Matthew Auld [Mon, 9 Oct 2017 11:00:24 +0000 (12:00 +0100)]
drm/i915: s/sg_mask/sg_page_sizes/
It's a little unclear what the sg_mask actually is, so prefer the more
meaningful name of sg_page_sizes.
Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Matthew Auld <matthew.auld@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/20171009110024.29114-1-matthew.auld@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Mon, 9 Oct 2017 08:44:01 +0000 (09:44 +0100)]
drm/i915: Early rejection of mappable GGTT pin attempts for large bo
Currently, we reject attempting to pin a large bo into the mappable
aperture, but only after trying to create the vma. Under debug kernels,
repeatedly creating and freeing that vma for an oversized bo consumes
one-third of the runtime for pwrite/pread tests as it is spent on
kmalloc/kfree tracking. If we move the rejection to before creating that
vma, we lose some accuracy of checking against the fence_size as opposed
to object size, though the fence can never be smaller than the object.
Note that the vma creation itself will reject an attempt to create a vma
larger than the GTT so we can remove one redundant test.
Chris Wilson [Mon, 9 Oct 2017 08:44:00 +0000 (09:44 +0100)]
drm/i915: Avoid evicting user fault mappable vma for pread/pwrite
Both pread/pwrite GTT paths provide a fast fallback in case we cannot
map the whole object at a time. Currently, we use the fallback for very
large objects and for active objects that would require remapping, but
we can also add active fault mappable objects to the list that we want
to avoid evicting. The rationale is that such fault mappable objects are
in active use and to evict requires tearing down the CPU PTE and forcing
a page fault on the next access; more costly, and intefers with other
processes, than our per-page GTT fallback.
Chris Wilson [Mon, 9 Oct 2017 08:43:59 +0000 (09:43 +0100)]
drm/i915: Try a minimal attempt to insert the whole object for relocations
As we have a lightweight fallback to insert a single page into the
aperture, try to avoid any heavier evictions when attempting to insert
the entire object.
Chris Wilson [Mon, 9 Oct 2017 08:43:58 +0000 (09:43 +0100)]
drm/i915: Check PIN_NONFAULT overlaps in evict_for_node
If the caller says that he doesn't want to evict any other faulting
vma, honour that flag. The logic was used in evict_something, but not
the more specific evict_for_node, now being used as a preliminary probe
since commit 606fec956c0e ("drm/i915: Prefer random replacement before
eviction search").
Fixes: 606fec956c0e ("drm/i915: Prefer random replacement before eviction search") Fixes: 821188778b9b ("drm/i915: Choose not to evict faultable objects from the GGTT")
References: https://bugs.freedesktop.org/show_bug.cgi?id=102490 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171009084401.29090-4-chris@chris-wilson.co.uk Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Chris Wilson [Mon, 9 Oct 2017 08:43:57 +0000 (09:43 +0100)]
drm/i915: Track user GTT faulting per-vma
We don't wish to refault the entire object (other vma) when unbinding
one partial vma. To do this track which vma have been faulted into the
user's address space.
v2: Use a local vma_offset to tidy up a multiline unmap_mapping_range().
Chris Wilson [Mon, 9 Oct 2017 08:43:56 +0000 (09:43 +0100)]
drm/i915: Consolidate get_fence with pin_fence
Following the pattern now used for obj->mm.pages, use just pin_fence and
unpin_fence to control access to the fence registers. I.e. instead of
calling get_fence(); pin_fence(), we now just need to call pin_fence().
This will make it easier to reduce the locking requirements around
fence registers.
Chris Wilson [Mon, 9 Oct 2017 08:43:55 +0000 (09:43 +0100)]
drm/i915: Pin fence for iomap
Acquire the fence register for the iomap in i915_vma_pin_iomap() on
behalf of the caller.
We probably want for the caller to specify whether the fence should be
pinned for their usage, but at the moment all callers do want the
associated fence, or none, so take it on their behalf.
Chris Wilson [Mon, 9 Oct 2017 11:03:01 +0000 (12:03 +0100)]
drm/i915: Provide an assert for when we expect forcewake to be held
Add assert_forcewakes_active() (the complementary function to
assert_forcewakes_inactive) that documents the requirement of a
function for its callers to be holding the forcewake ref (i.e. the
function is part of a sequence over which RC6 must be prevented).
One such example is during ringbuffer reset, where RC6 must be held
across the whole reinitialisation sequence.
v2: Include debug information in the WARN so we know which fw domain is
missing.
