Tvrtko Ursulin [Fri, 26 Feb 2016 16:58:32 +0000 (16:58 +0000)]
drm/i915: Execlists small cleanups and micro-optimisations
Assorted changes in the areas of code cleanup, reduction of
invariant conditional in the interrupt handler and lock
contention and MMIO access optimisation.
* Remove needless initialization.
* Improve cache locality by reorganizing code and/or using
branch hints to keep unexpected or error conditions out
of line.
* Favor busy submit path vs. empty queue.
* Less branching in hot-paths.
v2:
* Avoid mmio reads when possible. (Chris Wilson)
* Use natural integer size for csb indices.
* Remove useless return value from execlists_update_context.
* Extract 32-bit ppgtt PDPs update so it is out of line and
shared with two callers.
* Grab forcewake across all mmio operations to ease the
load on uncore lock and use chepear mmio ops.
v3:
* Removed some more pointless u8 data types.
* Removed unused return from execlists_context_queue.
* Commit message updates.
v4:
* Unclumsify the unqueue if statement. (Chris Wilson)
* Hide forcewake from the queuing function. (Chris Wilson)
Version 3 now makes the irq handling code path ~20% smaller on
48-bit PPGTT hardware, and a little bit less elsewhere. Hot
paths are mostly in-line now and hammering on the uncore
spinlock is greatly reduced together with mmio traffic to an
extent.
Benchmarking with "gem_latency -n 100" (keep submitting
batches with 100 nop instruction) shows approximately 4% higher
throughput, 2% less CPU time and 22% smaller latencies. This was
on a big-core while small-cores could benefit even more.
Most likely reason for the improvements are the MMIO
optimization and uncore lock traffic reduction.
One odd result is with "gem_latency -n 0" (dispatching empty
batches) which shows 5% more throughput, 8% less CPU time,
25% better producer and consumer latencies, but 15% higher
dispatch latency which is yet unexplained.
In addition to calculating final watermarks, let's also pre-calculate a
set of intermediate watermark values at atomic check time. These
intermediate watermarks are a combination of the watermarks for the old
state and the new state; they should satisfy the requirements of both
states which means they can be programmed immediately when we commit the
atomic state (without waiting for a vblank). Once the vblank does
happen, we can then re-program watermarks to the more optimal final
value.
v2: Significant rebasing/rewriting.
v3:
- Move 'need_postvbl_update' flag to CRTC state (Daniel)
- Don't forget to check intermediate watermark values for validity
(Maarten)
- Don't due async watermark optimization; just do it at the end of the
atomic transaction, after waiting for vblanks. We do want it to be
async eventually, but adding that now will cause more trouble for
Maarten's in-progress work. (Maarten)
- Don't allocate space in crtc_state for intermediate watermarks on
platforms that don't need it (gen9+).
- Move WaCxSRDisabledForSpriteScaling:ivb into intel_begin_crtc_commit
now that ilk_update_wm is gone.
v4:
- Add a wm_mutex to cover updates to intel_crtc->active and the
need_postvbl_update flag. Since we don't have async yet it isn't
terribly important yet, but might as well add it now.
- Change interface to program watermarks. Platforms will now expose
.initial_watermarks() and .optimize_watermarks() functions to do
watermark programming. These should lock wm_mutex, copy the
appropriate state values into intel_crtc->active, and then call
the internal program watermarks function.
v5:
- Skip intermediate watermark calculation/check during initial hardware
readout since we don't trust the existing HW values (and don't have
valid values of our own yet).
- Don't try to call .optimize_watermarks() on platforms that don't have
atomic watermarks yet. (Maarten)
v6:
- Rebase
v7:
- Further rebase
v8:
- A few minor indentation and line length fixes
v9:
- Yet another rebase since Maarten's patches reworked a bunch of the
code (wm_pre, wm_post, etc.) that this was previously based on.
v10:
- Move wm_mutex to dev_priv to protect against racing commits against
disjoint CRTC sets. (Maarten)
- Drop unnecessary clearing of cstate->wm.need_postvbl_update (Maarten)
v11:
- Now that we've moved to atomic watermark updates, make sure we call
the proper function to program watermarks in
{ironlake,haswell}_crtc_enable(); the failure to do so on the
previous patch iteration led to us not actually programming the
watermarks before turning on the CRTC, which was the cause of the
underruns that the CI system was seeing.
