FreeRDP/libfreerdp/primitives/prim_copy.c
Martin Fleisz 343d210a75 primitives: separating optimized functions into their own .c files.
This way we use certain compiler flags (like -msse3) only on files
containing optimized code. This avoids problems that occured when
using these flags compiling generic code and running it on platforms
that don't support these optimizations (i.e. NEON optimization on
ARM platforms).
2013-02-21 02:45:10 -08:00

181 lines
4.9 KiB
C

/* FreeRDP: A Remote Desktop Protocol Client
* Copy operations.
* vi:ts=4 sw=4:
*
* (c) Copyright 2012 Hewlett-Packard Development Company, L.P.
* 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.
*/
#ifdef HAVE_CONFIG_H
#include "config.h"
#endif
#include <string.h>
#include <freerdp/types.h>
#include <freerdp/primitives.h>
#ifdef WITH_IPP
# include <ipps.h>
# include <ippi.h>
#endif /* WITH_IPP */
#include "prim_internal.h"
/* ------------------------------------------------------------------------- */
/*static inline BOOL memory_regions_overlap_1d(*/
static BOOL memory_regions_overlap_1d(
const BYTE *p1,
const BYTE *p2,
size_t bytes)
{
const ULONG_PTR p1m = (const ULONG_PTR) p1;
const ULONG_PTR p2m = (const ULONG_PTR) p2;
if (p1m <= p2m)
{
if (p1m + bytes > p2m) return TRUE;
}
else
{
if (p2m + bytes > p1m) return TRUE;
}
/* else */
return FALSE;
}
/* ------------------------------------------------------------------------- */
/*static inline BOOL memory_regions_overlap_2d( */
static BOOL memory_regions_overlap_2d(
const BYTE *p1, int p1Step, int p1Size,
const BYTE *p2, int p2Step, int p2Size,
int width, int height)
{
ULONG_PTR p1m = (ULONG_PTR) p1;
ULONG_PTR p2m = (ULONG_PTR) p2;
if (p1m <= p2m)
{
ULONG_PTR p1mEnd = p1m + (height-1)*p1Step + width*p1Size;
if (p1mEnd > p2m) return TRUE;
}
else
{
ULONG_PTR p2mEnd = p2m + (height-1)*p2Step + width*p2Size;
if (p2mEnd > p1m) return TRUE;
}
/* else */
return FALSE;
}
/* ------------------------------------------------------------------------- */
pstatus_t general_copy_8u(
const BYTE *pSrc,
BYTE *pDst,
INT32 len)
{
if (memory_regions_overlap_1d(pSrc, pDst, (size_t) len))
{
memmove((void *) pDst, (const void *) pSrc, (size_t) len);
}
else
{
memcpy((void *) pDst, (const void *) pSrc, (size_t) len);
}
return PRIMITIVES_SUCCESS;
}
/* ------------------------------------------------------------------------- */
/* Copy a block of pixels from one buffer to another.
* The addresses are assumed to have been already offset to the upper-left
* corners of the source and destination region of interest.
*/
pstatus_t general_copy_8u_AC4r(
const BYTE *pSrc, INT32 srcStep,
BYTE *pDst, INT32 dstStep,
INT32 width, INT32 height)
{
primitives_t *prims = primitives_get();
const BYTE *src = (const BYTE *) pSrc;
BYTE *dst = (BYTE *) pDst;
int rowbytes = width * sizeof(UINT32);
if ((width == 0) || (height == 0)) return PRIMITIVES_SUCCESS;
if (memory_regions_overlap_2d(pSrc, srcStep, sizeof(UINT32),
pDst, dstStep, sizeof(UINT32), width, height))
{
do {
prims->copy(src, dst, rowbytes);
src += srcStep;
dst += dstStep;
} while (--height);
}
else
{
/* TODO: do it in one operation when the rowdata is adjacent. */
do {
/* If we find a replacement for memcpy that is consistently
* faster, this could be replaced with that.
*/
memcpy(dst, src, rowbytes);
src += srcStep;
dst += dstStep;
} while (--height);
}
return PRIMITIVES_SUCCESS;
}
#ifdef WITH_IPP
/* ------------------------------------------------------------------------- */
/* This is just ippiCopy_8u_AC4R without the IppiSize structure parameter. */
static pstatus_t ippiCopy_8u_AC4r(
const BYTE *pSrc, INT32 srcStep,
BYTE *pDst, INT32 dstStep,
INT32 width, INT32 height)
{
IppiSize roi;
roi.width = width;
roi.height = height;
return (pstatus_t) ippiCopy_8u_AC4R(pSrc, srcStep, pDst, dstStep, roi);
}
#endif /* WITH_IPP */
/* ------------------------------------------------------------------------- */
void primitives_init_copy(
const primitives_hints_t *hints,
primitives_t *prims)
{
/* Start with the default. */
prims->copy_8u = general_copy_8u;
prims->copy_8u_AC4r = general_copy_8u_AC4r;
/* Pick tuned versions if possible. */
#ifdef WITH_IPP
prims->copy_8u = (__copy_8u_t) ippsCopy_8u;
prims->copy_8u_AC4r = (__copy_8u_AC4r_t) ippiCopy_8u_AC4r;
#endif
/* Performance with an SSE2 version with no prefetch seemed to be
* all over the map vs. memcpy.
* Sometimes it was significantly faster, sometimes dreadfully slower,
* and it seemed to vary a lot depending on block size and processor.
* Hence, no SSE version is used here unless once can be written that
* is consistently faster than memcpy.
*/
/* This is just an alias with void* parameters */
prims->copy = (__copy_t) (prims->copy_8u);
}
/* ------------------------------------------------------------------------- */
void primitives_deinit_copy(
primitives_t *prims)
{
/* Nothing to do. */
}