10000 [pull] master from postgres:master by pull[bot] · Pull Request #139 · Haofei/postgres · GitHub
[go: up one dir, main page]

Skip to content

[pull] master from postgres:master #139

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 5 commits into from
Jul 2, 2025
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
Add GetNamedDSA() and GetNamedDSHash().
Presently, the dynamic shared memory (DSM) registry only provides
GetNamedDSMSegment(), which allocates a fixed-size segment.  To use
the DSM registry for more sophisticated things like dynamic shared
memory areas (DSAs) or a hash table backed by a DSA (dshash), users
need to create a DSM segment that stores various handles and LWLock
tranche IDs and to write fairly complicated initialization code.
Furthermore, there is likely little variation in this
initialization code between libraries.

This commit introduces functions that simplify allocating a DSA or
dshash within the DSM registry.  These functions are very similar
to GetNamedDSMSegment().  Notable differences include the lack of
an initialization callback parameter and the prohibition of calling
the functions more than once for a given entry in each backend
(which should be trivially avoidable in most circumstances).  While
at it, this commit bumps the maximum DSM registry entry name length
from 63 bytes to 127 bytes.

Also note that even though one could presumably detach/destroy the
DSAs and dshashes created in the registry, such use-cases are not
yet well-supported, if for no other reason than the associated DSM
registry entries cannot be removed.  Adding such support is left as
a future exercise.

The test_dsm_registry test module contains tests for the new
functions and also serves as a complete usage example.

Reviewed-by: Dagfinn Ilmari Mannsåker <ilmari@ilmari.org>
Reviewed-by: Sami Imseih <samimseih@gmail.com>
Reviewed-by: Florents Tselai <florents.tselai@gmail.com>
Reviewed-by: Rahila Syed <rahilasyed90@gmail.com>
Discussion: https://postgr.es/m/aEC8HGy2tRQjZg_8%40nathan
  • Loading branch information
nathan-bossart committed Jul 2, 2025
commit fe07100e82b096d3c848cace790d4b4daf0c4131
265 changes: 251 additions & 14 deletions src/backend/storage/ipc/dsm_registry.c
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,20 @@
* current backend. This function guarantees that only one backend
* initializes the segment and that all other backends just attach it.
*
* A DSA can be created in or retrieved from the registry by calling
* GetNamedDSA(). As with GetNamedDSMSegment(), if a DSA with the provided
* name does not yet exist, it is created. Otherwise, GetNamedDSA()
* ensures the DSA is attached to the current backend. This function
* guarantees that only one backend initializes the DSA and that all other
* backends just attach it.
*
* A dshash table can be created in or retrieved from the registry by
* calling GetNamedDSHash(). As with GetNamedDSMSegment(), if a hash
* table with the provided name does not yet exist, it is created.
* Otherwise, GetNamedDSHash() ensures the hash table is attached to the
* current backend. This function guarantees that only one backend
* initializes the table and that all other backends just attach it.
*
* Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
Expand All @@ -32,6 +46,12 @@
#include "storage/shmem.h"
#include "utils/memutils.h"

#define DSMR_NAME_LEN 128

#define DSMR_DSA_TRANCHE_SUFFIX " DSA"
#define DSMR_DSA_TRANCHE_SUFFIX_LEN (sizeof(DSMR_DSA_TRANCHE_SUFFIX) - 1)
#define DSMR_DSA_TRANCHE_NAME_LEN (DSMR_NAME_LEN + DSMR_DSA_TRANCHE_SUFFIX_LEN)

typedef struct DSMRegistryCtxStruct
{
dsa_handle dsah;
Expand All @@ -40,15 +60,48 @@ typedef struct DSMRegistryCtxStruct

static DSMRegistryCtxStruct *DSMRegistryCtx;

typedef struct DSMRegistryEntry
typedef struct NamedDSMState
{
char name[64];
dsm_handle handle;
size_t size;
} NamedDSMState;

typedef struct NamedDSAState
{
dsa_handle handle;
int tranche;
char tranche_name[DSMR_DSA_TRANCHE_NAME_LEN];
} NamedDSAState;

typedef struct NamedDSHState
{
NamedDSAState dsa;
dshash_table_handle handle;
int tranche;
char tranche_name[DSMR_NAME_LEN];
} NamedDSHState;

typedef enum DSMREntryType
{
DSMR_ENTRY_TYPE_DSM,
DSMR_ENTRY_TYPE_DSA,
DSMR_ENTRY_TYPE_DSH,
} DSMREntryType;

typedef struct DSMRegistryEntry
{
char name[DSMR_NAME_LEN];
DSMREntryType type;
union
{
NamedDSMState dsm;
NamedDSAState dsa;
NamedDSHState dsh;
} data;
} DSMRegistryEntry;

static const dshash_parameters dsh_params = {
offsetof(DSMRegistryEntry, handle),
offsetof(DSMRegistryEntry, type),
sizeof(DSMRegistryEntry),
dshash_strcmp,
dshash_strhash,
Expand Down Expand Up @@ -141,7 +194,7 @@ GetNamedDSMSegment(const char *name, size_t size,
ereport(ERROR,
(errmsg("DSM segment name cannot be empty")));

if (strlen(name) >= offsetof(DSMRegistryEntry, handle))
if (strlen(name) >= offsetof(DSMRegistryEntry, type))
ereport(ERROR,
(errmsg("DSM segment name too long")));

