The design is for the user to create a cairo_font_options_t object with
cairo_font_options_create() and then is free to use it with any Cairo
operation. This requires us to check when we may be about to overwrite
the read-only nil object.
Original work by Jorn Baayen <jorn@openedhand.com>,
2715f20981
We use a small cache of size 16 for surfaces created for solid patterns.
This mainly helps with the X backends where we don't have to create a
pattern for every operation, so we save a lot on X traffic. Xft uses a
similar cache, so cairo's text rendering traffic with the xlib backend
now completely matches that of Xft.
The cache uses an static index variable, which itself acts like a cache of
size 1, remembering the most recently used solid pattern. So repeated
lookups for the same pattern hit immediately. If that fails, the cache is
searched linearly, and if that fails too, a new surface is created and a
random member of the cache is evicted.
A cached surface can only be reused if it is similar to the destination.
In order to check for similar surfaces a new test is introduced for the
backends to determine that the cached surface is as would be returned by
a _create_similar() call for the destination and content.
As surfaces are in general complex encapsulation of graphics state we
only return unshared cached surfaces and reset them (to clear any error
conditions and graphics state). In practice this makes little difference
to the efficacy of the cache during various benchmarks. However, in order
to transparently share solid surfaces it would be possible to implement a
COW scheme.
Cache hit rates: (hit same index + hit in cache) / lookups
cairo-perf: (42346 + 28480) / 159600 = 44.38%
gtk-theme-torturer: (3023 + 3502) / 6528 = 99.95%
gtk-perf: (8270 + 3190) / 21504 = 53.29%
This translates into a reduction of about 25% of the XRENDER traffic during
cairo-perf.
This allows for the surface acquired from the pattern to have the
same content. In particular, in a case such as cairo_paint_with_alpha
we can now acquire an A8 mask surface instead of an ARGB32 mask
surface which can be rendered much more efficiently. This results
in a 4x speedup when using the OVER operator with the recently
added paint-with-alpha test:
Speedups
========
image-rgb paint-with-alpha_image_rgb_over-256 2.25 -> 0.60: 4.45x speedup
███▌
It does slowdown the same test when using the SOURCE operator, but
I don't think we care. Performing SOURCE with a mask is already a very
slow operation, (hitting compositeGeneral), so the slowdown here is
likely from having to convert from A8 back to ARGB32 before the
generalized compositing. So if someone cares about this slowdown,
(though SOURCE with cairo_paint_with_alpha doesn't seem extremely
useful), they will probably be motivated enough to contribute a
customized compositing function to replace compositeGeneral in which
case this slowdown should go away:
image-rgba paint-with-alpha_image_rgb_source-256 3.84 -> 8.86%: 1.94x slowdown
█
_cairo_surface_create_similar_solid() creates a fresh pattern to wrap
color, however sometimes the caller already has that pattern available.
In those circumstances we can pass the pattern as well as the color and
avoid the extra allocation.
Fixing this uncovered a leak of a CAIRO_INT_STATUS_UNSUPPORTED value
up to cairo_show_page, (and similarly to cairo_copy_page). There was
really no good reason for _cairo_surface_show_page and
_cairo_surface_copy_page to be returning cairo_int_status_t. Fix
this by simply handling the UNSUPPORTED return at the surface layer
instead of the gstate layer.
Most of the time pixman_region_init is called without any extents, and
followed by a pixman_region_union_rect, used to used to initialize
rectangular regions. pixman_region_union_rect is not that cheap, but
the sequence is called quite often. So it should be worth introducing
a specialized and fast function for this sequence.
This introduces pixman_region_init_rect. This new function makes
_cairo_region_init_from_rectangle obsolete.
Also removes the extent argument from pixman_region_init as it was
called with NULL most of the time. A pixman_region_init_with_extents
is added for the general case.
This is necessary to avoid many portability problems as cairoint.h includes
config.h. Without a test, we will regress again, hence add it.
The inclusion idiom for cairo now is:
#include "cairoint.h"
#include "cairo-something.h"
#include "cairo-anotherthing-private.h"
#include <some-library.h>
#include <other-library/other-file.h>
Moreover, some standard headers files are included from cairoint.h and need
not be included again.
All mutex declarations have been moved to cairo-mutex-list.h.
This should avoid breaking of less frequently tested backends,
when mutexes are introduced or when existing mutexes are renamed.
Instead of initializing mutexes on library startup, mutexes are
lazily initialized within the few entry points of now by calling
CAIRO_MUTEX_INITIALIZE(). Currently only the OS/2 backend takes
care about releasing global mutexes. Therefore there is no counter
part of that macro for finalizing all global mutexes yet - but
as cairo-backend-os2.c shows such a function would be quite
easy to implement.
We detect an error in the surface before calling into private surface-modifying
functions, (such as _cairo_surface_set_font_options), that don't have the
nil-surface protection of public functions.
This should fix the problem reported (again) in this bug report:
cairo crashes in cairo_create_simular if nil surface returned by other->backend->create_similar
https://bugs.freedesktop.org/show_bug.cgi?id=9844
Though I haven't succeeded in replicating the bug yet, (perhaps a system
difference in allowing writes to read-only memory or not, or something
like that).
When a single function accepts pointers for multiple return values,
the convention is that it's legal for the user to pass NULL for
those pointers in case the user is only interested in some subset
of the values.
This was already properly implemented for functions such as
cairo_pattern_get_rgba, etc.
