|
|
|
|
/*
|
|
|
|
|
* Tagged type definition (duk_tval) and accessor macros.
|
|
|
|
|
*
|
|
|
|
|
* Access all fields through the accessor macros, as the representation
|
|
|
|
|
* is quite tricky.
|
|
|
|
|
*
|
|
|
|
|
* There are two packed type alternatives: an 8-byte representation
|
|
|
|
|
* based on an IEEE double (preferred for compactness), and a 12-byte
|
|
|
|
|
* representation (portability). The latter is needed also in e.g.
|
|
|
|
|
* 64-bit environments (it usually pads to 16 bytes per value).
|
|
|
|
|
*
|
|
|
|
|
* Selecting the tagged type format involves many trade-offs (memory
|
|
|
|
|
* use, size and performance of generated code, portability, etc).
|
|
|
|
|
*
|
|
|
|
|
* NB: because macro arguments are often expressions, macros should
|
|
|
|
|
* avoid evaluating their argument more than once.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#if !defined(DUK_TVAL_H_INCLUDED)
|
|
|
|
|
#define DUK_TVAL_H_INCLUDED
|
|
|
|
|
|
|
|
|
|
/* sanity */
|
|
|
|
|
#if !defined(DUK_USE_DOUBLE_LE) && !defined(DUK_USE_DOUBLE_ME) && !defined(DUK_USE_DOUBLE_BE)
|
|
|
|
|
#error unsupported: cannot determine byte order variant
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_PACKED_TVAL)
|
|
|
|
|
/* ======================================================================== */
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Packed 8-byte representation
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* use duk_double_union as duk_tval directly */
|
|
|
|
|
typedef union duk_double_union duk_tval;
|
|
|
|
|
typedef struct {
|
|
|
|
|
duk_uint16_t a;
|
|
|
|
|
duk_uint16_t b;
|
|
|
|
|
duk_uint16_t c;
|
|
|
|
|
duk_uint16_t d;
|
|
|
|
|
} duk_tval_unused;
|
|
|
|
|
|
|
|
|
|
/* tags */
|
|
|
|
|
#define DUK_TAG_NORMALIZED_NAN 0x7ff8UL /* the NaN variant we use */
|
|
|
|
|
/* avoid tag 0xfff0, no risk of confusion with negative infinity */
|
|
|
|
|
#define DUK_TAG_MIN 0xfff1UL
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TAG_FASTINT 0xfff1UL /* embed: integer value */
|
|
|
|
|
#endif
|
|
|
|
|
#define DUK_TAG_UNUSED 0xfff2UL /* marker; not actual tagged value */
|
|
|
|
|
#define DUK_TAG_UNDEFINED 0xfff3UL /* embed: nothing */
|
|
|
|
|
#define DUK_TAG_NULL 0xfff4UL /* embed: nothing */
|
|
|
|
|
#define DUK_TAG_BOOLEAN 0xfff5UL /* embed: 0 or 1 (false or true) */
|
|
|
|
|
/* DUK_TAG_NUMBER would logically go here, but it has multiple 'tags' */
|
|
|
|
|
#define DUK_TAG_POINTER 0xfff6UL /* embed: void ptr */
|
|
|
|
|
#define DUK_TAG_LIGHTFUNC 0xfff7UL /* embed: func ptr */
|
|
|
|
|
#define DUK_TAG_STRING 0xfff8UL /* embed: duk_hstring ptr */
|
|
|
|
|
#define DUK_TAG_OBJECT 0xfff9UL /* embed: duk_hobject ptr */
|
|
|
|
|
#define DUK_TAG_BUFFER 0xfffaUL /* embed: duk_hbuffer ptr */
|
|
|
|
|
#define DUK_TAG_MAX 0xfffaUL
|
|
|
|
|
|
|
|
|
|
/* for convenience */
|
|
|
|
|
#define DUK_XTAG_BOOLEAN_FALSE 0xfff50000UL
|
|
|
|
|
#define DUK_XTAG_BOOLEAN_TRUE 0xfff50001UL
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_IS_VALID_TAG(tv) \
|
|
|
|
|
(DUK_TVAL_GET_TAG((tv)) - DUK_TAG_MIN <= DUK_TAG_MAX - DUK_TAG_MIN)
|
|
|
|
|
|
|
|
|
|
/* DUK_TVAL_UNUSED initializer for duk_tval_unused, works for any endianness. */
|
|
|
|
|
#define DUK_TVAL_UNUSED_INITIALIZER() \
|
|
|
|
|
{ DUK_TAG_UNUSED, DUK_TAG_UNUSED, DUK_TAG_UNUSED, DUK_TAG_UNUSED }
|
|
|
|
|
|
|
|
|
|
/* two casts to avoid gcc warning: "warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]" */
|
|
|
|
|
#if defined(DUK_USE_64BIT_OPS)
|
|
|
|
|
#if defined(DUK_USE_DOUBLE_ME)
|
|
|
|
|
#define DUK__TVAL_SET_TAGGEDPOINTER(tv,h,tag) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) (tag)) << 16) | (((duk_uint64_t) (duk_uint32_t) (h)) << 32); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK__TVAL_SET_TAGGEDPOINTER(tv,h,tag) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) (tag)) << 48) | ((duk_uint64_t) (duk_uint32_t) (h)); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#endif
|
|
|
|
|