- Fix inverted logic for determining when to optimize watermarks. We
were needlessly optimizing when the intermediate/optimal values were
the same (harmless), but not actually optimizing when they differed
(also harmless, but wasteful from a power/bandwidth perspective).
Chris Wilson [Fri, 26 Feb 2016 11:22:31 +0000 (11:22 +0000)]
drm/i915: Execlists cannot pin a context without the object
Given that the intel_lr_context_pin cannot succeed without the object,
we cannot reach intel_lr_context_unpin() without first allocating that
object - so we can remove the redundant test.
Chris Wilson [Fri, 26 Feb 2016 11:03:19 +0000 (11:03 +0000)]
drm/i915: Rename vma->*_list to *_link for consistency
Elsewhere we have adopted the convention of using '_link' to denote
elements in the list (and '_list' for the actual list_head itself), and
that the name should indicate which list the link belongs to (and
preferrably not just where the link is being stored).
s/vma_link/obj_link/ (we iterate over obj->vma_list)
s/mm_list/vm_link/ (we iterate over vm->[in]active_list)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
drm/i915: Add helper to get a display power ref if it was already enabled
left the rpm wakelock assertions unbalanced if CONFIG_PM was disabled as
intel_runtime_pm_get_if_in_use() would return true without incrementing
the local bookkeeping required for the assertions.
drm/i915: Kill off intel_crtc->atomic.wait_vblank, v6.
Currently we perform our own wait in post_plane_update,
but the atomic core performs another one in wait_for_vblanks.
This means that 2 vblanks are done when a fb is changed,
which is a bit overkill.
Merge them by creating a helper function that takes a crtc mask
for the planes to wait on.
The broadwell vblank workaround may look gone entirely but this is
not the case. pipe_config->wm_changed is set to true
when any plane is turned on, which forces a vblank wait.
Changes since v1:
- Removing the double vblank wait on broadwell moved to its own commit.
Changes since v2:
- Move out POWER_DOMAIN_MODESET handling to its own commit.
Changes since v3:
- Do not wait for vblank on legacy cursor updates. (Ville)
- Move broadwell vblank workaround comment to page_flip_finished. (Ville)
Changes since v4:
- Compile fix, legacy_cursor_flip -> *_update.
Changes since v5:
- Kill brackets.
- Add WARN_ON when wait_for_vblanks fails.
- Remove extra newlines.
- Split the checks whether vblank is needed to a separate function,
with comments why a vblank is needed.
drm/i915: Pass crtc state to modeset_get_crtc_power_domains.
Use our newly created encoder_mask to iterate over the encoders.
This makes it possible to get the crtc power domains from the
crtc_state at any time, without any locks or having to look at
the legacy state.
Ville Syrjälä [Fri, 19 Feb 2016 18:47:30 +0000 (20:47 +0200)]
drm/i915: Make sure pipe interrupts are processed before turning off power well on BDW+
Starting from BDW the DE_PIPE interrupts for pipe B and C belong to the
relevant display power well. So we should make sure we've finished
processing them before turning off the power well.
The pipe interrupts shouldn't really happen at this point anymore since
we've already shut down the planes/pipes/whatnot, but being a bit
paranoid shouldn't hurt.
Ville Syrjälä [Fri, 19 Feb 2016 16:41:52 +0000 (18:41 +0200)]
drm/i915: synchronize_irq() before turning off disp2d power well on VLV/CHV
After we've told the irq code we don't want to handle display irqs
anymore, we must make sure any display irq handling already
kicked off has finished before we actually turn off the power well.
I wouldn't expect PIPESTAT based interrupts to occur anymore since
vblanks/page flips/gmbus/etc should all be quiescent at this point.
But at least hotplug interrupts could still occur. Hotplug
interrupts may also kick off the workqueue based hotplug processing,
but that code should take the required power domain references
itself, so there shouldn't be any need to synchronize with the
hotplug processing from the power well code.
