Memory Pool System Reference Manual
Contents
1. Introduction
This is the reference manual for the Memory Pool System.
This document is quite incomplete. At present it consists simply of reference descriptions of a number of MPS symbols (section 3). Many MPS symbols are not described here (see section 4 for a list). There are also no overview or protocol-oriented sections.
2. Overview
3. Reference
-
MPS_ARCH_AL
- function
mps_fix
- macro
MPS_FIX1
- macro
MPS_FIX12
- macro
MPS_FIX2
- macro
MPS_FIX_CALL
- constant
MPS_RES_LIMIT
- constant
MPS_RES_MEMORY
-
MPS_RES_PARAM
-
MPS_RM_CONST
-
MPS_RM_PROT
- function
mps_sac_alloc
- macro
MPS_SAC_ALLOC_FAST
- constant
MPS_SAC_CLASS_LIMIT
- function
mps_sac_free
- macro
MPS_SAC_FREE_FAST
- macro
MPS_SCAN_BEGIN
- macro
MPS_SCAN_END
- type
MPS_T_WORD
- constant
MPS_WORD_SHIFT
- constant
MPS_WORD_WIDTH
- type
mps_addr_t
- type
mps_align_t
- function
mps_alloc
-
mps_alloc_pattern_ramp
-
mps_alloc_pattern_ramp_collect_all
-
mps_amc_apply
-
mps_ap_alloc_pattern_begin
-
mps_ap_alloc_pattern_end
- function
mps_ap_alloc_pattern_reset
- function
mps_ap_frame_pop
- function
mps_ap_frame_push
- function
mps_arena_clamp
- function
mps_arena_class_cl
- type
mps_arena_class_t
- function
mps_arena_class_vm
- function
mps_arena_class_vmnz
- function
mps_arena_collect
- function
mps_arena_commit_limit
- function
mps_arena_commit_limit_set
- Function
mps_arena_committed
- function
mps_arena_create
- function
mps_arena_create_v
- function
mps_arena_expose
- function
mps_arena_formatted_objects_walk
- function
mps_arena_park
- function
mps_arena_release
- function
mps_arena_roots_walk
- function
mps_arena_spare_commit_limit
- function
mps_arena_spare_commit_limit_set
- function
mps_arena_spare_committed
- function
mps_bool_t
- function
mps_class_amc
- function
mps_class_mvff
- function
mps_class_snc
-
mps_class_mvt
- Type
mps_class_t
- function
mps_finalize
- type
mps_fmt_A_s
- type
mps_fmt_A_t
- type
mps_fmt_B_s
- Type
mps_fmt_B_t
- type
mps_fmt_auto_header_s
- type
mps_fmt_class_t
- type
mps_fmt_copy_t
- function
mps_fmt_create_A
- function
mps_fmt_create_B
- function
mps_fmt_create_auto_header
- type
mps_fmt_fwd_t
- type
mps_fmt_isfwd_t
- type
mps_fmt_pad_t
- type
mps_fmt_scan_t
- type
mps_fmt_skip_t
- type
mps_fmt_t
- type
mps_formatted_objects_stepper_t
- function
mps_free
- function
mps_lib_memcmp
- function
mps_lib_memcpy
- function
mps_lib_memset
-
mps_lib_telemetry_control
- function
mps_message_discard
- function
mps_message_finalization_ref
- function
mps_message_gc_condemned_size
- function
mps_message_gc_live_size
- function
mps_message_gc_not_condemned_size
- function
mps_message_get
-
mps_message_poll
- function
mps_message_queue_type
- type
mps_message_t
- function
mps_message_type
- function
mps_message_type_disable
- function
mps_message_type_enable
- function
mps_message_type_finalization
- function
mps_message_type_gc
- type
mps_message_type_t
- function
mps_pool_check_fenceposts
- structure
mps_pool_debug_option_s
- function
mps_rank_ambig
- function
mps_rank_exact
- type
mps_rank_t
- function
mps_rank_weak
- type
mps_reg_scan_t
- type
mps_res_t
- function
mps_root_create
- function
mps_root_create_fmt
- function
mps_root_create_reg
- function
mps_root_create_table
- function
mps_root_create_table_masked
- type
mps_root_scan_t
- type
mps_roots_stepper_t
- type
mps_sac_class_s
- function
mps_sac_create
- function
mps_sac_destroy
- function
mps_sac_flush
- type
mps_sac_t
- function
mps_stack_scan_ambig
- function
mps_telemetry_control
- function
mps_telemetry_flush
- function
mps_telemetry_intern
- function
mps_telemetry_label
- type
mps_thr_t
MPS_ARCH_AL
Name
Summary
MPS_ARCH_AL
is a C
preprocessor macro that indicates, if defined, that the target
processor architecture of the compilation is a member of the DEC Alpha
family. It is defined, if appropriate, by "mpstd.h".
Associated Protocols
None.
Resources
mpstd.h.
Description
See summary.
Example
#ifdef MPS_ARCH_AL typedef struct RegisterFile { unsigned long v0; unsigned long t0, t1, t2, t3, t4, t5, t6, t7; unsigned long s0, s1, s2, s3, s4, s5; unsigned long fp; unsigned long a0, a1, a2, a3, a4, a5; unsigned long t8, t9, t10, t11; unsigned long ra; unsigned long t12; unsigned long at, gp, sp, zero; unsigned long fir; unsigned psr; } RegisterFile; #endif /* MPS_ARCH_AL */
Error Handling
Not applicable.
See Also
MPS_PF_*
,
MPS_OS_*
,
MPS_BUILD_*
,
MPS_ARCH_*
Notes
Internal Notes
I'm not sure that the user ought to be using these symbols. GavinM 1997-05-01
function mps_fix
Name
Summary
The function mps_fix
is the
part of the scanning protocol used to indicate references to the
MPS. It may only be called from within a scanning function.
Associated Protocols
Scanning.
Syntax
mps_res_t mps_fix(mps_ss_t mps_ss, mps_addr_t *ref_io)
Arguments
mps_ss
the scan state argument that was passed to the
scanning function
ref_io
a pointer to a reference within the object
being scanned
Returned Values
Returns a result code, see ERROR HANDLING.
If the reference rank of the object being scanned is not MPS_RANK_AMBIG
then the reference
pointed to by ref_io
may be modified by mps_fix
.
Resources
mps.h
Description
This function is the part of the scanning protocol used to indicate
references. Scanning functions apply it, or MPS_FIX12
, or MPS_FIX1
and MPS_FIX2
tothe references in the object
being scanned.
It may only be called from within a scanning function. If it is
called within a MPS_SCAN_BEGIN
block, MPS_FIX_CALL
must be used (yes,
really).
This function does not perform any particular operation. The MPS
may call scanning functions for a number of reasons, and mps_fix
may take different actions
depending on those reasons.
Example
mps_res_t scan_array(mps_ss_t ss, mps_addr_t object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; for(i = 0; i < length; ++i) { res = mps_fix(ss, &array[i]); if(res != MPS_RES_OK) return res; } return res; }
Error Handling
The function returns MPS_RES_OK
if it was successful, in
which case the scanning function should continue to scan the rest of
the object, applying mps_fix
to
the remaining references. If mps_fix
returns a value other than MPS_RES_OK
, the scanning function must
return that value, and may return without scanning further
references. Generally, it is better if it returns as soon
as possible.
See Also
mps_ss_t
,
mps_root_scan_t
,
mps_fmt_scan_t
,
mps_reg_scan_t
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
macro MPS_FIX1
Name
Summary
The macro MPS_FIX1
is the part
of the scanning protocol used to indicate references to the MPS. It
may only be used from within MPS_SCAN_BEGIN
and MPS_SCAN_END
.
Associated Protocols
Format.
Syntax
MPS_FIX1(mps_ss, ref)
Arguments
mps_ss
the scan state argument
that was passed to the scanning function
ref
a reference within the object being scanned, type
mps_addr_t
Returned Values
Returns a truth value (type mps_bool_t
) indicating whether the
reference is likely to be interesting to the MPS.
Resources
mps.h.
Description
MPS_FIX1
and MPS_FIX2
are a trick to speed up scanning
by splitting MPS_FIX12
into two
macros. MPS_FIX1
is a fast test
to see if the reference is likely to be interesting to the MPS; if it
returns false, the scanner can proceed to the next reference. If it
returns true, the scan method must invoke MPS_FIX2
, which does the actual
fixing.
This macro may only be used in code textually between MPS_SCAN_BEGIN
and MPS_SCAN_END
.
Example
mps_res_t scan_array(mps_ss_t ss, Array object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; MPS_SCAN_BEGIN(ss) for(i = 0; i < length; ++i) { mps_addr_t ref = array[i]; if(MPS_FIX1(ss, ref)) { /* if(((Object*)ref)->type == ScannableType) { */ /* You can do something here, but in the end, you must call MPS_FIX2. */ res = MPS_FIX2(ss, &array[i]); if(res != MPS_RES_OK) return res; /* } */ } } MPS_SCAN_END(ss); return res; }
See Also
MPS_FIX12
,
MPS_FIX2
,
mps_fix
,
mps_ss_t
,
mps_root_scan_t
,
mps_fmt_scan_t
,
mps_reg_scan_t
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX_CALL
macro MPS_FIX12
Name
Summary
The macro MPS_FIX12
is the
part of the scanning protocol used to indicate references to the
MPS. It may only be used from within MPS_SCAN_BEGIN
and MPS_SCAN_END
.
Associated Protocols
Scanning.
Syntax
MPS_FIX12(mps_ss, ref_io);
Arguments
mps_ss
the scan state argument that was passed to the scanning function
ref_io
a pointer to a reference within the object
being scanned, type mps_addr_t *
Returned Values
Returns a result code, see ERROR HANDLING.
If the reference rank of the object being scanned is not MPS_RANK_AMBIG
then the reference
pointed to by ref_io
may be modified by MPS_FIX12
.
Resources
mps.h
Description
This macro is used in the scanning protocol to indicate
references. Scanning functions apply it or mps_fix
or MPS_FIX1
and MPS_FIX2
to the references in the object
being scanned.
It may only be used in code textually between MPS_SCAN_BEGIN
and MPS_SCAN_END
.
It is permitted for the reference (*ref_io
) to point
outside the MPS arena being scanned, or to be NULL; in that case, it
is simply ignored.
This macro does not perform any particular operation. The MPS may
call scanning functions for a number of reasons, and MPS_FIX
may take different actions
depending on those reasons.
Example
mps_res_t scan_array(mps_ss_t ss, mps_addr_t object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; MPS_SCAN_BEGIN(ss) for(i = 0; i < length; ++i) { res = MPS_FIX(ss, &array[i]); if(res != MPS_RES_OK) return res; } MPS_SCAN_END(ss); return res; }
Error Handling
The macro returns MPS_RES_OK
if it was successful, in which case the scanning function should
continue to scan the rest of the object, fixing the remaining
references. If MPS_FIX12
returns
a value other than MPS_RES_OK
,
the scanning function must return that value, and may return without
scanning further references. Generally, it is better if it returns as
soon as possible.
See Also
mps_fix
,
mps_ss_t
,
mps_root_scan_t
,
mps_fmt_scan_t
,
mps_reg_scan_t
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
Notes
MPS_FIX12
is so called, as it basically performs the work of both MPS_FIX1
and MPS_FIX2
.
macro MPS_FIX2
Name
Summary
The macro MPS_FIX2
, together with MPS_FIX1
, is the part of the scanning protocol used to indicate references to the MPS. It may only be used from within MPS_SCAN_BEGIN
and MPS_SCAN_END
.
Associated Protocols
Scanning.
Syntax
MPS_FIX2(mps_ss, ref_io);
Arguments
mps_ss
the scan state argument
that was passed to the scanning function
ref_io
a pointer to a reference within the object
being scanned, type mps_addr_t *
Returned Values
Returns a result code, see ERROR HANDLING.
If the reference rank of the object being scanned is not MPS_RANK_AMBIG
then the reference pointed to by ref_io
may be modified by MPS_FIX2
.
Resources
mps.h.
Description
MPS_FIX1
and MPS_FIX2
are a trick to speed up scanning
by splitting MPS_FIX12
into two
macros. MPS_FIX1
is a fast test
to see if the reference is likely to be interesting to the MPS; if it
returns false, the scanner can proceed to the next reference. If it
returns true, the scan method must invoke MPS_FIX2
, which does the actual
fixing.
This macro may only be used in code textually between MPS_SCAN_BEGIN
and MPS_SCAN_END
.
This macro does not perform any particular operation. The MPS may call scanning functions for a number of reasons, and MPS_FIX2
may take different actions depending on those reasons.
Example
mps_res_t scan_array(mps_ss_t ss, Array object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; MPS_SCAN_BEGIN(ss) for(i = 0; i < length; ++i) { mps_addr_t ref = array[i]; if(MPS_FIX1(ss, ref)) { /* if(((Object*)ref)->type == ScannableType) { */ /* You can do something here, but in the end, you must call MPS_FIX2. */ res = MPS_FIX2(ss, &array[i]); if(res != MPS_RES_OK) return res; /* } */ } } MPS_SCAN_END(ss); return res; }
Error Handling
The macro returns MPS_RES_OK
if it was successful, in which case the scanning function should
continue to scan the rest of the object, fixing the remaining
references. IfMPS_FIX2
returns a
value other than MPS_RES_OK
,
the scanning function must return that value, and may return without
scanning further references. Generally, it is better if it returns as
soon as possible.
See Also
MPS_FIX12
,
MPS_FIX1
,
mps_fix
,
mps_ss_t
,
mps_root_scan_t
,
mps_fmt_scan_t
,
mps_reg_scan_t
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX_CALL
macro MPS_FIX_CALL
Name
Summary
MPS_FIX_CALL
is used to call a scanning function from within MPS_SCAN_BEGIN
and MPS_SCAN_END
.
Associated Protocols
Scanning.
Syntax
MPS_FIX_CALL(ss, call);
Arguments
mps_ss
the scan state argument that was passed to the scanning function
call
an expression (containing a call to a scanning function)
Returned Values
None.
Resources
mps.h.
Description
When using the MPS_SCAN_BEGIN
and MPS_SCAN_END
macros, you can't
directly call a separate function to do part of the scanning, because
between MPS_SCAN_BEGIN
and
MPS_SCAN_END
, the
scan_state
parameter is in a strange state, so you
shouldn't pass it as an argument to a function. However, if you
really want to do it (say, because you have an embedded structure
shared between two scan methods), you can pass the scan state
correctly using MPS_FIX_CALL
.
Note that you must receive the return value of the scanning function called, and pass it on as described in the ERROR HANDLING section.
Example
mps_res_t foo_scan(mps_ss_t scan_state, mps_addr_t base, mps_addr_t limit) { Object *obj; Object *obj_limit; mps_res_t res; obj_limit = limit; MPS_SCAN_BEGIN(scan_state) for(obj = base; obj < obj_limit; obj++) { if(MPS_FIX12(scan_state, &obj->left) != MPS_RES_OK) return res; MPS_FIX_CALL(scan_state, res = scan_data(scan_state, &obj->data)); if(res != MPS_RES_OK) return res; if(MPS_FIX12(scan_state, &obj->right) != MPS_RES_OK) return res; } MPS_SCAN_END(scan_state); return MPS_RES_OK; }
Error Handling
You must receive the return value of the function called. Like all
scanning functions, itreturns MPS_RES_OK
if it was successful, in
which case the caller should continue to scan the rest of the object,
fixing the remaining references. If it returns a value other
thanMPS_RES_OK
, the calling
scanning function must return that value, and may return without
scanning further references. Generally, it is better if it returns as
soon as possible.
See Also
mps_fix
,
mps_ss_t
,
mps_root_scan_t
,
mps_fmt_scan_t
,
mps_reg_scan_t
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
constant MPS_RES_LIMIT
Name
Summary
MPS_RES_LIMIT
is a result
code, indicating that an operation failed because an internal limit
was reached.
Resources
mps.h.
Description
This result code is returned if an operation could not be completed as requested because of an internal limitation of the MPS. The precise meaning depends on the function that returned the code. Refer to the documentation of that function for details.
Example
switch(mps_alloc(&(mps_addr_t)object, pool, size)) { case MPS_RES_LIMIT: bomb("The MPS has reached an internal limit"); break; /* ... */ }
See Also
constant MPS_RES_MEMORY
Name
Summary
MPS_RES_MEMORY
is a result code, indicating that an operation failed because it ran out of memory.
Associated Protocols
All.
Resources
mps.h
Description
This result code is returned if an operation could not be completed because there wasn't enough memory available. You need to deallocate something or allow the garbage collector to reclaim something to free enough memory, or expand the arena (if you're using an arena for which that does not happen automatically).
Note that failing to acquire enough memory because the arena commit
limit would have been exceeded is indicated by returning MPS_RES_COMMIT_LIMIT
, not
MPS_RES_MEMORY
.
Note that running out of address space (as might happen in virtual
memory systems) is indicated by returning MPS_RES_RESOURCE
, not MPS_RES_MEMORY
.
Example
See Also
mps_res_t
,
MPS_RES_RESOURCE
,
MPS_RES_COMMIT_LIMIT
MPS_RES_PARAM
Name
Summary
MPS_RES_PARAM
is a result
code, indicating that an operation failed because an invalid parameter
was specified for the operation.
Associated Protocols
All.
Type
Resources
mps.h.
Description
This result code is returned if an operation could not be completed as requested because an invalid parameter was specified for the operation. The precise meaning depends on the function that returned the code. Refer to the documentation of that function for details.
Example
switch( res = mps_pool_create_v(&pool, arena, class, params) ) { case MPS_RES_PARAM: bomb("Can't make a pool with those specifications"); break; /* ... */ }
See Also
Notes
Internal Notes
MPS_RM_CONST
Name
Summary
MPS_RM_CONST
is a constant used in root mode arguments to indicate constant roots.
Associated Protocols
Root.
Type
Integral constant.
Resources
mps.h
Description
MPS_RM_CONST
is a
preprocessor macro defining a constant that can be OR'ed with other
MPS_RM_*
constants, and passed as the root mode argument
to certain root creation functions (mps_root_create
, mps_root_create_fmt
, mps_root_create_table
,
mps_root_create_table_masked
,
mps_root_create_reg
).
Passing MPS_RM_CONST
means
that the client program will not change the root after it is
declared. I.e., scanning the root will produce the same set of
references every time. Furthermore, for formatted and table roots, the
client program may not write to the root at all.
Example
res = mps_root_create_table(&mmRoot, arena, MPS_RANK_EXACT, MPS_RM_CONST, (mps_addr_t)&Objects, rootCOUNT);
See Also
mps_root_create
,
mps_root_create_fmt
,
mps_root_create_table
,
mps_root_create_table_masked
,
mps_root_create_reg
Notes
Internal Notes
Currently ignored. -- drj 1997-12-18.
MPS_RM_PROT
Name
Summary
MPS_RM_PROT
is a constant used in root mode arguments to indicate protectable roots.
Associated Protocols
Root.
Type
Integral constant.
Resources
mps.h
Description
MPS_RM_PROT
is a
preprocessor macro defining a constant that can be OR'ed with other
MPS_RM_*
contants, and passed as the root mode argument
to certain root creation functions (mps_root_create_fmt
, mps_root_create_table
,
mps_root_create_table_masked
).
Passing MPS_RM_PROT
means
that the MPS may place a hardware write barrier on any pages which any
part of the root covers. Format methods and any scanning function
(except for the one for this root) may not write data in this
root. They may read it.
You mustn't specify MPS_RM_PROT
on a root allocated from
the MPS.
Example
res = mps_root_create_table(&mmRoot, arena, MPS_RANK_EXACT, MPS_RM_PROT, (mps_addr_t)&Objects, rootCOUNT);
See Also
mps_root_create_fmt
,
mps_root_create_table
,
mps_root_create_table_masked
.
Notes
No page may contain parts of two or more roots with MPS_RM_PROT
[how does one prevent
that?]. You mustn't specify MPS_RM_PROT
if the client program or
anything other than (this instance of) the MPS is going to protect or
unprotect the relevant pages.
Internal Notes
Future meaning: The MPS may place a hardware read and/or write
barrier on any pages which any part of the root covers. Format methods
and scanning functions (except for the one for thisroot) may not read
or write data in this root. You may specify MPS_RM_PROT
on a root allocated from
the MPS, as long as it's not from a GCd pool. - drj 1997-12-18
This feature is far too technical for most of our clients: we should think about producing some guidelines on how to use it. - pekka 1998-01-27
There may be problems if the client wants the OS to access the root. Lots of OSes can't cope with writing to protected pages. So we'll need to document that caveat too. drj 1998-05-20
function mps_sac_alloc
Name
Summary
This function allocates a block using the segregated allocation cache given.
