Imre Deak [Thu, 7 Mar 2019 10:32:35 +0000 (12:32 +0200)]
drm/i915/icl: Prevent incorrect DBuf enabling
Pretend that we have only 1 DBuf slice and that 1 slice is always
enabled, until we have a proper way for on-demand toggling of the second
slice. Currently we'll try to incorrectly enable DBuf even when all
pipes are disabled and we are already runtime suspended (as the computed
number of DBuf slices will be 1 in that case).
This also means we'll leave the second slice enabled redundantly (except
when suspended), but that's an acceptable tradeoff until we have a
proper solution.
Ville Syrjälä [Wed, 6 Mar 2019 20:35:44 +0000 (22:35 +0200)]
drm/i915: Fix DRAM size reporting for BXT
The BXT DUNIT register tells us the size of each DRAM device
in Gb. We want to report the size of the whole DIMM in GB, so
that it matches how we report it for non-LP platforms.
v2: Deobfuscate the math (Chris)
s/GB/GBIT/ in the register bit definitions (Jani)
Ville Syrjälä [Wed, 6 Mar 2019 20:35:42 +0000 (22:35 +0200)]
drm/i915: Polish skl_is_16gb_dimm()
Pass the dimm struct to skl_is_16gb_dimm() rather than passing each
value separately. And let's replace the hardcoded set of values with
some simple arithmetic.
Also fix the byte vs. bit inconsistency in the debug message,
and polish the wording otherwise as well.
Now with the watermarks fixes merged, Icelake is stable enough to
have the alpha support protection flag removed.
We have a few ICL machines in our CI and it is mostly green with
failures in tests that will not impact future linux installations.
Also there is no warnings, errors, flickering or any visual defects
while doing ordinary tasks like browsing and editing documents in a
dual monitor setup.
As a reminder i915.alpha_support was created to protect
future linux installation's iso images that might contain a
kernel from the enabling time of the new platform. Without this
protection most of linux installation was recommending
nomodeset option during installation that was getting stick
there after installation.
Specifically, alpha support says nothing about the development
state of the hardware, and everything about the state of the
driver in a kernel release.
This is semantically no different from the old
preliminary_hw_support flag, but the old one was all too often
interpreted as (preliminary hw) support instead of the intended
(preliminary) hw support, and it was misleading for everyone.
Hence the rename.
Reference: https://intel-gfx-ci.01.org/tree/drm-tip/fi-icl-y.html
Reference: https://intel-gfx-ci.01.org/tree/drm-tip/shard-iclb.html Cc: James Ausmus <james.ausmus@intel.com> Cc: Jani Saarinen <jani.saarinen@intel.com> Cc: Paulo Zanoni <paulo.r.zanoni@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: José Roberto de Souza <jose.souza@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190305221153.359-1-jose.souza@intel.com
Chris Wilson [Thu, 7 Mar 2019 10:45:30 +0000 (10:45 +0000)]
drm/i915: Force GPU idle on suspend
To facilitate the next patch to allow preemptible kernels not to incur
the wrath of hangcheck, we need to ensure that we can still suspend and
shutdown. That is we will not be able to rely on hangcheck to terminate
a blocking kernel and instead must manually do so ourselves. The
advantage is that we can apply more pressure!
As we now perform a GPU reset to clean up any residual kernels, we leave
the GPU in an unknown state and in particular can not talk to the GuC
before we reinitialise it following resume. For example, we no longer
need to tell the GuC to suspend itself, as it is already reset.
Chris Wilson [Thu, 7 Mar 2019 10:45:29 +0000 (10:45 +0000)]
drm/i915: Make I915_GEM_IDLE_TIMEOUT into a macro
Currently we use HZ/5 for detecting a dead gpu on startup, and we will
wish to reuse this value for detecting a dead gpu on suspend, so convert
it into a macro for later convenience.
At some point people have started to assume that
pipe_offsets[] & co. are only populated for pipes and whatnot
that actually exist. That is in fact not currently true, but
we can easily make it so.
Ville Syrjälä [Tue, 5 Mar 2019 19:24:00 +0000 (21:24 +0200)]
drm/i915: Simplify i830 DVO 2x clock handling
Let's just always enable the DVO 2x clock on i830. This way we don't
have to track if DVO is being used or not. The spec does suggest we
should disable the clock when it isn't needed, but this does appear
to work just fine.
This removes another crtc->config usage.
v2: Split the DPLL enable sequence change to a separate patch
Ville Syrjälä [Tue, 5 Mar 2019 19:23:59 +0000 (21:23 +0200)]
drm/i915: Do not temporarily disable the DPLL on i830
The current code clears the DPLL register entirely when re-enabling
VGA mode temporarily during the DPLL enable sequence. On i830 we want to
keep the DPLLs on all the time, so let's not do this temporary
disabling.
The current code does work, so this doesn't seem super important.
