profile/ivi/pixman.git
12 years agoUse the ARRAY_LENGTH() macro when possible
Andrea Canciani [Thu, 3 Nov 2011 09:51:27 +0000 (10:51 +0100)]
Use the ARRAY_LENGTH() macro when possible

This patch has been generated by the following Coccinelle semantic patch:

// Use the ARRAY_LENGTH() macro when possible
//
// Replace open-coded array length computations with the
// ARRAY_LENGTH() macro

@@
type T;
T[] E;
@@
- (sizeof(E)/sizeof(T))
+ ARRAY_LENGTH (E)

12 years agotest: Cleanup includes
Andrea Canciani [Thu, 3 Nov 2011 09:40:24 +0000 (10:40 +0100)]
test: Cleanup includes

All the tests are linked to libutil, hence it makes sence to always
include utils.h and reuse what it provides (config.h inclusion, access
to private pixman APIs, ARRAY_LENGTH, ...).

12 years agoRemove useless checks for NULL before freeing
Andrea Canciani [Thu, 3 Nov 2011 09:21:41 +0000 (10:21 +0100)]
Remove useless checks for NULL before freeing

This patch has been generated by the following Coccinelle semantic patch:

// Remove useless checks for NULL before freeing
//
// free (NULL) is a no-op, so there is no need to avoid it

@@
expression E;
@@
+ free (E);
+ E = NULL;
- if (unlikely (E != NULL)) {
-   free(E);
(
-   E = NULL;
|
-   E = 0;
)
   ...
- }

@@
expression E;
@@
+ free (E);
- if (unlikely (E != NULL)) {
-   free (E);
- }

12 years agoPost-release version bump to 0.25.1
Søren Sandmann Pedersen [Sun, 6 Nov 2011 21:36:01 +0000 (16:36 -0500)]
Post-release version bump to 0.25.1

12 years agoPre-release version bump to 0.24.0
Søren Sandmann Pedersen [Sun, 6 Nov 2011 21:10:33 +0000 (16:10 -0500)]
Pre-release version bump to 0.24.0

12 years agoChange MMX ldq_u to return _m64 instead of forcing all callers to cast
Alan Coopersmith [Sun, 30 Oct 2011 16:12:06 +0000 (09:12 -0700)]
Change MMX ldq_u to return _m64 instead of forcing all callers to cast

Sun/Oracle Studio compilers allow the pointers to be cast, but not the
non-pointer forms, causing pixman compiles to fail with many errors of:
"pixman-mmx.c", line 1411: invalid cast expression

Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
12 years agoAdd definitions of INT64_MIN and INT64_MAX
Jeff Muizelaar [Wed, 2 Nov 2011 22:49:58 +0000 (18:49 -0400)]
Add definitions of INT64_MIN and INT64_MAX

12 years agoPost-release version bump to 0.23.9
Søren Sandmann Pedersen [Sat, 29 Oct 2011 09:51:44 +0000 (05:51 -0400)]
Post-release version bump to 0.23.9

12 years agoPre-release version bump to 0.23.8
Søren Sandmann Pedersen [Sat, 29 Oct 2011 09:33:44 +0000 (05:33 -0400)]
Pre-release version bump to 0.23.8

12 years agoFix use of uninitialized fields reported by valgrind
Søren Sandmann Pedersen [Tue, 25 Oct 2011 12:45:34 +0000 (08:45 -0400)]
Fix use of uninitialized fields reported by valgrind

In pixman-noop.c and pixman-sse2.c, we are accessing
image->bits.width/height without first making sure the image is a bits
image. The warning is harmless because we never act on this
information without checking that the image is a8r8g8b8, but valgrind
does warn about it.

In pixman-noop.c, just reorder the clauses in the if statement; in
pixman-sse2.c require images to have the FAST_PATH_BITS_IMAGE flag
set.

12 years agoMerge branch 'gradients'
Søren Sandmann Pedersen [Thu, 20 Oct 2011 13:13:12 +0000 (09:13 -0400)]
Merge branch 'gradients'

12 years agoARM: NEON: Fix assembly typo error in src_n_8_8888
Taekyun Kim [Tue, 18 Oct 2011 12:50:18 +0000 (21:50 +0900)]
ARM: NEON: Fix assembly typo error in src_n_8_8888

Binutils 2.21 does not complain about missing comma between ARM
register and alignement specifier in vld/vst instructions which
causes build error on binutils 2.20.

12 years agoARM: NEON: Standard fast path src_n_8_8
Taekyun Kim [Mon, 26 Sep 2011 09:33:27 +0000 (18:33 +0900)]
ARM: NEON: Standard fast path src_n_8_8

Performance numbers of before/after on cortex-a8 @ 1GHz

- before
L1:  28.05  L2:  28.26  M: 26.97 (  4.48%)  HT: 19.79  VT: 19.14  R: 17.61  RT:  9.88 ( 101Kops/s)

- after
L1:1430.28  L2:1252.10  M:421.93 ( 75.48%)  HT:170.16  VT:138.03  R:145.86  RT: 35.51 ( 255Kops/s)

12 years agoARM: NEON: Standard fast path src_n_8_8888
Taekyun Kim [Mon, 26 Sep 2011 08:03:54 +0000 (17:03 +0900)]
ARM: NEON: Standard fast path src_n_8_8888

Performance numbers of before/after on cortex-a8 @ 1GHz

- before
L1:  32.39  L2:  31.79  M: 30.84 ( 13.77%)  HT: 21.58  VT: 19.75  R: 18.83  RT: 10.46 ( 106Kops/s)

- after
L1: 516.25  L2: 372.00  M:193.49 ( 85.59%)  HT:136.93  VT:109.10  R:104.48  RT: 34.77 ( 253Kops/s)

12 years agoARM: NEON: Instruction scheduling of bilinear over_8888_8_8888
Taekyun Kim [Mon, 26 Sep 2011 10:04:53 +0000 (19:04 +0900)]
ARM: NEON: Instruction scheduling of bilinear over_8888_8_8888

Instructions are reordered to eliminate pipeline stalls and get
better memory access.

