Based on an original patch by Zhenyu Wang, this initializes the BLT ring for
SandyBridge and enables support for user execbuffers.

Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

... before someone tries to use it. The code both calls
intel_ring_begin/advance() and open-codes the bookkeeping performed by
those two functions.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

In commit 8187a2b7, the number of dwords used in the ringbuffer for
executing the batch buffer was erroneously changed from 2 to 4.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

If the userspace driver is using a constant relocation array with a
static buffer, they will pass the same relocation array back to the
kernel. So we *do* need to update the presumed offset value in those
relocations to reflect the current object so that they remain correct
with future batchbuffers and we avoid the necessity of having to suspend
execution and perform redundant relocations.

Fixes the regression introduced by 12f889c for applications using
absolute addressing on trees of buffer (i.e. the current consumers of
libdrm_intel.so).

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=30996Reported-by: NWang, Jinjin <jinjin.wang@intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

To handle retirements, we need per-ring tracking of active objects.
To handle evictions, we need global tracking of active objects.

As we enable more rings, rebuilding the global list from the individual
per-ring lists quickly grows tiresome and overly complicated. Tracking the
active objects in two lists is the lesser of two evils.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

... by always initialising the empty ringbuffer it is always then safe
to check whether it is active.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

The most frequent relocation within a batchbuffer is a contiguous sequence
of vertex buffer relocations, for which we can virtually eliminate the
drm_gem_object_lookup() overhead by caching the last handle to object
translation.

In doing so we refactor the pin and relocate retry loop out of
do_execbuffer into its own helper function and so improve the error
paths.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

One of the primarily consumers of the i915 driver is X, a large signal
driven application. Frequently when writing into the buffers, there is a
pending signal which causes us not to take the interruptible lock but
then we need to take that same lock around the object unreference. By
rearranging the code to do the interruptible lock as the first check, we
can avoid the frequent additional locking around the unreference.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

"userpace" -> "userspace"
Signed-off-by: NAndrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

This was a missing piece from 41a51428 that dropped recognition of the
AGP module for the second B43 variant.
Reported-by: NStefan Bader <stefan.bader@canonical.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: stable@kernel.org

... to avoid the double acquisition along fast[er] paths.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

... to avoid reacquiring it to drop the object reference count on
exit. Note we have to make sure we now drop (and reacquire) the lock
around acquiring the mm semaphore on the slow paths.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

... in the hope that it makes the atomic fast paths more likely.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

After allocation a handle for the fresh object, we know that we can
safely drop the refcnt without triggering a free so we do not need the
mutex. Strangely, this mutex acquisition is the one that appears on
driver profiles.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Avoid an early eviction of the batch buffer into the uncached GTT
domain, and so do the relocation fixup in cacheable memory.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

... perform an access validation check up front instead and copy them in
on-demand, during i915_gem_object_pin_and_relocate(). As around 20% of
the CPU overhead may be spent inside vmalloc for the relocation entries
when submitting an execbuffer [for x11perf -aa10text], the savings are
considerable and result in around a 10% throughput increase [for glyphs].
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Allow the user to override the detection of the sink's audio capabilities
from EDID. Not all sinks support the required EDID level to specify
whether they handle audio over the display connection, so allow the user
to enable it manually.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Allow the user to override the detection of the sink's audio capabilities
from EDID. Not all sinks support the required EDID level to specify
whether they handle audio over the display connection, so allow the user
to enable it manually.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Allow the user to override the detection of the sink's audio capabilities
from EDID. Not all sinks support the required EDID level to specify
whether they handle audio over the display connection, so allow the user
to enable it manually.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Rely on monitor's audio capability to turn on audio output for HDMI.
Tested-by: NWu Fengguang <fengguang.wu@intel.com>
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

This will turn on DP audio output by checking monitor's audio
capability.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
[ickle: rebase onto recent changes and rearranged for clarity]
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

To help to determine if digital display port needs to enable
audio output or not. This one adds a helper to get monitor's
audio capability via EDID CEA extension block.
Tested-by: NWu Fengguang <fengguang.wu@intel.com>
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Reviewed-by: NAdam Jackson <ajax@redhat.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

The time between start of the pixel clock and backlight enable is a basic
panel timing constraint.  If the Panel Power On/Off registers are found
to be 0, assume we are booting without VBIOS initialization and set these
registers to something reasonable.

Change-Id: Ibed6cc10d46bf52fd92e0beb25ae3525b5eef99d
Signed-off-by: NBryan Freed <bfreed@chromium.org>
[ickle: rearranged into a separate function to distinguish its role from
simply parsing the VBIOS tables.]
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

If userspace is submitting so many long running batches that the ring
becomes full, throttle by sleeping for a 1ms before checking for free
space. Simply yielding was causing excessive scheduler overhead whilst
making no progress.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

In i2c GPIO fallback, index 6 is reserved for nothing.
Signed-off-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

FDI_PLL_BIOS_0 register is for Ironlake only, don't apply to
Sandybridge.
Original-patch-by: NZhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Since the PLL may still be on, and the training pattern may not be
correct.  Fixes suspend/resume on my PCH eDP test system.
Signed-off-by: NJesse Barnes <jbarnes@virtuousgeek.org>
[ickle: minor merge conflict and silence the compiler]
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Freeing the Hardware Status Page was writing to the HWS register in
order to disable the GPU writing to the HWS page. Unfortunately, we were
writing to the mmio register after unmapping the register space, hence
the oops.
Signed-off-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>

This reverts commit 6939a5ac.