Expand All @@ -158,32 +211,39 @@ GetNamedDSMSegment(const char *name, size_t size,
entry = dshash_find_or_insert(dsm_registry_table, name, found);
if (!(*found))
{
NamedDSMState *state = &entry->data.dsm;
dsm_segment *seg;

entry->type = DSMR_ENTRY_TYPE_DSM;

/* Initialize the segment. */
dsm_segment *seg = dsm_create(size, 0);
seg = dsm_create(size, 0);

dsm_pin_segment(seg);
dsm_pin_mapping(seg);
entry->handle = dsm_segment_handle(seg);
entry->size = size;
state->handle = dsm_segment_handle(seg);
state->size = size;
ret = dsm_segment_address(seg);

if (init_callback)
(*init_callback) (ret);
}
else if (entry->size != size)
{
else if (entry->type != DSMR_ENTRY_TYPE_DSM)
ereport(ERROR,
(errmsg("requested DSM segment size does not match size of "
"existing segment")));
}
(errmsg("requested DSM segment does not match type of existing entry")));
else if (entry->data.dsm.size != size)
ereport(ERROR,
(errmsg("requested DSM segment size does not match size of existing segment")));
else
{
dsm_segment *seg = dsm_find_mapping(entry->handle);
NamedDSMState *state = &entry->data.dsm;
dsm_segment *seg;

/* If the existing segment is not already attached, attach it now. */
seg = dsm_find_mapping(state->handle);
if (seg == NULL)
{
seg = dsm_attach(entry->handle);
seg = dsm_attach(state->handle);
if (seg == NULL)
elog(ERROR, "could not map dynamic shared memory segment");

Expand All @@ -198,3 +258,180 @@ GetNamedDSMSegment(const char *name, size_t size,

return ret;
}

/*
* Initialize or attach a named DSA.
*
* This routine returns a pointer to the DSA. A new LWLock tranche ID will be
* generated if needed. Note that the lock tranche will be registered with the
* provided name. Also note that this should be called at most once for a
* given DSA in each backend.
*/
dsa_area *
GetNamedDSA(const char *name, bool *found)
{
DSMRegistryEntry *entry;
MemoryContext oldcontext;
dsa_area *ret;

Assert(found);

if (!name || *name == '\0')
ereport(ERROR,
(errmsg("DSA name cannot be empty")));

if (strlen(name) >= offsetof(DSMRegistryEntry, type))
ereport(ERROR,
(errmsg("DSA name too long")));

/* Be sure any local memory allocated by DSM/DSA routines is persistent. */
oldcontext = MemoryContextSwitchTo(TopMemoryContext);

/* Connect to the registry. */
init_dsm_registry();

entry = dshash_find_or_insert(dsm_registry_table, name, found);
if (!(*found))
{
NamedDSAState *state = &entry->data.dsa;

entry->type = DSMR_ENTRY_TYPE_DSA;

/* Initialize the LWLock tranche for the DSA. */
state->tranche = LWLockNewTrancheId();
strcpy(state->tranche_name, name);
LWLockRegisterTranche(state->tranche, state->tranche_name);

/* Initialize the DSA. */
ret = dsa_create(state->tranche);
dsa_pin(ret);
dsa_pin_mapping(ret);

/* Store handle for other backends to use. */
state->handle = dsa_get_handle(ret);
}
else if (entry->type != DSMR_ENTRY_TYPE_DSA)
ereport(ERROR,
(errmsg("requested DSA does not match type of existing entry")));
else
{
NamedDSAState *state = &entry->data.dsa;

if (dsa_is_attached(state->handle))
ereport(ERROR,
(errmsg("requested DSA already attached to current process")));

/* Initialize existing LWLock tranche for the DSA. */
LWLockRegisterTranche(state->tranche, state->tranche_name);

/* Attach to existing DSA. */
ret = dsa_attach(state->handle);
dsa_pin_mapping(ret);
}

dshash_release_lock(dsm_registry_table, entry);
MemoryContextSwitchTo(oldcontext);

return ret;
}

/*
* Initialize or attach a named dshash table.
*
* This routine returns the address of the table. The tranche_id member of
* params is ignored; new tranche IDs will be generated if needed. Note that
* the DSA lock tranche will be registered with the provided name with " DSA"
* appended. The dshash lock tranche will be registered with the provided
* name. Also note that this should be called at most once for a given table
* in each backend.
*/
dshash_table *
GetNamedDSHash(const char *name, const dshash_parameters *params, bool *found)
{
DSMRegistryEntry *entry;
MemoryContext oldcontext;
dshash_table *ret;

Assert(params);
Assert(found);

if (!name || *name == '\0')
ereport(ERROR,
(errmsg("DSHash name cannot be empty")));

if (strlen(name) >= offsetof(DSMRegistryEntry, type))
ereport(ERROR,
(errmsg("DSHash name too long")));

