cleanup - remove old cpu_memcpy and cpu_i64FromDouble that are no longer needed

This was SVN commit r8517.

source/graphics/Terrain.cpp

@@ -81,7 +81,7 @@ bool CTerrain::Initialize(ssize_t patchesPerSide,const u16* data)
 	// given a heightmap?
 	if (data) {
 		// yes; keep a copy of it
-		cpu_memcpy(m_Heightmap,data,m_MapSize*m_MapSize*sizeof(u16));
+		memcpy(m_Heightmap,data,m_MapSize*m_MapSize*sizeof(u16));
 	} else {
 		// build a flat terrain
 		memset(m_Heightmap,0,m_MapSize*m_MapSize*sizeof(u16));
@@ -364,7 +364,7 @@ void CTerrain::Resize(ssize_t size)
 	u16* dst=newHeightmap;
 	ssize_t copysize=std::min(newMapSize, m_MapSize);
 	for (ssize_t j=0;j<copysize;j++) {
-		cpu_memcpy(dst,src,copysize*sizeof(u16));
+		memcpy(dst,src,copysize*sizeof(u16));
 		dst+=copysize;
 		src+=m_MapSize;
 		if (newMapSize>m_MapSize) {
@@ -381,7 +381,7 @@ void CTerrain::Resize(ssize_t size)
 		src=newHeightmap+((m_MapSize-1)*newMapSize);
 		dst=src+newMapSize;
 		for (ssize_t i=0;i<newMapSize-m_MapSize;i++) {
-			cpu_memcpy(dst,src,newMapSize*sizeof(u16));
+			memcpy(dst,src,newMapSize*sizeof(u16));
 			dst+=newMapSize;
 		}
 	}
@@ -391,7 +391,7 @@ void CTerrain::Resize(ssize_t size)
 		for (ssize_t i=0;i<size;i++) {
 			// copy over texture data from existing tiles, if possible
 			if (i<m_MapSizePatches && j<m_MapSizePatches) {
-				cpu_memcpy(newPatches[j*size+i].m_MiniPatches,m_Patches[j*m_MapSizePatches+i].m_MiniPatches,sizeof(CMiniPatch)*PATCH_SIZE*PATCH_SIZE);
+				memcpy(newPatches[j*size+i].m_MiniPatches,m_Patches[j*m_MapSizePatches+i].m_MiniPatches,sizeof(CMiniPatch)*PATCH_SIZE*PATCH_SIZE);
 			}
 		}
 
@@ -460,7 +460,7 @@ void CTerrain::InitialisePatches()
 void CTerrain::SetHeightMap(u16* heightmap)
 {
 	// keep a copy of the given heightmap
-	cpu_memcpy(m_Heightmap,heightmap,m_MapSize*m_MapSize*sizeof(u16));
+	memcpy(m_Heightmap,heightmap,m_MapSize*m_MapSize*sizeof(u16));
 
 	// recalculate patch bounds, invalidate vertices
 	for (ssize_t j=0;j<m_MapSizePatches;j++) {

source/i18n/StrImmutable.h

@@ -54,7 +54,7 @@ namespace I18n
 			ref = new strImW_data;
 			size_t len = wcslen(s)+1;
 			ref->data = new wchar_t[len];
-			cpu_memcpy((void*)ref->data, s, len*sizeof(wchar_t));
+			memcpy((void*)ref->data, s, len*sizeof(wchar_t));
 		}
 
 		StrImW(const char* s)

source/lib/allocators/dynarray.cpp

@@ -189,7 +189,7 @@ LibError da_read(DynArray* da, void* data, size_t size)
 	if(da->pos+size > da->cur_size)
 		WARN_RETURN(ERR::FAIL);
 
-	cpu_memcpy(data, da->base+da->pos, size);
+	memcpy(data, da->base+da->pos, size);
 	da->pos += size;
 	return INFO::OK;
 }
@@ -198,7 +198,7 @@ LibError da_read(DynArray* da, void* data, size_t size)
 LibError da_append(DynArray* da, const void* data, size_t size)
 {
 	RETURN_ERR(da_reserve(da, size));
-	cpu_memcpy(da->base+da->pos, data, size);
+	memcpy(da->base+da->pos, data, size);
 	da->pos += size;
 	return INFO::OK;
 }

source/lib/allocators/string_pool.cpp

@@ -28,7 +28,6 @@
 #include "lib/allocators/string_pool.h"
 
