GuC needs to know which registers and how they will be saved and
restored during event such as engine reset or power state changes.
For now only the base address of reg state is initialized. The
detail register table probably will be setup in future GuC TDR or
Preemption patch series.
Signed-off-by: Alex Dai <yu.dai@intel.com>
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450468812-4882-5-git-send-email-yu.dai@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
GuC supports different scheduling policies for its four internal
queues. Currently these have been set to the same default values
as KMD_NORMAL queue.
Particularly POLICY_MAX_NUM_WI is set to 15 to match GuC internal
maximum submit queue numbers to avoid an out-of-space problem.
This value indicates max number of work items allowed to be queued
for one DPC process. A smaller value will let GuC schedule more
frequently while a larger number may increase chances to optimize
cmds (such as collapse cmds from same lrc) with risks that keeps
CS idle.
v1: tidy up code
Signed-off-by: Alex Dai <yu.dai@intel.com>
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450468812-4882-4-git-send-email-yu.dai@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The GuC firmware uses this for various purposes. The ADS itself is
a chunk of memory created by driver to share with GuC. Its members
are usually addresses telling where GuC to access them, including
things like scheduler policies, register list that will be saved
and restored during reset etc.
This is the first patch of a series to enable GuC ADS. For now, we
only create the ADS obj whilst keep it disabled.
v1: remove dead code checking return of kmap_atomic (Chris Wilson)
v2: use kmap instead of the atomic version of it.
Signed-off-by: Alex Dai <yu.dai@intel.com>
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450468812-4882-3-git-send-email-yu.dai@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The GuC code needs to know the size of a logical context, so we
expose get_lr_context_size(), renaming it intel_lr_context__size()
to fit the naming conventions for nonstatic functions.
For: VIZ-2021
Signed-off-by: Dave Gordon <david.s.gordon@intel.com>
Signed-off-by: Alex Dai <yu.dai@intel.com>
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450468812-4882-2-git-send-email-yu.dai@intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Split GuC work queue space checking from submission and move it to
ring_alloc_request_extras. The reason is that failure in later
i915_add_request() won't be handled. In the case timeout happens,
driver can return early in order to handle the error.
v1: Move wq_reserve_space to ring_reserve_space
v2: Move wq_reserve_space to alloc_request_extras (Chris Wilson)
v3: The work queue head pointer is cached by driver now. So we can
quickly return if space is available.
s/reserve/check/g (Dave Gordon)
v4: Update cached wq head after ring doorbell; check wq space before
ring doorbell in case unexpected error happens; call wq space
check only when GuC submission is enabled. (Dave Gordon)
Signed-off-by: Alex Dai <yu.dai@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450295155-10050-1-git-send-email-yu.dai@intel.com
Reviewed-by: Dave Gordon <david.s.gordon@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If the system has no available swap pages, we cannot make forward
progress in the shrinker by releasing active pages, only by releasing
purgeable pages which are immediately reaped. Take total_swap_pages into
account when counting up available objects to be shrunk and subsequently
shrinking them. By doing so, we avoid unbinding objects that cannot be
shrunk and so wasting CPU cycles flushing those objects from the GPU to
the system and then immediately back again (as they will more than
likely be reused shortly after).
Based on a patch by Akash Goel.
v2: frontswap registers extra swap pages available for the system, so it
is already include in the count of available swap pages.
v3: Use get_nr_swap_pages() to query the currently available amount of
swap space. This should also stop us from shrinking the GPU buffers if
we ever run out of swap space. Though at that point, we would expect the
oom-notifier to be running and failing miserably...