Associated Protocols
Allocation cache
Type
mps_res_t mps_sac_alloc(mps_addr_t *p_o, mps_sac_t sac, size_t size, mps_bool_t has_reservoir_permit);
Arguments
p_o a pointer to a variable to hold the address of the new block
sac the segregated allocation cache
size the size of the block requested
has_reservoir_permit
regulates access to the reservoir
Returned Values
If the return value is MPS_RES_OK
, the address of a new block
is *p_o
.
Resources
mps.h
Description
This function allocates a block using the cache given. If no
suitable block exists in the cache, it will ask for more memory from
the associated pool. size
does not have to be one of the
class sizes of the cache; it does not have to be aligned.
The client is responsible for synchronising the access to the cache, but if the cache decides to access the pool, the MPS will properly synchronize with any other threads that might be accessing the same pool.
has_reservoir_permit
regulates whether the pool has
permission to get more memory from the reservoir to satisfy this
request.
Example
void *p; Foo *foo; res = mps_sac_alloc(&p, sac, FooSIZE, is_in_panic); if (res != MPS_RES_OK) { printf("Failed to alloc foo!\n"); exit(1); } foo = p; /* use foo */ mps_sac_free(sac, p, FooSIZE);
Error Handling
mps_sac_alloc
returns
MPS_RES_MEMORY
when it
fails to find enough memory; see the documentation for this return
code for recovery options. It returns MPS_RES_COMMIT_LIMIT
if it
can't allocate without exceeding the arena commit limit; Free
something to make more space or increase the limit using mps_arena_commit_limit_set
. It
returns MPS_RES_RESOURCE
if it has run out of swap space; Free something or terminate other
processes on the same machine.
See Also
MPS_SAC_ALLOC_FAST
,
mps_sac_free
,
MPS_SAC_FREE_FAST
,
mps_sac_t
,
mps_reservoir_limit_set
,
mps_arena_commit_limit_set
,
MPS_RES_MEMORY
,
MPS_RES_COMMIT_LIMIT
,
MPS_RES_RESOURCE
Notes
There's also a macro called MPS_SAC_ALLOC_FAST
, that does
the same thing. The macro is faster, but generates more code and does
less checking.
The block allocated can be larger than requested. Blocks not matching any class size are allocated from the next largest class, and blocks larger than the largest class size are simply allocated at the requested size (rounded up to alignment, as usual).
Objects allocated through a segregated allocation cache should only
be freed through a segregated allocation cache with the same class
structure. Using mps_free
on
them can cause memory leaks, because the size of the block might be
larger than you think. Naturally, the cache must also be attached to
the same pool.
macro MPS_SAC_ALLOC_FAST
Name
Summary
This macro allocates a block using the segregated allocation cache given.
Associated Protocols
Allocation cache
Syntax
MPS_SAC_ALLOC_FAST(res_o, p_o, sac, size, has_reservoir_permit)
Arguments
res_o |
mps_res_t |
an lvalue to hold the result code |
p_o |
mps_addr_t |
an lvalue to hold the address of the new block |
sac |
mps_sac_t |
the segregated allocation cache |
size |
size_t |
the size of the block requested |
has_reservoir_permit |
mps_bool_t |
regulates access to the reservoir |
Returned Values
res_o
will be set to the return code. If this is
MPS_RES_OK
, the address of the
new block is in p_o
.
Resources
mps.h
Description
This macro allocates a block using the cache given. If no suitable
block exists in the cache, it will ask for more memory from the
associated pool. size
does not have to be one of the
class sizes of the cache; it does not have to be aligned.
The client is responsible for synchronizing the access to the cache, but if the cache decides to access the pool, the MPS will properly synchronize with any other threads that might be accessing the same pool.
has_reservoir_permit
regulates whether the pool has
permission to get more memory from the reservoir to satisfy this
request.
Example
void *p; Foo *foo; mps_res_t res; MPS_SAC_ALLOC_FAST(res, p, sac, FooSIZE, is_in_panic); if (res != MPS_RES_OK) { printf("Failed to alloc foo!\n"); exit(1); } foo = p; /* use foo */ MPS_SAC_FREE_FAST(sac, p, FooSIZE);
Error Handling
MPS_SAC_ALLOC_FAST
returns MPS_RES_MEMORY
when
it fails to find enough memory; see the documentation for this return
code for recovery options. It returns MPS_RES_COMMIT_LIMIT
if it
can't allocate without exceeding the arena commit limit; free something
to make more space or increase the limit using mps_arena_commit_limit_set
. It
returns MPS_RES_RESOURCE
if it has run out of swap space; free something or terminate other
processes on the same machine.
See Also
mps_sac_alloc
,
MPS_SAC_FREE_FAST
,
mps_sac_free
,
mps_sac_t
,
mps_reservoir_limit_set
,
mps_arena_commit_limit_set
,
MPS_RES_MEMORY
,
MPS_RES_COMMIT_LIMIT
,
MPS_RES_RESOURCE
Notes
There's also a function called mps_sac_alloc
, that does the same
thing.
The block allocated can be larger than requested. Blocks not matching any class size are allocated from the next largest class, and blocks larger than the largest class size are simply allocated at the requested size (rounded up to alignment, as usual).
Objects allocated through a segregated allocation cache should only
be freed through a segregated allocation cache with the same class
structure. Using mps_free
on them
can cause memory leaks or assertions, because the size of the block
might be larger than you think. Naturally, the cache must also be
attached to the same pool.
The macro doesn't evaluate has_reservoir_permit
,
unless it decides to access the pool.
constant MPS_SAC_CLASS_LIMIT
Name
Summary
MPS_SAC_CLASS_LIMIT
specifies how many classes mps_sac_create
is guaranteed to accept.
Type
size_t
Associated Protocols
Allocation cache
Resources
mps.h
Description
MPS_SAC_CLASS_LIMIT
specifies a lower limit on the maximum number of classesthat can be described in a call to mps_sac_create
, i.e., the MPS guarantees to acceptat least this many classes. More might be accepted -- in fact, there might not be any limit in theimplementation on the maximum number of classes, but if you specify more than this, you should beprepared to handle the error.
MPS_SAC_CLASS_LIMIT
is a macro suitable for use in a constant expression, bothin a #if directive and wherever else constant expressions may be used.
Example
mps_sac_t sac; mps_sac_class_s classes[3] = { {8, 38, 1}, {136, 19, 3}, {512, 4, 1} }; #if (MPS_SAC_CLASS_LIMIT < 3) # error "Too many classes!" #endif res = mps_sac_create(&sac, pool, 3, classes); if (res != MPS_RES_OK) { printf("Failed to create the allocation cache!"); exit(1); }
See Also
Notes
If you ask for too many size classes, mps_sac_create
returns MPS_RES_LIMIT
; you can recover by combining some small adjacent classes.
function mps_sac_free
Name
Summary
This function frees an object using the segregated allocation cache given.
Associated Protocols
Allocation cache
Type
void mps_sac_free(mps_sac_t sac, mps_addr_t p, size_t size);
Arguments
sac the segregated allocation cache
p a pointer to the block being freed
size the size of the block being freed
Returned Values
None.
Resources
mps.h
Description
This function frees an object using the cache given. If the cache would become too full,some blocks are returned to the associated pool.
size
should be the size that wasspecified when the object was allocated (the cache knows what the real size of the block is). Theobject must have been allocated through a segregated allocation cache with the same class structure,attached to the same pool.
The client is responsible for synchronising the access to the cache, but if the cachedecides to access the pool, the MPS will properly synchronize with any other threads that might beaccessing the same pool.
Example
void *p; Foo *foo; res = mps_sac_alloc(&p, sac, FooSIZE, is_in_panic); if (res != MPS_RES_OK) { printf("Failed to alloc foo!\n"); exit(1); } foo = p; /* use foo */ mps_sac_free(sac, p, FooSIZE);
See Also
MPS_SAC_FREE_FAST
,
mps_sac_alloc
,
MPS_SAC_ALLOC_FAST
,
mps_sac_t
Notes
Usually, you'd use the same cache to allocate and deallocate an object.
There's also a macro called MPS_SAC_FREE_FAST
, that does the same thing. The macro is faster, but generates more code and does no checking.
Note that mps_sac_free
does very little checking; it's optimized for speed.Double frees and other mistakes will only be detected when the cache is flushed (which can happen bydemand through mps_sac_flush
or automatically), unless intervening operations have obscured symptom.
macro MPS_SAC_FREE_FAST
Name
Summary
MPS_SAC_FREE_FAST
frees an object using the segregated allocation cache given.
Associated Protocols
Allocation cache
Syntax
MPS_SAC_FREE_FAST(sac, p, size)
Arguments
sac
|
mps_sac_t
|
the segregated allocation cache |
p
|
mps_addt_t
|
the address of the object to be freed |
size
|
size_t
|
the size of the object |
Returned Values
None.
Resources
mps.h
Description
This macro frees an object using the cache given. If the cache would become too full, someblocks are returned to the associated pool.
size
should be the size that was specifiedwhen the object was allocated (the cache knows what the real size of the block is). The objects musthave been allocated through a segregated allocation cache with the same class structure, attachedto the same pool.
The client is responsible for synchronizing the access to the cache, but if the cachedecides to access the pool, the MPS will properly synchronize with any other threads that might beaccessing the same pool.
Example
void *p; Foo *foo; mps_res_t res; MPS_SAC_ALLOC_FAST(res, p, sac, FooSIZE, is_in_panic); if (res != MPS_RES_OK) { printf("Failed to alloc foo!\n"); exit(1); } foo = p; /* use foo */ MPS_SAC_FREE_FAST(sac, p, FooSIZE);
See Also
mps_sac_free
,
MPS_SAC_ALLOC_FAST
,
mps_sac_alloc
,
mps_sac_t
Notes
Usually, you'd use the same cache to allocate and deallocate an object.
There's also a function called mps_sac_free
, that does the same thing. Themacro is faster, but generates more code and does no checking.
Note that MPS_SAC_FREE_FAST
doesn't do any checking; it's optimized for speed. Double frees and other mistakes will only be detected when the cache is flushed (which can happen by demand through mps_sac_flush
or automatically), unless intervening operations have obscured symptom.
macro MPS_SCAN_BEGIN
Name
Summary
The macro MPS_SCAN_BEGIN
is part of the scanning protocol; together with MPS_SCAN_END
, it sets up local information used by MPS_FIX*
.
Associated Protocols
Scanning.
Syntax
MPS_SCAN_BEGIN(ss)
Arguments
ss the scan state argument that was passed to the scanning function
Resources
mps.h
Description
This macro is used in the scanning protocol. Together with MPS_SCAN_END
, it sets up local information used by the fast MPS_FIX*
macros.
Example
mps_res_t scan_array(mps_ss_t ss, mps_addr_t object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; MPS_SCAN_BEGIN(ss) for(i = 0; i < length; ++i) { res = MPS_FIX12(ss, &array[i]); if(res != MPS_RES_OK) return res; } MPS_SCAN_END(ss); return res; }
See Also
MPS_SCAN_END
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
Notes
Between MPS_SCAN_BEGIN
and MPS_SCAN_END
, you may not call another scanning functiondirectly, because the scan state parameter is in a strange state, so you shouldn't pass it as anargument to a function. However, you can pass the scan state using MPS_FIX_CALL
. You also cannotnest MPS_SCAN_BEGIN
textually within another MPS_SCAN_BEGIN
-- MPS_SCAN_END
pair.
macro MPS_SCAN_END
Name
Summary
The macro MPS_SCAN_END
is part of the scanning protocol; it terminates a blockstarted by MPS_SCAN_BEGIN
.
Associated Protocols
Scanning.
Syntax
MPS_SCAN_END(ss);
Arguments
ss the scan state argument that was passed to the scanning function
Resources
mps.h
Description
This macro is used in the scanning protocol. Together with MPS_SCAN_BEGIN
, itsets up local information used by the fast MPS_FIX*
macros. Note that MPS_SCAN_END
completes the scanning, so successful termination of the scanning mustinvoke it (error branches can return without passing through).
Example
mps_res_t scan_array(mps_ss_t ss, mps_addr_t object, size_t length) { size_t i; mps_res_t res; mps_addr_t *array = (mps_addr_t *)object; MPS_SCAN_BEGIN(ss) for(i = 0; i < length; ++i) { res = MPS_FIX12(ss, &array[i]); if(res != MPS_RES_OK) return res; } MPS_SCAN_END(ss); return res; }
See Also
MPS_SCAN_BEGIN
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
Notes
Between MPS_SCAN_BEGIN
and MPS_SCAN_END
, you may not call another scanning functiondirectly, because the scan state parameter is in a strange state, so you shouldn't pass it as anargument to a function. However, you can pass the scan state using MPS_FIX_CALL
.
type MPS_T_WORD
Name
Summary
MPS_T_WORD
an unsigned integral type that is the same size as an object pointer.
Resources
mpstd.h.
Description
MPS_T_WORD
is a preprocessor macro defined in "mpstd.h". It is the name of an unsignedintegral type that is the same size as an object pointer (so sizeof(MPS_T_WORD) == sizeof(void*)
).The exact identity of the type is platform-dependent.
Example
See Also
MPS_WORD_SHIFT
,
MPS_WORD_WIDTH
constant MPS_WORD_SHIFT
Name
Summary
MPS_WORD_SHIFT
is log base 2 of MPS_WORD_WIDTH
.
Type
Integral constant.
Resources
mpstd.h.
Description
MPS_WORD_SHIFT
is a preprocessor macro defined in "mpstd.h". It is the logarithm in base 2 of MPS_WORD_WIDTH
(so 1 << MPS_WORD_SHIFT == MPS_WORD_WIDTH
). The value of MPS_WORD_SHIFT
is platform-dependent. Typical values are 5 and 6.
Example
See Also
constant MPS_WORD_WIDTH
Name
Summary
MPS_WORD_WIDTH
is the width in bits of the type MPS_T_WORD
.
Type
Integral constant.
Resources
mpstd.h.
Description
MPS_WORD_WIDTH
is a preprocessor macro defined in "mpstd.h" to be the width in bits of the type MPS_T_WORD
(so MPS_WORD_WIDTH == sizeof(MPS_T_WORD) * CHAR_BIT
).
This value is required for the use of the MPS C interface and the interpretation of "mps.h".It is platform-dependent. It is a power of 2; typical values are 32 and 64. It may be defined byincluding "mpstd.h" on a supported platform, or by defining it to be the width of MPS_T_WORD
inbits.
Example
#define MPS_WORD_WIDTH 32
See Also
type mps_addr_t
Name
Summary
mps_addr_t
is the type of addresses managed by the MPS, and also the type of references too bjects.
Associated Protocols
Allocation, allocation point, format, root, location dependency.
Type
typedef void *mps_addr_t;
Resources
mps.h
Description
mps_addr_t
is the type of addresses managed by the MPS, and also the type of referencesto objects. It is used in the MPS C interface where the MPS needs to pass a pointer to memory thatis under the control of the MPS.
In accordance with standard C practice, the value NULL of type mps_addr_t
will never beused to represent the address of an object.
Example
{ mps_addr_t new_block; mps_res_t res; thingy *tp; res = mps_pool_alloc(&new_block, pool, sizeof(thingy)); if(res != MPS_RES_OK) return res; tp = new_block; /* ... */ }
Error Handling
Not applicable.
See Also
type mps_align_t
Name
Summary
mps_align_t
is the type of an alignment.
Associated Protocols
Format, pool, allocation.
Type
typedef size_t mps_align_t;
Resources
mps.h
Description
An alignment specifies the address modulus to which all objects in an object format must bealigned. That is, if an alignment of 4 is specified, then the address of any object in that formatmodulo 4 will always be 0.
mps_align_t
is a transparent type equivalent to the C type "size_t" and must be a positivepower of 2.
Some pools and allocation protocols accept an alignment as an option that can be used toensure that objects in the pool or objects allocated observe a stricter alignment than that of theobject format.
Example
mps_align_t floatAlign = 4; mps_align_t doubleFloatAlign = 8;
See Also
function mps_alloc
Name
Summary
mps_alloc
allocates a block of memory in a pool.
Associated Protocols
Allocation.
Syntax
mps_res_t mps_alloc(mps_addr_t *p, mps_pool_t pool, size_t size, ...);
Arguments
p output parameter for a pointer to the block allocated
pool the pool to allocate in
size the size of the block to allocate in bytes
... (some pools can take additional arguments)
Returned Values
A return code.
Resources
mps.h
Description
mps_alloc
allocates a block of memory in the given pool.
Example
mps_res_t res; mps_addr_t p; res = mps_alloc(&p, pool, size); if(res != MPS_RES_OK) { /* p hasn't been touched in this case. */ handle error; } /* p now contains the result, which is the address of the new block */ /* in this case. */
Error Handling
See Also
mps_alloc_pattern_ramp
Name
Summary
Returns a allocation pattern type indicating that allocation will follow a ramp pattern.
Associated Protocols
alloc-pattern-ramp
Syntax
mps_alloc_pattern_t mps_alloc_pattern_ramp();
Arguments
none
Returned Values
Returns the allocation pattern type for ramps.
Description
When declaring an allocation pattern for an AP, if the calls to mps_ap_alloc_pattern_begin
use ramp allocation patterns (such as the result of mps_alloc_pattern_ramp
), then the MPS willtake this as an indication that most of the objects allocated after the call to mps_ap_alloc_pattern_begin
are likely to be dead by the corresponding call to mps_ap_alloc_pattern_end
.
This permits the client to indicate useful points for GC with minimal perturbation of the GCstrategy.
Example
{ mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); do_lots_of_work(); mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); }
Error Handling
Cannot fail.
See Also
mps_alloc_pattern_ramp_collect_all
,
mps_ap_alloc_pattern_begin
Notes
mps_alloc_pattern_ramp_collect_all
Name
mps_alloc_pattern_ramp_collect_all
Summary
Returns a ramp allocation pattern type indicating that a full GC should be done.
Associated Protocols
GC, AP, ramps
Syntax
mps_alloc_pattern_t mps_alloc_pattern_ramp_collect_all();
Arguments
none
Returned Values
Returns the allocation pattern type for full collection ramps.
Description
This yields an allocation pattern for an AP that is similar to that returned by mps_alloc_pattern_ramp
, in that it declares a ramp allocation pattern, but additionally indicates tothe MPS that the next collection following the ramp should be a full GC.
This permits the client to indicate useful points for a full GC, either because most of theheap is likely to be dead, or because accurate statistics are required, with minimal perturbation ofthe GC strategy.
As usual, this allocation pattern should be used in matching mps_ap_alloc_pattern_begin
and mps_ap_alloc_pattern_end
pairs. It may nest with, but should not otherwise overlap with allocationpatterns of type mps_alloc_pattern_ramp
. In this case, the MPS may defer the full GC until after allramp allocation patterns have ended.
Example
{ mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp_collect_all()); do_lots_of_work(); mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp_collect_all()); wait_for_collection_statistics_while_doing_other_allocation(); }
Error Handling
Cannot fail.
See Also
mps_alloc_pattern_ramp
,
mps_ap_alloc_pattern_begin
Notes
Internal Notes
mps_amc_apply
Name
Summary
mps_amc_apply
is used to inspect objects in an AMC pool. You may only call it when the
arena is parked (for example, after mps_arena_collect
).
Associated Protocols
Arena.
Syntax
extern void mps_amc_apply(mps_pool_t, void(*f)(mps_addr_t, void *, size_t), void *, size_t);
Arguments
mps_pool_t
the pool whose objects you want to inpect
f a supplied function
mps_addr_t
the address of the object you want to inspect
*
size_t
Returned Values
Not applicable.
Resources
Not applicable.
Description
mps_amc_apply
is used to inspect objects in an AMC pool. You may only call it when thearena is parked (for example, after mps_arena_collect
). When called, mps_amc_apply
calls thesupplied function "f" once for each object in the pool, with the address of the object as its firstargument. "f" is given as its second and third arguments whatever values were given as the third andfourth arguments to mps_amc_apply
. (This is intended to make it easy to pass, for example, anarray and its size as parameters.)
"f" will be called on both data and pad objects, and it is "f"'s job to distinguish, ifrequired, between the two. Note (c.f. mps_arena_collect
, above) that it may be called onunreachable objects that the collector has not recycled or has not been able to recycle.
The function "f" may not allocate memory or access any automatically-managed memory exceptthe object at which it is pointed and, in the case of objects in Dylan Container Format, thatobject's wrapper.
Example
Error Handling
See Also
Functions that park the arena:
mps_arena_park
,
mps_arena_collect
.
A more general heap walker which inspects all formatted objects in
the arena:
mps_arena_formatted_objects_walk
Notes
There is no equivalent function for other pool classes, but there is a more general heapwalker: mps_arena_formatted_objects_walk
.
Internal Notes
Does "You must call it when the arena is parked" mean that (a) parking an arena requires that you call this function, or (b) you can only call this function when the arena is in the parked state? LMB
(b). Changed "must call" to "may only call" drj 1998-08-25
mps_ap_alloc_pattern_begin
Name
Summary
Indicates the start of allocation following a particular pattern.
Associated Protocols
AP, Allocation, Alloc-pattern-ramp.
