/* Subroutines used for code generation on IA-32.
Copyright (C) 1988, 1992, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
- 2002, 2003, 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
+ 2002, 2003, 2004, 2005, 2006, 2007, 2008
+ Free Software Foundation, Inc.
This file is part of GCC.
#include "tm-constrs.h"
#include "params.h"
+static int x86_builtin_vectorization_cost (bool);
+
#ifndef CHECK_STACK_LIMIT
#define CHECK_STACK_LIMIT (-1)
#endif
{{rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}},
{rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}}},
{{rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}},
- {rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}}}
+ {rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}}},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 1, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 1, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
/* Processor costs (relative to an add) */
DUMMY_STRINGOP_ALGS},
{{rep_prefix_1_byte, {{-1, rep_prefix_1_byte}}},
DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{rep_prefix_4_byte, {{-1, rep_prefix_4_byte}}},
DUMMY_STRINGOP_ALGS},
{{rep_prefix_4_byte, {{-1, rep_prefix_4_byte}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{256, rep_prefix_4_byte}, {-1, libcall}}},
DUMMY_STRINGOP_ALGS},
{{libcall, {{-1, rep_prefix_4_byte}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
DUMMY_STRINGOP_ALGS},
{{rep_prefix_4_byte, {{1024, unrolled_loop},
{8192, rep_prefix_4_byte}, {-1, libcall}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{256, rep_prefix_4_byte}, {-1, libcall}}},
DUMMY_STRINGOP_ALGS},
{{libcall, {{256, rep_prefix_4_byte}, {-1, libcall}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{256, rep_prefix_4_byte}, {-1, libcall}}},
DUMMY_STRINGOP_ALGS},
{{libcall, {{256, rep_prefix_4_byte}, {-1, libcall}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{2048, rep_prefix_4_byte}, {-1, libcall}}},
DUMMY_STRINGOP_ALGS},
{{libcall, {{2048, rep_prefix_4_byte}, {-1, libcall}}},
- DUMMY_STRINGOP_ALGS}
+ DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
to limit number of prefetches at all, as their execution also takes some
time). */
100, /* number of parallel prefetches */
- 5, /* Branch cost */
+ 3, /* Branch cost */
COSTS_N_INSNS (4), /* cost of FADD and FSUB insns. */
COSTS_N_INSNS (4), /* cost of FMUL instruction. */
COSTS_N_INSNS (19), /* cost of FDIV instruction. */
{libcall, {{16, loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
{{libcall, {{8, loop}, {24, unrolled_loop},
{2048, rep_prefix_4_byte}, {-1, libcall}}},
- {libcall, {{48, unrolled_loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}}
+ {libcall, {{48, unrolled_loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
+ 4, /* scalar_stmt_cost. */
+ 2, /* scalar load_cost. */
+ 2, /* scalar_store_cost. */
+ 5, /* vec_stmt_cost. */
+ 0, /* vec_to_scalar_cost. */
+ 2, /* scalar_to_vec_cost. */
+ 2, /* vec_align_load_cost. */
+ 3, /* vec_unalign_load_cost. */
+ 3, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 2, /* cond_not_taken_branch_cost. */
};
struct processor_costs amdfam10_cost = {
to limit number of prefetches at all, as their execution also takes some
time). */
100, /* number of parallel prefetches */
- 5, /* Branch cost */
+ 2, /* Branch cost */
COSTS_N_INSNS (4), /* cost of FADD and FSUB insns. */
COSTS_N_INSNS (4), /* cost of FMUL instruction. */
COSTS_N_INSNS (19), /* cost of FDIV instruction. */
{libcall, {{16, loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
{{libcall, {{8, loop}, {24, unrolled_loop},
{2048, rep_prefix_4_byte}, {-1, libcall}}},
- {libcall, {{48, unrolled_loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}}
+ {libcall, {{48, unrolled_loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
+ 4, /* scalar_stmt_cost. */
+ 2, /* scalar load_cost. */
+ 2, /* scalar_store_cost. */
+ 6, /* vec_stmt_cost. */
+ 0, /* vec_to_scalar_cost. */
+ 2, /* scalar_to_vec_cost. */
+ 2, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 2, /* vec_store_cost. */
+ 2, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{6, loop_1_byte}, {48, loop}, {20480, rep_prefix_4_byte},
{-1, libcall}}},
DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{6, loop_1_byte}, {48, loop}, {20480, rep_prefix_4_byte},
{-1, libcall}}},
{libcall, {{24, loop}, {64, unrolled_loop},
- {8192, rep_prefix_8_byte}, {-1, libcall}}}}
+ {8192, rep_prefix_8_byte}, {-1, libcall}}}},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
static const
{{libcall, {{8, loop}, {15, unrolled_loop},
{2048, rep_prefix_4_byte}, {-1, libcall}}},
{libcall, {{24, loop}, {32, unrolled_loop},
- {8192, rep_prefix_8_byte}, {-1, libcall}}}}
+ {8192, rep_prefix_8_byte}, {-1, libcall}}}},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
/* Generic64 should produce code tuned for Nocona and K8. */
{DUMMY_STRINGOP_ALGS,
{libcall, {{32, loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
{DUMMY_STRINGOP_ALGS,
- {libcall, {{32, loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}}
+ {libcall, {{32, loop}, {8192, rep_prefix_8_byte}, {-1, libcall}}}},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
/* Generic32 should produce code tuned for Athlon, PPro, Pentium4, Nocona and K8. */
DUMMY_STRINGOP_ALGS},
{{libcall, {{32, loop}, {8192, rep_prefix_4_byte}, {-1, libcall}}},
DUMMY_STRINGOP_ALGS},
+ 1, /* scalar_stmt_cost. */
+ 1, /* scalar load_cost. */
+ 1, /* scalar_store_cost. */
+ 1, /* vec_stmt_cost. */
+ 1, /* vec_to_scalar_cost. */
+ 1, /* scalar_to_vec_cost. */
+ 1, /* vec_align_load_cost. */
+ 2, /* vec_unalign_load_cost. */
+ 1, /* vec_store_cost. */
+ 3, /* cond_taken_branch_cost. */
+ 1, /* cond_not_taken_branch_cost. */
};
const struct processor_costs *ix86_cost = &pentium_cost;
#define m_ATHLON (1<<PROCESSOR_ATHLON)
#define m_ATHLON_K8 (m_K8 | m_ATHLON)
#define m_AMDFAM10 (1<<PROCESSOR_AMDFAM10)
-#define m_ATHLON_K8_AMDFAM10 (m_K8 | m_ATHLON | m_AMDFAM10)
+#define m_AMD_MULTIPLE (m_K8 | m_ATHLON | m_AMDFAM10)
#define m_GENERIC32 (1<<PROCESSOR_GENERIC32)
#define m_GENERIC64 (1<<PROCESSOR_GENERIC64)
negatively, so enabling for Generic64 seems like good code size
tradeoff. We can't enable it for 32bit generic because it does not
work well with PPro base chips. */
- m_386 | m_K6_GEODE | m_ATHLON_K8_AMDFAM10 | m_CORE2 | m_GENERIC64,
+ m_386 | m_K6_GEODE | m_AMD_MULTIPLE | m_CORE2 | m_GENERIC64,
/* X86_TUNE_PUSH_MEMORY */
- m_386 | m_K6_GEODE | m_ATHLON_K8_AMDFAM10 | m_PENT4
+ m_386 | m_K6_GEODE | m_AMD_MULTIPLE | m_PENT4
| m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_ZERO_EXTEND_WITH_AND */
m_386,
/* X86_TUNE_UNROLL_STRLEN */
- m_486 | m_PENT | m_PPRO | m_ATHLON_K8_AMDFAM10 | m_K6 | m_CORE2 | m_GENERIC,
+ m_486 | m_PENT | m_PPRO | m_AMD_MULTIPLE | m_K6 | m_CORE2 | m_GENERIC,
/* X86_TUNE_DEEP_BRANCH_PREDICTION */
- m_PPRO | m_K6_GEODE | m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_GENERIC,
+ m_PPRO | m_K6_GEODE | m_AMD_MULTIPLE | m_PENT4 | m_GENERIC,
/* X86_TUNE_BRANCH_PREDICTION_HINTS: Branch hints were put in P4 based
on simulation result. But after P4 was made, no performance benefit
/* X86_TUNE_MOVX: Enable to zero extend integer registers to avoid
partial dependencies. */
- m_ATHLON_K8_AMDFAM10 | m_PPRO | m_PENT4 | m_NOCONA
+ m_AMD_MULTIPLE | m_PPRO | m_PENT4 | m_NOCONA
| m_CORE2 | m_GENERIC | m_GEODE /* m_386 | m_K6 */,
/* X86_TUNE_PARTIAL_REG_STALL: We probably ought to watch for partial
m_386 | m_486 | m_K6_GEODE,
/* X86_TUNE_USE_SIMODE_FIOP */
- ~(m_PPRO | m_ATHLON_K8_AMDFAM10 | m_PENT | m_CORE2 | m_GENERIC),
+ ~(m_PPRO | m_AMD_MULTIPLE | m_PENT | m_CORE2 | m_GENERIC),
/* X86_TUNE_USE_MOV0 */
m_K6,
~(m_PENT | m_PPRO),
/* X86_TUNE_PROMOTE_QIMODE */
- m_K6_GEODE | m_PENT | m_386 | m_486 | m_ATHLON_K8_AMDFAM10 | m_CORE2
+ m_K6_GEODE | m_PENT | m_386 | m_486 | m_AMD_MULTIPLE | m_CORE2
| m_GENERIC /* | m_PENT4 ? */,
/* X86_TUNE_FAST_PREFIX */
m_PPRO,
/* X86_TUNE_ADD_ESP_4: Enable if add/sub is preferred over 1/2 push/pop. */
- m_ATHLON_K8_AMDFAM10 | m_K6_GEODE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
+ m_AMD_MULTIPLE | m_K6_GEODE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_ADD_ESP_8 */
- m_ATHLON_K8_AMDFAM10 | m_PPRO | m_K6_GEODE | m_386
+ m_AMD_MULTIPLE | m_PPRO | m_K6_GEODE | m_386
| m_486 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_SUB_ESP_4 */
- m_ATHLON_K8_AMDFAM10 | m_PPRO | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
+ m_AMD_MULTIPLE | m_PPRO | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_SUB_ESP_8 */
- m_ATHLON_K8_AMDFAM10 | m_PPRO | m_386 | m_486
+ m_AMD_MULTIPLE | m_PPRO | m_386 | m_486
| m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_INTEGER_DFMODE_MOVES: Enable if integer moves are preferred
for DFmode copies */
- ~(m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2
+ ~(m_AMD_MULTIPLE | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2
| m_GENERIC | m_GEODE),
/* X86_TUNE_PARTIAL_REG_DEPENDENCY */
- m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
+ m_AMD_MULTIPLE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_SSE_PARTIAL_REG_DEPENDENCY: In the Generic model we have a
conflict here in between PPro/Pentium4 based chips that thread 128bit
m_ATHLON_K8,
/* X86_TUNE_SSE_TYPELESS_STORES */
- m_ATHLON_K8_AMDFAM10,
+ m_AMD_MULTIPLE,
/* X86_TUNE_SSE_LOAD0_BY_PXOR */
m_PPRO | m_PENT4 | m_NOCONA,
/* X86_TUNE_MEMORY_MISMATCH_STALL */
- m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
+ m_AMD_MULTIPLE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_PROLOGUE_USING_MOVE */
m_ATHLON_K8 | m_PPRO | m_CORE2 | m_GENERIC,
~m_486,
/* X86_TUNE_USE_FFREEP */
- m_ATHLON_K8_AMDFAM10,
+ m_AMD_MULTIPLE,
/* X86_TUNE_INTER_UNIT_MOVES */
- ~(m_ATHLON_K8_AMDFAM10 | m_GENERIC),
+ ~(m_AMD_MULTIPLE | m_GENERIC),
+
+ /* X86_TUNE_INTER_UNIT_CONVERSIONS */
+ ~(m_AMDFAM10),
/* X86_TUNE_FOUR_JUMP_LIMIT: Some CPU cores are not able to predict more
than 4 branch instructions in the 16 byte window. */
- m_PPRO | m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
+ m_PPRO | m_AMD_MULTIPLE | m_PENT4 | m_NOCONA | m_CORE2 | m_GENERIC,
/* X86_TUNE_SCHEDULE */
- m_PPRO | m_ATHLON_K8_AMDFAM10 | m_K6_GEODE | m_PENT | m_CORE2 | m_GENERIC,
+ m_PPRO | m_AMD_MULTIPLE | m_K6_GEODE | m_PENT | m_CORE2 | m_GENERIC,
/* X86_TUNE_USE_BT */
- m_ATHLON_K8_AMDFAM10,
+ m_AMD_MULTIPLE,
/* X86_TUNE_USE_INCDEC */
~(m_PENT4 | m_NOCONA | m_GENERIC),
/* X86_TUNE_PAD_RETURNS */
- m_ATHLON_K8_AMDFAM10 | m_CORE2 | m_GENERIC,
+ m_AMD_MULTIPLE | m_CORE2 | m_GENERIC,
/* X86_TUNE_EXT_80387_CONSTANTS */
m_K6_GEODE | m_ATHLON_K8 | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC,
operand that cannot be represented using a modRM byte. The XOR
replacement is long decoded, so this split helps here as well. */
m_K6,
+
+ /* X86_TUNE_USE_VECTOR_CONVERTS: Prefer vector packed SSE conversion
+ from integer to FP. */
+ m_AMDFAM10,
};
/* Feature tests against the various architecture variations. */
};
static const unsigned int x86_accumulate_outgoing_args
- = m_ATHLON_K8_AMDFAM10 | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC;
+ = m_AMD_MULTIPLE | m_PENT4 | m_NOCONA | m_PPRO | m_CORE2 | m_GENERIC;
static const unsigned int x86_arch_always_fancy_math_387
- = m_PENT | m_PPRO | m_ATHLON_K8_AMDFAM10 | m_PENT4
+ = m_PENT | m_PPRO | m_AMD_MULTIPLE | m_PENT4
| m_NOCONA | m_CORE2 | m_GENERIC;
static enum stringop_alg stringop_alg = no_stringop;
\f
static struct machine_function * ix86_init_machine_status (void);
-static rtx ix86_function_value (tree, tree, bool);
-static int ix86_function_regparm (tree, tree);
+static rtx ix86_function_value (const_tree, const_tree, bool);
+static int ix86_function_regparm (const_tree, const_tree);
static void ix86_compute_frame_layout (struct ix86_frame *);
static bool ix86_expand_vector_init_one_nonzero (bool, enum machine_mode,
rtx, rtx, int);
was set or cleared on the command line. */
static int ix86_isa_flags_explicit;
-/* Define a set of ISAs which aren't available for a given ISA. MMX
- and SSE ISAs are handled separately. */
+/* Define a set of ISAs which are available when a given ISA is
+ enabled. MMX and SSE ISAs are handled separately. */
+
+#define OPTION_MASK_ISA_MMX_SET OPTION_MASK_ISA_MMX
+#define OPTION_MASK_ISA_3DNOW_SET \
+ (OPTION_MASK_ISA_3DNOW | OPTION_MASK_ISA_MMX_SET)
+
+#define OPTION_MASK_ISA_SSE_SET OPTION_MASK_ISA_SSE
+#define OPTION_MASK_ISA_SSE2_SET \
+ (OPTION_MASK_ISA_SSE2 | OPTION_MASK_ISA_SSE_SET)
+#define OPTION_MASK_ISA_SSE3_SET \
+ (OPTION_MASK_ISA_SSE3 | OPTION_MASK_ISA_SSE2_SET)
+#define OPTION_MASK_ISA_SSSE3_SET \
+ (OPTION_MASK_ISA_SSSE3 | OPTION_MASK_ISA_SSE3_SET)
+#define OPTION_MASK_ISA_SSE4_1_SET \
+ (OPTION_MASK_ISA_SSE4_1 | OPTION_MASK_ISA_SSSE3_SET)
+#define OPTION_MASK_ISA_SSE4_2_SET \
+ (OPTION_MASK_ISA_SSE4_2 | OPTION_MASK_ISA_SSE4_1_SET)
+
+/* SSE4 includes both SSE4.1 and SSE4.2. -msse4 should be the same
+ as -msse4.2. */
+#define OPTION_MASK_ISA_SSE4_SET OPTION_MASK_ISA_SSE4_2_SET
+
+#define OPTION_MASK_ISA_SSE4A_SET \
+ (OPTION_MASK_ISA_SSE4A | OPTION_MASK_ISA_SSE3_SET)
+#define OPTION_MASK_ISA_SSE5_SET \
+ (OPTION_MASK_ISA_SSE5 | OPTION_MASK_ISA_SSE4A_SET)
+
+/* Define a set of ISAs which aren't available when a given ISA is
+ disabled. MMX and SSE ISAs are handled separately. */
#define OPTION_MASK_ISA_MMX_UNSET \
- (OPTION_MASK_ISA_3DNOW | OPTION_MASK_ISA_3DNOW_UNSET)
-#define OPTION_MASK_ISA_3DNOW_UNSET OPTION_MASK_ISA_3DNOW_A
+ (OPTION_MASK_ISA_MMX | OPTION_MASK_ISA_3DNOW_UNSET)
+#define OPTION_MASK_ISA_3DNOW_UNSET \
+ (OPTION_MASK_ISA_3DNOW | OPTION_MASK_ISA_3DNOW_A_UNSET)
+#define OPTION_MASK_ISA_3DNOW_A_UNSET OPTION_MASK_ISA_3DNOW_A
#define OPTION_MASK_ISA_SSE_UNSET \
- (OPTION_MASK_ISA_SSE2 | OPTION_MASK_ISA_SSE2_UNSET)
+ (OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_SSE2_UNSET)
#define OPTION_MASK_ISA_SSE2_UNSET \
- (OPTION_MASK_ISA_SSE3 | OPTION_MASK_ISA_SSE3_UNSET)
+ (OPTION_MASK_ISA_SSE2 | OPTION_MASK_ISA_SSE3_UNSET)
#define OPTION_MASK_ISA_SSE3_UNSET \
- (OPTION_MASK_ISA_SSSE3 | OPTION_MASK_ISA_SSSE3_UNSET)
+ (OPTION_MASK_ISA_SSE3 \
+ | OPTION_MASK_ISA_SSSE3_UNSET \
+ | OPTION_MASK_ISA_SSE4A_UNSET )
#define OPTION_MASK_ISA_SSSE3_UNSET \
- (OPTION_MASK_ISA_SSE4_1 | OPTION_MASK_ISA_SSE4_1_UNSET)
+ (OPTION_MASK_ISA_SSSE3 | OPTION_MASK_ISA_SSE4_1_UNSET)
#define OPTION_MASK_ISA_SSE4_1_UNSET \
- (OPTION_MASK_ISA_SSE4_2 | OPTION_MASK_ISA_SSE4_2_UNSET)
-#define OPTION_MASK_ISA_SSE4_2_UNSET OPTION_MASK_ISA_SSE4A
+ (OPTION_MASK_ISA_SSE4_1 | OPTION_MASK_ISA_SSE4_2_UNSET)
+#define OPTION_MASK_ISA_SSE4_2_UNSET OPTION_MASK_ISA_SSE4_2
-/* SSE4 includes both SSE4.1 and SSE4.2. -msse4 should be the same
- as -msse4.1 -msse4.2. -mno-sse4 should the same as -mno-sse4.1. */
-#define OPTION_MASK_ISA_SSE4 \
- (OPTION_MASK_ISA_SSE4_1 | OPTION_MASK_ISA_SSE4_2)
+/* SSE4 includes both SSE4.1 and SSE4.2. -mno-sse4 should the same
+ as -mno-sse4.1. */
#define OPTION_MASK_ISA_SSE4_UNSET OPTION_MASK_ISA_SSE4_1_UNSET
-#define OPTION_MASK_ISA_SSE4A_UNSET OPTION_MASK_ISA_SSE4
+#define OPTION_MASK_ISA_SSE4A_UNSET \
+ (OPTION_MASK_ISA_SSE4A | OPTION_MASK_ISA_SSE5_UNSET)
+
+#define OPTION_MASK_ISA_SSE5_UNSET OPTION_MASK_ISA_SSE5
+
+/* Vectorization library interface and handlers. */
+tree (*ix86_veclib_handler)(enum built_in_function, tree, tree) = NULL;
+static tree ix86_veclibabi_acml (enum built_in_function, tree, tree);
/* Implement TARGET_HANDLE_OPTION. */
switch (code)
{
case OPT_mmmx:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_MMX;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_MMX_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_MMX_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_MMX_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_MMX_UNSET;
return true;
case OPT_m3dnow:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_3DNOW;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_3DNOW_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_3DNOW_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_3DNOW_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_3DNOW_UNSET;
return false;
case OPT_msse:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE_UNSET;
return true;
case OPT_msse2:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE2;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE2_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE2_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE2_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE2_UNSET;
return true;
case OPT_msse3:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE3;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE3_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE3_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE3_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE3_UNSET;
return true;
case OPT_mssse3:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSSE3;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSSE3_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSSE3_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSSE3_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSSE3_UNSET;
return true;
case OPT_msse4_1:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_1;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE4_1_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_1_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE4_1_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_1_UNSET;
return true;
case OPT_msse4_2:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_2;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE4_2_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_2_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE4_2_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_2_UNSET;
return true;
case OPT_msse4:
- ix86_isa_flags |= OPTION_MASK_ISA_SSE4;
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4;
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE4_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4_SET;
return true;
case OPT_mno_sse4:
return true;
case OPT_msse4a:
- ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4A;
- if (!value)
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE4A_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4A_SET;
+ }
+ else
{
ix86_isa_flags &= ~OPTION_MASK_ISA_SSE4A_UNSET;
ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE4A_UNSET;
}
return true;
+ case OPT_msse5:
+ if (value)
+ {
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE5_SET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE5_SET;
+ }
+ else
+ {
+ ix86_isa_flags &= ~OPTION_MASK_ISA_SSE5_UNSET;
+ ix86_isa_flags_explicit |= OPTION_MASK_ISA_SSE5_UNSET;
+ }
+ return true;
+
default:
return true;
}
{&amdfam10_cost, 32, 24, 32, 7, 32}
};
- static const char * const cpu_names[] = TARGET_CPU_DEFAULT_NAMES;
+ static const char *const cpu_names[TARGET_CPU_DEFAULT_max] =
+ {
+ "generic",
+ "i386",
+ "i486",
+ "pentium",
+ "pentium-mmx",
+ "pentiumpro",
+ "pentium2",
+ "pentium3",
+ "pentium4",
+ "pentium-m",
+ "prescott",
+ "nocona",
+ "core2",
+ "geode",
+ "k6",
+ "k6-2",
+ "k6-3",
+ "athlon",
+ "athlon-4",
+ "k8",
+ "amdfam10"
+ };
+
enum pta_flags
{
PTA_SSE = 1 << 0,
PTA_SSE4A = 1 << 12,
PTA_NO_SAHF = 1 << 13,
PTA_SSE4_1 = 1 << 14,
- PTA_SSE4_2 = 1 << 15
+ PTA_SSE4_2 = 1 << 15,
+ PTA_SSE5 = 1 << 16
};
static struct pta
{"pentium4", PROCESSOR_PENTIUM4, PTA_MMX |PTA_SSE | PTA_SSE2},
{"pentium4m", PROCESSOR_PENTIUM4, PTA_MMX | PTA_SSE | PTA_SSE2},
{"prescott", PROCESSOR_NOCONA, PTA_MMX | PTA_SSE | PTA_SSE2 | PTA_SSE3},
- {"nocona", PROCESSOR_NOCONA, (PTA_64BIT
+ {"nocona", PROCESSOR_NOCONA, (PTA_64BIT
| PTA_MMX | PTA_SSE | PTA_SSE2 | PTA_SSE3
| PTA_CX16 | PTA_NO_SAHF)},
{"core2", PROCESSOR_CORE2, (PTA_64BIT
ix86_tune_string = ix86_arch_string;
if (!ix86_tune_string)
{
- ix86_tune_string = cpu_names [TARGET_CPU_DEFAULT];
+ ix86_tune_string = cpu_names[TARGET_CPU_DEFAULT];
ix86_tune_defaulted = 1;
}
ix86_arch_string = TARGET_64BIT ? "x86-64" : "i386";
else
ix86_arch_specified = 1;
-
+
if (!strcmp (ix86_arch_string, "generic"))
error ("generic CPU can be used only for -mtune= switch");
if (!strncmp (ix86_arch_string, "generic", 7))
if (processor_alias_table[i].flags & PTA_SSE4A
&& !(ix86_isa_flags_explicit & OPTION_MASK_ISA_SSE4A))
ix86_isa_flags |= OPTION_MASK_ISA_SSE4A;
+ if (processor_alias_table[i].flags & PTA_SSE5
+ && !(ix86_isa_flags_explicit & OPTION_MASK_ISA_SSE5))
+ ix86_isa_flags |= OPTION_MASK_ISA_SSE5;
if (processor_alias_table[i].flags & PTA_ABM)
x86_abm = true;
if (!TARGET_80387)
target_flags |= MASK_NO_FANCY_MATH_387;
- /* Turn on SSE4.1 builtins for -msse4.2. */
- if (TARGET_SSE4_2)
- ix86_isa_flags |= OPTION_MASK_ISA_SSE4_1;
-
- /* Turn on SSSE3 builtins for -msse4.1. */
- if (TARGET_SSE4_1)
- ix86_isa_flags |= OPTION_MASK_ISA_SSSE3;
-
- /* Turn on SSE3 builtins for -mssse3. */
- if (TARGET_SSSE3)
- ix86_isa_flags |= OPTION_MASK_ISA_SSE3;
-
- /* Turn on SSE3 builtins for -msse4a. */
- if (TARGET_SSE4A)
- ix86_isa_flags |= OPTION_MASK_ISA_SSE3;
-
- /* Turn on SSE2 builtins for -msse3. */
- if (TARGET_SSE3)
- ix86_isa_flags |= OPTION_MASK_ISA_SSE2;
-
- /* Turn on SSE builtins for -msse2. */
- if (TARGET_SSE2)
- ix86_isa_flags |= OPTION_MASK_ISA_SSE;
-
/* Turn on MMX builtins for -msse. */
if (TARGET_SSE)
{
x86_prefetch_sse = true;
}
- /* Turn on MMX builtins for 3Dnow. */
- if (TARGET_3DNOW)
- ix86_isa_flags |= OPTION_MASK_ISA_MMX;
-
/* Turn on popcnt instruction for -msse4.2 or -mabm. */
if (TARGET_SSE4_2 || TARGET_ABM)
x86_popcnt = true;
if (!TARGET_80387)
target_flags &= ~MASK_FLOAT_RETURNS;
+ /* Use external vectorized library in vectorizing intrinsics. */
+ if (ix86_veclibabi_string)
+ {
+ if (strcmp (ix86_veclibabi_string, "acml") == 0)
+ ix86_veclib_handler = ix86_veclibabi_acml;
+ else
+ error ("unknown vectorization library ABI type (%s) for "
+ "-mveclibabi= switch", ix86_veclibabi_string);
+ }
+
if ((x86_accumulate_outgoing_args & ix86_tune_mask)
&& !(target_flags_explicit & MASK_ACCUMULATE_OUTGOING_ARGS)
&& !optimize_size)
target_flags |= MASK_ACCUMULATE_OUTGOING_ARGS;
}
+ /* If stack probes are required, the space used for large function
+ arguments on the stack must also be probed, so enable
+ -maccumulate-outgoing-args so this happens in the prologue. */
+ if (TARGET_STACK_PROBE
+ && !(target_flags & MASK_ACCUMULATE_OUTGOING_ARGS))
+ {
+ if (target_flags_explicit & MASK_ACCUMULATE_OUTGOING_ARGS)
+ warning (0, "stack probing requires -maccumulate-outgoing-args "
+ "for correctness");
+ target_flags |= MASK_ACCUMULATE_OUTGOING_ARGS;
+ }
+
/* For sane SSE instruction set generation we need fcomi instruction.