Ville Syrjälä [Thu, 18 Feb 2016 19:54:26 +0000 (21:54 +0200)]
drm/i915: Skip PIPESTAT reads from irq handler on VLV/CHV when power well is down
PIPESTAT registers live in the display power well on VLV/CHV, so we
shouldn't access them when things are powered down. Let's check
whether the display interrupts are on or off before accessing the
PIPESTAT registers.
Another option would be to read the PIPESTAT registers only when
the IIR register indicates that there's a pending pipe event. But
that would mean we might miss even more underrun reports than we
do now, because the underrun status bit lives in PIPESTAT but doesn't
actually generate an interrupt.
Mika Kuoppala [Thu, 18 Feb 2016 15:58:09 +0000 (17:58 +0200)]
drm/i915/gen9: Verify and enforce dc6 state writes
It has been observed that sometimes disabling the dc6 fails
and dc6 state pops back up, brief moment after disabling. This
has to be dmc save/restore timing issue or other bug in the
way dc states are handled.
Try to work around this issue as we don't have firmware fix
yet available. Verify that the value we wrote for the dmc sticks,
and also enforce it by rewriting it, if it didn't.
v2: Zero rereads on rewrite for extra paranoia (Imre)
Testcase: kms_flip/basic-flip-vs-dpms
References: https://bugs.freedesktop.org/show_bug.cgi?id=93768 Cc: Patrik Jakobsson <patrik.jakobsson@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Imre Deak <imre.deak@intel.com> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1455811089-27884-1-git-send-email-mika.kuoppala@intel.com
Paulo Zanoni [Tue, 16 Feb 2016 20:47:21 +0000 (18:47 -0200)]
drm/i915/fbc: enable FBC by default on HSW and BDW
These platforms should be fine now.
FBC can allow very significant power savings for screen-on idle
systems, but it is worth mentioning that a lot of people won't get
significant power savings by enabling this feature because they may
have something else preventing the system from getting into the
deepest sleep states. Examples may include a hungry wifi device or a
max_performance SATA link power management policy. You can check your
PC state residencies on the powertop "Idle stats" tab. I recommend
trying to run "sudo powertop --auto-tune" and then seeing if the
residencies improve.
Oh, and in case you - the person reading this commit message - found
this commit through git bisect, please do the following:
- Check your dmesg and see if there are error messages mentioning
underruns around the time your problem started happening.
- Download intel-gpu-tools, compile it, and run:
$ sudo ./tests/kms_frontbuffer_tracking --run-subtest '*fbc-*' 2>&1 | tee fbc.txt
Then send us the fbc.txt file, especially if you get a failure.
This will really maximize your chances of getting the bug fixed
quickly.
- Try to find a reliable way to reproduce the problem, and tell us.
- Boot with drm.debug=0xe, reproduce the problem, then send us the
dmesg file.
Jani Nikula [Tue, 16 Feb 2016 11:18:12 +0000 (13:18 +0200)]
drm/i915: drop write perm from module params which don't support changing
We've given write permissions to dynamically change some module
parameters through /sys/module/i915/parameters although they only
support setting on module load. Fix the permissions.
Deepak M [Thu, 11 Feb 2016 15:03:27 +0000 (20:33 +0530)]
drm/i915/dsi: Using the bpp value wrt the pixel format
The bpp value which is used while calulating the txbyteclkhs values
should be wrt the pixel format value. Currently bpp is coming
from pipe config to calculate txbyteclkhs. Fix it in this patch.
V2: dsi_pixel_format_bpp is used to retrieve the bpp from pixel_format
[Review: Jani]
Imre Deak [Wed, 17 Feb 2016 14:31:29 +0000 (16:31 +0200)]
drm/i915/skl: Ensure HW is powered during DDB HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Ville Syrjälä [Wed, 17 Feb 2016 19:28:45 +0000 (21:28 +0200)]
drm/i915: Extract intel_encoder_has_connectors()
We have an open coded loop which tries to see if the encoder has any
connectors linked to it. Let's extract that to a helper similar to
intel_crtc_has_encoders().
drm/i915: Use atomic state to obtain load detection crtc, v3.
