We were using the last_fence to track the last request that used this
vma that might be interpreted by a fence register and forced ourselves
to wait for this request before modifying any fence register that
overlapped our vma. Due to requirement that we need to track any XY_BLT
command, linear or tiled, this in effect meant that we have to track the
vma for its active lifespan anyway, so we can forgo the explicit
last_fence tracking and just use the whole vma->active.
Another solution would be to pipeline the register updates, and would
help resolve some long running stalls for gen3 (but only gen 2 and 3!)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190812174804.26180-1-chris@chris-wilson.co.uk
The shrinker cannot touch objects used by the contexts (logical state
and ring). Currently we mark those as "pin_global" to let the shrinker
skip over them, however, if we remove them from the shrinker lists
entirely, we don't event have to include them in our shrink accounting.
By keeping the unshrinkable objects in our shrinker tracking, we report
a large number of objects available to be shrunk, and leave the shrinker
deeply unsatisfied when we fail to reclaim those. The shrinker will
persist in trying to reclaim the unavailable objects, forcing the system
into a livelock (not even hitting the dread oomkiller).
v2: Extend unshrinkable protection for perma-pinned scratch and guc
allocations (Tvrtko)
v3: Notice that we should be pinned when marking unshrinkable and so the
link cannot be empty; merge duplicate paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190802212137.22207-1-chris@chris-wilson.co.uk
As the fence registers only apply to regions inside the GGTT is makes
more sense that we track these as part of the i915_ggtt and not the
general mm. In the next patch, we will then pull the register locking
underneath the i915_ggtt.mutex.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190613073254.24048-1-chris@chris-wilson.co.uk
Use i915_gem_object_lock() to guard the LUT and active reference to
allow us to break free of struct_mutex for handling GEM_CLOSE.
Testcase: igt/gem_close_race
Testcase: igt/gem_exec_parallel
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190606112320.9704-1-chris@chris-wilson.co.uk
Use the per-object local lock to control the cache domain of the
individual GEM objects, not struct_mutex. This is a huge leap forward
for us in terms of object-level synchronisation; execbuffers are
coordinated using the ww_mutex and pread/pwrite is finally fully
serialised again.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190528092956.14910-10-chris@chris-wilson.co.uk
Currently the code for manipulating the pages on an object is still
residing in i915_gem.c, move it to i915_gem_object.c
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190528092956.14910-3-chris@chris-wilson.co.uk
To overcome display engine stride limits we'll want to remap the
pages in the GTT. To that end we need a new gtt_view type which
is just like the "rotated" type except not rotated.
v2: Use intel_remapped_plane_info base type
s/unused/unused_mbz/ (Chris)
Separate BUILD_BUG_ON()s (Chris)
Use I915_GTT_PAGE_SIZE (Chris)
v3: Use i915_gem_object_get_dma_address() (Chris)
Trim the sg (Tvrtko)
v4: Actually trim this time. Limit the max length
to one row of pages to keep things simple
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190509122159.24376-2-ville.syrjala@linux.intel.com
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Rather than manually add every new global into each hook, use
i915_global_register() function and keep a list of registered globals to
invoke instead.
However, I haven't found a way for random drivers to add an .init table
to avoid having to manually add ourselves to i915_globals_init() each
time.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20190305213830.18094-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Looking forward, we need to break the struct_mutex dependency on
i915_gem_active. In the meantime, external use of i915_gem_active is
quite beguiling, little do new users suspect that it implies a barrier
as each request it tracks must be ordered wrt the previous one. As one
of many, it can be used to track activity across multiple timelines, a
shared fence, which fits our unordered request submission much better. We
need to steer external users away from the singular, exclusive fence
imposed by i915_gem_active to i915_active instead. As part of that
process, we move i915_gem_active out of i915_request.c into
i915_active.c to start separating the two concepts, and rename it to
i915_active_request (both to tie it to the concept of tracking just one
request, and to give it a longer, less appealing name).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190205130005.2807-5-chris@chris-wilson.co.uk
We currently track GPU memory usage inside VMA, such that we never
release memory used by the GPU until after it has finished accessing it.