Performance of before/after on cortex-a8 @ 1GHz

<< 2000 x 2000 with scale factor close to 1.x >>
before : 40.53 Mpix/s
after  : 50.76 Mpix/s

12 years agoARM: NEON: Instruction scheduling of bilinear over_8888_8888
Taekyun Kim [Wed, 21 Sep 2011 06:52:13 +0000 (15:52 +0900)]
ARM: NEON: Instruction scheduling of bilinear over_8888_8888

Instructions are reordered to eliminate pipeline stalls and get
better memory access.

Performance of before/after on cortex-a8 @ 1GHz

<< 2000 x 2000 with scale factor close to 1.x >>
before : 50.43 Mpix/s
after  : 61.09 Mpix/s

12 years agoARM: NEON: Replace old bilinear scanline generator with new template
Taekyun Kim [Thu, 22 Sep 2011 15:03:22 +0000 (00:03 +0900)]
ARM: NEON: Replace old bilinear scanline generator with new template

Bilinear scanline functions in pixman-arm-neon-asm-bilinear.S can
be replaced with new template just by wrapping existing macros.

12 years agoARM: NEON: Bilinear macro template for instruction scheduling
Taekyun Kim [Tue, 20 Sep 2011 12:32:35 +0000 (21:32 +0900)]
ARM: NEON: Bilinear macro template for instruction scheduling

This macro template takes 6 code blocks.

1. process_last_pixel
2. process_two_pixels
3. process_four_pixels
4. process_pixblock_head
5. process_pixblock_tail
6. process_pixblock_tail_head

process_last_pixel does not need to update horizontal weight. This
is done by the template. two and four code block should update
horizontal weight inside of them. head/tail/tail_head blocks
consist unrolled core loop. You can apply instruction scheduling
to the tail_head blocks.

You can also specify size of the pixel block. Supported size is 4
and 8. If you want to use mask, give BILINEAR_FLAG_USE_MASK flags
to the template, then you can use register MASK. When using d8~d15
registers, give BILINEAR_FLAG_USE_ALL_NEON_REGS to make sure
registers are properly saved on the stack and later restored.

12 years agoARM: NEON: Some cleanup of bilinear scanline functions
Taekyun Kim [Tue, 20 Sep 2011 10:46:25 +0000 (19:46 +0900)]
ARM: NEON: Some cleanup of bilinear scanline functions

Use STRIDE and initial horizontal weight update is done before
entering interpolation loop. Cache preload for mask and dst.

12 years agoSimplify gradient_walker_reset()
Søren Sandmann Pedersen [Fri, 14 Oct 2011 13:04:48 +0000 (09:04 -0400)]
Simplify gradient_walker_reset()

The code that searches for the closest color stop to the given
position is duplicated across the various repeat modes. Replace the
switch with two if/else constructions, and put the search code between
them.

12 years agoUse sentinels instead of special casing first and last stops
Søren Sandmann Pedersen [Fri, 14 Oct 2011 13:02:14 +0000 (09:02 -0400)]
Use sentinels instead of special casing first and last stops

When storing the gradient stops internally, allocate two more stops,
one before the beginning of the stop list and one after the
end. Initialize those stops based on the repeat property of the
gradient.

This allows gradient_walker_reset() to be simplified because it can
now simply pick the two closest stops to the position without special
casing the first and last stops.

12 years agogradient walker: Correct types and fix formatting
Søren Sandmann Pedersen [Fri, 14 Oct 2011 11:42:00 +0000 (07:42 -0400)]
gradient walker: Correct types and fix formatting

The type of pos in gradient_walker_reset() and gradient_walker_pixel()
is pixman_fixed_48_16_t and not pixman_fixed_32_32. The types of the
positions in the walker struct are pixman_fixed_t and not int32_t, and
need_reset is a boolean, not an integer. The spread field should be
called repeat and have the type pixman_repeat_t.

Also fix some formatting issues, make gradient_walker_reset() static,
and delete the pointless PIXMAN_GRADIENT_WALKER_NEED_RESET() macro.

12 years agoAdd stable release / development snapshot to draft release notes
Søren Sandmann Pedersen [Tue, 11 Oct 2011 20:12:24 +0000 (16:12 -0400)]
Add stable release / development snapshot to draft release notes

This will hopefully serve as a reminder to me that I should put this
information in the release notes.

12 years agoPost-release version bump to 0.23.7
Søren Sandmann Pedersen [Tue, 11 Oct 2011 10:10:39 +0000 (06:10 -0400)]
Post-release version bump to 0.23.7

12 years agoPre-release version bump to 0.23.6
Søren Sandmann Pedersen [Tue, 11 Oct 2011 10:00:51 +0000 (06:00 -0400)]
Pre-release version bump to 0.23.6

12 years agoSimple repeat: Extend too short source scanlines into temporary buffer
Taekyun Kim [Thu, 22 Sep 2011 09:42:38 +0000 (18:42 +0900)]
Simple repeat: Extend too short source scanlines into temporary buffer

Too short scanlines can cause repeat handling overhead and optimized
pixman composite functions usually process a bunch of pixels in a
single loop iteration it might be beneficial to pre-extend source
scanlines. The temporary buffers will usually reside in cache, so
accessing them should be quite efficient.

12 years agoSimple repeat fast path
Taekyun Kim [Mon, 29 Aug 2011 12:44:36 +0000 (21:44 +0900)]
Simple repeat fast path

We can implement simple repeat by stitching existing fast path
functions. First lookup COVER_CLIP function for given input and
then stitch horizontally using the function.