But I prefer that we make the behaviour 100% consistent.
Tvrtko Ursulin [Tue, 5 Mar 2019 11:04:08 +0000 (11:04 +0000)]
drm/i915: Relax mmap VMA check
Legacy behaviour was to allow non-page-aligned mmap requests, as does the
linux mmap(2) implementation by virtue of automatically rounding up for
the caller.
To avoid breaking legacy userspace relax the newly introduced fix.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Fixes: 5c4604e757ba ("drm/i915: Prevent a race during I915_GEM_MMAP ioctl with WC set") Reported-by: Guenter Roeck <linux@roeck-us.net> Cc: Adam Zabrocki <adamza@microsoft.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: <stable@vger.kernel.org> # v4.0+ Cc: Akash Goel <akash.goel@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: intel-gfx@lists.freedesktop.org Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20190305110409.28633-1-tvrtko.ursulin@linux.intel.com
drm/i915/selftests: Upgrade printing test/subtest name to pr_info
We're using pr_debug for things that we don't really want to see in the
CI log, but we may find useful during test development.
Let's upgrade the test name printer - we do want to see those in CI log.
Chris Wilson [Wed, 6 Mar 2019 08:47:04 +0000 (08:47 +0000)]
drm/i915: Pass around the intel_context
Instead of passing the gem_context and engine to find the instance of
the intel_context to use, pass around the intel_context instead. This is
useful for the next few patches, where the intel_context is no longer a
direct lookup.
drm/i915/icl: Default to Thread Group preemption for compute workloads
We assumed that the default preemption granularity is fine for ICL.
Unfortunately, it turns out that some drivers don't support mid-thread
preemption for compute workloads.
If a workload that doesn't support mid-thread preemption gets mid-thread
preempted, we're going to observe a GPU hang.
While I'm here, let's also update the "workaround" naming.
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com> Cc: Anuj Phogat <anuj.phogat@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Cc: Rafael Antognolli <rafael.antognolli@intel.com> Tested-by: Anuj Phogat <anuj.phogat@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Acked-by: Rafael Antognolli <rafael.antognolli@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20190305124827.23446-1-michal.winiarski@intel.com
Chris Wilson [Tue, 5 Mar 2019 18:03:32 +0000 (18:03 +0000)]
drm/i915: Move find_active_request() to the engine
To find the active request, we need only search along the individual
engine for the right request. This does not require touching any global
GEM state, so move it into the engine compartment.
Chris Wilson [Tue, 5 Mar 2019 18:03:31 +0000 (18:03 +0000)]
drm/i915/gtt: Mark ALL_ENGINES as dirty on ppGTT modification
Small simplification to set all bits in the dirty mask rather than
lookup the exact mask of populated engines. The bits for the engines
that do not exist are unused and so can safely set and then ignored.
Chris Wilson [Tue, 5 Mar 2019 18:03:30 +0000 (18:03 +0000)]
drm/i915: Store the BIT(engine->id) as the engine's mask
In the next patch, we are introducing a broad virtual engine to encompass
multiple physical engines, losing the 1:1 nature of BIT(engine->id). To
reflect the broader set of engines implied by the virtual instance, lets
store the full bitmask.
v2: Use intel_engine_mask_t (s/ring_mask/engine_mask/)
v3: Tvrtko voted for moah churn so teach everyone to not mention ring
and use $class$instance throughout.
v4: Comment upon the disparity in bspec for using VCS1,VCS2 in gen8 and
VCS[0-4] in later gen. We opt to keep the code consistent and use
0-index naming throughout.
Chris Wilson [Tue, 5 Mar 2019 16:26:43 +0000 (16:26 +0000)]
drm/i915: Remove last traces of exec-id (GEM_BUSY)
As we allow per-context engine allows the legacy concept of
I915_EXEC_RING no longer applies universally. We are still exposing the
unrelated exec-id in GEM_BUSY, so transition this ioctl (once more
slightly changing its ABI, but no one cares) over to only reporting the
uabi-class (not instance as we can not foreseeably fit those into the
small bitmask).
The only user of the extended ring information from GEM_BUSY is ddx/sna,
which tries to use the non-rcs business information to guide which
engine to use for subsequent operations on foreign bo. All that matters
for it is the decision between rcs and !rcs, so it is unaffected by the
change in higher bits.
Chris Wilson [Tue, 5 Mar 2019 15:09:13 +0000 (15:09 +0000)]
drm/i915: Just check the vebox IIR regardless
As we don't unmask and enable the vebox interrupts if the engine is not
being used, we will never generate the vebox interrupts as part of the
IIR and so can unconditionally check IIR without fear of chasing into
the vebox.
Chris Wilson [Tue, 5 Mar 2019 13:54:27 +0000 (13:54 +0000)]
drm/i915/gtt: Store scratch page size alongside not in the common struct
As the scratch page is the only one to be allocated with variable size,
rather than keep an unused slot in all i915_page_table structs, store it
alongside the vm->scratch_page.