It is safe to enable all CMOVE instructions. */
if (TARGET_SSE)
set_param_value ("l1-cache-size", ix86_cost->l1_cache_size);
if (!PARAM_SET_P (PARAM_L2_CACHE_SIZE))
set_param_value ("l2-cache-size", ix86_cost->l2_cache_size);
+
+ /* If using typedef char *va_list, signal that __builtin_va_start (&ap, 0)
+ can be optimized to ap = __builtin_next_arg (0). */
+ if (!TARGET_64BIT || TARGET_64BIT_MS_ABI)
+ targetm.expand_builtin_va_start = NULL;
}
\f
/* Return true if this goes in large data/bss. */
flag_omit_frame_pointer = 2;
flag_pcc_struct_return = 2;
flag_asynchronous_unwind_tables = 2;
+ flag_vect_cost_model = 1;
#ifdef SUBTARGET_OPTIMIZATION_OPTIONS
SUBTARGET_OPTIMIZATION_OPTIONS;
#endif
warning to be generated). */
static int
-ix86_comp_type_attributes (tree type1, tree type2)
+ix86_comp_type_attributes (const_tree type1, const_tree type2)
{
/* Check for mismatch of non-default calling convention. */
const char *const rtdstr = TARGET_RTD ? "cdecl" : "stdcall";
- if (TREE_CODE (type1) != FUNCTION_TYPE)
+ if (TREE_CODE (type1) != FUNCTION_TYPE
+ && TREE_CODE (type1) != METHOD_TYPE)
return 1;
/* Check for mismatched fastcall/regparm types. */
or considering a libcall. */
static int
-ix86_function_regparm (tree type, tree decl)
+ix86_function_regparm (const_tree type, const_tree decl)
{
tree attr;
int regparm = ix86_regparm;
if (decl && TREE_CODE (decl) == FUNCTION_DECL
&& flag_unit_at_a_time && !profile_flag)
{
- struct cgraph_local_info *i = cgraph_local_info (decl);
+ /* FIXME: remove this CONST_CAST when cgraph.[ch] is constified. */
+ struct cgraph_local_info *i = cgraph_local_info (CONST_CAST_TREE(decl));
if (i && i->local)
{
int local_regparm, globals = 0, regno;
struct function *f;
/* Make sure no regparm register is taken by a
- global register variable. */
- for (local_regparm = 0; local_regparm < 3; local_regparm++)
- if (global_regs[local_regparm])
+ fixed register variable. */
+ for (local_regparm = 0; local_regparm < REGPARM_MAX; local_regparm++)
+ if (fixed_regs[local_regparm])
break;
/* We can't use regparm(3) for nested functions as these use
TYPE_ATTRIBUTES (TREE_TYPE (decl)))))
local_regparm = 2;
- /* Each global register variable increases register preassure,
- so the more global reg vars there are, the smaller regparm
- optimization use, unless requested by the user explicitly. */
- for (regno = 0; regno < 6; regno++)
- if (global_regs[regno])
+ /* Each fixed register usage increases register pressure,
+ so less registers should be used for argument passing.
+ This functionality can be overriden by an explicit
+ regparm value. */
+ for (regno = 0; regno <= DI_REG; regno++)
+ if (fixed_regs[regno])
globals++;
+
local_regparm
= globals < local_regparm ? local_regparm - globals : 0;
indirectly or considering a libcall. Otherwise return 0. */
static int
-ix86_function_sseregparm (tree type, tree decl)
+ix86_function_sseregparm (const_tree type, const_tree decl, bool warn)
{
gcc_assert (!TARGET_64BIT);
{
if (!TARGET_SSE)
{
- if (decl)
- error ("Calling %qD with attribute sseregparm without "
- "SSE/SSE2 enabled", decl);
- else
- error ("Calling %qT with attribute sseregparm without "
- "SSE/SSE2 enabled", type);
+ if (warn)
+ {
+ if (decl)
+ error ("Calling %qD with attribute sseregparm without "
+ "SSE/SSE2 enabled", decl);
+ else
+ error ("Calling %qT with attribute sseregparm without "
+ "SSE/SSE2 enabled", type);
+ }
return 0;
}
(and DFmode for SSE2) arguments in SSE registers. */
if (decl && TARGET_SSE_MATH && flag_unit_at_a_time && !profile_flag)
{
- struct cgraph_local_info *i = cgraph_local_info (decl);
+ /* FIXME: remove this CONST_CAST when cgraph.[ch] is constified. */
+ struct cgraph_local_info *i = cgraph_local_info (CONST_CAST_TREE(decl));
if (i && i->local)
return TARGET_SSE2 ? 2 : 1;
}
return REGNO_REG_SET_P (df_get_live_out (ENTRY_BLOCK_PTR), 0);
}
-/* Return true if TYPE has a variable argument list. */
-
-static bool
-type_has_variadic_args_p (tree type)
-{
- tree n, t = TYPE_ARG_TYPES (type);
-
- if (t == NULL)
- return false;
-
- while ((n = TREE_CHAIN (t)) != NULL)
- t = n;
-
- return TREE_VALUE (t) != void_type_node;
-}
-
/* Value is the number of bytes of arguments automatically
popped when returning from a subroutine call.
FUNDECL is the declaration node of the function (as a tree),
|| lookup_attribute ("fastcall", TYPE_ATTRIBUTES (funtype)))
rtd = 1;
- if (rtd && ! type_has_variadic_args_p (funtype))
+ if (rtd && ! stdarg_p (funtype))
return size;
}
}
/* RAX is used as hidden argument to va_arg functions. */
- if (!TARGET_64BIT_MS_ABI && regno == 0)
+ if (!TARGET_64BIT_MS_ABI && regno == AX_REG)
return true;
if (TARGET_64BIT_MS_ABI)
/* Return if we do not know how to pass TYPE solely in registers. */
static bool
-ix86_must_pass_in_stack (enum machine_mode mode, tree type)
+ix86_must_pass_in_stack (enum machine_mode mode, const_tree type)
{
if (must_pass_in_stack_var_size_or_pad (mode, type))
return true;
rtx libname, /* SYMBOL_REF of library name or 0 */
tree fndecl)
{
+ struct cgraph_local_info *i = fndecl ? cgraph_local_info (fndecl) : NULL;
memset (cum, 0, sizeof (*cum));
/* Set up the number of registers to use for passing arguments. */
cum->mmx_nregs = MMX_REGPARM_MAX;
cum->warn_sse = true;
cum->warn_mmx = true;
+
+ /* Because type might mismatch in between caller and callee, we need to
+ use actual type of function for local calls.
+ FIXME: cgraph_analyze can be told to actually record if function uses
+ va_start so for local functions maybe_vaarg can be made aggressive
+ helping K&R code.
+ FIXME: once typesytem is fixed, we won't need this code anymore. */
+ if (i && i->local)
+ fntype = TREE_TYPE (fndecl);
cum->maybe_vaarg = (fntype
- ? (!TYPE_ARG_TYPES (fntype)
- || type_has_variadic_args_p (fntype))
+ ? (!prototype_p (fntype) || stdarg_p (fntype))
: !libname);
if (!TARGET_64BIT)
/* Set up the number of SSE registers used for passing SFmode
and DFmode arguments. Warn for mismatching ABI. */
- cum->float_in_sse = ix86_function_sseregparm (fntype, fndecl);
+ cum->float_in_sse = ix86_function_sseregparm (fntype, fndecl, true);
}
}
the middle-end decides to do with these vector types. */
static enum machine_mode
-type_natural_mode (tree type)
+type_natural_mode (const_tree type)
{
enum machine_mode mode = TYPE_MODE (type);
*/
static int
-classify_argument (enum machine_mode mode, tree type,
+classify_argument (enum machine_mode mode, const_tree type,
enum x86_64_reg_class classes[MAX_CLASSES], int bit_offset)
{
HOST_WIDE_INT bytes =
}
/* for V1xx modes, just use the base mode */
- if (VECTOR_MODE_P (mode)
+ if (VECTOR_MODE_P (mode) && mode != V1DImode
&& GET_MODE_SIZE (GET_MODE_INNER (mode)) == bytes)
mode = GET_MODE_INNER (mode);
classes[0] = X86_64_SSE_CLASS;
classes[1] = X86_64_SSEUP_CLASS;
return 2;
+ case V1DImode:
case V2SFmode:
case V2SImode:
case V4HImode:
/* Examine the argument and return set number of register required in each
class. Return 0 iff parameter should be passed in memory. */
static int
-examine_argument (enum machine_mode mode, tree type, int in_return,
+examine_argument (enum machine_mode mode, const_tree type, int in_return,
int *int_nregs, int *sse_nregs)
{
enum x86_64_reg_class regclass[MAX_CLASSES];
static rtx
construct_container (enum machine_mode mode, enum machine_mode orig_mode,
- tree type, int in_return, int nintregs, int nsseregs,
+ const_tree type, int in_return, int nintregs, int nsseregs,
const int *intreg, int sse_regno)
{
/* The following variables hold the static issued_error state. */
case V4HImode:
case V2SImode:
case V2SFmode:
+ case V1DImode:
if (!type || !AGGREGATE_TYPE_P (type))
{
cum->mmx_words += words;
int regno = cum->regno;
/* Fastcall allocates the first two DWORD (SImode) or
- smaller arguments to ECX and EDX. */
+ smaller arguments to ECX and EDX if it isn't an
+ aggregate type . */
if (cum->fastcall)
{
- if (mode == BLKmode || mode == DImode)
+ if (mode == BLKmode
+ || mode == DImode
+ || (type && AGGREGATE_TYPE_P (type)))
break;
/* ECX not EAX is the first allocated register. */
- if (regno == 0)
- regno = 2;
+ if (regno == AX_REG)
+ regno = CX_REG;
}
return gen_rtx_REG (mode, regno);
}
case V4HImode:
case V2SImode:
case V2SFmode:
+ case V1DImode:
if (!type || !AGGREGATE_TYPE_P (type))
{
if (!TARGET_MMX && !warnedmmx && cum->warn_mmx)
static bool
ix86_pass_by_reference (CUMULATIVE_ARGS *cum ATTRIBUTE_UNUSED,
enum machine_mode mode ATTRIBUTE_UNUSED,
- tree type, bool named ATTRIBUTE_UNUSED)
+ const_tree type, bool named ATTRIBUTE_UNUSED)
{
if (TARGET_64BIT_MS_ABI)
{
static rtx
function_value_32 (enum machine_mode orig_mode, enum machine_mode mode,
- tree fntype, tree fn)
+ const_tree fntype, const_tree fn)
{
unsigned int regno;
regno = FIRST_FLOAT_REG;
else
/* Most things go in %eax. */
- regno = 0;
-
+ regno = AX_REG;
+
/* Override FP return register with %xmm0 for local functions when
SSE math is enabled or for functions with sseregparm attribute. */
if ((fn || fntype) && (mode == SFmode || mode == DFmode))
{
- int sse_level = ix86_function_sseregparm (fntype, fn);
+ int sse_level = ix86_function_sseregparm (fntype, fn, false);
if ((sse_level >= 1 && mode == SFmode)
|| (sse_level == 2 && mode == DFmode))
regno = FIRST_SSE_REG;
static rtx
function_value_64 (enum machine_mode orig_mode, enum machine_mode mode,
- tree valtype)
+ const_tree valtype)
{
rtx ret;
case TCmode:
return NULL;
default:
- return gen_rtx_REG (mode, 0);
+ return gen_rtx_REG (mode, AX_REG);
}
}
/* For zero sized structures, construct_container returns NULL, but we
need to keep rest of compiler happy by returning meaningful value. */
if (!ret)
- ret = gen_rtx_REG (orig_mode, 0);
+ ret = gen_rtx_REG (orig_mode, AX_REG);
return ret;
}
static rtx
function_value_ms_64 (enum machine_mode orig_mode, enum machine_mode mode)
{
- unsigned int regno = 0;
+ unsigned int regno = AX_REG;
if (TARGET_SSE)
{
}
static rtx
-ix86_function_value_1 (tree valtype, tree fntype_or_decl,
+ix86_function_value_1 (const_tree valtype, const_tree fntype_or_decl,
enum machine_mode orig_mode, enum machine_mode mode)
{
- tree fn, fntype;
+ const_tree fn, fntype;
fn = NULL_TREE;
if (fntype_or_decl && DECL_P (fntype_or_decl))
}
static rtx
-ix86_function_value (tree valtype, tree fntype_or_decl,
+ix86_function_value (const_tree valtype, const_tree fntype_or_decl,
bool outgoing ATTRIBUTE_UNUSED)
{
enum machine_mode mode, orig_mode;
/* Return true iff type is returned in memory. */
static int
-return_in_memory_32 (tree type, enum machine_mode mode)
+return_in_memory_32 (const_tree type, enum machine_mode mode)
{
HOST_WIDE_INT size;
}
static int
-return_in_memory_64 (tree type, enum machine_mode mode)
+return_in_memory_64 (const_tree type, enum machine_mode mode)
{
int needed_intregs, needed_sseregs;
return !examine_argument (mode, type, 1, &needed_intregs, &needed_sseregs);
}
static int
-return_in_memory_ms_64 (tree type, enum machine_mode mode)
+return_in_memory_ms_64 (const_tree type, enum machine_mode mode)
{
HOST_WIDE_INT size = int_size_in_bytes (type);
/* __m128 and friends are returned in xmm0. */
- if (size == 16 && VECTOR_MODE_P (mode))
+ if (!COMPLEX_MODE_P (mode) && size == 16 && VECTOR_MODE_P (mode))
return 0;
- /* Otherwise, the size must be exactly in [1248]. */
- return (size != 1 && size != 2 && size != 4 && size != 8);
+ /* Otherwise, the size must be exactly in [1248]. But not for complex. */
+ return (size != 1 && size != 2 && size != 4 && size != 8)
+ || COMPLEX_MODE_P (mode);
}
int
-ix86_return_in_memory (tree type)
+ix86_return_in_memory (const_tree type)
{
- enum machine_mode mode = type_natural_mode (type);
+ const enum machine_mode mode = type_natural_mode (type);
if (TARGET_64BIT_MS_ABI)
return return_in_memory_ms_64 (type, mode);
but differs notably in that when MMX is available, 8-byte vectors
are returned in memory, rather than in MMX registers. */
-int
-ix86_sol10_return_in_memory (tree type)
+int
+ix86_sol10_return_in_memory (const_tree type)
{
int size;
enum machine_mode mode = type_natural_mode (type);
ix86_save_varrargs_registers = 1;
/* We need 16-byte stack alignment to save SSE registers. If user
asked for lower preferred_stack_boundary, lets just hope that he knows
- what he is doing and won't varargs SSE values.
+ what he is doing and won't varargs SSE values.
We also may end up assuming that only 64bit values are stored in SSE
register let some floating point program work. */
label - 5*eax + nnamed_sse_arguments*5 */
tmp_reg = gen_reg_rtx (Pmode);
nsse_reg = gen_reg_rtx (Pmode);
- emit_insn (gen_zero_extendqidi2 (nsse_reg, gen_rtx_REG (QImode, 0)));
+ emit_insn (gen_zero_extendqidi2 (nsse_reg, gen_rtx_REG (QImode, AX_REG)));
emit_insn (gen_rtx_SET (VOIDmode, tmp_reg,
gen_rtx_MULT (Pmode, nsse_reg,
GEN_INT (4))));
{
CUMULATIVE_ARGS next_cum;
tree fntype;
- int stdarg_p;
/* This argument doesn't appear to be used anymore. Which is good,
because the old code here didn't suppress rtl generation. */
return;
fntype = TREE_TYPE (current_function_decl);
- stdarg_p = (TYPE_ARG_TYPES (fntype) != 0
- && (TREE_VALUE (tree_last (TYPE_ARG_TYPES (fntype)))
- != void_type_node));
/* For varargs, we do not want to skip the dummy va_dcl argument.
For stdargs, we do want to skip the last named argument. */
next_cum = *cum;
- if (stdarg_p)
+ if (stdarg_p (fntype))
function_arg_advance (&next_cum, mode, type, 1);
if (TARGET_64BIT_MS_ABI)
/* Implement va_start. */
-void
+static void
ix86_va_start (tree valist, rtx nextarg)
{
HOST_WIDE_INT words, n_gpr, n_fpr;
ix86_force_align_arg_pointer_string);
return virtual_incoming_args_rtx;
}
- cfun->machine->force_align_arg_pointer = gen_rtx_REG (Pmode, 2);
+ cfun->machine->force_align_arg_pointer = gen_rtx_REG (Pmode, CX_REG);
return copy_to_reg (cfun->machine->force_align_arg_pointer);
}
else
allocate += frame.nregs * UNITS_PER_WORD;
/* When using red zone we may start register saving before allocating
- the stack frame saving one cycle of the prologue. */
- if (TARGET_RED_ZONE && frame.save_regs_using_mov)
+ the stack frame saving one cycle of the prologue. However I will
+ avoid doing this if I am going to have to probe the stack since
+ at least on x86_64 the stack probe can turn into a call that clobbers
+ a red zone location */
+ if (TARGET_RED_ZONE && frame.save_regs_using_mov
+ && (! TARGET_STACK_PROBE || allocate < CHECK_STACK_LIMIT))
ix86_emit_save_regs_using_mov (frame_pointer_needed ? hard_frame_pointer_rtx
: stack_pointer_rtx,
-frame.nregs * UNITS_PER_WORD);
else
{
/* Only valid for Win32. */
- rtx eax = gen_rtx_REG (Pmode, 0);
+ rtx eax = gen_rtx_REG (Pmode, AX_REG);
bool eax_live;
rtx t;
}
}
- if (frame.save_regs_using_mov && !TARGET_RED_ZONE)
+ if (frame.save_regs_using_mov
+ && !(TARGET_RED_ZONE
+ && (! TARGET_STACK_PROBE || allocate < CHECK_STACK_LIMIT)))
{
if (!frame_pointer_needed || !frame.to_allocate)
ix86_emit_save_regs_using_mov (stack_pointer_rtx, frame.to_allocate);
{
if (ix86_cmodel == CM_LARGE_PIC)
{
- rtx tmp_reg = gen_rtx_REG (DImode,
- FIRST_REX_INT_REG + 3 /* R11 */);
+ rtx tmp_reg = gen_rtx_REG (DImode, R11_REG);
rtx label = gen_label_rtx ();
emit_label (label);
LABEL_PRESERVE_P (label) = 1;
insn = emit_insn (gen_set_got (pic_offset_table_rtx));
}
- /* Prevent function calls from be scheduled before the call to mcount.
+ /* Prevent function calls from being scheduled before the call to mcount.
In the pic_reg_used case, make sure that the got load isn't deleted. */
if (current_function_profile)
{
if (current_function_pops_args >= 65536)
{
- rtx ecx = gen_rtx_REG (SImode, 2);
+ rtx ecx = gen_rtx_REG (SImode, CX_REG);
/* There is no "pascal" calling convention in any 64bit ABI. */
gcc_assert (!TARGET_64BIT);
/* Special case: on K6, [%esi] makes the instruction vector decoded.
Avoid this by transforming to [%esi+0]. */
- if (ix86_tune == PROCESSOR_K6 && !optimize_size
+ if (TARGET_K6 && !optimize_size
&& base_reg && !index_reg && !disp
&& REG_P (base_reg)
&& REGNO_REG_CLASS (REGNO (base_reg)) == SIREG)
if (TARGET_64BIT && ! TARGET_GNU2_TLS)
{
- rtx rax = gen_rtx_REG (Pmode, 0), insns;
+ rtx rax = gen_rtx_REG (Pmode, AX_REG), insns;
start_sequence ();
emit_call_insn (gen_tls_global_dynamic_64 (rax, x));
if (TARGET_64BIT && ! TARGET_GNU2_TLS)
{
- rtx rax = gen_rtx_REG (Pmode, 0), insns, note;
+ rtx rax = gen_rtx_REG (Pmode, AX_REG), insns, note;
start_sequence ();
emit_call_insn (gen_tls_local_dynamic_base_64 (rax));
name = IDENTIFIER_POINTER (DECL_ASSEMBLER_NAME (decl));
name = targetm.strip_name_encoding (name);
- if (name[0] == FASTCALL_PREFIX)
- {
- name++;
- prefix = "*__imp_";
- }
- else
- prefix = "*__imp__";
-
+ prefix = name[0] == FASTCALL_PREFIX ? "*__imp_": "*__imp__";
namelen = strlen (name);
prefixlen = strlen (prefix);
imp_name = (char *) alloca (namelen + prefixlen + 1);
set_mem_alias_set (rtl, ix86_GOT_alias_set ());
SET_DECL_RTL (to, rtl);
+ SET_DECL_ASSEMBLER_NAME (to, get_identifier (name));
return to;
}
fputs ("@PLTOFF", file);
break;
case UNSPEC_GOTPCREL:
- fputs ("@GOTPCREL(%rip)", file);
+ fputs (ASSEMBLER_DIALECT == ASM_ATT ?