Instead of restoring dpms and a flag for whether a temp fb is allocated duplicate
an atomic state before the new state is committed, and commit it the old state
in intel_release_load_detect_pipe.
Changes since v1:
- Use a real atomic state. (Ville)
Changes since v2:
- Do not preserve shared_dpll any more, no need to do so. (Ville)
Rodrigo Vivi [Fri, 12 Feb 2016 12:08:13 +0000 (04:08 -0800)]
drm/i915: Enable PSR by default on Haswell and Broadwell.
With a reliable frontbuffer tracking and all instability corner cases
on Haswell and Broadwell solved let's re-enabled PSR by default on
these platforms.
In case a new issue is found and PSR is the main suspect, please check
if i915.enable_psr=0 really makes your problem go away. If this is the case
PSR is the culprit so after that please check if i915.enable_psr=2
or i915.enable_psr=3 solves your issue and please let us know.
There are many panels out there and not all implementations apparently
work as we would expect.
In case you needed to force it on standby or disabled or in case of any
PSR related bug please report it at bugs.freedesktop.org.
In a bugzilla entry for PSR is desirable:
- dmesg (drm.debug=0xe)
- output of /sys/kernel/debug/dri/0/i915_edp_psr_status
- Platform information. Vendor, model, id, pci id.
- Graphical environment: Gnome, KDE, openbox, etc...
- Details how to reproduce.
- Also good if you could run PSR test cases of Intel-gpu-tools
- Please mention if forcing main link standby or main link off helps you.
There are Intel-gpu-tools test cases that can be helpful to
determine if PSR is working as expected:
kms_psr_sink_crc and kms_psr_frontbuffer_tracking.
Rodrigo Vivi [Fri, 12 Feb 2016 12:08:12 +0000 (04:08 -0800)]
drm/i915: Enable PSR by default on Valleyview and Cherryview.
With a reliable frontbuffer tracking and all instability corner cases
solved for this platform let's re-enabled PSR by default.
In case a new issue is found and PSR is the main suspect, please check
if i915.enable_psr=0 really makes your problem go away,
please report it at bugs.freedesktop.org.
In a bugzilla entry for PSR is desirable:
- dmesg (drm.debug=0xe)
- output of /sys/kernel/debug/dri/0/i915_edp_psr_status
- Platform information. Vendor, model, id, pci id.
- Graphical environment: Gnome, KDE, openbox, etc...
- Details how to reproduce.
- Also good if you could run PSR test cases of Intel-gpu-tools
- Please mention if forcing main link standby or main link off helps you.
There are Intel-gpu-tools test cases that can be helpful to
determine if PSR is working as expected:
kms_psr_sink_crc and kms_psr_frontbuffer_tracking.
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rodrigo Vivi [Fri, 12 Feb 2016 12:08:11 +0000 (04:08 -0800)]
drm/i915: Change i915.enable_psr parameter to use per platform default.
This will give us flexibility to enable PSR by default independently so
issues and corner cases in one platform won't affect others were we have
it working properly.
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Imre Deak [Fri, 12 Feb 2016 16:55:21 +0000 (18:55 +0200)]
drm/i915/lvds: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:20 +0000 (18:55 +0200)]
drm/i915/hdmi: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:19 +0000 (18:55 +0200)]
drm/i915/dsi: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:18 +0000 (18:55 +0200)]
drm/i915/dp: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:17 +0000 (18:55 +0200)]
drm/i915: Ensure the HW is powered when accessing the CRC HW block
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
While at it also add the missing reference around the HW access in
i915_interrupt_info().
