When we setup buffer for display plane, we'll check any pending
required GPU flush and possible make interruptible wait for flush
complete. But that wait would be most possibly to fail in case of
signals received for X process, which will then fail modeset process
and put display engine in unconsistent state. The result could be
blank screen or CPU hang, and DDX driver would always turn on outputs
DPMS after whatever modeset fails or not.

So this one creates new helper for setup display plane buffer, and
when needing flush using uninterruptible wait for that.

This one should fix bug like https://bugs.freedesktop.org/show_bug.cgi?id=24009.
Also fixing mode switch stress test on Ironlake.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Without this fix, some modes couldn't find appropriate clocks.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>
Tested-by: NMatthew Garrett <mjg@redhat.com>

For any given clock we can use the find_pll to get the corresponding DPLL
setting. It is unnecessary to use the find_reduce_pll callback function
to calculate the DPLL parameter for LVDS downclock in order to get the same
divider factor(P) for the normal and downclock.

In theory when the LVDS downclock is supported by LVDS panel, we should get the
same DPLL divider factor(P) for the normal clock and reduced downclock.
If we get the diferent divider factor(P) for normal clock and reduced downclock,
it means that the found downclock is incorrect and should be discarded.

So we should use find_pll callback to calculate the DPLL parameter for the
LVDS reduced downclock as for the normal clock. Then we can do the cleanup
about find_reduced_pll.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
cc: Jesse Barnes <jbarnes@virtuousgeek.org>
cc: Matthew Garrett <mjg@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

For some clocks, the old Ironlake DPLL calculator wold give m/n/p
combinations that didn't match the spreadsheet of what HW validation
tests.  Instead, use the G4X DPLL calculator, which does a better job
at it.

So we use the intel_g4x_find_best_pll to calculate the DPLL for CRT/HDMI/LVDS
on ironlake. At the same time to consider the dpll setting for display port, we
add the display port DPLL limit on ironlake, which will directly use the
function of intel_find_pll_ironlake_dp to get the corresponding dpll setting.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Select the correct BPC for LVDS on Ironlake. If it is 18-bit LVDS panel,
the BPC will be 6. When it is 24-bit LVDS panel, the BPC will 8.
At the same time the BPC will be 8 when the output device is CRT/HDMI/DP.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Make the BPC in FDI rx/transcoder be consistent with that in pipeconf on Ironlake.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Enable/disable the dithering for LVDS based on VBT setting. On the 965/g4x
platform the dithering flag is defined in LVDS register. And on the ironlake
the dithering flag is defined in pipeconf register.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Hugh found an error path where we were attempting to unref a bo without
holding the struct mutex:

  [drm:intel_init_clock_gating] *ERROR* failed to pin power context: -16
  ------------[ cut here ]------------
  WARNING: at drivers/gpu/drm/drm_gem.c:438 drm_gem_object_free+0x20/0x5e()
  Hardware name: ESPRIMO Mobile V5505
  Modules linked in: snd_pcm_oss snd_mixer_oss snd_seq snd_seq_device
  Pid: 3793, comm: s2ram Not tainted 2.6.33-rc2 #4
  Call Trace:
   [<7815298e>] warn_slowpath_common+0x59/0x6b
   [<781529b3>] warn_slowpath_null+0x13/0x18
   [<78317c1a>] ? drm_gem_object_free+0x20/0x5e
   [<78317c1a>] drm_gem_object_free+0x20/0x5e
   [<78317bfa>] ? drm_gem_object_free+0x0/0x5e
   [<7829df11>] kref_put+0x38/0x45
   [<7833a5f0>] intel_init_clock_gating+0x232/0x271
   [<78317bfa>] ? drm_gem_object_free+0x0/0x5e
   [<7832c307>] i915_restore_state+0x21a/0x2b3
   [<7832379d>] i915_resume+0x3c/0xbb
   [<78174fe5>] ? trace_hardirqs_on_caller+0xfc/0x123
   [<7831c756>] ? drm_class_resume+0x0/0x3e
   [<7831c78d>] drm_class_resume+0x37/0x3e
   [<78351e0a>] legacy_resume+0x1e/0x51
   [<78351ece>] device_resume+0x91/0xab
   [<7831c756>] ? drm_class_resume+0x0/0x3e
   [<78352226>] dpm_resume+0x58/0x10f
   [<783522fb>] dpm_resume_end+0x1e/0x2c
   [<78180f80>] suspend_devices_and_enter+0x61/0x84
   [<78180ff8>] enter_state+0x55/0x83
   [<7818091c>] state_store+0x94/0xaa
   [<7829d09e>] kobj_attr_store+0x1e/0x23
   [<782098e0>] sysfs_write_file+0x66/0x99
   [<781cd2f0>] vfs_write+0x8a/0x108
   [<781cd408>] sys_write+0x3c/0x63
   [<78125c10>] sysenter_do_call+0x12/0x36
  ---[ end trace a343537f29950fda ]---