+ "@GOTPCREL(%rip)" : "@GOTPCREL[rip]", file);
break;
case UNSPEC_GOTTPOFF:
/* FIXME: This might be @TPOFF in Sun ld too. */
break;
case UNSPEC_GOTNTPOFF:
if (TARGET_64BIT)
- fputs ("@GOTTPOFF(%rip)", file);
+ fputs (ASSEMBLER_DIALECT == ASM_ATT ?
+ "@GOTTPOFF(%rip)": "@GOTTPOFF[rip]", file);
else
fputs ("@GOTNTPOFF", file);
break;
void
print_reg (rtx x, int code, FILE *file)
{
- gcc_assert (REGNO (x) != ARG_POINTER_REGNUM
- && REGNO (x) != FRAME_POINTER_REGNUM
- && REGNO (x) != FLAGS_REG
- && REGNO (x) != FPSR_REG
- && REGNO (x) != FPCR_REG);
+ gcc_assert (x == pc_rtx
+ || (REGNO (x) != ARG_POINTER_REGNUM
+ && REGNO (x) != FRAME_POINTER_REGNUM
+ && REGNO (x) != FLAGS_REG
+ && REGNO (x) != FPSR_REG
+ && REGNO (x) != FPCR_REG));
- if (ASSEMBLER_DIALECT == ASM_ATT || USER_LABEL_PREFIX[0] == 0)
+ if (ASSEMBLER_DIALECT == ASM_ATT)
putc ('%', file);
+ if (x == pc_rtx)
+ {
+ gcc_assert (TARGET_64BIT);
+ fputs ("rip", file);
+ return;
+ }
+
if (code == 'w' || MMX_REG_P (x))
code = 2;
else if (code == 'b')
X -- don't print any sort of PIC '@' suffix for a symbol.
& -- print some in-use local-dynamic symbol name.
H -- print a memory address offset by 8; used for sse high-parts
+ Y -- print condition for SSE5 com* instruction.
+ + -- print a branch hint as 'cs' or 'ds' prefix
+ ; -- print a semicolon (after prefixes due to bug in older gas).
*/
void
}
return;
}
+
+ case 'Y':
+ switch (GET_CODE (x))
+ {
+ case NE:
+ fputs ("neq", file);
+ break;
+ case EQ:
+ fputs ("eq", file);
+ break;
+ case GE:
+ case GEU:
+ fputs (INTEGRAL_MODE_P (GET_MODE (x)) ? "ge" : "unlt", file);
+ break;
+ case GT:
+ case GTU:
+ fputs (INTEGRAL_MODE_P (GET_MODE (x)) ? "gt" : "unle", file);
+ break;
+ case LE:
+ case LEU:
+ fputs ("le", file);
+ break;
+ case LT:
+ case LTU:
+ fputs ("lt", file);
+ break;
+ case UNORDERED:
+ fputs ("unord", file);
+ break;
+ case ORDERED:
+ fputs ("ord", file);
+ break;
+ case UNEQ:
+ fputs ("ueq", file);
+ break;
+ case UNGE:
+ fputs ("nlt", file);
+ break;
+ case UNGT:
+ fputs ("nle", file);
+ break;
+ case UNLE:
+ fputs ("ule", file);
+ break;
+ case UNLT:
+ fputs ("ult", file);
+ break;
+ case LTGT:
+ fputs ("une", file);
+ break;
+ default:
+ gcc_unreachable ();
+ }
+ return;
+
+ case ';':
+#if TARGET_MACHO
+ fputs (" ; ", file);
+#else
+ fputc (' ', file);
+#endif
+ return;
+
default:
output_operand_lossage ("invalid operand code '%c'", code);
}
else if (MEM_P (x))
{
- /* No `byte ptr' prefix for call instructions. */
- if (ASSEMBLER_DIALECT == ASM_INTEL && code != 'X' && code != 'P')
+ /* No `byte ptr' prefix for call instructions or BLKmode operands. */
+ if (ASSEMBLER_DIALECT == ASM_INTEL && code != 'X' && code != 'P'
+ && GET_MODE (x) != BLKmode)
{
const char * size;
switch (GET_MODE_SIZE (GET_MODE (x)))
case 4: size = "DWORD"; break;
case 8: size = "QWORD"; break;
case 12: size = "XWORD"; break;
- case 16: size = "XMMWORD"; break;
+ case 16:
+ if (GET_MODE (x) == XFmode)
+ size = "XWORD";
+ else
+ size = "XMMWORD";
+ break;
default:
gcc_unreachable ();
}
break;
case SEG_FS:
case SEG_GS:
- if (USER_LABEL_PREFIX[0] == 0)
+ if (ASSEMBLER_DIALECT == ASM_ATT)
putc ('%', file);
fputs ((parts.seg == SEG_FS ? "fs:" : "gs:"), file);
break;
gcc_unreachable ();
}
+ /* Use one byte shorter RIP relative addressing for 64bit mode. */
+ if (TARGET_64BIT && !base && !index)
+ {
+ rtx symbol = disp;
+
+ if (GET_CODE (disp) == CONST
+ && GET_CODE (XEXP (disp, 0)) == PLUS
+ && CONST_INT_P (XEXP (XEXP (disp, 0), 1)))
+ symbol = XEXP (XEXP (disp, 0), 0);
+
+ if (GET_CODE (symbol) == LABEL_REF
+ || (GET_CODE (symbol) == SYMBOL_REF
+ && SYMBOL_REF_TLS_MODEL (symbol) == 0))
+ base = pc_rtx;
+ }
if (!base && !index)
{
/* Displacement only requires special attention. */
if (CONST_INT_P (disp))
{
if (ASSEMBLER_DIALECT == ASM_INTEL && parts.seg == SEG_DEFAULT)
- {
- if (USER_LABEL_PREFIX[0] == 0)
- putc ('%', file);
- fputs ("ds:", file);
- }
+ fputs ("ds:", file);
fprintf (file, HOST_WIDE_INT_PRINT_DEC, INTVAL (disp));
}
else if (flag_pic)
output_pic_addr_const (file, disp, 0);
else
output_addr_const (file, disp);
-
- /* Use one byte shorter RIP relative addressing for 64bit mode. */
- if (TARGET_64BIT)
- {
- if (GET_CODE (disp) == CONST
- && GET_CODE (XEXP (disp, 0)) == PLUS
- && CONST_INT_P (XEXP (XEXP (disp, 0), 1)))
- disp = XEXP (XEXP (disp, 0), 0);
- if (GET_CODE (disp) == LABEL_REF
- || (GET_CODE (disp) == SYMBOL_REF
- && SYMBOL_REF_TLS_MODEL (disp) == 0))
- fputs ("(%rip)", file);
- }
}
else
{
case UNSPEC_GOTNTPOFF:
output_addr_const (file, op);
if (TARGET_64BIT)
- fputs ("@GOTTPOFF(%rip)", file);
+ fputs (ASSEMBLER_DIALECT == ASM_ATT ?
+ "@GOTTPOFF(%rip)" : "@GOTTPOFF[rip]", file);
else
fputs ("@GOTNTPOFF", file);
break;
/* This predicate should match that for movsi_xor and movdi_xor_rex64. */
if (reload_completed && (!TARGET_USE_MOV0 || optimize_size))
{
- rtx clob = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCmode, 17));
+ rtx clob = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCmode, FLAGS_REG));
tmp = gen_rtx_PARALLEL (VOIDmode, gen_rtvec (2, tmp, clob));
}
void
ix86_expand_move (enum machine_mode mode, rtx operands[])
{
- int strict = (reload_in_progress || reload_completed);
rtx op0, op1;
enum tls_model model;
/* Force large constants in 64bit compilation into register
to get them CSEed. */
- if (TARGET_64BIT && mode == DImode
+ if (can_create_pseudo_p ()
+ && (mode == DImode) && TARGET_64BIT
&& immediate_operand (op1, mode)
&& !x86_64_zext_immediate_operand (op1, VOIDmode)
&& !register_operand (op0, mode)
- && optimize && !reload_completed && !reload_in_progress)
+ && optimize)
op1 = copy_to_mode_reg (mode, op1);
- if (FLOAT_MODE_P (mode))
+ if (can_create_pseudo_p ()
+ && FLOAT_MODE_P (mode)
+ && GET_CODE (op1) == CONST_DOUBLE)
{
/* If we are loading a floating point constant to a register,
force the value to memory now, since we'll get better code
out the back end. */
- if (strict)
- ;
- else if (GET_CODE (op1) == CONST_DOUBLE)
+ op1 = validize_mem (force_const_mem (mode, op1));
+ if (!register_operand (op0, mode))
{
- op1 = validize_mem (force_const_mem (mode, op1));
- if (!register_operand (op0, mode))
- {
- rtx temp = gen_reg_rtx (mode);
- emit_insn (gen_rtx_SET (VOIDmode, temp, op1));
- emit_move_insn (op0, temp);
- return;
- }
+ rtx temp = gen_reg_rtx (mode);
+ emit_insn (gen_rtx_SET (VOIDmode, temp, op1));
+ emit_move_insn (op0, temp);
+ return;
}
}
}
the instructions used to build constants modify the upper 64 bits
of the register, once we have that information we may be able
to handle some of them more efficiently. */
- if ((reload_in_progress | reload_completed) == 0
+ if (can_create_pseudo_p ()
&& register_operand (op0, mode)
&& (CONSTANT_P (op1)
|| (GET_CODE (op1) == SUBREG
&& standard_sse_constant_p (op1) <= 0)
op1 = validize_mem (force_const_mem (mode, op1));
- /* TDmode values are passed as TImode on the stack. Timode values
+ /* TDmode values are passed as TImode on the stack. TImode values
are moved via xmm registers, and moving them to stack can result in
unaligned memory access. Use ix86_expand_vector_move_misalign()
if memory operand is not aligned correctly. */
ix86_expand_vector_extract (false, target, fp_xmm, 0);
}
+/* Not used, but eases macroization of patterns. */
+void
+ix86_expand_convert_uns_sixf_sse (rtx target ATTRIBUTE_UNUSED,
+ rtx input ATTRIBUTE_UNUSED)
+{
+ gcc_unreachable ();
+}
+
/* Convert an unsigned SImode value into a DFmode. Only currently used
for SSE, but applicable anywhere. */
rtx operands[])
{
rtx mask, set, use, clob, dst, src;
- bool matching_memory;
bool use_sse = false;
bool vector_mode = VECTOR_MODE_P (mode);
enum machine_mode elt_mode = mode;
dst = operands[0];
src = operands[1];
- /* If the destination is memory, and we don't have matching source
- operands or we're using the x87, do things in registers. */
- matching_memory = false;
- if (MEM_P (dst))
- {
- if (use_sse && rtx_equal_p (dst, src))
- matching_memory = true;
- else
- dst = gen_reg_rtx (mode);
- }
- if (MEM_P (src) && !matching_memory)
- src = force_reg (mode, src);
-
if (vector_mode)
{
set = gen_rtx_fmt_ee (code == NEG ? XOR : AND, mode, src, mask);
else
emit_insn (set);
}
-
- if (dst != operands[0])
- emit_move_insn (operands[0], dst);
}
/* Expand a copysign operation. Special case operand 0 being a constant. */
ix86_fp_comparison_codes (code, &bypass_code, &first_code, &second_code);
/* Do fcomi/sahf based test when profitable. */
- if ((TARGET_CMOVE || TARGET_SAHF)
+ if (ix86_fp_comparison_arithmetics_cost (code) > cost
&& (bypass_code == UNKNOWN || bypass_test)
- && (second_code == UNKNOWN || second_test)
- && ix86_fp_comparison_arithmetics_cost (code) > cost)
+ && (second_code == UNKNOWN || second_test))
{
+ tmp = gen_rtx_COMPARE (fpcmp_mode, op0, op1);
+ tmp = gen_rtx_SET (VOIDmode, gen_rtx_REG (fpcmp_mode, FLAGS_REG),
+ tmp);
if (TARGET_CMOVE)
- {
- tmp = gen_rtx_COMPARE (fpcmp_mode, op0, op1);
- tmp = gen_rtx_SET (VOIDmode, gen_rtx_REG (fpcmp_mode, FLAGS_REG),
- tmp);
- emit_insn (tmp);
- }
+ emit_insn (tmp);
else
{
- tmp = gen_rtx_COMPARE (fpcmp_mode, op0, op1);
- tmp2 = gen_rtx_UNSPEC (HImode, gen_rtvec (1, tmp), UNSPEC_FNSTSW);
+ gcc_assert (TARGET_SAHF);
+
if (!scratch)
scratch = gen_reg_rtx (HImode);
- emit_insn (gen_rtx_SET (VOIDmode, scratch, tmp2));
- emit_insn (gen_x86_sahf_1 (scratch));
+ tmp2 = gen_rtx_CLOBBER (VOIDmode, scratch);
+
+ emit_insn (gen_rtx_PARALLEL (VOIDmode, gen_rtvec (2, tmp, tmp2)));
}
/* The FP codes work out to act like unsigned. */
/* Check whether we will use the natural sequence with one jump. If
so, we can expand jump early. Otherwise delay expansion by
creating compound insn to not confuse optimizers. */
- if (bypass_code == UNKNOWN && second_code == UNKNOWN
- && TARGET_CMOVE)
+ if (bypass_code == UNKNOWN && second_code == UNKNOWN)
{
ix86_split_fp_branch (code, ix86_compare_op0, ix86_compare_op1,
gen_rtx_LABEL_REF (VOIDmode, label),
vec = rtvec_alloc (3 + !use_fcomi);
RTVEC_ELT (vec, 0) = tmp;
RTVEC_ELT (vec, 1)
- = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCFPmode, 18));
+ = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCFPmode, FPSR_REG));
RTVEC_ELT (vec, 2)
- = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCFPmode, 17));
+ = gen_rtx_CLOBBER (VOIDmode, gen_rtx_REG (CCFPmode, FLAGS_REG));
if (! use_fcomi)
RTVEC_ELT (vec, 3)
= gen_rtx_CLOBBER (VOIDmode, gen_rtx_SCRATCH (HImode));
/* Otherwise, if we are doing less-than or greater-or-equal-than,
op1 is a constant and the low word is zero, then we can just
- examine the high word. */
+ examine the high word. Similarly for low word -1 and
+ less-or-equal-than or greater-than. */
- if (CONST_INT_P (hi[1]) && lo[1] == const0_rtx)
+ if (CONST_INT_P (hi[1]))
switch (code)
{
case LT: case LTU: case GE: case GEU:
- ix86_compare_op0 = hi[0];
- ix86_compare_op1 = hi[1];
- ix86_expand_branch (code, label);
- return;
+ if (lo[1] == const0_rtx)
+ {
+ ix86_compare_op0 = hi[0];
+ ix86_compare_op1 = hi[1];
+ ix86_expand_branch (code, label);
+ return;
+ }
+ break;
+ case LE: case LEU: case GT: case GTU:
+ if (lo[1] == constm1_rtx)
+ {
+ ix86_compare_op0 = hi[0];
+ ix86_compare_op1 = hi[1];
+ ix86_expand_branch (code, label);
+ return;
+ }
+ break;
default:
break;
}
enum machine_mode mode =
GET_MODE (op0) != VOIDmode ? GET_MODE (op0) : GET_MODE (op1);
- /* Do not handle DImode compares that go through special path.
- Also we can't deal with FP compares yet. This is possible to add. */
+ /* Do not handle DImode compares that go through special path. */
if (mode == (TARGET_64BIT ? TImode : DImode))
return false;
code = swap_condition (code);
}
- /* Try to expand the comparison and verify that we end up with carry flag
- based comparison. This is fails to be true only when we decide to expand
- comparison using arithmetic that is not too common scenario. */
+ /* Try to expand the comparison and verify that we end up with
+ carry flag based comparison. This fails to be true only when
+ we decide to expand comparison using arithmetic that is not
+ too common scenario. */
start_sequence ();
compare_op = ix86_expand_fp_compare (code, op0, op1, NULL_RTX,
&second_test, &bypass_test);
if (second_test || bypass_test)
return false;
+
if (GET_MODE (XEXP (compare_op, 0)) == CCFPmode
|| GET_MODE (XEXP (compare_op, 0)) == CCFPUmode)
code = ix86_fp_compare_code_to_integer (GET_CODE (compare_op));
else
code = GET_CODE (compare_op);
+
if (code != LTU && code != GEU)
return false;
+
emit_insn (compare_seq);
*pop = compare_op;
return true;
}
+
if (!INTEGRAL_MODE_P (mode))
return false;
+
switch (code)
{
case LTU:
enum machine_mode mode = GET_MODE (dest);
rtx t2, t3, x;
- if (op_false == CONST0_RTX (mode))
+ if (TARGET_SSE5)
+ {
+ rtx pcmov = gen_rtx_SET (mode, dest,
+ gen_rtx_IF_THEN_ELSE (mode, cmp,
+ op_true,
+ op_false));
+ emit_insn (pcmov);
+ }
+ else if (op_false == CONST0_RTX (mode))
{
op_true = force_reg (mode, op_true);
x = gen_rtx_AND (mode, cmp, op_true);
emit_insn (unpack (dest, src));
}
-/* Expand conditional increment or decrement using adb/sbb instructions.
- The default case using setcc followed by the conditional move can be
- done by generic code. */
-int
-ix86_expand_int_addcc (rtx operands[])
-{
- enum rtx_code code = GET_CODE (operands[1]);
- rtx compare_op;
- rtx val = const0_rtx;
- bool fpcmp = false;
- enum machine_mode mode = GET_MODE (operands[0]);
+/* This function performs the same task as ix86_expand_sse_unpack,
+ but with amdfam15 instructions. */
- if (operands[3] != const1_rtx
- && operands[3] != constm1_rtx)
- return 0;
- if (!ix86_expand_carry_flag_compare (code, ix86_compare_op0,
- ix86_compare_op1, &compare_op))
- return 0;
- code = GET_CODE (compare_op);
+#define PPERM_SRC 0x00 /* copy source */
+#define PPERM_INVERT 0x20 /* invert source */
+#define PPERM_REVERSE 0x40 /* bit reverse source */
+#define PPERM_REV_INV 0x60 /* bit reverse & invert src */
+#define PPERM_ZERO 0x80 /* all 0's */
+#define PPERM_ONES 0xa0 /* all 1's */
+#define PPERM_SIGN 0xc0 /* propagate sign bit */
+#define PPERM_INV_SIGN 0xe0 /* invert & propagate sign */
- if (GET_MODE (XEXP (compare_op, 0)) == CCFPmode
- || GET_MODE (XEXP (compare_op, 0)) == CCFPUmode)
- {
- fpcmp = true;
- code = ix86_fp_compare_code_to_integer (code);
- }
+#define PPERM_SRC1 0x00 /* use first source byte */
+#define PPERM_SRC2 0x10 /* use second source byte */
- if (code != LTU)
+void
+ix86_expand_sse5_unpack (rtx operands[2], bool unsigned_p, bool high_p)
+{
+ enum machine_mode imode = GET_MODE (operands[1]);
+ int pperm_bytes[16];
+ int i;
+ int h = (high_p) ? 8 : 0;
+ int h2;
+ int sign_extend;
+ rtvec v = rtvec_alloc (16);
+ rtvec vs;
+ rtx x, p;
+ rtx op0 = operands[0], op1 = operands[1];
+
+ switch (imode)
{
- val = constm1_rtx;
- if (fpcmp)
- PUT_CODE (compare_op,
- reverse_condition_maybe_unordered
- (GET_CODE (compare_op)));
- else
- PUT_CODE (compare_op, reverse_condition (GET_CODE (compare_op)));
+ case V16QImode:
+ vs = rtvec_alloc (8);
+ h2 = (high_p) ? 8 : 0;
+ for (i = 0; i < 8; i++)
+ {
+ pperm_bytes[2*i+0] = PPERM_SRC | PPERM_SRC2 | i | h;
+ pperm_bytes[2*i+1] = ((unsigned_p)
+ ? PPERM_ZERO
+ : PPERM_SIGN | PPERM_SRC2 | i | h);
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ for (i = 0; i < 8; i++)
+ RTVEC_ELT (vs, i) = GEN_INT (i + h2);
+
+ p = gen_rtx_PARALLEL (VOIDmode, vs);
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ if (unsigned_p)
+ emit_insn (gen_sse5_pperm_zero_v16qi_v8hi (op0, op1, p, x));
+ else
+ emit_insn (gen_sse5_pperm_sign_v16qi_v8hi (op0, op1, p, x));
+ break;
+
+ case V8HImode:
+ vs = rtvec_alloc (4);
+ h2 = (high_p) ? 4 : 0;
+ for (i = 0; i < 4; i++)
+ {
+ sign_extend = ((unsigned_p)
+ ? PPERM_ZERO
+ : PPERM_SIGN | PPERM_SRC2 | ((2*i) + 1 + h));
+ pperm_bytes[4*i+0] = PPERM_SRC | PPERM_SRC2 | ((2*i) + 0 + h);
+ pperm_bytes[4*i+1] = PPERM_SRC | PPERM_SRC2 | ((2*i) + 1 + h);
+ pperm_bytes[4*i+2] = sign_extend;
+ pperm_bytes[4*i+3] = sign_extend;
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ for (i = 0; i < 4; i++)
+ RTVEC_ELT (vs, i) = GEN_INT (i + h2);
+
+ p = gen_rtx_PARALLEL (VOIDmode, vs);
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ if (unsigned_p)
+ emit_insn (gen_sse5_pperm_zero_v8hi_v4si (op0, op1, p, x));
+ else
+ emit_insn (gen_sse5_pperm_sign_v8hi_v4si (op0, op1, p, x));
+ break;
+
+ case V4SImode:
+ vs = rtvec_alloc (2);
+ h2 = (high_p) ? 2 : 0;
+ for (i = 0; i < 2; i++)
+ {
+ sign_extend = ((unsigned_p)
+ ? PPERM_ZERO
+ : PPERM_SIGN | PPERM_SRC2 | ((4*i) + 3 + h));
+ pperm_bytes[8*i+0] = PPERM_SRC | PPERM_SRC2 | ((4*i) + 0 + h);
+ pperm_bytes[8*i+1] = PPERM_SRC | PPERM_SRC2 | ((4*i) + 1 + h);
+ pperm_bytes[8*i+2] = PPERM_SRC | PPERM_SRC2 | ((4*i) + 2 + h);
+ pperm_bytes[8*i+3] = PPERM_SRC | PPERM_SRC2 | ((4*i) + 3 + h);
+ pperm_bytes[8*i+4] = sign_extend;
+ pperm_bytes[8*i+5] = sign_extend;
+ pperm_bytes[8*i+6] = sign_extend;
+ pperm_bytes[8*i+7] = sign_extend;
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ for (i = 0; i < 2; i++)
+ RTVEC_ELT (vs, i) = GEN_INT (i + h2);
+
+ p = gen_rtx_PARALLEL (VOIDmode, vs);
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ if (unsigned_p)
+ emit_insn (gen_sse5_pperm_zero_v4si_v2di (op0, op1, p, x));
+ else
+ emit_insn (gen_sse5_pperm_sign_v4si_v2di (op0, op1, p, x));
+ break;
+
+ default:
+ gcc_unreachable ();
+ }
+
+ return;
+}
+
+/* Pack the high bits from OPERANDS[1] and low bits from OPERANDS[2] into the
+ next narrower integer vector type */
+void
+ix86_expand_sse5_pack (rtx operands[3])
+{
+ enum machine_mode imode = GET_MODE (operands[0]);
+ int pperm_bytes[16];
+ int i;
+ rtvec v = rtvec_alloc (16);
+ rtx x;
+ rtx op0 = operands[0];
+ rtx op1 = operands[1];
+ rtx op2 = operands[2];
+
+ switch (imode)
+ {
+ case V16QImode:
+ for (i = 0; i < 8; i++)
+ {
+ pperm_bytes[i+0] = PPERM_SRC | PPERM_SRC1 | (i*2);
+ pperm_bytes[i+8] = PPERM_SRC | PPERM_SRC2 | (i*2);
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ emit_insn (gen_sse5_pperm_pack_v8hi_v16qi (op0, op1, op2, x));
+ break;
+
+ case V8HImode:
+ for (i = 0; i < 4; i++)
+ {
+ pperm_bytes[(2*i)+0] = PPERM_SRC | PPERM_SRC1 | ((i*4) + 0);
+ pperm_bytes[(2*i)+1] = PPERM_SRC | PPERM_SRC1 | ((i*4) + 1);
+ pperm_bytes[(2*i)+8] = PPERM_SRC | PPERM_SRC2 | ((i*4) + 0);
+ pperm_bytes[(2*i)+9] = PPERM_SRC | PPERM_SRC2 | ((i*4) + 1);
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ emit_insn (gen_sse5_pperm_pack_v4si_v8hi (op0, op1, op2, x));
+ break;
+
+ case V4SImode:
+ for (i = 0; i < 2; i++)
+ {
+ pperm_bytes[(4*i)+0] = PPERM_SRC | PPERM_SRC1 | ((i*8) + 0);
+ pperm_bytes[(4*i)+1] = PPERM_SRC | PPERM_SRC1 | ((i*8) + 1);
+ pperm_bytes[(4*i)+2] = PPERM_SRC | PPERM_SRC1 | ((i*8) + 2);
+ pperm_bytes[(4*i)+3] = PPERM_SRC | PPERM_SRC1 | ((i*8) + 3);
+ pperm_bytes[(4*i)+8] = PPERM_SRC | PPERM_SRC2 | ((i*8) + 0);
+ pperm_bytes[(4*i)+9] = PPERM_SRC | PPERM_SRC2 | ((i*8) + 1);
+ pperm_bytes[(4*i)+10] = PPERM_SRC | PPERM_SRC2 | ((i*8) + 2);
+ pperm_bytes[(4*i)+11] = PPERM_SRC | PPERM_SRC2 | ((i*8) + 3);
+ }
+
+ for (i = 0; i < 16; i++)
+ RTVEC_ELT (v, i) = GEN_INT (pperm_bytes[i]);
+
+ x = force_reg (V16QImode, gen_rtx_CONST_VECTOR (V16QImode, v));
+ emit_insn (gen_sse5_pperm_pack_v2di_v4si (op0, op1, op2, x));
+ break;
+
+ default:
+ gcc_unreachable ();
+ }
+
+ return;
+}
+
+/* Expand conditional increment or decrement using adb/sbb instructions.