v2:
- update the commit message mentioning that this also fixes the
HW access in the interrupt info debugfs entry (Daniel)
Imre Deak [Fri, 12 Feb 2016 16:55:16 +0000 (18:55 +0200)]
drm/i915/ddi: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:15 +0000 (18:55 +0200)]
drm/i915/crt: Ensure the HW is powered during HW state readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:14 +0000 (18:55 +0200)]
drm/i915: Ensure the HW is powered during HW access in assert_pipe
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:13 +0000 (18:55 +0200)]
drm/i915: Ensure the HW is powered when disabling VGA
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:12 +0000 (18:55 +0200)]
drm/i915/ibx: Ensure the HW is powered during PLL HW readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Fri, 12 Feb 2016 16:55:11 +0000 (18:55 +0200)]
drm/i915: Ensure the HW is powered during display pipe HW readout
The assumption when adding the intel_display_power_is_enabled() checks
was that if it returns success the power can't be turned off afterwards
during the HW access, which is guaranteed by modeset locks. This isn't
always true, so make sure we hold a dedicated reference for the time of
the access.
Imre Deak [Wed, 17 Feb 2016 12:17:42 +0000 (14:17 +0200)]
drm/i915: Add helper to get a display power ref if it was already enabled
We have many places in the code where we check if a given display power
domain is enabled and if so access registers backed by this power
domain. We assumed that some modeset lock will prevent the power
reference from vanishing in the middle of the HW access, but this
assumption doesn't always hold. In such cases we get either the wakeref
not held, or an unclaimed register access error message. To fix this in
a future-proof way that's independent of other locks wrap any such
access with a get_ref_if_enabled()/put_ref() pair.
Kudos to Ville and Joonas for the ideas of this new interface.
v2:
- init the power_domains ptr when declaring it everywhere (Joonas)
v3:
- don't report the device to be powered if runtime PM is disabled
CC: Mika Kuoppala <mika.kuoppala@intel.com> CC: Chris Wilson <chris@chris-wilson.co.uk> CC: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> CC: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1455711462-7442-1-git-send-email-imre.deak@intel.com
Chris Wilson [Mon, 11 Jan 2016 09:16:14 +0000 (09:16 +0000)]
drm/i915: Add an optional selection from i915 of CONFIG_MMU_NOTIFIER
userptr requires mmu-notifier for full unprivileged support. Most
systems have mmu-notifier support already enabled as a requirement for
virtualisation support, but we should make the option for i915 to take
advantage of mmu-notifiers explicit (and enable by default so that
regular userspace can take advantage of passing client memory to the
GPU.)
Ben Widawsky [Tue, 9 Feb 2016 19:44:12 +0000 (11:44 -0800)]
drm/i915: Check for get_pages instead of shmem (filp)
This behavior of checking for a shmem backed GEM object was introduced here:
commit 4c914c0c7c787b8f730128a8cdcca9c50b0784ab
Author: Brad Volkin <bradley.d.volkin@intel.com>
Date: Tue Feb 18 10:15:45 2014 -0800
drm/i915: Refactor shmem pread setup
It is possible for an object to not be a shmem backed GEM object (for example
userptr objects). An example of how we hit this failure can be found through
copy_batch() in the command parser because we allocate a userptr object for the
batch which contains privileged instructions. Userptr calls
drm_gem_private_object_init() which explicitly sets the filp to none.
NOTE: I manually retyped this from a test machine. So I haven't even compiled
this exact patch.
v2: Use same logic as from a2a4f916c2f (Kristian, Dave Gordon)
Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Kristian Høgsberg <krh@bitplanet.net> Cc: Dave Gordon <david.s.gordon@intel.com> Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com> Tested-by: Jordan Justen <jordan.l.justen@intel.com> (v1) Reviewed-by: Jordan Justen <jordan.l.justen@intel.com> (v1) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455047053-2644-1-git-send-email-benjamin.widawsky@intel.com
Ville Syrjälä [Thu, 11 Feb 2016 14:44:28 +0000 (16:44 +0200)]
drm/i915: Add missing 'else' to intel_digital_port_connected()
intel_digital_port_connected() lacks one 'else'. There's no
actual harm in not having it since each branch has an unconditional
return, so it can't accidentally end up in taking two branches instead
of just the one. But let's be consistent and add the 'else' anyway.
Ville Syrjälä [Wed, 10 Feb 2016 17:59:05 +0000 (19:59 +0200)]
drm/i915: Fix hpd live status bits for g4x
Looks like g4x hpd live status bits actually agree with the spec. At
least they do on the machine I have, and apparently on Nick Bowler's
g4x as well.