12 years agoMove _pixman_lookup_composite_function() to pixman-utils.c
Taekyun Kim [Thu, 22 Sep 2011 07:33:02 +0000 (16:33 +0900)]
Move _pixman_lookup_composite_function() to pixman-utils.c

12 years agoAdd src, mask, and dest flags to the composite args struct.
Søren Sandmann Pedersen [Mon, 27 Jun 2011 21:17:04 +0000 (21:17 +0000)]
Add src, mask, and dest flags to the composite args struct.

These flags are useful in the various compositing routines, and the
flags stored in the image structs are missing some bits of information
that can only be computed when pixman_image_composite() is called.

12 years agoAdd new fast path flag FAST_PATH_BITS_IMAGE
Taekyun Kim [Thu, 22 Sep 2011 07:26:55 +0000 (16:26 +0900)]
Add new fast path flag FAST_PATH_BITS_IMAGE

This fast path flag indicate that type of the image is bits image.

12 years agoinit/fini functions for pixman_image_t
Taekyun Kim [Thu, 22 Sep 2011 07:20:03 +0000 (16:20 +0900)]
init/fini functions for pixman_image_t

pixman_image_t itself can be on stack or heap. So segregating
init/fini from create/unref can be useful when we want to use
pixman_image_t on stack or other memory.

12 years agosse2: Bilinear scaled over_8888_8_8888
Taekyun Kim [Wed, 7 Sep 2011 14:00:29 +0000 (23:00 +0900)]
sse2: Bilinear scaled over_8888_8_8888

12 years agosse2: Bilinear scaled over_8888_8888
Taekyun Kim [Wed, 7 Sep 2011 13:57:29 +0000 (22:57 +0900)]
sse2: Bilinear scaled over_8888_8888

12 years agosse2: Macros for assembling bilinear interpolation code fractions
Taekyun Kim [Wed, 7 Sep 2011 13:51:46 +0000 (22:51 +0900)]
sse2: Macros for assembling bilinear interpolation code fractions

Primitive bilinear interpolation code is reusable to implement other
bilinear functions.

BILINEAR_DECLARE_VARIABLES
- Declare variables needed to interpolate src pixels.

BILINEAR_INTERPOLATE_ONE_PIXEL
- Interpolate one pixel and advance to next pixel

BILINEAR_SKIP_ONE_PIXEL
- Skip interpolation and just advance to next pixel
  This is useful for skipping zero mask

12 years agoCorrect the minimum gcc version needed for iwmmxt
Matt Turner [Thu, 6 Oct 2011 21:56:09 +0000 (17:56 -0400)]
Correct the minimum gcc version needed for iwmmxt

Spotted by Søren Sandmann.

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agoMake sure iwMMXt is only detected on ARM
Matt Turner [Thu, 6 Oct 2011 02:54:36 +0000 (22:54 -0400)]
Make sure iwMMXt is only detected on ARM

iwMMXt is incorrectly detected on x86 and amd64. This happens because
the test uses standard _mm_* intrinsic functions which it compiles with
-march=iwmmxt, but when the user has set CFLAGS=-march=k8 for instance,
no error is generated from -march=iwmmxt, even though it's not a valid
flag on x86/amd64. Passing CFLAGS=-march=native does not override the
-march=iwmmxt flag though, which is why it wasn't noticed before.

So, just #error out in the test if the __arm__ preprocessor directive
isn't defined.

Fixes https://bugs.gentoo.org/show_bug.cgi?id=385179

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agoDon't include stdint.h in scaling-helpers-test.
Søren Sandmann Pedersen [Tue, 27 Sep 2011 15:32:13 +0000 (11:32 -0400)]
Don't include stdint.h in scaling-helpers-test.

Fixes bug 41257.

12 years agobuild: replace @VAR@ with $(VAR) in makefiles
Benjamin Otte [Wed, 14 Sep 2011 15:52:03 +0000 (17:52 +0200)]
build: replace @VAR@ with $(VAR) in makefiles

12 years agotests: Add PNG_CFLAGS/LIBS to tests
Benjamin Otte [Wed, 14 Sep 2011 15:01:51 +0000 (17:01 +0200)]
tests: Add PNG_CFLAGS/LIBS to tests

PNG flags were accidentally included by gdk-pixbuf. This has been fixed
recently, so we need to make sure to include it ourselves.

12 years agommx: optimize unaligned 64-bit ARM/iwmmxt loads
Matt Turner [Thu, 22 Sep 2011 19:28:00 +0000 (15:28 -0400)]
mmx: optimize unaligned 64-bit ARM/iwmmxt loads

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: compile on ARM for iwmmxt optimizations
Matt Turner [Mon, 1 Aug 2011 02:42:24 +0000 (22:42 -0400)]
mmx: compile on ARM for iwmmxt optimizations

Check in configure for at least gcc-4.6, since gcc-4.7 (and hopefully
4.6) will be the eariest version capable of compiling the _mm_*
intrinsics on ARM/iwmmxt. Even for suitable compile versions I use
_mm_srli_si64 which is known to cause unpatched compilers to fail.

Select iwmmxt at runtime only after NEON, since we expect the NEON
optimizations to be more capable and faster than iwmmxt.

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: prepare pixman-mmx.c to be compiled for ARM/iwmmxt
Matt Turner [Sun, 4 Sep 2011 18:11:46 +0000 (14:11 -0400)]
mmx: prepare pixman-mmx.c to be compiled for ARM/iwmmxt

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: fix unaligned accesses
Matt Turner [Thu, 8 Sep 2011 18:33:45 +0000 (20:33 +0200)]
mmx: fix unaligned accesses

Simply return *p in the unaligned access functions, since alignment
constraints are very relaxed on x86 and this allows us to generate
identical code as before.