However, we may want to track other resources aside from VMA, or we may
want to split a VMA into multiple independent regions and track each
separately. For this purpose, generalise our request tracking (akin to
struct reservation_object) so that we can embed it into other objects.
v2: Tweak error handling during selftest setup.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190205130005.2807-2-chris@chris-wilson.co.uk
Previously we only accommodated having a vma pinned by a small number of
users, with the maximum being pinned for use by the display engine. As
such, we used a small bitfield only large enough to allow the vma to
be pinned twice (for back/front buffers) in each scanout plane. Keeping
the maximum permissible pin_count small allows us to quickly catch a
potential leak. However, as we want to split a 4096B page into 64
different cachelines and pin each cacheline for use by a different
timeline, we will exceed the current maximum permissible vma->pin_count
and so time has come to enlarge it.
Whilst we are here, try to pull together the similar bits:
Address/layout specification:
- bias, mappable, zone_4g: address limit specifiers
- fixed: address override, limits still apply though
- high: not strictly an address limit, but an address direction to search
Search controls:
- nonblock, nonfault, noevict
v2: Rewrite the guideline comment on bit consumption.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: John Harrison <john.C.Harrison@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190128181812.22804-2-chris@chris-wilson.co.uk
Remove the struct_mutex requirement for looking up the vma for an
object.
v2: Highlight how the race for duplicate vma creation is resolved on
reacquiring the lock with a short comment.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190128102356.15037-3-chris@chris-wilson.co.uk
gcc is too smart for us and doesn't evaluate BUILD_BUG_ON()s in
unused static inlines. Collect them up in one static inline and
actually call it to make sure gcc sees it.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180828133723.18505-1-ville.syrjala@linux.intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Removing the pin bias from GuC allows us to not check for GuC every time
we pin a context, which fixes the assertion error on unresolved GuC
platform default in mock contexts selftest.
It also seems that we were using uninitialized WOPCM variables when
setting the GuC pin bias. The pin bias has to be set after the WOPCM,
but before the call to i915_gem_contexts_init where the first contexts
are pinned.
v2:
This also makes it so that there's no need to set GuC variables from
within the WOPCM init function or to move the WOPCM init, while keeping
the correct initialization order. Also for mock tests the pin bias is
left at 0 and we make sure that the pin bias with GuC will not be
smaller than without GuC.
v3:
Avoid unused i915 in intel_guc_ggtt_offset if debug is disabled.
v4:
Squash with WOPCM init reordering.
Moved the i915_ggtt_pin_bias helper to this patch, and made some
functions use it instead of directly dereferencing i915->ggtt.
v5:
Since we now don't use wopcm.guc.base for the pin bias there's no need to
validate it. It also has already been verified in WOPCM init.
v6:
Deleted the now unnecessarily introduced includes from previous versions.
Dropped naming changes from dev_priv to i915 for better patch readability.
v7:
Changed some comments to make more sense in the context they're in.
v8:
Moved and renamed the function which now returns the wopcm.guc.size to
intel_guc.c:intel_guc_reserved_gtt_size to avoid any possible confusion
with the pin_bias in ggtt, which should be used for pinning.
Fixed patch not applying or the most recent upstream.
Fixes: f7dc0157e4 ("drm/i915/uc: Fetch GuC/HuC firmwares from guc/huc specific init")
Testcase: igt/drv_selftest/mock_contexts #GuC
Signed-off-by: Jakub Bartmiński <jakub.bartminski@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20180727141148.30874-3-jakub.bartminski@intel.com
A reasonably common operation is to pin the map of the vma alongside the
vma itself for the lifetime of the vma, and so release both pins at the
same time as destroying the vma. It is common enough to pull into the
release function, making that central function more attractive to a
couple of other callsites.
The continual ulterior motive is to sweep over errors on module load
aborting...
Testcase: igt/drv_module_reload/basic-reload-inject
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180721125037.20127-1-chris@chris-wilson.co.uk
Using a VMA on more than one timeline concurrently is the exception
rather than the rule (using it concurrently on multiple engines). As we
expect to only use one active tracker, store the most recently used
tracker inside the i915_vma itself and only fallback to the rbtree if
we need a second or more concurrent active trackers.
v2: Comments on how we overwrite any existing last_active cache.
v3: __list_del_entry() before list_replace_init() is confusing and, much
more important, entirely redundant.
v4: Note that both last_active and the rbtree may be simultaneously
tracking this timeline, albeit with different requests, and so the vma
may be retired twice for the same timeline.
v5: No, that list_del is required!
