If any of the perf tests run into 0 time, not only are we liable to
divide by zero, but the result would be highly questionable.
Nevertheless, let's not have a div-by-zero error.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Andi Shyti <andi.shyti@intel.com>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210108204026.20682-2-chris@chris-wilson.co.uk

On error we unpin and free the wa_ctx.vma, but do not clear any of the
derived flags. During lrc_init, we look at the flags and attempt to
dereference the wa_ctx.vma if they are set. To protect the error path
where we try to limp along without the wa_ctx, make sure we clear those
flags!
Reported-by: NMatt Roper <matthew.d.roper@intel.com>
Fixes: 604a8f6f ("drm/i915/lrc: Only enable per-context and per-bb buffers if set")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: <stable@vger.kernel.org> # v4.15+
Reviewed-by: NMatt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210108204026.20682-1-chris@chris-wilson.co.uk

In the next^W future patch, we remove the strict priority system and
continuously re-evaluate the relative priority of tasks. As such we need
to enable the timeslice whenever there is more than one context in the
pipeline. This simplifies the decision and removes some of the tweaks to
suppress timeslicing, allowing us to lift the timeslice enabling to a
common spot at the end of running the submission tasklet.

One consequence of the suppression is that it was reducing fairness
between virtual engines on an over saturated system; undermining the
principle for timeslicing.

v2: Commentary
v3: Commentary for the right cancel_timer()
v4: Add tracing for why we need a timeslice

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2802
Testcase: igt/gem_exec_balancer/fairslice
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210107132322.28373-1-chris@chris-wilson.co.uk

Refactor our timer_list.expires checking into its own timer_active()
helper.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210107123541.17153-1-chris@chris-wilson.co.uk

Use memset_io() on the iomem, and silence sparse as we copy from the
iomem to normal system pages.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210106123939.18435-2-chris@chris-wilson.co.uk

AFter detecting a register mismatch between the protocontext and the
image generated by HW, immediately break out of the double loop.
Otherwise we end up with a second confusing error message.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210106123939.18435-1-chris@chris-wilson.co.uk

In generating the reference LRC, we want a page-aligned address for
simplicity in computing the offsets within. This then shares the
computation for the HW LRC which is mapped and so page aligned, making
the comparison straightforward. It seems that kmalloc(4k) is not always
returning from a 4k-aligned slab cache (which would give us a page aligned
address) so force alignment by explicitly allocating a page.
Reported-by: N"Gote, Nitin R" <nitin.r.gote@intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: "Gote, Nitin R" <nitin.r.gote@intel.com>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200923114156.17749-1-chris@chris-wilson.co.uk

If another sibling is able to claim the virtual request, by the time we
inspect the request under the lock it may no longer match the local
engine.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2877Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104115145.24460-4-chris@chris-wilson.co.uk

If the engine reset fails, we will attempt to resume with the current
inflight submissions. When that happens, we cannot assert that the
engine reset cleared the pending submission, so do not.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2878
Fixes: 16f2941a ("drm/i915/gt: Replace direct submit with direct call to tasklet")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104115145.24460-3-chris@chris-wilson.co.uk

Older platforms use rawclk to derive the CS clock. rawclk is being
determined during intel_device_info_init(), and so that needs to be
pushed slightly earlier.

Fixes: f170523a ("drm/i915/gt: Consolidate the CS timestamp clocks")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104115145.24460-2-chris@chris-wilson.co.uk

A few missed PTR_ERR() upon create_gang() errors.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104115145.24460-1-chris@chris-wilson.co.uk

In the near future, upstream will introduce a SZ_8G macro that is
slightly different to our own. Employ a temporary ifndef to avoid
compilation failure until we have backmerged.

References: 8b0fac44 ("sizes.h: add SZ_8G/SZ_16G/SZ_32G macros")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NJosé Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104171511.32684-1-chris@chris-wilson.co.uk

Some rcs0 workarounds were being incorrectly applied to the GT, and so
we failed to restore the expected register settings after a reset.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104114914.30165-2-chris@chris-wilson.co.uk

Some rcs0 workarounds were being incorrectly applied to the GT, and so
we failed to restore the expected register settings after a reset.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210104114914.30165-1-chris@chris-wilson.co.uk

Randconfig builds on 32-bit machines show lots of warnings for
the i915 driver for passing a 32-bit value into __const_hweight64():

drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c:2584:9: error: shift count >= width of type [-Werror,-Wshift-count-overflow]
        return hweight64(VDBOX_MASK(&i915->gt));
               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/asm-generic/bitops/const_hweight.h:29:49: note: expanded from macro 'hweight64'
 #define hweight64(w) (__builtin_constant_p(w) ? __const_hweight64(w) : __arch_hweight64(w))

