7a15726687
Using pointer sized variables (e.g. size_t / ptrdiff_t) when the variables are used as array indices allows accessing larger arrays, and can be a slight performance improvement due to no need for sign or zero extending, or masking. Regtested on x86_64-pc-linux-gnu. libgfortran/ChangeLog: 2018-01-31 Janne Blomqvist <jb@gcc.gnu.org> * generated/cshift1_16.c (cshift1): Regenerated. * generated/cshift1_4.c (cshift1): Regenerated. * generated/cshift1_8.c (cshift1): Regenerated. * generated/eoshift1_16.c (eoshift1): Regenerated. * generated/eoshift1_4.c (eoshift1): Regenerated. * generated/eoshift1_8.c (eoshift1): Regenerated. * generated/eoshift3_16.c (eoshift3): Regenerated. * generated/eoshift3_4.c (eoshift3): Regenerated. * generated/eoshift3_8.c (eoshift3): Regenerated. * generated/in_pack_c10.c (internal_pack_c10): Regenerated. * generated/in_pack_c16.c (internal_pack_c16): Regenerated. * generated/in_pack_c4.c (internal_pack_c4): Regenerated. * generated/in_pack_c8.c (internal_pack_c8): Regenerated. * generated/in_pack_i1.c (internal_pack_1): Regenerated. * generated/in_pack_i16.c (internal_pack_16): Regenerated. * generated/in_pack_i2.c (internal_pack_2): Regenerated. * generated/in_pack_i4.c (internal_pack_4): Regenerated. * generated/in_pack_i8.c (internal_pack_8): Regenerated. * generated/in_pack_r10.c (internal_pack_r10): Regenerated. * generated/in_pack_r16.c (internal_pack_r16): Regenerated. * generated/in_pack_r4.c (internal_pack_r4): Regenerated. * generated/in_pack_r8.c (internal_pack_r8): Regenerated. * generated/in_unpack_c10.c (internal_unpack_c10): Regenerated. * generated/in_unpack_c16.c (internal_unpack_c16): Regenerated. * generated/in_unpack_c4.c (internal_unpack_c4): Regenerated. * generated/in_unpack_c8.c (internal_unpack_c8): Regenerated. * generated/in_unpack_i1.c (internal_unpack_1): Regenerated. * generated/in_unpack_i16.c (internal_unpack_16): Regenerated. * generated/in_unpack_i2.c (internal_unpack_2): Regenerated. * generated/in_unpack_i4.c (internal_unpack_4): Regenerated. * generated/in_unpack_i8.c (internal_unpack_8): Regenerated. * generated/in_unpack_r10.c (internal_unpack_r10): Regenerated. * generated/in_unpack_r16.c (internal_unpack_r16): Regenerated. * generated/in_unpack_r4.c (internal_unpack_r4): Regenerated. * generated/in_unpack_r8.c (internal_unpack_r8): Regenerated. * generated/reshape_c10.c (reshape_c10): Regenerated. * generated/reshape_c16.c (reshape_c16): Regenerated. * generated/reshape_c4.c (reshape_c4): Regenerated. * generated/reshape_c8.c (reshape_c8): Regenerated. * generated/reshape_i16.c (reshape_16): Regenerated. * generated/reshape_i4.c (reshape_4): Regenerated. * generated/reshape_i8.c (reshape_8): Regenerated. * generated/reshape_r10.c (reshape_r10): Regenerated. * generated/reshape_r16.c (reshape_r16): Regenerated. * generated/reshape_r4.c (reshape_r4): Regenerated. * generated/reshape_r8.c (reshape_r8): Regenerated. * generated/shape_i1.c (shape_1): Regenerated. * generated/shape_i16.c (shape_16): Regenerated. * generated/shape_i2.c (shape_2): Regenerated. * generated/shape_i4.c (shape_4): Regenerated. * generated/shape_i8.c (shape_8): Regenerated. * generated/spread_c10.c (spread_scalar_c10): Regenerated. * generated/spread_c16.c (spread_scalar_c16): Regenerated. * generated/spread_c4.c (spread_scalar_c4): Regenerated. * generated/spread_c8.c (spread_scalar_c8): Regenerated. * generated/spread_i1.c (spread_scalar_i1): Regenerated. * generated/spread_i16.c (spread_scalar_i16): Regenerated. * generated/spread_i2.c (spread_scalar_i2): Regenerated. * generated/spread_i4.c (spread_scalar_i4): Regenerated. * generated/spread_i8.c (spread_scalar_i8): Regenerated. * generated/spread_r10.c (spread_scalar_r10): Regenerated. * generated/spread_r16.c (spread_scalar_r16): Regenerated. * generated/spread_r4.c (spread_scalar_r4): Regenerated. * generated/spread_r8.c (spread_scalar_r8): Regenerated. * intrinsics/random.c (jump): Use size_t for array index in loop. (getosrandom): Likewise. (arandom_r4): Make n an index_type. (arandom_r8): Likewise. (arandom_r10): Likewise. (arandom_r16): Likewise. (scramble_seed): Use size_t for array index in loop. * m4/cshift1.m4: Make i an index_type. * m4/eoshift1.m4: Likewise. * m4/eoshift3.m4: Likewise. * m4/in_pack.m4: Make n an index_type. * m4/in_unpack.m4: Likewise. * m4/reshape.m4: Make n and dim index_type's. * m4/shape.m4: Make n an index_type. * m4/spread.m4: Likewise, use index_type argument rather than copying to int. * runtime/bounds.c (bounds_ifunction_return): Make n an index_type. * runtime/in_pack_generic.c (internal_pack): Likewise. * runtime/in_unpack_generic.c (internal_unpack): Make n and size index_type's. From-SVN: r257234
225 lines
6.2 KiB
C
225 lines
6.2 KiB
C
/* Generic helper function for repacking arrays.