Reported-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: linux-mm@kvack.org
Cc: Akash Goel <akash.goel@intel.com>
Cc: sourab.gupta@intel.com
Link: http://patchwork.freedesktop.org/patch/msgid/1449244734-25733-2-git-send-email-chris@chris-wilson.co.uk
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Some modules, like i915.ko, use swappable objects and may try to swap
them out under memory pressure (via the shrinker). Before doing so, they
want to check using get_nr_swap_pages() to see if any swap space is
available as otherwise they will waste time purging the object from the
device without recovering any memory for the system. This requires the
nr_swap_pages counter to be exported to the modules.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Goel, Akash" <akash.goel@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: linux-mm@kvack.org
Link: http://patchwork.freedesktop.org/patch/msgid/1449244734-25733-1-git-send-email-chris@chris-wilson.co.uk
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
According to PRM, some parts of HW require the addresses to be in
a canonical form, where bits [63:48] == [47]. Let's convert addresses to
canonical form prior to relocating and return converted offsets to
userspace. We also need to make sure that userspace is using addresses
in canonical form in case of softpin.
v2: Whitespace fixup, gen8_canonical_addr description (Chris, Ville)
v3: Rebase on top of softpin, fix a hole in relocate_entry,
s/expect/require (Chris)
v4: Handle softpin in validate_exec_list (Chris)
v5: Convert back to canonical form at copy_to_user time (Chris)
v6: Don't use struct exec_object2 in place of exec_object
v7: Use sign_extend64 for converting to canonical form (Joonas),
reject non-canonical and non-page-aligned offset for softpin (Chris)
v8: Convert back to non-canonical form in a function,
split the test for EXEC_OBJECT_PINNED (Chris)
v9: s/canonial/canonical, drop accidental double newline (Chris)
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Michał Winiarski <michal.winiarski@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1451409892-13708-1-git-send-email-michal.winiarski@intel.com
Testcase: igt/gem_bad_reloc/negative-reloc-blt
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92699
Cc: drm-intel-fixes@lists.freedesktop.org
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The total delay of HDMI hotplug detecting with 30ms is sometimes not
enoughtfor HDMI live status up with specific HDMI monitors in BSW platform.
After doing experiments for following monitors, it needs 80ms at least
for those worst cases.
Lenovo L246 1xwA (4 failed, necessary hot-plug delay: 58/40/60/40ms)
Philips HH2AP (9 failed, necessary hot-plug delay: 80/50/50/60/46/40/58/58/39ms)
BENQ ET-0035-N (6 failed, necessary hot-plug delay: 60/50/50/80/80/40ms)
DELL U2713HM (2 failed, necessary hot-plug delay: 58/59ms)
HP HP-LP2475w (5 failed, necessary hot-plug delay: 70/50/40/60/40ms)
It looks like 70-80 ms is BSW platform needs in some bad cases of the
monitors at this end (8 times delay at most). Keep less than 100ms for
HDCP pulse HPD low (with at least 100ms) to respond a plug out.
Reviewed-by: Cooper Chiou <cooper.chiou@intel.com>
Tested-by: Gary Wang <gary.c.wang@intel.com>
Cc: Gavin Hindman <gavin.hindman@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Shobhit Kumar <shobhit.kumar@intel.com>
Signed-off-by: Gary Wang <gary.c.wang@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450858295-12804-1-git-send-email-gary.c.wang@intel.com
Tested-by: Shobhit Kumar <shobhit.kumar@intel.com>
Cc: drm-intel-fixes@lists.freedesktop.org
Fixes: 237ed86c69 ("drm/i915: Check live status before reading edid")
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When the crtc is configured but not active we currently clip to (0,0)x(0,0).
This results in differences in calculations depending on dpms setting.
When the crtc is enabled but not active run check_plane as if it were on,
but afterwards set plane_state->visible = false for the checks.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1447945645-32005-13-git-send-email-maarten.lankhorst@linux.intel.com
On skylake when calculating plane visibility with the crtc in
dpms off mode the real cdclk may be different from what it would be
if the crtc was active. This may result in a WARN_ON(cdclk < crtc_clock)
from skl_max_scale. The fix is to keep a atomic_cdclk that would be true
if all crtc's were active.
This is required to get the same calculations done correctly regardless
of dpms mode.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1447945645-32005-12-git-send-email-maarten.lankhorst@linux.intel.com
Parallel modesets are still not allowed, but this will allow updating
a different crtc during a modeset if the clock is not changed.
Additionally when all pipes are DPMS off the cdclk will be lowered
to the minimum allowed.
Changes since v1:
- Add dev_priv->active_crtcs for tracking which crtcs are active.
- Rename min_cdclk to min_pixclk and move to dev_priv.