Syntax
mps_res_t mps_ap_alloc_pattern_begin(mps_ap_t ap, mps_alloc_pattern_t alloc_pattern)
Arguments
ap The allocation point in which the patterned allocation will occur
alloc_pattern The pattern of the allocation
Returned Values
Result code indicating whether start of the allocation pattern was successfully registered.
Resources
mps.h
Description
This function is used, together with mps_ap_alloc_pattern_end
, to indicate periods ofallocation in an allocation point that follow some pattern of lifetime.
The nesting/overlapping restrictions on allocation patterns may vary depending on theparticular allocation pattern type, but in general, if mps_ap_alloc_pattern_begin
is used multipletimes on the same allocation point without intervening calls to mps_ap_alloc_pattern_end
, the callsmatch in a stack-like way, outermost and innermost; that is, allocation patterns may nest, but nototherwise overlap.
Example
{ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); do_some_work(); /* Leaves stuff lying around */ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); do_some_more_work(); /* Tidies up after itself */ res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); tidy_up_first_work(); res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); }
Error Handling
Currently doesn't fail, but may in future if certain allocation patterns are inappropriatefor that allocation point at that point in time.
See Also
mps_alloc_pattern_ramp
,
mps_alloc_pattern_ramp_collect_all
,
mps_ap_alloc_pattern_end
,
mps_ap_alloc_pattern_reset
mps_ap_alloc_pattern_end
Name
Summary
Indicates the end of allocation following a particular pattern.
Associated Protocols
AP, Allocation, Alloc-pattern-ramp.
Syntax
mps_res_t mps_ap_alloc_pattern_end(mps_ap_t ap, mps_alloc_pattern_t alloc_pattern)
Arguments
ap The allocation point in which the patterned allocation occurred
alloc_pattern The pattern of the allocation
Returned Values
Result code indicating whether end of the allocation pattern was successfully registered.
Description
This function is used, together with mps_ap_alloc_pattern_begin
, to indicate periods of allocation in an allocation point that follow some pattern of lifetime.
Example
{ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); do_some_work(); /* Leaves stuff lying around */ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); do_some_more_work(); /* Tidies up after itself */ res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); tidy_up_first_work(); res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); }
Error Handling
Will fail if there is no extant allocation pattern of that type. May fail in future ifcertain allocation patterns are inappropriate for that allocation point at that point in time.
See Also
mps_alloc_pattern_ramp
,
mps_alloc_pattern_ramp_collect_all
,
mps_ap_alloc_pattern_begin
,
mps_ap_alloc_pattern_reset
function mps_ap_alloc_pattern_reset
Name
Summary
Indicates the end of allocation pattern on an allocation point.
Associated Protocols
AP, Alloc-pattern-ramp
Syntax
mps_res_t mps_ap_alloc_pattern_reset(mps_ap_t ap);
Arguments
ap The allocation point in which the patterned allocation occurred
Returned Values
Result code indicating whether end of the allocation patterns was successfully registered.
Description
This function may be used in place of mps_ap_alloc_pattern_end
to end all extant allocationpatterns on an allocation point. It is anticipated that this may be used to recover from errorconditions.
Example
{ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); do_some_work(); /* Leaves stuff lying around */ res = mps_ap_alloc_pattern_begin(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); res = do_some_more_work(); /* Tidies up after itself */ if(res != mps_res_ok) { res = mps_ap_alloc_pattern_reset(ap); assert(res == mps_res_ok); return; } res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); tidy_up_first_work(); res = mps_ap_alloc_pattern_end(ap, mps_alloc_pattern_ramp()); assert(res == mps_res_ok); }
Error Handling
Cannot fail at present. May fail in future if certain allocation patterns cannot be endedfor that allocation point at that point in time.
See Also
mps_alloc_pattern_ramp
,
mps_alloc_pattern_ramp_collect_all
,
mps_ap_alloc_pattern_begin
,
mps_ap_alloc_pattern_end
function mps_ap_frame_pop
Name
Summary
Declares that a set of objects in a particular frame are dead or likely to be dead.
Associated Protocols
AP Stack Protocol
Type
mps_res_t (mps_ap_frame_pop)(mps_ap_t /* ap */, mps_frame_t /* frame */)
Arguments
mps_ap_t ap
The allocation point in which the frame was pushed.
frame
The frame.
Returned Values
A result code in the usual way.
Description
This function pops the specified frame making its parent the current frame (frames areimplicitly created using the push operation, see mps_ap_frame_push
). Poppinginvalidates the specified frame and all frames pushed since the specified frame. Popping the framemakes a declaration about the set of objects which were allocated in the specified frame and alsoall frames which were pushed since the specified frame. It can be used to declare a set of objectsdead or likely to be mostly dead; the exact interpretation of the declaration depends on pool classthat the allocation point is in (the same pool class that that objects are in). Typically poolclasses which are mostly manually managed will use this declaration to mean that the objects aredead and their space can be reclaimed immediately, whereas pool classes which are mostlyautomatically managed will use this declaration to mean that the objects are likely to be mostlydead (the pool class may use this declaration to alter its collection decisions). Consult the poolclass documentation for details.
In general a frame other than the current frame can be popped (all frames pushed morerecently will be invalidated as well, as described above), but a particular pool class may imposethe restriction that only the current frame may be popped. This restriction means that every pushmust have a corresponding pop. Consult the pool class documentation for details.
It is illegal to pass invalid frames to any MPS function. In particular it is not possibleto pop frames out of order (so the sequence "A = push, B = push, pop A, pop B" is illegal) or to popto the same frame twice (so the sequence "A = push, pop A, pop A" is illegal).
More comprehensive documentation is available in the protocol document (AP Stack Protocol).
Example
<example of how to use the symbol>
Error Handling
<how the client program should handle errors that the symbol returns, if applicable>
See Also
mps.protocol.alloc-point.stack,
mps_ap_frame_push
Notes
None.
function mps_ap_frame_push
Name
Summary
Declares a new frame as part of the AP stack protocol.
Associated Protocols
AP Stack Protocol
Type
mps_res_t (mps_ap_frame_push)(mps_frame_t * /* frameReturn */, mps_ap_t /* ap */);
Arguments
mps_frame_t *frameReturn
The frame return parameter. A new frame (declared by this function) is stored in thislocation if this function is successful.
mps_ap_t ap
The allocation point in which the new frame is declared.
Returned Values
A result code in the usual way. The creation of new frame objects (which is implicit in theaction of this function) can consume resources, so this function can fail because there areinsufficient resources. This function may fail if the correct protocol is not followed by theclient.
Description
This function declares a new frame in the specified allocation point, makes that new frame achild of the current frame, changes the current frame to be the newly created frame, and returns ahandle to the frame. Frames have two important features: A single frame identifies a set of objects(those objects that are "allocated in the frame") which can be destroyed (or declared dead) in a popoperation (see mps_ap_frame_pop
); They are arranged in a partially ordered sequence (this isimportant when the pop operation is used). A fuller and more useful description is found in the APstack protocol document (protocol.mps.alloc-point.stack).
Example
[missing]
Error Handling
Errors can either be because the client hasn't followed the correct protocol in which casethere isn't much that we can recommend or else because some needed resource isn't available. Theusual course of actions when short of resources is recommended.
See Also
mps_ap_frame_pop
,
protocol.mps.alloc-point.stack
Notes
function mps_arena_clamp
Name
Summary
mps_arena_clamp
puts the specified arena into the clamped state.
Associated Protocols
Arena.
Syntax
extern void mps_arena_clamp(mps_arena_t);
Arguments
arena -- the arena to be put into the clamped state
Returned Values
None.
Resources
mps.h
Description
mps_arena_clamp
puts the specified arena into the clamped state. In the clamped state, no
object motion will occur and the staleness of location dependencies will not change. All references
to objects loaded while the arena is clamped will keep the same binary representation until after itis released.
In a clamped arena, incremental collection may still occur, but it will not be visible to the mutator and no new collections will begin. Space used by unreachable objects will not be recycled until the arena becomes unclamped.
Example
Error Handling
See Also
mps_arena_park
,
mps_arena_release
Notes
function mps_arena_class_cl
Name
Summary
mps_arena_class_cl
returns the client arena class.
Associated Protocols
Arena.
Type
mps_arena_class_t mps_arena_class_cl(void)
Arguments
None.
Returned Values
Returns the client arena class.
Resources
mpsacl.h
Description
This function is used to get hold of the client arena class, for the purpose of passing itto mps_arena_create
.
Example
mps_arena_t arena; int main(void) { void *block; mps_res_t res; block = malloc(ARENA_SIZE); if(block == NULL) { printf("Not enough memory!"); exit(1); } res = mps_arena_create(&arena, mps_arena_class_cl(), ARENA_SIZE, block); if(res != MPS_RES_OK) { printf("ARENA_SIZE too small"); exit(2); } /* rest of program */ }
Error Handling
None.
See Also
Notes
A client arena gets its managed memory from the client. This memory block is passed when thearena is created. When creating a client arena, mps_arena_create
takes two extraarguments:
mps_res_t mps_arena_create(mps_arena_t *mps_arena_o, mps_arena_class_t mps_arena_class_cl,size_t size, void *block)
block
is the address of the memory block managed by the arena, andsize
is its size in bytes. If mps_arena_create
returnsMPS_RES_MEMORY
, then the block was too small to hold the internal arena structures.Allocate a (much) larger one, and try again. mps_arena_create
returnsMPS_RES_FAIL
, if the MPS library is copy-protected by a security device, such as adongle, and a valid security device cannot be found.
type mps_arena_class_t
Name
Summary
"m ps_arena_class_t " is the type of arena classes.
Associated Protocols
Arena.
Type
typedef struct mps_arena_s *mps_arena_t;
mps_arena_class_s
is an incomplete structure type used only to declare the opaque type mps_arena_class_t
.
Resources
mps.h
Description
mps_arena_class_t
is the type of arena classes. It is opaque.
Example
The definition of the client arena class in the "mpsacl.h" header:
extern mps_arena_class_t mps_arena_class_cl(void);
See Also
Notes
None.
function mps_arena_class_vm
Name
Summary
mps_arena_class_vm
returns the virtual memory arena class.
Associated Protocols
Arena.
Syntax
mps_arena_class_t mps_arena_class_vm(void)
Arguments
None.
Returned Values
Returns the virtual memory arena class.
Resources
mpsavm.h
Description
This function is used to get hold of the virtual memory arena class, for the purpose ofpassing it to mps_arena_create
. The VM arenas use the OS virtual memory interfaces toallocate memory. The chief consequence of this is that the arena can manage many more virtualaddresses than it needs to commit memory to. This gives it flexibility as to where to place objects,which reduces fragmentation and helps make garbage collection more efficient.
This class is similar to mps_arena_class_vmnz
but uses a more complex placementpolicy, which is more suited to copying garbage collection.
Example
mps_arena_t arena; int main(void) { mps_res_t res; res = mps_arena_create(&arena, mps_arena_class_vm(), ARENA_SIZE); if(res != MPS_RES_OK) { printf("Not enough memory!"); exit(1); } /* rest of program */ }
Error Handling
None.
See Also
mps_arena_create
,
mps_arena_class_vmnz
Notes
A virtual memory arena gets its managed memory from the operating system's virtual memoryservices. An initial address space size is passed when the arena is created. When creating a virtualmemory arena, mps_arena_create
takes one extra argument:
mps_res_t mps_arena_create(mps_arena_t *arena_o, mps_arena_class_t arena_class_vm, size_t size)
size
is the initial amount of virtual address space, in bytes, that the arenawill reserve (this space is initially reserved so that the arena can subsequently use it withoutinterference from other parts of the program, but most of it is not committed, so it don't requireany RAM or backing store). The arena may allocate more virtual address space beyond this initialreservation as and when it deems it necessary. The MPS is most efficient if you reserve an addressspace that is several times larger than your peak memory usage.
mps_arena_create
returns MPS_RES_RESOURCE
if it fails to reserveadequate address space to place the arena in; possibly other parts of the program are reserving toomuch virtual memory. It returns MPS_RES_MEMORY
when it fails to allocate memory for theinternal arena structures; either size
was far too small or you ran out of swap space.It returns MPS_RES_FAIL
, if the library is copy-protected by a security device, suchas a dongle, and a valid security device cannot be found.
Virtual memory arenas are not available on the Mac platforms, other than MacOS X. You willget a linking error, if you attempt to use this function.
function mps_arena_class_vmnz
Name
Summary
An arena class like mps_arena_class_vm
but with a different placement policy.
Associated Protocols
Arena.
Syntax
mps_arena_class_t mps_arena_class_vmnz(void);
Arguments
None.
Returned Values
Returns the VMNZ arena class.
Resources
mpsavm.h
Description
Returns the VMNZ arena class (stands for Virtual Memory No Zones, if you really care.) Thisclass can be passed to mps_arena_create
in order to create a VMNZ arena. The VMNZarenas use the OS virtual memory interfaces to allocate memory. The chief consequence of this isthat the arena can manage many more virtual addresses than it needs to commit memory to. This givesit flexibility as to where to place objects.
This class is similar to mps_arena_class_vm
but uses a simpler placementpolicy, that makes it slightly faster.
Example
mps_arena_t arena; int main(void) { mps_res_t res; res = mps_arena_create(&arena, mps_arena_class_vmnz(), ARENA_SIZE); if(res != MPS_RES_OK) { printf("Not enough memory!"); exit(1); } /* rest of program */ }
Error Handling
No errors.
See Also
mps_arena_create
,
mps_arena_class_vm
Notes
This class takes an extra argument when used in mps_arena_create
(see example).The extra parameter should be of type size_t
. It specifies the amount of virtualaddress space, in bytes, that this arena should use. The arena will reserve this amount of virtualaddress space from the OS during initialization. It will not subsequently use any more address space(compare with mps_arena_class_vm
which can grow).
mps_arena_create
returns MPS_RES_RESOURCE
if it fails to reserveadequate address space to place the arena in; possibly other parts of the program are reserving toomuch virtual memory. It returns MPS_RES_MEMORY
when it fails to allocate memory for theinternal arena structures; either size
was far too small or you ran out of swap space.It returns MPS_RES_FAIL
, if the library is copy-protected by a security device, suchas a dongle, and a valid security device cannot be found.
Virtual memory arenas are not available on the Mac platforms, other than MacOS X. You willget a linking error, if you attempt to use this function.
function mps_arena_collect
Name
Summary
mps_arena_collect
collects the arena and puts it in the parked state.
Associated Protocols
Arena.
Syntax
void mps_arena_collect(mps_arena_t arena);
Arguments
arena the arena to collect
Resources
mps.h
Description
mps_arena_collect
collects the arena and puts it in the parked state. Collecting the arenaattempts to recycle as many unreachable objects as possible and reduce the size of the arena as muchas possible (though in some cases it may increase because it becomes more fragmented). If you do notwant the arena to be in the parked state, you must explicitly call mps_arena_release
aftermps_arena_collect
.
Note that the collector may not be able to recycle some objects (such as those near thedestination of ambiguous references) even though they are not reachable.
Example
[missing]
Error Handling
No errors.
See Also
mps_arena_park
,
mps_arena_release
Notes
None.
function mps_arena_commit_limit
Name
Summary
Returns the current commit limit associated with the arena in bytes.
Associated Protocols
Arena
Type
size_t mps_arena_commit_limit(mps_arena_t arena)
Arguments
arena -- the arena
Returned Values
Returns the current commit limit as a number of bytes in a size_t
Resources
mps.h
Description
Returns the current commit limit associated with the arena in bytes. The commit limit can bechanged using the function mps_commit_limit_set
. The commit limit is used to control how much memorythe MPS can obtain from the OS. See Arena Protocol for details.
Example
limit = mps_arena_commit_limit(arena);
Error Handling
No errors.
See Also
mps_arena_committed
,
mps_arena_commit_limit_set
Notes
None.
function mps_arena_commit_limit_set
Name
Summary
Changes the current commit limit associated with the arena.
Associated Protocols
Arena
Type
mps_res_t mps_arena_commit_limit_set(mps_arena_t arena, size_t limit)
Arguments
arena -- the arena
limit -- the new commit limit in bytes
Returned Values
Returns a result code.
Resources
mps.h
Description
The commit limit of the arena is set to the limit given. The commit limit controls how muchmemory the MPS will obtain from the OS. See Arena Protocol for details. The commit limit cannot beset to a value that is lower than the number of bytes that the MPS is using. If an attempt is madeto set the commit limit to a value greater than or equal to that returned bymps_arena_committed
then it will succeed. If an attempt is made to set the commit limitto a value less than that returned by mps_arena_committed
then it will succeed only ifthe amount committed by the MPS can be reduced by reducing the amount of spare committed memory; insuch a case the spare committed memory will be reduced appropriately and the attempt will succeed.
Example
do { res = mps_arena_commit_limit_set(arena, limit - 100 * 1024); if(res != MPS_RES_OK) flush_caches(); } while(res != MPS_RES_OK);
Error Handling
Returns MPS_RES_OK
when successful, and some other result code when not.
See Also
mps_arena_committed
,
mps_arena_commit_limit
,
mps_arena_spare_commit_limit_set
Notes
mps_arena_commit_limit_set
puts a limit on all memory committed by the MPS. The"spare committed" memory can be limited separately with mps_arena_spare_commit_limit_set
. Note that "spare committed" memory is subject toboth limits; there cannot be more spare committed memory than the spare commit limit, and therecan't be so much spare committed memory that there is more committed memory than the commit limit.
Function mps_arena_committed
Name
Summary
mps_arena_committed
returns the amount of memory (backing store) in use by the arena, bothfor storing client objects and for its own data structures.
Associated Protocols
Arena.
Syntax
extern size_t mps_arena_committed(mps_arena_t arena)
Arguments
arena -- the arena
Returned Values
Returns a number of bytes (the amount of committed memory) as a size_t.
Resources
mps.h
Description
mps_arena_committed
returns the amount of memory (backing store) in use by the arena (alsoknown as "committed memory"). The value returned is a number of bytes.
Committed memory may be used both for storing client objects and for storing MPS datastructures. In addition the MPS maintains committed memory which is not being used (for either ofthe above purposes). This memory is known as "spare committed" memory (see mps_arena_spare_committed
). The amount of "spare committed" memory can change atany time, in particular in will be reduced as appropriate in order meet client requests.
The reasons that the committed memory (as return by this function) might be large than thesum of the sizes of client allocated objects are:
some memory is used internally by the MPS to manage its own data structures and to recordinformation about client objects (such as free lists, page tables, colour tables, statistics, etc).
operating systems (and hardware) typically restrict programs to requesting and releasingmemory with a certain granularity (for example, pages), so extra memory is committed when thisrounding is necessary.
there might be "spare committed" memory.
The amount of committed memory is a good measure of how much virtual memory resource ("swapspace") the MPS is using from the OS.
This function may be called whether the arena is unclamped, clamped or parked, if calledwhen the arena in unclamped then the value may change after this function returns. A possible usemight be to call it just after mps_arena_collect
to (over-)estimate the size of the heap.
If you want to know how much memory the MPS is using then you're probably interested in the value mps_arena_committed() - mps_arena_spare_committed()
.
The amount of committed memory can be limited with the function mps_arena_commit_limit
.
Example
Error Handling
See Also
mps_arena_collect
,
mps_arena_clamp
,
mps_arena_park
,
mps_arena_release
Notes
-
function mps_arena_create
Name
Summary
mps_arena_create
is used to create an arena.
Associated Protocols
Arena.
Syntax
mps_res_t mps_arena_create(mps_arena_t *mps_arena_o, mps_arena_class_t mps_arena_class, ...)
Arguments
mps_arena_o
pointer to a variable to store the new arena in
mps_arena_class
the arena class
...
initialization arguments for the arena class
Initial/Default Values
Different for each arena class. See mps_arena_class_*
.
Returned Values
If the return value is MPS_RES_OK
, the new arena is in *mps_arena_o
.
Resources
mps.h
Description
mps_arena_create
is used to create an arena.
Example
mps_arena_t arena; int main(void) { mps_res_t res; res = mps_arena_create(&arena, mps_arena_class_vm(), ARENA_SIZE); if(res != MPS_ RES_OK) { printf("Not enough memory!"); exit(1); } /* rest of program */ }
Error Handling
mps_arena_create
returns MPS_RES_FAIL
, if the MPS library iscopy-protected by a security device, such as a dongle, and a valid security device cannot be found.Other error codes are specific to each arena class. See mps_arena_class_*
.
See Also
mps_arena_create_v
,
mps_arena_class_*
,
mps_arena_destroy
function mps_arena_create_v
Name
Summary
mps_arena_create_v
is used to create an arena.
Associated Protocols
Arena.
Syntax
mps_res_t mps_arena_create_v(mps_arena_t *mps_arena_o, mps_arena_class_t mps_arena_class, va_list args)
Arguments
mps_arena_o
pointer to a variable to store the new arena in
mps_arena_class
the arena class
args
initialization arguments for the arena class
Initial/Default Values
Different for each arena class. See mps_arena_class_*
.
