Before changing the status of a buffer with a pending write we will await
upon a new flush for that buffer. So we can take advantage of any flushes
posted whilst the buffer is active and pending processing by the GPU, by
clearing its write_domain and updating its last_rendering_seqno -- thus
saving a potential flush in deep queues and improves flushing behaviour
upon eviction for both GTT space and fences.
In order to reduce the time spent searching the active list for matching
write_domains, we move those to a separate list whose elements are
the buffers belong to the active/flushing list with pending writes.
Orignal patch by Chris Wilson <chris@chris-wilson.co.uk>, forward-ported
by me.
In addition to better performance, this also fixes a real bug. Before
this changes, i915_gem_evict_everything didn't work as advertised. When
the gpu was actually busy and processing request, the flush and subsequent
wait would not move active and dirty buffers to the inactive list, but
just to the flushing list. Which triggered the BUG_ON at the end of this
function. With the more tight dirty buffer tracking, all currently busy and
dirty buffers get moved to the inactive list by one i915_gem_flush operation.
I've left the BUG_ON I've used to prove this in there.
References:
Bug 25911 - 2.10.0 causes kernel oops and system hangs
http://bugs.freedesktop.org/show_bug.cgi?id=25911
Bug 26101 - [i915] xf86-video-intel 2.10.0 (and git) triggers kernel oops
within seconds after login
http://bugs.freedesktop.org/show_bug.cgi?id=26101
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Tested-by: Adam Lantos <hege@playma.org>
Cc: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
Having missed the ENOMEM return via i915_gem_fault(), there are probably
other paths that I also missed. By not enabling NORETRY by default these
paths can run the shrinker and take memory from the system (but not from
our own inactive lists because our shrinker can not run whilst we hold
the struct mutex) and this may allow the system to survive a little longer
whilst our drivers consume all available memory.
References:
OOM killer unexpectedly called with kernel 2.6.32
http://bugzilla.kernel.org/show_bug.cgi?id=14933
v2: Pass gfp into page mapping.
v3: Use new read_cache_page_gfp() instead of open-coding.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Many platform support this feature, and it can provide significant
power savings when the reduced refresh rate is low. However, on some
platforms a secondary (reduced) timing is provided but not actually
supported by the hardware. This results in undesirable flicker at
runtime.
So disable the feature by default, but allow users to opt-in to the
reduced clock behavior with a new module parameter, lvds_downclock,
that can be set to 1 to enable the feature.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
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: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric 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: Andy Lutomirski <luto@mit.edu>
Tested-by: Jeff Chua <jeff.chua.linux@gmail.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
This patch adds a new execbuf ioctl, execbuf2, for use by clients that
want to control fence register allocation more finely. The buffer
passed in to the new ioctl includes a new relocation type to indicate
whether a given object needs a fence register assigned for the command
buffer in question.
Compatibility with the existing execbuf ioctl is implemented in terms
of the new code, preserving the assumption that fence registers are
required for pre-965 rendering commands.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[ickle: Remove pre-emptive clear_fence_reg()]
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
[anholt: Removed dmesg spam]
Signed-off-by: Eric Anholt <eric@anholt.net>
Instead of using the IS_I9XX etc macros that expand to a ton of
comparisons, use new struct intel_device_info to capture the
capabilities of the different chipsets. The drm_i915_private struct
will be initialized to point to the device info that correspond to
the actual device and this way, testing for a specific capability is
just a matter of checking a bit field.
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Dirk reports that nothing is displayed on LVDS when using ubuntu 9.1 after
close/reopen the LID. And I also reproduce this issue on another laptop.
After some tests and debug, it seems that it is related with that the
LVDS status is not updated in time in course of suspend/resume.
Now the LID state is used to check whether the LVDS is connected or
disconnected. And when the LID is closed, it means that the LVDS is
disconnected. When it is reopened, it means that the LVDS is connected.
At the same time on some distributions the LID event is also used to put
the system into suspend state. When the LID is closed, the system will enter
the suspend state. When the LID is reopened, the system will be resumed.