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180706123157.9645-1-chris@chris-wilson.co.uk
In the next patch, we will want to be able to use more flexible request
timelines that can hop between engines. From the vma pov, we can then
not rely on the binding of this request to an engine and so can not
ensure that different requests are ordered through a per-engine
timeline, and so we must track activity of all timelines. (We track
activity on the vma itself to prevent unbinding from HW before the HW
has finished accessing it.)
v2: Switch to a rbtree for 32b safety (since using u64 as a radixtree
index is fraught with aliasing of unsigned longs).
v3: s/lookup_active/active_instance/ because we can never agree on names
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180706103947.15919-5-chris@chris-wilson.co.uk
Currently all page directories are bound at creation using an
unevictable node in the GGTT. This severely limits us as we cannot
remove any inactive ppgtt for new contexts, or under aperture pressure.
To fix this we need to make the page directory into a first class and
unbindable vma. Hence, the creation of a custom vma to wrap the page
directory as opposed to a GEM object.
In this patch, we leave the page directories pinned upon creation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180612120446.13901-1-chris@chris-wilson.co.uk
To allow for future non-object backed vma, we need to be able to
specialise the callbacks for binding, et al, the vma. For example,
instead of calling vma->vm->bind_vma(), we now call
vma->ops->bind_vma(). This gives us the opportunity to later override the
operation for a custom vma.
v2: flip order of unbind/bind
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607154047.9171-2-chris@chris-wilson.co.uk
In order to allow ourselves to use VMA to wrap other entities other than
GEM objects, we need to allow for the vma->obj backpointer to be NULL.
In most cases, we know we are operating on a GEM object and its vma, but
we need the core code (such as i915_vma_pin/insert/bind/unbind) to work
regardless of the innards.
The remaining eyesore here is vma->obj->cache_level and related (but
less of an issue) vma->obj->gt_ro. With a bit of care we should mirror
those on the vma itself.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180607154047.9171-1-chris@chris-wilson.co.uk
When userspace is passing around swapbuffers using DRI, we frequently
have to open and close the same object in the foreign address space.
This shows itself as the same object being rebound at roughly 30fps
(with a second object also being rebound at 30fps), which involves us
having to rewrite the page tables and maintain the drm_mm range manager
every time.
However, since the object still exists and it is only the local handle
that disappears, if we are lazy and do not unbind the VMA immediately
when the local user closes the object but defer it until the GPU is
idle, then we can reuse the same VMA binding. We still have to be
careful to mark the handle and lookup tables as closed to maintain the
uABI, just allowing the underlying VMA to be resurrected if the user is
able to access the same object from the same context again.
If the object itself is destroyed (neither userspace keeping a handle to
it), the VMA will be reaped immediately as usual.
In the future, this will be even more useful as instantiating a new VMA
for use on the GPU will become heavier. A nuisance indeed, so nip it in
the bud.
v2: s/__i915_vma_final_close/i915_vma_destroy/ etc.
v3: Leave a hint as to why we deferred the unbind on close.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180503195115.22309-1-chris@chris-wilson.co.uk
We want to de-emphasize the link between the request (dependency,
execution and fence tracking) from GEM and so rename the struct from
drm_i915_gem_request to i915_request. That is we may implement the GEM
user interface on top of requests, but they are an abstraction for
tracking execution rather than an implementation detail of GEM. (Since
they are not tied to HW, we keep the i915 prefix as opposed to intel.)
In short, the spatch:
@@
@@
- struct drm_i915_gem_request
+ struct i915_request
A corollary to contracting the type name, we also harmonise on using
'rq' shorthand for local variables where space if of the essence and
repetition makes 'request' unwieldy. For globals and struct members,
'request' is still much preferred for its clarity.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180221095636.6649-1-chris@chris-wilson.co.uk
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Acked-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
In quite a few places, we have a list iteration over the vma on an
object that only want to inspect GGTT vma. By construction, these are
placed at the start of the list, so we have copied that knowledge into
many callsites. Pull that knowledge back to i915_vma.h and provide a
for_each_ggtt_vma() to tidy up the code.
v2: Add a backreference from vma_create() to remind ourselves why we put
ggtt vma at the head of the obj->vma_list (and ppgtt vma at the tail).
v3: Fixup s/vma/V/
Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20171207211407.31549-1-chris@chris-wilson.co.uk
As writes through the GTT and GGTT PTE updates do not share the same
path, they are not strictly ordered and so we must explicitly flush the
indirect writes prior to modifying the PTE. We do track outstanding GGTT
writes on the object itself, but since the object may have multiple GGTT
vma, that is overly coarse as we can track and flush individual vma as
required.
Whilst here, update the GGTT flushing behaviour for Cannonlake.
v2: Hard-code ring offset to allow use during unload (after RCS may have
been freed, or never existed!)