So gm45 may be the only platform where they don't agree. At least
that seems to be the case based on the (somewhat incomplete)
logs/dumps in [1], and Daniel has also tested this on his gm45
sometime in the past.
So let's change the bits to match the spec on g4x. That actually makes
the g4x bits identical to vlv/chv so we can just share the code
between those platforms, leaving gm45 as the special case.
Tvrtko Ursulin [Thu, 11 Feb 2016 10:27:32 +0000 (10:27 +0000)]
drm/i915/ilk: Move register read under spinlock
Code does read-modify-write but the read was outside the lock.
It is fine since the caller holds struct mutex, but if we
correct this we open up the opportunity for decreasing the
mutex duration time since the call to ironlake_enable_drps
does not need it any longer since it is covered by the
mchdev_lock lock.
Tvrtko Ursulin [Thu, 11 Feb 2016 10:27:30 +0000 (10:27 +0000)]
drm/i915: Fix struct mutex vs. RPS lock inversion
RPS lock must be taken before the struct_mutex to avoid
locking inversion. So stop grabbing it for the whole
powersave initialization and instead only take it during
the sections which need it.
Also, struct_mutex is not needed any more since dedicated
RPS lock was added in:
drm/i915: protect RPS/RC6 related accesses (including PCU) with a new mutex
Based on prototype patch by Chris Wilson and a subsequent
mailing list discussion involving Ville, Imre, Chris and
Daniel.
v2: More details in the commit.
v3: Use drm_gem_object_unreference_unlocked. (Chris Wilson)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Tvrtko Ursulin [Thu, 11 Feb 2016 10:27:29 +0000 (10:27 +0000)]
drm/i915: GEM operations need to be done under the big lock
VMA creation and GEM list management need the big lock.
v2:
Mutex unlock ended on the wrong path somehow. (0-day, Julia Lawall)
Not to mention drm_gem_object_unreference was there in existing
code with no mutex held.
v3:
Some callers of i915_gem_object_create_stolen_for_preallocated
already hold the lock so move the mutex into the other caller
as well.
v4:
Changed to lockdep_assert_held. (Chris Wilson)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
drm/i915: Clear shared dpll based on old state, v2.
Atomic resume was preserving the dpll state because it was required
for clearing pll state correctly. If we look at the old_crtc_state
for pll to clear this is not needed and the hack can be removed.
Changes since v1:
- Rename dpll variable to old_dpll. (Ville)
Chris retracted his reviewed-by (which I failed to notice) and somehow
it blows up (I did it again!) as reported by Mika with the below
backtrace on module reload:
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Dave Gordon <david.s.gordon@intel.com> Cc: Nick Hoath <nicholas.hoath@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Abhay Kumar [Sat, 23 Jan 2016 01:39:04 +0000 (17:39 -0800)]
drm/i915: edp resume/On time optimization.
Make resume/on codepath not to wait for panel_power_cycle_delay(t11_t12)
if this time is already spent in suspend/poweron time.
v2: Use CLOCK_BOOTTIME and remove jiffies for panel power cycle
delay calculation(Ville).
v3: Addressed below comments
1. Tracking time from where last powercycle is initiated.
2. Used ktime_get_bootime() wrapper for boottime clock.
3. Used ktime_ms_delta() to get time difference.
v4: Updated v3 change log in detail.
v5: Removed static from panel_power_on_time(Stéphane).
Arnd Bergmann [Thu, 11 Feb 2016 09:10:00 +0000 (10:10 +0100)]
drm/msm: remove unused variable
After the drm_device_is_unplugged() was removed, the 'dev' variable is now
unused, and we get a warning for that:
drivers/gpu/drm/msm/msm_fbdev.c: In function 'msm_fbdev_mmap':
drivers/gpu/drm/msm/msm_fbdev.c:65:21: error: unused variable 'dev' [-Werror=unused-variable]
With this there's no longer the problem that 2 drivers (fake agp
driver and the drm/i915 driver) fight over the same piece, which fixes
apparent dma leaks detected by CONFIG_DMA_API_DEBUG.