Daniel Vetter supplied a set of fixes for all the module unload bugs he
could trigger on his machines, so let the fun recommence!

The enter argument as implemented by commit 413d45d3 (drm, kdb, kms:
Add an enter argument to mode_set_base_atomic() API) should be more
descriptive as to what it does vs just passing 1 and 0 around.

There is no runtime behavior change as a result of this patch.
Reported-by: NJesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
CC: David Airlie <airlied@linux.ie>
CC: dri-devel@lists.freedesktop.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

When changing VTs non-atomically the kernel works in conjunction with
the Xserver in user space and receives the LUT information from the
Xserver via a system call.  When changing modes atomically for kdb,
this information must be saved and restored without disturbing user
space as if nothing ever happened.

There is a short cut used by this patch where gamma_store is used as
the save space.  If this turns out to be a problem in the future a
pre-allocated chunk of memory will be required for each crtc to save
and restore the LUT information.
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
CC: Jesse Barnes <jbarnes@virtuousgeek.org>
CC: David Airlie <airlied@linux.ie>
CC: dri-devel@lists.freedesktop.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

This reverts commit ff773714.

A generic solution is needed to save and retore the LUT information.

CC: Jesse Barnes <jbarnes@virtuousgeek.org>
CC: dri-devel@lists.freedesktop.org
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

TTM-based DRM drivers need to be able to bind user memory to the AGP
aperture. This patch fixes the "[TTM] AGP Bind memory failed." errors
and the subsequent fallout seen with the nouveau driver.
Signed-off-by: NFrancisco Jerez <currojerez@riseup.net>
Tested-by: NGrzesiek Sójka <pld@pfu.pl>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Function ttm_bo_wait_unreserved can be slightly simplified.
Signed-off-by: NJean Delvare <khali@linux-fr.org>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

This reverts commit f28488c2.

On my rv610 test machine the monitor failed to light up after this.
Signed-off-by: NDave Airlie <airlied@redhat.com>

We need the unlocked variant for the new codepath introduced to fix the
race condition in master recently.
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fixes cursor corruption in certain cases.
Signed-off-by: NAlex Deucher <alexdeucher@gmail.com>
Cc: stable@kernel.org
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fix SDIO suspend/resume regression introduced by 4c2ef25f "mmc: fix
all hangs related to mmc/sd card insert/removal during suspend/resume":

  PM: Syncing filesystems ... done.
  Freezing user space processes ... (elapsed 0.01 seconds) done.
  Freezing remaining freezable tasks ... (elapsed 0.01 seconds) done.
  Suspending console(s) (use no_console_suspend to debug)
  pm_op(): platform_pm_suspend+0x0/0x5c returns -38
  PM: Device pxa2xx-mci.0 failed to suspend: error -38
  PM: Some devices failed to suspend

4c2ef25f moved the card removal/insertion mechanism out of MMC's
suspend/resume path and into pm notifiers (mmc_pm_notify), and that
broke SDIO's expectation that mmc_suspend_host() will remove the card,
and squash the error, in case -ENOSYS is returned from the bus suspend
handler (mmc_sdio_suspend() in this case).

mmc_sdio_suspend() is using this whenever at least one of the card's SDIO
function drivers does not have suspend/resume handlers - in that case
it is agreed to force removal of the entire card.

This patch fixes this regression by trivially bringing back that part of
mmc_suspend_host(), which was removed by 4c2ef25f.
Reported-and-tested-by: NSven Neumann <s.neumann@raumfeld.com>
Signed-off-by: NOhad Ben-Cohen <ohad@wizery.com>
Cc: Maxim Levitsky <maximlevitsky@gmail.com>
Cc: <stable@kernel.org>
Acked-by: NNicolas Pitre <nico@fluxnic.net>
Signed-off-by: NChris Ball <cjb@laptop.org>

Commit 07934481 "DMAENGINE: generic channel status v2" changed the interface for
how dma channel progress is retrieved.  It inadvertently exported an internal
helper function ioat_tx_status() instead of ioat_dma_tx_status().  The latter
polls the hardware to get the latest completion state, while the helper just
evaluates the current state without touching hardware.  The effect is that we
end up waiting for completion timeouts or descriptor allocation errors before
the completion state is updated.

iperf (before fix):
[SUM]  0.0-41.3 sec   364 MBytes  73.9 Mbits/sec

iperf (after fix):
[SUM]  0.0- 4.5 sec   499 MBytes   940 Mbits/sec

This is a regression starting with 2.6.35.

Cc: <stable@kernel.org>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Jesse Brandeburg <jesse.brandeburg@intel.com>
Cc: Linus Walleij <linus.walleij@stericsson.com>
Cc: Maciej Sosnowski <maciej.sosnowski@intel.com>
Reported-by: NRichard Scobie <richard@sauce.co.nz>
Signed-off-by: NDan Williams <dan.j.williams@intel.com>