Returned Values
If the return value is MPS_RES_OK
, the new arena is in *mps_arena_o
.
Resources
mps.h
Description
mps_arena_create_v
is used to create an arena. It is exactly the same as mps_arena_create
, except that it takes the arena class initialization arguments in a va_list
.
Error Handling
mps_arena_create_v
returns MPS_RES_FAIL
, if the MPS library is copy-protected by a security device, such as a dongle, and a valid security device cannot be found. Other error codes are specific to each arena class. See mps_arena_class_*
.
See Also
mps_arena_create
,
mps_arena_class_*
,
mps_arena_destroy
function mps_arena_expose
Name
Summary
mps_arena_expose
ensures
that the MPS is not protecting any pages in the arena with read- or
write-memory protection barriers.
Associated Protocols
Arena, clamp, park, protection
Syntax
mps_arena_expose(mps_arena);
Type
extern void mps_arena_expose(mps_arena_t);
Arguments
(mps_arena_t mps_arena)
mps_arena
is an MPS arena object.
Returned Values
None.
Resources
mps.h
Description
This function will ensure that the MPS is not protecting (with memory
read/write barriers) any page in the arena.
This is expected to only be useful for debugging.
The arena is left in the clamped state (see mps_arena_clamp
).
Since barriers are used during a collection, calling this function has
the same effect as calling mps_arena_park
; all collections are
run to completion and the arena is clamped so that no new collections
begin. The MPS also uses barriers to maintain remembered sets (an
optimisation to help avoid scanning work); calling
this function will effectively destroy the remembered sets and any
optimisation gains. Not only is there a performance hit from having to
finish any active collections, but future collections will be a lot more
expensive until the remembered set information has been restored (which
will usually take one collection cycle).
Example
None.
Error Handling
There can be no errors.
See Also
mps_arena_clamp
, mps_arena_park
function mps_arena_formatted_objects_walk
Name
mps_arena_formatted_objects_walk
Summary
mps_arena_formatted_objects_walk
is used to iterate over all formatted objects in the MPS heap.
Associated Protocols
None.
Syntax
mps_arena_formatted_objects_walk(mps_arena, client_step_function, client_step_closure_p,client_step_closure_s);
Type
extern void mps_arena_formatted_objects_walk(mps_arena_t, mps_formatted_objects_stepper_t, void *, size_t);
Arguments
(mps_arena_t mps_arena, mps_formatted_objects_stepper_t stepper, void *p, size_t s)
mps_arena
is an MPS arena object.
stepper
is a client-supplied function (pointer) of
the right type (see mps_formatted_objects_stepper_t
).
This function is applied to every object in all formatted pools.
This function should take the argument list (mps_addr_t object,
mps_fmt_t format,mps_pool_t pool, void *p, size_t s)
and return
void
.
object
is the object to which the function is being
applied.
format
is the format (an MPS format object)
of the object. pool
is the pool in which the object
resides. p
and s
are copies of the
corresponding values that the client passed into mps_arena_formatted_objects_walk
originally.
p
and s
are passed into the function specified by the stepper argument whenever the MPS calls that function. See mps_formatted_objects_stepper_t
.
Returned Values
None.
Resources
mps.h
Description
mps_arena_formatted_objects_walk
is used to iterate over all formatted objects in the MPS heap. A
client-supplied function is called for every object in all formatted
pools; the object, the format, and the pool are passed to the user
supplied function, as well as user supplied closure variables.
Applies stepper function to a pool-class-specific collection of objects (that is, the pool class determines which objects in its instances get walked). Typically pool classes will arrange that all validly formatted objects are walked. During a trace this will in general be only the blackobjects, though the leaf pool class (LO), for example, will walk all objects since they are validly formatted whether they are black or white. Padding objects may be walked at the pool classes discretion, the client should handle this case.
The user supplied stepper function is called in a restricted context. It may not in general call any MPS function.
Example
[not yet]
Error Handling
There are none.
See Also
mps_amc_apply
(the
historical walker),
mps_formatted_objects_stepper_t
Notes
function mps_arena_park
Name
Summary
mps_arena_has_addr
tests
whether an address is managed by a particular arena.
Associated Protocols
Arena.
Syntax
extern mps_bool_t mps_arena_has_addr(mps_arena_t arena,
mps_addr_t addr);
Arguments
arena
an arena
addr
an address
Returned Values
A boolean. Returns true if the address is managed by the arena, false otherwise.
Resources
mps.h
Description
mps_arena_has_addr
determines
whether a particular address is managed by a particular arena. An arena
manages a portion of total address space available on the hardware
architecture. No two arenas overlap so for any particular address this
function will return true for at most one arena. In general not all the
architecture addresses are managed by some arena; some addreses will not
be managed by any arena. This is what allows the MPS to cooperate with
other memory managers, shared object loaders, memory mapped file I/O,
and such like - it does not steal the whole address space.
The results from this function are true only for the instant at which the function returned. In some circumstances the results may immediately become invalidated (for example, a garbage collection may occur, the address in question may become free, the arena may choose to unmap the address and return storage to the operating system); for reliable results call this function whilst the arena is parked.
Example
Error Handling
Can't fail.
See Also
mps_arena_park used to park an
arena
,
Notes
None.
function mps_arena_park
Name
Summary
mps_arena_park
puts the specified arena into the parked state.
Associated Protocols
Arena.
Syntax
extern void mps_arena_park(mps_arena_t arena);
Arguments
arena
the arena to park
Returned Values
None.
Resources
mps.h
Description
mps_arena_park
puts the specified arena into the parked state. While an arena is parked,no object motion will occur and the staleness of location dependencies will not change. Allreferences to objects loaded while the arena is parked will keep the same binary representationuntil after it is released.
Any current collection is run to completion before the arena is parked, and no newcollections will start. When an arena is in the parked state, it is necessarily not in the middle ofa collection.
Example
Error Handling
Can't fail.
See Also
mps_arena_clamp
,
mps_arena_release
Notes
None.
function mps_arena_release
Name
Summary
mps_arena_release
puts the specified arena into the unclamped state.
Associated Protocols
Arena.
Syntax
extern void mps_arena_release(mps_arena_t);
Arguments
Returned Values
None.
Resources
mps.h
Description
mps_arena_release
puts the specified arena into the unclamped state. While an arena isunclamped, garbage collection, object motion, and other background activity can take place.
Example
Error Handling
Can't fail.
See Also
mps_arena_clamp
,
mps_arena_park
Notes
None.
function mps_arena_roots_walk
Name
Summary
mps_arena_roots_walk
is used to iterate over all roots of the MPS heap.
Associated Protocols
None.
Syntax
mps_arena_roots_walk(mps_arena, client_step_function, client_step_closure_p,client_step_closure_s);
Type
extern void mps_arena_roots_walk(mps_arena_t, mps_roots_stepper_t, void *, size_t);
Arguments
(mps_arena_t mps_arena, mps_roots_stepper_t stepper, void *p, size_t s)
mps_arena
in an MPS arena object.
stepper
is a client-supplied function (pointer) of the right type (see mps_roots_stepper_t
). This function is applied to every reference to the heap fromevery root object registered with the arena. This function should take the argument list (mps_addr_t *ref, mps_root_t root, void *p, size_t s)
. ref
is the address of a root which references an object in the arena. root
is the registered root (an MPS root object) of which ref
is a single reference, p
and s
are copies of the corresponding values that the client passed into mps_arena_roots_walk
originally.
p
and s
are passed into the function specified by the stepper argument whenever the MPS calls that function. See mps_roots_stepper_t
Returned Values
None.
Resources
mps.h
Description
mps_arena_roots_walk
is used to iterate over all roots of the MPS heap. A client-suppliedfunction is called for every root reference which points to an object in any automatically managedpools; the address of the root reference and the MPS root object are passed to the user suppliedfunction, as well as some closure variables.
May only be called when the arena is in the parked state.
Applies stepper to each reference in any roots registered with the arena and which point toobjects in automatically managed pools. If the root has rank MPS_RANK_AMBIG
then the reference mightnot be to the start of an object; the client should handle this case. There is no guarantee that thereference corresponds to the actual location that holds the pointer to the object (since this mightbe a register, for example) - but the actual location will be passed if possible. This may aidanalysis of roots via a debugger.
Example
[not yet]
Error Handling
There are none.
See Also
mps_arena_formatted_objects_walk
Notes
function mps_arena_spare_commit_limit
Name
Summary
Retrieves the value of the spare commit limit (previously set with mps_arena_spare_commit_limit_set
).
Associated Protocols
Arena.
Type
extern size_t mps_arena_spare_commit_limit(mps_arena_t arena)
Arguments
mps_arena_t
arena
Specifies the arena to retrieve the spare commit limit of.
Returned Values
Returns, as a size_t, the value of the spare commit limit.
Resources
mps.h
Description
Returns the current value of the spare commit limit which is the value most recently setwith mps_arena_spare_commit_limit_set
. (See mps_arena_spare_commit_limit_set
fordetails).
Example
[missing]
Error Handling
There are no errors.
See Also
mps_arena_spare_commit_limit_set
Notes
None.
function mps_arena_spare_commit_limit_set
Name
mps_arena_spare_commit_limit_set
Summary
Sets the limit of the amount of spare committed memory.
Associated Protocols
Arena.
Type
extern void mps_arena_spare_commit_limit_set(mps_arena_t arena, size_t limit)
Arguments
mps_arena_t
arena
The arena to which the new limit should apply.
size_t limit
The value of the new limit (specified in bytes).
Resources
mps.h
Description
The limit argument specifies a new "spare commit limit". The spare commit limit specifiesthe maximum amount of bytes of "spare committed" memory the MPS is allowed to have. Setting it to avalue lower than the current amount of spare committed memory would immediately cause sufficientspare committed memory to be uncommitted so as to bring the value under the limit. In particularsetting to 0 will mean that the MPS will have no "spare committed" memory.
"spare committed" memory is the term for describing memory which the arena is managing asfree memory (so not in use by any pool and not otherwise in use for obscure internal reasons) butwhich remains committed (mapped from the OS). It is used by the arena to (attempt to) avoid callingthe OS to repeatedly unmap and map areas of VM. "spare committed" memory is counted as committedmemory as counted by mps_arena_committed
and restricted by mps_arena_commit_limit
.
Non-VM arenas do not have this concept, but they support the two functions mps_arena_spare_commit_limit
and mps_arena_spare_commit_limit_set
. The functions simply get andretrieve a value but do nothing else in that case.
Initially the value is some configuration-dependent value.
The value of the limit can be retrieved with mps_arena_spare_commit_limit
.
Example
[missing]
Error Handling
There are no errors.
See Also
Notes
None.
function mps_arena_spare_committed
Name
Summary
Returns the number of bytes of spare committed memory.
Associated Protocols
Memory
Type
size_t mps_arena_spare_committed(mps_arena_t);
Arguments
The arena to which the query applies.
Returned Values
Returns the number of bytes of spare committed memory.
Resources
mps.h
Description
"Spare committed" memory is the term for describing memory which the arena is committed fromthe OS but which is free (so not in use by any pool and not otherwise in use for obscure internalreasons). It is used by the arena to (attempt to) avoid calling the OS to repeatedly uncommit andcommit areas of VM (because calling the OS to commit and uncommit memory is typically expensive)."Spare committed" memory can be used for grant client requests; if this is done when the MPS wouldotherwise have had to call the OS to commit more memory then the MPS has avoid some OS calls.
"spare committed" memory is counted as part of committed memory. The amount of committedmemory can be retrieved with mps_arena_committed
(see mps_arena_committed
).
The amount of "spare committed" memory can be limited by using mps_arena_spare_commit_limit_set
(see mps_arena_spare_commit_limit_set
), and the valueof that limit can be retrieved with mps_arena_spare_commit_limit
(see mps_arena_spare_commit_limit
). This is analogous to the functions for limiting theamount of committed memory.
Example
[missing]
Error Handling
[missing]
See Also
mps_arena_spare_commit_limit_set
,
mps_arena_spare_commit_limit
Notes
None.
function mps_bool_t
Name
Summary
mps_bool_t
is a transparent type, equivalent to int
, that is used in the MPS C interfaceto indicate that a boolean value is intended.
Associated Protocols
Not applicable.
Syntax
Not applicable.
Structure
Not applicable.
Type
typedef int mps_bool_t;
Resources
mps.h
Description
When used as an input parameter to the MPS, a value of 0 indicates "false" and any othervalue indicates "true". As an output parameter or function return from the MPS, 0 indicates "false",and 1 indicates "true". Note that an mps_bool_t
value can be used in a conditional context, suchas in an "if" statement.
Example
if(mps_ld_isstale(&ld, space, obj)) { mps_ld_reset(&ld, space); mps_ld_add(&ld, space, obj); }
See Also
Notes
None.
function mps_class_amc
Name
Summary
mps_class_amc
returns the pool class object for the Automatic Mostly Copying pool class.
Associated Protocols
Pool
Syntax
mps_class_t mps_class_amc(void)
Arguments
No arguments.
Returned Values
Returns a pool class object.
Resources
mpscamc.h
Description
This function returns an object of type mps_class_t
which represents the Automatic MostlyCopying pool class.
This pool class requires an extra argument when used in mps_pool_create
:
res = mps_pool_create(&pool, arena, mps_class_amc(), format);
The extra argument, format, should be of type mps_fmt_t
and specifies the format of theobjects allocated in the pool.
An AMC pool is both scannable and collectable. Objects may contain exact references to otherobjects that will preserve such other objects. Objects may be reclaimed if they are not reachablefrom a root. Objects may move during collection, unless reachable via a (direct) ambiguousreference. Objects in an AMC pool may be registered for finalization. Exact (that is, non-ambiguous)references into an object in an AMC pool must be to the start of the object.
The AMC pool class exploits assumptions about object lifetimes and inter-connectionvariously referred to as "the generational hypothesis". In particular, the following tendencies willbe efficiently exploited by such a pool:
- Most objects die young;
- Objects that don't die young will live a long time;
- Most references are backwards in time.
mps_ap_frame_push
and mps_ap_frame_pop
may be used on an allocation point in an AMC pool.They do not declare the affected objects to be definitely dead (compare with the SNC pool class),but have an undefined effect on the collection strategy.
If an allocation point is created in an AMC pool, the call to mps_ap_create
will take noadditional parameters.
Example
See Also
mps_ap_frame_pop
,
mps_ap_frame_push
,
mps_ap_create
function mps_class_mvff
Name
Summary
Used as a parameter to mps_pool_create
to create an MVFF pool.
Associated Protocols
Pool, Allocation Points.
Type
mps_class_t mps_class_mvff(void)
Arguments
None.
Returned Values
The function returns a class object that can be passed to mps_pool_create
.
Resources
mpscmvff.h
Description
MVFF pools implement a first-fit policy. The pool requires six parameters to pool creation:
mps_size_t
extendBy -- The size of segment to allocate by default;mps_size_t
avgSize -- The average size of objects to be allocated;mps_align_t
alignment -- The alignment of addresses for allocation (and freeing) in thepool;mps_bool_t
slotHighmps_bool_t
arenaHighmps_bool_t
firstFit
The alignment is the alignment of ranges that can be allocated and freed. If an unalignedsize is passed to mps_alloc
or mps_free
, it will be rounded up to the pool's alignment. The minimumalignment supported by pools of this class is
sizeof(void *)
.
The three boolean parameters may be set to (0, 0, 1) or (1, 1, 1). No other settings of these parameters is currently recommended.
Buffered allocation (mps_reserve
and mps_commit
) is also supported, but in that case, thepolicy is rather different: buffers are filled worst-fit, and allocation is always upwards from thebase. The arenaHigh parameter regulates whether new segments are acquired at high or low addresses;the slotHigh and firstFit parameters do not affect buffered allocation. Buffered and unbufferedallocation can be used at the same time, but in that case, the first allocation point must becreated before any call to mps_alloc
.
Cached allocation ( MPS_SAC_ALLOC
and MPS_SAC_FREE
) is also supported, but in that case,the policy is a little different: allocation from the cache follows its own policy (typicallyfirst-fit), and only when the cache needs to acquire more blocks from the underlying MVFF pool doesit use the usual algorithm to choose blocks for the cache.
Example
if(mps_pool_create(&pool, arena, mps_class_mvff(), 8 * 1024, 135, 4, 0, 0, 1) != MPS_RES_OK) { printf("Error creating pool!"); exit(2); }
See Also
mps_pool_create
,
mps_reserve
,
mps_commit
.
Notes
It is usually not advisable to use buffered and unbuffered allocationat the same time,because the worst-fit policy of buffer filling will grab all the large blocks, leading to severefragmentation. Use two separate pools instead.
Note that using buffered allocation prevents (for obscure technical reasons) the pool fromallocating across segment boundaries. This can cause added external fragmentation if objects areallocated that are a significant fraction of the segment size. (This quirk will disappear in afuture version.)
function mps_class_snc
Name
Summary
Returns the pool class object (of type mps_class_t
) for the Stack No Check pool class.
Associated Protocols
Pool.
Syntax
mps_class_t mps_class_snc(void)
Arguments
No arguments.
Returned Values
Returns a pool class object.
Resources
mpscsnc.h
Description
This function returns an object of type mps_class_t
which represents the Stack No Check poolclass.
This pool class requires an extra argument when used in mps_pool_create
:
res = mps_pool_create(&pool, arena, mps_class_snc(), format);
The extra argument, format, should be of type mps_fmt_t
and specifies the format of theobjects allocated in the pool (in a similar way to mps_class_amc
). The format should provide atleast the methods: scan, skip, pad.
An SNC pool is scannable, in that objects may contain references to objects in other poolsthat will keep those objects alive (depending on rank). In this sense, an SNC pool is a de-factoroot.
Exact references may point to (the start of) objects in an SNC pool, but will have no effecton whether those objects are either scanned or kept alive.
If mps_ap_frame_pop
is used on an allocation point in an SNC pool (after a correspondingcall to mps_ap_frame_push
), then the objects affected by the pop are effectively declared dead, andmay be reclaimed by the collector. Extant references to such objects from reachable or de factoalive objects are safe, but such other objects should be dead; that is, such references must neverbe used.
If an allocation point is created in an SNC pool, then the call to mps_ap_create
will takeas an additional parameter the rank (of type mps_rank_t
) of references in the objects to be createdin that allocation point. Currently, only rank exact (mps_rank_exact
) is supported.
Objects in an SNC pool may not be registered for finalization.
Objects in an SNC pool will not move.
Example
Nya
Error Handling
Cannot fail.
See Also
mps_class_amc
,
mps_ap_frame_pop
,
mps_ap_frame_push
,
mps_ap_create
mps_class_mvt
Name
Summary
mps_class_mvt
is a function that returns the MVT pool class object.
Associated Protocols
Allocation point.
Syntax
mps_class_t mps_class_mvt(void);
Type
C function
Arguments
None.
Returned Values
The MVT pool class object.
Resources
mpscmv2.h
Description
The function mps_class_mvt
returns the MVT pool
class object, which can be used to create an MVT pool instance by
passing the class object as the mps_class_t
(third) argument to
mps_pool_create
.
The MVT pool class manually manages variable-sized, unformatted objects. The MVT pool usesan allocation policy termed "temporal fit". Temporal fit attempts to place consecutive allocationsnext to each other. It relies on delaying reuse as long as possible to permit freed blocks tocoalesce, thus maximizing the number of consecutive allocations that can be co-located. Temporal fitpermits a very fast allocator and a deallocator competitive in speed with all other known policies.
Temporal fit is intended to take advantage of knowledge of object lifetimes, either apriori knowledge or knowledge acquired by profiling. The best performance of the MVT poolwill be achieved by allocating objects with similar expected deathtimes together.
A simple policy can be implemented to take advantage of MVT: Object size is typicallywell-correlated with object life-expectancy, and birthtime plus lifetime gives deathtime, soallocating objects of similar size sequentially from the same pool instance should result in objectsallocated close to each other dying at about the same time.
An application that has several classes of objects of widely differing life expectancy willbest be served by creating a different MVT pool instance for each life-expectancy class. A moresophisticated policy can use either the programmer's knowledge of the expected lifetime of an objector any characteristic of objects that correlates with lifetime to choose an appropriate poolinstance to allocate in.
Allocating objects with unknown or very different deathtimes together will pessimize thespace performance of MVT.
Example
if(mps_pool_create(&pool, arena, mps_class_mvt(), 8, 32, 256, 70, 20) != MPS_RES_OK) { printf("Error creating pool!"); exit(2); }
Error Handling
mps_class_mvt
cannot result in an error.