It is in fact slightly more insiduous that first appears since we are
attempting to not just free the object without the lock, but are trying
to do the whole bo manipulation without holding the lock.
Reported-by: NHugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org
Signed-off-by: NEric Anholt <eric@anholt.net>

The C99 specification states in section 6.11.5:

The placement of a storage-class specifier other than at the beginning
of the declaration specifiers in a declaration is an obsolescent
feature.
Signed-off-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NEric Anholt <eric@anholt.net>

This code generally fails to adjust the render clock, and when it does,
it conflicts with some other register settings and can cause problems.

So remove this code altogether.  I'm reworking it now to do the right
thing, but the only bit it will share is the VBT check for whether
reclocking is supported, so I'm leaving that bit.

Reverts most of 652c393a ("add dynamic
clock frequency control"), though for many the regressions showed up
in the later 181a5336 ("Fix render
reclock availability detection").
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

We restored RC6 twice on resume, even with modesetting off.  Instead,
only restore it once and skip RC6 initialization entirely in non-KMS mode.
Signed-off-by: NAndy Lutomirski <luto@mit.edu>
Tested-by: NJeff Chua <jeff.chua.linux@gmail.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

This patch changes around our hotplug enable code a bit to only enable
it for ports we actually detect and initialize.  This prevents problems
with stuck or spurious interrupts on outputs that aren't actually wired
up, and is generally more correct.

Fixes FDO bug #23183.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NKristian Høgsberg <krh@bitplanet.net>
Signed-off-by: NEric Anholt <eric@anholt.net>

Rather than restoring just a few clock gating registers on resume,
just reinitialize the whole thing.
Signed-off-by: NAndy Lutomirski <luto@mit.edu>
[anholt: Fixed up for RC6 support landed since the patch was written]
Signed-off-by: NEric Anholt <eric@anholt.net>

Both radeon and nouveau can re-use this code so move it up a level
so they can. However the hw interfaces for aux ch are different
enough that the code to translate from mode, address, bytes
to actual hw interfaces isn't generic, so move that code into the
Intel driver.
Signed-off-by: NDave Airlie <airlied@redhat.com>

IGD* isn't a useful name.  Replace with the codenames, as sourced from
pci.ids.
Signed-off-by: NAdam Jackson <ajax@redhat.com>
[anholt: Fixed up for merge with pineview/ironlake changes]
Signed-off-by: NEric Anholt <eric@anholt.net>

If we don't detect a supported memory configuration, we can't enable
CxSR.  Warn the user in this case so they can file a bug.

This patch brings the tree up to date with some fixes that were in a
more recent version of the page flipping patch you applied.  It fixes
pre-965 flip support, removes a leftover hack that forced alignment,
and initializes the pipe & plane CRTC mappings.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

In current vblank-wait implementation, if we turn off VGA output,
drm_wait_vblank will still wait on the disabled pipe until timeout,
because vblank on the pipe is assumed be enabled. This would cause
slow system response on some system such as moblin.

This patch resolve the issue by adding a drm helper function
drm_vblank_off which explicitly clear vblank_enabled[crtc], wake up
any waiting queue and save last vblank counter before turning off
crtc. It also slightly change drm_vblank_get to ensure that we will
will return immediately if trying to wait on a disabled pipe.
Signed-off-by: NLi Peng <peng.li@intel.com>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
[anholt: hand-applied for conflicts with overlay changes]
Signed-off-by: NEric Anholt <eric@anholt.net>

Only update the render-clock on transition from busy to idle and vice
versa, or else we burn a significant percentage of the cpu just rewriting
the register -- not quite as power-friendly as intended ;-)
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NEric Anholt <eric@anholt.net>

Acked-by: NJakob Bornecrantz <jakob@vmware.com>
Acked-by: NThomas Hellström <thomas@shipmail.org>
Review-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NJesse "Orange Smoothie" Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NKristian Høgsberg <krh@bitplanet.net>
Signed-off-by: NEric Anholt <eric@anholt.net>

PineView only has 2 ports for LVDS and CRT. Don't enable other
ports for it.