Chris Wilson [Mon, 9 Oct 2017 11:03:00 +0000 (12:03 +0100)]
drm/i915/selftests: Hold the rpm wakeref for the reset tests
The lowlevel reset functions expect the caller to be holding the rpm
wakeref for the device access across the reset. We were not explicitly
doing this in the sefltest, so for simplicity acquire the wakeref for
the duration of all subtests.
Chris Wilson [Mon, 9 Oct 2017 11:02:59 +0000 (12:02 +0100)]
drm/i915: Hold forcewake for the duration of reset+restart
Resetting the engine requires us to hold the forcewake wakeref to
prevent RC6 trying to happen in the middle of the reset sequence. The
consequence of an unwanted RC6 event in the middle is that random state
is then saved to the powercontext and restored later, which may
overwrite the mmio state we need to preserve (e.g. PD_DIR_BASE in the
legacy ringbuffer reset_ring_common()).
This was noticed in the live_hangcheck selftests when Haswell would
sporadically fail to restart during igt_reset_queue().
Chris Wilson [Mon, 9 Oct 2017 11:02:57 +0000 (12:02 +0100)]
drm/i915: Make i915_engine_info pretty printer to standalone
We can use drm_printer to hide the differences between printk and
seq_printf, and so make the i915_engine_info pretty printer able to be
called from different contexts and not just debugfs. For instance, I
want to use the pretty printer to debug kselftests.
Chris Wilson [Mon, 9 Oct 2017 09:20:19 +0000 (10:20 +0100)]
drm/i915: Assert we do not try to expand VMA for hugepage inside GGTT
We only apply the hugepage PD redirection inside the ppGTT, so during
i915_vma_insert() we want to exclude the GGTT from the additional
alignment constraints (thereby avoiding the extra GTT pressure from
fragmentation). Add an assert to document that intention alongside the
comment.
v2: After discussion with Matthew, make it a blanket GGTT ban
(previously we allowed the expansion for appgtt, and so indirectly
ggtt). There are issues we need to fix before allowing the current
appgtt to be used with hugepages, and if we do, we probably want more
care over when to expand/align, as the mappable aperture inside the ggtt
is precious.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Matthew Auld <matthew.william.auld@gmail.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171009092019.20747-1-chris@chris-wilson.co.uk Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Ville Syrjälä [Fri, 1 Apr 2016 15:37:25 +0000 (18:37 +0300)]
drm/i915: Read timings from the correct transcoder in intel_crtc_mode_get()
intel_crtc->config->cpu_transcoder isn't yet filled out when
intel_crtc_mode_get() gets called during output probing, so we should
not use it there. Instead intel_crtc_mode_get() figures out the correct
transcoder on its own, and that's what we should use.
If the BIOS boots LVDS on pipe B, intel_crtc_mode_get() would actually
end up reading the timings from pipe A instead (since PIPE_A==0),
which clearly isn't what we want.
It looks to me like this may have been broken by
commit eccb140bca67 ("drm/i915: hw state readout&check support for cpu_transcoder")
as that one removed the early initialization of cpu_transcoder from
intel_crtc_init().
Cc: stable@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: Rob Kramer <rob@solution-space.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Reported-by: Rob Kramer <rob@solution-space.com> Fixes: eccb140bca67 ("drm/i915: hw state readout&check support for cpu_transcoder")
References: https://lists.freedesktop.org/archives/dri-devel/2016-April/104142.html Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: https://patchwork.freedesktop.org/patch/msgid/1459525046-19425-1-git-send-email-ville.syrjala@linux.intel.com
Matthew Auld [Fri, 6 Oct 2017 22:18:25 +0000 (23:18 +0100)]
drm/i915: add support for 64K scratch page
Before we can fully enable 64K pages, we need to first support a 64K
scratch page if we intend to support the case where we have object sizes
< 2M, since any scratch PTE must also point to a 64K region. Without
this our 64K usage is limited to objects which completely fill the
page-table, and therefore don't need any scratch.
Matthew Auld [Fri, 6 Oct 2017 22:18:21 +0000 (23:18 +0100)]
drm/i915: align 64K objects to 2M
We can't mix 64K and 4K pte's in the same page-table, so for now we
align 64K objects to 2M to avoid any potential mixing. This is
potentially wasteful but in reality shouldn't be too bad since this only
applies to the virtual address space of a 48b PPGTT.