See Also
Notes
Creation
The MVT pool class has five creation parameters:
mps_res_t mps_pool_create(mps_pool_t * pool, mps_arena_t arena, mps_class_t mvt_class, size_t minimum_size, size_t mean_size, size_t maximum_size, mps_count_t reserve_depth mps_count_t fragmentation_limit);
Sizes
minimum_size
, mean_size
, and maximum_size
are the minimum, mean, and maximum (typical) sizein bytes of objects expected to be allocated in the pool. Objects smaller than minimum size may beallocated, but the pool is not guaranteed to manage them space-efficiently. Objects larger thanmaximum_size may be allocated, but the pool is not guaranteed to manage them space-efficiently.Furthermore, partial freeing is not supported for objects larger than maximum size; doing so willresult in the storage of the object never being reused. Mean_size need not be an accurate mean,although the pool will manage mean_size objects more efficiently.
Reserve Depth
reserve_depth is the expected hysteresis of the object population. When pool objects arefreed, the pool will retain sufficient storage to allocate reserve_depth objects of mean_size fornear term allocations (rather than immediately making that storage available to other pools).
If a pool has a stable object population, one which only grows over the lifetime of thepool, or one which grows steadily and then shrinks steadily, use a reserve_depth of 0.
It is always safe to use a reserve depth of 0, but if the object population typicallyfluctuates in a range (e.g., the client program may repeatedly create and destroy a subset ofobjects in a loop), it is more efficient for the pool to retain enough storage to satisfy thatfluctuation. For example, if a pool has an object population that typically fluctuates between 8,000and 10,000, use a reserve_depth of 2,000.
The reserve will not normally be available to other pools for allocation, even when it isnot used by the pool. If this is undesirable, a reserve depth of 0 may be used for a pool whoseobject population does vary, at a slight cost in efficiency. The reserve does not guarantee anyparticular amount of allocation.
Fragmentation Limit
fragmentation_limit is a percentage in (0, 100] that can be used to set an upper limit onthe space overhead of MVT in case object deathtimes and allocations do not correlate well.
If the free space managed by the pool as a ratio of all the space managed by the poolexceeds the specified percentage, the pool will fall back to a first fit allocation policy,exploiting space more efficiently at a cost in time efficiency.
A fragmentation_limit of 0 would cause the pool to operate as a first-fit pool, at asignificant cost in time-efficiency, therefore is not permitted.
A fragmentation_limit of 100 will cause the pool to use temporal fit (unless resources areexhausted). If the objects allocated in the pool have similar lifetime expectancies, this mode willhave the best time- and space-efficiency. If the objects have widely varying lifetime expectancies,this mode will be time-efficient, but may be space-inefficient. An intermediate setting can be usedto limit the space-inefficiency of temporal fit due to varying object life expectancies.
Allocation
The MVT pool class only supports allocation through allocation points. See mps_ap_create
.
Deallocation
The MVT pool class supports explicit freeing. See mps_pool_free
.
Internal Notes
Need a life-expectancy parameter! How else will different instances choose their Loci?
Need an alignment parameter. Perhaps this is embedded in a format parameter (when all poolshave at least a null format).
It is conceivable that a client would want to mix manual and automatic pools with the manualpool being able to be a root for the automatic. To do so, MVT would need to support formattedobjects and scanning. This may be added someday.
Eventually the MM product will include profiling tools that will help determine objectcharacteristics that correlate with object lifetime and suggest how to configure the appropriatenumber of MVT pool instances and what characteritics to dispatch on when choosing which instance toallocate from.
[From mail.ptw.1998-08-19.02-33(0) ]
Remember Wilson's statement that the goal of a memory manager is to exploit the regularitiesin allocation patterns? My intent in the interface parameters is to accept measurable regularitiesin object populations, then the implementation can exploit them.
Perhaps the pool should accept some description of the mean and deviation of the objectsizes, object population, and object lifetimes. Is that what you are getting at? [Reserve_depth is in some sense a deviation.]
Type mps_class_t
Name
Summary
mps_class_t
is the type of pool classes.
Associated Protocols
Pool.
Resources
mps.h
Description
mps_class_t
is the abstract type of pool classes. It is opaque. A pool class may beobtained by calling the class function for the appropriate class, such as mps_class_amc
for theAMC class. A pool class is used when creating a pool with mps_pool_create
or mps_pool_create_v
.
Example
See Also
mps_pool_create
,
mps_pool_create_v
Notes
mps_class_s
is an incomplete structure type used only to define mps_class_t
.
function mps_finalize
Name
Summary
Registers an object for finalization.
Associated Protocols
Finalization, message.
Syntax
mps_res_t mps_finalize(mps_arena_t arena, mps_addr_t *object_ref)
Arguments
arena
-- the arena in which the object lives
object_ref
-- a pointer to a pointer to the object to be finalized
Returned Values
A result code.
Resources
mps.h
Description
This function registers the specified object for finalization. This object must be an objectallocated from a pool in the specified arena.
An object becomes finalizable if it is registered for finalization, and the collectorobserves that it would otherwise be reclaimable. Note that the subsequent creation of strongreferences to the object (from, say, weak references) may cause finalization to occur when an objectis not otherwise reclaimable.
When an object is finalizable, it may be finalized up to N times, where N is the number oftimes it has been registered for finalization. When an object is finalized, it is also deregisteredfor finalization (so that it cannot be finalized again from the same registration).
Finalization is performed by passing a finalization message to the client, containing anexact reference to the object. See the message protocol, mps_message_type_finalization
, and mps_message_finalization_ref
for details.
If an object is registered for finalization multiple times, then there may be multiplefinalization messages on the queue at the same time. It may also be necessary to discard previousfinalization messages for an object before all such messages are posted on the message queue.Clients performing multiple registrations must cope with both behaviors.
Note that there is no guarantee that finalization will be prompt, although the collectordoes attempt to do this.
Note that there will be no attempt to finalize objects in the context of mps_arena_destroy
or mps_pool_destroy
. mps_pool_destroy
should therefore not be invoked on pools containing objects registered for finalization.
Not all pools support finalization of objects in those pools. For more information, see thePool Class Catalog.
Example
[missing]
Error Handling
[missing]
See Also
mps_message_*
,
mps_arena_destroy
,
mps_pool_destroy
Notes
This function receives a pointer to a reference. This is to avoid placing the restriction onthe client that the C call stack be a root.
type mps_fmt_A_s
Name
Summary
mps_fmt_A_s
is a structure used to create object formats of variant A.
Associated Protocols
Format.
Type
typedef struct mps_fmt_A_s { mps_align_t align; mps_fmt_scan_t scan; mps_fmt_skip_t skip; mps_fmt_copy_t copy; mps_fmt_fwd_t fwd; mps_fmt_isfwd_t isfwd; mps_fmt_pad_t pad; } mps_fmt_A_s;
Resources
mps.h
Description
Objects of this type are intended to be used in the creation of object formats. Objectformats describe the layout of client objects.
mps_fmt_A_s
is a structure that represents the particular collection of methodsand values that describes an object format of variant A.
Broadly speaking, the object formats of this variant are suitable for use in copying ormoving memory managers.
mps_fmt_A_s
has the
following methods: scan
, skip
, copy
, fwd
, isfwd
, pad
, and the following value:align
.
align
is an integer value defines the alignment of objects allocated with thisformat. It should be large enough to satisfy the alignment requirements of any field in the objects,and it cannot be larger than the arena alignment. For details of the methods, consult the reference pages for the type of each method.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt my_format; mps_res_t res; mps_fmt_A_s my_format_A = { my_alignment, &my_scan, &my_skip, &my_copy, &my_fwd, &my_isfwd, &my_pad }; res = mps_fmt_create_A(&my_format, arena, &my_format_A); assert(res != MPS_RES_OK); return my_format; }
See Also
mps_fmt_create_A
,
mps_fmt_scan_t
,
mps_fmt_skip_t
mps_fmt_copy_t
,
mps_fmt_fwd_t
,
mps_isfwd_t
,
mps_pad_t
,
mps_align_t
,
mps_fmt_B_s
type mps_fmt_A_t
Name
Summary
mps_fmt_A_t
is the type pointer to mps_fmt_A_s
.
Associated Protocols
Format.
Type
typedef struct mps_fmt_A_s { mps_align_t align; mps_fmt_scan_t scan; mps_fmt_skip_t skip; mps_fmt_copy_t copy; mps_fmt_fwd_t fwd; mps_fmt_isfwd_t isfwd; mps_fmt_pad_t pad; } mps_fmt_A_s; typedef struct mps_fmt_A_s *mps_fmt_A_t;
Resources
mps.h
Description
mps_fmt_A_t
is the type pointer to mps_fmt_A_s
. A value of this type represents a collection of methods and values that can be used to create a format object of type mps_fmt_t
.This type represents a particular collection of methods and values; other collections arerepresented by other types.
Objects of type mps_fmt_A_t
are intended to be used in the creation of object formats.Object formats describe the layout of client objects. The function mps_fmt_create_A
takes an mps_fmt_A_t
as one of its arguments and creates an object of type mps_fmt_t
(an object format).
See the documentation of mps_fmt_A_s
for further details.
Example
See Also
mps_fmt_A_s
,
mps_fmt_t
,
mps_fmt_create_A
type mps_fmt_B_s
Name
Summary
mps_fmt_B_s
is a transparent structure used to create object formats of variantB.
Associated Protocols
Format.
Type
typedef struct mps_fmt_B_s { mps_align_t align; mps_fmt_scan_t scan; mps_fmt_skip_t skip; mps_fmt_copy_t copy; mps_fmt_fwd_t fwd; mps_fmt_isfwd_t isfwd; mps_fmt_pad_t pad; mps_fmt_class_t mps_class; } mps_fmt_B_s;
Resources
mps.h
Description
Objects of this type are intended to be used in the creation of object formats. Objectformats describe the layout of client objects. mps_fmt_B_s
is a structure thatrepresents the particular collection of methods and values that describes an object format ofvariant B.
mps_fmt_B_s
is the same as mps_fmt_A_s
except for the addition ofthe mps_class
method. Broadly speaking, the object formats of variety B are suitable for use incopying or moving memory managers (just like variety A); the addition of the class method allowsmore information to be passed to various support tools (such as graphical browsers).
mps_fmt_B_s
has the following methods: scan, skip, copy, fwd, isfwd, pad, mps_class
, and the following value: align.
align is an integer value defines the alignment of objects allocated with this format. Itshould be large enough to satisfy the alignment requirements of any field in the objects, and itcannot be larger than the arena alignment. For details of the methods, consult the reference pagesfor the type of each method.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt_B_s my_format_B = { my_alignment, &my_scan, &my_skip, &my_copy, &my_fwd, &my_isfwd, &my_pad, &my_class }; mps_fmt my_format; mps_res_t res; res = mps_fmt_create_B(&my_format, arena, &my_format_B); assert(res != MPS_RES_OK); return my_format; }
See Also
mps_fmt_create_B
,
mps_fmt_scan_t
,
mps_fmt_skip_t
,
mps_fmt_copy_t
,
mps_fmt_fwd_t
,
mps_isfwd_t
,
mps_pad_t
,
mps_align_t
,
mps_class_t
,
mps_fmt_A_s
Notes
The mps_class
field used to be called "class", but that was problematic forC++, so we changed it.
Type mps_fmt_B_t
Name
Summary
mps_fmt_B_t
is a type passed to mps_fmt_create_B
. It represents the collection of methodsand values used to create a mps_fmt_t
. You are expected to declare and create structures of thistype if you require an object of type mps_fmt_B_t
.
Associated Protocols
Format
Structure
typedef struct mps_fmt_B_s { mps_align_t align; mps_fmt_scan_t scan; mps_fmt_skip_t skip; mps_fmt_copy_t copy; mps_fmt_fwd_t fwd; mps_fmt_isfwd_t isfwd; mps_fmt_pad_t pad; mps_fmt_class_t class; } mps_fmt_B_s;
Type
typedef struct mps_fmt_B_s *mps_fmt_B_t;
Resources
mps.h
Description
mps_fmt_B_t
is the equivalent to mps_fmt_A_t
that should be passed tomps_fmt_create_B
. It is suitable for format variety A collectors that need to use tools that useclass information.
See the documentation for the symbol mps_fmt_B_s
for further details.
Example
See Also
mps_fmt_B_s
,
mps_fmt_t
,
mps_fmt_create_B
,
mps_fmt_A_t
Notes
None.
type mps_fmt_auto_header_s
Name
Summary
mps_fmt_auto_header_s
is a structure used to create objectformats of variant auto_header.
Associated Protocols
Format.
Type
typedef struct mps_fmt_auto_header_s { mps_align_t align; mps_fmt_scan_t scan; mps_fmt_skip_t skip; mps_fmt_fwd_t fwd; mps_fmt_isfwd_t isfwd; mps_fmt_pad_t pad; size_t mps_headerSize; } mps_fmt_auto_header_s;
Resources
mps.h
Description
Objects of this type are intended to be used in the creation of object formats. Objectformats describe the layout of client objects. mps_fmt_auto_header_s
isa structure that represents the particular collection of methods and values that describes an objectformat of variant auto_header.
Broadly speaking, the object formats of this variant are suitable for use in automaticmemory management for objects with headers (hence the name). More precisely, this variant is intended for formats where the client's pointers point some distance into the memory blockcontaining the object. This typically happens when the objects have a common header used for memorymanagement or class system purposes, but this situation also arises when the low bits of a pointerare used for a tag. The MPS does not care what the reason is, only about the offset of the pointerin relation to the memory block.
mps_fmt_auto_header_s
has the following methods: scan
, skip
, fwd
, isfwd
, pad
, and the following values: align
and mps_headerSize
.
align
is an integer value defines the alignment of objectsallocated with this format. It should be large enough to satisfy the alignment requirements of anyfield in the objects, and it cannot be larger than the arena alignment.
mps_headerSize
is the size of the header, i.e., the offset of aclient pointer from the base the memory block. For details of the methods, consult the referencepages for the type of each method.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt format; mps_res_t res; mps_fmt_auto_header_s format_desc = { my_alignment, &my_scan, &my_skip, &my_fwd, &my_isfwd, &my_pad, HEADER_SIZE }; res = mps_fmt_create_auto_header(&format, arena, &format_desc); assert(res != MPS_RES_OK); return format; }
See Also
mps_fmt_create_auto_header
,
mps_fmt_scan_t
,
mps_fmt_skip_t
,
mps_fmt_fwd_t
,
mps_isfwd_t
,
mps_pad_t
,
mps_align_t
,
mps_fmt_A_s
Notes
For technical reasons, client objects must be longer than the header, i.e., objectsconsisting of only a header are not supported. However, if the header size is larger than or equalto alignment, the pad method must still be able to create padding objects down to alignment size.
At the moment, this format only works with pool classes AMC and AMCZ.
type mps_fmt_class_t
Name
Summary
mps_fmt_class_t
is a function pointer type for the class method of a format.
Associated Protocols
Format. Telemetry.
Type
typedef mps_addr_t (*mps_fmt_class_t)(mps_addr_t addr);
Arguments
addr the address of the object whose class is of interest
Returned Values
Returns an address that the client associates with the class or type of the object.
Resources
mps.h
Description
mps_fmt_class_t
is t he type of a format's class method. A class methodreturns an address that is related to the class of the object, for passing on to various supporttools (such as graphical browsers).
A class method is provided by the client as part of a format (see Format Protocol).
The exact meaning of the return value is up to the client, but it would typically bear somerelation to class or type in the client program. The client may have objects that represent classesor types. These may be associated with strings via mps_telemetry_intern
and mps_telemetry_label
.
Example
mps_addr_t my_class_method(mps_addr_t object) { my_object_generic_t generic_object = object; return (mps_addr_t)(generic_object.class); }
Error Handling
A class method is not allowed to fail, but may return NULL.
See Also
Notes
It is recommended that NULL be returned for padding objects and forwarded objects.
type mps_fmt_copy_t
Name
Summary
mps_fmt_copy_t
is a function pointer type for the copy method of a format.
Associated Protocols
Format.
Type
typedef void (*mps_fmt_copy_t)(mps_addr_t old, mps_addr_t new);
Arguments
old -- the address of the object
new -- the address to which the object should be copied
Resources
mps.h
Description
mps_fmt_copy_t
is a function pointer type for the copy method of a format. A copy methodcopies an object to a new location. It may be called by the MPS as part of copying garbagecollection, for example.
A copy method is required in some formats (in particular formats A and B (see mps_fmt_A_t
and mps_fmt_B_t
)). A copy method takes the address of an object and another address, and copiesthe object to the new address. The new and the old locations are guaranteed not to overlap.
Example
void my_copy_method(mps_addr_t old, mps_addr_t new) { size_t length = (char*)my_skip_method(old) - (char *)old; memcpy(new, old, length); }
Error Handling
A copy method is not allowed to fail.
See Also
mps_fmt_t
,
mps_fmt_create_A
,
mps_fmt_A_t
,
mps_fmt_B_t
,
mps_fmt_create_B
Notes
Most pools will just ignore Copy methods, and do the copy themselves.
function mps_fmt_create_A
Name
Summary
Function for create a format of variety A.
Associated Protocols
Format.
Syntax
mps_res_t mps_fmt_create_A(mps_fmt_t *fmt_o, mps_arena_t arena, mps_fmt_A_s *fmt_A);
Arguments
fmt_o
- the address of a variable to hold the new format
arena
- the arena in which to create the format
fmt_A
- format description of variety A
Returned Values
Result status. If the return value is MPS_RES_OK
, the new format is in *fmt_o
.
Resources
mps.h
Description
This function creates a format from a user format specification of variety A.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt_A_s my_format_A = { my_alignment, &my_scan, &my_skip, &my_copy,&my_fwd, &my_isfwd, &my_pad }; mps_fmt my_format; mps_res_t res; res = mps_fmt_create_A(&my_format, arena, &my_format_A); if(res != MPS_RES_OK) { fprintf(stderr, "Couldn't create format.\n"); exit(1); } return my_format; }
Error Handling
The MPS may exhaust some resource in the course of mps_fmt_create_A
and willreturn an appropriate error code in such circumstances.
See Also
mps_fmt_A_s
,
mps_fmt_t
,
mps_fmt_create_B
function mps_fmt_create_B
Name
Summary
Function for create a format of variety B.
Associated Protocols
Format.
Syntax
mps_res_t mps_fmt_create_B(mps_fmt_t *fmt_o, mps_arena_t arena, mps_fmt_B_s *fmt_B);
Arguments
arena
- the arena in which to create the format
fmt_B
- format description of variety B
Returned Values
Result status. If the return value is MPS_RES_OK
, the new format is in*fmt_o
.
Resources
mps.h
Description
This function creates a format from a user format specification of variety B. It is verysimilar to mps_fmt_create_A
.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt_B_s my_format_B = { my_alignment, &my_scan, &my_skip, &my_copy, &my_fwd, &my_isfwd, &my_pad, &my_class }; mps_fmt my_format; mps_res_t res; res = mps_fmt_create_B(&my_format, arena, &my_format_B); assert(res != MPS_RES_OK); return my_format; }
Error Handling
The MPS may exhaust some resource in the course of mps_fmt_create_B
and willreturn an appropriate error code in such circumstances.
See Also
mps_fmt_B_s
,
mps_fmt_t
,
mps_fmt_create_A
function mps_fmt_create_auto_header
Name
Summary
Function for create a format of variety auto_header.
Associated Protocols
Format.
Syntax
mps_res_t mps_fmt_create_auto_header(mps_fmt_t *fmt_o, mps_arena_t arena, mps_fmt_auto_header_s *fmt_st);
Arguments
fmt_o
- the address of a variable to hold the new format
arena
- the arena in which to create the format
fmt_st
- format description of variety auto_header
Returned Values
Result status. If the return value is MPS_RES_OK
, the new format is in *fmt_o
.
Resources
mps.h
Description
This function creates a format from a user format specification of variety auto_header.
Example
mps_fmt_t create_format(mps_arena_t arena) { mps_fmt_auto_header_s format_desc = { my_alignment, &my_scan, &my_skip, &my_fwd, &my_isfwd, &my_pad, HEADER_SIZE }; mps_fmt format; mps_res_t res; res = mps_fmt_create_auto_header(&format, arena, &format_desc); assert(res != MPS_RES_OK); return format; }
Error Handling
The MPS may exhaust some resource in the course of mps_fmt_create_auto_header
and will return an appropriate error code in such circumstances.
See Also
mps_fmt_auto_header_s
,
mps_fmt_t
,
mps_fmt_create_A
type mps_fmt_fwd_t
Name
Summary
The type of a format's forward method.
Associated Protocols
Format.
Type
typedef void (*mps_fmt_fwd_t)(mps_addr_t old, mps_addr_t new);
Arguments
old
the address of an object
new
the address where the object has been moved
Returned Values
None.
Description
mps_fmt_fwd_t
is the type of a format's forward method. A forward method isused to store relocation information in a heap. It may be called by the MPS as part of copyinggarbage collection.