Tested with the test suite, lowlevel-blit-test, and cairo-perf-trace on
ARM and Alpha with no unaligned accesses found.

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: wrap x86/MMX inline assembly in ifdef USE_X86_MMX
Matt Turner [Thu, 22 Sep 2011 19:39:53 +0000 (15:39 -0400)]
mmx: wrap x86/MMX inline assembly in ifdef USE_X86_MMX

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: rename USE_MMX to USE_X86_MMX
Matt Turner [Sun, 31 Jul 2011 20:20:12 +0000 (20:20 +0000)]
mmx: rename USE_MMX to USE_X86_MMX

This will make upcoming ARM usage of pixman-mmx.c unambiguous.

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: convert while (w) to if (w) when possible
Matt Turner [Fri, 23 Sep 2011 18:10:52 +0000 (14:10 -0400)]
mmx: convert while (w) to if (w) when possible

gcc isn't able to see that w is no greater than 1, so it generates
unnecessary loop instructions with while (w).

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agommx: fix formats in commented code
Matt Turner [Fri, 9 Sep 2011 13:33:14 +0000 (15:33 +0200)]
mmx: fix formats in commented code

b8r8g8 is apparently no longer supported sometime since this code was
commented.

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agolowlevel-blt: add over_x888_8_8888
Matt Turner [Fri, 9 Sep 2011 13:34:04 +0000 (15:34 +0200)]
lowlevel-blt: add over_x888_8_8888

Signed-off-by: Matt Turner <mattst88@gmail.com>
12 years agoBILINEAR->NEAREST filter optimization for simple rotation and translation
Siarhei Siamashka [Sun, 22 May 2011 19:51:00 +0000 (22:51 +0300)]
BILINEAR->NEAREST filter optimization for simple rotation and translation

Simple rotation and translation are the additional cases when BILINEAR
filter can be safely reduced to NEAREST.

12 years agoStrength-reduce BILINEAR filter to NEAREST filter for identity transforms
Søren Sandmann Pedersen [Sun, 4 Sep 2011 06:53:39 +0000 (02:53 -0400)]
Strength-reduce BILINEAR filter to NEAREST filter for identity transforms

An image with a bilinear filter and an identity transform is
equivalent to one with a nearest filter, so there is no reason the
standard fast paths shouldn't be usable.

But because a BILINEAR filter samples a 2x2 pixel block in the source
image, FAST_PATH_SAMPLES_COVER_CLIP can't be set in the case where the
source area is the entire image, because some compositing operations
might then read pixels outside the image.

This patch fixes the problem by splitting the
FAST_PATH_SAMPLES_COVER_CLIP flag into two separate flags
FAST_PATH_SAMPLES_COVER_CLIP_NEAREST and
FAST_PATH_SAMPLES_COVER_CLIP_BILINEAR that indicate that the clip
covers the samples taking into account NEAREST/BILINEAR filters
respectively.

All the existing compositing operations that require
FAST_PATH_SAMPLES_COVER_CLIP then have their flags modified to pick
either COVER_CLIP_NEAREST or COVER_CLIP_BILINEAR depending on which
filter they depend on.

In compute_image_info() both COVER_CILP_NEAREST and
COVER_CLIP_BILINEAR can be set depending on how much room there is
around the clip rectangle.

Finally, images with an identity transform and a bilinear filter get
FAST_PATH_NEAREST_FILTER set as well as FAST_PATH_BILINEAR_FILTER.

Performance measurementas with render_bench against Xephyr:

Before

*** ROUND 1 ***
---------------------------------------------------------------
Test: Test Xrender doing non-scaled Over blends
Time: 5.720 sec.
---------------------------------------------------------------
Test: Test Xrender (offscreen) doing non-scaled Over blends
Time: 5.149 sec.
---------------------------------------------------------------
Test: Test Imlib2 doing non-scaled Over blends
Time: 6.237 sec.

After:

*** ROUND 1 ***
---------------------------------------------------------------
Test: Test Xrender doing non-scaled Over blends
Time: 4.947 sec.
---------------------------------------------------------------
Test: Test Xrender (offscreen) doing non-scaled Over blends
Time: 4.487 sec.
---------------------------------------------------------------
Test: Test Imlib2 doing non-scaled Over blends
Time: 6.235 sec.

12 years agotest: Occasionally use a BILINEAR filter in blitters-test
Søren Sandmann Pedersen [Mon, 5 Sep 2011 18:43:25 +0000 (14:43 -0400)]
test: Occasionally use a BILINEAR filter in blitters-test

To test that reductions of BILINEAR->NEAREST for identity
transformations happen correctly, occasionally use a bilinear filter
in blitters test.

12 years agotest: better coverage for BILINEAR->NEAREST filter optimization
Siarhei Siamashka [Sun, 22 May 2011 19:16:38 +0000 (22:16 +0300)]
test: better coverage for BILINEAR->NEAREST filter optimization

The upcoming optimization which is going to be able to replace BILINEAR filter
with NEAREST where appropriate needs to analyze the transformation matrix
and not to make any mistakes.

The changes to affine-test include:
1. Higher chance of using the same scale factor for x and y axes. This can help
   to stress some special cases (for example the case when both x and y scale
   factors are integer). The same applies to x/y translation.
2. Introduced a small chance for "corrupting" transformation matrix by flipping
   random bits. This supposedly can help to identify the cases when some of the
   fast paths or other code logic is wrongly activated due to insufficient checks.

12 years agoEliminate compute_sample_extents() function
Søren Sandmann Pedersen [Mon, 5 Sep 2011 04:19:51 +0000 (00:19 -0400)]
Eliminate compute_sample_extents() function

In analyze_extents(), instead of calling compute_sample_extents() call
compute_transformed_extents() and inline the remaining part of
compute_sample_extents(). The upcoming bilinear->nearest optimization
will do something different with these two pieces of code.