Cc: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Otherwise the chip may scribble over free memory.
Signed-off-by: NKristian Høgsberg <krh@bitplanet.net>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

This also extends the mutex to cover fbc disabling, which is safe.
Signed-off-by: NKristian Høgsberg <krh@bitplanet.net>
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

20ms delay is quite big and the routine isn't called in atomic context.
better use msleep to let other tasks run. This can reduce cpu time used
by Xorg, so potentially boost boot.
Signed-off-by: NShaohua Li <shaohua.li@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

In disable sequence, all output ports on PCH have to be disabled
before PCH transcoder, but LVDS port was left always enabled. This
one fixes that by disable LVDS port properly during pipe disable
process, and resolved stability issue seen on Ironlake. Also move
panel fitting disable time just after pipe disable to align with
the spec.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

The DPLL calculation logic for 9xx platform is changed in:
commit 652c393a
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date:   Mon Aug 17 13:31:43 2009 -0700

    drm/i915: add dynamic clock frequency control

Maybe we will get the different M/N/P combination with that by using the
previous dpll calculation logic.

So restore the DPLL calculation logic for 9xx platform.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

If more than one mode with the same resolution defined in EDID has different
refresh rate, it is thought that the downclock is found for LVDS.
We will program the different FPx0/1 register so that we can select dynamically
between the low and high frequency.

On the g4x platform we will use the CxSR feature to switch the different
refresh rate if the LVDS downclock feature is supported.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Reviewed-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Somehow the case for G33 got dropped while porting from ums code.
This made a 400MHz chip into a 133MHz one which resulted in the
unnecessary enabling of double wide pipe mode which in turn
screwed up the overlay code.

Nothing else (than the overlay code) seems to be affected.

This fixes fdo.org bug #24835
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NEric Anholt <eric@anholt.net>

Need to calculate the SR watermark and enable it.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

Signed-off-by: NKristian Høgsberg <krh@redhat.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Replace the DRM_DEBUG with DRM_DEBUG_KMS in output device code.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Replace the DRM_DEBUG with DRM_DEBUG_DRIVER in generic i915 driver.
Then the debug info can be obtained by adding the boot option of
"drm.debug=0x02".

At the same time the debug info in increase/decrease clock is also
printed by using DRM_DEBUG_DRIVER instead of DRM_DEBUG_KMS.
Signed-off-by: NZhao Yakui <yakui.zhao@intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

At least for the common case of userspace ioctls. When doing a
modeset operation, the wait is still uninterruptible. But considering
that failing to turn off the overlay when switching off the crtc it's
running on hangs the chip, it doesn't complicate matters _very_
much. There's just an unkillable X in addition to a black screen.
BUG() about it and explain in the code.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NEric Anholt <eric@anholt.net>

This implements intel overlay support for kms via a device-specific
ioctl. Thomas Hellstrom brought up the idea of a general ioctl (on
dri-devel). We've reached the conclusion that such an infrastructure
only makes sense when multiple kms overlay implementations exists,
which atm don't (and it doesn't look like this is gonna change).

Open issues:
- Runs in sync with the gpu, i.e. unnecessary waiting. I've decided
  to wait on this because the hw tends to hang when changing something
  in this area. I left some dummy functions as infrastructure.
- polyphase filtering uses a static table.
- uses uninterruptible sleeps. Unfortunately the alternatives may
  unnecessarily wedged the hw if/when we timeout too early (and
  userspace only overloaded the batch buffers with stuff worth a few
  secs of gpu time).

Changes since v1:
- fix off-by-one misconception on my side. This fixes fullscreen
  playback.
Changes since v2:
- add underrun detection as spec'ed for i965.
- flush caches properly, fixing visual corruptions.
Changes since v4:
- fix up cache flushing of overlay memory regs.
- killed require_pipe_a logic - it hangs the chip.

Tested-By: diego.abelenda@gmail.com (on a 865G)
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
[anholt: Resolved against the MADVISE ioctl going in before this one]
Signed-off-by: NEric Anholt <eric@anholt.net>

It is identical to I85X. Use that one instead.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
[anholt: fix conflicts against the display function pointer stuff]
Signed-off-by: NEric Anholt <eric@anholt.net>

Render standy allows the GPU to power down the render unit when idle.
In order for this to work, it needs a page of graphics memory to save
state.  This patch allocates that page and enables the feature on
supported chipsets.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>

When the ideal error range can't be reached, this will safely use
a most closed one. Clean up some dumb codes in DPLL function too.

This fixes DPLL clock issue against one monitor at 1680x1050@60hz.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NEric Anholt <eric@anholt.net>

Initialize clock.vco to silence gcc.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NEric Anholt <eric@anholt.net>