A forward method is provided by the client as part of a format (see Format Protocol ). TheMPS calls a forward method when it has relocated an object. The forward method when called mustreplace the object at 'old' with a forwarding marker that points to the address 'new'. Theforwarding marker must meet the following requirements:
it must be possible for the MPS to call other format methods with the address of aforwarding marker as the argument.
he forwarding marker must not be bigger than the original object.
t must be possible to distinguish the forwarding marker from ordinary objects using theisfwd method (see
mps_fmt_isfwd_t
), and the isfwd method must return the address'new'.
Example
/* define the function */ void example_fwd(mps_addr_t old, mps_addr_t new) { /* ... */ } /* also define example_scan, example_skip, etc */ /* store pointer to function in the format variant struct */ struct mps_fmt_B_s example_fmt_B = { 4, /* align */ example_scan, example_skip, example_copy, example_fwd, example_isfwd, example_pad, example_class }; /* The (address of the) example_fmt_B object can now be passed to */ /* mps_fmt_create_B to create a format. */
See Also
mps_fmt_A_s
,
mps_fmt_B_s
,
mps_fmt_auto_header_s
,
mps_fmt_isfwd_t
Notes
This method is never invoked by the GC on an object in a non-moving pool.
type mps_fmt_isfwd_t
Name
Summary
The type of a format's isfwd ("is forwarded") method.
Associated Protocols
Format.
Type
typedef mps_addr_t (*mps_fmt_isfwd_t)(mps_addr_t addr);
Arguments
addr
the address of a candidate object
Returned Values
Either a null pointer to indicate the object at
addr
has not been relocated, orthe new location of the object if there is a forwarding marker at
addr
indicating thatthe object has been relocated.
Description
The type of a format's isfwd ("is forwarded") method. An isfwd method is used to testwhether an object has been relocated using the format's forward method.
An isfwd method is provided by the client as part of a format (see protocol.mps.format(0) ).The MPS calls the isfwd method to determine whether an object in the heap has been relocated or not.Objects in the heap are relocated using the format's forward method (see mps_fmt_fwd_t
). When the isfwd method is called the parameter addr will be the address of either an object or aforwarding marker created with the forward method. If it is an object (so it has not been relocated)the method should return a null pointer; otherwise it is a forward marker indicating the address ofthe relocated object, the address of the relocated object should be returned (this should be thesame as the 'new' parameter that was passed to the forward method that created the forwardingmarker).
Example
<example of how to use the symbol>
See Also
mps_fmt_A_s
,
mps_fmt_B_s
,
mps_fmt_auto_header_s
,
mps_fmt_fwd_t
Notes
This method is never invoked by the GC on an object in a non-moving pool.
type mps_fmt_pad_t
Name
Summary
The type of a format's pad method.
Associated Protocols
Format.
Type
typedef void (*mps_fmt_pad_t)(mps_addr_t addr, size_t size);
Arguments
addr
The address at which to create a padding object.
size
The size (in bytes) of the padding object to be created.
Returned Values
None.
Description
The type of a format's pad method. A pad method is used to create padding objects.
A pad method is provided by the client as part of a format (see Format Protocol ). The MPScalls a pad method when it wants to create a padding object. Typically the MPS creates paddingobjects to fill in otherwise unused gaps in memory; they allow the MPS to pack objects in fixed-sizeunits (such as OS pages). The pad method should create a padding object of the specified size at thespecified address. The size can be any aligned (to the format alignment) size. A padding objectshould be acceptable to other methods in the format (scan, skip, isfwd, etc.).
Example
<example of how to use the symbol>
See Also
type mps_fmt_scan_t
Name
Summary
Type of the scan method of a format.
Associated Protocols
Format, Scanning.
Syntax
typedef mps_res_t (*mps_fmt_scan_t)(mps_ss_t scan_state, mps_addr_t base, mps_addr_t limit)
Arguments
scan_state
a scan state
base
a client pointer to the first object in the block to be scanned
limit
a client pointer to the object just beyond the end of the block
Returned Values
A result code.
Resources
mps.h
Description
This is the type of scanning functions provided by the client in some format variants and mps_root_create_fmt
. When the MPS needs to scan objects in an area of memory that thisscanning function has been registered for, it will be called with a scan state and the limits of theblock of objects to scan. It must then indicate references within the objects by usingmps_fix
or one of the alternatives.
The base
and limit
arguments are client pointers, as usual. Notethat there might not be any object at the location indicated by limit
.
Example
/* Scanner for a simple Scheme-like language with just two interesting types */ mps_res_t scan_objs(mps_ss_t ss, mps_addr_t base, mps_addr_t limit) { mps_res_t res; mps_addr_t obj; MPS_SCAN_BEGIN(ss) for(obj = base; obj < limit;) { /* obj maps over the objects to scan */ switch(((Object*)obj)->type) { case ArrayType: { size_t i; Array *array = (Array *)obj; for(i = 0; i < array->length; ++i) { /* fix each element */ res = MPS_FIX12(ss, &array->contents[i]); if(res != MPS_RES_OK) return res; } obj = AddrAdd(obj, ArraySize(array)); /* move to next object */ break; } case StackFrameType: { StackFrame *frame = (StackFrame *)obj; for(i = frame->size; i > 0; --i) { /* fix each local var */ res = MPS_FIX12(ss, &frame->locals[i]); if(res != MPS_RES_OK) return res; } res = MPS_FIX12(ss, &frame->next); if(res != MPS_RES_OK) return res; obj = AddrAdd(obj, StackFrameSize(frame)); break; } default: /* other types don't contain references */ obj = AddrAdd(obj, DefaultSize(obj)); break; } } MPS_SCAN_END(ss); return res; }
Error Handling
If a fixing operation returns a value other than MPS_RES_OK
, the scanning function must
return that value, and may return without scanning further
references. Generally, itis better if it returns as soon as
possible. If the scanning is completed successfully, the
function should return MPS_RES_OK
.
See Also
mps_fmt_A_s
,
mps_fmt_B_s
,
mps_fmt_auto_header_s
,
mps_root_create_fmt
,
mps_fix
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
type mps_fmt_skip_t
Name
Summary
mps_fmt_skip_t
is a function pointer type for the skip method of a format.
Associated Protocols
Format.
Type
typedef mps_addr_t (*mps_fmt_skip_t)(mps_addr_t obj);
Arguments
obj
the client pointer to the object to be skipped
Returned Values
The skip method should return the address of the next object.
Resources
mps.h
Description
mps_fmt_skip_t
is a function pointer type for the skip method of a format.
These methods are provided by the client as part of a format and invoked by the MPS (seeFormat Protocol). The skip method takes the client pointer to the object. The method should returnthe client pointer to the next object, whether there is one or not. With no headers, this is theaddress just past the end of this object; with headers, it's the address just past where the headerof next object would be. It is always the case that the difference between the argument and thereturn value is the size of the block containing the object.
Example
mps_addr_t my_skip_method(mps_addr_t object) { char *p = (char *)object; my_object_t my_object = (my_object_t)object; return((mps_addr_t)(p + my_object->length)); }
Error Handling
A skip method is not allowed to fail.
See Also
mps_fmt_A_s
,
mps_fmt_B_s
,
mps_fmt_auto_header_s
type mps_fmt_t
Name
Summary
mps_fmt_t
is the type of object formats.
Associated Protocols
Format.
Type
typedef struct mps_fmt_s *mps_fmt_t;
mps_fmt_s
is an incomplete structure type used only to declare the opaque type mps_fmt_t
.
Resources
mps.h
Description
mps_fmt_t
is the opaque type of object formats. An object format is a way for the MPS andclient programs to communicate regarding the layout of client objects. For more information, seeFormat Protocol.
Example
#include "mps.h" #include "mpscamc.h" #include <stdlib.h> struct mps_fmt_A_s fmt_A_s = { (mps_align_t)4, scan, skip, copy, move, isMoved, pad }; void go(mps_space_t space) { mps_fmt_t format; mps_res_t res; mps_pool_t pool; res = mps_fmt_create_A(&format, space, &mps_fmt_A_s); if(res != MPS_RES_OK) abort(); res = mps_pool_create(&pool, space, mps_class_amc(), format); if(res != MPS_RES_OK) abort(); /* do some stuff here */ mps_pool_destroy(pool); mps_format_destroy(format); }
See Also
mps_fmt_create_A
,
mps_fmt_create_B
,
mps_fmt_destroy
,
mps_fmt_A_t
type mps_formatted_objects_stepper_t
Name
mps_formatted_objects_stepper_t
Summary
Type of the client supplied heap walker function.
Associated Protocols
Heap walking.
Type
typedef void (*mps_formatted_objects_stepper_t)(mps_addr_t, mps_fmt_t, mps_pool_t, void *,size_t )
Arguments
mps_formatted_objects_stepper
is a type not a function so it doesn't take any arguments; however the
function pointed to by an object of this type does. Such functions
take the following argument list:
(mps_addr_t object, mps_fmt_t format, mps_pool_t pool, void *p, size_t s)
object
is a pointer to the (client) object.
format
is the MPS format of the client object.
pool
in the MPS pool in which the client object resides.
p
and s
are two closure values which are copies of the corresponding values which the client
passed into the heap walking function, mps_arena_formatted_objects_walk
.
Returned Values
The function pointed to by an object of type mps_formatted_objects_stepper_t
returns no
arguments.
Resources
mps.h
Description
This symbol describe the type of pointers passed into the heap
walking function
mps_arena_formatted_objects_walk
.
The heap walker arranges to apply this
function to all objects on the heap, see
mps_arena_formatted_objects_walk
for details.
Example
<example of how to use the symbol>
Error Handling
The function pointed to by an object of type mps_formatted_objects_stepper_t
have no way to
return an error code to the caller.
See Also
mps_arena_formatted_objects_arena_walk
Notes
function mps_free
Name
Summary
Frees a block of memory to a pool.
Associated Protocols
Allocation
Syntax
void mps_free(mps_pool_t pool, mps_addr_t p, size_t size);
Arguments
pool
the pool of the object to be freed
p
a pointer to the object to the freed
size
the size of the object to the freed in bytes
Returned Values
None.
Resources
mps.h
Description
Frees an object of memory, returning the memory block to the pool it was allocated from.The pool might then decide to make it available to other pools, but the way this happens depends onthe pool class and the current situation.
Example
See Also
Notes
mps_free
takes a size argument, because it is most efficient to do so. In practicalprograms, the type of an object is usually known at the point in the code that calls thedeallocation function, and hence the size is trivially available. In such cases. storing the size onthe MPS side would cost time and memory, and make it hard to get good virtual memory behaviour (asit is, the deallocation code doesn't have to touch the dead object at all).
Undoubtedly, one day, we'll get around to writing a pool that stores the size of eachobject.
function mps_lib_memcmp
Name
Summary
A plinth function similar to C's "memcmp".
Associated Protocols
Plinth
Syntax
int mps_lib_memcmp(const void *s1, const void *s2, size_t n);
Arguments
s1, s2 pointers to memory blocks to be compared
n length of the blocks, in bytes
Returned Values
An integer that is greater than, equal to, or less than zero, accordingly as the blockpointed to by "s1" is greater than, equal to, or less than the block pointer to by "s2".
Resources
mpslib.h
Description
This function is intended to have the same semantics as the "memcmp" function of the ANSI Cstandard (section 7.11.4.1).
Like other plinth features, it is used by the MPS and provided by the client (possibly usingthe ANSI plinth, mpsliban.c).
Example
None, clients don't use it.
Error Handling
None.
See Also
mps_lib_memset
,
mps_lib_memcpy
,
mpsliban.c
Notes
None.
function mps_lib_memcpy
Name
Summary
A plinth function similar to C's "memcpy".
Associated Protocols
Plinth
Syntax
void *mps_lib_memcpy(void *dest, const void *source, size_t n);
Arguments
dest destination of copy
source source of copy
n length of the blocks, in bytes
Returned Values
Returns the value of the dest argument.
Resources
mpslib.h
Description
This function is intended to have the same semantics as the "memcpy" function of the ANSI Cstandard (section 7.11.2.1).
Like other plinth features, it is used by the MPS and provided by the client (possibly usingthe ANSI plinth, mpsliban.c).
Example
None, clients don't use it.
Error Handling
None.
See Also
mps_lib_memset
,
mps_lib_memcmp
,
mpsliban.c
Notes
None.
function mps_lib_memset
Name
Summary
A plinth function similar to C's "memset".
Associated Protocols
Plinth
Syntax
void *mps_lib_memset(void *s, int c, size_t n);
Arguments
s destination of copy
c byte (when converted to an unsigned char) to copy
n length of the block, in bytes
Returned Values
Returns the value of s.
Resources
mpslib.h
Description
This function is intended to have the same semantics as the "memset" function of the ANSI Cstandard (section 7.11.6.1).
Like other plinth features, it is used by the MPS and provided by the client (possibly usingthe ANSI plinth, mpsliban.c).
Example
None, clients don't use it.
Error Handling
None.
See Also
mps_lib_memcpy
,
mps_lib_memcmp
,
mpsliban.c
Notes
None.
mps_lib_telemetry_control
Name
Summary
Plinth function to supply a default value for telemetry filters from environment.
Associated Protocols
Telemetry
Type
unsigned long mps_lib_telemetry_control();
Arguments
None.
Initial/Default Values
In the absence of environmental data, a default of zero is recommended.
Returned Values
The default value of the telemetry filter, as derived from the environment. It isrecommended that the environment be consulted for a symbol analagous to MPS_TELEMETRY_CONTROL
, subject to local restrictions.
Resources
Depends on access to the environment.
Description
See mps_telemetry_control
for more information on the significant of the values.
Example
See the supplied ANSI plinth for an example implementation.
See Also
function mps_message_discard
Name
Summary
mps_message_discard
is used to indicate that there is no further use for thespecified message.
Associated Protocols
Message.
Syntax
void mps_message_discard(mps_arena_t arena, mps_message_t message)
Arguments
arena
-- the arena
message
-- the message
Returned Values
None.
Resources
mps.h
Description
mps_message_discard
is used to indicate that the client has no further use for the specified message in the specified arena. After this, the message may not be passed as argumentto any message functions.
This procedure is called by the client program as a courtesy, to avoid the use of unboundedresources by the message queue itself and any other MPS-managed objects referred to by the message.
Example
[missing]
Error Handling
Can't fail.
See Also
Notes
A finalized object will not be reclaimed by the collector until the finalization message hasbeen discarded, because the message contains a reference to the object.
function mps_message_finalization_ref
Name
Summary
mps_message_finalization_ref
returns the "finalization reference" property of thespecified message in the specified arena.
Associated Protocols
Message, finalization.
Syntax
void mps_message_finalization_ref(mps_addr_t *object_ref, mps_arena_t arena, mps_message_tmessage)
Arguments
object_ref -- a reference to a reference to the object to which the message belongs
arena -- the arena that the message is in
message -- a message of a message type that supports this method
Returned Values
None.
Resources
mps.h
Description
This method returns the "finalization reference" property of the specified message in thespecified arena. The message must be of a message type that supports this method. Currently, theonly such type is that of finalization messages, as returned by mps_message_type_finalization
.
Note that the reference returned is subject to the normal constraints such as movingcollection, if appropriate. For this reason, it is returned indirectly via "object_ref" to enablethe client to place it directly into scanned memory, without imposing the restriction that the Cstack be a root.
Note that the invocation of this method does not affect the liveness of the specifiedmessage and hence the liveness of the object referred to by "object_ref. Use mps_message_discard
to discard messages.
Example
Error Handling
See Also
mps_message_*
,
mps_finalize
function mps_message_gc_condemned_size
Name
Summary
mps_message_gc_condemned_size
returns the "condemned size" property of the specified message in the specified arena.
Associated Protocols
Message, GC.
Syntax
size_t mps_message_gc_condemned_size(mps_arena_t arena, mps_message_tmessage)
Arguments
arena
-- the arena
message
-- a message of a message type that supports this method
Returned Values
An approximate size for the set of objects condemned in the collection that generated themessage.
Resources
mps.h
Description
Currently, the only type of message that supports this property is mps_message_type_gc
, such messages are generated whenever a garbage collectioncompletes. This method returns an approximation to the size of the set of objects that werecondemned in that collection.
Example
Error Handling
See Also
mps_message_*
function mps_message_gc_live_size
Name
Summary
mps_message_gc_live_size
returns the "live size" property of thespecified message in the specified arena.
Associated Protocols
Message, GC.
Syntax
size_t mps_message_gc_live_size(mps_arena_t arena, mps_message_t message)
Arguments
arena
-- the arena;
message
-- a message of a message type that supports thismethod.
Returned Values
The total size of the condemned objects that survived the collection that generated themessage.
Resources
mps.h
Description
Currently, the only type of message that supports this property is mps_message_type_gc
, such messages are generated whenever a garbage collectioncompletes. This method returns the size of the set of objects that were condemned in thatcollection, but survived.
Example
Error Handling
See Also
mps_message_*
function mps_message_gc_not_condemned_size
Name
mps_message_gc_not_condemned_size
Summary
mps_message_gc_not_condemned_size
returns the "not condemnedsize" property of the specified message in the specified arena.
Associated Protocols
Message, GC.
Syntax
size_t mps_message_gc_not_condemned_size(mps_arena_t arena,mps_message_t message)
Arguments
arena
-- the arena
message
-- a message of a message type that supports this method
Returned Values
An approximate size for the set of objects that were in collected pools, but were notcondemned in the collection that generated the message.
Resources
mps.h
Description
Currently, the only type of message that supports this property is mps_message_type_gc
; such messages are generated whenever a garbage collectioncompletes. This method returns an approximation to the size of the set of objects that were incollected pools (so potentially subject to garbage collection), but were not condemned in thatcollection.
Example
Error Handling
See Also
mps_message_*
function mps_message_get
Name
Summary
Gets a message of the specified type from a message queue.
Associated Protocols
Message.
Syntax
mps_bool_t mps_message_get(mps_message_t *message_return, mps_arena_t arena, mps_message_type_tmessage_type)
Arguments
message_return
-- the handle to the message that was removed from the queue
arena
-- the arena
message_type
-- the type of message
Returned Values
Returns true if a message has been removed from the queue, false if not.
Resources
mps.h
Description
If there is a message of the specified type on the message queue of the specified arena,then this function removes one such message from the queue, returns a handle to it via themessage_return
argument, and returns true. Otherwise it returns false.
Example
See Also
mps_message_*
mps_message_poll
Name
Summary
mps_message_poll
determines whether there are currently any messages on amessage queue.
Associated Protocols
Message.
Syntax
mps_bool_t mps_message_poll(mps_arena_t arena)
Arguments
arena
-- the arena whose message queue you are interested in
Returned Values
A flag to indicate whether there are any messages on the queue.
Resources
mps.h
Description
mps_message_poll
is used to determine whether there are currently any messageson the message queue of the specified arena.
Example
[missing]
Error Handling
Can't fail.
See Also
Notes
If you expect a particular type of message, it is usually more practical to just call mps_message_get
.
function mps_message_queue_type
Name
Summary
mps_message_queue_type
returns the type of the first message on a message queue.
Associated Protocols
Message.
Syntax
mps_bool_t mps_message_queue_type(mps_message_type_t *message_type_return, mps_arena_t arena)
Arguments
message_type_return -- the type of the first message on the queue of the specified arena
arena -- the arena
Returned Values
"True" if there are any messages on the queue of the specified arena, "false" if not.
Resources
mps.h
Description
If there are any messages on the queue of the specified arena, then this function returns"true", and also returns the type of the first message via "message_type_return". Otherwise itreturns "false".
Example
See Also
mps_message_*
type mps_message_t
Name
Summary
mps_message_t
is used as a handle on an individual message.
Associated Protocols
Message.
Type
typedef struct mps_message_s *mps_message_t
mps_message_s
is an incomplete structure type used only to declare the opaque type mps_message_t
.
Resources
mps.h
Description
The opaque type mps_message_t
is used as a handle on an individual message. Messages aremanually managed. They are created at the instigation of the MPS (but see mps_message_type_enable
), and are deleted by the client.
An mps_message_t
is a reference into MPS managed memory, and can safely be stored as suchin scannable memory.
Example
Error Handling
Not applicable.
See Also
mps_message_*
function mps_message_type
Name
Summary
mps_message_type
returns the type of a message.
Associated Protocols
Message.
Syntax
mps_message_type_t mps_message_type(mps_arena_t arena, mps_message_t message)
Arguments
arena -- the arena containing the message
message -- a valid message; that is, one previously returned by mps_message_get
, and notdiscarded via mps_message_discard
Returned Values
The type of the specified message.
Resources
mps.h
Description
mps_message_type
returns the type of a message.
Example
Error Handling
See Also
mps_message_*
function mps_message_type_disable
Name
Summary
mps_message_type_disable
restores the arena to the default state whereby messages of thespecified type are not generated.
This reverses the effect of an earlier call to "m ps_message_type_enable".
Associated Protocols
Message.
Syntax
void mps_message_type_disable(mps_arena_t arena, mps_message_type_t message_type)
Arguments
arena -- the arena
message_type -- the message type to be disabled
Returned Values
None.
Resources
mps.h
Description
This procedure may be used by the client to specify that messages of the specified typeshould not created for the specified arena.