- Add a active_crtcs mask which is updated atomically.
- Add intel_atomic_state->modeset which is set on modesets.
- Commit new pixclk/active_crtcs right after state swap.
Changes since v2:
- Make the changes related to max_pixel_rate calculations more readable.
Changes since v3:
- Add cherryview and missing WARN_ON to readout.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
This fixes a warning when the crtc is turned off. In that case fb
will be NULL, and crtc_clock will be 0. Because the crtc is no longer
active this is not a bug, and shouldn't trigger the WARN_ON.
Also remove handling a null crtc_state, with all transitional helpers
gone this can no longer happen.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1448360945-5723-2-git-send-email-maarten.lankhorst@linux.intel.com
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Atomic changes broke check_digital_port_conflicts(). It needs to look
at the global situation instead of just trying to find a conflict
within the current atomic state.
This bug made my HSW explode spectacularly after I had split the DDI
encoders into separate DP and HDMI encoders. With the fix, things
seem much more solid.
I hope holding the connection_mutex is enough protection that we can
actually walk the connectors even if they're not part of the current
atomic state...
v2: Regenerate the patch so that it actually applies (Jani)
Cc: stable@vger.kernel.org
Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Fixes: 5448a00d3f ("drm/i915: Don't use staged config in check_digital_port_conflicts()")
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1449764551-12466-1-git-send-email-ville.syrjala@linux.intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Using __stringify(x) instead of #x adds support for macros as
a parameter and compile-time concatenation reduces the runtime
overhead.
Slightly increases the .text size but should not matter.
v2:
- Define I915_STATE_WARN_ON though I915_STATE_WARN
(Bikeshed inspiration by Chris)
v3:
- More specific commit message
v4:
- Do not directly pass arbitary string as format, instead
guard with "%s" (Dave)
Cc: Rob Clark <robdclark@gmail.com>
Cc: Dave Gordon <david.s.gordon@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v3)
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450441647-23924-3-git-send-email-joonas.lahtinen@linux.intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Clean up after 0c82312f3f ("drm/i915: Pin the ifbdev for the
info->system_base GGTT mmapping"):
At each of the remaining "goto out" in intelfb_alloc(), fb can only be
either an ERR_PTR or NULL, so the call to drm_framebuffer_unreference()
is now obsolete.
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Link: http://patchwork.freedesktop.org/patch/msgid/56756c41.c306c20a.d0602.1830SMTPIN_ADDED_MISSING@mx.google.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Turns out CHV pipe C was glued on somewhat poorly, and there's something
wrong with the cursor. If the cursor straddles the left screen edge,
and is then moved away from the edge or disabled, the pipe will often
underrun. If enough underruns are triggered quickly enough the pipe
will fall over and die (it just scans out a solid color and reports
a constant underrun). We need to turn the disp2d power well off and
on again to recover the pipe.
None of that is very nice for the user, so let's just refuse to place
the cursor in the compromised position. The ddx appears to fall back
to swcursor when the ioctl returns an error, so theoretically there's
no loss of functionality for the user (discounting swcursor bugs).
I suppose most cursors images actually have the hotspot not exactly
at 0,0 so under typical conditions the fallback will in fact kick in
as soon as the cursor touches the left edge of the screen.
Any atomic compositor should anyway be prepared to fall back to
GPU composition when things don't work out, so there should be no
problem with those.
Other things that I tried to solve this include flipping all
display related clock gating knobs I could find, increasing the
minimum gtt alignment all the way up to 512k. I also tried to see
if there are more specific screen coordinates that hit the bug, but
the findings were somewhat inconclusive. Sometimes the failures
happen almost across the whole left edge, sometimes more at the very
top and around the bottom half. I wasn't able to find any real pattern
to these variations, so it seems our only choice is to just refuse
to straddle the left screen edge at all.