#else /* DUK_USE_64BIT_OPS */
|
|
|
|
|
#define DUK__TVAL_SET_TAGGEDPOINTER(tv,h,tag) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI0] = ((duk_uint32_t) (tag)) << 16; \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI1] = (duk_uint32_t) (h); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#endif /* DUK_USE_64BIT_OPS */
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_64BIT_OPS)
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
/* Double casting for pointer to avoid gcc warning (cast from pointer to integer of different size) */
|
|
|
|
|
#if defined(DUK_USE_DOUBLE_ME)
|
|
|
|
|
#define DUK__TVAL_SET_LIGHTFUNC(tv,fp,flags) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) DUK_TAG_LIGHTFUNC) << 16) | \
|
|
|
|
|
((duk_uint64_t) (flags)) | \
|
|
|
|
|
(((duk_uint64_t) (duk_uint32_t) (fp)) << 32); \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
} while (0)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK__TVAL_SET_LIGHTFUNC(tv,fp,flags) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) DUK_TAG_LIGHTFUNC) << 48) | \
|
|
|
|
|
(((duk_uint64_t) (flags)) << 32) | \
|
|
|
|
|
((duk_uint64_t) (duk_uint32_t) (fp)); \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
} while (0)
|
|
|
|
|
#endif
|
|
|
|
|
#else /* DUK_USE_64BIT_OPS */
|
|
|
|
|
#define DUK__TVAL_SET_LIGHTFUNC(tv,fp,flags) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI0] = (((duk_uint32_t) DUK_TAG_LIGHTFUNC) << 16) | ((duk_uint32_t) (flags)); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI1] = (duk_uint32_t) (fp); \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
} while (0)
|
|
|
|
|
#endif /* DUK_USE_64BIT_OPS */
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
/* Note: masking is done for 'i' to deal with negative numbers correctly */
|
|
|
|
|
#if defined(DUK_USE_DOUBLE_ME)
|
|
|
|
|
#define DUK__TVAL_SET_I48(tv,i) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI0] = ((duk_uint32_t) DUK_TAG_FASTINT) << 16 | (((duk_uint32_t) ((i) >> 32)) & 0x0000ffffUL); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI1] = (duk_uint32_t) (i); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK__TVAL_SET_U32(tv,i) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI0] = ((duk_uint32_t) DUK_TAG_FASTINT) << 16; \
|
|
|
|
|
duk__tv->ui[DUK_DBL_IDX_UI1] = (duk_uint32_t) (i); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK__TVAL_SET_I48(tv,i) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) DUK_TAG_FASTINT) << 48) | (((duk_uint64_t) (i)) & 0x0000ffffffffffffULL); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK__TVAL_SET_U32(tv,i) do { \
|
|
|
|
|
(tv)->ull[DUK_DBL_IDX_ULL0] = (((duk_uint64_t) DUK_TAG_FASTINT) << 48) | (duk_uint64_t) (i); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
/* This needs to go through a cast because sign extension is needed. */
|
|
|
|
|
#define DUK__TVAL_SET_I32(tv,i) do { \
|
|
|
|
|
duk_int64_t duk__tmp = (duk_int64_t) (i); \
|
|
|
|
|
DUK_TVAL_SET_I48((tv), duk__tmp); \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
/* XXX: Clumsy sign extend and masking of 16 topmost bits. */
|
|
|
|
|
#if defined(DUK_USE_DOUBLE_ME)
|
|
|
|
|
#define DUK__TVAL_GET_FASTINT(tv) (((duk_int64_t) ((((duk_uint64_t) (tv)->ui[DUK_DBL_IDX_UI0]) << 32) | ((duk_uint64_t) (tv)->ui[DUK_DBL_IDX_UI1]))) << 16 >> 16)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK__TVAL_GET_FASTINT(tv) ((((duk_int64_t) (tv)->ull[DUK_DBL_IDX_ULL0]) << 16) >> 16)
|
|
|
|
|
#endif
|
|
|
|
|
#define DUK__TVAL_GET_FASTINT_U32(tv) ((tv)->ui[DUK_DBL_IDX_UI1])
|
|
|
|
|
#define DUK__TVAL_GET_FASTINT_I32(tv) ((duk_int32_t) (tv)->ui[DUK_DBL_IDX_UI1])
|
|
|
|
|
#endif /* DUK_USE_FASTINT */
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_UNDEFINED(tv) do { \
|
|
|
|
|
(tv)->us[DUK_DBL_IDX_US0] = (duk_uint16_t) DUK_TAG_UNDEFINED; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_UNUSED(tv) do { \
|
|
|
|
|
(tv)->us[DUK_DBL_IDX_US0] = (duk_uint16_t) DUK_TAG_UNUSED; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_NULL(tv) do { \