+ The default case using setcc followed by the conditional move can be
+ done by generic code. */
+int
+ix86_expand_int_addcc (rtx operands[])
+{
+ enum rtx_code code = GET_CODE (operands[1]);
+ rtx compare_op;
+ rtx val = const0_rtx;
+ bool fpcmp = false;
+ enum machine_mode mode = GET_MODE (operands[0]);
+
+ if (operands[3] != const1_rtx
+ && operands[3] != constm1_rtx)
+ return 0;
+ if (!ix86_expand_carry_flag_compare (code, ix86_compare_op0,
+ ix86_compare_op1, &compare_op))
+ return 0;
+ code = GET_CODE (compare_op);
+
+ if (GET_MODE (XEXP (compare_op, 0)) == CCFPmode
+ || GET_MODE (XEXP (compare_op, 0)) == CCFPUmode)
+ {
+ fpcmp = true;
+ code = ix86_fp_compare_code_to_integer (code);
+ }
+
+ if (code != LTU)
+ {
+ val = constm1_rtx;
+ if (fpcmp)
+ PUT_CODE (compare_op,
+ reverse_condition_maybe_unordered
+ (GET_CODE (compare_op)));
+ else
+ PUT_CODE (compare_op, reverse_condition (GET_CODE (compare_op)));
}
PUT_MODE (compare_op, mode);
int *dynamic_check)
{
const struct stringop_algs * algs;
+ /* Algorithms using the rep prefix want at least edi and ecx;
+ additionally, memset wants eax and memcpy wants esi. Don't
+ consider such algorithms if the user has appropriated those
+ registers for their own purposes. */
+ bool rep_prefix_usable = !(fixed_regs[CX_REG] || fixed_regs[DI_REG]
+ || (memset
+ ? fixed_regs[AX_REG] : fixed_regs[SI_REG]));
+
+#define ALG_USABLE_P(alg) (rep_prefix_usable \
+ || (alg != rep_prefix_1_byte \
+ && alg != rep_prefix_4_byte \
+ && alg != rep_prefix_8_byte))
*dynamic_check = -1;
if (memset)
algs = &ix86_cost->memset[TARGET_64BIT != 0];
else
algs = &ix86_cost->memcpy[TARGET_64BIT != 0];
- if (stringop_alg != no_stringop)
+ if (stringop_alg != no_stringop && ALG_USABLE_P (stringop_alg))
return stringop_alg;
/* rep; movq or rep; movl is the smallest variant. */
else if (optimize_size)
{
if (!count || (count & 3))
- return rep_prefix_1_byte;
+ return rep_prefix_usable ? rep_prefix_1_byte : loop_1_byte;
else
- return rep_prefix_4_byte;
+ return rep_prefix_usable ? rep_prefix_4_byte : loop;
}
/* Very tiny blocks are best handled via the loop, REP is expensive to setup.
*/
enum stringop_alg alg = libcall;
for (i = 0; i < NAX_STRINGOP_ALGS; i++)
{
- gcc_assert (algs->size[i].max);
+ /* We get here if the algorithms that were not libcall-based
+ were rep-prefix based and we are unable to use rep prefixes
+ based on global register usage. Break out of the loop and
+ use the heuristic below. */
+ if (algs->size[i].max == 0)
+ break;
if (algs->size[i].max >= expected_size || algs->size[i].max == -1)
{
- if (algs->size[i].alg != libcall)
- alg = algs->size[i].alg;
+ enum stringop_alg candidate = algs->size[i].alg;
+
+ if (candidate != libcall && ALG_USABLE_P (candidate))
+ alg = candidate;
/* Honor TARGET_INLINE_ALL_STRINGOPS by picking
- last non-libcall inline algorithm. */
+ last non-libcall inline algorithm. */
if (TARGET_INLINE_ALL_STRINGOPS)
{
/* When the current size is best to be copied by a libcall,
- but we are still forced to inline, run the heuristic bellow
+ but we are still forced to inline, run the heuristic below
that will pick code for medium sized blocks. */
if (alg != libcall)
return alg;
break;
}
- else
- return algs->size[i].alg;
+ else if (ALG_USABLE_P (candidate))
+ return candidate;
}
}
- gcc_assert (TARGET_INLINE_ALL_STRINGOPS);
+ gcc_assert (TARGET_INLINE_ALL_STRINGOPS || !rep_prefix_usable);
}
/* When asked to inline the call anyway, try to pick meaningful choice.
We look for maximal size of block that is faster to copy by hand and
If this turns out to be bad, we might simply specify the preferred
choice in ix86_costs. */
if ((TARGET_INLINE_ALL_STRINGOPS || TARGET_INLINE_STRINGOPS_DYNAMICALLY)
- && algs->unknown_size == libcall)
+ && (algs->unknown_size == libcall || !ALG_USABLE_P (algs->unknown_size)))
{
int max = -1;
enum stringop_alg alg;
int i;
+ bool any_alg_usable_p = true;
for (i = 0; i < NAX_STRINGOP_ALGS; i++)
- if (algs->size[i].alg != libcall && algs->size[i].alg)
- max = algs->size[i].max;
+ {
+ enum stringop_alg candidate = algs->size[i].alg;
+ any_alg_usable_p = any_alg_usable_p && ALG_USABLE_P (candidate);
+
+ if (candidate != libcall && candidate
+ && ALG_USABLE_P (candidate))
+ max = algs->size[i].max;
+ }
+ /* If there aren't any usable algorithms, then recursing on
+ smaller sizes isn't going to find anything. Just return the
+ simple byte-at-a-time copy loop. */
+ if (!any_alg_usable_p)
+ {
+ /* Pick something reasonable. */
+ if (TARGET_INLINE_STRINGOPS_DYNAMICALLY)
+ *dynamic_check = 128;
+ return loop_1_byte;
+ }
if (max == -1)
max = 4096;
alg = decide_alg (count, max / 2, memset, dynamic_check);
*dynamic_check = max;
return alg;
}
- return algs->unknown_size;
+ return ALG_USABLE_P (algs->unknown_size) ? algs->unknown_size : libcall;
+#undef ALG_USABLE_P
}
/* Decide on alignment. We know that the operand is already aligned to ALIGN
}
/* Expand string move (memcpy) operation. Use i386 string operations when
- profitable. expand_clrmem contains similar code. The code depends upon
+ profitable. expand_setmem contains similar code. The code depends upon
architecture, block size and alignment, but always has the same
overall structure:
if (CONST_INT_P (expected_size_exp) && count == 0)
expected_size = INTVAL (expected_size_exp);
+ /* Make sure we don't need to care about overflow later on. */
+ if (count > ((unsigned HOST_WIDE_INT) 1 << 30))
+ return 0;
+
/* Step 0: Decide on preferred algorithm, desired alignment and
size of chunks to be copied by main loop. */
/* Alignment code needs count to be in register. */
if (CONST_INT_P (count_exp) && desired_align > align)
- {
- enum machine_mode mode = SImode;
- if (TARGET_64BIT && (count & ~0xffffffff))
- mode = DImode;
- count_exp = force_reg (mode, count_exp);
- }
+ count_exp = force_reg (counter_mode (count_exp), count_exp);
gcc_assert (desired_align >= 1 && align >= 1);
/* Ensure that alignment prologue won't copy past end of block. */
Make sure it is power of 2. */
epilogue_size_needed = smallest_pow2_greater_than (epilogue_size_needed);
- label = gen_label_rtx ();
- emit_cmp_and_jump_insns (count_exp,
- GEN_INT (epilogue_size_needed),
- LTU, 0, counter_mode (count_exp), 1, label);
- if (GET_CODE (count_exp) == CONST_INT)
- ;
- else if (expected_size == -1 || expected_size < epilogue_size_needed)
- predict_jump (REG_BR_PROB_BASE * 60 / 100);
+ if (CONST_INT_P (count_exp))
+ {
+ if (UINTVAL (count_exp) < (unsigned HOST_WIDE_INT)epilogue_size_needed)
+ goto epilogue;
+ }
else
- predict_jump (REG_BR_PROB_BASE * 20 / 100);
+ {
+ label = gen_label_rtx ();
+ emit_cmp_and_jump_insns (count_exp,
+ GEN_INT (epilogue_size_needed),
+ LTU, 0, counter_mode (count_exp), 1, label);
+ if (expected_size == -1 || expected_size < epilogue_size_needed)
+ predict_jump (REG_BR_PROB_BASE * 60 / 100);
+ else
+ predict_jump (REG_BR_PROB_BASE * 20 / 100);
+ }
}
+
/* Emit code to decide on runtime whether library call or inline should be
used. */
if (dynamic_check != -1)
{
- rtx hot_label = gen_label_rtx ();
- jump_around_label = gen_label_rtx ();
- emit_cmp_and_jump_insns (count_exp, GEN_INT (dynamic_check - 1),
- LEU, 0, GET_MODE (count_exp), 1, hot_label);
- predict_jump (REG_BR_PROB_BASE * 90 / 100);
- emit_block_move_via_libcall (dst, src, count_exp, false);
- emit_jump (jump_around_label);
- emit_label (hot_label);
+ if (CONST_INT_P (count_exp))
+ {
+ if (UINTVAL (count_exp) >= (unsigned HOST_WIDE_INT)dynamic_check)
+ {
+ emit_block_move_via_libcall (dst, src, count_exp, false);
+ count_exp = const0_rtx;
+ goto epilogue;
+ }
+ }
+ else
+ {
+ rtx hot_label = gen_label_rtx ();
+ jump_around_label = gen_label_rtx ();
+ emit_cmp_and_jump_insns (count_exp, GEN_INT (dynamic_check - 1),
+ LEU, 0, GET_MODE (count_exp), 1, hot_label);
+ predict_jump (REG_BR_PROB_BASE * 90 / 100);
+ emit_block_move_via_libcall (dst, src, count_exp, false);
+ emit_jump (jump_around_label);
+ emit_label (hot_label);
+ }
}
/* Step 2: Alignment prologue. */
}
/* Step 4: Epilogue to copy the remaining bytes. */
-
+ epilogue:
if (label)
{
/* When the main loop is done, COUNT_EXP might hold original count,
if (CONST_INT_P (expected_size_exp) && count == 0)
expected_size = INTVAL (expected_size_exp);
+ /* Make sure we don't need to care about overflow later on. */
+ if (count > ((unsigned HOST_WIDE_INT) 1 << 30))
+ return 0;
+
/* Step 0: Decide on preferred algorithm, desired alignment and
size of chunks to be copied by main loop. */
/* Avoid branch in fixing the byte. */
tmpreg = gen_lowpart (QImode, tmpreg);
emit_insn (gen_addqi3_cc (tmpreg, tmpreg, tmpreg));
- cmp = gen_rtx_LTU (Pmode, gen_rtx_REG (CCmode, 17), const0_rtx);
+ cmp = gen_rtx_LTU (Pmode, gen_rtx_REG (CCmode, FLAGS_REG), const0_rtx);
if (TARGET_64BIT)
emit_insn (gen_subdi3_carry_rex64 (out, out, GEN_INT (3), cmp));
else
else
{
rtx unspec;
+
+ /* Can't use this if the user has appropriated eax, ecx, or edi. */
+ if (fixed_regs[AX_REG] || fixed_regs[CX_REG] || fixed_regs[DI_REG])
+ return false;
+
scratch2 = gen_reg_rtx (Pmode);
scratch3 = gen_reg_rtx (Pmode);
scratch4 = force_reg (Pmode, constm1_rtx);
if (TARGET_64BIT && INTVAL (callarg2) >= 0)
{
- rtx al = gen_rtx_REG (QImode, 0);
+ rtx al = gen_rtx_REG (QImode, AX_REG);
emit_move_insn (al, callarg2);
use_reg (&use, al);
}
assign_386_stack_local (enum machine_mode mode, enum ix86_stack_slot n)
{
struct stack_local_entry *s;
+ int align;
gcc_assert (n < MAX_386_STACK_LOCALS);
if (s->mode == mode && s->n == n)
return copy_rtx (s->rtl);
+ /* Align DImode slots to their natural alignment
+ to avoid store forwarding stalls. */
+ if (mode == DImode
+ && (GET_MODE_BITSIZE (mode) > GET_MODE_ALIGNMENT (mode)))
+ align = GET_MODE_BITSIZE (mode);
+ else
+ align = 0;
+
s = (struct stack_local_entry *)
ggc_alloc (sizeof (struct stack_local_entry));
s->n = n;
s->mode = mode;
- s->rtl = assign_stack_local (mode, GET_MODE_SIZE (mode), 0);
+ s->rtl = assign_stack_local (mode, GET_MODE_SIZE (mode), align);
s->next = ix86_stack_locals;
ix86_stack_locals = s;
static int
ia32_multipass_dfa_lookahead (void)
{
- if (ix86_tune == PROCESSOR_PENTIUM)
- return 2;
+ switch (ix86_tune)
+ {
+ case PROCESSOR_PENTIUM:
+ return 2;
- if (ix86_tune == PROCESSOR_PENTIUMPRO
- || ix86_tune == PROCESSOR_K6)
- return 1;
+ case PROCESSOR_PENTIUMPRO:
+ case PROCESSOR_K6:
+ return 1;
- else
- return 0;
+ default:
+ return 0;
+ }
}
\f
int
ix86_constant_alignment (tree exp, int align)
{
- if (TREE_CODE (exp) == REAL_CST)
+ if (TREE_CODE (exp) == REAL_CST || TREE_CODE (exp) == VECTOR_CST
+ || TREE_CODE (exp) == INTEGER_CST)
{
if (TYPE_MODE (TREE_TYPE (exp)) == DFmode && align < 64)
return 64;
IX86_BUILTIN_RCPPS,
IX86_BUILTIN_RCPSS,
IX86_BUILTIN_RSQRTPS,
+ IX86_BUILTIN_RSQRTPS_NR,
IX86_BUILTIN_RSQRTSS,
IX86_BUILTIN_RSQRTF,
IX86_BUILTIN_SQRTPS,
+ IX86_BUILTIN_SQRTPS_NR,
IX86_BUILTIN_SQRTSS,
IX86_BUILTIN_UNPCKHPS,
IX86_BUILTIN_FABSQ,
IX86_BUILTIN_COPYSIGNQ,
+ /* SSE5 instructions */
+ IX86_BUILTIN_FMADDSS,
+ IX86_BUILTIN_FMADDSD,
+ IX86_BUILTIN_FMADDPS,
+ IX86_BUILTIN_FMADDPD,
+ IX86_BUILTIN_FMSUBSS,
+ IX86_BUILTIN_FMSUBSD,
+ IX86_BUILTIN_FMSUBPS,
+ IX86_BUILTIN_FMSUBPD,
+ IX86_BUILTIN_FNMADDSS,
+ IX86_BUILTIN_FNMADDSD,
+ IX86_BUILTIN_FNMADDPS,
+ IX86_BUILTIN_FNMADDPD,
+ IX86_BUILTIN_FNMSUBSS,
+ IX86_BUILTIN_FNMSUBSD,
+ IX86_BUILTIN_FNMSUBPS,
+ IX86_BUILTIN_FNMSUBPD,
+ IX86_BUILTIN_PCMOV_V2DI,
+ IX86_BUILTIN_PCMOV_V4SI,
+ IX86_BUILTIN_PCMOV_V8HI,
+ IX86_BUILTIN_PCMOV_V16QI,
+ IX86_BUILTIN_PCMOV_V4SF,
+ IX86_BUILTIN_PCMOV_V2DF,
+ IX86_BUILTIN_PPERM,
+ IX86_BUILTIN_PERMPS,
+ IX86_BUILTIN_PERMPD,
+ IX86_BUILTIN_PMACSSWW,
+ IX86_BUILTIN_PMACSWW,
+ IX86_BUILTIN_PMACSSWD,
+ IX86_BUILTIN_PMACSWD,
+ IX86_BUILTIN_PMACSSDD,
+ IX86_BUILTIN_PMACSDD,
+ IX86_BUILTIN_PMACSSDQL,
+ IX86_BUILTIN_PMACSSDQH,
+ IX86_BUILTIN_PMACSDQL,
+ IX86_BUILTIN_PMACSDQH,
+ IX86_BUILTIN_PMADCSSWD,
+ IX86_BUILTIN_PMADCSWD,
+ IX86_BUILTIN_PHADDBW,
+ IX86_BUILTIN_PHADDBD,
+ IX86_BUILTIN_PHADDBQ,
+ IX86_BUILTIN_PHADDWD,
+ IX86_BUILTIN_PHADDWQ,
+ IX86_BUILTIN_PHADDDQ,
+ IX86_BUILTIN_PHADDUBW,
+ IX86_BUILTIN_PHADDUBD,
+ IX86_BUILTIN_PHADDUBQ,
+ IX86_BUILTIN_PHADDUWD,
+ IX86_BUILTIN_PHADDUWQ,
+ IX86_BUILTIN_PHADDUDQ,
+ IX86_BUILTIN_PHSUBBW,
+ IX86_BUILTIN_PHSUBWD,
+ IX86_BUILTIN_PHSUBDQ,
+ IX86_BUILTIN_PROTB,
+ IX86_BUILTIN_PROTW,
+ IX86_BUILTIN_PROTD,
+ IX86_BUILTIN_PROTQ,
+ IX86_BUILTIN_PROTB_IMM,
+ IX86_BUILTIN_PROTW_IMM,
+ IX86_BUILTIN_PROTD_IMM,
+ IX86_BUILTIN_PROTQ_IMM,
+ IX86_BUILTIN_PSHLB,
+ IX86_BUILTIN_PSHLW,
+ IX86_BUILTIN_PSHLD,
+ IX86_BUILTIN_PSHLQ,
+ IX86_BUILTIN_PSHAB,
+ IX86_BUILTIN_PSHAW,
+ IX86_BUILTIN_PSHAD,
+ IX86_BUILTIN_PSHAQ,
+ IX86_BUILTIN_FRCZSS,
+ IX86_BUILTIN_FRCZSD,
+ IX86_BUILTIN_FRCZPS,
+ IX86_BUILTIN_FRCZPD,
+ IX86_BUILTIN_CVTPH2PS,
+ IX86_BUILTIN_CVTPS2PH,
+
+ IX86_BUILTIN_COMEQSS,
+ IX86_BUILTIN_COMNESS,
+ IX86_BUILTIN_COMLTSS,
+ IX86_BUILTIN_COMLESS,
+ IX86_BUILTIN_COMGTSS,
+ IX86_BUILTIN_COMGESS,
+ IX86_BUILTIN_COMUEQSS,
+ IX86_BUILTIN_COMUNESS,
+ IX86_BUILTIN_COMULTSS,
+ IX86_BUILTIN_COMULESS,
+ IX86_BUILTIN_COMUGTSS,
+ IX86_BUILTIN_COMUGESS,
+ IX86_BUILTIN_COMORDSS,
+ IX86_BUILTIN_COMUNORDSS,
+ IX86_BUILTIN_COMFALSESS,
+ IX86_BUILTIN_COMTRUESS,
+
+ IX86_BUILTIN_COMEQSD,
+ IX86_BUILTIN_COMNESD,
+ IX86_BUILTIN_COMLTSD,
+ IX86_BUILTIN_COMLESD,
+ IX86_BUILTIN_COMGTSD,
+ IX86_BUILTIN_COMGESD,
+ IX86_BUILTIN_COMUEQSD,
+ IX86_BUILTIN_COMUNESD,
+ IX86_BUILTIN_COMULTSD,
+ IX86_BUILTIN_COMULESD,
+ IX86_BUILTIN_COMUGTSD,
+ IX86_BUILTIN_COMUGESD,
+ IX86_BUILTIN_COMORDSD,
+ IX86_BUILTIN_COMUNORDSD,
+ IX86_BUILTIN_COMFALSESD,
+ IX86_BUILTIN_COMTRUESD,
+
+ IX86_BUILTIN_COMEQPS,
+ IX86_BUILTIN_COMNEPS,
+ IX86_BUILTIN_COMLTPS,
+ IX86_BUILTIN_COMLEPS,
+ IX86_BUILTIN_COMGTPS,
+ IX86_BUILTIN_COMGEPS,
+ IX86_BUILTIN_COMUEQPS,
+ IX86_BUILTIN_COMUNEPS,
+ IX86_BUILTIN_COMULTPS,
+ IX86_BUILTIN_COMULEPS,
+ IX86_BUILTIN_COMUGTPS,
+ IX86_BUILTIN_COMUGEPS,
+ IX86_BUILTIN_COMORDPS,
+ IX86_BUILTIN_COMUNORDPS,
+ IX86_BUILTIN_COMFALSEPS,
+ IX86_BUILTIN_COMTRUEPS,
+
+ IX86_BUILTIN_COMEQPD,
+ IX86_BUILTIN_COMNEPD,
+ IX86_BUILTIN_COMLTPD,
+ IX86_BUILTIN_COMLEPD,
+ IX86_BUILTIN_COMGTPD,
+ IX86_BUILTIN_COMGEPD,
+ IX86_BUILTIN_COMUEQPD,
+ IX86_BUILTIN_COMUNEPD,
+ IX86_BUILTIN_COMULTPD,
+ IX86_BUILTIN_COMULEPD,
+ IX86_BUILTIN_COMUGTPD,
+ IX86_BUILTIN_COMUGEPD,
+ IX86_BUILTIN_COMORDPD,
+ IX86_BUILTIN_COMUNORDPD,
+ IX86_BUILTIN_COMFALSEPD,
+ IX86_BUILTIN_COMTRUEPD,
+
+ IX86_BUILTIN_PCOMEQUB,
+ IX86_BUILTIN_PCOMNEUB,
+ IX86_BUILTIN_PCOMLTUB,
+ IX86_BUILTIN_PCOMLEUB,
+ IX86_BUILTIN_PCOMGTUB,
+ IX86_BUILTIN_PCOMGEUB,
+ IX86_BUILTIN_PCOMFALSEUB,
+ IX86_BUILTIN_PCOMTRUEUB,
+ IX86_BUILTIN_PCOMEQUW,
+ IX86_BUILTIN_PCOMNEUW,
+ IX86_BUILTIN_PCOMLTUW,
+ IX86_BUILTIN_PCOMLEUW,
+ IX86_BUILTIN_PCOMGTUW,
+ IX86_BUILTIN_PCOMGEUW,
+ IX86_BUILTIN_PCOMFALSEUW,
+ IX86_BUILTIN_PCOMTRUEUW,
+ IX86_BUILTIN_PCOMEQUD,
+ IX86_BUILTIN_PCOMNEUD,
+ IX86_BUILTIN_PCOMLTUD,
+ IX86_BUILTIN_PCOMLEUD,
+ IX86_BUILTIN_PCOMGTUD,
+ IX86_BUILTIN_PCOMGEUD,
+ IX86_BUILTIN_PCOMFALSEUD,
+ IX86_BUILTIN_PCOMTRUEUD,
+ IX86_BUILTIN_PCOMEQUQ,
+ IX86_BUILTIN_PCOMNEUQ,
+ IX86_BUILTIN_PCOMLTUQ,
+ IX86_BUILTIN_PCOMLEUQ,
+ IX86_BUILTIN_PCOMGTUQ,
+ IX86_BUILTIN_PCOMGEUQ,
+ IX86_BUILTIN_PCOMFALSEUQ,
+ IX86_BUILTIN_PCOMTRUEUQ,
+
+ IX86_BUILTIN_PCOMEQB,
+ IX86_BUILTIN_PCOMNEB,
+ IX86_BUILTIN_PCOMLTB,
+ IX86_BUILTIN_PCOMLEB,
+ IX86_BUILTIN_PCOMGTB,
+ IX86_BUILTIN_PCOMGEB,
+ IX86_BUILTIN_PCOMFALSEB,
+ IX86_BUILTIN_PCOMTRUEB,
+ IX86_BUILTIN_PCOMEQW,
+ IX86_BUILTIN_PCOMNEW,
+ IX86_BUILTIN_PCOMLTW,
+ IX86_BUILTIN_PCOMLEW,
+ IX86_BUILTIN_PCOMGTW,
+ IX86_BUILTIN_PCOMGEW,
+ IX86_BUILTIN_PCOMFALSEW,
+ IX86_BUILTIN_PCOMTRUEW,
+ IX86_BUILTIN_PCOMEQD,
+ IX86_BUILTIN_PCOMNED,
+ IX86_BUILTIN_PCOMLTD,
+ IX86_BUILTIN_PCOMLED,
+ IX86_BUILTIN_PCOMGTD,
+ IX86_BUILTIN_PCOMGED,
+ IX86_BUILTIN_PCOMFALSED,
+ IX86_BUILTIN_PCOMTRUED,
+ IX86_BUILTIN_PCOMEQQ,
+ IX86_BUILTIN_PCOMNEQ,
+ IX86_BUILTIN_PCOMLTQ,
+ IX86_BUILTIN_PCOMLEQ,
+ IX86_BUILTIN_PCOMGTQ,
+ IX86_BUILTIN_PCOMGEQ,
+ IX86_BUILTIN_PCOMFALSEQ,
+ IX86_BUILTIN_PCOMTRUEQ,
+
IX86_BUILTIN_MAX
};
static const struct builtin_description bdesc_ptest[] =
{
/* SSE4.1 */
- { OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestz128", IX86_BUILTIN_PTESTZ, EQ, 0 },
- { OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestc128", IX86_BUILTIN_PTESTC, LTU, 0 },
- { OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestnzc128", IX86_BUILTIN_PTESTNZC, GTU, 0 },
+ { OPTION_MASK_ISA_ROUND, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestz128", IX86_BUILTIN_PTESTZ, EQ, 0 },
+ { OPTION_MASK_ISA_ROUND, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestc128", IX86_BUILTIN_PTESTC, LTU, 0 },
+ { OPTION_MASK_ISA_ROUND, CODE_FOR_sse4_1_ptest, "__builtin_ia32_ptestnzc128", IX86_BUILTIN_PTESTNZC, GTU, 0 },
};
static const struct builtin_description bdesc_pcmpestr[] =
{ OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_mpsadbw, "__builtin_ia32_mpsadbw128", IX86_BUILTIN_MPSADBW128, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_pblendvb, "__builtin_ia32_pblendvb128", IX86_BUILTIN_PBLENDVB128, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_pblendw, "__builtin_ia32_pblendw128", IX86_BUILTIN_PBLENDW128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_roundsd, 0, IX86_BUILTIN_ROUNDSD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_roundss, 0, IX86_BUILTIN_ROUNDSS, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_ROUND, CODE_FOR_sse4_1_roundsd, 0, IX86_BUILTIN_ROUNDSD, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_ROUND, CODE_FOR_sse4_1_roundss, 0, IX86_BUILTIN_ROUNDSS, UNKNOWN, 0 },
};
static const struct builtin_description bdesc_2arg[] =