Lucas De Marchi [Fri, 22 Feb 2019 23:02:54 +0000 (15:02 -0800)]
drm/i915: allow platforms without eDP transcoder
Define a HAS_TRANSCODER_EDP() macro that checks if we have defined an
offset for this transcoder. This allows platforms to be defined without
eDP transcoder.
drm/i915: Forcing a modeset when resetting HDMI link
With fastboot enabled in gen9+ it broke the HDMI reset as just
setting mode_changed to true causes a fastset and here we want a full
modeset that will disable and then enable the encoder of this HDMI
link actually, so setting connectors_changed instead that will cause
modeset as desired.
drm/i915: Fix atomic state leak when resetting HDMI link
Atomic state needs to be put even if the commit was successful.
Fixes: dba14b27dd3c ("drm/i915: Reinitialize sink scrambling/TMDS clock ratio on HPD") Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Lyude Paul <lyude@redhat.com> Signed-off-by: José Roberto de Souza <jose.souza@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190302003349.19189-1-jose.souza@intel.com
Rodrigo Vivi [Fri, 1 Mar 2019 17:27:03 +0000 (09:27 -0800)]
drm/i915: Yet another if/else sort of newer to older platforms.
No functional change. Just a reorg to match the preferred
behavior.
When rebasing internal branch on top of latest sort I noticed
few more cases that needs to get reordered.
Let's do in a bundle this time and hoping there's no other
missing places.
v2: Check for HSW/BDW ULT before generic IS_HASWELL or
IS_BROADWELL or it doesn't work as pointed by Ville.
But also ULT came afterwards anyway.
v3: Accepting suggestions from Lucas:
Sort CNL/CFL, KBL/SKL, and use <= 8 removing chv and bdw.
Chris Wilson [Mon, 4 Mar 2019 11:41:13 +0000 (11:41 +0000)]
drm/i915: Acquire breadcrumb ref before cancelling
We may race the interrupt signaling with retirement, in which case the
order in which we acquire the reference inside the interrupt is vital to
provide the correct barrier against the request being freed in
retirement, i.e. we need to acquire our reference before marking the
breadcrumb as cancelled (as soon as the breadcrumb is cancelled
retirement may drop its reference to the request without serialisation
with the interrupt handler).
<3>[ 683.372226] BUG i915_request (Tainted: G U ): Object already free
<3>[ 683.372269] -----------------------------------------------------------------------------
Lucas De Marchi [Tue, 26 Feb 2019 00:49:00 +0000 (16:49 -0800)]
drm/i915: extract AUX mask assignment to separate function
No change in behavior, this only allows to more easily follow the flow
of gen8_de_irq_handler without the mask assignments for each platform.
This also re-organizes the branches a little bit, so the one-off case
for CNL_WITH_PORT_F is separate from the generic gen >= 11.
Lucas De Marchi [Fri, 22 Feb 2019 23:23:22 +0000 (15:23 -0800)]
drm/i915/icl: move MG pll hw_state readout
Let the MG plls have their own hooks since it shares very little with
other PLL types. It's also better so the platform info contains the info
if the PLL is for MG PHY rather than relying on the PLL ids.
Chris Wilson [Fri, 1 Mar 2019 14:03:47 +0000 (14:03 +0000)]
drm/i915: Fix I915_EXEC_RING_MASK
This was supposed to be a mask of all known rings, but it is being used
by execbuffer to filter out invalid rings, and so is instead mapping high
unused values onto valid rings. Instead of a mask of all known rings,
we need it to be the mask of all possible rings.
We don't want to busywait on the GPU if we have other work to do. If we
give non-busywaiting workloads higher (initial) priority than workloads
that require a busywait, we will prioritise work that is ready to run
immediately. We then also have to be careful that we don't give earlier
semaphores an accidental boost because later work doesn't wait on other
rings, hence we keep a history of semaphore usage of the dependency chain.
v2: Stop rolling the bits into a chain and just use a flag in case this
request or any of our dependencies use a semaphore. The rolling around
was contagious as Tvrtko was heard to fall off his chair.
Chris Wilson [Fri, 1 Mar 2019 17:09:00 +0000 (17:09 +0000)]
drm/i915: Use HW semaphores for inter-engine synchronisation on gen8+
Having introduced per-context seqno, we now have a means to identity
progress across the system without feel of rollback as befell the
global_seqno. That is we can program a MI_SEMAPHORE_WAIT operation in
advance of submission safe in the knowledge that our target seqno and
address is stable.
However, since we are telling the GPU to busy-spin on the target address
until it matches the signaling seqno, we only want to do so when we are
sure that busy-spin will be completed quickly. To achieve this we only
submit the request to HW once the signaler is itself executing (modulo
preemption causing us to wait longer), and we only do so for default and
above priority requests (so that idle priority tasks never themselves
hog the GPU waiting for others).