Note that the leak isn't real since intel-gtt refcounts and will tear
down eventually. But the debug code assumes that when the i915 driver
unbinds from the pci device everything should be gone. Which isn't the
case if we have intel-agp enabled - userspace might need it. But by
ditching this intel-gtt setup and teardown is completely tied to the
livetime of the "real" driver.
While at it untangle the init ordering a bit - the fake agp wouldn't
be initialized correctly if i915.ko loads first. Which isn't a problem
since when i915 loads in kms mode you won't need the fake agp support
needed by the ums driver ...
Mika Kahola [Tue, 2 Feb 2016 13:16:43 +0000 (15:16 +0200)]
drm/i915: TV pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
Mika Kahola [Tue, 2 Feb 2016 13:16:42 +0000 (15:16 +0200)]
drm/i915: CRT pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
Mika Kahola [Tue, 2 Feb 2016 13:16:41 +0000 (15:16 +0200)]
drm/i915: SDVO pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
Mika Kahola [Tue, 2 Feb 2016 13:16:40 +0000 (15:16 +0200)]
drm/i915: DisplayPort-MST pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
Mika Kahola [Tue, 2 Feb 2016 13:16:39 +0000 (15:16 +0200)]
drm/i915: HDMI pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
This patch applies to HDMI.
V2:
- removed computation for max dot clock
V3:
- cleanup by removing unnecessary lines
V4:
- max_pixclk variable renamed as max_dotclk
- check for stereo mode added
Mika Kahola [Tue, 2 Feb 2016 13:16:38 +0000 (15:16 +0200)]
drm/i915: DisplayPort pixel clock check
It is possible the we request to have a mode that has
higher pixel clock than our HW can support. This patch
checks if requested pixel clock is lower than the one
supported by the HW. The requested mode is discarded
if we cannot support the requested pixel clock.
Lukas Wunner [Wed, 10 Feb 2016 13:17:41 +0000 (14:17 +0100)]
apple-gmux: Fix build breakage if !CONFIG_ACPI
The DRM drivers i915, nouveau and radeon may be compiled with
CONFIG_ACPI not set, in which case acpi_dev_present() is undefined.
Add a no-op stub for apple_gmux_present() which is used if
CONFIG_APPLE_GMUX is not enabled to avoid build breakage.
(CONFIG_APPLE_GMUX depends on CONFIG_ACPI.)
drm/i915: check that rpm ref is held when accessing ringbuf in stolen mem
While running some tests on the scheduler patches with rpm enabled I
came across a corruption in the ringbuffer, which was root-caused to
the GPU being suspended while commands were being emitted to the
ringbuffer. The access to memory was failing because the GPU needs to
be awake when accessing stolen memory (where my ringbuffer was located).
Since we have this constraint it looks like a sensible idea to check
that we hold a refcount when we access the rungbuffer.
v2: move the check from ring_begin to ringbuffer iomap time (Chris)
v3: update comment (Chris)
Daniel Vetter [Wed, 27 Jan 2016 13:37:59 +0000 (14:37 +0100)]
drm/i915: Stop depending upon CONFIG_AGP_INTEL
The AGP_INTEL driver provides an interface for very old userspace to
control the GART (though the GART itself was only ever emulated on Intel
systems). The pci bridge discovery code is also used by the i915.ko
driver to set up the GTT on old systems, but it does not require the
old userspace interface. When i915.ko selects the old interface, it
binds another user to the core GTT routines, and in particular creates a
second reference to the scratch pages allocated. This hinders resource
leak debugging for when we unload i915.ko as we want to assert that all
DMA pages have been released, but we appear to leak because of the
secondary interface which persists after i915.ko unloads.
All i915.ko users do not require the old /dev/agpgart interface so stop
selecting it and simplify our debugging by dropping the historical
baggage.
Note that by selecting AGP=n it was already possible to unselect
AGP_INTEL. But since we've dropped support for any of the AGP stuff
long ago there's really no point for this any more.
