1
0
Fork 0
mirror of https://github.com/ruby/ruby.git synced 2022-11-09 12:17:21 -05:00
ruby--ruby/vm_insnhelper.h
Jeremy Evans d2c41b1bff Reduce allocations for keyword argument hashes
Previously, passing a keyword splat to a method always allocated
a hash on the caller side, and accepting arbitrary keywords in
a method allocated a separate hash on the callee side.  Passing
explicit keywords to a method that accepted a keyword splat
did not allocate a hash on the caller side, but resulted in two
hashes allocated on the callee side.

This commit makes passing a single keyword splat to a method not
allocate a hash on the caller side.  Passing multiple keyword
splats or a mix of explicit keywords and a keyword splat still
generates a hash on the caller side.  On the callee side,
if arbitrary keywords are not accepted, it does not allocate a
hash.  If arbitrary keywords are accepted, it will allocate a
hash, but this commit uses a callinfo flag to indicate whether
the caller already allocated a hash, and if so, the callee can
use the passed hash without duplicating it.  So this commit
should make it so that a maximum of a single hash is allocated
during method calls.

To set the callinfo flag appropriately, method call argument
compilation checks if only a single keyword splat is given.
If only one keyword splat is given, the VM_CALL_KW_SPLAT_MUT
callinfo flag is not set, since in that case the keyword
splat is passed directly and not mutable.  If more than one
splat is used, a new hash needs to be generated on the caller
side, and in that case the callinfo flag is set, indicating
the keyword splat is mutable by the callee.

In compile_hash, used for both hash and keyword argument
compilation, if compiling keyword arguments and only a
single keyword splat is used, pass the argument directly.

On the caller side, in vm_args.c, the callinfo flag needs to
be recognized and handled.  Because the keyword splat
argument may not be a hash, it needs to be converted to a
hash first if not.  Then, unless the callinfo flag is set,
the hash needs to be duplicated.  The temporary copy of the
callinfo flag, kw_flag, is updated if a hash was duplicated,
to prevent the need to duplicate it again.  If we are
converting to a hash or duplicating a hash, we need to update
the argument array, which can including duplicating the
positional splat array if one was passed.  CALLER_SETUP_ARG
and a couple other places needs to be modified to handle
similar issues for other types of calls.

This includes fairly comprehensive tests for different ways
keywords are handled internally, checking that you get equal
results but that keyword splats on the caller side result in
distinct objects for keyword rest parameters.

Included are benchmarks for keyword argument calls.
Brief results when compiled without optimization:

  def kw(a: 1) a end
  def kws(**kw) kw end
  h = {a: 1}

  kw(a: 1)       # about same
  kw(**h)        # 2.37x faster
  kws(a: 1)      # 1.30x faster
  kws(**h)       # 2.19x faster
  kw(a: 1, **h)  # 1.03x slower
  kw(**h, **h)   # about same
  kws(a: 1, **h) # 1.16x faster
  kws(**h, **h)  # 1.14x faster
2020-03-17 12:09:43 -07:00