Cc: stable@vger.kernel.org
Cc: Jason Plum <max@warheads.net>
Testcase: igt/kms_chv_cursor_fail
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92826
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450459479-16286-1-git-send-email-ville.syrjala@linux.intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Otherwise usage in the i915 debug macros yields problems due to
i915_drv.h <-> i915_trace.h <-> intel_drv.h include loops.
v2:
- Document not-so-obvious need for linux/cache.h (Chris)
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450436898-20408-2-git-send-email-joonas.lahtinen@linux.intel.com
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
It is unclear if this is even required on BXT.
v2: Make sure to set the default value to false. Uncertain how my compiler
doesn't complain with v1.
Cc: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450374597-7021-1-git-send-email-benjamin.widawsky@intel.com
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The total delay of HDMI hotplug detecting with 30ms have already
been split into a resolution of 3 retries of 10ms each, for the worst
cases. But it still suffered from only waiting 10ms at most in
intel_hdmi_detect(). This patch corrects it by reading hotplug status
with 4 times at most for 30ms delay.
v2:
- straight up to loop execution for more clear in code readability
- mdelay will replace with msleep by Daniel's new patch
drm/i915: mdelay(10) considered harmful
- suggest to re-evaluate try times for being compatible to old HDMI monitor
Reviewed-by: Cooper Chiou <cooper.chiou@intel.com>
Tested-by: Gary Wang <gary.c.wang@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Gavin Hindman <gavin.hindman@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Cc: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Gary Wang <gary.c.wang@intel.com>
[danvet: fixup conflict with s/mdelay/msleep/ patch.]
Cc: drm-intel-fixes@lists.freedesktop.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There was a silent conflict between
commit 0a87871626
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu Oct 15 14:23:01 2015 +0200
drm/i915: restore ggtt double-bind avoidance
and
commit 5bab6f60cb
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Fri Oct 23 18:43:32 2015 +0100
drm/i915: Serialise updates to GGTT with access through GGTT on Braswell
thankfully caught by the extra WARN safegaurd in 0a878716. Since we now
override the GGTT insert_pages callback when installing the aliasing
ppgtt, we assert that the callback is the original ggtt routine.
However, on Braswell we now use a different insertion routine to
serialise access through the GGTT with updating the PTE and hence the
conflict. To avoid the conflict, move the custom insertion routine for
Braswell down a level.
Reported-by: Ville Syrjälä <ville.syrjala@linux.intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Michel Thierry <michel.thierry@intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1447859979-20107-1-git-send-email-chris@chris-wilson.co.uk
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: drm-intel-fixes@lists.freedesktop.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
commit 344df9809f ("drm/i915/skl: Disable coarse power gating up until F0")
failed to take into account that the same workaround is used in guc
when forcewake is sampled.
Wrap the condition check inside a macro and use it in both places
to fix the guc side scope.
Cc: Dave Gordon <david.s.gordon@intel.com>
Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450286318-6854-1-git-send-email-mika.kuoppala@intel.com
The workarounds for disabling hdc invalidation and also forcing
context to be non coherent, are advised to be used up until rev D0.
However as it was found that rev F0, without the
WaForceEnableNonCoherent might system hang if the mesa
tried to use coherent mode.
As these two workarounds are about non coherent access, are
grouped in scope and they point the same HSD, increase the
scope of both to set default behaviour to non coherent access.
References: HSD: gen9lp/2131413
References: http://lists.freedesktop.org/archives/mesa-dev/2015-November/101515.html
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450448093-22906-1-git-send-email-mika.kuoppala@intel.com
Limit busywaiting only to the request currently being processed by the
GPU. If the request is not currently being processed by the GPU, there
is a very low likelihood of it being completed within the 2 microsecond
spin timeout and so we will just be wasting CPU cycles.
v2: Check for logical inversion when rebasing - we were incorrectly
checking for this request being active, and instead busywaiting for
when the GPU was not yet processing the request of interest.
v3: Try another colour for the seqno names.
v4: Another colour for the function names.
v5: Remove the forced coherency when checking for the active request. On
reflection and plenty of recent experimentation, the issue is not a
cache coherency problem - but an irq/seqno ordering problem (timing issue).