12 years agoSplit computation of sample area into own function
Søren Sandmann Pedersen [Sun, 4 Sep 2011 21:43:29 +0000 (17:43 -0400)]
Split computation of sample area into own function

compute_sample_extents() have two parts: one that computes the
transformed extents, and one that checks whether the computed extents
fit within the 16.16 coordinate space.

Split the first part into its own function
compute_transformed_extents().

12 years agoRemove x and y coordinates from analyze_extents() and compute_sample_extents()
Søren Sandmann Pedersen [Sun, 4 Sep 2011 21:17:53 +0000 (17:17 -0400)]
Remove x and y coordinates from analyze_extents() and compute_sample_extents()

These coordinates were only ever used for subtracting from the extents
box to put it into the coordinate space of the image, so we might as
well do this coordinate translation only once before entering the
functions.

12 years agoUse MAKE_ACCESSORS() to generate accessors for paletted formats
Søren Sandmann Pedersen [Tue, 16 Aug 2011 10:13:59 +0000 (06:13 -0400)]
Use MAKE_ACCESSORS() to generate accessors for paletted formats

Add support in convert_pixel_from_a8r8g8b8() and
convert_pixel_to_a8r8g8b8() for conversion to/from paletted formats,
then use MAKE_ACCESSORS() to generate accessors for the indexed
formats: c8, g8, g4, c4, g1

12 years agoUse MAKE_ACCESSORS() to generate accessors for the a1 format.
Søren Sandmann Pedersen [Sun, 30 May 2010 16:36:58 +0000 (12:36 -0400)]
Use MAKE_ACCESSORS() to generate accessors for the a1 format.

Add FETCH_1 and STORE_1 macros and use them to add support for 1bpp
pixels to fetch_and_convert_pixel() and convert_and_store_pixel(),
then use MAKE_ACCESSORS() to generate the accessors for the a1
format. (Not the g1 format as it is indexed).

12 years agoUse MAKE_ACCESSORS() to generate accessors for 24bpp formats
Søren Sandmann Pedersen [Tue, 16 Aug 2011 18:38:44 +0000 (14:38 -0400)]
Use MAKE_ACCESSORS() to generate accessors for 24bpp formats

Add FETCH_24 and STORE_24 macros and use them to add support for 24bpp
pixels in fetch_and_convert_pixel() and
convert_and_store_pixel(). Then use MAKE_ACCESSORS() to generate
accessors for the 24 bpp formats:

    r8g8b8
    b8g8r8

12 years agoUse MAKE_ACCESSORS() to generate accessors for 4 bpp RGB formats
Søren Sandmann Pedersen [Thu, 18 Aug 2011 09:09:07 +0000 (05:09 -0400)]
Use MAKE_ACCESSORS() to generate accessors for 4 bpp RGB formats

Use FETCH_4 and STORE_4 macros to add support for 4bpp pixels to
fetch_and_convert_pixel() and convert_and_store_pixel(), then use
MAKE_ACCESSORS() to generate accessors for 4 bpp formats, except g4 and
c4 which are indexed:

    a4
    r1g2b1
    b1g2r1
    a1r1g1b1
    a1b1g1r1

12 years agoUse MAKE_ACCESSORS() to generate accessors for 8bpp RGB formats
Søren Sandmann Pedersen [Thu, 18 Aug 2011 12:13:58 +0000 (08:13 -0400)]
Use MAKE_ACCESSORS() to generate accessors for 8bpp RGB formats

Add support for 8 bpp formats to fetch_and_convert_pixel() and
convert_and_store_pixel(), then use MAKE_ACCESSORS() to generate the
accessors for all the 8 bpp formats, except g8 and c8, which are
indexed:

    a8
    r3g3b2
    b2g3r3
    a2r2g2b2
    a2b2g2r2
    x4a4

12 years agoUse MAKE_ACCESSORS() to generate accessors for all the 16bpp formats
Søren Sandmann Pedersen [Thu, 18 Aug 2011 12:13:44 +0000 (08:13 -0400)]
Use MAKE_ACCESSORS() to generate accessors for all the 16bpp formats

Add support for 16bpp pixels to fetch_and_convert_pixel() and
convert_and_store_pixel(), then use MAKE_ACCESSORS() to generate
accessors for all the 16bpp formats:

    r5g6b5
    b5g6r5
    a1r5g5b5
    x1r5g5b5
    a1b5g5r5
    x1b5g5r5
    a4r4g4b4
    x4r4g4b4
    a4b4g4r4
    x4b4g4r4

12 years agoUse MAKE_ACCESSORS() to generate all the 32 bit accessors
Søren Sandmann Pedersen [Thu, 18 Aug 2011 12:13:30 +0000 (08:13 -0400)]
Use MAKE_ACCESSORS() to generate all the 32 bit accessors

Add support for 32bpp formats in fetch_and_convert_pixel() and
convert_and_store_pixel(), then use MAKE_ACCESSORS() to generate
accessors for all the 32 bpp formats:

    a8r8g8b8
    x8r8g8b8
    a8b8g8r8
    x8b8g8r8
    x14r6g6b6
    b8g8r8a8
    b8g8r8x8
    r8g8b8x8
    r8g8b8a8

12 years agoAdd initial version of the MAKE_ACCESSORS() macro
Søren Sandmann Pedersen [Wed, 17 Aug 2011 21:27:58 +0000 (17:27 -0400)]
Add initial version of the MAKE_ACCESSORS() macro

This macro will eventually allow the fetchers and storers to be
generated automatically. For now, it's just a skeleton that doesn't
actually do anything.

