diff options
author | senorblanco@chromium.org <senorblanco@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-05-13 13:16:52 +0000 |
---|---|---|
committer | senorblanco@chromium.org <senorblanco@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-05-13 13:16:52 +0000 |
commit | d5282e72b2da0b27b2a487d8376c44ad795736dc (patch) | |
tree | 7296767f5fcfa3b1cdd46f122e14dfaca1dd7e4a /skia/config/SkUserConfig.h | |
parent | 3ac3f51f8261910dbc38ccfa4cdf3d491ba7ecdb (diff) | |
download | chromium_src-d5282e72b2da0b27b2a487d8376c44ad795736dc.zip chromium_src-d5282e72b2da0b27b2a487d8376c44ad795736dc.tar.gz chromium_src-d5282e72b2da0b27b2a487d8376c44ad795736dc.tar.bz2 |
This CL updates chrome to the latest version of skia, retrieved via DEPS, and
placed in third_party. All relevant skia changes (for all 3 platforms) have
been upstreamed.
Most of this CL is mind-numbingly repetitive. Things of interest are: skia.gyp
(now points at third_party versions), DEPS, and SkUserConfig.h. stdint.h: Skia
now requires C99 integer types, which MSVC doesn't support natively. I have put
typedefs in config/win/stdint.h.
Note that the new version of skia appears to render rects whose coordinates
are "backwards" (ie., x2 < x1 or y2 < y1), which were formerly culled. There
were a couple obvious instances of this in the code which I fixed, but there may
be more.
There were ~35 layout test failures due to minor pixel differences which I
rebaselined on Windows and Linux, and 8 genuine failures related to masks and
stroked text, which I have put in text_expectations.txt and assigned to
myself. (There was another change which broke ~1700 tests on each platform,
but I put that change behind an #ifdef for now).
R=brettw
Review URL: http://codereview.chromium.org/65012
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@15949 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'skia/config/SkUserConfig.h')
-rw-r--r-- | skia/config/SkUserConfig.h | 222 |
1 files changed, 222 insertions, 0 deletions
diff --git a/skia/config/SkUserConfig.h b/skia/config/SkUserConfig.h new file mode 100644 index 0000000..f9ba042 --- /dev/null +++ b/skia/config/SkUserConfig.h @@ -0,0 +1,222 @@ +/* + * Copyright (C) 2006 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef SkUserConfig_DEFINED +#define SkUserConfig_DEFINED + +/* SkTypes.h, the root of the public header files, does the following trick: + + #include <SkPreConfig.h> + #include <SkUserConfig.h> + #include <SkPostConfig.h> + + SkPreConfig.h runs first, and it is responsible for initializing certain + skia defines. + + SkPostConfig.h runs last, and its job is to just check that the final + defines are consistent (i.e. that we don't have mutually conflicting + defines). + + SkUserConfig.h (this file) runs in the middle. It gets to change or augment + the list of flags initially set in preconfig, and then postconfig checks + that everything still makes sense. + + Below are optional defines that add, subtract, or change default behavior + in Skia. Your port can locally edit this file to enable/disable flags as + you choose, or these can be delared on your command line (i.e. -Dfoo). + + By default, this include file will always default to having all of the flags + commented out, so including it will have no effect. +*/ + +/////////////////////////////////////////////////////////////////////////////// + +/* Scalars (the fractional value type in skia) can be implemented either as + floats or 16.16 integers (fixed). Exactly one of these two symbols must be + defined. +*/ +//#define SK_SCALAR_IS_FLOAT +//#define SK_SCALAR_IS_FIXED + + +/* Somewhat independent of how SkScalar is implemented, Skia also wants to know + if it can use floats at all. Naturally, if SK_SCALAR_IS_FLOAT is defined, + then so muse SK_CAN_USE_FLOAT, but if scalars are fixed, SK_CAN_USE_FLOAT + can go either way. + */ +//#define SK_CAN_USE_FLOAT + +/* For some performance-critical scalar operations, skia will optionally work + around the standard float operators if it knows that the CPU does not have + native support for floats. If your environment uses software floating point, + define this flag. + */ +//#define SK_SOFTWARE_FLOAT + + +/* Skia has lots of debug-only code. Often this is just null checks or other + parameter checking, but sometimes it can be quite intrusive (e.g. check that + each 32bit pixel is in premultiplied form). This code can be very useful + during development, but will slow things down in a shipping product. + + By default, these mutually exclusive flags are defined in SkPreConfig.h, + based on the presence or