Here, we do not need the w/a to force ordering of the read with an
interrupt.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Cc: "Rantala, Valtteri" <valtteri.rantala@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1449833608-22125-4-git-send-email-chris@chris-wilson.co.uk
When waiting for high frequency requests, the finite amount of time
required to set up the irq and wait upon it limits the response rate. By
busywaiting on the request completion for a short while we can service
the high frequency waits as quick as possible. However, if it is a slow
request, we want to sleep as quickly as possible. The tradeoff between
waiting and sleeping is roughly the time it takes to sleep on a request,
on the order of a microsecond. Based on measurements of synchronous
workloads from across big core and little atom, I have set the limit for
busywaiting as 10 microseconds. In most of the synchronous cases, we can
reduce the limit down to as little as 2 miscroseconds, but that leaves
quite a few test cases regressing by factors of 3 and more.
The code currently uses the jiffie clock, but that is far too coarse (on
the order of 10 milliseconds) and results in poor interactivity as the
CPU ends up being hogged by slow requests. To get microsecond resolution
we need to use a high resolution timer. The cheapest of which is polling
local_clock(), but that is only valid on the same CPU. If we switch CPUs
because the task was preempted, we can also use that as an indicator that
the system is too busy to waste cycles on spinning and we should sleep
instead.
__i915_spin_request was introduced in
commit 2def4ad99b [v4.2]
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Apr 7 16:20:41 2015 +0100
drm/i915: Optimistically spin for the request completion
v2: Drop full u64 for unsigned long - the timer is 32bit wraparound safe,
so we can use native register sizes on smaller architectures. Mention
the approximate microseconds units for elapsed time and add some extra
comments describing the reason for busywaiting.
v3: Raise the limit to 10us
v4: Now 5us.
Reported-by: Jens Axboe <axboe@kernel.dk>
Link: https://lkml.org/lkml/2015/11/12/621
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Cc: "Rantala, Valtteri" <valtteri.rantala@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1449833608-22125-3-git-send-email-chris@chris-wilson.co.uk
The busywait in __i915_spin_request() does not respect pending signals
and so may consume the entire timeslice for the task instead of
returning to userspace to handle the signal.
In the worst case this could cause a delay in signal processing of 20ms,
which would be a noticeable jitter in cursor tracking. If a higher
resolution signal was being used, for example to provide fairness of a
server timeslices between clients, we could expect to detect some
unfairness between clients (i.e. some windows not updating as fast as
others). This issue was noticed when inspecting a report of poor
interactivity resulting from excessively high __i915_spin_request usage.
Fixes regression from
commit 2def4ad99b [v4.2]
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Apr 7 16:20:41 2015 +0100
drm/i915: Optimistically spin for the request completion
v2: Try to assess the impact of the bug
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc; "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: Eero Tamminen <eero.t.tamminen@intel.com>
Cc: "Rantala, Valtteri" <valtteri.rantala@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1449833608-22125-2-git-send-email-chris@chris-wilson.co.uk
pm_runtime_{use,dont_use}_autosuspend() controls whether the device's
sysfs power/autosuspend_delay_ms file is writeable or returns -EIO on
access to user space. Since
commit 25b181b46e
Author: Imre Deak <imre.deak@intel.com>
Date: Thu Dec 17 13:44:56 2015 +0200
drm/i915: get a permanent RPM reference on platforms w/o RPM support
this sysfs file is writeable also on platforms without RPM support, but
userspace (at least IGT) depends on this file being unchangable to
determine whether the device supports runtime PM at all. So restore the
old behavior.
This gets rid of igt/pm_rpm failures on old platforms without RPM
support, where the test should be skipped.
Testcase: igt/pm_rpm/basic-rte
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: David Weinehall <david.weinehall@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450371873-878-1-git-send-email-imre.deak@intel.com
If we fail to reconstruct the BIOS fb (e.g., because the FB is too
large), we'll be left with plane state that indicates the primary plane
is visible yet has a NULL fb. This mismatch causes problems later on
(e.g., for the watermark code). Since we've failed to reconstruct the
BIOS FB, the best solution is to just disable the primary plane and
pretend the BIOS never had it enabled.
v2: Add intel_pre_disable_primary() call (Maarten)
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: drm-intel-fixes@lists.freedesktop.org
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1449171462-30763-2-git-send-email-matthew.d.roper@intel.com
A long time ago (before 3.14) we relied on a permanent pinning of the
ifbdev to lock the fb in place inside the GGTT. However, the
introduction of stealing the BIOS framebuffer and reusing its address in
the GGTT for the fbdev has muddied waters and we use an inherited fb.