As might be reasonably expected, HW semaphores excel in inter-engine
synchronisation microbenchmarks (where the 3x reduced latency / increased
throughput more than offset the power cost of spinning on a second ring)
and have significant improvement (can be up to ~10%, most see no change)
for single clients that utilize multiple engines (typically media players
and transcoders), without regressing multiple clients that can saturate
the system or changing the power envelope dramatically.
v3: Drop the older NEQ branch, now we pin the signaler's HWSP anyway.
v4: Tell the world and include it as part of scheduler caps.
Chris Wilson [Fri, 1 Mar 2019 17:08:59 +0000 (17:08 +0000)]
drm/i915: Keep timeline HWSP allocated until idle across the system
In preparation for enabling HW semaphores, we need to keep in flight
timeline HWSP alive until its use across entire system has completed,
as any other timeline active on the GPU may still refer back to the
already retired timeline. We both have to delay recycling available
cachelines and unpinning old HWSP until the next idle point.
An easy option would be to simply keep all used HWSP until the system as
a whole was idle, i.e. we could release them all at once on parking.
However, on a busy system, we may never see a global idle point,
essentially meaning the resource will be leaked until we are forced to
do a GC pass. We already employ a fine-grained idle detection mechanism
for vma, which we can reuse here so that each cacheline can be freed
immediately after the last request using it is retired.
v3: Keep track of the activity of each cacheline.
v4: cacheline_free() on canceling the seqno tracking
v5: Finally with a testcase to exercise wraparound
v6: Pack cacheline into empty bits of page-aligned vaddr
v7: Use i915_utils to hide the pointer casting around bit manipulation
Chris Wilson [Fri, 1 Mar 2019 17:08:58 +0000 (17:08 +0000)]
drm/i915/execlists: Suppress redundant preemption
On unwinding the active request we give it a small (limited to internal
priority levels) boost to prevent it from being gazumped a second time.
However, this means that it can be promoted to above the request that
triggered the preemption request, causing a preempt-to-idle cycle for no
change. We can avoid this if we take the boost into account when
checking if the preemption request is valid.
v2: After preemption the active request will be after the preemptee if
they end up with equal priority.
v3: Tvrtko pointed out that this, the existing logic, makes
I915_PRIORITY_WAIT non-preemptible. Document this interesting quirk!
v4: Prove Tvrtko was right about WAIT being non-preemptible and test it.
v5: Except not all priorities were made equal, and the WAIT not preempting
is only if we start off as !NEWCLIENT.
v6: More commentary after coming to an understanding about what I had
forgotten to say.
Ville Syrjälä [Thu, 28 Feb 2019 17:36:39 +0000 (19:36 +0200)]
drm/i915: Finalize Wa_1408961008:icl
The icl wm1+ underrun w/a has been added to the spec. It changed
slightly from the previous incarnation by requiring that we mirror
the lines watermark and the ignore lines bit from WM0 into WM1.
Chris Wilson [Fri, 1 Mar 2019 11:05:44 +0000 (11:05 +0000)]
drm/i915: Introduce i915_timeline.mutex
A simple mutex used for guarding the flow of requests in and out of the
timeline. In the short-term, it will be used only to guard the addition
of requests into the timeline, taken on alloc and released on commit so
that only one caller can construct a request into the timeline
(important as the seqno and ring pointers must be serialised). This will
be used by observers to ensure that the seqno/hwsp is stable. Later,
when we have reduced retiring to only operate on a single timeline at a
time, we can then use the mutex as the sole guard required for retiring.
Chris Wilson [Thu, 28 Feb 2019 22:06:39 +0000 (22:06 +0000)]
drm/i915/execlists: Suppress mere WAIT preemption
WAIT is occasionally suppressed by virtue of preempted requests being
promoted to NEWCLIENT if they have not all ready received that boost.
Make this consistent for all WAIT boosts that they are not allowed to
preempt executing contexts and are merely granted the right to be at the
front of the queue for the next execution slot. This is in keeping with
the desire that the WAIT boost be a minor tweak that does not give
excessive promotion to its user and open ourselves to trivial abuse.
The problem with the inconsistent WAIT preemption becomes more apparent
as the preemption is propagated across the engines, where one engine may
preempt and the other not, and we be relying on the exact execution
order being consistent across engines (e.g. using HW semaphores to
coordinate parallel execution).
v2: Also protect GuC submission from false preemption loops.
v3: Build bug safeguards and better debug messages for st.
v4: Do the priority bumping in unsubmit (i.e. on preemption/reset
unwind), applying it earlier during submit causes out-of-order execution
combined with execute fences.
v5: Call sw_fence_fini for our dummy request (Matthew)
Chris Wilson [Thu, 28 Feb 2019 10:20:35 +0000 (10:20 +0000)]
drm/i915: Remove second level open-coded rcu work
We currently use a worker queued from an rcu callback to determine when
a how grace period has elapsed while we remained idle. We use this idle
delay to infer that we will be idle for a while and this is a suitable
point at which we can trim our global memory caches.