{ OPTION_MASK_ISA_SSE, CODE_FOR_addv4sf3, "__builtin_ia32_addps", IX86_BUILTIN_ADDPS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_subv4sf3, "__builtin_ia32_subps", IX86_BUILTIN_SUBPS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_mulv4sf3, "__builtin_ia32_mulps", IX86_BUILTIN_MULPS, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE, CODE_FOR_divv4sf3, "__builtin_ia32_divps", IX86_BUILTIN_DIVPS, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE, CODE_FOR_sse_divv4sf3, "__builtin_ia32_divps", IX86_BUILTIN_DIVPS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_vmaddv4sf3, "__builtin_ia32_addss", IX86_BUILTIN_ADDSS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_vmsubv4sf3, "__builtin_ia32_subss", IX86_BUILTIN_SUBSS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_vmmulv4sf3, "__builtin_ia32_mulss", IX86_BUILTIN_MULSS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_addv8qi3, "__builtin_ia32_paddb", IX86_BUILTIN_PADDB, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_addv4hi3, "__builtin_ia32_paddw", IX86_BUILTIN_PADDW, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_addv2si3, "__builtin_ia32_paddd", IX86_BUILTIN_PADDD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_mmx_adddi3, "__builtin_ia32_paddq", IX86_BUILTIN_PADDQ, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE2, CODE_FOR_mmx_addv1di3, "__builtin_ia32_paddq", IX86_BUILTIN_PADDQ, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_subv8qi3, "__builtin_ia32_psubb", IX86_BUILTIN_PSUBB, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_subv4hi3, "__builtin_ia32_psubw", IX86_BUILTIN_PSUBW, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_subv2si3, "__builtin_ia32_psubd", IX86_BUILTIN_PSUBD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_mmx_subdi3, "__builtin_ia32_psubq", IX86_BUILTIN_PSUBQ, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE2, CODE_FOR_mmx_subv1di3, "__builtin_ia32_psubq", IX86_BUILTIN_PSUBQ, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ssaddv8qi3, "__builtin_ia32_paddsb", IX86_BUILTIN_PADDSB, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ssaddv4hi3, "__builtin_ia32_paddsw", IX86_BUILTIN_PADDSW, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_cvtsi2ss, 0, IX86_BUILTIN_CVTSI2SS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_64BIT, CODE_FOR_sse_cvtsi2ssq, 0, IX86_BUILTIN_CVTSI642SS, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashlv4hi3, 0, IX86_BUILTIN_PSLLW, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashlv4hi3, 0, IX86_BUILTIN_PSLLWI, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashlv2si3, 0, IX86_BUILTIN_PSLLD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashlv2si3, 0, IX86_BUILTIN_PSLLDI, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashldi3, 0, IX86_BUILTIN_PSLLQ, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashldi3, 0, IX86_BUILTIN_PSLLQI, UNKNOWN, 0 },
-
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrv4hi3, 0, IX86_BUILTIN_PSRLW, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrv4hi3, 0, IX86_BUILTIN_PSRLWI, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrv2si3, 0, IX86_BUILTIN_PSRLD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrv2si3, 0, IX86_BUILTIN_PSRLDI, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrdi3, 0, IX86_BUILTIN_PSRLQ, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_lshrdi3, 0, IX86_BUILTIN_PSRLQI, UNKNOWN, 0 },
-
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashrv4hi3, 0, IX86_BUILTIN_PSRAW, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashrv4hi3, 0, IX86_BUILTIN_PSRAWI, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashrv2si3, 0, IX86_BUILTIN_PSRAD, UNKNOWN, 0 },
- { OPTION_MASK_ISA_MMX, CODE_FOR_mmx_ashrv2si3, 0, IX86_BUILTIN_PSRADI, UNKNOWN, 0 },
-
{ OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, CODE_FOR_mmx_psadbw, 0, IX86_BUILTIN_PSADBW, UNKNOWN, 0 },
{ OPTION_MASK_ISA_MMX, CODE_FOR_mmx_pmaddwd, 0, IX86_BUILTIN_PMADDWD, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE2, CODE_FOR_umulv8hi3_highpart, "__builtin_ia32_pmulhuw128", IX86_BUILTIN_PMULHUW128, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_psadbw, 0, IX86_BUILTIN_PSADBW128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_umulsidi3, 0, IX86_BUILTIN_PMULUDQ, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_umulv1siv1di3, 0, IX86_BUILTIN_PMULUDQ, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_umulv2siv2di3, 0, IX86_BUILTIN_PMULUDQ128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_ashlv8hi3, 0, IX86_BUILTIN_PSLLWI128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_ashlv4si3, 0, IX86_BUILTIN_PSLLDI128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_ashlv2di3, 0, IX86_BUILTIN_PSLLQI128, UNKNOWN, 0 },
-
- { OPTION_MASK_ISA_SSE2, CODE_FOR_lshrv8hi3, 0, IX86_BUILTIN_PSRLWI128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_lshrv4si3, 0, IX86_BUILTIN_PSRLDI128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_lshrv2di3, 0, IX86_BUILTIN_PSRLQI128, UNKNOWN, 0 },
-
- { OPTION_MASK_ISA_SSE2, CODE_FOR_ashrv8hi3, 0, IX86_BUILTIN_PSRAWI128, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE2, CODE_FOR_ashrv4si3, 0, IX86_BUILTIN_PSRADI128, UNKNOWN, 0 },
-
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_pmaddwd, 0, IX86_BUILTIN_PMADDWD128, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_cvtsi2sd, 0, IX86_BUILTIN_CVTSI2SD, UNKNOWN, 0 },
static const struct builtin_description bdesc_1arg[] =
{
+ /* SSE */
{ OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, CODE_FOR_mmx_pmovmskb, 0, IX86_BUILTIN_PMOVMSKB, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_movmskps, 0, IX86_BUILTIN_MOVMSKPS, UNKNOWN, 0 },
- { OPTION_MASK_ISA_SSE, CODE_FOR_sqrtv4sf2, 0, IX86_BUILTIN_SQRTPS, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE, CODE_FOR_sse_sqrtv4sf2, 0, IX86_BUILTIN_SQRTPS, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE, CODE_FOR_sqrtv4sf2, 0, IX86_BUILTIN_SQRTPS_NR, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_rsqrtv4sf2, 0, IX86_BUILTIN_RSQRTPS, UNKNOWN, 0 },
+ { OPTION_MASK_ISA_SSE, CODE_FOR_rsqrtv4sf2, 0, IX86_BUILTIN_RSQRTPS_NR, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_rcpv4sf2, 0, IX86_BUILTIN_RCPPS, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_cvtps2pi, 0, IX86_BUILTIN_CVTPS2PI, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE, CODE_FOR_sse_cvttss2si, 0, IX86_BUILTIN_CVTTSS2SI, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_64BIT, CODE_FOR_sse_cvttss2siq, 0, IX86_BUILTIN_CVTTSS2SI64, UNKNOWN, 0 },
+ /* SSE2 */
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_pmovmskb, 0, IX86_BUILTIN_PMOVMSKB128, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE2, CODE_FOR_sse2_movmskpd, 0, IX86_BUILTIN_MOVMSKPD, UNKNOWN, 0 },
{ OPTION_MASK_ISA_SSE4_1, CODE_FOR_sse4_1_roundps, 0, IX86_BUILTIN_ROUNDPS, UNKNOWN, 0 },
};
+/* SSE5 */
+enum multi_arg_type {
+ MULTI_ARG_UNKNOWN,
+ MULTI_ARG_3_SF,
+ MULTI_ARG_3_DF,
+ MULTI_ARG_3_DI,
+ MULTI_ARG_3_SI,
+ MULTI_ARG_3_SI_DI,
+ MULTI_ARG_3_HI,
+ MULTI_ARG_3_HI_SI,
+ MULTI_ARG_3_QI,
+ MULTI_ARG_3_PERMPS,
+ MULTI_ARG_3_PERMPD,
+ MULTI_ARG_2_SF,
+ MULTI_ARG_2_DF,
+ MULTI_ARG_2_DI,
+ MULTI_ARG_2_SI,
+ MULTI_ARG_2_HI,
+ MULTI_ARG_2_QI,
+ MULTI_ARG_2_DI_IMM,
+ MULTI_ARG_2_SI_IMM,
+ MULTI_ARG_2_HI_IMM,
+ MULTI_ARG_2_QI_IMM,
+ MULTI_ARG_2_SF_CMP,
+ MULTI_ARG_2_DF_CMP,
+ MULTI_ARG_2_DI_CMP,
+ MULTI_ARG_2_SI_CMP,
+ MULTI_ARG_2_HI_CMP,
+ MULTI_ARG_2_QI_CMP,
+ MULTI_ARG_2_DI_TF,
+ MULTI_ARG_2_SI_TF,
+ MULTI_ARG_2_HI_TF,
+ MULTI_ARG_2_QI_TF,
+ MULTI_ARG_2_SF_TF,
+ MULTI_ARG_2_DF_TF,
+ MULTI_ARG_1_SF,
+ MULTI_ARG_1_DF,
+ MULTI_ARG_1_DI,
+ MULTI_ARG_1_SI,
+ MULTI_ARG_1_HI,
+ MULTI_ARG_1_QI,
+ MULTI_ARG_1_SI_DI,
+ MULTI_ARG_1_HI_DI,
+ MULTI_ARG_1_HI_SI,
+ MULTI_ARG_1_QI_DI,
+ MULTI_ARG_1_QI_SI,
+ MULTI_ARG_1_QI_HI,
+ MULTI_ARG_1_PH2PS,
+ MULTI_ARG_1_PS2PH
+};
+
+static const struct builtin_description bdesc_multi_arg[] =
+{
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfmaddv4sf4, "__builtin_ia32_fmaddss", IX86_BUILTIN_FMADDSS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfmaddv2df4, "__builtin_ia32_fmaddsd", IX86_BUILTIN_FMADDSD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fmaddv4sf4, "__builtin_ia32_fmaddps", IX86_BUILTIN_FMADDPS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fmaddv2df4, "__builtin_ia32_fmaddpd", IX86_BUILTIN_FMADDPD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfmsubv4sf4, "__builtin_ia32_fmsubss", IX86_BUILTIN_FMSUBSS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfmsubv2df4, "__builtin_ia32_fmsubsd", IX86_BUILTIN_FMSUBSD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fmsubv4sf4, "__builtin_ia32_fmsubps", IX86_BUILTIN_FMSUBPS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fmsubv2df4, "__builtin_ia32_fmsubpd", IX86_BUILTIN_FMSUBPD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfnmaddv4sf4, "__builtin_ia32_fnmaddss", IX86_BUILTIN_FNMADDSS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfnmaddv2df4, "__builtin_ia32_fnmaddsd", IX86_BUILTIN_FNMADDSD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fnmaddv4sf4, "__builtin_ia32_fnmaddps", IX86_BUILTIN_FNMADDPS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fnmaddv2df4, "__builtin_ia32_fnmaddpd", IX86_BUILTIN_FNMADDPD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfnmsubv4sf4, "__builtin_ia32_fnmsubss", IX86_BUILTIN_FNMSUBSS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_vmfnmsubv2df4, "__builtin_ia32_fnmsubsd", IX86_BUILTIN_FNMSUBSD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fnmsubv4sf4, "__builtin_ia32_fnmsubps", IX86_BUILTIN_FNMSUBPS, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5i_fnmsubv2df4, "__builtin_ia32_fnmsubpd", IX86_BUILTIN_FNMSUBPD, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v2di, "__builtin_ia32_pcmov", IX86_BUILTIN_PCMOV_V2DI, 0, (int)MULTI_ARG_3_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v2di, "__builtin_ia32_pcmov_v2di", IX86_BUILTIN_PCMOV_V2DI, 0, (int)MULTI_ARG_3_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v4si, "__builtin_ia32_pcmov_v4si", IX86_BUILTIN_PCMOV_V4SI, 0, (int)MULTI_ARG_3_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v8hi, "__builtin_ia32_pcmov_v8hi", IX86_BUILTIN_PCMOV_V8HI, 0, (int)MULTI_ARG_3_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v16qi, "__builtin_ia32_pcmov_v16qi",IX86_BUILTIN_PCMOV_V16QI,0, (int)MULTI_ARG_3_QI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v2df, "__builtin_ia32_pcmov_v2df", IX86_BUILTIN_PCMOV_V2DF, 0, (int)MULTI_ARG_3_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcmov_v4sf, "__builtin_ia32_pcmov_v4sf", IX86_BUILTIN_PCMOV_V4SF, 0, (int)MULTI_ARG_3_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pperm, "__builtin_ia32_pperm", IX86_BUILTIN_PPERM, 0, (int)MULTI_ARG_3_QI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_permv4sf, "__builtin_ia32_permps", IX86_BUILTIN_PERMPS, 0, (int)MULTI_ARG_3_PERMPS },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_permv2df, "__builtin_ia32_permpd", IX86_BUILTIN_PERMPD, 0, (int)MULTI_ARG_3_PERMPD },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacssww, "__builtin_ia32_pmacssww", IX86_BUILTIN_PMACSSWW, 0, (int)MULTI_ARG_3_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacsww, "__builtin_ia32_pmacsww", IX86_BUILTIN_PMACSWW, 0, (int)MULTI_ARG_3_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacsswd, "__builtin_ia32_pmacsswd", IX86_BUILTIN_PMACSSWD, 0, (int)MULTI_ARG_3_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacswd, "__builtin_ia32_pmacswd", IX86_BUILTIN_PMACSWD, 0, (int)MULTI_ARG_3_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacssdd, "__builtin_ia32_pmacssdd", IX86_BUILTIN_PMACSSDD, 0, (int)MULTI_ARG_3_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacsdd, "__builtin_ia32_pmacsdd", IX86_BUILTIN_PMACSDD, 0, (int)MULTI_ARG_3_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacssdql, "__builtin_ia32_pmacssdql", IX86_BUILTIN_PMACSSDQL, 0, (int)MULTI_ARG_3_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacssdqh, "__builtin_ia32_pmacssdqh", IX86_BUILTIN_PMACSSDQH, 0, (int)MULTI_ARG_3_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacsdql, "__builtin_ia32_pmacsdql", IX86_BUILTIN_PMACSDQL, 0, (int)MULTI_ARG_3_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmacsdqh, "__builtin_ia32_pmacsdqh", IX86_BUILTIN_PMACSDQH, 0, (int)MULTI_ARG_3_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmadcsswd, "__builtin_ia32_pmadcsswd", IX86_BUILTIN_PMADCSSWD, 0, (int)MULTI_ARG_3_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pmadcswd, "__builtin_ia32_pmadcswd", IX86_BUILTIN_PMADCSWD, 0, (int)MULTI_ARG_3_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_rotlv2di3, "__builtin_ia32_protq", IX86_BUILTIN_PROTQ, 0, (int)MULTI_ARG_2_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_rotlv4si3, "__builtin_ia32_protd", IX86_BUILTIN_PROTD, 0, (int)MULTI_ARG_2_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_rotlv8hi3, "__builtin_ia32_protw", IX86_BUILTIN_PROTW, 0, (int)MULTI_ARG_2_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_rotlv16qi3, "__builtin_ia32_protb", IX86_BUILTIN_PROTB, 0, (int)MULTI_ARG_2_QI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_rotlv2di3, "__builtin_ia32_protqi", IX86_BUILTIN_PROTQ_IMM, 0, (int)MULTI_ARG_2_DI_IMM },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_rotlv4si3, "__builtin_ia32_protdi", IX86_BUILTIN_PROTD_IMM, 0, (int)MULTI_ARG_2_SI_IMM },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_rotlv8hi3, "__builtin_ia32_protwi", IX86_BUILTIN_PROTW_IMM, 0, (int)MULTI_ARG_2_HI_IMM },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_rotlv16qi3, "__builtin_ia32_protbi", IX86_BUILTIN_PROTB_IMM, 0, (int)MULTI_ARG_2_QI_IMM },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_ashlv2di3, "__builtin_ia32_pshaq", IX86_BUILTIN_PSHAQ, 0, (int)MULTI_ARG_2_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_ashlv4si3, "__builtin_ia32_pshad", IX86_BUILTIN_PSHAD, 0, (int)MULTI_ARG_2_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_ashlv8hi3, "__builtin_ia32_pshaw", IX86_BUILTIN_PSHAW, 0, (int)MULTI_ARG_2_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_ashlv16qi3, "__builtin_ia32_pshab", IX86_BUILTIN_PSHAB, 0, (int)MULTI_ARG_2_QI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_lshlv2di3, "__builtin_ia32_pshlq", IX86_BUILTIN_PSHLQ, 0, (int)MULTI_ARG_2_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_lshlv4si3, "__builtin_ia32_pshld", IX86_BUILTIN_PSHLD, 0, (int)MULTI_ARG_2_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_lshlv8hi3, "__builtin_ia32_pshlw", IX86_BUILTIN_PSHLW, 0, (int)MULTI_ARG_2_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_lshlv16qi3, "__builtin_ia32_pshlb", IX86_BUILTIN_PSHLB, 0, (int)MULTI_ARG_2_QI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmfrczv4sf2, "__builtin_ia32_frczss", IX86_BUILTIN_FRCZSS, 0, (int)MULTI_ARG_2_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmfrczv2df2, "__builtin_ia32_frczsd", IX86_BUILTIN_FRCZSD, 0, (int)MULTI_ARG_2_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_frczv4sf2, "__builtin_ia32_frczps", IX86_BUILTIN_FRCZPS, 0, (int)MULTI_ARG_1_SF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_frczv2df2, "__builtin_ia32_frczpd", IX86_BUILTIN_FRCZPD, 0, (int)MULTI_ARG_1_DF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_cvtph2ps, "__builtin_ia32_cvtph2ps", IX86_BUILTIN_CVTPH2PS, 0, (int)MULTI_ARG_1_PH2PS },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_cvtps2ph, "__builtin_ia32_cvtps2ph", IX86_BUILTIN_CVTPS2PH, 0, (int)MULTI_ARG_1_PS2PH },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddbw, "__builtin_ia32_phaddbw", IX86_BUILTIN_PHADDBW, 0, (int)MULTI_ARG_1_QI_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddbd, "__builtin_ia32_phaddbd", IX86_BUILTIN_PHADDBD, 0, (int)MULTI_ARG_1_QI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddbq, "__builtin_ia32_phaddbq", IX86_BUILTIN_PHADDBQ, 0, (int)MULTI_ARG_1_QI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddwd, "__builtin_ia32_phaddwd", IX86_BUILTIN_PHADDWD, 0, (int)MULTI_ARG_1_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddwq, "__builtin_ia32_phaddwq", IX86_BUILTIN_PHADDWQ, 0, (int)MULTI_ARG_1_HI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phadddq, "__builtin_ia32_phadddq", IX86_BUILTIN_PHADDDQ, 0, (int)MULTI_ARG_1_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddubw, "__builtin_ia32_phaddubw", IX86_BUILTIN_PHADDUBW, 0, (int)MULTI_ARG_1_QI_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddubd, "__builtin_ia32_phaddubd", IX86_BUILTIN_PHADDUBD, 0, (int)MULTI_ARG_1_QI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddubq, "__builtin_ia32_phaddubq", IX86_BUILTIN_PHADDUBQ, 0, (int)MULTI_ARG_1_QI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phadduwd, "__builtin_ia32_phadduwd", IX86_BUILTIN_PHADDUWD, 0, (int)MULTI_ARG_1_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phadduwq, "__builtin_ia32_phadduwq", IX86_BUILTIN_PHADDUWQ, 0, (int)MULTI_ARG_1_HI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phaddudq, "__builtin_ia32_phaddudq", IX86_BUILTIN_PHADDUDQ, 0, (int)MULTI_ARG_1_SI_DI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phsubbw, "__builtin_ia32_phsubbw", IX86_BUILTIN_PHSUBBW, 0, (int)MULTI_ARG_1_QI_HI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phsubwd, "__builtin_ia32_phsubwd", IX86_BUILTIN_PHSUBWD, 0, (int)MULTI_ARG_1_HI_SI },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_phsubdq, "__builtin_ia32_phsubdq", IX86_BUILTIN_PHSUBDQ, 0, (int)MULTI_ARG_1_SI_DI },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comeqss", IX86_BUILTIN_COMEQSS, EQ, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comness", IX86_BUILTIN_COMNESS, NE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comneqss", IX86_BUILTIN_COMNESS, NE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comltss", IX86_BUILTIN_COMLTSS, LT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comless", IX86_BUILTIN_COMLESS, LE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comgtss", IX86_BUILTIN_COMGTSS, GT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comgess", IX86_BUILTIN_COMGESS, GE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comueqss", IX86_BUILTIN_COMUEQSS, UNEQ, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comuness", IX86_BUILTIN_COMUNESS, LTGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comuneqss", IX86_BUILTIN_COMUNESS, LTGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comunltss", IX86_BUILTIN_COMULTSS, UNLT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comunless", IX86_BUILTIN_COMULESS, UNLE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comungtss", IX86_BUILTIN_COMUGTSS, UNGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comungess", IX86_BUILTIN_COMUGESS, UNGE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comordss", IX86_BUILTIN_COMORDSS, ORDERED, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv4sf3, "__builtin_ia32_comunordss", IX86_BUILTIN_COMUNORDSS, UNORDERED, (int)MULTI_ARG_2_SF_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comeqsd", IX86_BUILTIN_COMEQSD, EQ, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comnesd", IX86_BUILTIN_COMNESD, NE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comneqsd", IX86_BUILTIN_COMNESD, NE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comltsd", IX86_BUILTIN_COMLTSD, LT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comlesd", IX86_BUILTIN_COMLESD, LE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comgtsd", IX86_BUILTIN_COMGTSD, GT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comgesd", IX86_BUILTIN_COMGESD, GE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comueqsd", IX86_BUILTIN_COMUEQSD, UNEQ, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comunesd", IX86_BUILTIN_COMUNESD, LTGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comuneqsd", IX86_BUILTIN_COMUNESD, LTGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comunltsd", IX86_BUILTIN_COMULTSD, UNLT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comunlesd", IX86_BUILTIN_COMULESD, UNLE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comungtsd", IX86_BUILTIN_COMUGTSD, UNGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comungesd", IX86_BUILTIN_COMUGESD, UNGE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comordsd", IX86_BUILTIN_COMORDSD, ORDERED, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_vmmaskcmpv2df3, "__builtin_ia32_comunordsd", IX86_BUILTIN_COMUNORDSD, UNORDERED, (int)MULTI_ARG_2_DF_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comeqps", IX86_BUILTIN_COMEQPS, EQ, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comneps", IX86_BUILTIN_COMNEPS, NE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comneqps", IX86_BUILTIN_COMNEPS, NE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comltps", IX86_BUILTIN_COMLTPS, LT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comleps", IX86_BUILTIN_COMLEPS, LE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comgtps", IX86_BUILTIN_COMGTPS, GT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comgeps", IX86_BUILTIN_COMGEPS, GE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comueqps", IX86_BUILTIN_COMUEQPS, UNEQ, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comuneps", IX86_BUILTIN_COMUNEPS, LTGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comuneqps", IX86_BUILTIN_COMUNEPS, LTGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comunltps", IX86_BUILTIN_COMULTPS, UNLT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comunleps", IX86_BUILTIN_COMULEPS, UNLE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comungtps", IX86_BUILTIN_COMUGTPS, UNGT, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comungeps", IX86_BUILTIN_COMUGEPS, UNGE, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comordps", IX86_BUILTIN_COMORDPS, ORDERED, (int)MULTI_ARG_2_SF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4sf3, "__builtin_ia32_comunordps", IX86_BUILTIN_COMUNORDPS, UNORDERED, (int)MULTI_ARG_2_SF_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comeqpd", IX86_BUILTIN_COMEQPD, EQ, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comnepd", IX86_BUILTIN_COMNEPD, NE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comneqpd", IX86_BUILTIN_COMNEPD, NE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comltpd", IX86_BUILTIN_COMLTPD, LT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comlepd", IX86_BUILTIN_COMLEPD, LE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comgtpd", IX86_BUILTIN_COMGTPD, GT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comgepd", IX86_BUILTIN_COMGEPD, GE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comueqpd", IX86_BUILTIN_COMUEQPD, UNEQ, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comunepd", IX86_BUILTIN_COMUNEPD, LTGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comuneqpd", IX86_BUILTIN_COMUNEPD, LTGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comunltpd", IX86_BUILTIN_COMULTPD, UNLT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comunlepd", IX86_BUILTIN_COMULEPD, UNLE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comungtpd", IX86_BUILTIN_COMUGTPD, UNGT, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comungepd", IX86_BUILTIN_COMUGEPD, UNGE, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comordpd", IX86_BUILTIN_COMORDPD, ORDERED, (int)MULTI_ARG_2_DF_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2df3, "__builtin_ia32_comunordpd", IX86_BUILTIN_COMUNORDPD, UNORDERED, (int)MULTI_ARG_2_DF_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomeqb", IX86_BUILTIN_PCOMEQB, EQ, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomneb", IX86_BUILTIN_PCOMNEB, NE, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomneqb", IX86_BUILTIN_PCOMNEB, NE, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomltb", IX86_BUILTIN_PCOMLTB, LT, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomleb", IX86_BUILTIN_PCOMLEB, LE, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomgtb", IX86_BUILTIN_PCOMGTB, GT, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv16qi3, "__builtin_ia32_pcomgeb", IX86_BUILTIN_PCOMGEB, GE, (int)MULTI_ARG_2_QI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomeqw", IX86_BUILTIN_PCOMEQW, EQ, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomnew", IX86_BUILTIN_PCOMNEW, NE, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomneqw", IX86_BUILTIN_PCOMNEW, NE, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomltw", IX86_BUILTIN_PCOMLTW, LT, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomlew", IX86_BUILTIN_PCOMLEW, LE, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomgtw", IX86_BUILTIN_PCOMGTW, GT, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv8hi3, "__builtin_ia32_pcomgew", IX86_BUILTIN_PCOMGEW, GE, (int)MULTI_ARG_2_HI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomeqd", IX86_BUILTIN_PCOMEQD, EQ, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomned", IX86_BUILTIN_PCOMNED, NE, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomneqd", IX86_BUILTIN_PCOMNED, NE, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomltd", IX86_BUILTIN_PCOMLTD, LT, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomled", IX86_BUILTIN_PCOMLED, LE, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomgtd", IX86_BUILTIN_PCOMGTD, GT, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv4si3, "__builtin_ia32_pcomged", IX86_BUILTIN_PCOMGED, GE, (int)MULTI_ARG_2_SI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomeqq", IX86_BUILTIN_PCOMEQQ, EQ, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomneq", IX86_BUILTIN_PCOMNEQ, NE, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomneqq", IX86_BUILTIN_PCOMNEQ, NE, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomltq", IX86_BUILTIN_PCOMLTQ, LT, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomleq", IX86_BUILTIN_PCOMLEQ, LE, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomgtq", IX86_BUILTIN_PCOMGTQ, GT, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmpv2di3, "__builtin_ia32_pcomgeq", IX86_BUILTIN_PCOMGEQ, GE, (int)MULTI_ARG_2_DI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v16qi3,"__builtin_ia32_pcomequb", IX86_BUILTIN_PCOMEQUB, EQ, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v16qi3,"__builtin_ia32_pcomneub", IX86_BUILTIN_PCOMNEUB, NE, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v16qi3,"__builtin_ia32_pcomnequb", IX86_BUILTIN_PCOMNEUB, NE, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv16qi3, "__builtin_ia32_pcomltub", IX86_BUILTIN_PCOMLTUB, LTU, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv16qi3, "__builtin_ia32_pcomleub", IX86_BUILTIN_PCOMLEUB, LEU, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv16qi3, "__builtin_ia32_pcomgtub", IX86_BUILTIN_PCOMGTUB, GTU, (int)MULTI_ARG_2_QI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv16qi3, "__builtin_ia32_pcomgeub", IX86_BUILTIN_PCOMGEUB, GEU, (int)MULTI_ARG_2_QI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v8hi3, "__builtin_ia32_pcomequw", IX86_BUILTIN_PCOMEQUW, EQ, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v8hi3, "__builtin_ia32_pcomneuw", IX86_BUILTIN_PCOMNEUW, NE, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v8hi3, "__builtin_ia32_pcomnequw", IX86_BUILTIN_PCOMNEUW, NE, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv8hi3, "__builtin_ia32_pcomltuw", IX86_BUILTIN_PCOMLTUW, LTU, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv8hi3, "__builtin_ia32_pcomleuw", IX86_BUILTIN_PCOMLEUW, LEU, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv8hi3, "__builtin_ia32_pcomgtuw", IX86_BUILTIN_PCOMGTUW, GTU, (int)MULTI_ARG_2_HI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv8hi3, "__builtin_ia32_pcomgeuw", IX86_BUILTIN_PCOMGEUW, GEU, (int)MULTI_ARG_2_HI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v4si3, "__builtin_ia32_pcomequd", IX86_BUILTIN_PCOMEQUD, EQ, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v4si3, "__builtin_ia32_pcomneud", IX86_BUILTIN_PCOMNEUD, NE, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v4si3, "__builtin_ia32_pcomnequd", IX86_BUILTIN_PCOMNEUD, NE, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv4si3, "__builtin_ia32_pcomltud", IX86_BUILTIN_PCOMLTUD, LTU, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv4si3, "__builtin_ia32_pcomleud", IX86_BUILTIN_PCOMLEUD, LEU, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv4si3, "__builtin_ia32_pcomgtud", IX86_BUILTIN_PCOMGTUD, GTU, (int)MULTI_ARG_2_SI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv4si3, "__builtin_ia32_pcomgeud", IX86_BUILTIN_PCOMGEUD, GEU, (int)MULTI_ARG_2_SI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v2di3, "__builtin_ia32_pcomequq", IX86_BUILTIN_PCOMEQUQ, EQ, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v2di3, "__builtin_ia32_pcomneuq", IX86_BUILTIN_PCOMNEUQ, NE, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_uns2v2di3, "__builtin_ia32_pcomnequq", IX86_BUILTIN_PCOMNEUQ, NE, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv2di3, "__builtin_ia32_pcomltuq", IX86_BUILTIN_PCOMLTUQ, LTU, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv2di3, "__builtin_ia32_pcomleuq", IX86_BUILTIN_PCOMLEUQ, LEU, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv2di3, "__builtin_ia32_pcomgtuq", IX86_BUILTIN_PCOMGTUQ, GTU, (int)MULTI_ARG_2_DI_CMP },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_maskcmp_unsv2di3, "__builtin_ia32_pcomgeuq", IX86_BUILTIN_PCOMGEUQ, GEU, (int)MULTI_ARG_2_DI_CMP },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv4sf3, "__builtin_ia32_comfalsess", IX86_BUILTIN_COMFALSESS, COM_FALSE_S, (int)MULTI_ARG_2_SF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv4sf3, "__builtin_ia32_comtruess", IX86_BUILTIN_COMTRUESS, COM_TRUE_S, (int)MULTI_ARG_2_SF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv4sf3, "__builtin_ia32_comfalseps", IX86_BUILTIN_COMFALSEPS, COM_FALSE_P, (int)MULTI_ARG_2_SF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv4sf3, "__builtin_ia32_comtrueps", IX86_BUILTIN_COMTRUEPS, COM_TRUE_P, (int)MULTI_ARG_2_SF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv2df3, "__builtin_ia32_comfalsesd", IX86_BUILTIN_COMFALSESD, COM_FALSE_S, (int)MULTI_ARG_2_DF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv2df3, "__builtin_ia32_comtruesd", IX86_BUILTIN_COMTRUESD, COM_TRUE_S, (int)MULTI_ARG_2_DF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv2df3, "__builtin_ia32_comfalsepd", IX86_BUILTIN_COMFALSEPD, COM_FALSE_P, (int)MULTI_ARG_2_DF_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_com_tfv2df3, "__builtin_ia32_comtruepd", IX86_BUILTIN_COMTRUEPD, COM_TRUE_P, (int)MULTI_ARG_2_DF_TF },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv16qi3, "__builtin_ia32_pcomfalseb", IX86_BUILTIN_PCOMFALSEB, PCOM_FALSE, (int)MULTI_ARG_2_QI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv8hi3, "__builtin_ia32_pcomfalsew", IX86_BUILTIN_PCOMFALSEW, PCOM_FALSE, (int)MULTI_ARG_2_HI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv4si3, "__builtin_ia32_pcomfalsed", IX86_BUILTIN_PCOMFALSED, PCOM_FALSE, (int)MULTI_ARG_2_SI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv2di3, "__builtin_ia32_pcomfalseq", IX86_BUILTIN_PCOMFALSEQ, PCOM_FALSE, (int)MULTI_ARG_2_DI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv16qi3, "__builtin_ia32_pcomfalseub",IX86_BUILTIN_PCOMFALSEUB,PCOM_FALSE, (int)MULTI_ARG_2_QI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv8hi3, "__builtin_ia32_pcomfalseuw",IX86_BUILTIN_PCOMFALSEUW,PCOM_FALSE, (int)MULTI_ARG_2_HI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv4si3, "__builtin_ia32_pcomfalseud",IX86_BUILTIN_PCOMFALSEUD,PCOM_FALSE, (int)MULTI_ARG_2_SI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv2di3, "__builtin_ia32_pcomfalseuq",IX86_BUILTIN_PCOMFALSEUQ,PCOM_FALSE, (int)MULTI_ARG_2_DI_TF },
+
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv16qi3, "__builtin_ia32_pcomtrueb", IX86_BUILTIN_PCOMTRUEB, PCOM_TRUE, (int)MULTI_ARG_2_QI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv8hi3, "__builtin_ia32_pcomtruew", IX86_BUILTIN_PCOMTRUEW, PCOM_TRUE, (int)MULTI_ARG_2_HI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv4si3, "__builtin_ia32_pcomtrued", IX86_BUILTIN_PCOMTRUED, PCOM_TRUE, (int)MULTI_ARG_2_SI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv2di3, "__builtin_ia32_pcomtrueq", IX86_BUILTIN_PCOMTRUEQ, PCOM_TRUE, (int)MULTI_ARG_2_DI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv16qi3, "__builtin_ia32_pcomtrueub", IX86_BUILTIN_PCOMTRUEUB, PCOM_TRUE, (int)MULTI_ARG_2_QI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv8hi3, "__builtin_ia32_pcomtrueuw", IX86_BUILTIN_PCOMTRUEUW, PCOM_TRUE, (int)MULTI_ARG_2_HI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv4si3, "__builtin_ia32_pcomtrueud", IX86_BUILTIN_PCOMTRUEUD, PCOM_TRUE, (int)MULTI_ARG_2_SI_TF },
+ { OPTION_MASK_ISA_SSE5, CODE_FOR_sse5_pcom_tfv2di3, "__builtin_ia32_pcomtrueuq", IX86_BUILTIN_PCOMTRUEUQ, PCOM_TRUE, (int)MULTI_ARG_2_DI_TF },
+};
+
/* Set up all the MMX/SSE builtins. This is not called if TARGET_MMX
is zero. Otherwise, if TARGET_SSE is not set, only expand the MMX
builtins. */
tree V16QI_type_node = build_vector_type_for_mode (char_type_node, V16QImode);
tree V2SI_type_node = build_vector_type_for_mode (intSI_type_node, V2SImode);
+ tree V1DI_type_node
+ = build_vector_type_for_mode (long_long_integer_type_node, V1DImode);
tree V2SF_type_node = build_vector_type_for_mode (float_type_node, V2SFmode);
tree V2DI_type_node
= build_vector_type_for_mode (long_long_integer_type_node, V2DImode);
tree v4hi_ftype_v4hi_int
= build_function_type_list (V4HI_type_node,
V4HI_type_node, integer_type_node, NULL_TREE);
- tree v4hi_ftype_v4hi_di
- = build_function_type_list (V4HI_type_node,
- V4HI_type_node, long_long_unsigned_type_node,
- NULL_TREE);
- tree v2si_ftype_v2si_di
+ tree v2si_ftype_v2si_int
= build_function_type_list (V2SI_type_node,
- V2SI_type_node, long_long_unsigned_type_node,
- NULL_TREE);
+ V2SI_type_node, integer_type_node, NULL_TREE);
+ tree v1di_ftype_v1di_int
+ = build_function_type_list (V1DI_type_node,
+ V1DI_type_node, integer_type_node, NULL_TREE);
+
tree void_ftype_void
= build_function_type (void_type_node, void_list_node);
tree void_ftype_unsigned
tree v2si_ftype_v2si_v2si
= build_function_type_list (V2SI_type_node,
V2SI_type_node, V2SI_type_node, NULL_TREE);
- tree di_ftype_di_di
- = build_function_type_list (long_long_unsigned_type_node,
- long_long_unsigned_type_node,
- long_long_unsigned_type_node, NULL_TREE);
+ tree v1di_ftype_v1di_v1di
+ = build_function_type_list (V1DI_type_node,
+ V1DI_type_node, V1DI_type_node, NULL_TREE);
tree di_ftype_di_di_int
= build_function_type_list (long_long_unsigned_type_node,
tree v4si_ftype_v8hi_v8hi
= build_function_type_list (V4SI_type_node,
V8HI_type_node, V8HI_type_node, NULL_TREE);
- tree di_ftype_v8qi_v8qi
- = build_function_type_list (long_long_unsigned_type_node,
+ tree v1di_ftype_v8qi_v8qi
+ = build_function_type_list (V1DI_type_node,
V8QI_type_node, V8QI_type_node, NULL_TREE);
- tree di_ftype_v2si_v2si
- = build_function_type_list (long_long_unsigned_type_node,
+ tree v1di_ftype_v2si_v2si
+ = build_function_type_list (V1DI_type_node,
V2SI_type_node, V2SI_type_node, NULL_TREE);
tree v2di_ftype_v16qi_v16qi
= build_function_type_list (V2DI_type_node,
V16QI_type_node,
integer_type_node,
NULL_TREE);
+
+ /* SSE5 instructions */
+ tree v2di_ftype_v2di_v2di_v2di
+ = build_function_type_list (V2DI_type_node,
+ V2DI_type_node,
+ V2DI_type_node,
+ V2DI_type_node,
+ NULL_TREE);
+
+ tree v4si_ftype_v4si_v4si_v4si
+ = build_function_type_list (V4SI_type_node,
+ V4SI_type_node,
+ V4SI_type_node,
+ V4SI_type_node,
+ NULL_TREE);
+
+ tree v4si_ftype_v4si_v4si_v2di
+ = build_function_type_list (V4SI_type_node,
+ V4SI_type_node,
+ V4SI_type_node,
+ V2DI_type_node,
+ NULL_TREE);
+
+ tree v8hi_ftype_v8hi_v8hi_v8hi
+ = build_function_type_list (V8HI_type_node,
+ V8HI_type_node,
+ V8HI_type_node,
+ V8HI_type_node,
+ NULL_TREE);
+
+ tree v8hi_ftype_v8hi_v8hi_v4si
+ = build_function_type_list (V8HI_type_node,
+ V8HI_type_node,
+ V8HI_type_node,
+ V4SI_type_node,
+ NULL_TREE);
+
+ tree v2df_ftype_v2df_v2df_v16qi
+ = build_function_type_list (V2DF_type_node,
+ V2DF_type_node,
+ V2DF_type_node,
+ V16QI_type_node,
+ NULL_TREE);
+
+ tree v4sf_ftype_v4sf_v4sf_v16qi
+ = build_function_type_list (V4SF_type_node,
+ V4SF_type_node,
+ V4SF_type_node,
+ V16QI_type_node,
+ NULL_TREE);
+
+ tree v2di_ftype_v2di_si
+ = build_function_type_list (V2DI_type_node,
+ V2DI_type_node,
+ integer_type_node,
+ NULL_TREE);
+
+ tree v4si_ftype_v4si_si
+ = build_function_type_list (V4SI_type_node,
+ V4SI_type_node,
+ integer_type_node,
+ NULL_TREE);
+
+ tree v8hi_ftype_v8hi_si
+ = build_function_type_list (V8HI_type_node,
+ V8HI_type_node,
+ integer_type_node,
+ NULL_TREE);
+
+ tree v16qi_ftype_v16qi_si
+ = build_function_type_list (V16QI_type_node,
+ V16QI_type_node,
+ integer_type_node,
+ NULL_TREE);
+ tree v4sf_ftype_v4hi