Matthew Auld [Fri, 6 Oct 2017 22:18:20 +0000 (23:18 +0100)]
drm/i915: align the vma start to the largest gtt page size
For the 48b PPGTT try to align the vma start address to the required
page size boundary to guarantee we use said page size in the gtt. If we
are dealing with multiple page sizes, we can't guarantee anything and
just align to the largest. For soft pinning and objects which need to be
tightly packed into the lower 32bits we don't force any alignment.
v2: various improvements suggested by Chris
v3: use set_pages and better placement of page_sizes
Matthew Auld [Fri, 6 Oct 2017 22:18:19 +0000 (23:18 +0100)]
drm/i915: introduce vm set_pages/clear_pages
Move the setting/clearing of the vma->pages to a vm operation. Doing so
neatens things up a little, but more importantly gives us a sane place
to also set/clear the vma->pages_sizes, which we introduce later in
preparation for supporting huge-pages.
v2: remove redundant vma->pages check
v3: GEM_BUG_ON(vma->pages) following i915_vma_remove
Matthew Auld [Fri, 6 Oct 2017 22:18:18 +0000 (23:18 +0100)]
drm/i915: introduce page_size members
In preparation for supporting huge gtt pages for the ppgtt, we introduce
page size members for gem objects. We fill in the page sizes by
scanning the sg table.
v2: pass the sg_mask to set_pages
v3: calculate the sg_mask inline with populating the sg_table where
possible, and pass to set_pages along with the pages.
Matthew Auld [Fri, 6 Oct 2017 22:18:17 +0000 (23:18 +0100)]
drm/i915: push set_pages down to the callers
Each backend is now responsible for calling __i915_gem_object_set_pages
upon successfully gathering its backing storage. This eliminates the
inconsistency between the async and sync paths, which stands out even
more when we start throwing around an sg_mask in a later patch.
Matthew Auld [Fri, 6 Oct 2017 22:18:16 +0000 (23:18 +0100)]
drm/i915: introduce page_sizes field to dev_info
In preparation for huge gtt pages expose page_sizes as part of the
device info, to indicate the page sizes supported by the HW. Currently
only 4K is supported.
Matthew Auld [Fri, 6 Oct 2017 22:18:14 +0000 (23:18 +0100)]
drm/i915: introduce simple gemfs
Not a fully blown gemfs, just our very own tmpfs kernel mount. Doing so
moves us away from the shmemfs shm_mnt, and gives us the much needed
flexibility to do things like set our own mount options, namely huge=
which should allow us to enable the use of transparent-huge-pages for
our shmem backed objects.
v2: various improvements suggested by Joonas
v3: move gemfs instance to i915.mm and simplify now that we have
file_setup_with_mnt
v4: fallback to tmpfs shm_mnt upon failure to setup gemfs
Matthew Auld [Fri, 6 Oct 2017 22:18:13 +0000 (23:18 +0100)]
mm/shmem: introduce shmem_file_setup_with_mnt
We are planning to use our own tmpfs mnt in i915 in place of the
shm_mnt, such that we can control the mount options, in particular
huge=, which we require to support huge-gtt-pages. So rather than roll
our own version of __shmem_file_setup, it would be preferred if we could
just give shmem our mnt, and let it do the rest.
Chris Wilson [Thu, 5 Oct 2017 19:10:05 +0000 (20:10 +0100)]
drm/i915/execlists: Add a comment for the extra MI_ARB_ENABLE
Michel Thierry noticed that we were applying WaDisableCtxRestoreArbitration
even to gen9, which does not require the w/a. The rationale is that we
need to enable MI arbitration for execlists to work, and to be safe we
do that before every batch (in addition to every context switch into the
batch). Since this is not clear from the single line comment suggesting
the MI_ARB_ENABLE is solely for the w/a, add a little more detail.
Chris Wilson [Fri, 6 Oct 2017 11:56:17 +0000 (12:56 +0100)]
drm/i915: Order two completing nop_submit_request
If two nop's (requests in-flight following a wedged device) complete at
the same time, the global_seqno value written to the HWSP is undefined
as the two threads are not serialized.
v2: Use irqsafe spinlock. We expect the callback may be called from
inside another irq spinlock, so we can't unconditionally restore irqs.
Chris Wilson [Mon, 2 Oct 2017 10:04:16 +0000 (11:04 +0100)]
drm/i915: Silence compiler warning for hsw_power_well_enable()
Not all compilers are able to determine that pg is guarded by wait_fuses
and so may think that pg is used uninitialized.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org> Fixes: b2891eb2531e ("drm/i915/hsw+: Add has_fuses power well attribute") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Imre Deak <imre.deak@intel.com> Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171002100416.25865-1-chris@chris-wilson.co.uk Reviewed-by: Imre Deak <imre.deak@intel.com>
Chris Wilson [Fri, 6 Oct 2017 10:40:38 +0000 (11:40 +0100)]
drm/i915: Try harder to finish the idle-worker
If a worker requeues itself, it may switch to a different kworker pool,
which flush_work() considers as complete. To be strict, we then need to
keep flushing the work until it is no longer pending.