In such case when the LID is closed, user-space script will receive the LID
notification event and detect the LVDS as disconnected. Then the system will
enter the suspended state. When the LID is reopened, the system will be
resumed. As the LVDS status is not updated in course of resume, it will cause
that the LVDS connector is marked as unused and disabled. After the resume is
finished,user-space script will try to configure the display mode for LVDS.
But unfortunately as the LVDS status is not updated in time and it is still
marked as disconnected, the LVDS and its corresponding CRTC will be disabled
again in the function of drm_helper_disable_unused_functions after changing
mode for LVDS.
So we had better check and update the status of LVDS connector after receiving
the LID notication event. Then after the system is resumed from suspended
state, we can set the display mode for LVDS correctly.
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reported-by: Dirk Hohndel <hohndel@infradead.org>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
CC: stable@kernel.org
Signed-off-by: Eric Anholt <eric@anholt.net>
Rather than restoring just a few clock gating registers on resume,
just reinitialize the whole thing.
Signed-off-by: Andy Lutomirski <luto@mit.edu>
[anholt: Fixed up for RC6 support landed since the patch was written]
Signed-off-by: Eric Anholt <eric@anholt.net>
This merges the upstream Intel tree and fixes up numerous conflicts
due to patches merged into Linus tree later in -rc cycle.
Conflicts:
drivers/char/agp/intel-agp.c
drivers/gpu/drm/drm_dp_i2c_helper.c
drivers/gpu/drm/i915/i915_irq.c
drivers/gpu/drm/i915/i915_suspend.c
IGD* isn't a useful name. Replace with the codenames, as sourced from
pci.ids.
Signed-off-by: Adam Jackson <ajax@redhat.com>
[anholt: Fixed up for merge with pineview/ironlake changes]
Signed-off-by: Eric Anholt <eric@anholt.net>
Acked-by: Jakob Bornecrantz <jakob@vmware.com>
Acked-by: Thomas Hellström <thomas@shipmail.org>
Review-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse "Orange Smoothie" Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Eric 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: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
On some laptops there is no HDMI/DP. But the xrandr still reports
several disconnected HDMI/display ports. In such case the user will be
confused.
>DVI1 disconnected (normal left inverted right x axis y axis)
>DP1 disconnected (normal left inverted right x axis y axis)
>DVI2 disconnected (normal left inverted right x axis y axis)
>DP2 disconnected (normal left inverted right x axis y axis)
>DP3 disconnected (normal left inverted right x axis y axis)
This patch set is to use the child device parsed in VBT to decide whether
the HDMI/DP/LVDS/TV should be initialized.
Parse the child device from VBT.
The device class type is also added for LFP, TV, HDMI, DP output.
https://bugs.freedesktop.org/show_bug.cgi?id=22785
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
if no VBT is present, crt_ddc_bus will be left at 0, and cause us
to use that for the GPIO register offset. That's never a valid register
offset, so let the "undefined" value be 0 instead of -1.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
[anholt: clarified the commit message a bit]
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: Zhao Yakui <yakui.zhao@intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Add more display registers save/restore to fix unstable issues
during S4 testing on Ironlake. And DPLL_B_MD should not be restored
on Ironlake.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Add the support of ACPI opregion on Ironlake so that the backlight
brightness can be adjusted by using ACPI interface
>/sys/class/backlight/acpi_video0/brightness
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
Tested-by: Zhao Yakui <yakui.zhao@intel.com>
[zhenyuw: cleanups, fix typo for checking GSE irq and convert to
current irq handling logic.]
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
As long as the gpu can keep up, neither the cpu (waiting for gpu)
nore the gpu (waiting for vblank to do an overlay flip) stalls.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric 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: Daniel Vetter <daniel.vetter@ffwll.ch>
[anholt: Resolved against the MADVISE ioctl going in before this one]
Signed-off-by: Eric Anholt <eric@anholt.net>
It is identical to I85X. Use that one instead.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
[anholt: fix conflicts against the display function pointer stuff]
Signed-off-by: Eric Anholt <eric@anholt.net>
This just waits until the hw passed the current ring position with
cmd execution. This slightly changes the existing i915_wait_request
function to make uninterruptible waiting possible - no point in
returning to userspace while mucking around with the overlay, that
piece of hw is just too fragile.
