* 'drm-intel-fixes' of git://people.freedesktop.org/~keithp/linux: (25 commits)
  drm/i915: Fix inconsistent backlight level during disabled
  drm, i915: Fix memory leak in i915_gem_busy_ioctl().
  drm/i915: Use DPCD value for max DP lanes.
  drm/i915: Initiate DP link training only on the lanes we'll be using
  drm/i915: Remove trailing white space
  drm/i915: Try harder during dp pattern 1 link training
  drm/i915: Make DP prepare/commit consistent with DP dpms
  drm/i915: Let panel power sequencing hardware do its job
  drm/i915: Treat PCH eDP like DP in most places
  drm/i915: Remove link_status field from intel_dp structure
  drm/i915: Move common PCH_PP_CONTROL setup to ironlake_get_pp_control
  drm/i915: Module parameters using '-1' as default must be signed type
  drm/i915: Turn on another required clock gating bit on gen6.
  drm/i915: Turn on a required 3D clock gating bit on Sandybridge.
  drm/i915: enable cacheable objects on Ivybridge
  drm/i915: add constants to size fence arrays and fields
  drm/i915: Ivybridge still has fences!
  drm/i915: forcewake warning fixes in debugfs
  drm/i915: Fix object refcount leak on mmappable size limit error path.
  drm/i915: Use mode_config.mutex in ironlake_panel_vdd_work
  ...

This adds a new optional chunk to the CS ioctl that specifies optional flags
to the CS parser. Why this is useful is explained below. Note that some regs
no longer need the NOP relocation packet if this feature is enabled.
Tested on r300g and r600g with this flag disabled and enabled.

Assume there are two contexts sharing the same mipmapped tiled texture.
One context wants to render into the first mipmap and the other one
wants to render into the last mipmap. As you probably know, the hardware
has a MACRO_SWITCH feature, which turns off macro tiling for small mipmaps,
but that only applies to samplers.
(at least on r300-r500, though later hardware likely behaves the same)

So we want to just re-set the tiling flags before rendering (writing
packets), right? ... No. The contexts run in parallel, so they may
set the tiling flags simultaneously and then fire their command streams
also simultaneously. The last one setting the flags wins, the other one
loses.

Another problem is when one context wants to render into the first and
the last mipmap in one CS. Impossible. It must flush before changing
tiling flags and do the rendering into the smaller mipmaps in another CS.

Yet another problem is that writing copy_blit in userspace would be a mess
involving re-setting tiling flags to please the kernel, and causing races
with other contexts at the same time.

The only way out of this is to send tiling flags with each CS, ideally
with each relocation. But we already do that through the registers.
So let's just use what we have in the registers.
Signed-off-by: Marek Olšák <maraeo@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

When the brightness property is inquired while the backlight is disabled,
the driver returns a wrong value (zero) because it probes the value after
the backlight was turned off.  This caused a black screen even after the
backlight is enabled again.  It should return the internal backlight_level
instead, so that it won't be influenced by the backlight-enable state.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=41926
BugLink: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/872652Tested-by: Kamal Mostafa <kamal@canonical.com>
Cc: Alex Davis <alex14641@yahoo.com>
Cc: <stable@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Keith Packard <keithp@keithp.com>

A call to i915_add_request() has been made in function i915_gem_busy_ioctl(). i915_add_request can fail,
so in it's exit path previously allocated memory needs to be freed.
Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>

The BIOS VBT value for an eDP panel has been shown to be incorrect on
one machine, and we haven't found any machines where the DPCD value
was wrong, so we'll use the DPCD value everywhere.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Limit the link training setting command to the lanes needed for the
current mode. It seems vaguely possible that a monitor will try to
train the other lanes and fail in some way, so this seems like the
safer plan.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Found a couple of bare tabs in intel_dp.c
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Instead of going through the sequence just once, run through the whole
set up to 5 times to see if something can work. This isn't part of the
DP spec, but the BIOS seems to do it, and given that link training
failure is so bad, it seems reasonable to follow suit.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Make sure the sequence of operations in all three functions makes
sense:

 1) The backlight must be off unless the screen is running
 2) The link must be running to turn the eDP panel on/off
 3) The CPU eDP PLL must be running until everything is off
Signed-off-by: Keith Packard <keithp@keithp.com>

The panel power sequencing hardware tracks the stages of panel power
sequencing and signals when the panel is completely on or off. Instead
of blindly assuming the panel timings will work, poll the panel power
status register until it shows the correct values.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

PCH eDP has many of the same needs as regular PCH DP connections,
including the DP_CTl bit settings, the TRANS_DP_CTL register.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

No persistent data was ever stored here, so link_status is instead
allocated on the stack as needed.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>

Every usage of PCH_PP_CONTROL sets the PANEL_UNLOCK_REGS value to
ensure that writes will be respected, move this to a common function
to make the driver cleaner.

No functional changes.
Signed-off-by: Keith Packard <keithp@keithp.com>

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Fixes i2c test failures when i2c_algo_bit.bit_test=1.

The hw doesn't actually require a mask, so just set it
to the default mask bits for r1xx-r4xx radeon ddc.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Noticed by Egbert.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Cc: Egbert Eich <eich@suse.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>

I assumed all PCI buses had a bridge, but playing with qemu recently, I
discovered vgaarb bug where it wasn't detecting both devices shared a bridge
at the root level.

Don't check for NULL, if two buses have a NULL bridge, assume they share the
root bus.
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

commit 27641c3f (drm/vblank: Add support for precise vblank
timestamping) adds preempt_disable()/enable() around a spin locked
section with the comments:

 * Disable preemption, so vblank_time_lock is held as short as
 * possible, even under a kernel with PREEMPT_RT patches.