Joonas Lahtinen [Fri, 6 Oct 2017 08:49:40 +0000 (11:49 +0300)]
drm/i915: Unify uC variable types to avoid flooding checkpatch.pl
With the code motion mostly done, convert all the uC code away
from uint??_t at once (only a couple dozen variables), so that
reading the checkpatch.pl output should actually pinpoint if
a new uint??_t was accidentally introduced.
v2: - Include intel_uc_fw.h too (Sagar)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michał Winiarski <michal.winiarski@intel.com> Cc: Sagar Arun Kamble <sagar.a.kamble@intel.com> Cc: Anusha Srivatsa <anusha.srivatsa@intel.com> Cc: Sujaritha Sundaresan <sujaritha.sundaresan@intel.com> Reviewed-by: Sagar Arun Kamble <sagar.a.kamble@intel.com> Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171006084940.15910-1-joonas.lahtinen@linux.intel.com
drm/i915: Use crtc_state_is_legacy_gamma in intel_color_check
crtc_state_is_legacy_gamma also checks for CTM, which was missing from
intel_color_check. By using the same condition for commit and check
we reduce the chance of mismatches.
This was spotted by KASAN while trying to rework kms_color igt test.
Oscar Mateo [Wed, 4 Oct 2017 15:39:52 +0000 (08:39 -0700)]
drm/i915/cnl: Do not add an extra page for precaution in the Gen10 LRC size
BSpec indicates exactly 16752 DWORDs (17 pages), plus one page for PPHWSP. Please
notice that, when looking at the BSpec context image table, the right filter has
to be applied (e.g. "CNL") as some rows are excluded for specific GENs.
BSpec: 1383
v2: Update count and add BSpec tag (Joonas)
v3: Warning about filters in the commit message (Joonas)
Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Fixes: 7fd0b1a ("drm/i915/cnl: Add Gen10 LRC size") Signed-off-by: Oscar Mateo <oscar.mateo@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Ben Widawsky <ben@bwidawsk.net> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/1507131592-29209-1-git-send-email-oscar.mateo@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Arnd Bergmann [Thu, 5 Oct 2017 12:08:26 +0000 (14:08 +0200)]
drm/i915: avoid potential uninitialized variable use
One of the recent changes introduced a warning about
undefined behavior in the sanity checking:
drivers/gpu/drm/i915/intel_ddi.c: In function 'intel_ddi_hdmi_level':
drivers/gpu/drm/i915/intel_ddi.c:654:6: error: 'n_hdmi_entries' may be used uninitialized in this function [-Werror=maybe-uninitialized]
It seems that the new cnl specific get_buf_trans functions
can return uninitialized data if the voltage level is set
to an unexpected value. This changes the code to always return
'1' in that error case, which seems like the safest choice
as we use one less than the number as an array index later on.
Arnd Bergmann [Thu, 5 Oct 2017 12:07:22 +0000 (14:07 +0200)]
drm/i915/selftests: fix check for intel IOMMU
An earlier bugfix tried to work around this build failure:
drivers/gpu/drm/i915/selftests/mock_gem_device.c: In function 'mock_gem_device':
drivers/gpu/drm/i915/selftests/mock_gem_device.c:151:20: error: 'struct dev_archdata' has no member named 'iommu'
Checking for CONFIG_IOMMU_API is not sufficient as a compile-time
test since that may be enabled in configurations that have neither
INTEL_IOMMU not AMD_IOMMU enabled. This changes the check to
INTEL_IOMMU instead, as this is the only case we actually care about.
Chris Wilson [Thu, 5 Oct 2017 10:59:27 +0000 (11:59 +0100)]
drm/i915/selftests: Remember to create the fake preempt context
For the fake device we have our own set of mock contexts that need to
match the real contexts we normally create. Currently this requires us
to manually instantiate them for the selftests, which I forgot.
Reported-by: Matthew Auld <matthew.william.auld@gmail.com> Fixes: e7af3116836f ("drm/i915: Introduce a preempt context") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michał Winiarski <michal.winiarski@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Matthew Auld <matthew.william.auld@gmail.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171005105927.22991-1-chris@chris-wilson.co.uk Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Manasi Navare [Tue, 3 Oct 2017 23:37:25 +0000 (16:37 -0700)]
drm/i915/edp: Increase the T12 delay quirk to 1300ms
For this specific PCI device, the eDP panel requires a higher panel
power cycle delay of 1300ms where the minimum spec requirement of panel
power cycle delay is 500ms. This fix in combination with correct
timestamp at which we get the panel power off time fixes the dP AUX CH
timeouts seen on various IGT tests.