Also replace a magic 0 with the symbolic constant (and kill the then
superflous comment) while I was looking at the code.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Eric 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: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel:
drm/i915: Ironlake suspend/resume support
drm/i915: kill warning in intel_find_pll_g4x_dp
drm/i915: update watermarks before enabling PLLs
drm/i915: add FIFO watermark support for G4x
drm/i915: quiet DP i2c init
drm/i915: fix panel fitting filter coefficient select for Ironlake
drm/i915: fix to setup display reference clock control on Ironlake
drm/i915: Install a fence register for fbc on g4x
drm/i915: save/restore BLC histogram control reg across suspend/resume
drm/i915: Fix FDI M/N setting according with correct color depth
drm/i915: disable powersave feature for Ironlake currently
drm/i915: Fix render reclock availability detection.
drm/i915: Save and restore the GM45 FBC regs on suspend and resume.
drm/i915: Set the LVDS_BORDER when using LVDS scaling mode
drm/i915: disable FBC for Pineview, fixing a boot hang.
In commit c1c7af6089 ("drm/i915: force
mode set at lid open time") the intel graphics driver was taught to
restore the LVDS mode on lid open.
That caused problems with interaction with the suspend/resume code,
which commonly runs at the same time (suspend is often caused by the lid
close event, while lid open is commonly a resume event), which was
worked around with in commit 06891e27a9
("drm/i915: fix suspend/resume breakage in lid notifier").
However, in the meantime the lid event code had also grown a user event
notifier (commit 06324194ee: "drm/i915:
generate a KMS uevent at lid open/close time"), and now _that_ causes
problems with suspend/resume and some versions of Xorg reacting to those
uevents by setting the mode.
So this effectively reverts that commit 06324194ee, and makes the lid
open protection logic against suspend/resume more explicit. This fixes
at least one laptop. See
http://bugzilla.kernel.org/show_bug.cgi?id=14484
for more details.
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Riccardo Magliocchetti <riccardo.magliocchetti@gmail.com>
Cc: Eric Anholt <eric@anholt.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This adds registers save/restore for Ironlake to make suspend work.
Signed-off-by: Guo, Chaohong <chaohong.guo@intel.com>
[zhenyuw: some code re-orgnization, and add more save/restore for
FDI link and transcoder registers, also fix palette register for Ironlake]
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Turns out some machines, like the ThinkPad X40 don't come back if you
don't save/restore this register.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Until we figure out the right setting for powersave features on
Ironlake, disable it for now. Also disable watermark update,
which has new registers for it on Ironlake too.
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
[anholt: Resolved against the Pineview FBC changes]
Signed-off-by: Eric Anholt <eric@anholt.net>
This hasn't fixed the regressions we were testing against, but clearly
should be required.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
According to the spec the LVDS_BORDER_ENABLE bit decides whether the border
data should be included in the active display and data sent to the panel.
Border should be used when in VGA centered (un-scaled) mode or when scaling
a 4:3 source image to a wide screen panel (typical 16:9).
So when the LVDS scaling is used, decide whether the LVDS_BORDER should be
enabled or not according to the current scaling mode.
At the same time fix the typo error in LVDS center scaling mode.
https://bugs.freedesktop.org/show_bug.cgi?id=23789
Signed-off-by: Zhao Yakui <yakui.zhao@intel.com>
tested-by: Zhao Jian <jian.zhao@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Pineview doesn't have this FBC mechanism, so this code doesn't apply.
Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
If we trigger a tracepoint for batch buffer submission, it is a reasonable
assumption that we wish to also trace the batch buffer completion. So in
order to capture the completion events, we need to enable irqs... However,
we cannot rely on the completion event to disable the irq later, so we
defer the irq disable to the retire request.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
* 'drm-intel-next' of git://git.kernel.org/pub/scm/linux/kernel/git/anholt/drm-intel: (57 commits)
drm/i915: Handle ERESTARTSYS during page fault
drm/i915: Warn before mmaping a purgeable buffer.
drm/i915: Track purged state.
drm/i915: Remove eviction debug spam
drm/i915: Immediately discard any backing storage for uneeded objects
drm/i915: Do not mis-classify clean objects as purgeable
drm/i915: Whitespace correction for madv
drm/i915: BUG_ON page refleak during unbind
drm/i915: Search harder for a reusable object
drm/i915: Clean up evict from list.
drm/i915: Add tracepoints
drm/i915: framebuffer compression for GM45+
drm/i915: split display functions by chip type
drm/i915: Skip the sanity checks if the current relocation is valid
drm/i915: Check that the relocation points to within the target
drm/i915: correct FBC update when pipe base update occurs
drm/i915: blacklist Acer AspireOne lid status
ACPI: make ACPI button funcs no-ops if not built in
drm/i915: prevent FIFO calculation overflows on 32 bits with high dotclocks
drm/i915: intel_display.c handle latency variable efficiently
...
Fix up trivial conflicts in drivers/gpu/drm/i915/{i915_dma.c|i915_drv.h}
Add support for framebuffer compression on GM45 and above. Removes
some unnecessary I915_HAS_FBC checks as well (this is now part of the
FBC display function).
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
This patch splits out several of the display functions into a separate
display function table to avoid tons of chipset specific if..else
if..else if blocks all over. There are more opportunities for this
(some noted in the structure defintition); so more cleanup patches will
follow.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
VGA arb requires DRM support for non-kms drivers, to turn on/off
irqs when disabling the mem/io regions.
VGA arb requires KMS support for GPUs where we can turn off VGA
decoding. Currently we know how to do this for intel and radeon
kms drivers, which allows them to be removed from the arbiter.
This patch comes from Fedora rawhide kernel.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Similar to the madvise() concept, the application may wish to mark some
data as volatile. That is in the event of memory pressure the kernel is
free to discard such buffers safe in the knowledge that the application
can recreate them on demand, and is simply using these as a cache.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This should help GEM handle memory pressure sitatuions more gracefully.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is no need to store the gtt_alignment as it is either explicitly
set according to the hardware requirements (e.g. scanout) or the
minimum alignment is computed on demand.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Due to a bogus FBC support check and failing to check for FBC support
in the right places, mode setting on non-mobile platforms could fail
and hang in the FBC disable routine. Fix it up.
This fix highlights the need for cleanups in this area (function
pointers and better feature support checks). Patches for that to
follow.
Tested-by: Kenny Graunke <kenny@whitecape.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We now unconditionally restore the mode at lid open time since some
platforms turn off the panel, pipes or other display elements when the
lid is closed. There's a problem with doing this at resume time
however.
At resume time, we'll get a lid event, but restoring the mode at that
time may not be safe (e.g. if we get the lid event before global state
has been restored), so check the suspended state and make sure our
restore is locked against other mode updates.
Tested-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
There is a very real possibility that multiple CPUs will notice that the
GPU is wedged. This introduces all sorts of potential race conditions.
Make the wedged flag atomic to mitigate this risk.
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
This patch puts in place the machinery to attempt to reset the GPU. This
will be used when attempting to recover from a GPU hang.
Signed-off-by: Owain G. Ainsworth <oga@openbsd.org>
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We set a periodic timer to check on the GPU, resetting it every time a
batch is completed. If the timer elapses, we check acthd. If acthd
hasn't changed in two timer periods, we assume the chip is wedged.
This is implemented in such a way that it leaves the option open to
employ adaptive timer intervals in the future. One could wait until
several timer periods have elapsed before declaring the chip dead. If
the chip comes back after several periods but before the "dead"
threshold, the timer interval or dead threshold could be raised.
It is important to note that while checking for active requests, we need
to account for the fact that requests are removed from the list (i.e.
retired) in a deferred work queue handler. This means that merely
checking for an empty request_list is insufficient; the list could be
non-empty yet the GPU still idle, causing the hangcheck timer to
incorrectly mark the GPU as wedged (it took me a while to figure that
out---sigh...)
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
We'll need it in i915_irq.c for checking whether there are outstanding
requests. Also, the function really ought to return a bool, not an int.
Signed-off-by: Ben Gamari <bgamari.foss@gmail.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
