1
Fork 0
mirror of https://git.savannah.gnu.org/git/guile.git synced 2025-06-28 05:50:30 +02:00
guile/lightening/aarch64.c
Andy Wingo 434fe2b4aa Merge branch 'callr-fix3' into 'main'
Fix some problems with callr and calli.

See merge request wingo/lightening!19
2024-06-03 13:43:19 +00:00

269 lines
8.2 KiB
C

/*
* Copyright (C) 2013-2020, 2024 Free Software Foundation, Inc.
*
* This file is part of GNU lightning.
*
* GNU lightning is free software; you can redistribute it and/or modify it
* under the terms of the GNU Lesser General Public License as published
* by the Free Software Foundation; either version 3, or (at your option)
* any later version.
*
* GNU lightning is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
* or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
* License for more details.
*
* Authors:
* Paulo Cesar Pereira de Andrade
*/
/* libgcc */
extern void __clear_cache(void *, void *);
static int has_lse_atomics;
static inline int32_t
read_signed_bitfield(uint32_t word, uint8_t width, uint8_t shift)
{
return ((int32_t)word) << (32 - width - shift) >> (32 - width);
}
static inline uint32_t
read_unsigned_bitfield(uint32_t word, uint8_t width, uint8_t shift)
{
return word << (32 - width - shift) >> (32 - width);
}
static inline int
in_signed_range(ptrdiff_t diff, uint8_t bits)
{
return (-1 << (bits - 1)) <= diff && diff < (1 << (bits - 1));
}
static inline int
in_unsigned_range(uint32_t val, uint8_t bits)
{
ASSERT(bits < __WORDSIZE);
return val < (1 << bits);
}
static inline uint32_t
write_unsigned_bitfield(uint32_t word, uint32_t val, uint8_t width, uint8_t shift)
{
ASSERT(read_unsigned_bitfield(word, width, shift) == 0);
ASSERT(in_unsigned_range(val, width));
return word | (val << shift);
}
static inline int32_t
write_signed_bitfield(uint32_t word, ptrdiff_t val, uint8_t width, uint8_t shift)
{
ASSERT(read_signed_bitfield(word, width, shift) == 0);
ASSERT(in_signed_range(val, width));
return word | ((val & ((1 << width) - 1)) << shift);
}
#define DEFINE_ENCODER(name, width, shift, kind, val_t) \
static const uint8_t name##_width = width; \
static const uint8_t name##_shift = shift; \
static uint32_t \
write_##name##_bitfield(uint32_t word, val_t val) \
{ \
return write_##kind##_bitfield(word, val, name##_width, name##_shift); \
}
DEFINE_ENCODER(Rd, 5, 0, unsigned, uint32_t)
DEFINE_ENCODER(Rm, 5, 16, unsigned, uint32_t)
DEFINE_ENCODER(Rn, 5, 5, unsigned, uint32_t)
DEFINE_ENCODER(cond2, 4, 0, unsigned, uint32_t)
DEFINE_ENCODER(simm9, 9, 12, signed, ptrdiff_t)
DEFINE_ENCODER(imm12, 12, 10, unsigned, uint32_t)
DEFINE_ENCODER(imm16, 16, 5, unsigned, uint32_t)
DEFINE_ENCODER(simm19, 19, 5, signed, ptrdiff_t)
DEFINE_ENCODER(simm26, 26, 0, signed, ptrdiff_t)
DEFINE_ENCODER(immr, 6, 16, unsigned, uint32_t)
DEFINE_ENCODER(imms, 6, 10, unsigned, uint32_t)
DEFINE_ENCODER(size, 2, 22, unsigned, uint32_t)
#define DEFINE_PATCHABLE_INSTRUCTION(name, kind, RELOC, rsh) \
static int32_t \
read_##name##_offset(uint32_t *loc) \
{ \
return read_signed_bitfield(*loc, kind##_width, kind##_shift); \
} \
static int offset_in_##name##_range(ptrdiff_t diff, int flags) maybe_unused; \
static int \
offset_in_##name##_range(ptrdiff_t diff, int flags) \
{ \
return in_signed_range(diff, kind##_width); \
} \
static void \
patch_##name##_offset(uint32_t *loc, ptrdiff_t diff) \
{ \
*loc = write_##kind##_bitfield(*loc, diff); \
} \
static jit_reloc_t \
emit_##name(jit_state_t *_jit, uint32_t inst) \