Fixes: c99a259b4b5192ba ("drm/i915/edp: Add a T12 panel delay quirk to fix
DP AUX CH timeouts")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101144
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101518 Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Ville Syrjala <ville.syrjala@linux.intel.com> Signed-off-by: Manasi Navare <manasi.d.navare@intel.com> Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/1507073845-13420-2-git-send-email-manasi.d.navare@intel.com
Manasi Navare [Wed, 4 Oct 2017 16:48:26 +0000 (09:48 -0700)]
drm/i915/edp: Get the Panel Power Off timestamp after panel is off
Kernel stores the time in jiffies at which the eDP panel is turned
off. This should be obtained after the panel is off (after the
wait_panel_off). When we next attempt to turn the panel on, we use the
difference between the timestamp at which we want to turn the panel on
and timestamp at which panel was turned off to ensure that this is equal
to panel power cycle delay and if not we wait for the remaining
time. Not waiting for the panel power cycle delay can cause the panel to
not turn on giving rise to AUX timeouts for the attempted AUX
transactions.
v2:
* Separate lines for bugzilla (Jani Nikula)
* Suggested by tag (Daniel Vetter)
Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: stable@vger.kernel.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101518
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=101144 Suggested-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Manasi Navare <manasi.d.navare@intel.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Jani Nikula <jani.nikula@linux.intel.com> Signed-off-by: Jani Nikula <jani.nikula@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/1507135706-17147-1-git-send-email-manasi.d.navare@intel.com
drm/i915/mst: Use MST sideband message transactions for dpms control
Use the POWER_DOWN_PHY and POWER_UP_PHY sideband message transactions to
set power states for downstream sinks. Apart from giving us the ability
to set power state for individual sinks, this fixes the below test for
me.
Chris Wilson [Tue, 3 Oct 2017 20:34:53 +0000 (21:34 +0100)]
drm/i915/scheduler: Support user-defined priorities
Use a priority stored in the context as the initial value when
submitting a request. This allows us to change the default priority on a
per-context basis, allowing different contexts to be favoured with GPU
time at the expense of lower importance work. The user can adjust the
context's priority via I915_CONTEXT_PARAM_PRIORITY, with more positive
values being higher priority (they will be serviced earlier, after their
dependencies have been resolved). Any prerequisite work for an execbuf
will have its priority raised to match the new request as required.
Normal users can specify any value in the range of -1023 to 0 [default],
i.e. they can reduce the priority of their workloads (and temporarily
boost it back to normal if so desired).
Privileged users can specify any value in the range of -1023 to 1023,
[default is 0], i.e. they can raise their priority above all overs and
so potentially starve the system.
Note that the existing schedulers are not fair, nor load balancing, the
execution is strictly by priority on a first-come, first-served basis,
and the driver may choose to boost some requests above the range
available to users.
This priority was originally based around nice(2), but evolved to allow
clients to adjust their priority within a small range, and allow for a
privileged high priority range.
For example, this can be used to implement EGL_IMG_context_priority
https://www.khronos.org/registry/egl/extensions/IMG/EGL_IMG_context_priority.txt
EGL_CONTEXT_PRIORITY_LEVEL_IMG determines the priority level of
the context to be created. This attribute is a hint, as an
implementation may not support multiple contexts at some
priority levels and system policy may limit access to high
priority contexts to appropriate system privilege level. The
default value for EGL_CONTEXT_PRIORITY_LEVEL_IMG is
EGL_CONTEXT_PRIORITY_MEDIUM_IMG."
so we can map
PRIORITY_HIGH -> 1023 [privileged, will failback to 0]
PRIORITY_MED -> 0 [default]
PRIORITY_LOW -> -1023
They also map onto the priorities used by VkQueue (and a VkQueue is
essentially a timeline, our i915_gem_context under full-ppgtt).
v2: s/CAP_SYS_ADMIN/CAP_SYS_NICE/
v3: Report min/max user priorities as defines in the uapi, and rebase
internal priorities on the exposed values.
Chris Wilson [Tue, 3 Oct 2017 20:34:52 +0000 (21:34 +0100)]
drm/i915/execlists: Preemption!
When we write to ELSP, it triggers a context preemption at the earliest
arbitration point (3DPRIMITIVE, some PIPECONTROLs, a few other
operations and the explicit MI_ARB_CHECK). If this is to the same
context, it triggers a LITE_RESTORE where the RING_TAIL is merely
updated (used currently to chain requests from the same context
together, avoiding bubbles). However, if it is to a different context, a
full context-switch is performed and it will start to execute the new
context saving the image of the old for later execution.
