This turns an !!!ERROR!!! for scale-offset-similar with xcb-fallback into a
failed test and might fix other problems. However, since the problem here partly
is a race, those other problems might only be hit sometimes.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Seeing the relative amounts of time spent in each operation and the
slowest one of each, gives further insight into the peculiarities of a
trace. And hopefully point out areas of improvement.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
A clip with only a single path or can be reduced to a single polygon are
easier to handle than a clip containing a mixture of paths (typically
ANTIALIAS_NONE vs ANTIALIAS_DEFAULT).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Mucking around in 99fa5ff6c2, I tweaked the order to mark the
snapshot as detached before calling the callback. xcb relies on the old
ordering so that it can correctly update its fallbacks.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
As these tend to be the quickest, and putting them first keeps the reports
are in an estimated fast->slow order.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
In step 1 of speeding up stroking, we introduce contours as a means for
tracking the connected edges around the stroke. By keeping track of
these chains, we can analyse the edges as we proceed and eliminate
redundant vertices speeding up rasterisation.
Coincidentally fixes line-width-tolerance (looks like a combination of
using spline tangent vectors and tolerance).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Adjusting the _cairo_surface_is_recording() to check the value
in the backend exposed us to a potential NULL pointer dereference
on validating user-input. So add an explicit status check which
has the added bonus of being more correct!
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The mime surface is a user-callback surface designed for interfacing
cairo with an opaque data source. For instance, in a web browser, the
incoming page may be laid out and rendered to a recording surface before
all the image data has finished being downloaded. In this circumstance
we need to pass a place holder to cairo and to supply the image data
later upon demand.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Another logging passthrough surface that records the style of operations
performed trying to categorise what is slow/fast/important.
In combination with perf/cairo-analyse-trace it is very useful for
understanding what a trace does. The next steps for this tool would be
to identify the slow operations that the trace does. Baby steps.
This should be generally useful in similar situations outside of perf/
and should be extensible to become an online performance probe.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
So that we remember to forget the old image when the surface is dirtied.
Fixes filter-bilinear-extents.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Fixes push-group-offset which reduces to a replay of a recording surface
inside a recording surface and forgot to offset the clip imposed by the
extents of the first recording surface into device space.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If a function expects a surface but receives a context, it automatically
queries the context's target. We can take advantage of this to short-cut
a few operations.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
We clear past the end of the row so that we don't trigger valgrind
warning leaving harmless uninitialised bits inside the input image.
However, for RGB24 the input rowlen is 3*width, whereas we write 4*width
of data, so we need to take account of that and ensure we clear beyond
the end of the written data, not the read data.
Fixes reading of RGB24 input.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The high-level function creates a new lazy snapshot which is not what we
what - we want a snapshot now!
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Thanks to subsurface recursion. There's a pattern here, but no clean
solution has yet presented itself.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Ouch, a nasty bug surfaces after rearranging code to fix the others.
Another self-copy loop this time through a subsurface of a recording
surface.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
And not the backend directly as this bypasses the extra steps taken in
the higher level to do common actions such as detach snapshots.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
This is the root cause of the issue why we never succeeded in
implementing deferred snapshot correctly; that is we decoupled the
source from the target in the upper layers before we make the coupling
inside the lowest level of recording surface. By deferring the copy, we
never saw the detach-snapshot in time.
Fortunately this was only an issue for backends that implemented strong
immutable source semantics! The oversight implies that we need to push
down a similar flush mechanism into all backends.
Fixes self-copy.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
It was supposed to be the centre point of e849e7c92, but I had a little
battle with git and lost...
Reported-by: James Cloos <cloos@jhcloos.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
As handling joins/caps between line segments shorter than
half_line_width is tricky.
Rather than also fixing the bug in traps, remove that code. The plan is
to avoiding hitting the traps code, short-circuiting several steps along
the fast rectangular paths.
Fixes line-width-overlap.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