Since we wrote that, this mechanism now exists as rcu_work, and having
converted the idle shrinkers over to using that, we can remove our own
variant.
v2: Say goodbye to gt.epoch as well.
v3: Remove the misplaced and redundant comment before parking globals
Chris Wilson [Thu, 28 Feb 2019 10:20:33 +0000 (10:20 +0000)]
drm/i915: Make request allocation caches global
As kmem_caches share the same properties (size, allocation/free behaviour)
for all potential devices, we can use global caches. While this
potential has worse fragmentation behaviour (one can argue that
different devices would have different activity lifetimes, but you can
also argue that activity is temporal across the system) it is the
default behaviour of the system at large to amalgamate matching caches.
The benefit for us is much reduced pointer dancing along the frequent
allocation paths.
v2: Defer shrinking until after a global grace period for futureproofing
multiple consumers of the slab caches, similar to the current strategy
for avoiding shrinking too early.
Chris Wilson [Wed, 27 Feb 2019 21:41:59 +0000 (21:41 +0000)]
drm/i915: Report engines are idle if already parked
If we have parked, then we must have passed an idleness test and still
be idle. We chose not to use this shortcut in the past so that we could
use the idleness test at any time and inspect HW. However, some HW like
Sandybridge, doesn't like being woken up frivolously, so avoid doing so.
CI reports that this is not as reliable as it first appears, with
failures starting to sporadically occur in selftests.
Fixes: 0b702dca2658 ("drm/i915: Avoid waking the engines just to check if they are idle") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190227204654.14907-1-chris@chris-wilson.co.uk
Chris Wilson [Wed, 27 Feb 2019 11:45:50 +0000 (11:45 +0000)]
drm/i915: Avoid waking the engines just to check if they are idle
Exploit that reads of the ring registers return 0 from the engine when
it is idle and we do not apply forcewake to know that if the engine is
idle then both reads will be identical (and so we interpret the ring as
idle).
The ulterior motive is to try and reduce the number of spurious wakeups
to avoid untimely death, such as:
Chris Wilson [Tue, 26 Feb 2019 10:23:54 +0000 (10:23 +0000)]
drm/i915: Skip scanning for signalers if we are already inflight
When a request has its priority changed, we traverse the graph of all of
its signalers to raise their priorities to match (priority inheritance).
If the request has already started executing its payload, we know that
all of its signalers must have signaled and we do not need to process
our list of signalers.
Ville Syrjälä [Mon, 25 Feb 2019 17:41:04 +0000 (19:41 +0200)]
drm/i915/sdvo: Read out HDMI infoframes
Read the HDMI infoframes from the hbuf and unpack them into
the crtc state.
Well, actually just AVI infoframe for now but let's write the
infoframe readout code in a more generic fashion in case we
expand this later.
Note that Daniel was sceptical about the benefit if this and
also concerned about the potential for crappy sdvo encoders not
implementing the hbuf read commands. My (admittedly limited)
experience is that such encoders don't implement even the
get/set hdmi encoding commands and thus would always be treated
as dvi only. Hence I believe this is safe, and also IMO preferable
having quirks to deal with missing readout support. The readout
support is neatly isolated in the sdvo code whereas the quirk
would leak to other parts of the driver (state checker, fastboot,
etc.) thus complicating the lives of other people.
Ville Syrjälä [Mon, 25 Feb 2019 17:41:03 +0000 (19:41 +0200)]
drm/i915/sdvo: Precompute HDMI infoframes
As with regular HDMI encoders, let's precompute the infoframes
(actually just AVI infoframe for the time being) with SDVO HDMI
encoders.
v2: Drop the WARN_ON() from drm_hdmi_avi_infoframe_from_display_mode()
return since that could genuinely fail due to user asking
for incompatible aspect ratio
v3: .compute_config() now returns int
Ville Syrjälä [Mon, 25 Feb 2019 17:41:01 +0000 (19:41 +0200)]
drm/i915: Precompute HDMI infoframes
Store the infoframes in the crtc state and precompute them in
.compute_config(). While precomputing we'll also fill out the
inforames.enable bitmask appropriately.
v2: Drop the null packet stuff (Daniel)
Add a FIXME for lspcon
v3: .compute_config() now returns int
Ville Syrjälä [Mon, 25 Feb 2019 17:41:00 +0000 (19:41 +0200)]
drm/i915: Store mask of enabled infoframes in the crtc state
Store the mask of enabled infoframes in the crtc state. We'll start
with just the readout for HDMI encoder, and we'll expand this
to compute the bitmask in .compute_config() later. SDVO will also
follow later.