/* Be sure any local memory allocated by DSM/DSA routines is persistent. */
oldcontext = MemoryContextSwitchTo(TopMemoryContext);

/* Connect to the registry. */
init_dsm_registry();

entry = dshash_find_or_insert(dsm_registry_table, name, found);
if (!(*found))
{
NamedDSAState *dsa_state = &entry->data.dsh.dsa;
NamedDSHState *dsh_state = &entry->data.dsh;
dshash_parameters params_copy;
dsa_area *dsa;

entry->type = DSMR_ENTRY_TYPE_DSH;

/* Initialize the LWLock tranche for the DSA. */
dsa_state->tranche = LWLockNewTrancheId();
sprintf(dsa_state->tranche_name, "%s%s", name, DSMR_DSA_TRANCHE_SUFFIX);
LWLockRegisterTranche(dsa_state->tranche, dsa_state->tranche_name);

/* Initialize the LWLock tranche for the dshash table. */
dsh_state->tranche = LWLockNewTrancheId();
strcpy(dsh_state->tranche_name, name);
LWLockRegisterTranche(dsh_state->tranche, dsh_state->tranche_name);

/* Initialize the DSA for the hash table. */
dsa = dsa_create(dsa_state->tranche);
dsa_pin(dsa);
dsa_pin_mapping(dsa);

/* Initialize the dshash table. */
memcpy(&params_copy, params, sizeof(dshash_parameters));
params_copy.tranche_id = dsh_state->tranche;
ret = dshash_create(dsa, &params_copy, NULL);

/* Store handles for other backends to use. */
dsa_state->handle = dsa_get_handle(dsa);
dsh_state->handle = dshash_get_hash_table_handle(ret);
}
else if (entry->type != DSMR_ENTRY_TYPE_DSH)
ereport(ERROR,
(errmsg("requested DSHash does not match type of existing entry")));
else
{
NamedDSAState *dsa_state = &entry->data.dsh.dsa;
NamedDSHState *dsh_state = &entry->data.dsh;
dsa_area *dsa;

/* XXX: Should we verify params matches what table was created with? */

if (dsa_is_attached(dsa_state->handle))
ereport(ERROR,
(errmsg("requested DSHash already attached to current process")));

/* Initialize existing LWLock tranches for the DSA and dshash table. */
LWLockRegisterTranche(dsa_state->tranche, dsa_state->tranche_name);
LWLockRegisterTranche(dsh_state->tranche, dsh_state->tranche_name);

/* Attach to existing DSA for the hash table. */
dsa = dsa_attach(dsa_state->handle);
dsa_pin_mapping(dsa);

/* Attach to existing dshash table. */
ret = dshash_attach(dsa, params, dsh_state->handle, NULL);
}

dshash_release_lock(dsm_registry_table, entry);
MemoryContextSwitchTo(oldcontext);

return ret;
}
15 changes: 15 additions & 0 deletions src/backend/utils/mmgr/dsa.c
Original file line number Diff line number Diff line change
Expand Up @@ -531,6 +531,21 @@ dsa_attach(dsa_handle handle)
return area;
}

/*
* Returns whether the area with the given handle was already attached by the
* current process. The area must have been created with dsa_create (not
* dsa_create_in_place).
*/
bool
dsa_is_attached(dsa_handle handle)
{
/*
* An area handle is really a DSM segment handle for the first segment, so
* we can just search for that.
*/
return dsm_find_mapping(handle) != NULL;
}

/*
* Attach to an area that was created with dsa_create_in_place. The caller
* must somehow know the location in memory that was used when the area was
Expand Down
7 changes: 6 additions & 1 deletion src/include/storage/dsm_registry.h
Original file line number Diff line number Diff line change
Expand Up @@ -13,10 +13,15 @@
#ifndef DSM_REGISTRY_H
#define DSM_REGISTRY_H

#include "lib/dshash.h"

extern void *GetNamedDSMSegment(const char *name, size_t size,
void (*init_callback) (void *ptr),
bool *found);

extern dsa_area *GetNamedDSA(const char *name, bool *found);
extern dshash_table *GetNamedDSHash(const char *name,
const dshash_parameters *params,
bool *found);
extern Size DSMRegistryShmemSize(void);
extern void DSMRegistryShmemInit(void);

Expand Down
1 change: 1 addition & 0 deletions src/include/utils/dsa.h
Original file line number Diff line number Diff line change
Expand Up @@ -145,6 +145,7 @@ extern dsa_area *dsa_create_in_place_ext(void *place, size_t size,
size_t init_segment_size,
size_t max_segment_size);
extern dsa_area *dsa_attach(dsa_handle handle);
extern bool dsa_is_attached(dsa_handle handle);
extern dsa_area *dsa_attach_in_place(void *place, dsm_segment *segment);
extern void dsa_release_in_place(void *place);
extern void dsa_on_dsm_detach_release_in_place(dsm_segment *, Datum);
Expand Down
Loading
0