Put some markers into the debug log as those functions create many
proxies causing a lot of debug noise.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
For SNB, in case you really, really want to use GPU detiling and not
incur the ring switch. Tweaking when to just mmap the target seems to
gain most anyway...
The ulterior motive is that this provides fallback paths for avoiding
the use of TILING_Y with GTT mmaps which is broken on 855gm.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We only want to create huge pwrite buffers when populating the inactive
cache for mmapped uploads. In the absence of using mmap for upload, be
more conservative with the alignment value so as not to simply waste
valuable aperture and memory.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If we steal a write buffer for creating a pixmap for read back, then we
need to be careful as we will have set the used amount to 0 and then try
to incorrectly decrease by the last row. Fortunately, we do not yet have
any code that attempts to create a 2d buffer for reading.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The batch may legitimately be submitted prior to the attachment of the
read buffer, if, for example, we need to switch rings. Therefore update
the assertion to only check that the bo remains in existence via either
a reference from the exec or from the user
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When backporting the patches from gen6, I didn't notice the memset that
came later, and this wasn't along the paths checked by rendercheck.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Whilst we can render to and blend with an depth 15 target, we cannot use
it as a texture with the sampling engine.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The hw wants it as demonstrated by the '>' in KDE's menus. Why is it
always KDE that demonstrates coherency problems...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
One of the side-effects of emitting the composite state is that it
tags the destination surface as dirty as a result of the *forthcoming*
operation. So emitting the flush after emitting the composite state
clears that tag, so we need to restore it for future coherency.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When creating native glamor pixmaps we will get much better performance
than using the textured-drm pixmap, this commit is to make that the
default behaviour when configured to use glamor. Another advantage
of this commit is that we reduce the risk of encountering the
"incompatible region exists for this name" and the associated
render corruption. And since we now never intentionally allocate
a reusable pixmap we could just make all (intel_glamor) allocations
non-reusable without incurring too great an overhead.
A side effect is that those glamor pixmaps do not have a
valid BO attached to them and thus it fails to get a DRI drawable. This
commit also fixes that problem by adjusting the fixup_shadow mechanism
to recreate a textured-drm pixmap from the native glamor pixmap. I tested
this with mutter, and it works fine.
The performance gain to apply this patch is about 10% to 20% with
different workload.
Signed-off-by: Zhigang Gong <zhigang.gong@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
If the GPU and CPU caches are shared and coherent, we can use a cached
mapping for linear bo in the CPU domain with no penalty and so avoid the
penalty of using WC/UC mappings through the GTT (and any aperture
pressure). We presume that the bo for such mappings are indeed LLC
cached...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In how many different ways can we check that the scanout is allocated
before we start decoding video?
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The idea being that they facilitate copying to and from the CPU, but
also we want to avoid stalling on any pixels help by the CPU bo.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Similar to the desire to flush the next batch after an overflow, we do
not want to incur any lag in the midst of drawing, even if that lag is
mitigated by GPU semaphores.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Using the gpu to do the detiling just incurs extra latency and an extra
copy, so go back to using a fence and GTT mapping for the common path.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Be sure the mask picture has a valid format even though it points to the
same pixels as the valid source. And also be wary if the source was
converted to a solid, but the mask is not.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Rather than pollute the cache with bo that are known not to be in the
GTT and are no longer useful, drop the bo after we read from it.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
And instead derive a power-of-two alignment value for partial buffer
sizes from the mappable aperture size and use that during
kgem_create_buffer()
Reported-by: Clemens Eisserer <linuxhippy@gmail.com>
References: https://bugs.freedesktop.org/show_bug.cgi?id=44682
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Use a single idiom and reuse the check built into the state emission,
for both spans/boxes paths.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Typically we will be bound to the RENDER ring as once engaged we try not
to switch. However, with semaphores enabled we may switch more freely
and there it is advantageous to use as much of the faster BLT as is
feasible.
The most contentious point here is the choice of whether to use BLT for
copies by default. microbenchmarks (compwinwin) benefit from the
coallescing performed in the render batch, but the more complex traces
seem to prefer utilizing the blitter. The debate will continue...
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If the kernel uses GPU semaphores for its coherency mechanism between
rings rather than CPU waits, allow the ring to be chosen on the basis
of the subsequent operation following a submission of batch. (However,
since batches are likely to be submitted in the middle of a draw, then
the likelihood is for ddx to remain on one ring until forced to switch
for an operation or idle, which is the same situation as before and so
the difference is miniscule.)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Initially, the batch->mode was only set upon an actual mode switch,
batch submission would not reset the mode. However, to facilitate fast
ring switching with semaphores, reseting the mode upon batch submission
is desired which means that if we submit the batch in the middle of an
operation we must redeclare its mode before continuing.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The glyph cache grew to accommodate the fallback pixman image for mask
generation, and is equally applicable along the full fallback path.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Processing more than a single rectangle using the CA path on ILK is
extremely hit-or-miss, often resulting in the absence of the second
primitive (ie. the glyphs are cleared but not added.) This is
reminiscent of the complete breakage of the BRW shaders, none of which
can handle more than a single rectangle.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In the common case, we expect a very small number of vertices which will
fit into the batch along with the commands. However, in full flow we
overflow the on-stack buffer and likely several continuation buffers.
Streaming those straight into the GTT seems like a good idea, with the
usual caveats over aperture pressure. (Since these are linear we could
use snoopable bo for the architectures that support such for vertex
buffers and if we had kernel support.)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The goal is to reuse the most recently bound GTT mapping in the hope
that is still mappable at the time of reuse.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This includes the condition where the pixmap is too large, as well as
being too small, to be allocatable on the GPU. It is only a hint set
during creation, and may be overridden if required.
This fixes the regression in ocitysmap which decided to render glyphs
into a GPU mask for a destination that does not fit into the aperture.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Hide the noise under another level of debugging so that hopefully the
reason why it chose a particular path become clear.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In a few places, we can stream the source into the GTT and so upload in
place through the WC mapping. Notably, in many other places we want to
rasterise on a partial in cacheable memory. So we need to notify the
backend of the intended usage for the buffer and when we think it is
appropriate we can allocate a GTT mapped pointer for zero-copy upload.
The biggest improvement tends to be in the PutComposite style of
microbenchmark, yet throughput for trapezoid masks seems to suffer (e.g.
swfdec-giant-steps on i3 and gen2 in general). As expected, the culprit
of the regression is the aperture pressure causing eviction stalls, which
the pwrite paths sidesteps by doing a cached copy when there is no GTT
space. This could be alleviated with an is-mappable ioctl predicting when
use of the buffer would block and so falling back in those cases to
pwrite. However, I suspect that this will improve dispatch latency in
the common idle case for which I have no good metric.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson
Zhigang Gong