 #include "lib/rand.h"
-#include "lib/sysdep/cpu.h"	// cpu_memcpy
 
 
 StringPool::StringPool(size_t maxSize)
@@ -63,7 +62,7 @@ const char* StringPool::UniqueCopy(const char* string)
 	char* uniqueCopy = (char*)pool_alloc(&m_pool, length+1);
 	if(!uniqueCopy)
 		throw std::bad_alloc();
-	cpu_memcpy((void*)uniqueCopy, string, length);
+	memcpy((void*)uniqueCopy, string, length);
 	uniqueCopy[length] = '\0';
 
 	m_map.insert(uniqueCopy, uniqueCopy);

source/lib/debug.cpp

@@ -72,7 +72,7 @@ void debug_wprintf_mem(const wchar_t* fmt, ...)
 	{
 		const size_t copySize = sizeof(wchar_t) * LOG_CHARS/2;
 		wchar_t* const middle = &debug_log[LOG_CHARS/2];
-		cpu_memcpy(debug_log, middle, copySize);
+		memcpy(debug_log, middle, copySize);
 		memset(middle, 0, copySize);
 		debug_log_pos -= LOG_CHARS/2;	// don't assign middle (may leave gap)
 	}

source/lib/file/archive/archive_zip.cpp

@@ -36,7 +36,6 @@
 #include "lib/fat_time.h"
 #include "lib/path_util.h"
 #include "lib/allocators/pool.h"
-#include "lib/sysdep/cpu.h"		// cpu_memcpy
 #include "lib/file/archive/archive.h"
 #include "lib/file/archive/codec_zlib.h"
 #include "lib/file/archive/stream.h"
@@ -81,7 +80,7 @@ public:
 		m_fn_len    = to_le16(u16_from_larger(pathnameLength));
 		m_e_len     = to_le16(0);
 
-		cpu_memcpy((char*)this + sizeof(LFH), pathname_c.string().c_str(), pathnameLength);
+		memcpy((char*)this + sizeof(LFH), pathname_c.string().c_str(), pathnameLength);
 	}
 
 	size_t Size() const
@@ -133,7 +132,7 @@ public:
 		m_x3        = to_le32(0);
 		m_lfh_ofs   = to_le32(u32_from_larger(ofs));
 
-		cpu_memcpy((char*)this + sizeof(CDFH), pathname_c.string().c_str(), pathnameLength);
+		memcpy((char*)this + sizeof(CDFH), pathname_c.string().c_str(), pathnameLength);
 	}
 
 	fs::wpath Pathname() const
@@ -332,7 +331,7 @@ private:
 		LFH_Copier* p = (LFH_Copier*)cbData;
 
 		debug_assert(size <= p->lfh_bytes_remaining);
-		cpu_memcpy(p->lfh_dst, block, size);
+		memcpy(p->lfh_dst, block, size);
 		p->lfh_dst += size;
 		p->lfh_bytes_remaining -= size;
 

source/lib/file/archive/codec_zlib.cpp

@@ -84,7 +84,7 @@ public:
 	virtual LibError Process(const u8* in, size_t inSize, u8* out, size_t outSize, size_t& inConsumed, size_t& outProduced)
 	{
 		const size_t transferSize = std::min(inSize, outSize);
-		cpu_memcpy(out, in, transferSize);
+		memcpy(out, in, transferSize);
 		inConsumed = outProduced = transferSize;
 		m_checksum = UpdateChecksum(m_checksum, out, outProduced);
 		return INFO::OK;

source/lib/file/io/io.cpp

@@ -24,7 +24,6 @@
 #include "lib/file/io/io.h"
 
 #include "lib/allocators/allocators.h"	// AllocatorChecker
-#include "lib/sysdep/cpu.h"	// cpu_memcpy
 #include "lib/file/file.h"
 #include "lib/file/common/file_stats.h"
 #include "lib/file/io/block_cache.h"
@@ -35,7 +34,7 @@ static const size_t ioDepth = 16;
 