261 lines
8.2 KiB
C

/**********************************************************************
insnhelper.h - helper macros to implement each instructions
$Author$
created at: 04/01/01 15:50:34 JST
Copyright (C) 2004-2007 Koichi Sasada
**********************************************************************/
#ifndef RUBY_INSNHELPER_H
#define RUBY_INSNHELPER_H
RUBY_SYMBOL_EXPORT_BEGIN
RUBY_EXTERN VALUE ruby_vm_const_missing_count;
RUBY_EXTERN rb_serial_t ruby_vm_global_method_state;
RUBY_EXTERN rb_serial_t ruby_vm_global_constant_state;
RUBY_EXTERN rb_serial_t ruby_vm_class_serial;
RUBY_SYMBOL_EXPORT_END
#if VM_COLLECT_USAGE_DETAILS
#define COLLECT_USAGE_INSN(insn) vm_collect_usage_insn(insn)
#define COLLECT_USAGE_OPERAND(insn, n, op) vm_collect_usage_operand((insn), (n), ((VALUE)(op)))
#define COLLECT_USAGE_REGISTER(reg, s) vm_collect_usage_register((reg), (s))
#else
#define COLLECT_USAGE_INSN(insn) /* none */
#define COLLECT_USAGE_OPERAND(insn, n, op) /* none */
#define COLLECT_USAGE_REGISTER(reg, s) /* none */
#endif
/**********************************************************/
/* deal with stack */
/**********************************************************/
#define PUSH(x) (SET_SV(x), INC_SP(1))
#define TOPN(n) (*(GET_SP()-(n)-1))
#define POPN(n) (DEC_SP(n))
#define POP() (DEC_SP(1))
#define STACK_ADDR_FROM_TOP(n) (GET_SP()-(n))
/**********************************************************/
/* deal with registers */
/**********************************************************/
#define VM_REG_CFP (reg_cfp)
#define VM_REG_PC (VM_REG_CFP->pc)
#define VM_REG_SP (VM_REG_CFP->sp)
#define VM_REG_EP (VM_REG_CFP->ep)
#define RESTORE_REGS() do { \
VM_REG_CFP = ec->cfp; \
} while (0)
#if VM_COLLECT_USAGE_DETAILS
enum vm_regan_regtype {
VM_REGAN_PC = 0,
VM_REGAN_SP = 1,
VM_REGAN_EP = 2,
VM_REGAN_CFP = 3,
VM_REGAN_SELF = 4,
VM_REGAN_ISEQ = 5
};
enum vm_regan_acttype {
VM_REGAN_ACT_GET = 0,
VM_REGAN_ACT_SET = 1
};
#define COLLECT_USAGE_REGISTER_HELPER(a, b, v) \
(COLLECT_USAGE_REGISTER((VM_REGAN_##a), (VM_REGAN_ACT_##b)), (v))
#else
#define COLLECT_USAGE_REGISTER_HELPER(a, b, v) (v)
#endif
/* PC */
#define GET_PC() (COLLECT_USAGE_REGISTER_HELPER(PC, GET, VM_REG_PC))
#define SET_PC(x) (VM_REG_PC = (COLLECT_USAGE_REGISTER_HELPER(PC, SET, (x))))
#define GET_CURRENT_INSN() (*GET_PC())
#define GET_OPERAND(n) (GET_PC()[(n)])
#define ADD_PC(n) (SET_PC(VM_REG_PC + (n)))
#define JUMP(dst) (SET_PC(VM_REG_PC + (dst)))
/* frame pointer, environment pointer */
#define GET_CFP() (COLLECT_USAGE_REGISTER_HELPER(CFP, GET, VM_REG_CFP))
#define GET_EP() (COLLECT_USAGE_REGISTER_HELPER(EP, GET, VM_REG_EP))
#define SET_EP(x) (VM_REG_EP = (COLLECT_USAGE_REGISTER_HELPER(EP, SET, (x))))
#define GET_LEP() (VM_EP_LEP(GET_EP()))
/* SP */
#define GET_SP() (COLLECT_USAGE_REGISTER_HELPER(SP, GET, VM_REG_SP))
#define SET_SP(x) (VM_REG_SP = (COLLECT_USAGE_REGISTER_HELPER(SP, SET, (x))))
#define INC_SP(x) (VM_REG_SP += (COLLECT_USAGE_REGISTER_HELPER(SP, SET, (x))))
#define DEC_SP(x) (VM_REG_SP -= (COLLECT_USAGE_REGISTER_HELPER(SP, SET, (x))))
#define SET_SV(x) (*GET_SP() = (x))
/* set current stack value as x */
/* instruction sequence C struct */
#define GET_ISEQ() (GET_CFP()->iseq)
/**********************************************************/
/* deal with variables */
/**********************************************************/
#define GET_PREV_EP(ep) ((VALUE *)((ep)[VM_ENV_DATA_INDEX_SPECVAL] & ~0x03))
/**********************************************************/
/* deal with values */
/**********************************************************/
#define GET_SELF() (COLLECT_USAGE_REGISTER_HELPER(SELF, GET, GET_CFP()->self))
/**********************************************************/
/* deal with control flow 2: method/iterator */
/**********************************************************/