References: https://bugs.freedesktop.org/show_bug.cgi?id=104002
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171206124914.19960-2-chris@chris-wilson.co.uk
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().
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171009084401.29090-3-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
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.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171009084401.29090-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
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.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171009084401.29090-1-chris@chris-wilson.co.uk
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
v4: prefer upper_32_bits()
v5: assign vma->page_sizes = vma->obj->page_sizes directly
prefer sizeof(vma->page_sizes)
v6: fixup checking of end to exclude GGTT (which are assumed to be
limited to 4G).
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>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171006145041.21673-9-matthew.auld@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20171006221833.32439-8-chris@chris-wilson.co.uk
By using drm_gem_flink/drm_gem_open on an object using the same fd, it
is possible for a client to create multiple handles pointing to the same
object (tied to the same contexts and VMA), as exemplified by
igt::gem_handle_to_libdrm_bo(). Since this duplication has been possible
since forever, we cannot assume that the handle:(fpriv, object) is
unique and so must handle the multiple users of a single VMA.
v2: Added commentary noise.
Testcase: igt/gem_close
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102355
Fixes: d1b48c1e71 ("drm/i915: Replace execbuf vma ht with an idr")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170822110517.22277-3-chris@chris-wilson.co.uk
Tested-by: Marta Lofstedt <marta.lofstedt@intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
This was the competing idea long ago, but it was only with the rewrite
of the idr as an radixtree and using the radixtree directly ourselves,
along with the realisation that we can store the vma directly in the
radixtree and only need a list for the reverse mapping, that made the
patch performant enough to displace using a hashtable. Though the vma ht
is fast and doesn't require any extra allocation (as we can embed the node
inside the vma), it does require a thread for resizing and serialization
and will have the occasional slow lookup. That is hairy enough to
investigate alternatives and favour them if equivalent in peak performance.
One advantage of allocating an indirection entry is that we can support a
single shared bo between many clients, something that was done on a
first-come first-serve basis for shared GGTT vma previously. To offset
the extra allocations, we create yet another kmem_cache for them.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170816085210.4199-5-chris@chris-wilson.co.uk
When userspace is doing most of the work, avoiding relocs (using
NO_RELOC) and opting out of implicit synchronisation (using ASYNC), we
still spend a lot of time processing the arrays in execbuf, even though
we now should have nothing to do most of the time. One issue that
becomes readily apparent in profiling anv is that iterating over the
large execobj[] is unfriendly to the loop prefetchers of the CPU and it
much prefers iterating over a pair of arrays rather than one big array.
v2: Clear vma[] on construction to handle errors during vma lookup
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20170816085210.4199-3-chris@chris-wilson.co.uk
After we detect a i915_vma pin overflow, we call __i915_vma_unpin to
cleanup. However, on an overflow the pin_count bitfield will be zero,
triggering an assertion, even though we the intention is to merely warn
and report the error back to the user (as historically the culprit has
be a leak in the display code).
Fixes: 20dfbde463 ("drm/i915: Wrap vma->pin_count accessors with small inline helpers")
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/20170721145037.25105-2-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
During execbuf, a mandatory step is that we add this request (this
fence) to each object's reservation_object. Inside execbuf, we track the
vma, and to add the fence to the reservation_object then means having to
first chase the obj, incurring another cache miss. We can reduce the
number of cache misses by stashing a pointer to the reservation_object
in the vma itself.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170616140525.6394-1-chris@chris-wilson.co.uk
The major scaling bottleneck in execbuffer is the processing of the
execobjects. Creating an auxiliary list is inefficient when compared to
using the execobject array we already have allocated.
Reservation is then split into phases. As we lookup up the VMA, we
try and bind it back into active location. Only if that fails, do we add
it to the unbound list for phase 2. In phase 2, we try and add all those
objects that could not fit into their previous location, with fallback
to retrying all objects and evicting the VM in case of severe
fragmentation. (This is the same as before, except that phase 1 is now
done inline with looking up the VMA to avoid an iteration over the
execobject array. In the ideal case, we eliminate the separate reservation
phase). During the reservation phase, we only evict from the VM between
passes (rather than currently as we try to fit every new VMA). In
testing with Unreal Engine's Atlantis demo which stresses the eviction
logic on gen7 class hardware, this speed up the framerate by a factor of
2.
The second loop amalgamation is between move_to_gpu and move_to_active.
As we always submit the request, even if incomplete, we can use the
current request to track active VMA as we perform the flushes and
synchronisation required.
