT:
Creates a new event of a specified type

- AMDKFD_IOC_DESTROY_EVENT:
Destroys an existing event

- AMDKFD_IOC_SET_EVENT:
Signal an existing event

- AMDKFD_IOC_RESET_EVENT:
Reset an existing event

- AMDKFD_IOC_WAIT_EVENTS:
Wait on event(s) until they are signaled

v2:

- Move the limit of the signal events to kfd_ioctl.h so it
  can be used by userspace

v3:
- Change all bool fields in struct kfd_memory_exception_failure
to uint32_t

Signed-off-by: Andrew Lewycky <Andrew.Lewycky@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Add interrupt handling module
Andrew Lewycky [Wed, 16 Jul 2014 22:37:30 +0000 (01:37 +0300)]
drm/amdkfd: Add interrupt handling module

This patch adds the interrupt handling module, kfd_interrupt.c, and its
related members in different data structures to the amdkfd driver.

The amdkfd interrupt module maintains an internal interrupt ring
per amdkfd device. The internal interrupt ring contains interrupts
that needs further handling. The extra handling is deferred to
a later time through a workqueue.

There's no acknowledgment for the interrupts we use. The hardware
simply queues a new interrupt each time without waiting.

The fixed-size internal queue means that it's possible for us to lose
interrupts because we have no back-pressure to the hardware.

However, only interrupts that are "wanted" by amdkfd, are copied into
the amdkfd s/w interrupt ring, in order to minimize the chances
for overflow of the ring.

Signed-off-by: Andrew Lewycky <Andrew.Lewycky@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/radeon: Add init interrupt kfd->kgd interface
Oded Gabbay [Thu, 5 Mar 2015 13:13:18 +0000 (15:13 +0200)]
drm/radeon: Add init interrupt kfd->kgd interface

This patch adds a new interface function to the kfd->kgd interface.
The function is kgd_init_interrupts() and its function is to
initialize a pipe's interrupts.

The function currently enables the timestamp interrupt and the
bad opcode interrupt.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agoMAINTAINERS: update amdkfd Oded's email address
Oded Gabbay [Sun, 10 May 2015 09:37:28 +0000 (12:37 +0300)]
MAINTAINERS: update amdkfd Oded's email address

Leaving AMD soon so need to update my email address to @gmail.com

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: make the sdma vm init to be asic specific
Oded Gabbay [Tue, 5 May 2015 08:51:39 +0000 (11:51 +0300)]
drm/amdkfd: make the sdma vm init to be asic specific

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
9 years agodrm/amdkfd: Use new struct for asic specific ops
Oded Gabbay [Tue, 5 May 2015 08:51:31 +0000 (11:51 +0300)]
drm/amdkfd: Use new struct for asic specific ops

This patch creates a new structure for asic specific operations, instead
of using the existing structure of operations.

This is done to make the code flow more logic, readable and maintainable.

The change is done only to the device queue manager module at this point.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
9 years agodrm/amdkfd: reformat some debug prints
Oded Gabbay [Mon, 4 May 2015 12:53:15 +0000 (15:53 +0300)]
drm/amdkfd: reformat some debug prints

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
9 years agodrm/amdkfd: Remove unessary void pointer cast
Firo Yang [Thu, 23 Apr 2015 09:58:05 +0000 (17:58 +0800)]
drm/amdkfd: Remove unessary void pointer cast

kmalloc() returns a void pointer - no need to cast it in
drivers/gpu/drm/amd/amdkfd/kfd_process.c::kfd_process_destroy_delayed()

Signed-off-by: Firo Yang <firogm@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@amd.com>
9 years agoMerge tag 'drm-intel-next-2015-05-08' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Tue, 19 May 2015 00:18:13 +0000 (10:18 +1000)]
Merge tag 'drm-intel-next-2015-05-08' of git://anongit.freedesktop.org/drm-intel into drm-next

- skl plane scaler support (Chandra Kondru)
- enable hsw cmd parser (Daniel and fix from Rebecca Palmer)
- skl dc5/6 support (low power display modes) from Suketu&Sunil
- dp compliance testing patches (Todd Previte)
- dp link training optimization (Mika Kahola)
- fixes to make skl resume work (Damien)
- rework modeset code to fully use atomic state objects (Ander&Maarten)
- pile of bxt w/a patchs from Nick Hoath
- (linear) partial gtt mmap support (Joonas Lahtinen)