{ \
while (1) { \
jit_reloc_t ret = jit_reloc (_jit, JIT_RELOC_##RELOC, 0, \
_jit->pc.uc, _jit->pc.uc, rsh); \
if (add_pending_literal(_jit, ret, kind##_width - 1)) { \
emit_u32(_jit, inst); \
return ret; \
} \
} \
}
#define DEFINE_PATCHABLE_INSTRUCTIONS(name, kind, RELOC, rsh) \
DEFINE_PATCHABLE_INSTRUCTION(name, kind, RELOC, rsh); \
DEFINE_PATCHABLE_INSTRUCTION(veneer_##name, kind, RELOC, rsh);
DEFINE_PATCHABLE_INSTRUCTIONS(jmp, simm26, JMP_WITH_VENEER, 2);
DEFINE_PATCHABLE_INSTRUCTIONS(jcc, simm19, JCC_WITH_VENEER, 2);
DEFINE_PATCHABLE_INSTRUCTION(load_from_pool, simm19, LOAD_FROM_POOL, 2);
struct veneer
{
uint32_t ldr;
uint32_t br;
uint64_t addr;
};
static void
patch_veneer(uint32_t *loc, jit_pointer_t addr)
{
struct veneer *v = (struct veneer*) loc;
v->addr = (uint64_t) addr;
}
#include "aarch64-cpu.c"
#include "aarch64-fpu.c"
static const jit_gpr_t abi_gpr_args[] = {
_X0, _X1, _X2, _X3, _X4, _X5, _X6, _X7
};
static const jit_fpr_t abi_fpr_args[] = {
_D0, _D1, _D2, _D3, _D4, _D5, _D6, _D7
};
static const int abi_gpr_arg_count = sizeof(abi_gpr_args) / sizeof(abi_gpr_args[0]);
static const int abi_fpr_arg_count = sizeof(abi_fpr_args) / sizeof(abi_fpr_args[0]);
struct abi_arg_iterator
{
const jit_operand_t *args;
size_t argc;
size_t arg_idx;
size_t gpr_idx;
size_t fpr_idx;
size_t stack_size;
size_t stack_padding;
};
static size_t page_size;
# define HWCAP_ATOMICS (1 << 8)
#ifdef __gnu_linux__
// See
// https://github.com/gcc-mirror/gcc/blob/master/libgcc/config/aarch64/lse-init.c.
# define AT_HWCAP 16
unsigned long __getauxval (unsigned long int);
static unsigned long get_hwcap(void)
{
return __getauxval (AT_HWCAP);
}
#elif defined(DARWIN)
static unsigned long get_hwcap(void)
{
// All Mac machines have LSE atomics. Most iOS have it too but generally JIT
// isn't allowed there, so assume that it's OK to say we always have LSE.
return HWCAP_ATOMICS;
}
#else
static unsigned long get_hwcap(void)
{
return 0;
}
#endif
jit_bool_t
jit_get_cpu(void)
{
page_size = sysconf(_SC_PAGE_SIZE);
unsigned long hwcap = get_hwcap();
has_lse_atomics = (hwcap & HWCAP_ATOMICS) != 0;
return 1;
}
jit_bool_t
jit_init(jit_state_t *_jit)
{
return 1;
}
static size_t
jit_initial_frame_size (void)
{
return 0;
}
static void
reset_abi_arg_iterator(struct abi_arg_iterator *iter, size_t argc,
const jit_operand_t *args)
{
memset(iter, 0, sizeof *iter);
iter->argc = argc;
iter->args = args;
}
static void
next_abi_arg(struct abi_arg_iterator *iter, jit_operand_t *arg)
{
ASSERT(iter->arg_idx < iter->argc);
enum jit_operand_abi abi = iter->args[iter->arg_idx].abi;
if (is_gpr_arg(abi) && iter->gpr_idx < abi_gpr_arg_count) {
*arg = jit_operand_gpr (abi, abi_gpr_args[iter->gpr_idx++]);
} else if (is_fpr_arg(abi) && iter->fpr_idx < abi_fpr_arg_count) {
*arg = jit_operand_fpr (abi, abi_fpr_args[iter->fpr_idx++]);
} else {
*arg = jit_operand_mem (abi, JIT_SP, iter->stack_size);
iter->stack_size += 8;
}
iter->arg_idx++;
}
static void
jit_flush(void *fptr, void *tptr)
{
jit_word_t f = (jit_word_t)fptr & -page_size;
jit_word_t t = (((jit_word_t)tptr) + page_size - 1) & -page_size;
__clear_cache((void *)f, (void *)t);
}
static inline size_t
jit_stack_alignment(void)
{
return 16;
}
static void
jit_try_shorten(jit_state_t *_jit, jit_reloc_t reloc, jit_pointer_t addr)
{
}
static void*
bless_function_pointer(void *ptr)
{
return ptr;
}
static jit_gpr_t
get_callr_temp (jit_state_t * _jit)
{
return _LR;
}