|
|
|
|
|
(tv)->us[DUK_DBL_IDX_US0] = (duk_uint16_t) DUK_TAG_NULL; \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_BOOLEAN(tv,val) DUK_DBLUNION_SET_HIGH32((tv), (((duk_uint32_t) DUK_TAG_BOOLEAN) << 16) | ((duk_uint32_t) (val)))
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_NAN(tv) DUK_DBLUNION_SET_NAN_FULL((tv))
|
|
|
|
|
|
|
|
|
|
/* Assumes that caller has normalized NaNs, otherwise trouble ahead. */
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TVAL_SET_DOUBLE(tv,d) do { \
|
|
|
|
|
duk_double_t duk__dblval; \
|
|
|
|
|
duk__dblval = (d); \
|
|
|
|
|
DUK_ASSERT_DOUBLE_IS_NORMALIZED(duk__dblval); \
|
|
|
|
|
DUK_DBLUNION_SET_DOUBLE((tv), duk__dblval); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_I48(tv,i) DUK__TVAL_SET_I48((tv), (i))
|
|
|
|
|
#define DUK_TVAL_SET_I32(tv,i) DUK__TVAL_SET_I32((tv), (i))
|
|
|
|
|
#define DUK_TVAL_SET_U32(tv,i) DUK__TVAL_SET_U32((tv), (i))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_FAST(tv,d) duk_tval_set_number_chkfast_fast((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(tv,d) duk_tval_set_number_chkfast_slow((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER(tv,d) DUK_TVAL_SET_DOUBLE((tv), (d))
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_FAST(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__d; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
if (DUK_TVAL_IS_DOUBLE(duk__tv)) { \
|
|
|
|
|
duk__d = DUK_TVAL_GET_DOUBLE(duk__tv); \
|
|
|
|
|
DUK_TVAL_SET_NUMBER_CHKFAST_FAST(duk__tv, duk__d); \
|
|
|
|
|
} \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_SLOW(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__d; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
if (DUK_TVAL_IS_DOUBLE(duk__tv)) { \
|
|
|
|
|
duk__d = DUK_TVAL_GET_DOUBLE(duk__tv); \
|
|
|
|
|
DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(duk__tv, duk__d); \
|
|
|
|
|
} \
|
|
|
|
|
} while (0)
|
|
|
|
|
#else /* DUK_USE_FASTINT */
|
|
|
|
|
#define DUK_TVAL_SET_DOUBLE(tv,d) do { \
|
|
|
|
|
duk_double_t duk__dblval; \
|
|
|
|
|
duk__dblval = (d); \
|
|
|
|
|
DUK_ASSERT_DOUBLE_IS_NORMALIZED(duk__dblval); \
|
|
|
|
|
DUK_DBLUNION_SET_DOUBLE((tv), duk__dblval); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_I48(tv,i) DUK_TVAL_SET_DOUBLE((tv), (duk_double_t) (i)) /* XXX: fast int-to-double */
|
|
|
|
|
#define DUK_TVAL_SET_I32(tv,i) DUK_TVAL_SET_DOUBLE((tv), (duk_double_t) (i))
|
|
|
|
|
#define DUK_TVAL_SET_U32(tv,i) DUK_TVAL_SET_DOUBLE((tv), (duk_double_t) (i))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_FAST(tv,d) DUK_TVAL_SET_DOUBLE((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(tv,d) DUK_TVAL_SET_DOUBLE((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER(tv,d) DUK_TVAL_SET_DOUBLE((tv), (d))
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_FAST(tv) do { } while (0)
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_SLOW(tv) do { } while (0)
|
|
|
|
|
#endif /* DUK_USE_FASTINT */
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_FASTINT(tv,i) DUK_TVAL_SET_I48((tv), (i)) /* alias */
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_LIGHTFUNC(tv,fp,flags) DUK__TVAL_SET_LIGHTFUNC((tv), (fp), (flags))
|
|
|
|
|
#define DUK_TVAL_SET_STRING(tv,h) DUK__TVAL_SET_TAGGEDPOINTER((tv), (h), DUK_TAG_STRING)
|
|
|
|
|
#define DUK_TVAL_SET_OBJECT(tv,h) DUK__TVAL_SET_TAGGEDPOINTER((tv), (h), DUK_TAG_OBJECT)
|
|
|
|
|
#define DUK_TVAL_SET_BUFFER(tv,h) DUK__TVAL_SET_TAGGEDPOINTER((tv), (h), DUK_TAG_BUFFER)
|
|
|
|
|
#define DUK_TVAL_SET_POINTER(tv,p) DUK__TVAL_SET_TAGGEDPOINTER((tv), (p), DUK_TAG_POINTER)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_TVAL(tv,x) do { *(tv) = *(x); } while (0)
|
|
|
|
|
|
|
|
|
|
/* getters */
|
|
|
|
|
#define DUK_TVAL_GET_BOOLEAN(tv) ((duk_small_int_t) (tv)->us[DUK_DBL_IDX_US1])
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TVAL_GET_DOUBLE(tv) ((tv)->d)
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT(tv) DUK__TVAL_GET_FASTINT((tv))