/* set fastpath when cached method is *NOT* protected
* because inline method cache does not care about receiver.
*/
static inline void
CC_SET_FASTPATH(const struct rb_callcache *cc, vm_call_handler func, bool enabled)
{
if (LIKELY(enabled)) {
vm_cc_call_set(cc, func);
}
}
#define GET_BLOCK_HANDLER() (GET_LEP()[VM_ENV_DATA_INDEX_SPECVAL])
/**********************************************************/
/* deal with control flow 3: exception */
/**********************************************************/
/**********************************************************/
/* deal with stack canary */
/**********************************************************/
#if VM_CHECK_MODE > 0
#define SETUP_CANARY() \
VALUE *canary; \
if (leaf) { \
canary = GET_SP(); \
SET_SV(vm_stack_canary); \
} \
else {\
SET_SV(Qfalse); /* cleanup */ \
}
#define CHECK_CANARY() \
if (leaf) { \
if (*canary == vm_stack_canary) { \
*canary = Qfalse; /* cleanup */ \
} \
else { \
vm_canary_is_found_dead(INSN_ATTR(bin), *canary); \
} \
}
#else
#define SETUP_CANARY() /* void */
#define CHECK_CANARY() /* void */
#endif
/**********************************************************/
/* others */
/**********************************************************/
#ifndef MJIT_HEADER
#define CALL_SIMPLE_METHOD() do { \
rb_snum_t x = leaf ? INSN_ATTR(width) : 0; \
rb_snum_t y = attr_width_opt_send_without_block(0); \
rb_snum_t z = x - y; \
ADD_PC(z); \
DISPATCH_ORIGINAL_INSN(opt_send_without_block); \
} while (0)
#endif
#define PREV_CLASS_SERIAL() (ruby_vm_class_serial)
#define NEXT_CLASS_SERIAL() (++ruby_vm_class_serial)
#define GET_GLOBAL_METHOD_STATE() (ruby_vm_global_method_state)
#define INC_GLOBAL_METHOD_STATE() (++ruby_vm_global_method_state)
#define GET_GLOBAL_CONSTANT_STATE() (ruby_vm_global_constant_state)
#define INC_GLOBAL_CONSTANT_STATE() (++ruby_vm_global_constant_state)
static inline struct vm_throw_data *
THROW_DATA_NEW(VALUE val, const rb_control_frame_t *cf, int st)
{
struct vm_throw_data *obj = (struct vm_throw_data *)rb_imemo_new(imemo_throw_data, val, (VALUE)cf, 0, 0);
obj->throw_state = st;
return obj;
}
static inline VALUE
THROW_DATA_VAL(const struct vm_throw_data *obj)
{
VM_ASSERT(THROW_DATA_P(obj));
return obj->throw_obj;
}
static inline const rb_control_frame_t *
THROW_DATA_CATCH_FRAME(const struct vm_throw_data *obj)
{
VM_ASSERT(THROW_DATA_P(obj));
return obj->catch_frame;
}
static inline int
THROW_DATA_STATE(const struct vm_throw_data *obj)
{
VM_ASSERT(THROW_DATA_P(obj));
return obj->throw_state;
}
static inline int
THROW_DATA_CONSUMED_P(const struct vm_throw_data *obj)
{
VM_ASSERT(THROW_DATA_P(obj));
return obj->flags & THROW_DATA_CONSUMED;
}
static inline void
THROW_DATA_CATCH_FRAME_SET(struct vm_throw_data *obj, const rb_control_frame_t *cfp)
{
VM_ASSERT(THROW_DATA_P(obj));
obj->catch_frame = cfp;
}
static inline void
THROW_DATA_STATE_SET(struct vm_throw_data *obj, int st)
{
VM_ASSERT(THROW_DATA_P(obj));
obj->throw_state = st;
}
static inline void
THROW_DATA_CONSUMED_SET(struct vm_throw_data *obj)
{
if (THROW_DATA_P(obj) &&
THROW_DATA_STATE(obj) == TAG_BREAK) {
obj->flags |= THROW_DATA_CONSUMED;
}
}
#define IS_ARGS_SPLAT(ci) (vm_ci_flag(ci) & VM_CALL_ARGS_SPLAT)
#define IS_ARGS_KEYWORD(ci) (vm_ci_flag(ci) & VM_CALL_KWARG)
#define IS_ARGS_KW_SPLAT(ci) (vm_ci_flag(ci) & VM_CALL_KW_SPLAT)
#define IS_ARGS_KW_OR_KW_SPLAT(ci) (vm_ci_flag(ci) & (VM_CALL_KWARG | VM_CALL_KW_SPLAT))
#define IS_ARGS_KW_SPLAT_MUT(ci) (vm_ci_flag(ci) & VM_CALL_KW_SPLAT_MUT)
/* If this returns true, an optimized function returned by `vm_call_iseq_setup_func`
can be used as a fastpath. */
static bool
vm_call_iseq_optimizable_p(const struct rb_callinfo *ci, const struct rb_callcache *cc)
{
return !IS_ARGS_SPLAT(ci) && !IS_ARGS_KEYWORD(ci) &&
!(METHOD_ENTRY_VISI(vm_cc_cme(cc)) == METHOD_VISI_PROTECTED);
}
#endif /* RUBY_INSNHELPER_H */