Chris Wilson [Tue, 26 Feb 2019 09:49:22 +0000 (09:49 +0000)]
drm/i915/selftests: Exercise resetting during non-user payloads
In selftests/live_hangcheck, we have a lot of tests for resetting simple
spinners, but nothing quite prepared us for how the GPU reacted to
triggering a reset outside of the safe spinner. These two subtests fill
the ring with plain old empty, non-spinning requests, and then triggers
a reset. Without a user-payload to blame, these requests will exercise
the 'non-started' paths and mostly be replayed verbatim.
Chris Wilson [Tue, 26 Feb 2019 09:49:21 +0000 (09:49 +0000)]
drm/i915: Remove i915_request.global_seqno
Having weaned the interrupt handling off using a single global execution
queue, we no longer need to emit a global_seqno. Note that we still have
a few assumptions about execution order along engine timelines, but this
removes the most obvious artefact!
Chris Wilson [Tue, 26 Feb 2019 09:49:19 +0000 (09:49 +0000)]
drm/i915: Replace global_seqno with a hangcheck heartbeat seqno
To determine whether an engine has 'stuck', we simply check whether or
not is still on the same seqno for several seconds. To keep this simple
mechanism intact over the loss of a global seqno, we can simply add a
new global heartbeat seqno instead. As we cannot know the sequence in
which requests will then be completed, we use a primitive random number
generator instead (with a cycle long enough to not matter over an
interval of a few thousand requests between hangcheck samples).
The alternative to using a dedicated seqno on every request is to issue
a heartbeat request and query its progress through the system. Sadly
this requires us to reduce struct_mutex so that we can issue requests
without requiring that bkl.
v2: And without the extra CS_STALL for the hangcheck seqno -- we don't
need strict serialisation with what comes later, we just need to be sure
we don't write the hangcheck seqno before our batch is flushed.
drm/i915: Call MG_DP_MODE() macro with the right parameters order
The commit that this patch fixes changed the order of the parameters
of MG_DP_MODE() but din't update the callers, breaking type-c on ICL.
Fixes: 58106b7d816e ("drm/i915: Make MG PHY macros semantically consistent") Cc: Clint Taylor <clinton.a.taylor@intel.com> Cc: Imre Deak <imre.deak@intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Aditya Swarup <aditya.swarup@intel.com> Cc: Manasi navare <manasi.d.navare@intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by: José Roberto de Souza <jose.souza@intel.com> Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com> Signed-off-by: Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20190222202437.6575-1-jose.souza@intel.com
Chris Wilson [Sat, 23 Feb 2019 00:01:02 +0000 (00:01 +0000)]
drm/i915/pmu: Always sample an active ringbuffer
As we no longer have a precise indication of requests queued to an
engine, make no presumptions and just sample the ring registers to see
if the engine is busy.
v2: Report busy while the ring is idling on a semaphore/event.
v3: Give the struct a name!
v4: Always 0 outside the powerwell; trusting the powerwell is
accurate enough for our sampling pmu.
v5: Protect against gen7 mmio madness and try to improve grammar
Chris Wilson [Thu, 21 Feb 2019 10:29:19 +0000 (10:29 +0000)]
drm/i915: Reorder struct_mutex-vs-reset_lock in i915_gem_fault()
Annoyingly, struct_mutex was not entirely eliminated from the reset
pathway; for reasons of its own, intel_display_resume() requires
struct_mutex to prepare the planes it already captured. To avoid the
immediate problem of a deadlock between the struct_mutex and the reset
srcu, we have to acquire the reset_lock before struct_mutex in
i915_gem_fault(). Now any wait underneath struct_mutex will result us in
having to forcibly reset all inflight rendering, less than ideal, but
better than a deadlock (and will do for the short term).
Uma Shankar [Wed, 20 Feb 2019 19:05:19 +0000 (00:35 +0530)]
drm/i915/icl: Drop redundant gamma mode mask
gamma mode mask was not considering the 30th and 31st bits.
Due to this state readout was masking these bits, causing a
mismatch and false warning, even though the registers were
updated correctly. Dropped the gamma mode mask as it is
redundant and ideally entire register content should be
matching. This resolves the state mismatch warnings.
drm/i915/guc: Calling guc_disable_communication in all suspend paths
This aim of this patch is to call guc_disable_communication in all
suspend paths. The reason to introduce this is to resolve a bug that
occurred due to suspend late not being called in the hibernate devices
path.
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Sujaritha Sundaresan <sujaritha.sundaresan@intel.com> Reviewed-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20190220013927.9488-3-sujaritha.sundaresan@intel.com
Chris Wilson [Wed, 20 Feb 2019 22:55:56 +0000 (22:55 +0000)]
drm/i915: Prevent user context creation while wedged
Introduce a new ABI method for detecting a wedged driver by reporting
-EIO from DRM_IOCTL_I915_GEM_CONTEXT_CREATE.