Previously we avoided preemption by only submitting a new context when
the old was idle. But now we wish embrace it, and if the new request has
a higher priority than the currently executing request, we write to the
ELSP regardless, thus triggering preemption, but we tell the GPU to
switch to our special preemption context (not the target). In the
context-switch interrupt handler, we know that the previous contexts
have finished execution and so can unwind all the incomplete requests
and compute the new highest priority request to execute.
It would be feasible to avoid the switch-to-idle intermediate by
programming the ELSP with the target context. The difficulty is in
tracking which request that should be whilst maintaining the dependency
change, the error comes in with coalesced requests. As we only track the
most recent request and its priority, we may run into the issue of being
tricked in preempting a high priority request that was followed by a
low priority request from the same context (e.g. for PI); worse still
that earlier request may be our own dependency and the order then broken
by preemption. By injecting the switch-to-idle and then recomputing the
priority queue, we avoid the issue with tracking in-flight coalesced
requests. Having tried the preempt-to-busy approach, and failed to find
a way around the coalesced priority issue, Michal's original proposal to
inject an idle context (based on handling GuC preemption) succeeds.
The current heuristic for deciding when to preempt are only if the new
request is of higher priority, and has the privileged priority of
greater than 0. Note that the scheduler remains unfair!
v2: Disable for gen8 (bdw/bsw) as we need additional w/a for GPGPU.
Since, the feature is now conditional and not always available when we
have a scheduler, make it known via the HAS_SCHEDULER GETPARAM (now a
capability mask).
v3: Stylistic tweaks.
v4: Appease Joonas with a snippet of kerneldoc, only to fuel to fire of
the preempt vs preempting debate.
Suggested-by: Michal Winiarski <michal.winiarski@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michal Winiarski <michal.winiarski@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Ben Widawsky <benjamin.widawsky@intel.com> Cc: Zhenyu Wang <zhenyuw@linux.intel.com> Cc: Zhi Wang <zhi.a.wang@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171003203453.15692-8-chris@chris-wilson.co.uk
Chris Wilson [Tue, 3 Oct 2017 20:34:51 +0000 (21:34 +0100)]
drm/i915: Expand I915_PARAM_HAS_SCHEDULER into a capability bitmask
In the next few patches, we wish to enable different features for the
scheduler, some which may subtlety change ABI (e.g. allow requests to be
reordered under different circumstances). So we need to make sure
userspace is cognizant of the changes (if they care), by which we employ
the usual method of a GETPARAM. We already have an
I915_PARAM_HAS_SCHEDULER (which notes the existing ability to reorder
requests to avoid bubbles), and now we wish to extend that to be a
bitmask to describe the different capabilities implemented.
Chris Wilson [Tue, 3 Oct 2017 20:34:50 +0000 (21:34 +0100)]
drm/i915/execlists: Keep request->priority for its lifetime
With preemption, we will want to "unsubmit" a request, taking it back
from the hw and returning it to the priority sorted execution list. In
order to know where to insert it into that list, we need to remember
its adjust priority (which may change even as it was being executed).
This also affects reset for execlists as we are now unsubmitting the
requests following the reset (rather than directly writing the ELSP for
the inflight contexts). This turns reset into an accidental preemption
point, as after the reset we may choose a different pair of contexts to
submit to hw.
GuC is not updated as this series doesn't add preemption to the GuC
submission, and so it can keep benefiting from the early pruning of the
DFS inside execlists_schedule() for a little longer. We also need to
find a way of reducing the cost of that DFS...
Chris Wilson [Tue, 3 Oct 2017 20:34:48 +0000 (21:34 +0100)]
drm/i915: Introduce a preempt context
Add another perma-pinned context for using for preemption at any time.
We cannot just reuse the existing kernel context, as first and foremost
we need to ensure that we can preempt the kernel context itself, so
require a distinct context id. Similar to the kernel context, we may
want to interrupt execution and switch to the preempt context at any
time, and so it needs to be permanently pinned and available.
To compensate for yet another permanent allocation, we shrink the
existing context and the new context by reducing their ringbuffer to the
minimum.
v2: Assert that we never allocate a request from the preemption context.
v3: Limit perma-pin to engines that may preempt.
v4: Onion cleanup for early driver death
v5: Onion ordering in main driver cleanup as well.
drm/i915/preempt: Default to disabled mid-command preemption levels
Supporting fine-granularity preemption levels may require changes in
userspace batch buffer programming. Therefore, we need to fallback to
safe default values, rather that use hardware defaults. Userspace is
still able to enable fine-granularity, since we're whitelisting the
register controlling it in WaEnablePreemptionGranularityControlByUMD.
v2: Extend w/a to cover Cannonlake
v3: Fix commentary to include both fake w/a names.