Change it to hweight_long() to avoid the warning.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20210103135158.3591442-1-arnd@kernel.org

This allows us to remove pin_map from state allocation, which saves
us a few retry loops. We won't need this until first pin, anyway.
Signed-off-by: NMaarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: NThomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20201231170405.22843-1-chris@chris-wilson.co.uk

Let's modify the "workaround lost" error message slightly to make it
more clear what the various numbers represent.  Also, the 'expected'
value needs to be &'d with wa->read so that it doesn't include the mask
bits for masked registers (those bits are write-only in the hardware and
will usually always read out as 0's).
Signed-off-by: NMatt Roper <matthew.d.roper@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20201231191103.854519-1-matthew.d.roper@intel.com

Since we use a flag within i915_request.flags to indicate when we have
boosted the request (so that we only apply the boost) once, this can be
used as the serialisation with i915_request_retire() to avoid having to
explicitly take the i915_request.lock which is more heavily contended.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201231093149.19086-1-chris@chris-wilson.co.uk

We only need to evaluate the current status of the context when it is
scheduled in, we will force a reschedule when the context is closed
propagating the change to inflight contexts.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201231093946.11649-1-chris@chris-wilson.co.uk

Since we process schedule-in of a context after submitting the request,
if we decide to reset the context at that time, we also have to cancel
the requets we have marked for submission.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201230220028.17089-1-chris@chris-wilson.co.uk

Declare that, under extreme circumstances, the shrinker may need to wait
upon a request, in which case reset must not itself deadlock in order to
ensure forward progress of the driver. That is since the shrinker may
depend upon a reset, any reset cannot touch the shrinker.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201229141626.4773-1-chris@chris-wilson.co.uk

If supported by the backend, we can quickly look at the context's
inflight engine rather than search along the active list to confirm.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NAndi Shyti <andi.shyti@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201229144114.31686-1-chris@chris-wilson.co.uk

When splitting the Coffeelake define to also identify Cometlakes, I
missed the double fw_def for Coffeelake. That is only newer Cometlakes
use the cml specific guc firmware, older Cometlakes should use kbl
firmware.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2859
Fixes: 5f4ae270 ("drm/i915: Identify Cometlake platform")
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: <stable@vger.kernel.org> # v5.9+
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201229120828.29931-1-chris@chris-wilson.co.uk

Pull the individual strands of creating a custom heartbeat requests into
a pair of common functions. This will reduce the number of changes we
will need to make in future.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224160213.29521-1-chris@chris-wilson.co.uk

The reloc batch is short lived but can exist in the user visible ppGTT,
and since it's backed by an internal object, which lacks page clearing,
we should take care to clear it upfront.
Signed-off-by: NMatthew Auld <matthew.auld@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224151358.401345-2-matthew.auld@intel.com
Cc: stable@vger.kernel.org

The shadow batch is an internal object, which doesn't have any page
clearing, and since the batch_len can be smaller than the object, we
should take care to clear it.

Testcase: igt/gen9_exec_parse/shadow-peek
Fixes: 4f7af194 ("drm/i915: Support ro ppgtt mapped cmdparser shadow buffers")
Signed-off-by: NMatthew Auld <matthew.auld@intel.com>
Reviewed-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224151358.401345-1-matthew.auld@intel.com
Cc: stable@vger.kernel.org

Since schedule-in and schedule-out are now both always under the tasklet
bitlock, we can reduce the individual atomic operations to simple
instructions and worry less.

This notably eliminates the race observed with intel_context_inflight in
__engine_unpark().

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2583Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NTvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-9-chris@chris-wilson.co.uk

Now that the tasklet completely controls scheduling of the requests, and
we postpone scheduling out the old requests, we can keep a hanging
virtual request bound to the engine on which it hung, and remove it from
te queue. On release, it will be returned to the same engine and remain
in its queue until it is scheduled; after which point it will become
eligible for transfer to a sibling. Instead, we could opt to resubmit the
request along the virtual engine on unhold, making it eligible for load
balancing immediately -- but that seems like a pointless optimisation
for a hanging context.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-8-chris@chris-wilson.co.uk

Having recognised that we do not change the sibling until we schedule
out, we can then defer the decision to resubmit the virtual engine from
the unwind of the active queue to scheduling out of the virtual context.
This improves our resilence in virtual engine scheduling, and should
eliminate the rare cases of gem_exec_balance failing.

By keeping the unwind order intact on the local engine, we can preserve
data dependency ordering while doing a preempt-to-busy pass until we
have determined the new ELSP. This means that if we try to timeslice
between a virtual engine and a data-dependent ordinary request, the pair
will maintain their relative ordering and we will avoid the
resubmission, cancelling the timeslicing until further change.