This came up in considering how to handle context recovery from
userspace. There we wish to create a new context after the original is
banned (the clients opts into the no recovery after reset strategy) in
order to rebuild the mesa context from scratch. In doing so, if the
device was wedged and not the context banned, we would fall into a loop
of permanently trying to recreate the context and never making forward
progress. This patch would inform the client that we are no longer able
to create a context, and the client would have no choice but to abort
(or at least inform its callers about the lost device for anv).
Ville Syrjälä [Thu, 14 Feb 2019 19:22:19 +0000 (21:22 +0200)]
drm/i915: Extend skl+ crc sources with more planes
On skl the crc registers were extended to provide plane crcs
for up to 7 planes. Add the new crc sources.
The current code uses the ivb+ register definitions for skl+
which does happen to work as the plane1, plane2, and dmux/pf
bits happen the match what ivb+ had. So no bug in the current
code.
Ville Syrjälä [Thu, 14 Feb 2019 19:22:18 +0000 (21:22 +0200)]
drm/i915: Remove the broken DP CRC support for g4x
DP CRCs don't really work on g4x. If you want any CRCs on DP you must
select the CRC source before the port is enabled, otherwise the CRC
source select bits simply ignore any writes to them. And once the port
is enabled we mustn't change the CRC source select until the port is
disabled. That almost works, but not quite :( Eventually the CRC source
select bits get permanently stuck one way or the other, and after that
a reboot (or possibly a display reset) is needed to get working CRCs
on that pipe (not matter which CRC source we try to use).
Additionally the DFT scrambler reset bits we're trying to use don't
seem to exist on g4x. There are some potentially relevant looking bits
in the pipe registers, but when I tried it I got stable looking CRCs
without setting any bits for this.
If there is a way to make DP CRCs work reliably on g4x, I wasn't
able to find it. So let's just remove the broken code we have.
Ville Syrjälä [Thu, 14 Feb 2019 19:22:17 +0000 (21:22 +0200)]
drm/i915: Use named initializers for the crc source name array
We assume that the index of the string in the crc source names
array matches the enum value for the crc source. Let's use named
initializers to make sure that is indeed the case even if someone
rearranges either the enum or the array.
Chris Wilson [Tue, 19 Feb 2019 12:22:03 +0000 (12:22 +0000)]
drm/i915: Reduce the RPS shock
Limit deboosting and boosting to keep ourselves at the extremes
when in the respective power modes (i.e. slowly decrease frequencies
while in the HIGH_POWER zone and slowly increase frequencies while
in the LOW_POWER zone). On idle, we will hit the timeout and drop
to the next level quickly, and conversely if busy we expect to
hit a waitboost and rapidly switch into max power.
This should improve the UX experience by keeping the GPU clocks higher
than they ostensibly should be (based on simple busyness) by switching
into the INTERACTIVE mode (due to waiting for pageflips) and increasing
clocks via waitboosting. This will incur some additional power, our
saving grace should be rc6 and powergating to keep the extra current
draw in check.
Food for future thought would be deadline scheduling? If we know certain
contexts (high priority compositors) absolutely must hit the next vblank
then we can raise the frequencies ahead of time. Part of this is covered
by per-context frequencies, where userspace is given control over the
frequency range they want the GPU to execute at (for largely the same
problem as this, where the workload is very latency sensitive but at the
EI level appears mostly idle). Indeed, the per-context series does
extend the modeset boosting to include a frequency range tweak which
seems applicable to solving this jittery UX behaviour.