12 years agoAdd general pixel converter
Søren Sandmann Pedersen [Mon, 15 Aug 2011 22:42:38 +0000 (18:42 -0400)]
Add general pixel converter

This function can convert between any <= 32 bpp formats. Nothing uses
it yet.

12 years agoAdd a generic unorm_to_unorm() conversion utility
Søren Sandmann Pedersen [Mon, 15 Aug 2011 14:22:05 +0000 (10:22 -0400)]
Add a generic unorm_to_unorm() conversion utility

This function can convert between normalized numbers of different
depths. When converting to higher bit depths, it will replicate the
existing bits, when converting to lower bit depths, it will simply
truncate.

This function replaces the expand16() function in pixman-utils.c

12 years agoA few tweaks to a comment in pixman-combine.c.template
Søren Sandmann Pedersen [Mon, 19 Sep 2011 13:08:33 +0000 (09:08 -0400)]
A few tweaks to a comment in pixman-combine.c.template

Include a link to

http://marc.info/?l=xfree-render&m=99792000027857&w=2

where Keith explains how the disjoint/conjoint operators work.

12 years agoFix build on cygwin after commit efdf65c0c4fff551fb3cd9104deda9adb6261e22
Jon TURNEY [Mon, 19 Sep 2011 10:17:58 +0000 (06:17 -0400)]
Fix build on cygwin after commit efdf65c0c4fff551fb3cd9104deda9adb6261e22

libutils depends on pixman and so needs to preceed it in the link order

Found by tinderbox, see [1]

[1] http://tinderbox.freedesktop.org/builds/2011-09-15-0005/logs/pixman/#build

Signed-off-by: Jon TURNEY <jon.turney at dronecode.org.uk>
12 years agotest: Use smaller boxes in region_contains_test()
Søren Sandmann Pedersen [Tue, 13 Sep 2011 03:17:39 +0000 (23:17 -0400)]
test: Use smaller boxes in region_contains_test()

The boxes used region_contains_test() sometimes overflow causing

    *** BUG ***
    In pixman_region32_union_rect: Invalid rectangle passed
    Set a breakpoint on '_pixman_log_error' to debug

messages to be printed when pixman is compiled with DEBUG. Fix this by
dividing the x, y, w, h coordinates by 4 to prevent overflows.

12 years agobuild-win32: Add 'check' target
Andrea Canciani [Sun, 4 Sep 2011 19:33:05 +0000 (21:33 +0200)]
build-win32: Add 'check' target

On win32 the tests are built but they are not run automatically by the
build system.

A minimal 'check' target (depending on the tests being built) can
simply run them and log to the console their success/failure.

12 years agotest: Do not include config.h unless HAVE_CONFIG_H is defined
Andrea Canciani [Sun, 4 Sep 2011 20:52:53 +0000 (13:52 -0700)]
test: Do not include config.h unless HAVE_CONFIG_H is defined

The win32 build system does not generate config.h and correctly runs
the compiler without defining HAVE_CONFIG_H. Nevertheless some files
include config.h without checking for its availability, breaking the
build from a clean directory:

test\utils.h(2) : fatal error C1083: Cannot open include file:
'config.h': No such file or directory
...

12 years agobuild-win32: Add root Makefile.win32
Andrea Canciani [Sun, 4 Sep 2011 19:56:20 +0000 (21:56 +0200)]
build-win32: Add root Makefile.win32

Add Makefile.win32 to the pixman root. This makefile can recursively
run the other ones to compile the library or the test suite.

12 years agobuild-win32: Share targets and variables across win32 makefiles
Andrea Canciani [Sun, 4 Sep 2011 16:00:38 +0000 (18:00 +0200)]
build-win32: Share targets and variables across win32 makefiles

The win32 build system repeatedly defines some basic variables
(notably program names and flags) and C sources compilation rules.

They can be factored out to a common Makefile, to be included in every
other Makefile.win32.

12 years agobuild: Reuse test sources
Andrea Canciani [Sun, 4 Sep 2011 18:07:42 +0000 (20:07 +0200)]
build: Reuse test sources

Makefile.am and Makefile.win32 should not duplicate content, as this
leads to breaking the build when they are not kept in sync.

This can be avoided by listing sources, headers and common build
variables/rules in a Makefile.sources file.

In order to further simplify the test makefiles, the utility functions
are now in a static library, which gets linked to all the tests and
benchmarks.

12 years agobuild: Reuse sources and pixman-combine build rules
Andrea Canciani [Sun, 4 Sep 2011 16:41:41 +0000 (09:41 -0700)]
build: Reuse sources and pixman-combine build rules

Makefile.am and Makefile.win32 should not duplicate content, as this
leads to breaking the build when they are not kept in sync.

This can be avoided by listing sources, headers and common build
variables/rules in a Makefile.sources file.

12 years agotest: Fix compilation on win32
Andrea Canciani [Sun, 4 Sep 2011 18:07:57 +0000 (20:07 +0200)]
test: Fix compilation on win32

Adding scaling-helpers-test to the testsuite on win32 makes MSVC
complain about int64_t being used as an expression:

scaling-helpers-test.c(27) : error C2275: 'int64_t' : illegal use of
this type as an expression

12 years agoUse pkg-config to determine the flags to use with libpng
Søren Sandmann Pedersen [Sun, 11 Sep 2011 23:44:06 +0000 (19:44 -0400)]
Use pkg-config to determine the flags to use with libpng

Previously we would unconditionally link with -lpng leading to build
failures on systems without libpng.

12 years agotest: New function to save a pixman image to .png
Søren Sandmann Pedersen [Tue, 22 Feb 2011 10:20:36 +0000 (05:20 -0500)]
test: New function to save a pixman image to .png

When debugging it is often very useful to be able to save an image as
a png file. This commit adds a function "write_png()" that does that.

