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>
All the requests at the same priority are executed in FIFO order. They
do not need to be stored in the rbtree themselves, as they are a simple
list within a level. If we move the requests at one priority into a list,
we can then reduce the rbtree to the set of priorities. This should keep
the height of the rbtree small, as the number of active priorities can not
exceed the number of active requests and should be typically only a few.
Currently, we have ~2k possible different priority levels, that may
increase to allow even more fine grained selection. Allocating those in
advance seems a waste (and may be impossible), so we opt for allocating
upon first use, and freeing after its requests are depleted. To avoid
the possibility of an allocation failure causing us to lose a request,
we preallocate the default priority (0) and bump any request to that
priority if we fail to allocate it the appropriate plist. Having a
request (that is ready to run, so not leading to corruption) execute
out-of-order is better than leaking the request (and its dependency
tree) entirely.
There should be a benefit to reducing execlists_dequeue() to principally
using a simple list (and reducing the frequency of both rbtree iteration
and balancing on erase) but for typical workloads, request coalescing
should be small enough that we don't notice any change. The main gain is
from improving PI calls to schedule, and the explicit list within a
level should make request unwinding simpler (we just need to insert at
the head of the list rather than the tail and not have to make the
rbtree search more complicated).
v2: Avoid use-after-free when deleting a depleted priolist
v3: Michał found the solution to handling the allocation failure
gracefully. If we disable all priority scheduling following the
allocation failure, those requests will be executed in fifo and we will
ensure that this request and its dependencies are in strict fifo (even
when it doesn't realise it is only a single list). Normal scheduling is
restored once we know the device is idle, until the next failure!
Suggested-by: Michał Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Michał Winiarski <michal.winiarski@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Michał Winiarski <michal.winiarski@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170517121007.27224-8-chris@chris-wilson.co.uk
Rebrand the current (pointer | bits) pack/unpack utility macros as
explicit bit twiddling for PAGE_SIZE so that we can use the more
flexible underlying macros for different bits.
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/20170517121007.27224-4-chris@chris-wilson.co.uk
ptr_unpack_bits() is a function-like macro, as such it is meant to be
replaceable by a function. In this case, we should be passing in the
out-param as a pointer.
Bizarrely this does affect code generation:
function old new delta
i915_gem_object_pin_map 409 389 -20
An improvement(?) in this case, but one can't help wonder what
strict-aliasing optimisations we are preventing.
The generated code looks identical in using ptr_unpack_bits (no extra
motions to stack, the pointer and bits appear to be kept in registers),
the difference appears to be code ordering and with a reorder it is able
to use smaller forward jumps.
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/20170517121007.27224-3-chris@chris-wilson.co.uk
In order to allow use of e.g. forcewake_domains in a other feature headers
included from the top of i915_drv.h, move all uncore related definitions
into their own header.
v2: move __mask_next_bit macro to utils header (Mika)
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Suggested-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Mika Kuoppala <mika.kuoppala@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
We can't sometimes use these macros in other headers due to
include and definition order. As i915_utils.h already contains
other helper macros move these macros there.
v2: checkpatch cleanup for WARN() macro.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170328084513.174200-1-michal.wajdeczko@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Manual pointer manipulation is error prone. Let compiler calculate
right offsets for us in case we need to change ads layout.
v2: don't call it object (Chris)
v3: restyle offset assignments (Chris)
v4: stylistic reductions
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Oscar Mateo <oscar.mateo@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@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/20170314133309.126432-1-michal.wajdeczko@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Just do a quick check that the stolen memory address range doesn't
overflow our chosen integer type.
v2: Add add_overflows() to utils with the promise that gcc7 can do this
better than C and then maybe it will have a proper definition in core.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170130134721.5159-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Now that we have split out a header file for simple macros (that maybe
we can promote into a core header), move a few macros across from
i915_drv.h
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170105164148.26875-1-chris@chris-wilson.co.uk
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
In order to defeat some circular dependencies between headers to allow use
of e.g. range_overflows() in a header, move the simple independent macros
into their own header.
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/20170105153023.30575-4-chris@chris-wilson.co.uk