/* Disable preemption while holding vblank_time_lock. Do
 * it explicitely to guard against PREEMPT_RT kernel.

Just that this has never been tested on a RT kernel which would have
granted that nonsense with a might_sleep() warning because
dev->vblank_time_lock is converted to a "sleeping" spinlock on RT.

So this is activly wrong on RT and superflous on mainline. Remove it.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Mario Kleiner <mario.kleiner@tuebingen.mpg.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>

I missed the combios path when I updated the atombios pm code.

Reported by amarsh04 on IRC.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

On newer chips the number of clock modes per power state varies.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Avoid a lot of extra loops through the pm state array.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

The new power tables need to be handled differently when setting
up the profiles.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

It's already called via the DPMS functions.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Signed-off-by: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Fix kconfig unmet dependency warning.  BACKLIGHT_CLASS_DEVICE depends on
BACKLIGHT_LCD_SUPPORT, so select the latter along with the former.

warning: (DRM_RADEON_KMS && DRM_I915 && STUB_POULSBO && FB_BACKLIGHT && PANEL_SHARP_LS037V7DW01 && PANEL_ACX565AKM && USB_APPLEDISPLAY && FB_OLPC_DCON && ASUS_LAPTOP && SONY_LAPTOP && THINKPAD_ACPI && EEEPC_LAPTOP && ACPI_ASUS && ACPI_CMPC && SAMSUNG_Q10) selects BACKLIGHT_CLASS_DEVICE which has unmet direct dependencies (HAS_IOMEM && BACKLIGHT_LCD_SUPPORT)
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Cc: David Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

* 'drm-nouveau-fixes' of git://git.freedesktop.org/git/nouveau/linux-2.6:
  drm/nouveau: Fix bandwidth calculation for DisplayPort
  drm/nouveau: by default use low bpp framebuffer on low memory cards
  drm/nv10: Change the BO size threshold determining the memory placement range.
  drm/nvc0: enable acceleration for nvc1 by default
  drm/nvc0/gr: fixup the mmio list register writes for 0xc1
  drm/nvc1: hacky workaround to fix accel issues
  drm/nvc0/gr: fix some bugs in grctx generation
  drm/nvc0: enable acceleration on 0xc8 by default
  drm/nvc0/vram: skip disabled PBFB subunits
  drm/nv40/pm: fix issues on igp chipsets, which don't have memory
  drm/nouveau: testing the wrong variable
  drm/nvc0/vram: storage type 0xc3 is not compressed
  drm/nv50: fix stability issue on NV86.
  drm/nouveau: initialize chan->fence.lock before use
  drm/nv50/vram: fix incorrect detection of bank count on newer chipsets
  drm/nv50/gr: typo fix, how about we not reset fifo during graph init?
  drm/nv50/bios: fixup mpll programming from the init table parser
  drm/nouveau: fix oops if i2c bus not found in nouveau_i2c_identify()
  drm: make sure drm_vblank_init() has been called before touching vbl_lock

during the review of the fix for locks problems in drm_wait_vblank,
a couple of false concerns were raised about how the drm_vblank_get
and drm_vblank_put are used in this function; it turned out that the
code is correct and that it cannot be simplified

add a few comments to explain non-obvious flows in the code,
to prevent "false alarms" in the future

v2: incorporate comments received from Daniel Vetter
Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>

radeon_benchmark_do_move() returns an int so "time" should be int
too.  Making it unsigned breaks the error handling.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drm_wait_vblank must be DRM_UNLOCKED because otherwise it
will grab the drm_global_mutex and then go to sleep until the vblank
event it is waiting for. That can wreck havoc in the windowing system
because if one process issues this ioctl, it will block all other
processes for the duration of all vblanks between the current and the
one it is waiting for. In some cases it can block the entire windowing
system.

v2: incorporate comments received from Daniel Vetter and
    Michel Daenzer.

v3/v4: after a lengty discussion with Daniel Vetter, it was concluded
       that the only thing not yet protected with locks and atomic
       ops is the write to dev->last_vblank_wait. It's only used in a
       debug file in proc, and the current code already employs no
       correct locking: the proc file only takes dev->struct_mutex,
       whereas drm_wait_vblank implicitly took the drm_global_mutex.
       Given all this, it's not worth bothering to try to fix
       the locks at this time.
Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>

As Exynos DRM is merged at mainline. Also update the maintainer entry.
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

We restore the CRTC, encoder, and connector configurations, but if the
mode set failed, the attached display may have been turned off, so we
need to try set_config again to restore things to the way they were.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Can happen when there is no DP panel attached, confusing
users.  Make it debug only.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>

slow-work got killed in commit 181a51f6. This means that since v2.6.36
there is no Kconfig symbol SLOW_WORK. Apparently selecting that symbol
is a nop. Drop that select.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Nouveau, when configured with debugfs, creates debugfs files for every
channel, so structure holding list of files needs to be protected from
simultaneous changes by multiple threads.

Without this patch it's possible to hit kernel oops in
drm_debugfs_remove_files just by running a couple of xterms with
looped glxinfo.
Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This hunk seems to have gotten lost when I rebased the patch.
Reported-by: Sylvain Bertrand <sylvain.bertrand@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This was only the case if the GPU reset was triggered from the CS ioctl,
otherwise other processes could happily enter the CS ioctl and wreak havoc
during the GPU reset.

This is a little complicated because the GPU reset can be triggered from the
CS ioctl, in which case we're already holding the mutex, or from other call
paths, in which case we need to lock the mutex. AFAICT the mutex API doesn't
allow recursive locking or finding out the mutex owner, so we need to handle
this with helper functions which allow recursive locking from the same
process.
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Fixes Coverity buffer not null terminated defect.
Signed-off-by: Vinson Lee <vlee@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Snooping code expects this to be the case.
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>