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT_U32(tv) DUK__TVAL_GET_FASTINT_U32((tv))
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT_I32(tv) DUK__TVAL_GET_FASTINT_I32((tv))
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) duk_tval_get_number_packed((tv))
|
|
|
|
|
#else
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) ((tv)->d)
|
|
|
|
|
#define DUK_TVAL_GET_DOUBLE(tv) ((tv)->d)
|
|
|
|
|
#endif
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC(tv,out_fp,out_flags) do { \
|
|
|
|
|
(out_flags) = (tv)->ui[DUK_DBL_IDX_UI0] & 0xffffUL; \
|
|
|
|
|
(out_fp) = (duk_c_function) (tv)->ui[DUK_DBL_IDX_UI1]; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC_FUNCPTR(tv) ((duk_c_function) ((tv)->ui[DUK_DBL_IDX_UI1]))
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC_FLAGS(tv) (((duk_small_int_t) (tv)->ui[DUK_DBL_IDX_UI0]) & 0xffffUL)
|
|
|
|
|
#define DUK_TVAL_GET_STRING(tv) ((duk_hstring *) (tv)->vp[DUK_DBL_IDX_VP1])
|
|
|
|
|
#define DUK_TVAL_GET_OBJECT(tv) ((duk_hobject *) (tv)->vp[DUK_DBL_IDX_VP1])
|
|
|
|
|
#define DUK_TVAL_GET_BUFFER(tv) ((duk_hbuffer *) (tv)->vp[DUK_DBL_IDX_VP1])
|
|
|
|
|
#define DUK_TVAL_GET_POINTER(tv) ((void *) (tv)->vp[DUK_DBL_IDX_VP1])
|
|
|
|
|
#define DUK_TVAL_GET_HEAPHDR(tv) ((duk_heaphdr *) (tv)->vp[DUK_DBL_IDX_VP1])
|
|
|
|
|
|
|
|
|
|
/* decoding */
|
|
|
|
|
#define DUK_TVAL_GET_TAG(tv) ((duk_small_uint_t) (tv)->us[DUK_DBL_IDX_US0])
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_IS_UNDEFINED(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_UNDEFINED)
|
|
|
|
|
#define DUK_TVAL_IS_UNUSED(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_UNUSED)
|
|
|
|
|
#define DUK_TVAL_IS_NULL(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_NULL)
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_BOOLEAN)
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN_TRUE(tv) ((tv)->ui[DUK_DBL_IDX_UI0] == DUK_XTAG_BOOLEAN_TRUE)
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN_FALSE(tv) ((tv)->ui[DUK_DBL_IDX_UI0] == DUK_XTAG_BOOLEAN_FALSE)
|
|
|
|
|
#define DUK_TVAL_IS_LIGHTFUNC(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_LIGHTFUNC)
|
|
|
|
|
#define DUK_TVAL_IS_STRING(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_STRING)
|
|
|
|
|
#define DUK_TVAL_IS_OBJECT(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_OBJECT)
|
|
|
|
|
#define DUK_TVAL_IS_BUFFER(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_BUFFER)
|
|
|
|
|
#define DUK_TVAL_IS_POINTER(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_POINTER)
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
/* 0xfff0 is -Infinity */
|
|
|
|
|
#define DUK_TVAL_IS_DOUBLE(tv) (DUK_TVAL_GET_TAG((tv)) <= 0xfff0UL)
|
|
|
|
|
#define DUK_TVAL_IS_FASTINT(tv) (DUK_TVAL_GET_TAG((tv)) == DUK_TAG_FASTINT)
|
|
|
|
|
#define DUK_TVAL_IS_NUMBER(tv) (DUK_TVAL_GET_TAG((tv)) <= 0xfff1UL)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK_TVAL_IS_NUMBER(tv) (DUK_TVAL_GET_TAG((tv)) <= 0xfff0UL)
|
|
|
|
|
#define DUK_TVAL_IS_DOUBLE(tv) DUK_TVAL_IS_NUMBER((tv))
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
/* This is performance critical because it appears in every DECREF. */
|
|
|
|
|
#define DUK_TVAL_IS_HEAP_ALLOCATED(tv) (DUK_TVAL_GET_TAG((tv)) >= DUK_TAG_STRING)
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
DUK_INTERNAL_DECL duk_double_t duk_tval_get_number_packed(duk_tval *tv);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#else /* DUK_USE_PACKED_TVAL */
|
|
|
|
|
/* ======================================================================== */
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Portable 12-byte representation
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* Note: not initializing all bytes is normally not an issue: Duktape won't
|
|
|
|
|
* read or use the uninitialized bytes so valgrind won't issue warnings.
|
|
|
|
|
* In some special cases a harmless valgrind warning may be issued though.