 // the underlying aio implementation likes buffer and offset to be
 // sector-aligned; if not, the transfer goes through an align buffer,
-// and requires an extra cpu_memcpy.
+// and requires an extra memcpy.
 //
 // if the user specifies an unaligned buffer, there's not much we can
 // do - we can't assume the buffer contains padding. therefore,
@@ -138,7 +137,7 @@ public:
 				// copy from cache into user buffer
 				if(alignedBuf)
 				{
-					cpu_memcpy(alignedBuf, m_cachedBlock.get(), BLOCK_SIZE);
+					memcpy(alignedBuf, m_cachedBlock.get(), BLOCK_SIZE);
 					m_alignedBuf = alignedBuf;
 				}
 				// return cached block

source/lib/file/io/write_buffer.cpp

@@ -41,11 +41,11 @@ void WriteBuffer::Append(const void* data, size_t size)
 	{
 		m_capacity = round_up_to_pow2(m_size + size);
 		shared_ptr<u8> newData = io_Allocate(m_capacity);
-		cpu_memcpy(newData.get(), m_data.get(), m_size);
+		memcpy(newData.get(), m_data.get(), m_size);
 		m_data = newData;
 	}
 
-	cpu_memcpy(m_data.get() + m_size, data, size);
+	memcpy(m_data.get() + m_size, data, size);
 	m_size += size;
 }
 
@@ -53,7 +53,7 @@ void WriteBuffer::Append(const void* data, size_t size)
 void WriteBuffer::Overwrite(const void* data, size_t size, size_t offset)
 {
 	debug_assert(offset+size < m_size);
-	cpu_memcpy(m_data.get()+offset, data, size);
+	memcpy(m_data.get()+offset, data, size);
 }
 
 
@@ -93,7 +93,7 @@ LibError UnalignedWriter::Append(const u8* data, size_t size) const
 		}
 
 		const size_t chunkSize = std::min(size, BLOCK_SIZE-m_bytesUsed);
-		cpu_memcpy(m_alignedBuf.get()+m_bytesUsed, data, chunkSize);
+		memcpy(m_alignedBuf.get()+m_bytesUsed, data, chunkSize);
 		m_bytesUsed += chunkSize;
 		data += chunkSize;
 		size -= chunkSize;

source/lib/sysdep/acpi.cpp

@@ -248,7 +248,7 @@ static void* UnsafeAllocateAndCopyTable(PCV_u8 mem, size_t numBytes, void* arg)
 	if(!copy)
 		return FAILED;
 
-	cpu_memcpy((void*)copy, (const void*)mem, tableSize);
+	memcpy((void*)copy, (const void*)mem, tableSize);
 	return (void*)copy;
 }
 

source/lib/sysdep/arch/amd64/amd64.cpp

@@ -34,9 +34,4 @@ void cpu_ConfigureFloatingPoint()
 	// *guess* we don't need to do anything...
 }
 
-void* cpu_memcpy(void* RESTRICT dst, const void* RESTRICT src, size_t size)
-{
-	return memcpy(dst, src, size);
-}
-
 #endif // ARCH_AMD64

source/lib/sysdep/arch/ia32/ia32.cpp

@@ -145,9 +145,3 @@ void cpu_ConfigureFloatingPoint()
 	// results were changed significantly, so it had to be disabled.
 	//ia32_asm_control87(IA32_RC_CHOP, IA32_MCW_RC);
 }
-
-
-void* cpu_memcpy(void* RESTRICT dst, const void* RESTRICT src, size_t size)
-{
-	return memcpy(dst, src, size);
-}

source/lib/sysdep/cpu.h

@@ -110,23 +110,12 @@ inline void cpu_Pause()
 //-----------------------------------------------------------------------------
 // misc
 
-/**
- * drop-in replacement for POSIX memcpy().
- **/
-LIB_API void* cpu_memcpy(void* RESTRICT dst, const void* RESTRICT src, size_t size);
-
-
 /**
  * set the FPU control word to "desirable" values (see implementation)
  **/
 LIB_API void cpu_ConfigureFloatingPoint();
 
-// convert float to int much faster than _ftol2, which would normally be
-// used by (int) casts.
-// VC8 and GCC with -ffast-math now manage to generate SSE instructions,
-// so our implementation is no longer needed.
-#define cpu_i32FromFloat(f)  ((i32)(f))
-#define cpu_i32FromDouble(d) ((i32)(d))
-#define cpu_i64FromDouble(d) ((i64)(d))
+// NB: cpu_i64FromDouble et al. were faster than _ftol2, but are obsolete
+// since VC8 and GCC (with -ffast-math) generate SSE instructions.
 