* tag 'drm-intel-next-2015-05-08' of git://anongit.freedesktop.org/drm-intel: (103 commits)
  drm/i915: Update DRIVER_DATE to 20150508
  drm/i915: Only wait for required lanes in vlv_wait_port_ready()
  drm/i915: Fix possible security hole in command parsing
  drm/edid: Kerneldoc for newly added edid_corrupt
  drm/i915: Reject huge tiled objects
  Revert "drm/i915: Hack to tie both common lanes together on chv"
  drm/i915: Work around DISPLAY_PHY_CONTROL register corruption on CHV
  drm/i915: Implement chv display PHY lane stagger setup
  drm/i915/vlv: remove wait for previous GFX clk disable request
  drm/i915: Set crtc_state->active to false when CRTC is disabled (v2)
  drm/i915/skl: Re-indent part of skl_ddi_calculate_wrpll()
  drm/i915: Use partial view in mmap fault handler
  drm/i915: Add a partial GGTT view type
  drm/i915: Consider object pinned if any VMA is pinned
  drm/i915: Do not make assumptions on GGTT VMA sizes
  drm/i915/bxt: Mark WaCcsTlbPrefetchDisable as for Broxton also.
  drm/i915/bxt: Mark WaDisablePartialResolveInVc as for Broxton also.
  drm/i915/bxt: Mark Wa4x4STCOptimizationDisable as for Broxton also.
  drm/i915/bxt: Move WaForceEnableNonCoherent to Skylake only
  drm/i915/bxt: Enable WaEnableYV12BugFixInHalfSliceChicken7 for Broxton
  ...

9 years agoLinux 4.1-rc4
Linus Torvalds [Mon, 18 May 2015 17:13:47 +0000 (10:13 -0700)]
Linux 4.1-rc4

9 years agowatchdog: Fix merge 'conflict'
Peter Zijlstra [Mon, 18 May 2015 09:31:50 +0000 (11:31 +0200)]
watchdog: Fix merge 'conflict'

Two watchdog changes that came through different trees had a non
conflicting conflict, that is, one changed the semantics of a variable
but no actual code conflict happened. So the merge appeared fine, but
the resulting code did not behave as expected.

Commit 195daf665a62 ("watchdog: enable the new user interface of the
watchdog mechanism") changes the semantics of watchdog_user_enabled,
which thereafter is only used by the functions introduced by
b3738d293233 ("watchdog: Add watchdog enable/disable all functions").

There further appears to be a distinct lack of serialization between
setting and using watchdog_enabled, so perhaps we should wrap the
{en,dis}able_all() things in watchdog_proc_mutex.

This patch fixes a s2r failure reported by Michal; which I cannot
readily explain. But this does make the code internally consistent
again.

Reported-and-tested-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge tag 'for-linus-20150516' of git://git.infradead.org/linux-mtd
Linus Torvalds [Mon, 18 May 2015 17:01:54 +0000 (10:01 -0700)]
Merge tag 'for-linus-20150516' of git://git.infradead.org/linux-mtd

Pull MTD fixes from Brian Norris:
 "Two MTD fixes for 4.1:

   - readtest: the signal-handling code was clobbering the error codes
     we should be handling/reporting in this test, rendering it useless.
     Noticed by Coverity.

   - the common SPI NOR flash DT binding (merged for 4.1-rc1) is being
     revised, so let's change that before 4.1 is minted"

* tag 'for-linus-20150516' of git://git.infradead.org/linux-mtd:
  Documentation: dt: mtd: replace "nor-jedec" binding with "jedec, spi-nor"
  mtd: readtest: don't clobber error reports

9 years agodrm/atomic: Allow drivers to subclass drm_atomic_state, v3
Maarten Lankhorst [Mon, 18 May 2015 08:06:40 +0000 (10:06 +0200)]
drm/atomic: Allow drivers to subclass drm_atomic_state, v3

Drivers may need to store the state of shared resources, such as PLLs
or FIFO space, into the atomic state. Allow this by making it possible
to subclass drm_atomic_state.

Changes since v1:
- Change member names for functions to atomic_state_(alloc,clear)
- Change __drm_atomic_state_new to drm_atomic_state_init
- Allow free function to be overridden too, in case extra memory is
  allocated in alloc.
Changes since v2:
- Rename *_default_free to default_release, to make clear it doesn't
  free the state object itself.

Cc: dri-devel@lists.freedesktop.org
Acked-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>