|
|
|
|
|
* For example, the DumpHeap debugger command writes out a compiled function's
|
|
|
|
|
* 'data' area as is, including any uninitialized bytes, which causes a
|
|
|
|
|
* valgrind warning.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
typedef struct duk_tval_struct duk_tval;
|
|
|
|
|
|
|
|
|
|
struct duk_tval_struct {
|
|
|
|
|
duk_small_uint_t t;
|
|
|
|
|
duk_small_uint_t v_extra;
|
|
|
|
|
union {
|
|
|
|
|
duk_double_t d;
|
|
|
|
|
duk_small_int_t i;
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
duk_int64_t fi; /* if present, forces 16-byte duk_tval */
|
|
|
|
|
#endif
|
|
|
|
|
void *voidptr;
|
|
|
|
|
duk_hstring *hstring;
|
|
|
|
|
duk_hobject *hobject;
|
|
|
|
|
duk_hcompfunc *hcompfunc;
|
|
|
|
|
duk_hnatfunc *hnatfunc;
|
|
|
|
|
duk_hthread *hthread;
|
|
|
|
|
duk_hbuffer *hbuffer;
|
|
|
|
|
duk_heaphdr *heaphdr;
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
duk_c_function lightfunc;
|
|
|
|
|
} v;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
|
duk_small_uint_t t;
|
|
|
|
|
duk_small_uint_t v_extra;
|
|
|
|
|
/* The rest of the fields don't matter except for debug dumps and such
|
|
|
|
|
* for which a partial initializer may trigger out-ot-bounds memory
|
|
|
|
|
* reads. Include a double field which is usually as large or larger
|
|
|
|
|
* than pointers (not always however).
|
|
|
|
|
*/
|
|
|
|
|
duk_double_t d;
|
|
|
|
|
} duk_tval_unused;
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_UNUSED_INITIALIZER() \
|
|
|
|
|
{ DUK_TAG_UNUSED, 0, 0.0 }
|
|
|
|
|
|
|
|
|
|
#define DUK_TAG_MIN 0
|
|
|
|
|
#define DUK_TAG_NUMBER 0 /* DUK_TAG_NUMBER only defined for non-packed duk_tval */
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TAG_FASTINT 1
|
|
|
|
|
#endif
|
|
|
|
|
#define DUK_TAG_UNDEFINED 2
|
|
|
|
|
#define DUK_TAG_NULL 3
|
|
|
|
|
#define DUK_TAG_BOOLEAN 4
|
|
|
|
|
#define DUK_TAG_POINTER 5
|
|
|
|
|
#define DUK_TAG_LIGHTFUNC 6
|
|
|
|
|
#define DUK_TAG_UNUSED 7 /* marker; not actual tagged type */
|
|
|
|
|
#define DUK_TAG_STRING 8 /* first heap allocated, match bit boundary */
|
|
|
|
|
#define DUK_TAG_OBJECT 9
|
|
|
|
|
#define DUK_TAG_BUFFER 10
|
|
|
|
|
#define DUK_TAG_MAX 10
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_IS_VALID_TAG(tv) \
|
|
|
|
|
(DUK_TVAL_GET_TAG((tv)) - DUK_TAG_MIN <= DUK_TAG_MAX - DUK_TAG_MIN)
|
|
|
|
|
|
|
|
|
|
/* DUK_TAG_NUMBER is intentionally first, as it is the default clause in code
|
|
|
|
|
* to support the 8-byte representation. Further, it is a non-heap-allocated
|
|
|
|
|
* type so it should come before DUK_TAG_STRING. Finally, it should not break
|
|
|
|
|
* the tag value ranges covered by case-clauses in a switch-case.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* setters */
|
|
|
|
|
#define DUK_TVAL_SET_UNDEFINED(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_UNDEFINED; \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_UNUSED(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_UNUSED; \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_NULL(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_NULL; \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_BOOLEAN(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_BOOLEAN; \
|
|
|
|
|
duk__tv->v.i = (val); \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TVAL_SET_DOUBLE(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__dblval; \
|
|
|
|
|
duk__dblval = (val); \
|
|
|
|
|
DUK_ASSERT_DOUBLE_IS_NORMALIZED(duk__dblval); /* nop for unpacked duk_tval */ \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_NUMBER; \
|
|
|
|
|
duk__tv->v.d = duk__dblval; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_I48(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_FASTINT; \
|
|
|
|
|
duk__tv->v.fi = (val); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_U32(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_FASTINT; \
|
|
|
|
|
duk__tv->v.fi = (duk_int64_t) (val); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_I32(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_FASTINT; \
|
|
|
|
|
duk__tv->v.fi = (duk_int64_t) (val); \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_FAST(tv,d) \
|
|
|
|
|
