dhewg
661e8f453b
Stub out win32 CPU functions for MinGW
...
Most of these use MSVC style asm.
Binaries compiled with MinGW will not use any SIMD code for now.
2011-12-14 18:20:30 +01:00
dhewg
ed349d71cc
Fix Sys_QueMouseEvents for MinGW
...
DIMOFS_[X|Y|Z] can't be used in a switch statement.
2011-12-14 18:20:30 +01:00
dhewg
b8dc464dc7
Replace one unnecessary afxwin.h
...
No MFC required in win_taskkeyhook.cpp.
2011-12-14 18:20:30 +01:00
dhewg
f54752c209
Case corrections for includes on case sensitive fs'
2011-12-14 18:20:29 +01:00
dhewg
5c3a28e027
Basic infrastructure to build with MinGW
...
Add WIN32 support to CMake (core and games, no dedicated for now).
Only use MFC when it is available.
2011-12-14 18:19:38 +01:00
dhewg
36c5b05f1f
Remove the unused and ugly HackChkStk()
2011-12-14 02:40:55 +01:00
dhewg
5052f42c9e
Add a CMake build system
...
Tested on FreeBSD, Linux, and a ghetto rigged OSX i686 cross
compiler.
Find[OGG|Vorbis|VorbisFile].cmake borrowed from osgAudio
2011-12-14 02:40:49 +01:00
dhewg
41a1122a82
Remove the unused and useless CallStack code
...
This was never enabled and only implemented for Windows.
2011-12-14 02:39:42 +01:00
dhewg
a3cd1cf54b
Get rid of all #pragma mark
...
Fixes error: stray '\245' in program
2011-12-13 19:26:16 +01:00
dhewg
f81d32918f
Implement Sys_GetProcessorId for x86 and x86_64
...
Detect CPU features at runtime via cpuid - code borrowed from
libavutil.
Availability of cpuid is not checked since pentium3 supports it
and that was the minimum requirement anyway.
Only features enabled at compile time will be available.
Forced MMX/SSE/SSE2/SSE3 and it passed all tests via:
./doom3.x86_64 +disconnect +set s_noSound 1 +testSIMD
2011-12-13 19:26:16 +01:00
dhewg
18762483e6
Change return type of [Sys_]GetProcessorId to int
...
Fixes compile error: invalid conversion from 'int' to 'cpuid_t'
2011-12-13 18:57:30 +01:00
dhewg
8023bc5d56
Always compile all SIMD code
...
Protect all SIMD implementations with the according defines and
let the compiler decide if it supports the intructions.
Linux will still use Simd_Generic because CPU feature runtime
detection is missing.
2011-12-13 18:56:38 +01:00
dhewg
c15e761bae
Get rid of frustum code specialized for gcc
...
The functions are exactly the same ;)
Maybe splitted into two files because of an ICE on some ancient
gcc?
2011-12-13 16:26:16 +01:00
dhewg
4c38aaae78
Get rid of dmap code specialized for gcc
...
According to the file this existed because of a gcc 3.5.5
optimization bug, which we do not care anymore about.
2011-12-13 16:25:08 +01:00
dhewg
74acc03147
Enable OpenAL for all platforms
...
New scons variable 'OPENAL' (defaults to '1') to toggle the
OpenAL sound backend.
Enable via "s_useOpenAL 1".
2011-12-11 17:07:13 +01:00
dhewg
db65f7a61d
Use system OpenAL
2011-12-11 17:07:13 +01:00
dhewg
690967c276
Split EAX usage off OpenAL
...
New define ID_OPENAL_EAX to enable the deprecated EAX on top
of OpenAL.
OpenAL Soft can now be used.
2011-12-11 16:56:11 +01:00
dhewg
b54bc73532
Move Windows only OpenAL files to sys/win32/
2011-12-11 16:00:31 +01:00
dhewg
bd1fe41b0e
Move the OpenAL stub to sys/stub/
2011-12-11 14:44:30 +01:00
dhewg
a8b1d3a364
Fix POSIX net code for x86_64
...
Multiplayer working from 32bit dedicated server on FreeBSD with a
64bit Linux client.
2011-12-10 19:24:30 +01:00
dhewg
bdb21d20c8
Fix wrong usage of memset()
2011-12-10 15:36:17 +01:00
dhewg
82aaccc1ad
Use the local OSS headers
2011-12-10 15:36:16 +01:00
dhewg
1b8777a008
Get rid of GL_HARDLINK, always link against libGL
...
This kills the funky GL logging stubs, which unnecessarily
complicates the build process (think future cmake).