Here we fix four functions to follow the same convention:
cairo_stroke_extents
cairo_fill_extents
cairo_clip_extents
cairo_surface_get_device_offset
On some architectures, gcc will emit a memcpy for structure copies which will
produce a valgrind warning when the source and destination pointers are the
same. Workaround this issue by explicitly checking the source and destination
for inequality before doing the structure assignment.
This reverts the following commits:
2715f2098167e3b3c53b
See this thread for an analysis of the problems it caused:
http://lists.freedesktop.org/archives/cairo/2007-February/009825.html
In short, a single cache for all backends doesn't work, as one thread
using any backend can cause an unused xlib pattern to be evicted from
the cache, and trigger an xlib call while the display is being used
from another thread. Xlib is not prepared for this.
for surface patterns.
Right before releasing cairo 1.2, Carl and I decided to return error on
CAIRO_EXTEND_PAD and CAIRO_EXTEND_REFLECT for surface patterns, as they
are not implemented and one was causing crashes. Well, that was probably
the worst decision we could make (other than ignoring the problem). A
much better decision would have been to make them act like
CAIRO_EXTEND_NONE and CAIRO_EXTEND_REPEAT respectively. Anyway, remove
the error paths.
Previous commit broke cairo_surface_finish, since it was checking for
ref_count == CAIRO_REF_COUNT_INVALID and bailing. But, that condition
was reached from destroy, so finish was bailing out early.
user_data setters/getters were added to public refcounted objects
that were missing them (cairo_t, pattern, scaled_font). Also,
a refcount getter (cairo_*_get_reference_count) was added to all
public refcounted objects.
We use a small cache of size 16 for surfaces created for solid patterns.
This mainly helps with the X backends where we don't have to create a
pattern for every operation, so we save a lot on X traffic. Xft uses a
similar cache, so cairo's text rendering traffic with the xlib backend
now completely matches that of Xft.
The cache uses an static index variable, which itself acts like a cache of
size 1, remembering the most recently used solid pattern. So repeated
lookups for the same pattern hit immediately. If that fails, the cache is
searched linearly, and if that fails too, a new surface is created and a
random member of the cache is evicted.
Only surfaces that are "compatible" are used. The definition of compatible
is backend specific. For the xlib backend, it means that the two surfaces
are allocated on the same display. Implementations for compatibility are
provided for all backends that it makes sense.
A cairo_scaled_font_t can be implicitly shared among multiple threads
as the same cairo_scaled_font_t can be returned from different calls
to cairo_scaled_font_create. To retain the illusion that these
different calls produce distinct objects, cairo must internally lock
access when modifying them.
Each glyph in the scaled font is represented by a cairo_surface_t
which is used when rendering the glyph. Instead of attempting to push
fine-grained locking of these surfaces down to the backend rendering
functions, a simple per-cairo_scaled_font_t lock has been introduced
which protects the entire rendering path against re-entrancy.
Some care was required to ensure that existing re-entrancy was handled
appropriately; these cases are in the wrapping surfaces
(cairo-paginated, test-meta and test-paginated).
Thanks to Vladimir Vukicev and Peter Weilbacher for testing/providing
the mutex definitions for win32 and os2 (respectively).
Calling cairo_surface_finish from cairo_surface_destroy was
triggering an error due to finish being called twice. The
error was usually hidden as the surface would still eventually
be destroyed. But it did clutter things up quite badly if a
user was trying to break on _cairo_error to track down a
problem.
Thanks again to Stuart Parmenter <stuartp@gmail.com>
for pointing out the problem.
The rule is: cairo_glyph_t* is always passed as const for measurement
purposes. This was not reflected in our public api previously. Fixed
Showing glyphs used to have cairo_glyph_t* always as const. With this
changed, it is only const on cairo_t and cairo_gstate_t operations.
cairo_surface_t, cairo_scaled_font_t, and individual backends receive
cairo_glyph_t* as non-const. The desired semantics is that they may modify
the contents of the array as long as they do not return
CAIRO_STATUS_UNSUPPORTED. This makes it possible to avoid copying the glyph
array again and again, and edit it in-place. Backends are in fact free to use
the array as a generic buffer as they see fit.
This fixes a huge performance bug (entire image was being pushed to X
server in order to copy a tiny piece of it). I see up to 50x improvement
from subimage_copy (which was designed to expose this problem) but also
a 5x improvement in some text performance cases.
xlib-rgba subimage_copy-512 3.93 2.46% -> 0.07 2.71%: 52.91x faster
███████████████████████████████████████████████████▉
xlib-rgb subimage_copy-512 4.03 1.97% -> 0.09 2.61%: 44.74x faster
███████████████████████████████████████████▊
xlib-rgba subimage_copy-256 1.02 2.25% -> 0.07 0.56%: 14.42x faster
█████████████▍
xlib-rgba text_image_rgb_over-256 63.21 1.53% -> 11.87 2.17%: 5.33x faster
████▍
xlib-rgba text_image_rgba_over-256 62.31 0.72% -> 11.87 2.82%: 5.25x faster
████▎
xlib-rgba text_image_rgba_source-256 67.97 0.85% -> 16.48 2.23%: 4.13x faster
███▏
xlib-rgba text_image_rgb_source-256 68.82 0.55% -> 16.93 2.10%: 4.07x faster
███▏
xlib-rgba subimage_copy-128 0.19 1.72% -> 0.06 0.85%: 3.10x faster
██▏
New internal function _cairo_surface_set_font_options is used to set them.
cairo_surface_create_similar propagates the font options of the other
surface into the newly created surface. Fixes bugs with font options in
fallback images and bug 4106.