duk_tval_set_number_chkfast_fast((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(tv,d) \
|
|
|
|
|
duk_tval_set_number_chkfast_slow((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER(tv,val) \
|
|
|
|
|
DUK_TVAL_SET_DOUBLE((tv), (val))
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_FAST(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__d; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
if (DUK_TVAL_IS_DOUBLE(duk__tv)) { \
|
|
|
|
|
duk__d = DUK_TVAL_GET_DOUBLE(duk__tv); \
|
|
|
|
|
DUK_TVAL_SET_NUMBER_CHKFAST_FAST(duk__tv, duk__d); \
|
|
|
|
|
} \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_SLOW(tv) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__d; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
if (DUK_TVAL_IS_DOUBLE(duk__tv)) { \
|
|
|
|
|
duk__d = DUK_TVAL_GET_DOUBLE(duk__tv); \
|
|
|
|
|
DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(duk__tv, duk__d); \
|
|
|
|
|
} \
|
|
|
|
|
} while (0)
|
|
|
|
|
#else /* DUK_USE_FASTINT */
|
|
|
|
|
#define DUK_TVAL_SET_DOUBLE(tv,d) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_I48(tv,val) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (duk_double_t) (val)) /* XXX: fast int-to-double */
|
|
|
|
|
#define DUK_TVAL_SET_U32(tv,val) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (duk_double_t) (val))
|
|
|
|
|
#define DUK_TVAL_SET_I32(tv,val) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (duk_double_t) (val))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER(tv,val) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk_double_t duk__dblval; \
|
|
|
|
|
duk__dblval = (val); \
|
|
|
|
|
DUK_ASSERT_DOUBLE_IS_NORMALIZED(duk__dblval); /* nop for unpacked duk_tval */ \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_NUMBER; \
|
|
|
|
|
duk__tv->v.d = duk__dblval; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_FAST(tv,d) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (d))
|
|
|
|
|
#define DUK_TVAL_SET_NUMBER_CHKFAST_SLOW(tv,d) \
|
|
|
|
|
DUK_TVAL_SET_NUMBER((tv), (d))
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_FAST(tv) do { } while (0)
|
|
|
|
|
#define DUK_TVAL_CHKFAST_INPLACE_SLOW(tv) do { } while (0)
|
|
|
|
|
#endif /* DUK_USE_FASTINT */
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_FASTINT(tv,i) \
|
|
|
|
|
DUK_TVAL_SET_I48((tv), (i)) /* alias */
|
|
|
|
|
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
#define DUK_TVAL_SET_POINTER(tv,hptr) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_POINTER; \
|
|
|
|
|
duk__tv->v.voidptr = (hptr); \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_LIGHTFUNC(tv,fp,flags) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_LIGHTFUNC; \
|
|
|
|
|
duk__tv->v_extra = (flags); \
|
|
|
|
|
duk__tv->v.lightfunc = (duk_c_function) (fp); \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_STRING(tv,hptr) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_STRING; \
|
|
|
|
|
duk__tv->v.hstring = (hptr); \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_OBJECT(tv,hptr) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_OBJECT; \
|
|
|
|
|
duk__tv->v.hobject = (hptr); \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_BUFFER(tv,hptr) do { \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_BUFFER; \
|
|
|
|
|
duk__tv->v.hbuffer = (hptr); \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_NAN(tv) do { \
|
|
|
|
|
/* in non-packed representation we don't care about which NaN is used */ \
|
|
|
|
|
duk_tval *duk__tv; \
|
|
|
|
|
duk__tv = (tv); \
|
|
|
|
|
duk__tv->t = DUK_TAG_NUMBER; \
|
|
|
|
|
duk__tv->v.d = DUK_DOUBLE_NAN; \
|
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_TVAL(tv,x) do { *(tv) = *(x); } while (0)
|
|
|
|
|
|
|
|
|
|
/* getters */
|
|
|
|
|
#define DUK_TVAL_GET_BOOLEAN(tv) ((tv)->v.i)
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TVAL_GET_DOUBLE(tv) ((tv)->v.d)
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT(tv) ((tv)->v.fi)
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT_U32(tv) ((duk_uint32_t) ((tv)->v.fi))
|
|
|
|
|
#define DUK_TVAL_GET_FASTINT_I32(tv) ((duk_int32_t) ((tv)->v.fi))
|
|
|
|
|
#if 0
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) (DUK_TVAL_IS_FASTINT((tv)) ? \