As for logging GL calls: Use apitrace for that
https://github.com/apitrace/apitrace
2011-12-10 15:36:16 +01:00
dhewg
9678020677
Fix compilation for *BSD
...
Change most __linux__ defines to __unix__ for all UNIX like systems.
2011-12-10 15:36:16 +01:00
dhewg
56bd47ac43
Fix Sys_FileTimeStamp return type
...
Sync with its header.
2011-12-10 15:36:15 +01:00
dhewg
5ee2e5b564
Fix and cleanup endianess defines
...
Use various defines set by the compiler to determine the
endianess of the target arch.
2011-12-10 15:36:15 +01:00
dhewg
04832863e8
Basic infrastructure to build on *BSD
2011-12-10 15:36:15 +01:00
dhewg
98eed69848
Fix -Wreturn-type warnings in the GL stub
...
no return statement in function returning non-void
2011-12-10 15:36:15 +01:00
dhewg
5c9973773b
Force exit on double segfault
...
All bets are off in this scenario, don't deinit anything.
2011-12-10 15:36:14 +01:00
dhewg
f9dd2f93eb
Fix platform dependent thread function definitions
...
Thread return types are different between platforms, and its
probably not a good idea to return something of a different size,
cast the callback and expect it to not crash.
2011-12-10 15:36:13 +01:00
dhewg
0c84e4bff3
Fix shared library loading for x86_64
...
Library handles are "void *" on posix.
2011-12-10 15:36:10 +01:00
dhewg
ab201cfc06
Stub Sys_GetClockTicks for x86_64
...
This is only used for debugging/stats.
2011-12-10 15:36:09 +01:00
dhewg
6b1e27b157
Trivial pointer cast fixes for x86_64
2011-12-10 15:36:09 +01:00
dhewg
2995bcab30
Basic infrastructure for native x86_64 builds
...
Let scons link directly in the build folder.
Multiple archs can coexist next to each other.
New scons variable "X86" to cross compile x86 binaries on x86_64.
2011-12-10 15:36:08 +01:00
dhewg
49859d747e
Comment XFlush() to prevent a crash on exit
2011-12-10 15:36:08 +01:00
dhewg
541f8f967a
Fix typo in define
...
Gets rid of the warning:
XTRHEADS not defined -- this libXNVCtrl.a will not be thread safe!
2011-12-10 15:36:07 +01:00
dhewg
6e89dac99e
Fix -Warray-bounds warning
2011-12-10 15:36:06 +01:00
dhewg
2556a3923f
Fix -Wunused-result warnings
...
ignoring return value of
Implement some error checking.
Use [put|get]char in the posix tty code for ease of use.
2011-12-10 15:36:05 +01:00
dhewg
e4771f3a5f
Fix -Wunused-variable warnings
...
unused variable
2011-12-10 15:36:03 +01:00
dhewg
7f4bcf94ca
Fix -Wunused-function warnings
...
defined but not used
2011-12-10 15:36:02 +01:00
dhewg
fc447ec426
Fix -Waddress warning
...
the address of ‘hwparams’ will always evaluate as ‘true’
This was only in idAudioHardwareALSA::Initialize(), add the
missing return statement too while I'm at it.
2011-12-10 15:36:01 +01:00
dhewg
09f03c531a
Fix -Wmissing-braces warnings
...
missing braces around initializer for ‘const char* [2]’
2011-12-10 15:36:01 +01:00
dhewg
79ad905e05
Fix all whitespace errors
...
Excluding 3rd party files.
2011-12-10 15:35:54 +01:00
dhewg
ff493f6847
Fix quoting in GPL headers
2011-12-10 15:34:48 +01:00
dhewg
adebc145e3
Delete foreign build systems
...
Eventually a cmake based build system will replace all these.
Also, interference with `git grep` can not be tolerated.
2011-12-10 15:34:41 +01:00
dhewg
afcd0a93ea
Use system libvorbisfile
2011-12-10 15:34:41 +01:00
dhewg
e704911fed
Use system libjpeg8
...
The unused WriteLossless() was using libjpeg internals found in
jpegint.h. This header is available with libjpeg8, but not shipped
by most distros. Disable the relevant code - no functional
changes.
2011-12-10 15:33:48 +01:00
dhewg
1b858a774f
Use system libcurl
2011-12-10 11:00:03 +01:00
Timothee 'TTimo' Besset
15597777fc
towards XCode 4 support - need those converted
2011-11-22 17:32:53 -06:00
Timothee 'TTimo' Besset
fb1609f554
hello world
2011-11-22 15:28:15 -06:00