 #endif	// #ifndef INCLUDED_CPU

source/lib/sysdep/os/win/wdll_delay_load.cpp

@@ -254,7 +254,7 @@ PinhFromImageBase(HMODULE hmod) {
 
 static inline void WINAPI
 OverlayIAT(PImgThunkData pitdDst, PCImgThunkData pitdSrc) {
-    cpu_memcpy(pitdDst, pitdSrc, CountOfImports(pitdDst) * sizeof IMAGE_THUNK_DATA);
+    memcpy(pitdDst, pitdSrc, CountOfImports(pitdDst) * sizeof IMAGE_THUNK_DATA);
     }
 
 static inline DWORD WINAPI

source/lib/sysdep/os/win/whrt/whrt.cpp

@@ -265,7 +265,7 @@ static inline LibError InitUpdateThread()
 	// make sure our interval isn't too long
 	// (counterBits can be 64 => Bit() would overflow => calculate period/2)
 	const double period_2 = Bit<u64>(counterBits-1) / nominalFrequency;
-	const size_t rolloversPerInterval = size_t(UPDATE_INTERVAL_MS / cpu_i64FromDouble(period_2*2.0*1000.0));
+	const size_t rolloversPerInterval = size_t(UPDATE_INTERVAL_MS / i64(period_2*2.0*1000.0));
 	debug_assert(rolloversPerInterval <= 1);
 
 	hExitEvent = CreateEvent(0, TRUE, FALSE, 0);	// manual reset, initially false

source/lib/sysdep/os/win/wposix/wtime.cpp

@@ -32,7 +32,6 @@
 #include "lib/sysdep/os/win/wposix/wtime.h"
 
 #include "lib/sysdep/os/win/wposix/wposix_internal.h"
-#include "lib/sysdep/cpu.h"				// cpu_i64FromDouble
 #include "lib/sysdep/os/win/whrt/whrt.h"
 
 WINIT_REGISTER_MAIN_INIT(wtime_Init);	// whrt -> wtime
@@ -96,7 +95,7 @@ static void LatchInitialSystemTime()
 // algorithm: add current HRT value to the startup system time
 static i64 CurrentSystemTime_ns()
 {
-	const i64 ns = stInitial_ns + cpu_i64FromDouble(whrt_Time() * _1e9);
+	const i64 ns = stInitial_ns + i64(whrt_Time() * _1e9);
 	return ns;
 }
 
@@ -134,7 +133,7 @@ int clock_getres(clockid_t clock, struct timespec* ts)
 	debug_assert(clock == CLOCK_REALTIME || clock == CLOCK_MONOTONIC);
 
 	const double resolution = whrt_Resolution();
-	const i64 ns = cpu_i64FromDouble(resolution * 1e9);
+	const i64 ns = i64(resolution * 1e9);
 	*ts = TimespecFromNs(ns);
 	return 0;
 }

source/lib/sysdep/tests/test_sysdep.h

@@ -40,24 +40,6 @@
 class TestSysdep : public CxxTest::TestSuite 
 {
 public:
-	void test_float_int()
-	{
-		TS_ASSERT_EQUALS(cpu_i32FromFloat(0.99999f), 0);
-		TS_ASSERT_EQUALS(cpu_i32FromFloat(1.0f), 1);
-		TS_ASSERT_EQUALS(cpu_i32FromFloat(1.01f), 1);
-		TS_ASSERT_EQUALS(cpu_i32FromFloat(5.6f), 5);
-
-		TS_ASSERT_EQUALS(cpu_i32FromDouble(0.99999), 0);
-		TS_ASSERT_EQUALS(cpu_i32FromDouble(1.0), 1);
-		TS_ASSERT_EQUALS(cpu_i32FromDouble(1.01), 1);
-		TS_ASSERT_EQUALS(cpu_i32FromDouble(5.6), 5);
-
-		TS_ASSERT_EQUALS(cpu_i64FromDouble(0.99999), 0LL);
-		TS_ASSERT_EQUALS(cpu_i64FromDouble(1.0), 1LL);
-		TS_ASSERT_EQUALS(cpu_i64FromDouble(1.01), 1LL);
-		TS_ASSERT_EQUALS(cpu_i64FromDouble(5.6), 5LL);
-	}
-
 	void test_round()
 	{
 		TS_ASSERT_EQUALS(rintf(0.99999f), 1.0f);

source/lib/tex/tex.cpp

@@ -191,7 +191,7 @@ static void create_level(size_t level, size_t level_w, size_t level_h, const u8*
 	if(level == 0)
 	{
 		debug_assert(level_data_size == cld->prev_level_data_size);
-		cpu_memcpy(dst, src, level_data_size);
+		memcpy(dst, src, level_data_size);
 	}
 	else
 	{
@@ -356,7 +356,7 @@ TIMER_ACCRUE(tc_plain_transform);
 	else
 	{
 		src = (const u8*)newData.get();
-		cpu_memcpy(newData.get(), data, data_size);
+		memcpy(newData.get(), data, data_size);
 	}
 