|
|
|
|
|
(duk_double_t) DUK_TVAL_GET_FASTINT((tv)) : \
|
|
|
|
|
DUK_TVAL_GET_DOUBLE((tv)))
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) duk_tval_get_number_unpacked((tv))
|
|
|
|
|
#else
|
|
|
|
|
/* This seems reasonable overall. */
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) (DUK_TVAL_IS_FASTINT((tv)) ? \
|
|
|
|
|
duk_tval_get_number_unpacked_fastint((tv)) : \
|
|
|
|
|
DUK_TVAL_GET_DOUBLE((tv)))
|
|
|
|
|
#endif
|
|
|
|
|
#else
|
|
|
|
|
#define DUK_TVAL_GET_NUMBER(tv) ((tv)->v.d)
|
|
|
|
|
#define DUK_TVAL_GET_DOUBLE(tv) ((tv)->v.d)
|
|
|
|
|
#endif /* DUK_USE_FASTINT */
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
#define DUK_TVAL_GET_POINTER(tv) ((tv)->v.voidptr)
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC(tv,out_fp,out_flags) do { \
|
|
|
|
|
(out_flags) = (duk_uint32_t) (tv)->v_extra; \
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
(out_fp) = (tv)->v.lightfunc; \
|
|
|
|
|
} while (0)
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC_FUNCPTR(tv) ((tv)->v.lightfunc)
|
|
|
|
|
#define DUK_TVAL_GET_LIGHTFUNC_FLAGS(tv) ((duk_uint32_t) ((tv)->v_extra))
|
|
|
|
|
#define DUK_TVAL_GET_STRING(tv) ((tv)->v.hstring)
|
|
|
|
|
#define DUK_TVAL_GET_OBJECT(tv) ((tv)->v.hobject)
|
|
|
|
|
#define DUK_TVAL_GET_BUFFER(tv) ((tv)->v.hbuffer)
|
|
|
|
|
#define DUK_TVAL_GET_HEAPHDR(tv) ((tv)->v.heaphdr)
|
|
|
|
|
|
|
|
|
|
/* decoding */
|
|
|
|
|
#define DUK_TVAL_GET_TAG(tv) ((tv)->t)
|
|
|
|
|
#define DUK_TVAL_IS_UNDEFINED(tv) ((tv)->t == DUK_TAG_UNDEFINED)
|
|
|
|
|
#define DUK_TVAL_IS_UNUSED(tv) ((tv)->t == DUK_TAG_UNUSED)
|
|
|
|
|
#define DUK_TVAL_IS_NULL(tv) ((tv)->t == DUK_TAG_NULL)
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN(tv) ((tv)->t == DUK_TAG_BOOLEAN)
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN_TRUE(tv) (((tv)->t == DUK_TAG_BOOLEAN) && ((tv)->v.i != 0))
|
|
|
|
|
#define DUK_TVAL_IS_BOOLEAN_FALSE(tv) (((tv)->t == DUK_TAG_BOOLEAN) && ((tv)->v.i == 0))
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_TVAL_IS_DOUBLE(tv) ((tv)->t == DUK_TAG_NUMBER)
|
|
|
|
|
#define DUK_TVAL_IS_FASTINT(tv) ((tv)->t == DUK_TAG_FASTINT)
|
|
|
|
|
#define DUK_TVAL_IS_NUMBER(tv) ((tv)->t == DUK_TAG_NUMBER || \
|
|
|
|
|
(tv)->t == DUK_TAG_FASTINT)
|
|
|
|
|
#else
|
|
|
|
|
#define DUK_TVAL_IS_NUMBER(tv) ((tv)->t == DUK_TAG_NUMBER)
|
|
|
|
|
#define DUK_TVAL_IS_DOUBLE(tv) DUK_TVAL_IS_NUMBER((tv))
|
|
|
|
|
#endif /* DUK_USE_FASTINT */
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
#define DUK_TVAL_IS_POINTER(tv) ((tv)->t == DUK_TAG_POINTER)
|
|
|
|
|
#define DUK_TVAL_IS_LIGHTFUNC(tv) ((tv)->t == DUK_TAG_LIGHTFUNC)
|
|
|
|
|
#define DUK_TVAL_IS_STRING(tv) ((tv)->t == DUK_TAG_STRING)
|
|
|
|
|
#define DUK_TVAL_IS_OBJECT(tv) ((tv)->t == DUK_TAG_OBJECT)
|
|
|
|
|
#define DUK_TVAL_IS_BUFFER(tv) ((tv)->t == DUK_TAG_BUFFER)
|
|
|
|
|
|
|
|
|
|
/* This is performance critical because it's needed for every DECREF.
|
|
|
|
|
* Take advantage of the fact that the first heap allocated tag is 8,
|
|
|
|
|
* so that bit 3 is set for all heap allocated tags (and never set for
|
|
|
|
|
* non-heap-allocated tags).
|
|
|
|
|
*/
|
|
|
|
|
#if 0
|
|
|
|
|
#define DUK_TVAL_IS_HEAP_ALLOCATED(tv) ((tv)->t >= DUK_TAG_STRING)
|
|
|
|
|
#endif
|
|
|
|
|
#define DUK_TVAL_IS_HEAP_ALLOCATED(tv) ((tv)->t & 0x08)
|
|
|
|
|
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#if 0
|
|
|
|
|
DUK_INTERNAL_DECL duk_double_t duk_tval_get_number_unpacked(duk_tval *tv);
|
|
|
|
|
#endif
|
|
|
|
|
DUK_INTERNAL_DECL duk_double_t duk_tval_get_number_unpacked_fastint(duk_tval *tv);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#endif /* DUK_USE_PACKED_TVAL */
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Convenience (independent of representation)
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#define DUK_TVAL_SET_BOOLEAN_TRUE(tv) DUK_TVAL_SET_BOOLEAN((tv), 1)
|
|
|
|
|
#define DUK_TVAL_SET_BOOLEAN_FALSE(tv) DUK_TVAL_SET_BOOLEAN((tv), 0)
|
|
|
|
|
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
/* Lightfunc flags packing and unpacking. */
|
|
|
|
|
/* Sign extend: 0x0000##00 -> 0x##000000 -> sign extend to 0xssssss##.
|
|
|
|
|
* Avoid signed shifts due to portability limitations.
