1b5f74e8be
This PR is caused by POLY_INT_CSTs being (necessarily) valid
in tree-level VECTOR_CSTs but CONST_POLY_INTs not being valid
in RTL CONST_VECTORs. I can't tell/remember how deliberate
that was, but I'm guessing not very. In particular,
valid_for_const_vector_p was added to guard against symbolic
constants rather than CONST_POLY_INTs.
I did briefly consider whether we should maintain the current
status anyway. However, that would then require a way of
constructing variable-length vectors from individiual elements
if, say, we have:
{ [2, 2], [3, 2], [4, 2], … }
So I'm chalking this up to an oversight. I think the intention
(and certainly the natural thing) is to have the same rules for
both trees and RTL.
The SVE CONST_VECTOR code should already be set up to handle
CONST_POLY_INTs. However, we need to add support for Advanced SIMD
CONST_VECTORs that happen to contain SVE-based values. The patch does
that by expanding such CONST_VECTORs in the same way as variable vectors.
gcc/
PR rtl-optimization/97141
PR rtl-optimization/98726
* emit-rtl.c (valid_for_const_vector_p): Return true for
CONST_POLY_INT_P.
* rtx-vector-builder.h (rtx_vector_builder::step): Return a
poly_wide_int instead of a wide_int.
(rtx_vector_builder::apply_set): Take a poly_wide_int instead
of a wide_int.
* rtx-vector-builder.c (rtx_vector_builder::apply_set): Likewise.
* config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Return
false for CONST_VECTORs that cannot be forced to memory.
* config/aarch64/aarch64-simd.md (mov<mode>): If a CONST_VECTOR
is too complex to force to memory, build it up from individual
elements instead.
gcc/testsuite/
PR rtl-optimization/97141
PR rtl-optimization/98726
* gcc.c-torture/compile/pr97141.c: New test.
* gcc.c-torture/compile/pr98726.c: Likewise.
* gcc.target/aarch64/sve/pr97141.c: Likewise.
* gcc.target/aarch64/sve/pr98726.c: Likewise.
125 lines
3.8 KiB
C++
125 lines
3.8 KiB
C++
/* A class for building vector rtx constants.
|
|
Copyright (C) 2017-2021 Free Software Foundation, Inc.
|
|
|
|
This file is part of GCC.
|
|
|
|
GCC 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, or (at your option) any later
|
|
version.
|
|
|
|
GCC 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.
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
along with GCC; see the file COPYING3. If not see
|
|
<http://www.gnu.org/licenses/>. */
|
|
|
|
#ifndef GCC_RTX_VECTOR_BUILDER_H
|
|
#define GCC_RTX_VECTOR_BUILDER_H
|
|
|
|
#include "vector-builder.h"
|
|
|
|
/* This class is used to build VECTOR_CSTs from a sequence of elements.
|
|
See vector_builder for more details. */
|
|
class rtx_vector_builder : public vector_builder<rtx, machine_mode,
|
|
rtx_vector_builder>
|
|
{
|
|
typedef vector_builder<rtx, machine_mode, rtx_vector_builder> parent;
|
|
friend class vector_builder<rtx, machine_mode, rtx_vector_builder>;
|
|
|
|
public:
|
|
rtx_vector_builder () : m_mode (VOIDmode) {}
|
|
rtx_vector_builder (machine_mode, unsigned int, unsigned int);
|
|
rtx build (rtvec);
|
|
rtx build ();
|
|
|
|
machine_mode mode () const { return m_mode; }
|
|
|
|
void new_vector (machine_mode, unsigned int, unsigned int);
|
|
|
|
private:
|
|
bool equal_p (rtx, rtx) const;
|
|
bool allow_steps_p () const;
|
|
bool integral_p (rtx) const;
|
|
poly_wide_int step (rtx, rtx) const;
|
|
rtx apply_step (rtx, unsigned int, const poly_wide_int &) const;
|
|
bool can_elide_p (rtx) const { return true; }
|
|
void note_representative (rtx *, rtx) {}
|
|
|
|
static poly_uint64 shape_nelts (machine_mode mode)
|
|
{ return GET_MODE_NUNITS (mode); }
|
|
static poly_uint64 nelts_of (const_rtx x)
|
|
{ return CONST_VECTOR_NUNITS (x); }
|
|
static unsigned int npatterns_of (const_rtx x)
|
|
{ return CONST_VECTOR_NPATTERNS (x); }
|
|
static unsigned int nelts_per_pattern_of (const_rtx x)
|
|
{ return CONST_VECTOR_NELTS_PER_PATTERN (x); }
|
|
|
|
rtx find_cached_value ();
|
|
|
|
machine_mode m_mode;
|
|
};
|
|
|
|
/* Create a new builder for a vector of mode MODE. Initially encode the
|
|
value as NPATTERNS interleaved patterns with NELTS_PER_PATTERN elements
|
|
each. */
|
|
|
|
inline
|
|
rtx_vector_builder::rtx_vector_builder (machine_mode mode,
|
|
unsigned int npatterns,
|
|
unsigned int nelts_per_pattern)
|
|
{
|
|
new_vector (mode, npatterns, nelts_per_pattern);
|
|
}
|
|
|
|
/* Start building a new vector of mode MODE. Initially encode the value
|
|
as NPATTERNS interleaved patterns with NELTS_PER_PATTERN elements each. */
|
|
|
|
inline void
|
|
rtx_vector_builder::new_vector (machine_mode mode, unsigned int npatterns,
|
|
unsigned int nelts_per_pattern)
|
|
{
|
|
m_mode = mode;
|
|
parent::new_vector (GET_MODE_NUNITS (mode), npatterns, nelts_per_pattern);
|
|
}
|
|
|
|
/* Return true if elements ELT1 and ELT2 are equal. */
|
|
|
|
inline bool
|
|
rtx_vector_builder::equal_p (rtx elt1, rtx elt2) const
|
|
{
|
|
return rtx_equal_p (elt1, elt2);
|
|
}
|
|
|
|
/* Return true if a stepped representation is OK. We don't allow
|
|
linear series for anything other than integers, to avoid problems
|
|
with rounding. */
|
|
|
|
inline bool
|
|
rtx_vector_builder::allow_steps_p () const
|
|
{
|
|
return is_a <scalar_int_mode> (GET_MODE_INNER (m_mode));
|
|
}
|
|
|
|
/* Return true if element ELT can be interpreted as an integer. */
|
|
|
|
inline bool
|
|
rtx_vector_builder::integral_p (rtx elt) const
|
|
{
|
|
return CONST_SCALAR_INT_P (elt);
|
|
}
|
|
|
|
/* Return the value of element ELT2 minus the value of element ELT1.
|
|
Both elements are known to be CONST_SCALAR_INT_Ps. */
|
|
|
|
inline poly_wide_int
|
|
rtx_vector_builder::step (rtx elt1, rtx elt2) const
|
|
{
|
|
return (wi::to_poly_wide (elt2, GET_MODE_INNER (m_mode))
|
|
- wi::to_poly_wide (elt1, GET_MODE_INNER (m_mode)));
|
|
}
|
|
|
|
#endif
|