Messages are not generated by default, but the client may enable the generation of messageswith mps_message_type_enable
.
Any existing messages of the specified type are flushed from the message queue.
Example
[none]
Error Handling
Never fails.
See Also
mps_message_*
Notes
It is permitted to call this function when the message type is already disabled. Such a callwill have no effect.
function mps_message_type_enable
Name
Summary
mps_message_type_enable
allows messages of the specified type to be created for thespecified arena. Without such enabling, the MPS will, by default, not generate any messages of thattype.
Associated Protocols
Message.
Syntax
void mps_message_type_enable(mps_arena_t arena, mps_message_type_t message_type)
Arguments
arena -- the arena
message_type -- the message type to be enabled
Returned Values
None.
Resources
mps.h
Description
This procedure may be used by the client to specify that messages of the specified type maybe created for the specified arena. Without such enabling, the MPS will by default not generate anymessages of that type.
Note that the enabling of messages of a particular type implies that the client applicationwill handle and discard message of that type, or the message queue may consume unbounded resources.
The client may disable message generation again by means of an equivalent call to mps_message_type_disable
.
Example
[none]
Error Handling
Never fails.
See Also
mps_message_*
"Message Protocol"
Notes
It is permitted to call this function when the message type is already enabled. Such a callwill have no effect.
function mps_message_type_finalization
Name
Summary
mps_message_type_finalization
returns the type of finalization messages.
Associated Protocols
Message, Finalization.
Syntax
mps_message_type_t mps_message_type_finalization(void)
Arguments
None.
Returned Values
The type of finalization messages.
Resources
Not applicable.
Description
mps_message_type_finalization
returns the type of finalization messages. Finalizationmessages are used by the MPS see signal to the client that an object has become finalizable. Inaddition to the usual methods applicable to messages, finalization messages support the mps_message_finalization_ref
method which returns a reference to the registered object.
Example
{ mps_message_type_t type; if( mps_message_queue_type(&type, arena) ) { if(type == mps_message_type_finalization()) { process_finalization_message_from_queue(); } else { unknown_message_type(); } } }
See Also
mps_message_*
,
mps_finalize
function mps_message_type_gc
Name
Summary
mps_message_type_gc
returns the type of garbage collectionstatistic messages.
Associated Protocols
Message.
Syntax
mps_message_type_t mps_message_type_gc(void)
Arguments
None.
Returned Values
The type of garbage collection statistic messages.
Resources
mps.h
Description
mps_message_type_gc
returns the type of garbage collectionstatistic messages. Garbage collection statistic messages are used by the MPS to give the clientinformation about garbage collections that have occurred. Such information may be useful inanalysing the client's memory usage over time.
The access methods specific to a message of this type are:
mps_message_gc_live_size
-- gives the total size of thecondemned objects that survived the collection that generated the messagemps_message_gc_condemned_size
-- gives an approximate size for the set of objects condemned in the collection that generated the message.mps_message_gc_not_condemned_size
-- gives an approximate sizefor the set of objects that were in collected pools, but were not condemned in the collection thatgenerated the message.
Example
{ mps_message_t message; if(mps_message_get(&message, arena, mps_message_type_gc())) { size_t live, condemned, not_condemned; live = mps_message_gc_live_size(arena, message); condemned = mps_message_gc_condemned_size(arena, message); not_condemned = mps_message_gc_not_condemned_size(arena,message); mps_message_discard(arena, message); process_collection_stats(live, condemned, not_condemned); } }
Error Handling
Cannot fail.
See Also
mps_message_*
.
type mps_message_type_t
Name
Summary
mps_message_type_t
is the type of message types.
Associated Protocols
Message.
Resources
mps.h
Description
mps_message_type_t
is the type whose values are the various message types. It is opaque.
Example
See Also
mps_message_*
function mps_pool_check_fenceposts
Name
Summary
Check all the fenceposts in the pool.
Associated Protocols
Debug
Syntax
void mps_pool_check_fenceposts(mps_pool_t pool)
Arguments
pool the pool whose fenceposts are to be checked
Resources
mps.h
Description
This function is a debugging feature to check all the fenceposts in the pool. If a corruptedfencepost is found, an assert will fire. It is only useful to call this on a debug pool that hadfenceposting turned, it does nothing on other pools.
Example
mps_pool_check_fenceposts(gene_pool);
Error Handling
If a corrupted fencepost is found, an assert will fire. You will probably want to look at the problem with a debugger.
See Also
mps_class_*_debug
structure mps_pool_debug_option_s
Name
Summary
This structure is used to pass debug options to mps_pool_create
for debug classes.
Associated Protocols
Debug.
Type
typedef struct mps_pool_debug_option_s { void *fence_template; size_t fence_size; } mps_pool_debug_option_s;
Members
fence_template
the template for fencepost contents
fence_size
the size of the template in bytes
Resources
mps.h
Description
Structures of this type are used to pass debug options to mps_pool_create
when creatinginstances of debug classes.
Fenceposting is enabled by specifying a non-zero fence_size
; the size mustbe a multiple of the [pool/format] alignment. The content of fenceposts is given as a template thatis simply copied onto each fencepost (although sometimes the MPS will create fenceposts smaller thanthe given size, for example, to pad out some bit that was left unused because of alignmentrequirements).
Example
static mps_pool_debug_option_s debugOptions = { (void *)"postpost", 8 }; if(mps_pool_create(&pool, arena, mps_class_ams_debug(), &debugOptions, 8192, 135, 8) != MPS_RES_OK) { printf("Error creating pool!"); exit(2); }
See Also
Notes
Fencepost templates allow the client to specify complicated patterns that mimic illegal datavalues, that would cause an assert to fire if read by mistake, and that would never be written byany operation that writes at the wrong address by mistake.
Another trick is to make the pattern contain an instruction sequence that would cause theprogram to error or stop if executed by mistake.
function mps_rank_ambig
Name
Summary
Function returning the value representing "rank ambig".
Associated Protocols
Allocation, Root, Scanning.
Syntax
mps_rank_ambig()
Type
mps_rank_t mps_rank_ambig(void)
Arguments
None.
Returned Values
Returns a value of type mps_rank_t
representing "rank ambig".
Resources
mps.h
Description
Used to get a value for "rank ambig", which is used to denote that certain references (in aroot, for example) are ambiguous references.
Example
See Also
function mps_rank_exact
Name
Summary
Used to declare references which the client wishes to be exact references.
Associated Protocols
Allocation, Root, Scanning.
Type
mps_rank_t mps_rank_exact(void);
Arguments
No arguments.
Returned Values
Returns a rank (see mps_rank_t
) which can be used to declare references to be exactreferences.
Resources
mps.h
Description
Used to declare references which the client wishes to be exact, non-weak references.
Example
[missing]
See Also
mps_rank_t
,
mps_rank_ambig
,
mps_rank_weak
type mps_rank_t
Name
Summary
A type whose values are "reference ranks".
Associated Protocols
Allocation, Root.
Type
typedef unsigned int mps_rank_t;
Resources
mps.h
Description
mps_rank_t
is a concrete type. It is an alias (via the C typedef mechanism) for "unsignedint" provided for convenience and clarity. An object of type mps_rank_t
can store a valuerepresenting one reference rank. Reference ranks are used to conveniently express specific semanticsof particular references. See "MPS Scanning Protocol" for descriptions of these semantics, and mps_rank_*
for the actual ranks used to declare these semantics.
Example
(Probably won't be used explicitly, most likely to be seen in the prototype declaration forother MPS functions. For example, mps_root_create
.)
See Also
mps_rank_*
function mps_rank_weak
Name
Summary
Function to return a value used to represent "rank weak".
Associated Protocols
Allocation, Scanning.
Type
extern mps_rank_t mps_rank_weak(void);
Arguments
None.
Returned Values
Returns a value of type mps_rank_t
that represent "rank weak".
Resources
mps.h
Description
mps_rank_weak
returns a value used to represent "rank weak".
"Rank weak" is often used to denote that certain references (in a root or in objectsallocated in a pool) are weak references.
Example
<example of how to use the symbol>
See Also
type mps_reg_scan_t
Name
Summary
Type of root scanning functions for mps_root_create_reg
.
Associated Protocols
Root.
Syntax
typedef mps_res_t (*mps_reg_scan_t)( mps_ss_t scan_state, mps_thr_t thread, void *p, size_t s)
Arguments
scan_state a scan state
thread the thread
p a value passed through from root registration
s a value passed through from root registration
Returned Values
A result code.
Resources
mps.h
Description
This is the type of root scanning functions the client provides to mps_root_create_reg
.These functions will be called, whenever the root needs to be scanned, and passed the "p" and "s"values specified in the call to mps_root_create_reg
.
See Also
mps_root_create_reg
,
mps_stack_scan_ambig
Notes
Users are not expected to write any scanning functions of this type. The one functionsupplied with the MPS, mps_stack_scan_ambig
, should be enough for most purposes.
type mps_res_t
Name
Summary
mps_res_t
is the type of result codes returned by operations that may fail.
Type
typedef int mps_res_t;
Resources
mps.h
Description
A result code indicates the success or failure of an operation, along with the reason forfailure. Like UNIX error codes, the meaning of the code depends on the call that returned it. Referto the documentation of the function for the exact meaning. This documentation describes the broadcategories with mnemonic names for various sorts of problems.
MPS_RES_OK
: The operation succeeded. Out and in/out parameters will only be updated if OK isreturned, otherwise they will be left untouched. MPS_RES_OK
is zero.
MPS_RES_FAIL
: Something went wrong that does not fall into any of the other categories. Theexact meaning depends on the call. See the documentation of the function.
MPS_RES_RESOURCE
: A needed resource could not be obtained. Which resource, depends on thecall. Compare with MPS_RES_MEMORY
, which is a special case of this.
MPS_RES_MEMORY
: Needed memory (committed memory, not address space) could not be obtained.
MPS_RES_LIMIT
: An internal limitation was reached. For example, the maximum number ofsomething was reached.
MPS_RES_UNIMPL
: The operation, or some vital part of it, is unimplemented. This might bereturned by functions that are no longer supported, or by operations that are included for futureexpansion, but not yet supported.
MPS_RES_IO
: An I/O error occurred. Exactly what depends on the function.
MPS_RES_COMMIT_LIMIT
: The arena's commit limit would have been exceeded as a result of(explicit or implicit) allocation. See protocol.arena.commit.
MPS_RES_PARAM
: A parameter of the operation was invalid.
A
ny function that might fail will return a result code. Any other results of the function arebe passed back in "return" parameters. See MPS Interface Conventions for more information.
Example
mps_addr_t p; mps_res_t res; res = mps_alloc(&p, pool, sizeof(struct spong)); if(res != MPS_RES_OK) { handle_memory_error(res); abort(); }
For more examples, s ee doc.mps.ref-man.if-conv.
See Also
MPS_RES_*
function mps_root_create
Name
Summary
The function mps_root_create
declares a root that consists of all the references indicatedby a scanning function.
Associated Protocols
Root.
Syntax
mps_res_t mps_root_create(mps_root_t *root_o, mps_arena_t arena, mps_rank_t rank, mps_rm_trm, mps_root_scan_t scan, void *p, size_t s)
Arguments
root_o a pointer to a variable to store the new root structure
arena the arena
rank the rank of references in the root
rm the root mode
scan the scanning function
p a value to be passed to the scanning function
s a value to be passed to the scanning function
Returned Values
If the return value is MPS_RES_OK
, a new root structure in "*root_o".
Resources
mps.h.
Description
The client provides a scanning function, that will be called with a scan state and "p" and"s", whenever the root needs to be scanned. See mps_root_scan_t
for details.
If the rank of the root is not MPS_RANK_AMBIG
, the contents of the root have to be validwhenever a GC happens, i.e., they have to be references to actual objects or "NULL". If you're usingasynchronous GC, this could be right after the root is registered, so the root has to be valid whenit is registered. It's OK for a root to have entries which point to memory not managed by the MPS --they will simply be ignored.
Example
static mps_root_t mmRoot; int main(void) { mps_res_t res; /* ... */ res = mps_root_create(&mmRoot, arena, MPS_RANK_EXACT, (mps_rm_t)0, &rootScanner, NULL, 0); /* see doc of mps_root_scan_t for definition of rootScanner */ if(res != MPS_RES_OK) exit(1); /* ... */ }
Error Handling
mps_root_create
returns MPS_RES_MEMORY
when it fails to allocate memory for the internalroot structure; you need to deallocate or reclaim something to make enough space, or expand thearena.
See Also
mps_root_scan_t
,
mps_rm_t
,
mps_rank_t
,
mps_root_t
,
mps_root_create_fmt
,
mps_root_create_table
,
MPS_RM_CONST
Notes
"p" and "s" are just arbitrary data that scanning function can use. This is needed because Clacks local functions.
function mps_root_create_fmt
Name
Summary
The function mps_root_create_fmt
declares a root that consists of a block of objects, andprovides a scanning function for them.
Associated Protocols
Root.
Syntax
mps_res_t mps_root_create_fmt(mps_root_t *root_o, mps_arena_t arena, mps_rank_t rank, mps_rm_t rm, mps_fmt_scan_t scan, mps_addr_t base, mps_addr_t limit)
Arguments
root_o a pointer to a variable to store the new root structure
arena the arena
rank the rank of references in the root
rm the root mode
scan the scanning function
base the address of the start of the root
limit the address just beyond the end of the root
Returned Values
If the return value is MPS_RES_OK
, the new root in "*root_o".
Resources
mps.h
Description
The client provides a scanning function, that will be called with a scan state and an areaof memory, whenever the root needs to be scanned. See mps_fmt_scan_t
for details.
If the rank of the root is not MPS_RANK_AMBIG
, the contents of the root have to be validwhenever a GC happens, i.e., they have to be references to actual objects or "NULL". If you're usingasynchronous GC, this could be right after the root is registered, so the root has to be valid whenit is registered. It's OK for a root to have entries which point to memory not managed by the MPS --they will simply be ignored.
Example
static mps_root_t mmRoot; SegmentDescriptor DataSegment; int main(void) { mps_res_t res; /* ... */ res = mps_root_create_fmt(&mmRoot, arena, MPS_RANK_EXACT, (mps_rm_t)0, &scan_objs, (mps_addr_t)DataSegment.base, (mps_addr_t) (DataSegment.base + SegmentLength) ); /* see doc of mps_fmt_scan_t for definition of scan_objs */ if(res != MPS_RES_OK) exit( EXIT_FAILURE ); /* ... */ }
Error Handling
mps_root_create_fmt
returns MPS_RES_MEMORY
when it fails to allocate memory for theinternal root structure; you need to deallocate or reclaim something to make enough space, or expandthe arena.
See Also
mps_fmt_scan_t
,
mps_rm_t
,
mps_rank_t
,
mps_root_t
,
mps_root_create
,
mps_root_create_table
,
MPS_RM_PROT
,
MPS_RM_CONST
Notes
This is like mps_root_create_table
, except you get to supply your own scanning function.This is like mps_root_create
, except the scanning function has a slightly different argument list(and the MPS knows where the root is).
function mps_root_create_reg
Name
Summary
mps_root_create_reg
registers a thread as a root.
Associated Protocols
Root.
Syntax
mps_res_t mps_root_create_reg(mps_root_t * root_o, mps_arena_t arena, mps_rank_t rank, mps_rm_t rm, mps_thr_t thread, mps_reg_scan_t scan, void *p, size_t s)
Arguments
root_o a pointer to a variable to store the new root structure
arena the arena
rank the rank of references in the root
rm the root mode
thread the thread to the registered as a root
scan the scanning function
p a value to be passed to the scanning function
s a value to be passed to the scanning function
Returned Values
If the return value is MPS_RES_OK
, a new root structure in "*root_o".
Resources
mps.h
Description
mps_root_create_reg
declares the state of a thread as a root. The client provides a
scanning function that will be called and passed "p" and "s", whenever
the root needs to be scanned. See mps_reg_scan_t
for details.
If the rank of the root is not MPS_RANK_AMBIG
, the contents of the root have to be validwhenever a GC happens, i.e., they have to be references to actual objects or "NULL". If you're usingasynchronous GC, this could be right after the root is registered, so the root has to be valid whenit is registered. It's OK for a root to have entries which point to memory not managed by the MPS --they will simply be ignored.
Example
typedef struct { mps_root_t mmRoot; mps_thr_t thread; /* ... */ } ThreadLocals; void InitThread(ThreadLocals *thr) { /* This is a hack to find the bottom of the stack. */ void *stackBottom=&stackBottom; mps_thread_reg(&thr->thread, arena); mps_root_create_reg(&thr->mmRoot, arena, MPS_RANK_AMBIG, (mps_rm_t) 0, thr->thread, mps_stack_scan_ambig, stackBottom, 0); /* ... */ }
Error Handling
mps_root_create_reg
returns MPS_RES_MEMORY
when it fails to allocate memory for theinternal root structure; you need to deallocate or reclaim something to make enough space, or expandthe arena.
See Also
mps_stack_scan_ambig
,
mps_reg_scan_t
Notes
Only one suitable scanning function is supplied with the MPS, namely mps_stack_scan_ambig
.
function mps_root_create_table
Name
Summary
mps_root_create_table
create s a root that is a vector of references.
Associated Protocols
Root.
Syntax
mps_res_t mps_root_create_table(mps_root_t *root_o, mps_arena_t arena, mps_rank_t rank, mps_rm_t rm, mps_addr_t *base, size_t size)
Arguments
root_o a pointer to a variable for storing the new root structure in
arena the arena
rank the rank of the references in this root
rm the root mode
base a pointer to the vector of references that is being registered
size the number of references in the vector being registered
Returned Values
If the return value is MPS_RES_OK
, the new root in "*root_o".
Resources
mps.h
Description
This function declares a root that is a vector of references.
If the rank of the root is not MPS_RANK_AMBIG
, the contents of the root have to be validwhenever a GC happens, i.e., they have to be references to actual objects or "NULL". If you're usingasynchronous GC, this could be right after the root is registered, so the root has to be valid whenit is registered. It's OK for a root to have entries which point to memory not managed by the MPS --they will simply be ignored.
Example
static mps_root_t mmRoot; Object *Objects[rootCOUNT]; int main(void) { mps_res_t res; /* ... */ res = mps_root_create_table(&mmRoot, arena, MPS_RANK_EXACT, (mps_rm_t)0, (mps_addr_t) &Objects, rootCOUNT ); if(res != MPS_RES_OK) exit(1); /* ... */ }
Error Handling
mps_root_create_table
returns MPS_RES_MEMORY
when it fails to allocate memory for theinternal root structure; you need to deallocate or reclaim something to make enough space, or expandthe arena.
See Also
mps_root_create_table_masked
,
MPS_RM_PROT
,
MPS_RM_CONST
function mps_root_create_table_masked
Name
Summary
mps_root_create_table_masked
creates a root that is a vector of tagged values.
Associated Protocols
Root.
Syntax
mps_res_t mps_root_create_table_masked(mps_root_t *root_o, mps_arena_t arena, mps_rank_t rank, mps_rm_t rm, mps_addr_t *base, size_t size, mps_word_t mask);
Arguments
root_o a pointer to a variable for storing the new root structure in
arena the arena
rank the rank of the references in this root
rm the root mode
base a pointer to the vector of references that is being registered
size the number of references in the vector being registered
mask any element that has any of the bits in mask set is ignored
Returned Values
If the return value is MPS_RES_OK
, the new root in "*root_o".
Resources
mps.h
Description
mps_root_create_table_masked
creates a root that is a table of tagged values. The maskparameter indicates which bits of a pointer are tag bits. References are assumed to have a tag ofzero, values with other tags are ignored.
If the rank of the root is not MPS_RANK_AMBIG
, the contents of the root have to be validwhenever a GC happens, i.e., they have to be references to actual objects or "NULL". If you're usingasynchronous GC, this could be right after the root is registered, so the root has to be valid whenit is registered. It's OK for a root to have entries which point to memory not managed by the MPS --they will simply be ignored.
Example
#define tagMASK 0x0003 static mps_root_t mmRoot; Object *Objects[rootCOUNT]; int main(void) { mps_res_t res; /* ... */ res = mps_root_create_table_masked(&mmRoot, arena, MPS_RANK_EXACT, (mps_rm_t)0, (mps_addr_t)&Objects, rootCOUNT, (mps_word_t)tagMASK); if(res != MPS_RES_OK) exit(1); /* ... */ }
Error Handling
mps_root_create_table_masked
returns MPS_RES_MEMORY
when it fails to allocate memoryfor the internal root structure; you need to deallocate or reclaim something to make enough space,or expand the arena.
See Also
mps_root_create_table
,
MPS_RM_PROT
,
MPS_RM_CONST
type mps_root_scan_t
Name
Summary
Type of root scanning functions for mps_root_create
.
Associated Protocols
Root.