The dilemma though is that we then may end up in a situation where the
'demotion' of the virtual request to an ordinary request in the engine
queue results in filling the ELSP[] with virtual requests instead of
spreading the load across the engines. To compensate for this, we mark
each virtual request and refuse to resubmit a virtual request in the
secondary ELSP slots, thus forcing subsequent virtual requests to be
scheduled out after timeslicing. By delaying the decision until we
schedule out, we will avoid unnecessary resubmission.

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2079
Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2098Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-7-chris@chris-wilson.co.uk

Let's only wait for the list iterator when decoupling the virtual
breadcrumb, as the signaling of all the requests may take a long time,
during which we do not want to keep the tasklet spinning.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-6-chris@chris-wilson.co.uk

The issue with stale virtual breadcrumbs remain. Now we have the problem
that if the irq-signaler is still referencing the stale breadcrumb as we
transfer it to a new sibling, the list becomes spaghetti. This is a very
small window, but that doesn't stop it being hit infrequently. To
prevent the lists being tangled (the iterator starting on one engine's
b->signalers but walking onto another list), always decouple the virtual
breadcrumb on schedule-out and make sure that the walker has stepped out
of the lists.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-5-chris@chris-wilson.co.uk

Inside schedule_out, we do extra work upon idling the context, such as
updating the runtime, kicking off retires, kicking virtual engines.
However, if we are in a series of processing single requests per
contexts, we may find ourselves scheduling out the context, only to
immediately schedule it back in during dequeue. This is just extra work
that we can avoid if we keep the context marked as inflight across the
dequeue. This becomes more significant later on for minimising virtual
engine misses.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-4-chris@chris-wilson.co.uk

Once a virtual engine has been bound to a sibling, it will remain bound
until we finally schedule out the last active request. We can not rebind
the context to a new sibling while it is inflight as the context save
will conflict, hence we wait. As we cannot then use any other sibliing
while the context is inflight, only kick the bound sibling while it
inflight and upon scheduling out the kick the rest (so that we can swap
engines on timeslicing if the previously bound engine becomes
oversubscribed).
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-3-chris@chris-wilson.co.uk

Rather than going back and forth between the rb_node entry and the
virtual_engine type, store the ve local and reuse it. As the
container_of conversion from rb_node to virtual_engine requires a
variable offset, performing that conversion just once shaves off a bit
of code.

v2: Keep a single virtual engine lookup, for typical use.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-2-chris@chris-wilson.co.uk

Rather than having special case code for opportunistically calling
process_csb() and performing a direct submit while holding the engine
spinlock for submitting the request, simply call the tasklet directly.
This allows us to retain the direct submission path, including the CS
draining to allow fast/immediate submissions, without requiring any
duplicated code paths, and most importantly greatly simplifying the
control flow by removing reentrancy. This will enable us to close a few
races in the virtual engines in the next few patches.

The trickiest part here is to ensure that paired operations (such as
schedule_in/schedule_out) remain under consistent locking domains,
e.g. when pulled outside of the engine->active.lock

v2: Use bh kicking, see commit 3c53776e ("Mark HI and TASKLET
softirq synchronous").
v3: Update engine-reset to be tasklet aware
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201224135544.1713-1-chris@chris-wilson.co.uk

As we shrink an object, also see if we can prune the dma-resv of idle
fences it is maintaining a reference to.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201223122051.4624-2-chris@chris-wilson.co.uk

If we want to reuse a fence that is in active use by the GPU, we have to
wait an uncertain amount of time, but if we reuse an inactive fence, we
can change it right away. Loop through the list of available fences
twice, ignoring any active fences on the first pass.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201223122051.4624-1-chris@chris-wilson.co.uk

Pull the GT clock information [used to derive CS timestamps and PM
interval] under the GT so that is it local to the users. In doing so, we
consolidate the two references for the same information, of which the
runtime-info took note of a potential clock source override and scaling
factors.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201223122359.22562-2-chris@chris-wilson.co.uk

We assume that both timestamps are driven off the same clock [reported
to userspace as I915_PARAM_CS_TIMESTAMP_FREQUENCY]. Verify that this is
so by reading the timestamp registers around a busywait (on an otherwise
idle engine so there should be no preemptions).

v2: Icelake (not ehl, nor tgl) seems to be using a fixed 80ns interval
for, and only for, CTX_TIMESTAMP -- or it may be GPU frequency and the
test is always running at maximum frequency?. As far as I can tell, this
isolated change in behaviour is undocumented.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: NMika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201223122359.22562-1-chris@chris-wilson.co.uk

We just need the context image from the logical state to force eviction
of many contexts, so simplify by avoiding the GEM context container.
Signed-off-by: NChris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: NMatthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20201223154509.14155-1-chris@chris-wilson.co.uk