However, the inherited fb is only pinned whilst it is active and we no
longer have an explicit pin for the info->system_base mmapping used by
the fbdev. The result is that after some aperture pressure the fbdev may
be evicted, but we continue to write the fbcon into the same GGTT
address - overwriting anything else that may be put into that offset.
The effect is most pronounced across suspend/resume as
intel_fbdev_set_suspend() does a full clear over the whole scanout.
v2: Only unpin the intel_fb is we allocate it. If we inherit the fb from
the BIOS, we do not own the pinned vma (except for the reference we add
in this patch for our access via info->screen_base).
v3: Finish balancing the vma pinning for the normal !preallocated case.
v4: Try to simplify the pinning even further.
v5: Leak the VMA (cleaned up by object-free) to avoid complicated error paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Goel, Akash" <akash.goel@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Lukas Wunner <lukas@wunner.de>
Cc: drm-intel-fixes@lists.freedesktop.org
Link: http://patchwork.freedesktop.org/patch/msgid/1449245126-26158-1-git-send-email-chris@chris-wilson.co.uk
Tested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
As we mark the preallocated objects as bound, we should also flag them
correctly as being map-and-fenceable (if appropriate!) so that later
users do not get confused and try and rebind the pinned vma in order to
get a map-and-fenceable binding.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "Goel, Akash" <akash.goel@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: drm-intel-fixes@lists.freedesktop.org
Link: http://patchwork.freedesktop.org/patch/msgid/1448029000-10616-1-git-send-email-chris@chris-wilson.co.uk
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
I missed this myself when reviewing
commit 237ed86c69
Author: Sonika Jindal <sonika.jindal@intel.com>
Date: Tue Sep 15 09:44:20 2015 +0530
drm/i915: Check live status before reading edid
Long sleeps like this really shouldn't waste cpu cycles spinning.
Cc: Sonika Jindal <sonika.jindal@intel.com>
Cc: "Wang, Gary C" <gary.c.wang@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1449859455-32609-1-git-send-email-daniel.vetter@ffwll.ch
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The device should be on for the whole duration of the update, so check
for this.
v2:
- use the existing dev_priv directly everywhere (Ville)
v3:
- check also that we are in an RPM atomic section (Chris)
- add the assert to i915_ggtt_insert_entries/clear_range too (Chris)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1450203038-5150-11-git-send-email-imre.deak@intel.com
In some cases we want to check whether we hold an RPM wakelock reference
for the whole duration of a sequence. To achieve this add a new RPM
atomic sequence counter that we increment any time the wakelock refcount
drops to zero. Check whether the sequence number stays the same during
the atomic section and that we hold the wakelock at the beginning of the
section.
Motivated by Chris.
v2-v3:
- unchanged
v4:
- swap the order of atomic_read() and assert_rpm_wakelock_held() in
assert_rpm_atomic_begin() to avoid race
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v3)
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450203038-5150-10-git-send-email-imre.deak@intel.com
With this change we have the corresponding wake lock checks in both the
rpm get and put functions.
v2-v3:
- unchanged
v4:
- keep the corresponding check in the get helper (Chris)
v5:
- add a note to the commit message that with this change we have the
checks both in the rpm get and put functions (Joonas)
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/1450203038-5150-9-git-send-email-imre.deak@intel.com
Atm, we assert that the device is not suspended until the point when the
device is truly put to a suspended state. This is fine, but we can catch
more problems if we check that RPM refcount is non-zero. After that one
drops to zero we shouldn't access the device any more, even if the actual
device suspend may be delayed. Change assert_rpm_wakelock_held()
accordingly to check for a non-zero RPM refcount in addition to the
current device-not-suspended check.
For the new asserts to work we need to annotate every place explicitly in
the code where we expect that the device is powered. The places where we
only assume this, but may not hold an RPM reference:
- driver load
We assume the device to be powered until we enable RPM. Make this
explicit by taking an RPM reference around the load function.