Reported-by: Lyude Paul <lyude@redhat.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109408
References: 0d55babc8392 ("drm/i915: Drop stray clearing of rps->last_adj")
References: 60548c554be2 ("drm/i915: Interactive RPS mode") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Lyude Paul <lyude@redhat.com> Cc: Eero Tamminen <eero.t.tamminen@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Michel Thierry <michel.thierry@intel.com>
Quoting Lyude Paul:
> Before reverting 0d55babc8392754352f1058866dd4182ae587d11: [4.20]
>
> 35 measurements [of gnome-shell animations]
> Average: 33.65657142857143 FPS
> FPS observed: 20.8 - 46.87 FPS
> Percentage under 60 FPS: 100.0%
> Percentage under 55 FPS: 100.0%
> Percentage under 50 FPS: 100.0%
> Percentage under 45 FPS: 97.14285714285714%
> Percentage under 40 FPS: 97.14285714285714%
> Percentage under 35 FPS: 45.714285714285715%
> Percentage under 30 FPS: 11.428571428571429%
> Percentage under 25 FPS: 2.857142857142857%
>
> After reverting: [4.19 behaviour]
>
> 30 measurements
> Average: 49.833666666666666 FPS
> FPS observed: 33.85 - 60.0 FPS
> Percentage under 60 FPS: 86.66666666666667%
> Percentage under 55 FPS: 70.0%
> Percentage under 50 FPS: 53.333333333333336%
> Percentage under 45 FPS: 20.0%
> Percentage under 40 FPS: 6.666666666666667%
> Percentage under 35 FPS: 6.666666666666667%
> Percentage under 30 FPS: 0%
> Percentage under 25 FPS: 0%
>
> Patched:
> 42 measurements
> Average: 46.05428571428571 FPS
> FPS observed: 1.82 - 59.98 FPS
> Percentage under 60 FPS: 88.09523809523809%
> Percentage under 55 FPS: 61.904761904761905%
> Percentage under 50 FPS: 45.23809523809524%
> Percentage under 45 FPS: 35.714285714285715%
> Percentage under 40 FPS: 33.33333333333333%
> Percentage under 35 FPS: 19.047619047619047%
> Percentage under 30 FPS: 7.142857142857142%
> Percentage under 25 FPS: 4.761904761904762%
Ramalingam C [Sat, 16 Feb 2019 17:37:03 +0000 (23:07 +0530)]
drm/i915: Fix KBL HDCP2.2 encrypt status signalling
HDCP transmitter is supposed to indicate the HDCP encryption status of
the link through enc_en signals in a window of time called "window of
opportunity" defined by HDCP HDMI spec.
But on KBL this timing of signalling has an issue. To fix the issue this
WA of resetting the signalling is required.
v2:
WA is moved into the toggle_signalling [Daniel]
v3:
Commit msg is rewritten with more information
v4:
Reviewed-by Daniel.
Ramalingam C [Sat, 16 Feb 2019 17:37:02 +0000 (23:07 +0530)]
drm/i915: CP_IRQ handling for DP HDCP2.2 msgs
Implements the
Waitqueue is created to wait for CP_IRQ
Signaling the CP_IRQ arrival through atomic variable.
For applicable DP HDCP2.2 msgs read wait for CP_IRQ.
As per HDCP2.2 spec "HDCP Transmitters must process CP_IRQ interrupts
when they are received from HDCP Receivers"
Without CP_IRQ processing, DP HDCP2.2 H_Prime msg was getting corrupted
while reading it based on corresponding status bit. This creates the
random failures in reading the DP HDCP2.2 msgs.
v2:
CP_IRQ arrival is tracked based on the atomic val inc [daniel]
Recording the reviewed-by Daniel from IRC.
Ramalingam C [Sat, 16 Feb 2019 17:37:01 +0000 (23:07 +0530)]
drm/i915: Implement the HDCP2.2 support for HDMI
Implements the HDMI adaptation specific HDCP2.2 operations.
Basically these are DDC read and write for authenticating through
HDCP2.2 messages.
v2: Rebased.
v3:
No more special handling of Gmbus burst read for AKE_SEND_CERT.
Style fixed with few naming. [Uma]
%s/PARING/PAIRING
v4:
msg_sz is initialized at definition.
Lookup table is defined for HDMI HDCP2.2 msgs [Daniel].
v5: Rebased.
v6:
Make a function as inline [Uma]
%s/uintxx_t/uxx
v7:
Errors due to sinks are reported as DEBUG logs.
Adjust to the new mei interface.
v8:
ARRAY_SIZE for the # of array members [Jon & Daniel].
hdcp adaptation is added as a const in the hdcp_shim [Daniel]
Ramalingam C [Sat, 16 Feb 2019 17:37:00 +0000 (23:07 +0530)]
drm/i915: Implement the HDCP2.2 support for DP
Implements the DP adaptation specific HDCP2.2 functions.
These functions perform the DPCD read and write for communicating the
HDCP2.2 auth message back and forth.
v2:
wait for cp_irq is merged with this patch. Rebased.
v3:
wait_queue is used for wait for cp_irq [Chris Wilson]
v4:
Style fixed.
%s/PARING/PAIRING
Few style fixes [Uma]
v5:
Lookup table for DP HDCP2.2 msg details [Daniel].
Extra lines are removed.
v6: Rebased.
v7:
Fixed some regression introduced at v5. [Ankit]
Macro HDCP_2_2_RX_CAPS_VERSION_VAL is reused [Uma]
Converted a function to inline [Uma]
%s/uintxx_t/uxx
v8:
Error due to the sinks are reported as DEBUG logs.
Adjust to the new mei interface.
v9:
ARRAY_SIZE for no of array members [Jon & Daniel]
return of the wait_for_cp_irq is made as void [Daniel]
Wait for HDCP2.2 msg is done based on polling the reg bit than
CP_IRQ based. [Daniel]
hdcp adaptation is added as a const in the hdcp_shim [Daniel]
v10:
config_stream_type is redefined [Daniel]
DP Errata specific defines are moved into intel_dp.c.
projects / mirror_ubuntu-kernels.git / log