If libpng is not available, then the function becomes a noop.

12 years agoPost-release version bump to 0.23.5
Søren Sandmann Pedersen [Sat, 10 Sep 2011 03:59:20 +0000 (23:59 -0400)]
Post-release version bump to 0.23.5

12 years agoPre-release version bump to 0.23.4
Søren Sandmann Pedersen [Sat, 10 Sep 2011 03:51:11 +0000 (23:51 -0400)]
Pre-release version bump to 0.23.4

12 years agobits: optimise fetching width==1 repeats
Chris Wilson [Mon, 22 Aug 2011 14:29:25 +0000 (15:29 +0100)]
bits: optimise fetching width==1 repeats

Profiling ign.com, 20% of the entire render time was absorbed in this
single operation:

<< /content //COLOR_ALPHA /width 480 /height 800 >> surface context
<< /width 1 /height 677 /format //ARGB32 /source <|!!!@jGb!m5gD']#$jFHGWtZcK&2i)Up=!TuR9`G<8;ZQp[FQk;emL9ibhbEL&NTh-j63LhHo$E=mSG,0p71`cRJHcget4%<S\X+~> >> image pattern
  //EXTEND_REPEAT set-extend
  set-source
n 0 0 480 677 rectangle
fill+
pop

which is a simple composition of a single pixel wide image. Sadly this
is a workaround for lack of independent repeat-x/y handling in cairo and
pixman. Worse still is that the worst-case behaviour of the general repeat
path is for width 1 images...

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
12 years agoARM: NEON better instruction scheduling of over_n_8888
Taekyun Kim [Fri, 19 Aug 2011 12:20:08 +0000 (21:20 +0900)]
ARM: NEON better instruction scheduling of over_n_8888

New head, tail, tail/head blocks are added and instructions
are reordered to eliminate pipeline stalls

Performance numbers of before/after

- cortex a8 -
before : L1: 375.39  L2: 391.93  M:114.39 ( 40.99%)  HT: 99.37  VT: 98.20  R: 90.24  RT: 32.87 ( 240Kops/s)
after  : L1: 481.90  L2: 483.46  M:114.29 ( 40.69%)  HT:106.91  VT: 93.38  R: 90.74  RT: 29.51 ( 236Kops/s)

- cortex a9 -
before : L1: 324.50  L2: 332.79  M:155.55 ( 47.51%)  HT:111.93  VT: 93.58  R: 71.92  RT: 28.21 ( 233Kops/s)
after  : L1: 355.87  L2: 364.49  M:156.90 ( 47.59%)  HT:111.52  VT: 91.76  R: 72.16  RT: 28.22 ( 234Kops/s)

12 years agoARM: NEON better instruction scheduling of over_n_8_8888
Taekyun Kim [Tue, 23 Aug 2011 06:00:11 +0000 (15:00 +0900)]
ARM: NEON better instruction scheduling of over_n_8_8888

tail/head block is expanded and reordered to eliminate stalls

Performance numbers of before/after

- cortex a8 -
before : L1: 201.35  L2: 190.48  M:101.94 ( 54.85%)  HT: 78.41  VT: 63.83  R: 58.25  RT: 21.74 ( 191Kops/s)
after  : L1: 257.65  L2: 255.49  M:102.04 ( 55.33%)  HT: 79.19  VT: 65.46  R: 59.23  RT: 21.12 ( 189Kops/s)

- cortex a9 -
before : L1: 157.35  L2: 159.81  M:133.00 ( 60.94%)  HT: 82.44  VT: 63.64  R: 51.66  RT: 19.15 ( 179Kops/s)
after  : L1: 216.83  L2: 219.40  M:135.83 ( 61.80%)  HT: 85.60  VT: 64.80  R: 52.23  RT: 19.16 ( 179Kops/s)

12 years agoWorkaround bug in llvm-gcc
Andrea Canciani [Sat, 13 Aug 2011 14:18:17 +0000 (16:18 +0200)]
Workaround bug in llvm-gcc

llvm-gcc (shipped in Apple XCode 4.1.1 as the default compiler or in
the 2.9 release of LLVM) performs an invalid optimization which
unifies the empty_region and the bad_region structures because they
have the same content.

A bugreport has been filed against Apple Developers Tool for this
issue. This commit works around this bug by making one of the two
structures volatile, so that it cannot be merged.

Fixes region-contains-test.

12 years agowin32: Build benchmarks
Andrea Canciani [Wed, 29 Jun 2011 12:14:38 +0000 (14:14 +0200)]
win32: Build benchmarks

Add the makefile rules needed to compile lowlevel-blt-bench on win32
and fix the compilation errors.

12 years agoMove bilinear interpolation to pixman-inlines.h
Søren Sandmann Pedersen [Fri, 11 Mar 2011 22:09:34 +0000 (17:09 -0500)]
Move bilinear interpolation to pixman-inlines.h

12 years agoUse repeat() function from pixman-inlines.h in pixman-bits-image.c
Søren Sandmann Pedersen [Fri, 11 Mar 2011 21:09:21 +0000 (16:09 -0500)]
Use repeat() function from pixman-inlines.h in pixman-bits-image.c

The repeat() functionality was duplicated between pixman-bits-image.c
and pixman-inlines.h

12 years agoRename pixman-fast-path.h to pixman-inlines.h
Søren Sandmann Pedersen [Fri, 11 Mar 2011 21:07:24 +0000 (16:07 -0500)]
Rename pixman-fast-path.h to pixman-inlines.h

It is not really specific to pixman-fast-path.c.

12 years agoIn pixman_image_create_bits() allow images larger than 2GB
Søren Sandmann Pedersen [Thu, 11 Aug 2011 10:30:43 +0000 (06:30 -0400)]
In pixman_image_create_bits() allow images larger than 2GB

There is no reason for pixman_image_create_bits() to check that the
image size fits in int32_t. The correct check is against size_t since
that is what the argument to calloc() is.