Syntax
typedef mps_res_t (*mps_root_scan_t)(mps_ss_t scan_state, void * p, size_t s)
Arguments
scan_state
a scan state
p
an argument passed through from mps_root_create
s
an argument passed through from mps_root_create
Returned Values
A result code.
Resources
mps.h
Description
This is the type of root scanning functions the client provides tomps_root_create
. The MPS will call these functions whenever the root needs to bescanned, with a scan state (of type mps_ss_t
), and the p
ands
values specified in the call to mps_root_create
. Apart from theargument list, the scanning function works like the format scan methods: it needs to indicate allreferences using mps_fix
or MPS_FIX*
.
Example
static StackFrame *stackBottom; /* root scanner for an imaginary interpreter for a stack-oriented language */ static mps_res_t rootScanner(mps_ss_t ss, void * p, size_t s) { StackFrame *frame; size_t i; mps_res_t res; UNUSED(p); UNUSED(s); for(frame = stackBottom; frame != NULL; frame = frame->next) for(i = frame->size; i > 0; --i) { res = mps_fix(ss, &frame->locals[i]); if(res != MPS_RES_OK) return res; } return res; }
Error Handling
If a fixing operation returns a value other than MPS_RES_OK
, the scanning function must
return that value, and may return without scanning further
references. Generally, itis better if it returns as soon as
possible. If the scanning is completed successfully, the
function should return MPS_RES_OK
.
See Also
mps_root_create
,
mps_ss_t
,
mps_fix
,
MPS_SCAN_BEGIN
,
MPS_SCAN_END
,
MPS_FIX12
,
MPS_FIX1
,
MPS_FIX2
,
MPS_FIX_CALL
,
mps_fmt_scan_t
type mps_roots_stepper_t
Name
Summary
Type of the client-supplied root walker component.
Associated Protocols
None.
Type
typedef void (*mps_roots_stepper_t)( mps_addr_t *, mps_root_t, void *, size_t )
Arguments
The function pointed to by an object of type mps_roots_stepper_t
takes the followingargument list:
(mps_addr_t *ref, mps_root_t root, void *p, size_t s)
ref is the address of a root which references an object in the arena. It's a pointer to aroot which points to "something" in the client heap. That "something" will be an object if the rootis an exact root. But it might be an interior pointer to an object if the root is an ambiguous root.
root is the MPS root object which contains ref.
p and s are two closure values which are copies of the corresponding values which the clientpassed into mps_arena_roots_walk
.
Returned Values
he function pointed to by an object of type mps_roots_stepper_t
returns no values.
Resources
mps.h
Description
A pointer to a function is passed into the function mps_arena_roots_walk
; the pointer hasthis type. The root walker arranges to apply this function to all objects which are directlyreferenced from the roots.
Example
<example of how to use the symbol>
Error Handling
T
he function pointed to by an object of type mps_roots_stepper_t
has no way of signalling anerror to the caller.
See Also
Notes
type mps_sac_class_s
Name
Summary
A structure describing a size class to be passed as an argument to mps_sac_create
.
Associated Protocols
Allocation cache
Type
typedef struct mps_sac_class_s { size_t mps_block_size; size_t mps_cached_count; unsigned mps_frequency; } mps_sac_class_s;
Resources
mps.h
Description
mps_sac_class_s
is the element t ype of the array passed tomps_sac_create
to describ e the size classes. Each element of this array describes oneclass by specifying block_size
, the maximum size (in bytes) in this class;cached_count
, the number of objects of this class to cache ; andfrequency
, a number that describes the frequency of requests (allocation anddeallocation combined ) in this class relative to all the other classes. The classes should be givenin the order of ascending size.
block_size
s have to be aligned to the pool alignment. All sizes must bedifferent, and the smallest size must be large enough to hold a void *
.
cached_count
is advice to the MPS on how many blocks to cache, not an absolutelimit. The cache policy tries to accommodate fluctuations in the population and minimize the cost ofresponding to client requests; the purpose of this parameter is to limit how much memory the clientis willing to set aside for this purpose. However, a
cached_count
of zero prevents anycaching of blocks falling into that class.
The MPS automatically provides an "overlarge" class for arbitrarily large objects above thelargest class described. Allocations falling into the overlarge class are not cached.
Example
mps_sac_t sac; mps_sac_class_s classes[3] = { {8, 38, 1}, {136, 19, 3}, {512, 4, 1} }; res = mps_sac_create(&sac, pool, 3, classes); if (res != MPS_RES_OK) { printf("Failed to create the allocation cache!"); exit(1); }
See Also
Notes
Any blocks whose size falls between two classes are allocated from the larger class.
function mps_sac_create
Name
Summary
This function creates a segregated allocation cache.
Associated Protocols
Allocation cache
Type
mps_res_t mps_sac_create(mps_sac_t *sac_o, mps_pool_t pool, size_t classes_count,mps_sac_class_s *classes);
Arguments
sac_o a pointer to a variable to hold the cache created
pool the pool the cache is attached to
classes_count the number of the size classes
classes pointer to the first element of an array describing the size classes
Returned Values
If the return value is MPS_RES_OK
, a new cache in *sac_o
.
Resources
mps.h
Description
This function creates an allocation cache whose free-list is segregated into the given sizeclasses. The cache can get more memory from the given pool, or return memory to it.
Segregated allocation caches can be associated with any pool that supports mps_alloc
and mps_free
.
The size classes are described by an array of element type mps_sac_class_s
(q.v.). This array is used to initialize the cache, and is not needed aftermps_sac_create
returns. There might be a limit on how many classes can be described,but it will be no less than MPS_SAC_CLASS_LIMIT
. You must specify at least one class.The MPS automatically provides an "overlarge" class for arbitrarily large objects above the largestclass described. Allocations falling into the overlarge class are not cached.
Example
mps_sac_t sac; mps_sac_class_s classes[3] = { {8, 38, 1}, {136, 19, 3}, {512, 4, 1} }; res = mps_sac_create(&sac, pool, 3, classes); if (res != MPS_RES_OK) { printf("Failed to create the allocation cache!"); exit(1); }
Error Handling
mps_sac_create
returns MPS_RES_MEMORY
orMPS_RES_COMMIT_LIMIT
when it fails to allocate memory for the internal cache structure;see the documentation for those return codes for recovery options. It returnsMPS_RES_LIMIT
if you ask for too many size classes; combine some small adjacentclasses. It returns MPS_RES_PARAM
if the pool doesn't support segregated allocationcaches.
See Also
mps_sac_class_s
,
MPS_SAC_CLASS_LIMIT
,
mps_sac_destroy
,
MPS_RES_MEMORY
,
MPS_RES_COMMIT_LIMIT
,
MPS_RES_LIMIT
,
MPS_RES_PARAM
,
mps_sac_t
Notes
Too many classes will slow down allocation; too few classes waste more space in internalfragmentation. It is assumed that overlarge allocations are rare; otherwise, you would add anotherclass for them, or even create separate allocation caches or pools for them.
Some pools will work more efficiently with caches than others. In the future, the MPS mightoffer pools specially optimized for particular types of cache.
Segregated allocation caches work poorly with debug pool classes at the moment: the checkingonly happens when blocks are moved between the cache and the pool. This will be fixed, but the speedof allocation with a debug class will always be similar to mps_alloc
, rather than cached speed.
function mps_sac_destroy
Name
Summary
This function destroys a segregated allocation cache.
Associated Protocols
Allocation cache
Type
void mps_sac_destroy(mps_sac_t);
Arguments
sac the segregated allocation cache
Returned Values
None.
Resources
mps.h
Description
This function destroys a segregated allocation cache. All memory held in it is returned tothe associated pool.
Example
res = mps_sac_create(&sac, pool, 3, classes); if (res != MPS_RES_OK) { printf("Failed to create the allocation cache!"); exit(1); } /* Use sac. */ mps_sac_destroy(sac); mps_pool_destroy(pool);
See Also
Notes
Destroying the cache might well cause the pool to return some memory to the arena, butthat's up to the pool's usual policy.
Destroying the cache has no effect on objects allocated through it.
function mps_sac_flush
Name
Summary
This function flushes the segregated allocation cache given.
Associated Protocols
Allocation cache
Type
void mps_sac_flush(mps_sac_t sac);
Arguments
sac the segregated allocation cache
Returned Values
None.
Resources
mps.h
Description
This function flushes the segregated allocation cache given, returning all memory held in itto the associated pool.
The client is responsible for synchronising the access to the cache, but the MPS willproperly synchronize with any other threads that might be accessing the same pool.
Example
mps_sac_t sac_small, sac_large; res = mps_sac_create(&sac_small, pool, 3, small_classes); if (res != MPS_RES_OK) { printf("Failed to create the small allocation cache!"); exit(1); } res = mps_sac_create(&sac_large, pool, 3, large_classes); if (res != MPS_RES_OK) { printf("Failed to create the large allocation cache!"); exit(1); } /* Use sac_small. */ mps_sac_flush(sac_small); /* Use sac_large. */ mps_sac_flush(sac_large); /* Use sac_small. */
See Also
Notes
This is something that you'd typically do when you know you won't be using the cache for awhile, but want to hold on to the cache itself. Destroying a cache has the effect of flushing it,naturally.
Flushing the cache might well cause the pool to return some memory to the arena, but that'supto the pool's usual policy.
Note that the MPS might also decide to take memory from the cache without the clientrequesting a flush.
type mps_sac_t
Name
Summary
Type of segregated allocation caches.
Associated Protocols
Allocation cache
Type
typedef struct mps_sac_s *mps_sac_t;
Resources
mps.h
Description
A value of this type represents an allocation cache with segregated freelists. It is anopaque type.
Example
mps_sac_t sac; mps_sac_class_s classes[3] = { {8, 38, 1}, {136, 19, 3}, {512, 4, 1} }; res = mps_sac_create(&sac, pool, 3, classes); if (res != MPS_RES_OK) { printf("Failed to create the allocation cache!"); exit(1); }
See Also
mps_sac_create
,
mps_sac_destroy
,
MPS_SAC_ALLOC
,
mps_sac_alloc
,
MPS_SAC_FREE
,
mps_sac_free
,
mps_sac_flush
Notes
None.
function mps_stack_scan_ambig
Name
Summary
A scanning function for ambiguous scanning of thread states.
Associated Protocols
Root.
Syntax
mps_res_t mps_stack_scan_ambig(mps_ss_t scan_state, mps_thr_t thread, void *stack_bottom, size_t ignore)
Arguments
scan_state a scan state
thread the thread
stack_bottom a pointer to the bottom of the stack
ignore ignored
Returned Values
A result code.
Resources
mps.h
Description
This is a root scanning function of type mps_reg_scan_t
. It will scan allinteger registers and everything on the stack of the thread given, and can therefore only be usedwith roots of rank MPS_RANK_AMBIG
. It will only scan things at the given stack bottompointer or higher on the stack (that is, more recently added). References are assumed to berepresented as machine words, and are required to be 4-byte-aligned; unaligned values are ignored.
Clients don't call this function, it is used as an argument of mps_root_create_reg
.
Example
typedef struct { mps_root_t mmRoot; mps_thr_t thread; /* ... */ } ThreadLocals; void InitThread(ThreadLocals *thr) { /* This is a hack to find the bottom of the stack. */ void *stackBottom=&stackBottom; mps_thread_reg(&thr->thread, arena); mps_root_create_reg(&thr->mmRoot, arena, MPS_RANK_AMBIG, (mps_rm_t)0, thr->thread, mps_stack_scan_ambig, stackBottom, 0) /* ... */ }
See Also
mps_reg_scan_t
,
mps_root_create_reg
Notes
The MPS provides this function because it's hard to write (it's OS- andarchitecture-dependent and possibly compiler-dependent).
function mps_telemetry_control
Name
Summary
This function is used to read and change the filters on the telemetry stream.
Associated Protocols
Telemetry.
Syntax
mps_word_t mps_telemetry_control(mps_word_t reset_mask, mps_word_t flip_mask);
Arguments
reset_mask is a bit mask indicating the bits that should be reset, regardless of previousvalue.
flip_mask is a bit mask indicating the bits whose value should be flipped after theresetting.
Returned Values
The function returns the previous value of the telemetry filter control.
Description
This function is used to read and change the filters on the telemetry stream. It isgenerally for use by developers.
The parameters reset_mask and flip_mask allow specifying any binary operation on the filtercontrol. To use this function for typical operations, the parameters should be set as follows:
Operation reset_mask flip_mask
set(M) M M
reset(M) M 0
flip(M) 0 M
read() 0 0
The significance of the bits is liable to change, but the current values (number the leastsignificant bit as zero) are:
0 -- per space or arena
1 -- per pool
2 -- per trace or scan
3 -- per page (segment)
4 -- per reference or fix
5 -- per allocation or object
6 -- user events (e.g., mps_telemetry_intern
)
Example
See Also
function mps_telemetry_flush
Name
Summary
This function is used to flush the internal event buffers.
Associated Protocols
Telemetry.
Syntax
void mps_telemetry_flush(void);
Resources
mps.h
Description
This function is used to flush the internal event buffers into the event stream. Thisfunction also calls mps_lib_io_flush
on the event stream itself. This ensures that even the latestevents are now properly recorded, should the application terminate (uncontrollably as a result of abug, for example) or some interactive tool require access to the event data. You could even trycalling this from a debugger after a problem.
Example
mps_telemetry_flush();
See Also
function mps_telemetry_intern
Name
Summary
This function registers a string with the MPS, and receives a unique identifier in return.This identifier is suitable for use with mps_telemetry_label
.
Associated Protocols
Telemetry
Type
mps_word_t mps_telemetry_intern(char *)
Arguments
The function receives a name as a nul-terminated string in the usual C way. The string'slength should not exceed 256 characters, including nul terminating character. In appropriatevarieties this restriction is checked and will cause the MPS to issue an ASSERT. So don't do it.
Returned Values
The function returns a unique idenifier that may be used to represent the string in future.
Description
The intention of this function is to provide an immediate
identifier that can be used toconcisely represent a string for the
purposes of mps_telemetry_label
. Note that
the appropriatesettings must be made to the telemetry filter (via
mps_telemetry_control
) before
this function is invoked; the associate event is of the user kind.
Error Handling
The string's length should not exceed 256 characters, including nul terminating character.This will cause the MPS to issue an ASSERT in appropriate varieties.
See Also
function mps_telemetry_label
Name
Summary
This function associates an identifier returned from mps_telemetry_intern
, and hence astring, with an address, in the telemetry stream.
Associated Protocols
telemetry
Type
void mps_telemetry_label(mps_addr_t, mps_word_t);
Arguments
The function receives an address and an identifier. The identifier should be one returned by mps_telemetry_intern
in the same session.
Description
This function is intended to associate the address with an identifier in the telemetrystream. Note that the user kind must be set in the telemetry filter.
Example
Typical uses include:
- Label pools with a human-meaningful name;
- Label allocated objects with their type or class.
See Also
mps_telemetry_intern
,
mps_telemetry_control
,
mps_thr_t
type mps_thr_t
Name
Summary
mps_thr_t
is the type of thread records registered with the MPS.
Associated Protocols
Threads.
Type
typedef mps_thr_s *mps_thr_t;
Resources
mps.h
Description
An object of the opaque type mps_thr_t
is a thread registration. In a multi-threadedenvironment where incremental garbage collection is used, threads must be registered with the MPS sothat the MPS can examine their state.
An object of type mps_thr_t
is obtained using the thread registration function mps_thread_reg
.
Example
mps_thr_t this_thread; mps_res_t res; res = mps_thread_reg(&this_thread, space); if(res != MPS_RES_OK) return res;
See Also
mps_reg_t
,
mps_thread_reg
,
mps_thread_dereg
,
mps_reg_scan_t
,
mps_root_create_reg
,
mps_stack_scan_ambig
4. Undocumented Symbols
The following MPS symbols are used or defined in MPS header files, and intended for client use, but are not yet documented in this reference manual.
mps_arena_t mps_pool_t mps_chain_t mps_root_t mps_ap_t mps_ld_t mps_ss_t mps_alloc_pattern_t mps_frame_t mps_word_t mps_shift_t mps_rm_t MPS_RES_OK MPS_RES_FAIL MPS_RES_RESOURCE MPS_RES_UNIMPL MPS_RES_IO MPS_RES_COMMIT_LIMIT mps_ap_s mps_sac_freelist_block_s mps_sac_s mps_ld_s mps_ss_s mps_fmt_fixed_s MPS_BEGIN MPS_END mps_arena_step mps_arena_start_collect mps_arena_destroy mps_arena_reserved mps_arena_has_addr mps_arena_extend mps_arena_retract mps_fmt_create_fixed mps_fmt_destroy mps_pool_create mps_pool_create_v mps_pool_destroy mps_gen_param_s mps_chain_create mps_chain_destroy mps_alloc_v mps_ap_create mps_ap_create_v mps_ap_destroy mps_reserve mps_commit mps_ap_fill mps_ap_fill_with_reservoir_permit mps_ap_trip MPS_SAC_ALLOC MPS_SAC_FREE mps_reservoir_limit_set mps_reservoir_limit mps_reservoir_available mps_reserve_with_reservoir_permit MPS_RESERVE_BLOCK MPS_RESERVE_WITH_RESERVOIRf_PERMIT_BLOCK mps_root_destroy mps_tramp_t mps_tramp mps_thread_reg mps_thread_dereg mps_ld_reset mps_ld_add mps_ld_merge mps_ld_isstale mps_collections mps_definalize mps_pool_check_free_space mps_lib_get_EOF mps_lib_stream_s mps_lib_get_stderr mps_lib_get_stdout mps_lib_fputc mps_lib_fputs mps_lib_assert_fail mps_clock_t mps_clock mps_clocks_per_sec mps_class_amcz mps_class_ams mps_class_ams_debug mps_class_awl mps_class_lo mps_mv_free_size mps_mv_size mps_class_mv mps_class_mv_debug mps_mvt_free_size mps_mvt_size mps_mvff_free_size mps_mvff_size mps_class_mvff_debug mps_SEH_filter mps_SEH_handler mps_io_t mps_io_create mps_io_destroy mps_io_write mps_io_flush MPS_PF_STRING MPS_PF_ALIGN MPS_ARCH_60 MPS_ARCH_I3 MPS_ARCH_I4 MPS_ARCH_M2 MPS_ARCH_M4 MPS_ARCH_PP MPS_ARCH_S8 MPS_ARCH_S9 MPS_BUILD_AC MPS_BUILD_CC MPS_BUILD_GC MPS_BUILD_LC MPS_BUILD_MV MPS_BUILD_MW MPS_BUILD_SC MPS_OS_FR MPS_OS_I5 MPS_OS_IA MPS_OS_LI MPS_OS_O1 MPS_OS_S7 MPS_OS_SO MPS_OS_SU MPS_OS_W3 MPS_OS_XC MPS_PF_FRI4GC MPS_PF_I5M2CC MPS_PF_IAM4CC MPS_PF_LII4GC MPS_PF_LIPPGC MPS_PF_O1ALCC MPS_PF_O1ALGC MPS_PF_S760AC MPS_PF_S760MW MPS_PF_S7PPAC MPS_PF_S7PPMW MPS_PF_SOS8GC MPS_PF_SOS9SC MPS_PF_SUS8GC MPS_PF_SUS8LC MPS_PF_W3ALMV MPS_PF_W3I3MV MPS_PF_W3PPMV MPS_PF_XCPPGC
A. References
B. Document History
2002-05-27 | RB | Created from individual MPS reference pages, originally written and mainted in Lotus Notes by members of the Memory Management Group of Global Graphics (formerly Harlequin). I found many errors caused by the various conversions that this text has been through. There are probably many more. |
2002-06-17 | RB | Removed Global Graphics specific entries for confidential sources not included in open source release. |
2002-06-18 | NB | Added contents table to section 3. |
2002-06-20 | NB | Quite a bit of proof-reading, to insert missing spaces. Also reformatted for easier editing, including the "See Also" sections. |
2002-06-21 | NB | Removed obsolete symbols. |
C. Copyright and License
This document is copyright © 1997-2002 Ravenbrook Limited. All rights reserved. This is an open source license. Contact Ravenbrook for commercial licensing options.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
- Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
- Redistributions in any form must be accompanied by information on how to obtain complete source code for this software and any accompanying software that uses this software. The source code must either be included in the distribution or be available for no more than the cost of distribution plus a nominal fee, and must be freely redistributable under reasonable conditions. For an executable file, complete source code means the source code for all modules it contains. It does not include source code for modules or files that typically accompany the major components of the operating system on which the executable file runs.
This software is provided by the copyright holders and contributors "as is" and any express or implied warranties, including, but not limited to, the implied warranties of merchantability, fitness for a particular purpose, or non-infringement, are disclaimed. In no event shall the copyright holders and contributors be liable for any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited to, procurement of substitute goods or services; loss of use, data, or profits; or business interruption) however caused and on any theory of liability, whether in contract, strict liability, or tort (including negligence or otherwise) arising in any way out of the use of this software, even if advised of the possibility of such damage.