+ = build_function_type_list (V4SF_type_node,
+ V4HI_type_node,
+ NULL_TREE);
+
+ tree v4hi_ftype_v4sf
+ = build_function_type_list (V4HI_type_node,
+ V4SF_type_node,
+ NULL_TREE);
+
+ tree v2di_ftype_v2di
+ = build_function_type_list (V2DI_type_node, V2DI_type_node, NULL_TREE);
+
tree ftype;
/* The __float80 type. */
case V2SImode:
type = v2si_ftype_v2si_v2si;
break;
- case DImode:
- type = di_ftype_di_di;
+ case V1DImode:
+ type = v1di_ftype_v1di_v1di;
break;
default:
/* Add the remaining MMX insns with somewhat more complicated types. */
def_builtin (OPTION_MASK_ISA_MMX, "__builtin_ia32_emms", void_ftype_void, IX86_BUILTIN_EMMS);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllw", v4hi_ftype_v4hi_di, IX86_BUILTIN_PSLLW);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_pslld", v2si_ftype_v2si_di, IX86_BUILTIN_PSLLD);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllq", di_ftype_di_di, IX86_BUILTIN_PSLLQ);
-
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlw", v4hi_ftype_v4hi_di, IX86_BUILTIN_PSRLW);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrld", v2si_ftype_v2si_di, IX86_BUILTIN_PSRLD);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlq", di_ftype_di_di, IX86_BUILTIN_PSRLQ);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psraw", v4hi_ftype_v4hi_di, IX86_BUILTIN_PSRAW);
- def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrad", v2si_ftype_v2si_di, IX86_BUILTIN_PSRAD);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllwi", v4hi_ftype_v4hi_int, IX86_BUILTIN_PSLLWI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_pslldi", v2si_ftype_v2si_int, IX86_BUILTIN_PSLLDI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllqi", v1di_ftype_v1di_int, IX86_BUILTIN_PSLLQI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllw", v4hi_ftype_v4hi_v4hi, IX86_BUILTIN_PSLLW);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_pslld", v2si_ftype_v2si_v2si, IX86_BUILTIN_PSLLD);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psllq", v1di_ftype_v1di_v1di, IX86_BUILTIN_PSLLQ);
+
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlwi", v4hi_ftype_v4hi_int, IX86_BUILTIN_PSRLWI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrldi", v2si_ftype_v2si_int, IX86_BUILTIN_PSRLDI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlqi", v1di_ftype_v1di_int, IX86_BUILTIN_PSRLQI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlw", v4hi_ftype_v4hi_v4hi, IX86_BUILTIN_PSRLW);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrld", v2si_ftype_v2si_v2si, IX86_BUILTIN_PSRLD);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrlq", v1di_ftype_v1di_v1di, IX86_BUILTIN_PSRLQ);
+
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrawi", v4hi_ftype_v4hi_int, IX86_BUILTIN_PSRAWI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psradi", v2si_ftype_v2si_int, IX86_BUILTIN_PSRADI);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psraw", v4hi_ftype_v4hi_v4hi, IX86_BUILTIN_PSRAW);
+ def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_psrad", v2si_ftype_v2si_v2si, IX86_BUILTIN_PSRAD);
def_builtin_const (OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, "__builtin_ia32_pshufw", v4hi_ftype_v4hi_int, IX86_BUILTIN_PSHUFW);
def_builtin_const (OPTION_MASK_ISA_MMX, "__builtin_ia32_pmaddwd", v2si_ftype_v4hi_v4hi, IX86_BUILTIN_PMADDWD);
def_builtin (OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, "__builtin_ia32_sfence", void_ftype_void, IX86_BUILTIN_SFENCE);
- def_builtin_const (OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, "__builtin_ia32_psadbw", di_ftype_v8qi_v8qi, IX86_BUILTIN_PSADBW);
+ def_builtin_const (OPTION_MASK_ISA_SSE | OPTION_MASK_ISA_3DNOW_A, "__builtin_ia32_psadbw", v1di_ftype_v8qi_v8qi, IX86_BUILTIN_PSADBW);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rcpps", v4sf_ftype_v4sf, IX86_BUILTIN_RCPPS);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rcpss", v4sf_ftype_v4sf, IX86_BUILTIN_RCPSS);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rsqrtps", v4sf_ftype_v4sf, IX86_BUILTIN_RSQRTPS);
+ def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rsqrtps_nr", v4sf_ftype_v4sf, IX86_BUILTIN_RSQRTPS_NR);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rsqrtss", v4sf_ftype_v4sf, IX86_BUILTIN_RSQRTSS);
ftype = build_function_type_list (float_type_node,
float_type_node,
NULL_TREE);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_rsqrtf", ftype, IX86_BUILTIN_RSQRTF);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_sqrtps", v4sf_ftype_v4sf, IX86_BUILTIN_SQRTPS);
+ def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_sqrtps_nr", v4sf_ftype_v4sf, IX86_BUILTIN_SQRTPS_NR);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_sqrtss", v4sf_ftype_v4sf, IX86_BUILTIN_SQRTSS);
def_builtin_const (OPTION_MASK_ISA_SSE, "__builtin_ia32_shufps", v4sf_ftype_v4sf_v4sf_int, IX86_BUILTIN_SHUFPS);
def_builtin (OPTION_MASK_ISA_SSE2, "__builtin_ia32_loaddqu", v16qi_ftype_pcchar, IX86_BUILTIN_LOADDQU);
def_builtin (OPTION_MASK_ISA_SSE2, "__builtin_ia32_storedqu", void_ftype_pchar_v16qi, IX86_BUILTIN_STOREDQU);
- def_builtin_const (OPTION_MASK_ISA_SSE2, "__builtin_ia32_pmuludq", di_ftype_v2si_v2si, IX86_BUILTIN_PMULUDQ);
+ def_builtin_const (OPTION_MASK_ISA_SSE2, "__builtin_ia32_pmuludq", v1di_ftype_v2si_v2si, IX86_BUILTIN_PMULUDQ);
def_builtin_const (OPTION_MASK_ISA_SSE2, "__builtin_ia32_pmuludq128", v2di_ftype_v4si_v4si, IX86_BUILTIN_PMULUDQ128);
def_builtin_const (OPTION_MASK_ISA_SSE2, "__builtin_ia32_pslldqi128", v2di_ftype_v2di_int, IX86_BUILTIN_PSLLDQI128);
def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_pmovzxwq128", v2di_ftype_v8hi, IX86_BUILTIN_PMOVZXWQ128);
def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_pmovzxdq128", v2di_ftype_v4si, IX86_BUILTIN_PMOVZXDQ128);
def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_pmuldq128", v2di_ftype_v4si_v4si, IX86_BUILTIN_PMULDQ128);
- def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_roundpd", v2df_ftype_v2df_int, IX86_BUILTIN_ROUNDPD);
- def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_roundps", v4sf_ftype_v4sf_int, IX86_BUILTIN_ROUNDPS);
- def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_roundsd", v2df_ftype_v2df_v2df_int, IX86_BUILTIN_ROUNDSD);
- def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_roundss", v4sf_ftype_v4sf_v4sf_int, IX86_BUILTIN_ROUNDSS);
+
+ /* SSE4.1 and SSE5 */
+ def_builtin_const (OPTION_MASK_ISA_ROUND, "__builtin_ia32_roundpd", v2df_ftype_v2df_int, IX86_BUILTIN_ROUNDPD);
+ def_builtin_const (OPTION_MASK_ISA_ROUND, "__builtin_ia32_roundps", v4sf_ftype_v4sf_int, IX86_BUILTIN_ROUNDPS);
+ def_builtin_const (OPTION_MASK_ISA_ROUND, "__builtin_ia32_roundsd", v2df_ftype_v2df_v2df_int, IX86_BUILTIN_ROUNDSD);
+ def_builtin_const (OPTION_MASK_ISA_ROUND, "__builtin_ia32_roundss", v4sf_ftype_v4sf_v4sf_int, IX86_BUILTIN_ROUNDSS);
/* SSE4.2. */
ftype = build_function_type_list (unsigned_type_node,
intQI_type_node,
integer_type_node, NULL_TREE);
def_builtin_const (OPTION_MASK_ISA_SSE4_1, "__builtin_ia32_vec_set_v16qi", ftype, IX86_BUILTIN_VEC_SET_V16QI);
+
+ /* Add SSE5 multi-arg argument instructions */
+ for (i = 0, d = bdesc_multi_arg; i < ARRAY_SIZE (bdesc_multi_arg); i++, d++)
+ {
+ tree mtype = NULL_TREE;
+
+ if (d->name == 0)
+ continue;
+
+ switch ((enum multi_arg_type)d->flag)
+ {
+ case MULTI_ARG_3_SF: mtype = v4sf_ftype_v4sf_v4sf_v4sf; break;
+ case MULTI_ARG_3_DF: mtype = v2df_ftype_v2df_v2df_v2df; break;
+ case MULTI_ARG_3_DI: mtype = v2di_ftype_v2di_v2di_v2di; break;
+ case MULTI_ARG_3_SI: mtype = v4si_ftype_v4si_v4si_v4si; break;
+ case MULTI_ARG_3_SI_DI: mtype = v4si_ftype_v4si_v4si_v2di; break;
+ case MULTI_ARG_3_HI: mtype = v8hi_ftype_v8hi_v8hi_v8hi; break;
+ case MULTI_ARG_3_HI_SI: mtype = v8hi_ftype_v8hi_v8hi_v4si; break;
+ case MULTI_ARG_3_QI: mtype = v16qi_ftype_v16qi_v16qi_v16qi; break;
+ case MULTI_ARG_3_PERMPS: mtype = v4sf_ftype_v4sf_v4sf_v16qi; break;
+ case MULTI_ARG_3_PERMPD: mtype = v2df_ftype_v2df_v2df_v16qi; break;
+ case MULTI_ARG_2_SF: mtype = v4sf_ftype_v4sf_v4sf; break;
+ case MULTI_ARG_2_DF: mtype = v2df_ftype_v2df_v2df; break;
+ case MULTI_ARG_2_DI: mtype = v2di_ftype_v2di_v2di; break;
+ case MULTI_ARG_2_SI: mtype = v4si_ftype_v4si_v4si; break;
+ case MULTI_ARG_2_HI: mtype = v8hi_ftype_v8hi_v8hi; break;
+ case MULTI_ARG_2_QI: mtype = v16qi_ftype_v16qi_v16qi; break;
+ case MULTI_ARG_2_DI_IMM: mtype = v2di_ftype_v2di_si; break;
+ case MULTI_ARG_2_SI_IMM: mtype = v4si_ftype_v4si_si; break;
+ case MULTI_ARG_2_HI_IMM: mtype = v8hi_ftype_v8hi_si; break;
+ case MULTI_ARG_2_QI_IMM: mtype = v16qi_ftype_v16qi_si; break;
+ case MULTI_ARG_2_SF_CMP: mtype = v4sf_ftype_v4sf_v4sf; break;
+ case MULTI_ARG_2_DF_CMP: mtype = v2df_ftype_v2df_v2df; break;
+ case MULTI_ARG_2_DI_CMP: mtype = v2di_ftype_v2di_v2di; break;
+ case MULTI_ARG_2_SI_CMP: mtype = v4si_ftype_v4si_v4si; break;
+ case MULTI_ARG_2_HI_CMP: mtype = v8hi_ftype_v8hi_v8hi; break;
+ case MULTI_ARG_2_QI_CMP: mtype = v16qi_ftype_v16qi_v16qi; break;
+ case MULTI_ARG_2_SF_TF: mtype = v4sf_ftype_v4sf_v4sf; break;
+ case MULTI_ARG_2_DF_TF: mtype = v2df_ftype_v2df_v2df; break;
+ case MULTI_ARG_2_DI_TF: mtype = v2di_ftype_v2di_v2di; break;
+ case MULTI_ARG_2_SI_TF: mtype = v4si_ftype_v4si_v4si; break;
+ case MULTI_ARG_2_HI_TF: mtype = v8hi_ftype_v8hi_v8hi; break;
+ case MULTI_ARG_2_QI_TF: mtype = v16qi_ftype_v16qi_v16qi; break;
+ case MULTI_ARG_1_SF: mtype = v4sf_ftype_v4sf; break;
+ case MULTI_ARG_1_DF: mtype = v2df_ftype_v2df; break;
+ case MULTI_ARG_1_DI: mtype = v2di_ftype_v2di; break;
+ case MULTI_ARG_1_SI: mtype = v4si_ftype_v4si; break;
+ case MULTI_ARG_1_HI: mtype = v8hi_ftype_v8hi; break;
+ case MULTI_ARG_1_QI: mtype = v16qi_ftype_v16qi; break;
+ case MULTI_ARG_1_SI_DI: mtype = v2di_ftype_v4si; break;
+ case MULTI_ARG_1_HI_DI: mtype = v2di_ftype_v8hi; break;
+ case MULTI_ARG_1_HI_SI: mtype = v4si_ftype_v8hi; break;
+ case MULTI_ARG_1_QI_DI: mtype = v2di_ftype_v16qi; break;
+ case MULTI_ARG_1_QI_SI: mtype = v4si_ftype_v16qi; break;
+ case MULTI_ARG_1_QI_HI: mtype = v8hi_ftype_v16qi; break;
+ case MULTI_ARG_1_PH2PS: mtype = v4sf_ftype_v4hi; break;
+ case MULTI_ARG_1_PS2PH: mtype = v4hi_ftype_v4sf; break;
+ case MULTI_ARG_UNKNOWN:
+ default:
+ gcc_unreachable ();
+ }
+
+ if (mtype)
+ def_builtin_const (d->mask, d->name, mtype, d->code);
+ }
}
static void
return target;
}
-/* Subroutine of ix86_expand_builtin to take care of stores. */
+/* Subroutine of ix86_expand_builtin to take care of 2-4 argument insns. */
static rtx
-ix86_expand_store_builtin (enum insn_code icode, tree exp)
+ix86_expand_multi_arg_builtin (enum insn_code icode, tree exp, rtx target,
+ enum multi_arg_type m_type,
+ enum insn_code sub_code)
{
rtx pat;
- tree arg0 = CALL_EXPR_ARG (exp, 0);
- tree arg1 = CALL_EXPR_ARG (exp, 1);
- rtx op0 = expand_normal (arg0);
- rtx op1 = expand_normal (arg1);
- enum machine_mode mode0 = insn_data[icode].operand[0].mode;
- enum machine_mode mode1 = insn_data[icode].operand[1].mode;
+ int i;
+ int nargs;
+ bool comparison_p = false;
+ bool tf_p = false;
+ bool last_arg_constant = false;
+ int num_memory = 0;
+ struct {
+ rtx op;
+ enum machine_mode mode;
+ } args[4];
+
+ enum machine_mode tmode = insn_data[icode].operand[0].mode;
+
+ switch (m_type)
+ {
+ case MULTI_ARG_3_SF:
+ case MULTI_ARG_3_DF:
+ case MULTI_ARG_3_DI:
+ case MULTI_ARG_3_SI:
+ case MULTI_ARG_3_SI_DI:
+ case MULTI_ARG_3_HI:
+ case MULTI_ARG_3_HI_SI:
+ case MULTI_ARG_3_QI:
+ case MULTI_ARG_3_PERMPS:
+ case MULTI_ARG_3_PERMPD:
+ nargs = 3;
+ break;
+
+ case MULTI_ARG_2_SF:
+ case MULTI_ARG_2_DF:
+ case MULTI_ARG_2_DI:
+ case MULTI_ARG_2_SI:
+ case MULTI_ARG_2_HI:
+ case MULTI_ARG_2_QI:
+ nargs = 2;
+ break;
+
+ case MULTI_ARG_2_DI_IMM:
+ case MULTI_ARG_2_SI_IMM:
+ case MULTI_ARG_2_HI_IMM:
+ case MULTI_ARG_2_QI_IMM:
+ nargs = 2;
+ last_arg_constant = true;
+ break;
+
+ case MULTI_ARG_1_SF:
+ case MULTI_ARG_1_DF:
+ case MULTI_ARG_1_DI:
+ case MULTI_ARG_1_SI:
+ case MULTI_ARG_1_HI:
+ case MULTI_ARG_1_QI:
+ case MULTI_ARG_1_SI_DI:
+ case MULTI_ARG_1_HI_DI:
+ case MULTI_ARG_1_HI_SI:
+ case MULTI_ARG_1_QI_DI:
+ case MULTI_ARG_1_QI_SI:
+ case MULTI_ARG_1_QI_HI:
+ case MULTI_ARG_1_PH2PS:
+ case MULTI_ARG_1_PS2PH:
+ nargs = 1;
+ break;
+
+ case MULTI_ARG_2_SF_CMP:
+ case MULTI_ARG_2_DF_CMP:
+ case MULTI_ARG_2_DI_CMP:
+ case MULTI_ARG_2_SI_CMP:
+ case MULTI_ARG_2_HI_CMP:
+ case MULTI_ARG_2_QI_CMP:
+ nargs = 2;
+ comparison_p = true;
+ break;
+
+ case MULTI_ARG_2_SF_TF:
+ case MULTI_ARG_2_DF_TF:
+ case MULTI_ARG_2_DI_TF:
+ case MULTI_ARG_2_SI_TF:
+ case MULTI_ARG_2_HI_TF:
+ case MULTI_ARG_2_QI_TF:
+ nargs = 2;
+ tf_p = true;
+ break;
+
+ case MULTI_ARG_UNKNOWN:
+ default:
+ gcc_unreachable ();
+ }
+
+ if (optimize || !target
+ || GET_MODE (target) != tmode
+ || ! (*insn_data[icode].operand[0].predicate) (target, tmode))
+ target = gen_reg_rtx (tmode);
+
+ gcc_assert (nargs <= 4);
+
+ for (i = 0; i < nargs; i++)
+ {
+ tree arg = CALL_EXPR_ARG (exp, i);
+ rtx op = expand_normal (arg);
+ int adjust = (comparison_p) ? 1 : 0;
+ enum machine_mode mode = insn_data[icode].operand[i+adjust+1].mode;
+
+ if (last_arg_constant && i == nargs-1)
+ {
+ if (GET_CODE (op) != CONST_INT)
+ {
+ error ("last argument must be an immediate");
+ return gen_reg_rtx (tmode);
+ }
+ }
+ else
+ {
+ if (VECTOR_MODE_P (mode))
+ op = safe_vector_operand (op, mode);
+
+ /* If we aren't optimizing, only allow one memory operand to be
+ generated. */
+ if (memory_operand (op, mode))
+ num_memory++;
+
+ gcc_assert (GET_MODE (op) == mode || GET_MODE (op) == VOIDmode);
+
+ if (optimize
+ || ! (*insn_data[icode].operand[i+adjust+1].predicate) (op, mode)
+ || num_memory > 1)
+ op = force_reg (mode, op);
+ }
+
+ args[i].op = op;
+ args[i].mode = mode;
+ }
+
+ switch (nargs)
+ {
+ case 1:
+ pat = GEN_FCN (icode) (target, args[0].op);
+ break;
+
+ case 2:
+ if (tf_p)
+ pat = GEN_FCN (icode) (target, args[0].op, args[1].op,
+ GEN_INT ((int)sub_code));
+ else if (! comparison_p)
+ pat = GEN_FCN (icode) (target, args[0].op, args[1].op);
+ else
+ {
+ rtx cmp_op = gen_rtx_fmt_ee (sub_code, GET_MODE (target),
+ args[0].op,
+ args[1].op);
+
+ pat = GEN_FCN (icode) (target, cmp_op, args[0].op, args[1].op);
+ }
+ break;
+
+ case 3:
+ pat = GEN_FCN (icode) (target, args[0].op, args[1].op, args[2].op);
+ break;
+
+ default:
+ gcc_unreachable ();
+ }
+
+ if (! pat)
+ return 0;
+
+ emit_insn (pat);
+ return target;
+}
+
+/* Subroutine of ix86_expand_builtin to take care of stores. */
+
+static rtx
+ix86_expand_store_builtin (enum insn_code icode, tree exp)
+{
+ rtx pat;
+ tree arg0 = CALL_EXPR_ARG (exp, 0);
+ tree arg1 = CALL_EXPR_ARG (exp, 1);
+ rtx op0 = expand_normal (arg0);
+ rtx op1 = expand_normal (arg1);
+ enum machine_mode mode0 = insn_data[icode].operand[0].mode;
+ enum machine_mode mode1 = insn_data[icode].operand[1].mode;
if (VECTOR_MODE_P (mode1))
op1 = safe_vector_operand (op1, mode1);
emit_insn (pat);
return target;
- case IX86_BUILTIN_PSLLWI128:
- icode = CODE_FOR_ashlv8hi3;
- goto do_pshifti;
- case IX86_BUILTIN_PSLLDI128:
- icode = CODE_FOR_ashlv4si3;
- goto do_pshifti;
- case IX86_BUILTIN_PSLLQI128:
- icode = CODE_FOR_ashlv2di3;
- goto do_pshifti;
- case IX86_BUILTIN_PSRAWI128:
- icode = CODE_FOR_ashrv8hi3;
- goto do_pshifti;
- case IX86_BUILTIN_PSRADI128:
- icode = CODE_FOR_ashrv4si3;
- goto do_pshifti;
- case IX86_BUILTIN_PSRLWI128:
- icode = CODE_FOR_lshrv8hi3;
- goto do_pshifti;
- case IX86_BUILTIN_PSRLDI128:
- icode = CODE_FOR_lshrv4si3;
- goto do_pshifti;
- case IX86_BUILTIN_PSRLQI128:
- icode = CODE_FOR_lshrv2di3;
- goto do_pshifti;
- do_pshifti:
- arg0 = CALL_EXPR_ARG (exp, 0);
- arg1 = CALL_EXPR_ARG (exp, 1);
- op0 = expand_normal (arg0);
- op1 = expand_normal (arg1);
-
- if (!CONST_INT_P (op1))
- {
- error ("shift must be an immediate");
- return const0_rtx;
- }
- if (INTVAL (op1) < 0 || INTVAL (op1) > 255)
- op1 = GEN_INT (255);
-
- tmode = insn_data[icode].operand[0].mode;
- mode1 = insn_data[icode].operand[1].mode;
- if (! (*insn_data[icode].operand[1].predicate) (op0, mode1))
- op0 = copy_to_reg (op0);
-
- target = gen_reg_rtx (tmode);
- pat = GEN_FCN (icode) (target, op0, op1);
- if (!pat)
- return 0;
- emit_insn (pat);
- return target;
+ case IX86_BUILTIN_PSLLW:
+ case IX86_BUILTIN_PSLLWI:
+ icode = CODE_FOR_mmx_ashlv4hi3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSLLD:
+ case IX86_BUILTIN_PSLLDI:
+ icode = CODE_FOR_mmx_ashlv2si3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSLLQ:
+ case IX86_BUILTIN_PSLLQI:
+ icode = CODE_FOR_mmx_ashlv1di3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSRAW:
+ case IX86_BUILTIN_PSRAWI:
+ icode = CODE_FOR_mmx_ashrv4hi3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSRAD:
+ case IX86_BUILTIN_PSRADI:
+ icode = CODE_FOR_mmx_ashrv2si3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSRLW:
+ case IX86_BUILTIN_PSRLWI:
+ icode = CODE_FOR_mmx_lshrv4hi3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSRLD:
+ case IX86_BUILTIN_PSRLDI:
+ icode = CODE_FOR_mmx_lshrv2si3;
+ goto do_pshift;
+ case IX86_BUILTIN_PSRLQ:
+ case IX86_BUILTIN_PSRLQI:
+ icode = CODE_FOR_mmx_lshrv1di3;
+ goto do_pshift;
case IX86_BUILTIN_PSLLW128:
+ case IX86_BUILTIN_PSLLWI128:
icode = CODE_FOR_ashlv8hi3;
goto do_pshift;
case IX86_BUILTIN_PSLLD128:
+ case IX86_BUILTIN_PSLLDI128:
icode = CODE_FOR_ashlv4si3;
goto do_pshift;
case IX86_BUILTIN_PSLLQ128:
+ case IX86_BUILTIN_PSLLQI128:
icode = CODE_FOR_ashlv2di3;
goto do_pshift;
case IX86_BUILTIN_PSRAW128:
+ case IX86_BUILTIN_PSRAWI128:
icode = CODE_FOR_ashrv8hi3;
goto do_pshift;
case IX86_BUILTIN_PSRAD128:
+ case IX86_BUILTIN_PSRADI128:
icode = CODE_FOR_ashrv4si3;
goto do_pshift;
case IX86_BUILTIN_PSRLW128:
+ case IX86_BUILTIN_PSRLWI128:
icode = CODE_FOR_lshrv8hi3;
goto do_pshift;
case IX86_BUILTIN_PSRLD128:
+ case IX86_BUILTIN_PSRLDI128:
icode = CODE_FOR_lshrv4si3;
goto do_pshift;
case IX86_BUILTIN_PSRLQ128:
+ case IX86_BUILTIN_PSRLQI128:
icode = CODE_FOR_lshrv2di3;
- goto do_pshift;
+
do_pshift:
arg0 = CALL_EXPR_ARG (exp, 0);
arg1 = CALL_EXPR_ARG (exp, 1);
if (! (*insn_data[icode].operand[1].predicate) (op0, mode1))
op0 = copy_to_reg (op0);
- op1 = simplify_gen_subreg (TImode, op1, GET_MODE (op1), 0);
- if (! (*insn_data[icode].operand[2].predicate) (op1, TImode))
+ if (!CONST_INT_P (op1))