This patch fixes this by adding a new _pixman_multiply_overflows_size()
and using it in create_bits(). Also prepend an underscore to the names
of other similar functions since they are internal to pixman.

V2: Use int, not ssize_t for the arguments in create_bits() since
width/height are still limited to 32 bits, as pointed out by Chris
Wilson.

12 years agoDon't include stdint.h in lowlevel-blt-bench.c
Søren Sandmann Pedersen [Mon, 8 Aug 2011 14:18:07 +0000 (10:18 -0400)]
Don't include stdint.h in lowlevel-blt-bench.c

Some systems don't have the file, and the types are already defined in
pixman.h.

https://bugs.freedesktop.org//show_bug.cgi?id=37422

12 years agoUse find_box_for_y() in pixman_region_contains_point() too
Søren Sandmann Pedersen [Tue, 2 Aug 2011 07:03:48 +0000 (03:03 -0400)]
Use find_box_for_y() in pixman_region_contains_point() too

The same binary search from the previous commit can be used in this
function too.

V2: Remove check from loop that is not needed anymore, pointed out by
Andrea Canciani.

12 years agoSpeed up pixman_region{,32}_contains_rectangle()
Søren Sandmann Pedersen [Tue, 2 Aug 2011 02:32:09 +0000 (22:32 -0400)]
Speed up pixman_region{,32}_contains_rectangle()

When someone selects some text in Firefox under a non-composited X
server and initiates a drag, a shaped window is created with a complex
shape corresponding to the outline of the text. Then, on every mouse
movement pixman_region_contains_rectangle() is called many times on
that complicated region. And pixman_region_contains_rectangle() is
doing a linear scan through the rectangles in the region, although the
scan does exit when it finds the first box that can't possibly
intersect the passed-in rectangle.

This patch changes the loop so that it uses a binary search to skip
boxes that don't overlap the current y position.  The performance
improvement for the text dragging case is easily noticable.

V2: Use the binary search for the "getting up to speed or skippping
remainder of band" as well.

12 years agoNew test of pixman_region_contains_{rectangle,point}
Søren Sandmann Pedersen [Tue, 2 Aug 2011 05:32:15 +0000 (01:32 -0400)]
New test of pixman_region_contains_{rectangle,point}

This test generates random regions and checks whether random boxes and
points are contained within them. The results are combined and a CRC32
value is computed and compared to a known-correct one.

12 years agoFix lcg_rand_u32() to return 32 random bits.
Søren Sandmann Pedersen [Wed, 3 Aug 2011 22:38:20 +0000 (18:38 -0400)]
Fix lcg_rand_u32() to return 32 random bits.

The lcg_rand() function only returns 15 random bits, so lcg_rand_u32()
would always have 0 in bit 31 and bit 15. Fix that by calling
lcg_rand() three times, to generate 15, 15, and 2 random bits
respectively.

V2: Use the 10/11 most significant bits from the 3 lcg results and mix
them with the low ones from the adjacent one, as suggested by Andrea
Canciani.

12 years agoARM NEON: Standard fast path out_reverse_8_8888
Taekyun Kim [Thu, 4 Aug 2011 13:21:04 +0000 (22:21 +0900)]
ARM NEON: Standard fast path out_reverse_8_8888

This fast path is frequently used by cairo to do polygon rendering.
Existing NEON code generation framework is used.

12 years agoradial: Fix typos and trailing whitespace
Andrea Canciani [Mon, 18 Jul 2011 06:15:23 +0000 (08:15 +0200)]
radial: Fix typos and trailing whitespace

Correct a typo reported by James Cloos and some reported by automatic
spellchecking.

Remove trailing whitespace.

12 years agoARM: workaround binutils bug #12931 (code sections alignment)
Siarhei Siamashka [Fri, 22 Jul 2011 21:27:34 +0000 (00:27 +0300)]
ARM: workaround binutils bug #12931 (code sections alignment)

More details in binutils bugtracker:
  http://sourceware.org/bugzilla/show_bug.cgi?id=12931

The problem was encountered in the wild by Mozilla:
  https://bugzilla.mozilla.org/show_bug.cgi?id=672787

12 years agoC fast path for scaled src_x888_8888 with nearest filter
Siarhei Siamashka [Fri, 15 Jul 2011 20:35:21 +0000 (23:35 +0300)]
C fast path for scaled src_x888_8888 with nearest filter

The necessity is justified by a message in the pixman mailing list:
  http://lists.freedesktop.org/archives/pixman/2011-July/001330.html

NONE repeat is not supported, but could be added by tweaking
the interpretation and making use of 'fully_transparent_src'
scanline function argument.

13 years agoradial: Improve documentation and naming
Andrea Canciani [Fri, 15 Jul 2011 20:02:01 +0000 (22:02 +0200)]
radial: Improve documentation and naming

Add a comment to explain why the tests guarantee that the code always
computes the greatest valid root.

Rename "det" as "discr" to make it match the mathematical name
"discriminant".

Based on a patch by Jeff Muizelaar <jmuizelaar@mozilla.com>.

13 years agoMakefile.am: Add pixman@lists.freedesktop.org to RELEASE_ANNOUNCE_LIST
Søren Sandmann Pedersen [Mon, 4 Jul 2011 19:55:52 +0000 (15:55 -0400)]
Makefile.am: Add pixman@lists.freedesktop.org to RELEASE_ANNOUNCE_LIST

13 years agoPost-release version bump to 0.23.3
Søren Sandmann Pedersen [Mon, 4 Jul 2011 19:35:17 +0000 (15:35 -0400)]
Post-release version bump to 0.23.3