Chris Wilson [Wed, 4 Oct 2017 12:41:53 +0000 (13:41 +0100)]
drm/i915: Remove WA_(SET|CLR)_BIT
These macros are of dubious merit when coupled with the per-context w/a
set. Instead of tweaking the value in the context, they tweak the value
based on the mmio at the time of recording; they are almost by
definition not per-context! Having removed the last users, remove the
macros to avoid temptation in the future.
Chris Wilson [Wed, 4 Oct 2017 12:41:52 +0000 (13:41 +0100)]
drm/i915: Move MMCD_MISC_CTRL from context w/a to standard
Looking at gem_workarounds shows us that MMCD_MISC_CTRL is not restored
following a suspend-resume cycle. This implies that MMCD_MISC_CTRL is
not stored in the context, but is an ordinary register w/a that we need to
restore during init_hw.
Oscar Mateo [Thu, 28 Sep 2017 22:40:39 +0000 (15:40 -0700)]
drm/i915: Transform whitelisting WAs into a simple reg write
RING_FORCE_TO_NONPRIV registers do not live in the logical context. They are simply
global privileged MMIO registers that happen to be powercontext saved and restored
(meaning only they can survive RC6). Therefore, there is absolutely no need to save
them so that they can be restored everytime we create a new logical context.
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Oscar Mateo <oscar.mateo@intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/1506638439-6903-1-git-send-email-oscar.mateo@intel.com Acked-by: Michel Thierry <michel.thierry@intel.com> Tested-by: Chris Wilson <chris@chris-wilson.co.uk> #bxt Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
drm/i915: Unset legacy_cursor_update early in intel_atomic_commit, v3.
Commit b44d5c0c105a ("drm/i915: Always wait for flip_done, v2.") removed
the call to wait_for_vblanks and replaced it with flip_done.
Unfortunately legacy_cursor_update was unset too late, and the
replacement call drm_atomic_helper_wait_for_flip_done() was
a noop. Make sure that its unset before setup_commit() is
called to fix this issue.
Changes since v1:
- Force vblank wait for watermarks not yet converted to atomic too. (Ville)
- Use for_each_new_intel_crtc_in_state. (Ville)
Changes since v2:
- Move the optimization to a separate commit. (Ville)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Fixes: b44d5c0c105a ("drm/i915: Always wait for flip_done, v2.")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102675
Testcase: kms_cursor_crc Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Jani Nikula <jani.nikula@linux.intel.com> Reported-by: Marta Löfstedt <marta.lofstedt@intel.com> Cc: Marta Löfstedt <marta.lofstedt@intel.com> Tested-by: Marta Löfstedt <marta.lofstedt@intel.com> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170919121419.13708-1-maarten.lankhorst@linux.intel.com
Rodrigo Vivi [Tue, 3 Oct 2017 06:36:51 +0000 (23:36 -0700)]
drm/i915: Organize GEN features inheritance.
As Chris noticed the current organization is confusing
and inheritance is not clear.
So, let's split it in GEN<n>_FEATURES <cdn>_PLATFORM
where new GEN inherit features from previous gens and
Platforms only use gen features plus what ever is specific
for that platform and shouldn't be passed on.
Rodrigo Vivi [Tue, 3 Oct 2017 06:36:50 +0000 (23:36 -0700)]
drm/i915/skl: Fix has_ipc on skl and document WaDisableIPC.
According to Spec for SKL+: "Isochronous Priority Control.
If enabled, Display sends demoted requests once the transition
watermark is reached. If transition watermark is not enabled,
Display sends demoted requests when the display buffer is full."
The commit 'e57f1c02155f ("drm/i915/gen9+: Add has_ipc flag in
device info structure")' introduced that as gen9+ but missing many
SKL Skus.
I believe the reason for that is Spec also mentions workarounds for
SKL-ALL: "IPC (Isoch Priority Control) may cause underflows
WA: Do not enable IPC in register ARB_CTL2"
It seems lame to add the feature and forever disable it,
but it will avoid a mistake of enabling it when we are reorganizing
the feature definitions on i915_pci.c later.
It will also allow us to probably extend that workaround for
other platforms.
Imre Deak [Mon, 2 Oct 2017 13:53:07 +0000 (16:53 +0300)]
drm/i915: Fix DDI PHY init if it was already on
The common lane power down flag of a DPIO PHY has a funky semantic:
after the initial enabling of the PHY (so from a disabled state) this
flag will be clear. It will be set only after the PHY will be used for
the first time (for instance due to enabling the corresponding pipe) and
then become unused (due to disabling the pipe). During the initial PHY
enablement we don't know which of the above phases we are in, so move
the check for the flag where this is known, the HW readout code. This is
where the rest of lane power down status checks are done anyway.