+ op1 = simplify_gen_subreg (SImode, op1, GET_MODE (op1), 0);
+
+ if (! (*insn_data[icode].operand[2].predicate) (op1, SImode))
op1 = copy_to_reg (op1);
target = gen_reg_rtx (tmode);
if (d->code == fcode)
return ix86_expand_sse_pcmpistr (d, exp, target);
+ for (i = 0, d = bdesc_multi_arg; i < ARRAY_SIZE (bdesc_multi_arg); i++, d++)
+ if (d->code == fcode)
+ return ix86_expand_multi_arg_builtin (d->icode, exp, target,
+ (enum multi_arg_type)d->flag,
+ d->comparison);
+
gcc_unreachable ();
}
if (out_mode == DFmode && out_n == 2
&& in_mode == DFmode && in_n == 2)
return ix86_builtins[IX86_BUILTIN_SQRTPD];
- return NULL_TREE;
+ break;
case BUILT_IN_SQRTF:
if (out_mode == SFmode && out_n == 4
&& in_mode == SFmode && in_n == 4)
- return ix86_builtins[IX86_BUILTIN_SQRTPS];
- return NULL_TREE;
+ return ix86_builtins[IX86_BUILTIN_SQRTPS_NR];
+ break;
case BUILT_IN_LRINT:
if (out_mode == SImode && out_n == 4
&& in_mode == DFmode && in_n == 2)
return ix86_builtins[IX86_BUILTIN_VEC_PACK_SFIX];
- return NULL_TREE;
+ break;
case BUILT_IN_LRINTF:
if (out_mode == SImode && out_n == 4
&& in_mode == SFmode && in_n == 4)
return ix86_builtins[IX86_BUILTIN_CVTPS2DQ];
- return NULL_TREE;
+ break;
default:
;
}
+ /* Dispatch to a handler for a vectorization library. */
+ if (ix86_veclib_handler)
+ return (*ix86_veclib_handler)(fn, type_out, type_in);
+
return NULL_TREE;
}
+/* Handler for an ACML-style interface to a library with vectorized
+ intrinsics. */
+
+static tree
+ix86_veclibabi_acml (enum built_in_function fn, tree type_out, tree type_in)
+{
+ char name[20] = "__vr.._";
+ tree fntype, new_fndecl, args;
+ unsigned arity;
+ const char *bname;
+ enum machine_mode el_mode, in_mode;
+ int n, in_n;
+
+ /* The ACML is 64bits only and suitable for unsafe math only as
+ it does not correctly support parts of IEEE with the required
+ precision such as denormals. */
+ if (!TARGET_64BIT
+ || !flag_unsafe_math_optimizations)
+ return NULL_TREE;
+
+ el_mode = TYPE_MODE (TREE_TYPE (type_out));
+ n = TYPE_VECTOR_SUBPARTS (type_out);
+ in_mode = TYPE_MODE (TREE_TYPE (type_in));
+ in_n = TYPE_VECTOR_SUBPARTS (type_in);
+ if (el_mode != in_mode
+ || n != in_n)
+ return NULL_TREE;
+
+ switch (fn)
+ {
+ case BUILT_IN_SIN:
+ case BUILT_IN_COS:
+ case BUILT_IN_EXP:
+ case BUILT_IN_LOG:
+ case BUILT_IN_LOG2:
+ case BUILT_IN_LOG10:
+ name[4] = 'd';
+ name[5] = '2';
+ if (el_mode != DFmode
+ || n != 2)
+ return NULL_TREE;
+ break;
+
+ case BUILT_IN_SINF:
+ case BUILT_IN_COSF:
+ case BUILT_IN_EXPF:
+ case BUILT_IN_POWF:
+ case BUILT_IN_LOGF:
+ case BUILT_IN_LOG2F:
+ case BUILT_IN_LOG10F:
+ name[4] = 's';
+ name[5] = '4';
+ if (el_mode != SFmode
+ || n != 4)
+ return NULL_TREE;
+ break;
+
+ default:
+ return NULL_TREE;
+ }
+
+ bname = IDENTIFIER_POINTER (DECL_NAME (implicit_built_in_decls[fn]));
+ sprintf (name + 7, "%s", bname+10);
+
+ arity = 0;
+ for (args = DECL_ARGUMENTS (implicit_built_in_decls[fn]); args;
+ args = TREE_CHAIN (args))
+ arity++;
+
+ if (arity == 1)
+ fntype = build_function_type_list (type_out, type_in, NULL);
+ else
+ fntype = build_function_type_list (type_out, type_in, type_in, NULL);
+
+ /* Build a function declaration for the vectorized function. */
+ new_fndecl = build_decl (FUNCTION_DECL, get_identifier (name), fntype);
+ TREE_PUBLIC (new_fndecl) = 1;
+ DECL_EXTERNAL (new_fndecl) = 1;
+ DECL_IS_NOVOPS (new_fndecl) = 1;
+ TREE_READONLY (new_fndecl) = 1;
+
+ return new_fndecl;
+}
+
+
/* Returns a decl of a function that implements conversion of the
input vector of type TYPE, or NULL_TREE if it is not available. */
switch (fn)
{
/* Vectorized version of sqrt to rsqrt conversion. */
- case IX86_BUILTIN_SQRTPS:
- return ix86_builtins[IX86_BUILTIN_RSQRTPS];
+ case IX86_BUILTIN_SQRTPS_NR:
+ return ix86_builtins[IX86_BUILTIN_RSQRTPS_NR];
default:
return NULL_TREE;
where integer modes in MMX/SSE registers are not tieable
because of missing QImode and HImode moves to, from or between
MMX/SSE registers. */
- return MAX (ix86_cost->mmxsse_to_integer, 8);
+ return MAX (8, ix86_cost->mmxsse_to_integer);
if (MAYBE_FLOAT_CLASS_P (class1))
return ix86_cost->fp_move;
return 1;
else if (VALID_FP_MODE_P (mode))
return 1;
+ else if (VALID_DFP_MODE_P (mode))
+ return 1;
/* Lots of MMX code casts 8 byte vector modes to DImode. If we then go
on to use that value in smaller contexts, this can easily force a
pseudo to be allocated to GENERAL_REGS. Since this is no worse than
nbits = 7;
/* Compute costs correctly for widening multiplication. */
- if ((GET_CODE (op0) == SIGN_EXTEND || GET_CODE (op1) == ZERO_EXTEND)
+ if ((GET_CODE (op0) == SIGN_EXTEND || GET_CODE (op0) == ZERO_EXTEND)
&& GET_MODE_SIZE (GET_MODE (XEXP (op0, 0))) * 2
== GET_MODE_SIZE (mode))
{
}
static bool
-ix86_ms_bitfield_layout_p (tree record_type)
+ix86_ms_bitfield_layout_p (const_tree record_type)
{
return (TARGET_MS_BITFIELD_LAYOUT &&
!lookup_attribute ("gcc_struct", TYPE_ATTRIBUTES (record_type)))
{
tree type = TREE_TYPE (function);
bool aggr = aggregate_value_p (TREE_TYPE (type), type) != 0;
+ int nregs;
if (TARGET_64BIT)
{
return gen_rtx_REG (DImode, parm_regs[aggr]);
}
- if (ix86_function_regparm (type, function) > 0
- && !type_has_variadic_args_p (type))
+ nregs = ix86_function_regparm (type, function);
+
+ if (nregs > 0 && !stdarg_p (type))
{
- int regno = 0;
+ int regno;
+
if (lookup_attribute ("fastcall", TYPE_ATTRIBUTES (type)))
- regno = 2;
+ regno = aggr ? DX_REG : CX_REG;
+ else
+ {
+ regno = AX_REG;
+ if (aggr)
+ {
+ regno = DX_REG;
+ if (nregs == 1)
+ return gen_rtx_MEM (SImode,
+ plus_constant (stack_pointer_rtx, 4));
+ }
+ }
return gen_rtx_REG (SImode, regno);
}
/* Determine whether x86_output_mi_thunk can succeed. */
static bool
-x86_can_output_mi_thunk (tree thunk ATTRIBUTE_UNUSED,
+x86_can_output_mi_thunk (const_tree thunk ATTRIBUTE_UNUSED,
HOST_WIDE_INT delta ATTRIBUTE_UNUSED,
- HOST_WIDE_INT vcall_offset, tree function)
+ HOST_WIDE_INT vcall_offset, const_tree function)
{
/* 64-bit can handle anything. */
if (TARGET_64BIT)
{
/* Put the this parameter into %eax. */
xops[0] = this_param;
- xops[1] = this_reg = gen_rtx_REG (Pmode, 0);
+ xops[1] = this_reg = gen_rtx_REG (Pmode, AX_REG);
output_asm_insn ("mov{l}\t{%0, %1|%1, %0}", xops);
}
else
tmp = gen_rtx_REG (DImode, R10_REG);
else
{
- int tmp_regno = 2 /* ECX */;
+ int tmp_regno = CX_REG;
if (lookup_attribute ("fastcall",
TYPE_ATTRIBUTES (TREE_TYPE (function))))
- tmp_regno = 0 /* EAX */;
+ tmp_regno = AX_REG;
tmp = gen_rtx_REG (SImode, tmp_regno);
}
else
#endif /* TARGET_MACHO */
{
- tmp = gen_rtx_REG (SImode, 2 /* ECX */);
+ tmp = gen_rtx_REG (SImode, CX_REG);
output_set_got (tmp, NULL_RTX);
xops[1] = tmp;
if (X86_FILE_START_FLTUSED)
fputs ("\t.global\t__fltused\n", asm_out_file);
if (ix86_asm_dialect == ASM_INTEL)
- fputs ("\t.intel_syntax\n", asm_out_file);
+ fputs ("\t.intel_syntax noprefix\n", asm_out_file);
}
int
for (i = 0; i < n_elts; ++i)
{
x = XVECEXP (vals, 0, i);
- if (!CONSTANT_P (x))
+ if (!(CONST_INT_P (x)
+ || GET_CODE (x) == CONST_DOUBLE
+ || GET_CODE (x) == CONST_FIXED))
n_var++, one_var = i;
else if (x != CONST0_RTX (inner_mode))
all_const_zero = false;
/* a / b = a * rcp(b) * (2.0 - b * rcp(b)) */
- /* x0 = 1./b estimate */
+ /* x0 = rcp(b) estimate */
emit_insn (gen_rtx_SET (VOIDmode, x0,
gen_rtx_UNSPEC (mode, gen_rtvec (1, b),
UNSPEC_RCP)));
void ix86_emit_swsqrtsf (rtx res, rtx a, enum machine_mode mode,
bool recip)
{
- rtx x0, e0, e1, e2, e3, three, half, zero, mask;
+ rtx x0, e0, e1, e2, e3, mthree, mhalf;
+ REAL_VALUE_TYPE r;
x0 = gen_reg_rtx (mode);
e0 = gen_reg_rtx (mode);
e2 = gen_reg_rtx (mode);
e3 = gen_reg_rtx (mode);
- three = CONST_DOUBLE_FROM_REAL_VALUE (dconst3, SFmode);
- half = CONST_DOUBLE_FROM_REAL_VALUE (dconsthalf, SFmode);
+ real_arithmetic (&r, NEGATE_EXPR, &dconst3, NULL);
+ mthree = CONST_DOUBLE_FROM_REAL_VALUE (r, SFmode);
- mask = gen_reg_rtx (mode);
+ real_arithmetic (&r, NEGATE_EXPR, &dconsthalf, NULL);
+ mhalf = CONST_DOUBLE_FROM_REAL_VALUE (r, SFmode);
if (VECTOR_MODE_P (mode))
{
- three = ix86_build_const_vector (SFmode, true, three);
- half = ix86_build_const_vector (SFmode, true, half);
+ mthree = ix86_build_const_vector (SFmode, true, mthree);
+ mhalf = ix86_build_const_vector (SFmode, true, mhalf);
}
- three = force_reg (mode, three);
- half = force_reg (mode, half);
-
- zero = force_reg (mode, CONST0_RTX(mode));
+ /* sqrt(a) = -0.5 * a * rsqrtss(a) * (a * rsqrtss(a) * rsqrtss(a) - 3.0)
+ rsqrt(a) = -0.5 * rsqrtss(a) * (a * rsqrtss(a) * rsqrtss(a) - 3.0) */
- /* sqrt(a) = 0.5 * a * rsqrtss(a) * (3.0 - a * rsqrtss(a) * rsqrtss(a))
- 1.0 / sqrt(a) = 0.5 * rsqrtss(a) * (3.0 - a * rsqrtss(a) * rsqrtss(a)) */
-
- /* Compare a to zero. */
- emit_insn (gen_rtx_SET (VOIDmode, mask,
- gen_rtx_NE (mode, a, zero)));
-
- /* x0 = 1./sqrt(a) estimate */
+ /* x0 = rsqrt(a) estimate */
emit_insn (gen_rtx_SET (VOIDmode, x0,
gen_rtx_UNSPEC (mode, gen_rtvec (1, a),
UNSPEC_RSQRT)));
- /* Filter out infinity. */
- if (VECTOR_MODE_P (mode))
- emit_insn (gen_rtx_SET (VOIDmode, gen_lowpart (V4SFmode, x0),
- gen_rtx_AND (mode,
- gen_lowpart (V4SFmode, x0),
- gen_lowpart (V4SFmode, mask))));
- else
- emit_insn (gen_rtx_SET (VOIDmode, x0,
- gen_rtx_AND (mode, x0, mask)));
+
+ /* If (a == 0.0) Filter out infinity to prevent NaN for sqrt(0.0). */
+ if (!recip)
+ {
+ rtx zero, mask;
+
+ zero = gen_reg_rtx (mode);
+ mask = gen_reg_rtx (mode);
+
+ zero = force_reg (mode, CONST0_RTX(mode));
+ emit_insn (gen_rtx_SET (VOIDmode, mask,
+ gen_rtx_NE (mode, zero, a)));
+
+ emit_insn (gen_rtx_SET (VOIDmode, x0,
+ gen_rtx_AND (mode, x0, mask)));
+ }
/* e0 = x0 * a */
emit_insn (gen_rtx_SET (VOIDmode, e0,
/* e1 = e0 * x0 */
emit_insn (gen_rtx_SET (VOIDmode, e1,
gen_rtx_MULT (mode, e0, x0)));
- /* e2 = 3. - e1 */
+
+ /* e2 = e1 - 3. */
+ mthree = force_reg (mode, mthree);
emit_insn (gen_rtx_SET (VOIDmode, e2,
- gen_rtx_MINUS (mode, three, e1)));
+ gen_rtx_PLUS (mode, e1, mthree)));
+
+ mhalf = force_reg (mode, mhalf);
if (recip)
- /* e3 = .5 * x0 */
+ /* e3 = -.5 * x0 */
emit_insn (gen_rtx_SET (VOIDmode, e3,
- gen_rtx_MULT (mode, half, x0)));
+ gen_rtx_MULT (mode, x0, mhalf)));
else
- /* e3 = .5 * e0 */
+ /* e3 = -.5 * e0 */
emit_insn (gen_rtx_SET (VOIDmode, e3,
- gen_rtx_MULT (mode, half, e0)));
+ gen_rtx_MULT (mode, e0, mhalf)));
/* ret = e2 * e3 */
emit_insn (gen_rtx_SET (VOIDmode, res,
gen_rtx_MULT (mode, e2, e3)));
/* Return the mangling of TYPE if it is an extended fundamental type. */
static const char *
-ix86_mangle_type (tree type)
+ix86_mangle_type (const_tree type)
{
type = TYPE_MAIN_VARIANT (type);
/* load nextafter (0.5, 0.0) */
fmt = REAL_MODE_FORMAT (mode);
- real_2expN (&half_minus_pred_half, -(fmt->p) - 1);
+ real_2expN (&half_minus_pred_half, -(fmt->p) - 1, mode);
REAL_ARITHMETIC (pred_half, MINUS_EXPR, dconsthalf, half_minus_pred_half);
/* adj = copysign (0.5, op1) */
/* load nextafter (0.5, 0.0) */
fmt = REAL_MODE_FORMAT (mode);
- real_2expN (&half_minus_pred_half, -(fmt->p) - 1);
+ real_2expN (&half_minus_pred_half, -(fmt->p) - 1, mode);
REAL_ARITHMETIC (pred_half, MINUS_EXPR, dconsthalf, half_minus_pred_half);
/* xa = xa + 0.5 */
}
\f
+/* Validate whether a SSE5 instruction is valid or not.
+ OPERANDS is the array of operands.
+ NUM is the number of operands.
+ USES_OC0 is true if the instruction uses OC0 and provides 4 variants.
+ NUM_MEMORY is the maximum number of memory operands to accept. */
+bool
+ix86_sse5_valid_op_p (rtx operands[], rtx insn, int num, bool uses_oc0, int num_memory)
+{
+ int mem_mask;
+ int mem_count;
+ int i;
+
+ /* Count the number of memory arguments */
+ mem_mask = 0;
+ mem_count = 0;
+ for (i = 0; i < num; i++)
+ {
+ enum machine_mode mode = GET_MODE (operands[i]);
+ if (register_operand (operands[i], mode))
+ ;
+
+ else if (memory_operand (operands[i], mode))
+ {
+ mem_mask |= (1 << i);
+ mem_count++;
+ }
+
+ else
+ {
+ rtx pattern = PATTERN (insn);
+
+ /* allow 0 for pcmov */
+ if (GET_CODE (pattern) != SET
+ || GET_CODE (SET_SRC (pattern)) != IF_THEN_ELSE
+ || i < 2
+ || operands[i] != CONST0_RTX (mode))
+ return false;
+ }
+ }
+
+ /* If there were no memory operations, allow the insn */
+ if (mem_mask == 0)
+ return true;
+
+ /* Do not allow the destination register to be a memory operand. */
+ else if (mem_mask & (1 << 0))
+ return false;
+
+ /* If there are too many memory operations, disallow the instruction. While
+ the hardware only allows 1 memory reference, before register allocation
+ for some insns, we allow two memory operations sometimes in order to allow
+ code like the following to be optimized:
+
+ float fmadd (float *a, float *b, float *c) { return (*a * *b) + *c; }
+
+ or similar cases that are vectorized into using the fmaddss
+ instruction. */
+ else if (mem_count > num_memory)
+ return false;
+
+ /* Don't allow more than one memory operation if not optimizing. */
+ else if (mem_count > 1 && !optimize)
+ return false;
+
+ else if (num == 4 && mem_count == 1)
+ {
+ /* formats (destination is the first argument), example fmaddss:
+ xmm1, xmm1, xmm2, xmm3/mem
+ xmm1, xmm1, xmm2/mem, xmm3
+ xmm1, xmm2, xmm3/mem, xmm1
+ xmm1, xmm2/mem, xmm3, xmm1 */
+ if (uses_oc0)
+ return ((mem_mask == (1 << 1))
+ || (mem_mask == (1 << 2))
+ || (mem_mask == (1 << 3)));
+
+ /* format, example pmacsdd:
+ xmm1, xmm2, xmm3/mem, xmm1 */
+ else
+ return (mem_mask == (1 << 2));
+ }
+
+ else if (num == 4 && num_memory == 2)
+ {
+ /* If there are two memory operations, we can load one of the memory ops
+ into the destination register. This is for optimizing the
+ multiply/add ops, which the combiner has optimized both the multiply
+ and the add insns to have a memory operation. We have to be careful
+ that the destination doesn't overlap with the inputs. */
+ rtx op0 = operands[0];
+
+ if (reg_mentioned_p (op0, operands[1])
+ || reg_mentioned_p (op0, operands[2])
+ || reg_mentioned_p (op0, operands[3]))
+ return false;
+
+ /* formats (destination is the first argument), example fmaddss:
+ xmm1, xmm1, xmm2, xmm3/mem
+ xmm1, xmm1, xmm2/mem, xmm3
+ xmm1, xmm2, xmm3/mem, xmm1
+ xmm1, xmm2/mem, xmm3, xmm1
+
+ For the oc0 case, we will load either operands[1] or operands[3] into
+ operands[0], so any combination of 2 memory operands is ok. */
+ if (uses_oc0)
+ return true;
+
+ /* format, example pmacsdd:
+ xmm1, xmm2, xmm3/mem, xmm1
+
+ For the integer multiply/add instructions be more restrictive and
+ require operands[2] and operands[3] to be the memory operands. */
+ else
+ return (mem_mask == ((1 << 2) | (1 << 3)));
+ }
+
+ else if (num == 3 && num_memory == 1)
+ {
+ /* formats, example protb:
+ xmm1, xmm2, xmm3/mem
+ xmm1, xmm2/mem, xmm3 */
+ if (uses_oc0)
+ return ((mem_mask == (1 << 1)) || (mem_mask == (1 << 2)));
+
+ /* format, example comeq:
+ xmm1, xmm2, xmm3/mem */
+ else
+ return (mem_mask == (1 << 2));
+ }
+
+ else
+ gcc_unreachable ();
+
+ return false;
+}
+
+\f
+/* Fixup an SSE5 instruction that has 2 memory input references into a form the
+ hardware will allow by using the destination register to load one of the
+ memory operations. Presently this is used by the multiply/add routines to
+ allow 2 memory references. */
+
+void
+ix86_expand_sse5_multiple_memory (rtx operands[],
+ int num,
+ enum machine_mode mode)
+{
+ rtx op0 = operands[0];
+ if (num != 4
+ || memory_operand (op0, mode)
+ || reg_mentioned_p (op0, operands[1])
+ || reg_mentioned_p (op0, operands[2])
+ || reg_mentioned_p (op0, operands[3]))
+ gcc_unreachable ();
+
+ /* For 2 memory operands, pick either operands[1] or operands[3] to move into
+ the destination register. */
+ if (memory_operand (operands[1], mode))
+ {
+ emit_move_insn (op0, operands[1]);
+ operands[1] = op0;
+ }
+ else if (memory_operand (operands[3], mode))
+ {
+ emit_move_insn (op0, operands[3]);
+ operands[3] = op0;
+ }
+ else
+ gcc_unreachable ();
+
+ return;
+}
+
+\f
/* Table of valid machine attributes. */
static const struct attribute_spec ix86_attribute_table[] =
{
{ NULL, 0, 0, false, false, false, NULL }
};
+/* Implement targetm.vectorize.builtin_vectorization_cost. */
+static int
+x86_builtin_vectorization_cost (bool runtime_test)
+{
+ /* If the branch of the runtime test is taken - i.e. - the vectorized
+ version is skipped - this incurs a misprediction cost (because the
+ vectorized version is expected to be the fall-through). So we subtract
+ the latency of a mispredicted branch from the costs that are incured
+ when the vectorized version is executed.
+
+ TODO: The values in individual target tables have to be tuned or new
+ fields may be needed. For eg. on K8, the default branch path is the
+ not-taken path. If the taken path is predicted correctly, the minimum
+ penalty of going down the taken-path is 1 cycle. If the taken-path is
+ not predicted correctly, then the minimum penalty is 10 cycles. */
+
+ if (runtime_test)
+ {
+ return (-(ix86_cost->cond_taken_branch_cost));
+ }
+ else
+ return 0;
+}
+
/* Initialize the GCC target structure. */
#undef TARGET_ATTRIBUTE_TABLE
#define TARGET_ATTRIBUTE_TABLE ix86_attribute_table
#undef TARGET_CANNOT_FORCE_CONST_MEM
#define TARGET_CANNOT_FORCE_CONST_MEM ix86_cannot_force_const_mem
#undef TARGET_USE_BLOCKS_FOR_CONSTANT_P
-#define TARGET_USE_BLOCKS_FOR_CONSTANT_P hook_bool_mode_rtx_true
+#define TARGET_USE_BLOCKS_FOR_CONSTANT_P hook_bool_mode_const_rtx_true
#undef TARGET_DELEGITIMIZE_ADDRESS
#define TARGET_DELEGITIMIZE_ADDRESS ix86_delegitimize_address
#undef TARGET_BUILD_BUILTIN_VA_LIST
#define TARGET_BUILD_BUILTIN_VA_LIST ix86_build_builtin_va_list
+#undef TARGET_EXPAND_BUILTIN_VA_START
+#define TARGET_EXPAND_BUILTIN_VA_START ix86_va_start
+
#undef TARGET_MD_ASM_CLOBBERS
#define TARGET_MD_ASM_CLOBBERS ix86_md_asm_clobbers
#undef TARGET_PROMOTE_PROTOTYPES
-#define TARGET_PROMOTE_PROTOTYPES hook_bool_tree_true
+#define TARGET_PROMOTE_PROTOTYPES hook_bool_const_tree_true
#undef TARGET_STRUCT_VALUE_RTX
#define TARGET_STRUCT_VALUE_RTX ix86_struct_value_rtx
#undef TARGET_SETUP_INCOMING_VARARGS
#undef TARGET_FUNCTION_VALUE
#define TARGET_FUNCTION_VALUE ix86_function_value
+#undef TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST
+#define TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST x86_builtin_vectorization_cost
+
struct gcc_target targetm = TARGET_INITIALIZER;
\f
#include "gt-i386.h"
OSDN Git Service