|
|
|
|
|
*/
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
#define DUK_LFUNC_FLAGS_GET_MAGIC(lf_flags) \
|
|
|
|
|
((duk_int32_t) (duk_int8_t) (((duk_uint16_t) (lf_flags)) >> 8))
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
#define DUK_LFUNC_FLAGS_GET_LENGTH(lf_flags) \
|
|
|
|
|
(((lf_flags) >> 4) & 0x0f)
|
|
|
|
|
#define DUK_LFUNC_FLAGS_GET_NARGS(lf_flags) \
|
|
|
|
|
((lf_flags) & 0x0f)
|
|
|
|
|
#define DUK_LFUNC_FLAGS_PACK(magic,length,nargs) \
|
|
|
|
|
(((magic) & 0xff) << 8) | ((length) << 4) | (nargs)
|
|
|
|
|
|
|
|
|
|
#define DUK_LFUNC_NARGS_VARARGS 0x0f /* varargs marker */
|
|
|
|
|
#define DUK_LFUNC_NARGS_MIN 0x00
|
|
|
|
|
#define DUK_LFUNC_NARGS_MAX 0x0e /* max, excl. varargs marker */
|
|
|
|
|
#define DUK_LFUNC_LENGTH_MIN 0x00
|
|
|
|
|
#define DUK_LFUNC_LENGTH_MAX 0x0f
|
|
|
|
|
#define DUK_LFUNC_MAGIC_MIN (-0x80)
|
|
|
|
|
#define DUK_LFUNC_MAGIC_MAX 0x7f
|
First round of lightfunc changes
A lot of changes to add preliminary lightfunc support:
* Add LIGHTFUNC tagged type to duk_tval.h and API.
* Internal changes for preliminary to support lightfuncs in call handling
and other operations (FIXMEs left in obvious places where support is
still missing after this commit)
* Preliminary Ecmascript and API testcases for lightfuncs
Detailed notes:
* Because magic is signed, reading it back involves sign extension which is
quite verbose to do in C. Use macros for reading the magic value and other
bit fields encoded in the flags.
* Function.prototype.bind(): the 'length' property of a bound function now
comes out wrong. We could simply look up the virtual 'length' property
even if h_target is NULL: no extra code and binding is relatively rare in
hot paths. Rewrite more cleanly in any case.
* The use flag DUK_USE_LIGHTFUNC_BUILTINS controls the forced lightfunc
conversion of built-ins. This results in non-compliant built-ins but
significant memory savings in very memory poor environments.
* Reject eval(), Thread.yield/resume as lightfuncs. These functions have
current assertions that they must be called as fully fledged functions.
* Lightfuncs are serialized like ordinary functions for JSON, JX, and JC
by this diff.
* Add 'magic' to activation for lightfuncs. It will be needed for lightweight
functions: we don't have the duk_tval related to the lightfunc, so we must
copy the magic value to the activation when a call is made.
* When lightfuncs are used as property lookup base values, continue property
lookup from the Function.prototype object. This is necessary to allow e.g.
``func.call()`` and ``func.apply()`` to be used.
* Call handling had to be reworked for lightfuncs, especially how bound
function chains are handled. This is a relatively large change but is
necessary to support lightweight functions properly in bound function
resolution.
The current solution is not ideal. The bytecode executor will first try an
ecma-to-ecma call setup which resolves the bound function chain first. If
the final, unbound function is not viable (a native function) the call setup
returns with an error code. The caller will then perform a normal call.
Although bound function resolution has already been done, the normal call
handling code will re-do it (and detect there is nothing to do).
This situation could be avoided by decoupling bound function handling and
effective this binding computation from the actual call setup. The caller
could then to do this prestep first, and only then decide whether to use an
ecma-to-ecma call or an ordinary heavyweight call.
Remove duk__find_nonbound_function as unused.
* Use indirect magic to allow LIGHTFUNCs for Date. Most of the built-in
functions not directly eligible as lightfuncs are the Date built-in methods,
whose magic values contain too much information to fit into the 8-bit magic
of a LIGHTFUNC value.
To work around this, add an array (duk__date_magics[]) containing the
actual control flags needed by the built-ins, and make the Date built-in
magic value an index into this table. With this change Date built-ins are
successfully converted to lightfuncs.
Testcase fixes:
- Whitespace fixes
- Print error for indirect eval error to make diagnosis easier
- Fix error string to match errmsg updated in this branch
11 years ago
|
|
|
|
|
|
|
|
/* fastint constants etc */
|
|
|
|
|
#if defined(DUK_USE_FASTINT)
|
|
|
|
|
#define DUK_FASTINT_MIN (-0x800000000000LL)
|
|
|
|
|
#define DUK_FASTINT_MAX 0x7fffffffffffLL
|
|
|
|
|
#define DUK_FASTINT_BITS 48
|
|
|
|
|
|
|
|
|
|
DUK_INTERNAL_DECL void duk_tval_set_number_chkfast_fast(duk_tval *tv, duk_double_t x);
|
|
|
|
|
DUK_INTERNAL_DECL void duk_tval_set_number_chkfast_slow(duk_tval *tv, duk_double_t x);
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#endif /* DUK_TVAL_H_INCLUDED */
|