Also note that we still need INTEL_GTT, which is the underlying,
shared, driver for the graphics GART on gen1-5.
v2: Entirely new commit message (Chris, Ville).
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1453901881-26425-2-git-send-email-daniel.vetter@ffwll.ch
Daniel Vetter [Wed, 27 Jan 2016 13:37:58 +0000 (14:37 +0100)]
agp/intel-gtt: Don't leak the scratch page
Recently discovered by enabling CONFIG_DMA_API_DEBUG in our CI. By the
looks of it broken since forever.
v2: Don't forget to set the scratch page back to wb (Chris). Reuse
intel_gtt_teardown_scratch_page for that (and fix it up to treat
needs_dmar y/n correctly).
Nick Hoath [Fri, 22 Jan 2016 23:10:06 +0000 (23:10 +0000)]
drm/i915: fix context/engine cleanup order
Swap the order of context & engine cleanup, so that contexts are cleaned
up first, and *then* engines. This is a more sensible order anyway, but
in particular has become necessary since the 'intel_ring_initialized()
must be simple and inline' patch, which now uses ring->dev as an
'initialised' flag, so it can now be NULL after engine teardown. This
in turn can cause a problem in the context code, which (used to) check
the ring->dev->struct_mutex -- causing a fault if ring->dev was NULL.
Also rename the cleanup function to reflect what it actually does
(cleanup engines, not a ringbuffer), and fix an annoying whitespace issue.
v2: Also make the fix in i915_load_modeset_init, not just in
i915_driver_unload (Chris Wilson)
v3: Had extra stuff in it.
v4: Reverted extra stuff (so we're back to v2).
Rebased and updated commentary above (Dave Gordon).
Signed-off-by: Nick Hoath <nicholas.hoath@intel.com> Signed-off-by: David Gordon <david.s.gordon@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Dave Gordon <david.s.gordon@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1453504211-7982-2-git-send-email-david.s.gordon@intel.com
Gabriel Feceoru [Fri, 22 Jan 2016 11:28:45 +0000 (13:28 +0200)]
drm/i915: Handle PipeC fused off on IVB/HSW/BDW
Some Gen7/8 production parts may have the Display Pipe C fused off.
In this case, the display hardware will prevent the enable bit in
PIPE_CONF register (for Pipe C) from being set to 1.
Fixed by adjusting pipe_count to reflect this.
v2: Rename HSW_PIPE_C_DISABLE to IVB_PIPE_C_DISABLE as it already exists
on ivybridge (Ville)
v3: Remove unnecessary MMIO read, correct the description (Damien)
v4: Be more specific in description (Patrick)
Lyude [Thu, 4 Feb 2016 15:43:21 +0000 (10:43 -0500)]
drm/i915/skl: Fix typo in DPLL_CFGCR1 definition
We accidentally point both cfgcr registers for the second shared DPLL to
the same location in i915_reg.h. This results in a lot of hw pipe state
mismatches whenever we try to do a modeset that requires allocating the
DPLL to a CRTC:
[drm:intel_pipe_config_compare [i915]] *ERROR* mismatch in dpll_hw_state.cfgcr1 (expected 0x80000168, found 0x000004a5)
[drm:intel_pipe_config_compare [i915]] *ERROR* mismatch in base.adjusted_mode.crtc_clock (expected 108000, found 49500)
[drm:intel_pipe_config_compare [i915]] *ERROR* mismatch in port_clock (expected 108000, found 49500)
This usually ends up causing blank monitors, since the DPLL never can
get set to the right clock.
drm/gma500: fix error path in gma_intel_setup_gmbus()
The current code fails to call i2c_del_adapter on
dev_prev->gmbus[0].adapter, and if the for loop above failed already
at i==0, all hell breaks loose when we do the loop body for
i = -1,-2,...
LABBE Corentin [Thu, 4 Feb 2016 14:03:52 +0000 (15:03 +0100)]
drm: modes: add missing [drm] to message printing
The warning message in drm_mode_parse_command_line_for_connector miss
the [drm] at beginning.
This patch add it and take the opportunity to convert
printk(KERN_WARNING to pr_warn()
projects / linux-beck.git / log