|
|
Copyright (C) 2003-2018 Free Software Foundation, Inc.
|
|
Contributed by Paul Brook <paul@nowt.org>
|
|
|
|
This file is part of the GNU Fortran runtime library (libgfortran).
|
|
|
|
Libgfortran is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU General Public
|
|
License as published by the Free Software Foundation; either
|
|
version 3 of the License, or (at your option) any later version.
|
|
|
|
Libgfortran is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU General Public License for more details.
|
|
|
|
Under Section 7 of GPL version 3, you are granted additional
|
|
permissions described in the GCC Runtime Library Exception, version
|
|
3.1, as published by the Free Software Foundation.
|
|
|
|
You should have received a copy of the GNU General Public License and
|
|
a copy of the GCC Runtime Library Exception along with this program;
|
|
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
|
|
<http://www.gnu.org/licenses/>. */
|
|
|
|
#include "libgfortran.h"
|
|
#include <string.h>
|
|
|
|
extern void *internal_pack (gfc_array_char *);
|
|
export_proto(internal_pack);
|
|
|
|
void *
|
|
internal_pack (gfc_array_char * source)
|
|
{
|
|
index_type count[GFC_MAX_DIMENSIONS];
|
|
index_type extent[GFC_MAX_DIMENSIONS];
|
|
index_type stride[GFC_MAX_DIMENSIONS];
|
|
index_type stride0;
|
|
index_type dim;
|
|
index_type ssize;
|
|
const char *src;
|
|
char *dest;
|
|
void *destptr;
|
|
int packed;
|
|
index_type size;
|
|
index_type type_size;
|
|
|
|
if (source->base_addr == NULL)
|
|
return NULL;
|
|
|
|
type_size = GFC_DTYPE_TYPE_SIZE(source);
|
|
size = GFC_DESCRIPTOR_SIZE (source);
|
|
switch (type_size)
|
|
{
|
|
case GFC_DTYPE_INTEGER_1:
|
|
case GFC_DTYPE_LOGICAL_1:
|
|
return internal_pack_1 ((gfc_array_i1 *) source);
|
|
|
|
case GFC_DTYPE_INTEGER_2:
|
|
case GFC_DTYPE_LOGICAL_2:
|
|
return internal_pack_2 ((gfc_array_i2 *) source);
|
|
|
|
case GFC_DTYPE_INTEGER_4:
|
|
case GFC_DTYPE_LOGICAL_4:
|
|
return internal_pack_4 ((gfc_array_i4 *) source);
|
|
|
|
case GFC_DTYPE_INTEGER_8:
|
|
case GFC_DTYPE_LOGICAL_8:
|
|
return internal_pack_8 ((gfc_array_i8 *) source);
|
|
|
|
#if defined(HAVE_GFC_INTEGER_16)
|
|
case GFC_DTYPE_INTEGER_16:
|
|
case GFC_DTYPE_LOGICAL_16:
|
|
return internal_pack_16 ((gfc_array_i16 *) source);
|
|
#endif
|
|
case GFC_DTYPE_REAL_4:
|
|
return internal_pack_r4 ((gfc_array_r4 *) source);
|
|
|
|
case GFC_DTYPE_REAL_8:
|
|
return internal_pack_r8 ((gfc_array_r8 *) source);
|
|
|
|
/* FIXME: This here is a hack, which will have to be removed when
|
|
the array descriptor is reworked. Currently, we don't store the
|
|
kind value for the type, but only the size. Because on targets with
|
|
__float128, we have sizeof(logn double) == sizeof(__float128),
|
|
we cannot discriminate here and have to fall back to the generic
|
|
handling (which is suboptimal). */
|
|
#if !defined(GFC_REAL_16_IS_FLOAT128)
|
|
# if defined (HAVE_GFC_REAL_10)
|
|
case GFC_DTYPE_REAL_10:
|
|
return internal_pack_r10 ((gfc_array_r10 *) source);
|
|
# endif
|
|
|
|
# if defined (HAVE_GFC_REAL_16)
|
|
case GFC_DTYPE_REAL_16:
|