absence of NDEBUG, but that decision can be changed + here. + */ +//#define SK_DEBUG +//#define SK_RELEASE + + +/* If, in debugging mode, Skia needs to stop (presumably to invoke a debugger) + it will call SK_CRASH(). If this is not defined it, it is defined in + SkPostConfig.h to write to an illegal address + */ +//#define SK_CRASH() *(int *)(uintptr_t)0 = 0 + + +/* preconfig will have attempted to determine the endianness of the system, + but you can change these mutually exclusive flags here. + */ +//#define SK_CPU_BENDIAN +//#define SK_CPU_LENDIAN + + +/* Some compilers don't support long long for 64bit integers. If yours does + not, define this to the appropriate type. + */ +//#define SkLONGLONG int64_t + + +/* Some envorinments do not suport writable globals (eek!). If yours does not, + define this flag. + */ +//#define SK_USE_RUNTIME_GLOBALS + + +/* To write debug messages to a console, skia will call SkDebugf(...) following + printf conventions (e.g. const char* format, ...). If you want to redirect + this to something other than printf, define yours here + */ +//#define SkDebugf(...) MyFunction(__VA_ARGS__) + + +/* If SK_DEBUG is defined, then you can optionally define SK_SUPPORT_UNITTEST + which will run additional self-tests at startup. These can take a long time, + so this flag is optional. + */ +#ifdef SK_DEBUG +#define SK_SUPPORT_UNITTEST +#endif + +// ===== Begin Chrome-specific definitions ===== + +#define SK_SCALAR_IS_FLOAT +#undef SK_SCALAR_IS_FIXED + +// Log the file and line number for assertions. +#define SkDebugf(...) SkDebugf_FileLine(__FILE__, __LINE__, false, __VA_ARGS__) +void SkDebugf_FileLine(const char* file, int line, bool fatal, + const char* format, ...); + +// Marking the debug print as "fatal" will cause a debug break, so we don't need +// a separate crash call here. +#define SK_DEBUGBREAK(cond) do { if (!(cond)) { \ + SkDebugf_FileLine(__FILE__, __LINE__, true, \ + "%s:%d: failed assertion \"%s\"\n", \ + __FILE__, __LINE__, #cond); } } while (false) + +#if defined(SK_BUILD_FOR_WIN32) + +#define SK_BUILD_FOR_WIN + +// VC8 doesn't support stdint.h, so we define those types here. +#define SK_IGNORE_STDINT_DOT_H +typedef signed char int8_t; +typedef unsigned char uint8_t; +typedef short int16_t; +typedef unsigned short uint16_t; +typedef int int32_t; +typedef unsigned uint32_t; +#define SK_A32_SHIFT 24 +#define SK_R32_SHIFT 16 +#define SK_G32_SHIFT 8 +#define SK_B32_SHIFT 0 + +// VC doesn't support __restrict__, so make it a NOP. +#undef SK_RESTRICT +#define SK_RESTRICT + +// Skia uses this deprecated bzero function to fill zeros into a string. +#define bzero(str, len) memset(str, 0, len) + +#elif defined(SK_BUILD_FOR_MAC) + +#define SK_CPU_LENDIAN +#undef SK_CPU_BENDIAN +// we want (memory order) RGBA +#define SK_A32_SHIFT 24 +#define SK_R32_SHIFT 0 +#define SK_G32_SHIFT 8 +#define SK_B32_SHIFT 16 + +#elif defined(SK_BUILD_FOR_UNIX) + +#ifdef SK_CPU_BENDIAN +// Below we set the order for ARGB channels in registers. I suspect that, on +// big endian machines, you can keep this the same and everything will work. +// The in-memory order will be different, of course, but as long as everything +// is reading memory as words rather than bytes, it will all work. However, if +// you find that colours are messed up I thought that I would leave a helpful +// locator for you. Also see the comments in +// base/gfx/bitmap_platform_device_linux.h +#error Read the comment at this location +#endif + +// For Linux we want to match the most common X visual, which is +// ARGB (in registers) +#define SK_A32_SHIFT 24 +#define SK_R32_SHIFT 16 +#define SK_G32_SHIFT 8 +#define SK_B32_SHIFT 0 + +#endif + +// The default crash macro writes to badbeef which can cause some strange +// problems. Instead, pipe this through to the logging function as a fatal +// assertion. +#define SK_CRASH() SkDebugf_FileLine(__FILE__, __LINE__, true, "SK_CRASH") + +// TODO(brettw) bug 6373: Re-enable Skia assertions. This is blocked on fixing +// some of our transparency handling which generates purposely-invalid colors, +// in turn causing assertions. +//#ifndef NDEBUG +// #define SK_DEBUG +// #undef SK_RELEASE + #undef SK_SUPPORT_UNITTEST // This is only necessary in debug mode since + // we've disabled assertions. When we re-enable + // them, this line can be removed. +//#else + #define SK_RELEASE + #undef SK_DEBUG +//#endif + +// For now (and to avoid rebaselining 1700+ tests), we'll use the old version +// of SkAlpha255To256. +#define SK_USE_OLD_255_TO_256 + +// ===== End Chrome-specific definitions ===== + +#endif |