 	// no conversion necessary
@@ -366,7 +366,7 @@ TIMER_ACCRUE(tc_plain_transform);
 		{
 			for(size_t y = 0; y < h; y++)
 			{
-				cpu_memcpy(dst, src, pitch);
+				memcpy(dst, src, pitch);
 				dst += pitch;
 				src += row_ofs;
 			}

source/ps/FileIo.cpp

@@ -149,7 +149,7 @@ void CFileUnpacker::UnpackRaw(void* rawData, size_t rawDataSize)
 		throw PSERROR_File_UnexpectedEOF();
 
 	void* src = m_buf.get() + m_unpackPos;
-	cpu_memcpy(rawData, src, rawDataSize);
+	memcpy(rawData, src, rawDataSize);
 	m_unpackPos += rawDataSize;
 }
 

source/ps/Util.cpp

@@ -345,7 +345,7 @@ void WriteBigScreenshot(const std::wstring& extension, int tiles)
 			{
 				void* dest = (char*)img + ((tile_y*tile_h + y) * img_w + (tile_x*tile_w)) * bpp/8;
 				void* src = (char*)tile_data + y * tile_w * bpp/8;
-				cpu_memcpy(dest, src, tile_w * bpp/8);
+				memcpy(dest, src, tile_w * bpp/8);
 			}
 		}
 	}

source/ps/XML/XMLWriter.cpp

@@ -97,7 +97,7 @@ bool XMLWriter_File::StoreVFS(const PIVFS& vfs, const VfsPath& pathname)
 
 	const size_t size = m_Data.length();
 	shared_ptr<u8> data = io_Allocate(size);
-	cpu_memcpy(data.get(), m_Data.data(), size);
+	memcpy(data.get(), m_Data.data(), size);
 	LibError ret = vfs->CreateFile(pathname, data, size);
 	if (ret < 0)
 	{

source/renderer/VertexBuffer.cpp

@@ -207,7 +207,7 @@ void CVertexBuffer::AppendBatch(VBChunk* UNUSED(chunk),Handle texture,size_t num
 
 	// resize the chunk's batch to fit its indices
 	batch->m_IndexData.push_back(std::pair<size_t,u16*>(numIndices,indices));
-//	cpu_memcpy(&batch->m_Indices[0]+cursize,indices,sizeof(u16)*numIndices);
+//	memcpy(&batch->m_Indices[0]+cursize,indices,sizeof(u16)*numIndices);
 }
 
 
@@ -224,7 +224,7 @@ void CVertexBuffer::UpdateChunkVertices(VBChunk* chunk,void* data)
 		// if (glGetError() != GL_NO_ERROR) throw PSERROR_Renderer_VBOFailed();
 	} else {
 		debug_assert(m_SysMem);
-		cpu_memcpy(m_SysMem + chunk->m_Index * m_VertexSize, data, chunk->m_Count * m_VertexSize);
+		memcpy(m_SysMem + chunk->m_Index * m_VertexSize, data, chunk->m_Count * m_VertexSize);
 	}
 }
 

source/tools/atlas/GameInterface/Handlers/MiscHandlers.cpp

@@ -87,9 +87,9 @@ QUERYHANDLER(CinemaRecord)
 //* // TODO: BGR24 output doesn't need flipping, YUV420 and RGBA32 do
 		for (int y = 0; y < h/2; ++y)
 		{
-			cpu_memcpy(temp, &img[y*w*3], w*3);
-			cpu_memcpy(&img[y*w*3], &img[(h-1-y)*w*3], w*3);
-			cpu_memcpy(&img[(h-1-y)*w*3], temp, w*3);
+			memcpy(temp, &img[y*w*3], w*3);
+			memcpy(&img[y*w*3], &img[(h-1-y)*w*3], w*3);
+			memcpy(&img[(h-1-y)*w*3], temp, w*3);
 		}
 //*/