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

Name

MPS_ARCH_AL

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

mps_fix

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

MPS_FIX1

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

MPS_FIX12

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

MPS_FIX2

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

MPS_FIX_CALL

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

MPS_RES_LIMIT

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

mps_res_t

constant MPS_RES_MEMORY

Name

MPS_RES_MEMORY

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

MPS_RES_PARAM

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

mps_res_t

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

mps_res_t

Notes

Internal Notes

MPS_RM_CONST

Name

MPS_RM_CONST

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

MPS_RM_PROT

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

mps_sac_alloc

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

MPS_SAC_ALLOC_FAST

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

MPS_SAC_CLASS_LIMIT

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

mps_sac_create

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

mps_sac_free

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

MPS_SAC_FREE_FAST

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

MPS_SCAN_BEGIN

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

MPS_SCAN_END

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

MPS_T_WORD

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

MPS_WORD_SHIFT

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

MPS_T_WORD, MPS_WORD_WIDTH

constant MPS_WORD_WIDTH

Name

MPS_WORD_WIDTH

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

MPS_T_WORD, MPS_WORD_SHIFT

type mps_addr_t

Name

mps_addr_t

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

mps_align_t

type mps_align_t

Name

mps_align_t

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

mps_fmt_A_t, mps_addr_t

function mps_alloc

Name

mps_alloc

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_free

mps_alloc_pattern_ramp

Name

mps_alloc_pattern_ramp

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

mps_amc_apply

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

mps_ap_alloc_pattern_begin

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

mps_ap_alloc_pattern_end

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

mps_ap_alloc_pattern_reset

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

mps_ap_frame_pop

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

mps_ap_frame_push

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

mps_arena_clamp

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

mps_arena_class_cl

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

mps_arena_create

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

mps_arena_class_t

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

mps_arena_class_vm

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

mps_arena_class_vmnz

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

mps_arena_collect

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

mps_arena_commit_limit

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

mps_arena_commit_limit_set

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

mps_arena_committed

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

mps_arena_create

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

mps_arena_create_v

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

mps_arena_expose

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

mps_arena_has_addr

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

mps_arena_park

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

mps_arena_release

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

mps_arena_roots_walk

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_roots_stepper_t

mps_arena_formatted_objects_walk

Notes

function mps_arena_spare_commit_limit

Name

mps_arena_spare_commit_limit

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

mps_arena_spare_commit_limit

Notes

None.

function mps_arena_spare_committed

Name

mps_arena_spare_committed

Summary

Returns the number of bytes of spare committed memory.

Associated Protocols

Memory

Type

size_t mps_arena_spare_committed(mps_arena_t);

Arguments

mps_arena_t mps_arena

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

mps_bool_t

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

mps_class_amc

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

mps_class_mvff

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:

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

mps_class_snc

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

mps_class_mvt

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

mps_pool_create

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

mps_class_t

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

mps_finalize

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_destroyshould 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

mps_fmt_A_s

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

mps_fmt_A_t

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

mps_fmt_B_s

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

mps_fmt_B_t

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

mps_fmt_auto_header_s

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_shas 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

mps_fmt_class_t

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

mps_fmt_t, mps_fmt_create_B

Notes

It is recommended that NULL be returned for padding objects and forwarded objects.

type mps_fmt_copy_t

Name

mps_fmt_copy_t

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

mps_fmt_create_A

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

mps_fmt_create_B

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

mps_fmt_create_auto_header

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_headerand 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

mps_fmt_fwd_t

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

mps_fmt_isfwd_t

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

mps_fmt_pad_t

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

mps_fmt_A_s, mps_fmt_B_s

type mps_fmt_scan_t

Name

mps_fmt_scan_t

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

mps_fmt_skip_t

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

mps_fmt_t

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

mps_free

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

mps_alloc

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

mps_lib_memcmp

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

mps_lib_memcpy

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

mps_lib_memset

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

mps_lib_telemetry_control

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

mps_telemetry_control

function mps_message_discard

Name

mps_message_discard

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

mps_message_get

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

mps_message_finalization_ref

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

mps_message_gc_condemned_size

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

mps_message_gc_live_size

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

mps_message_get

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

mps_message_poll

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

mps_message_get

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

mps_message_queue_type

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

mps_message_t

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

mps_message_type

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

mps_message_type_disable

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

mps_message_type_enable

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

mps_message_type_finalization

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

mps_message_type_gc

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 message

  • mps_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

mps_message_type_t

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

mps_pool_check_fenceposts

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

mps_pool_debug_option_s

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

mps_pool_check_fenceposts

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

mps_rank_ambig

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

mps_rank_t, mps_rank_exact

function mps_rank_exact

Name

mps_rank_exact

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

mps_rank_t

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

mps_rank_weak

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

mps_rank_t, mps_rank_exact

type mps_reg_scan_t

Name

mps_reg_scan_t

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

mps_res_t

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

mps_root_create

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

mps_root_create_fmt

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

mps_root_create_reg

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

mps_root_create_table

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

mps_root_create_table_masked

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

mps_root_scan_t

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

mps_roots_stepper_t

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

mps_arena_roots_arena_walk

Notes

type mps_sac_class_s

Name

mps_sac_class_s

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

mps_sac_create

Notes

Any blocks whose size falls between two classes are allocated from the larger class.

function mps_sac_create

Name

mps_sac_create

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

mps_sac_destroy

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

mps_sac_create, mps_sac_t

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

mps_sac_flush

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

mps_sac_t

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

mps_sac_t

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

mps_stack_scan_ambig

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

mps_telemetry_control

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

mps_lib_telemetry_control

function mps_telemetry_flush

Name

mps_telemetry_flush

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

mps_lib_io_flush

function mps_telemetry_intern

Name

mps_telemetry_intern

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

mps_telemetry_label

mps_telemetry_control

function mps_telemetry_label

Name

mps_telemetry_label

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

mps_thr_t

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.