- system and runtime sudpend/resume handlers
These handlers are called when the RPM reference becomes 0 and know the
exact point after which the device can get powered off. Disable the
RPM-reference-held check for their duration.
- the IRQ, hangcheck and RPS work handlers
These handlers are flushed in the system/runtime suspend handler
before the device is powered off, so it's guaranteed that they won't
run while the device is powered off even though they don't hold any
RPM reference. Disable the RPM-reference-held check for their duration.
In all these cases we still check that the device is not suspended.
These explicit annotations also have the positive side effect of
documenting our assumptions better.
This caught additional WARNs from the atomic modeset path, those should
be fixed separately.
v2:
- remove the redundant HAS_RUNTIME_PM check (moved to patch 1) (Ville)
v3:
- use a new dedicated RPM wakelock refcount to also catch cases where
our own RPM get/put functions were not called (Chris)
- assert also that the new RPM wakelock refcount is 0 in the RPM
suspend handler (Chris)
- change the assert error message to be more meaningful (Chris)
- prevent false assert errors and check that the RPM wakelock is 0 in
the RPM resume handler too
- prevent false assert errors in the hangcheck work too
- add a device not suspended assert check to the hangcheck work
v4:
- rename disable/enable_rpm_asserts to disable/enable_rpm_wakeref_asserts
and wakelock_count to wakeref_count
- disable the wakeref asserts in the IRQ handlers and RPS work too
- update/clarify commit message
v5:
- mark places we plan to change to use proper RPM refcounting with
separate DISABLE/ENABLE_RPM_WAKEREF_ASSERTS aliases (Chris)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1450227139-13471-1-git-send-email-imre.deak@intel.com
As a preparation for follow-up patches add a new helper that checks
whether we hold an RPM reference, since this is what we want most of
the cases. Atm this helper will only check for the HW suspended state, a
follow-up patch will do the actual change to check the refcount instead.
One exception is the forcewake release timer function, where it's
guaranteed that the HW is on even though the RPM refcount drops to zero.
This guarantee is provided by flushing the timer in the runtime suspend
handler. So leave the assert_device_not_suspended check in place there.
Also rename assert_device_suspended for consistency and export these
helpers as a preparation for the follow-up patches.
No functional change.
v3:
- change the assert warning message to be more meaningful (Chris)
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1450203038-5150-6-git-send-email-imre.deak@intel.com
We don't really need to check this flag in the get/put/assert helpers,
as on platforms without RPM support we won't ever enable RPM. That means
pm.suspend will be always false and the assert will be always true.
Do this to simplify the code and to let us extend the RPM asserts to all
platforms for a better coverage.
Motivated by Ville.
v2-v3:
- unchanged
v4:
- remove the HAS_RUNTIME_PM check from intel_runtime_pm_enable() too
made possible by the previous two patches
v5:
- rebased on the previous new patch in the series that keeps
HAS_RUNTIME_PM() in intel_runtime_pm_enable() with a permanent
reference taken there
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v3)
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450352931-16498-1-git-send-email-imre.deak@intel.com
Currently we disable RPM functionality on platforms that doesn't support
this by not putting/getting the RPM reference we receive from the RPM
core during driver loading/unloading respectively. This is somewhat
obscure, so make it more explicit by keeping a reference dedicated for
this particular purpose whenever the driver is loaded. This makes it
possible to remove the HAS_RUNTIME_PM() special casing from every other
places in the next patch.
v2:
- fix intel_runtime_pm_get vs. intel_runtime_pm_put in
intel_power_domains_fini()
v3:
- take only a low level RPM reference so the ref tracking asserts
continue to work (Ville)
- update the commit message
- move the patch earlier for bisectability
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1450352696-16135-1-git-send-email-imre.deak@intel.com
We can make the RPM dependency on RC6 explcit in the code by taking an
actual RPM reference, instead of avoiding to drop the initial one. This
will also enable us to remove the HAS_RUNTIME_PM special casing from
more places in the next patch.
v2:
- fixed typo in commit message (Joonas)
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/1450203038-5150-4-git-send-email-imre.deak@intel.com
Jani Nikula