The next big advancement is to avoid copying back to the user any
execobjects and relocations that are not changed.
v2: Add a Theory of Operation spiel.
v3: Fall back to slow relocations in preparation for flushing userptrs.
v4: Document struct members, factor out eb_validate_vma(), add a few
more comments to explain some magic and hide other magic behind macros.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
The advent of full-ppgtt lead to an extra indirection between the object
and its binding. That extra indirection has a noticeable impact on how
fast we can convert from the user handles to our internal vma for
execbuffer. In order to bypass the extra indirection, we use a
resizable hashtable to jump from the object to the per-ctx vma.
rhashtable was considered but we don't need the online resizing feature
and the extra complexity proved to undermine its usefulness. Instead, we
simply reallocate the hastable on demand in a background task and
serialize it before iterating.
In non-full-ppgtt modes, multiple files and multiple contexts can share
the same vma. This leads to having multiple possible handle->vma links,
so we only use the first to establish the fast path. The majority of
buffers are not shared and so we should still be able to realise
speedups with multiple clients.
v2: Prettier names, more magic.
v3: Many style tweaks, most notably hiding the misuse of execobj[].rsvd2
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
i915_vma_destroy() is now not used outside of i915_vma.c so we can
remove the export and make the function static.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170616123508.12673-1-chris@chris-wilson.co.uk
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Currently the vma has one link member that is used for both holding its
place in the execbuf reservation list, and in any eviction list. This
dual property is quite tricky and error prone.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170615081435.17699-3-chris@chris-wilson.co.uk
With the introduce of i915_vma_instance() for obtaining the VMA
singleton for a (obj, vm, view) tuple, we can remove the
i915_vma_create() in favour of a single entry point. We do incur a
lookup onto an empty tree, but the i915_vma_create() were being called
infrequently and during initialisation, so the small overhead is
negligible.
v2: Drop the i915_ prefix from the now static vma_create() function
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170116152131.18089-4-chris@chris-wilson.co.uk
Whilst writing testcases to exercise the VMA API, some oddities came to
light, such as i915_gem_obj_lookup_or_create(). Joonas suggested
i915_vma_instance() as a neat replacement, so rename them, move them to
i915_vma.c and add some kerneldoc as a sugary bonus.
s/i915_gem_obj_to_vma/i915_vma_lookup/
s/i915_gem_obj_lookup_or_create_vma/i915_vma_instance/
Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170116152131.18089-2-chris@chris-wilson.co.uk
Reading the ggtt_views is much more pleasant without the extra
characters from specifying the union (i.e. ggtt_view.partial rather than
ggtt_view.params.partial). To make this work inside i915_vma_compare()
with only a single memcmp requires us to ensure that there are no
uninitialised bytes within each branch of the union (we make sure the
structs are packed) and we need to store the size of each branch.
v4: Rewrite changelog and add comments explaining the assert.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170114002827.31315-5-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
In preparation for the next patch to convert to using an anonymous union
and leaving the excess bytes in the union uninitialised, we first need
to make sure we do not compare using those uninitialised bytes. We also
want to preserve the compactness of the code, avoiding a second call to
memcmp or introducing a switch, so we take advantage of using the type
as an encoded size (as well as a unique identifier for each type of view).
v2: Add the rationale for why we encode size into ggtt_view.type as a
comment before the memcmp()
v3: Use a switch to also assert that no two i915_ggtt_view_type have the same
value.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170114002827.31315-3-chris@chris-wilson.co.uk
We currently check after the slow path that the vma is bound correctly,
but we don't currently check after the fast path. This is important in
case we accidentally take the fast path and leave the vma misplaced.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170111210937.29252-27-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
The fence size/alignment is a combination of the vma size plus object
tiling parameters. Those parameters are rarely changed, making the fence
size/alignemnt roughly constant for the lifetime of the VMA. We can
simplify subsequent calculations by precalculating the size/alignment
required for GGTT vma taking fencing into account (with an update if we
do change the tiling or stride).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170109161613.11881-4-chris@chris-wilson.co.uk
Cast VM pointers before substraction to save the compiler
doing a smart one which includes multiplication.
v2: Only keep the first optimisation and prettify it. (Chris Wilson)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1481639847-9214-1-git-send-email-tvrtko.ursulin@linux.intel.com
99% of the time we access i915_address_space->dev we want the i915
device and not the drm device, so let's store the drm_i915_private
backpointer instead. The only real complication here are the inlines
in i915_vma.h where drm_i915_private is not yet defined and so we have
to choose an alternate path for our asserts.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20161129095008.32622-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