|
return internal_pack_r16 ((gfc_array_r16 *) source);
|
|
# endif
|
|
#endif
|
|
|
|
case GFC_DTYPE_COMPLEX_4:
|
|
return internal_pack_c4 ((gfc_array_c4 *) source);
|
|
|
|
case GFC_DTYPE_COMPLEX_8:
|
|
return internal_pack_c8 ((gfc_array_c8 *) source);
|
|
|
|
/* FIXME: This here is a hack, which will have to be removed when
|
|
the array descriptor is reworked. Currently, we don't store the
|
|
kind value for the type, but only the size. Because on targets with
|
|
__float128, we have sizeof(logn double) == sizeof(__float128),
|
|
we cannot discriminate here and have to fall back to the generic
|
|
handling (which is suboptimal). */
|
|
#if !defined(GFC_REAL_16_IS_FLOAT128)
|
|
# if defined (HAVE_GFC_COMPLEX_10)
|
|
case GFC_DTYPE_COMPLEX_10:
|
|
return internal_pack_c10 ((gfc_array_c10 *) source);
|
|
# endif
|
|
|
|
# if defined (HAVE_GFC_COMPLEX_16)
|
|
case GFC_DTYPE_COMPLEX_16:
|
|
return internal_pack_c16 ((gfc_array_c16 *) source);
|
|
# endif
|
|
#endif
|
|
|
|
default:
|
|
break;
|
|
}
|
|
|
|
switch(GFC_DESCRIPTOR_SIZE (source))
|
|
{
|
|
case 1:
|
|
return internal_pack_1 ((gfc_array_i1 *) source);
|
|
|
|
case 2:
|
|
if (GFC_UNALIGNED_2(source->base_addr))
|
|
break;
|
|
else
|
|
return internal_pack_2 ((gfc_array_i2 *) source);
|
|
|
|
case 4:
|
|
if (GFC_UNALIGNED_4(source->base_addr))
|
|
break;
|
|
else
|
|
return internal_pack_4 ((gfc_array_i4 *) source);
|
|
|
|
case 8:
|
|
if (GFC_UNALIGNED_8(source->base_addr))
|
|
break;
|
|
else
|
|
return internal_pack_8 ((gfc_array_i8 *) source);
|
|
|
|
#ifdef HAVE_GFC_INTEGER_16
|
|
case 16:
|
|
if (GFC_UNALIGNED_16(source->base_addr))
|
|
break;
|
|
else
|
|
return internal_pack_16 ((gfc_array_i16 *) source);
|
|
#endif
|
|
default:
|
|
break;
|
|
}
|
|
|
|
dim = GFC_DESCRIPTOR_RANK (source);
|
|
ssize = 1;
|
|
packed = 1;
|
|
for (index_type n = 0; n < dim; n++)
|
|
{
|
|
count[n] = 0;
|
|
stride[n] = GFC_DESCRIPTOR_STRIDE(source,n);
|
|
extent[n] = GFC_DESCRIPTOR_EXTENT(source,n);
|
|
if (extent[n] <= 0)
|
|
{
|
|
/* Do nothing. */
|
|
packed = 1;
|
|
break;
|
|
}
|
|
|
|
if (ssize != stride[n])
|
|
packed = 0;
|
|
|
|
ssize *= extent[n];
|
|
}
|
|
|
|
if (packed)
|
|
return source->base_addr;
|
|
|
|
/* Allocate storage for the destination. */
|
|
destptr = xmallocarray (ssize, size);
|
|
dest = (char *)destptr;
|
|
src = source->base_addr;
|
|
stride0 = stride[0] * size;
|
|
|
|
while (src)
|
|
{
|
|
/* Copy the data. */
|
|
memcpy(dest, src, size);
|
|
/* Advance to the next element. */
|
|
dest += size;
|
|
src += stride0;
|
|
count[0]++;
|
|
/* Advance to the next source element. */
|
|
index_type n = 0;
|
|
while (count[n] == extent[n])
|
|
{
|
|
/* When we get to the end of a dimension, reset it and increment
|
|
the next dimension. */
|
|
count[n] = 0;
|
|
/* We could precalculate these products, but this is a less
|
|
frequently used path so probably not worth it. */
|
|
src -= stride[n] * extent[n] * size;
|
|
n++;
|
|
if (n == dim)
|
|
{
|
|
src = NULL;
|
|
break;
|
|
}
|
|
else
|
|
{
|
|
count[n]++;
|
|
src += stride[n] * size;
|
|
}
|
|
}
|
|
}
|
|
return destptr;
|
|
}
|