This fixes at least a problem on GLK where after module reloading, the
common lane power down flag of PHY1 is set, but the PHY is actually
powered-on and properly set up. The GRC readout code for other PHYs will
hence think that PHY1 is not powered initially and disable it after the
GRC readout. This will cause the AUX power well related to PHY1 to get
disabled in a stuck state, timing out when we try to enable it later.
Cc: Ville Syrjala <ville.syrjala@linux.intel.com> Fixes: e93da0a0137b ("drm/i915/bxt: Sanitiy check the PHY lane power down status")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102777 Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20171002135307.26117-1-imre.deak@intel.com
On GLK and CNL enabling a pipe with its pipe scaler enabled will result
in a FIFO underrun. This happens only once after driver loading or
system/runtime resume, more specifically after power well 1 gets
enabled; subsequent modesets seem to be free of underruns. The BSpec
workaround for this is to disable the pipe scaler clock gating for the
duration of modeset. Based on my tests disabling clock gating must be
done before enabling pipe scaling and we can re-enable it after the pipe
is enabled and one vblank has passed.
For consistency I also checked if plane scaling would cause the same
problem, but that doesn't seem to trigger this problem.
The patch is based on an earlier version from Ander.
v2 (Rodrigo):
- Set also CLKGATE_DIS_PSL bits 8 and 9.
- Add also the BSpec workaround ID.
David Weinehall [Tue, 8 Aug 2017 10:09:52 +0000 (13:09 +0300)]
drm/i915: Add has_psr-flag to gen9lp
While testing Jim Bride's latest batch of PSR patches I noticed
that gen9lp doesn't include the has_psr flag, and that our GLK
system thus reported PSR as unsupported.
Imre Deak [Thu, 28 Sep 2017 10:06:24 +0000 (13:06 +0300)]
drm/i915/gen8+: Init/reset display interrupts only if i915 IRQs are enabled
Only init / reset the display interrupts during power well enabling /
disabling if the i915 interrupts are enabled. So far we did the
init / reset during driver loading / resuming too, where
initialization / enabling of the i915 interrupts happens only at a later
point. This didn't cause a problem due to GEN8_MASTER_IRQ_CONTROL being
cleared, but triggered gen3_assert_iir_is_zero() in GEN8_IRQ_INIT_NDX().
Imre Deak [Wed, 16 Aug 2017 14:46:07 +0000 (17:46 +0300)]
drm/i915/gen9+: Set same power state before hibernation image save/restore
Atm, on GEN9 big core platforms before saving the hibernation image we
uninitialize the display, disabling power wells manually, while before
restoring the image we keep things powered (letting HW/DMC power down
things as needed). The state mismatch will trigger the following error:
DC state mismatch (0x0 -> 0x2)
While the restore handler knows how to initialize the display from an
unknown state (due to a different loader kernel or not having i915
loaded in the loader kernel) we should still use the same state for
consistency before image saving and restoring. Do this by uniniting the
display before restoring the image too.
Bugzilla: https://bugs.freedesktop.org/attachment.cgi?id=133376 Reported-and-tested-by: Wang Wendy <wendy.wang@intel.com> Reported-and-tested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Wang Wendy <wendy.wang@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Ville Syrjala <ville.syrjala@linux.intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170816144607.9935-1-imre.deak@intel.com
Chris Wilson [Thu, 28 Sep 2017 19:39:00 +0000 (20:39 +0100)]
drm/i915: Give the invalid priority a magic name
We use INT_MIN to denote the priority of a request that has not been
submitted to the scheduler; we treat INT_MIN as an invalid priority and
initialise the request to it. Give the value a name so it stands out.
Chris Wilson [Thu, 28 Sep 2017 19:38:59 +0000 (20:38 +0100)]
drm/i915/execlists: Move request unwinding to a separate function
In the future, we will want to unwind requests following a preemption
point. This requires the same steps as for unwinding upon a reset, so
extract the existing code to a separate function for later use.
drm/i915: Move i915_gem_restore_fences to i915_gem_resume
i915_gem_restore_fences is GEM resumption task hence it is moved to
i915_gem_resume from i915_restore_state.
Signed-off-by: Sagar Arun Kamble <sagar.a.kamble@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Michał Winiarski <michal.winiarski@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/1506661116-12106-1-git-send-email-sagar.a.kamble@intel.com Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>