2018-04-10 17:03:59 +02:00
|
|
|
/*
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
* Copyright(c) 2006 to 2019 ADLINK Technology Limited and others
|
2018-04-10 17:03:59 +02:00
|
|
|
*
|
|
|
|
|
* This program and the accompanying materials are made available under the
|
|
|
|
|
* terms of the Eclipse Public License v. 2.0 which is available at
|
|
|
|
|
* http://www.eclipse.org/legal/epl-2.0, or the Eclipse Distribution License
|
|
|
|
|
* v. 1.0 which is available at
|
|
|
|
|
* http://www.eclipse.org/org/documents/edl-v10.php.
|
|
|
|
|
*
|
|
|
|
|
* SPDX-License-Identifier: EPL-2.0 OR BSD-3-Clause
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
#ifndef DDS__HANDLES_H
|
|
|
|
|
#define DDS__HANDLES_H
|
2018-04-10 17:03:59 +02:00
|
|
|
|
2019-01-18 14:10:19 +01:00
|
|
|
#include "dds/ddsrt/time.h"
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
#include "dds/ddsrt/retcode.h"
|
|
|
|
|
#include "dds/ddsrt/atomics.h"
|
|
|
|
|
#include "dds/dds.h"
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
#if defined (__cplusplus)
|
|
|
|
|
extern "C" {
|
|
|
|
|
#endif
|
|
|
|
|
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
struct dds_entity;
|
|
|
|
|
|
2018-04-10 17:03:59 +02:00
|
|
|
/********************************************************************************************
|
|
|
|
|
*
|
|
|
|
|
* TODO CHAM-138: Header file improvements
|
|
|
|
|
* - Remove some internal design decisions and masks from the header file.
|
|
|
|
|
* - Improve function headers where needed.
|
|
|
|
|
*
|
|
|
|
|
********************************************************************************************/
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Short working ponderings.
|
|
|
|
|
*
|
|
|
|
|
* A handle will be created with a related object. If you want to protect that object by
|
|
|
|
|
* means of a mutex, it's your responsibility, not handleservers'.
|
|
|
|
|
*
|
|
|
|
|
* The handle server keeps an 'active claim' count. Every time a handle is claimed, the
|
|
|
|
|
* count is increase and decreased when the handle is released. A handle can only be
|
|
|
|
|
* deleted when there's no more active claim. You can use this to make sure that nobody
|
|
|
|
|
* is using the handle anymore, which should mean that nobody is using the related object
|
|
|
|
|
* anymore. So, when you can delete the handle, you can safely delete the related object.
|
|
|
|
|
*
|
|
|
|
|
* To prevent new claims (f.i. when you want to delete the handle), you can close the
|
|
|
|
|
* handle. This will not remove any information within the handleserver, it just prevents
|
|
|
|
|
* new claims. The delete will actually free handleserver internal memory.
|
|
|
|
|
*
|
|
|
|
|
* There's currently a global lock in the handle server that is used with every API call.
|
|
|
|
|
* Maybe the internals can be improved to circumvent the need for that...
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
typedef int32_t dds_handle_t;
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The handle link type.
|
|
|
|
|
*
|
|
|
|
|
* The lookup of an handle should be very quick.
|
|
|
|
|
* But to increase the performance just a little bit more, it is possible to
|
|
|
|
|
* acquire a handlelink. This can be used by the handleserver to circumvent
|
|
|
|
|
* the lookup and go straight to the handle related information.
|
|
|
|
|
*
|
|
|
|
|
* Almost every handleserver function supports the handlelink. You don't have
|
|
|
|
|
* to provide the link. When it is NULL, the normal lookup will be done.
|
|
|
|
|
*
|
|
|
|
|
* This handlelink is invalid after the related handle is deleted and should
|
|
|
|
|
* never be used afterwards.
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
struct dds_handle_link {
|
|
|
|
|
dds_handle_t hdl;
|
|
|
|
|
ddsrt_atomic_uint32_t cnt_flags;
|
|
|
|
|
};
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Initialize handleserver singleton.
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
DDS_EXPORT dds_return_t
|
2019-06-27 10:45:38 +02:00
|
|
|
dds_handle_server_init(void);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Destroy handleserver singleton.
|
|
|
|
|
* The handleserver is destroyed when fini() is called as often as init().
|
|
|
|
|
*/
|
2019-01-18 14:10:19 +01:00
|
|
|
DDS_EXPORT void
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
dds_handle_server_fini(void);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* This creates a new handle that contains the given type and is linked to the
|
|
|
|
|
* user data.
|
|
|
|
|
*
|
|
|
|
|
* A kind value != 0 has to be provided, just to make sure that no 0 handles
|
2019-04-16 10:23:32 +02:00
|
|
|
* will be created. It should also fit the DDSRT_HANDLE_KIND_MASK.
|
2018-04-10 17:03:59 +02:00
|
|
|
* In other words handle creation will fail if
|
2019-04-16 10:23:32 +02:00
|
|
|
* ((kind & ~DDSRT_HANDLE_KIND_MASK != 0) || (kind & DDSRT_HANDLE_KIND_MASK == 0)).
|
2018-04-10 17:03:59 +02:00
|
|
|
*
|
|
|
|
|
* It has to do something clever to make sure that a deleted handle is not
|
|
|
|
|
* re-issued very quickly after it was deleted.
|
|
|
|
|
*
|
|
|
|
|
* kind - The handle kind, provided by the client.
|
|
|
|
|
* arg - The user data linked to the handle (may be NULL).
|
|
|
|
|
*
|
|
|
|
|
* Valid handle when returned value is positive.
|
|
|
|
|
* Otherwise negative handle is returned.
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
DDS_EXPORT dds_handle_t
|
|
|
|
|
dds_handle_create(
|
|
|
|
|
struct dds_handle_link *link);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* This will close the handle. All information remains, only new claims will
|
|
|
|
|
* fail.
|
|
|
|
|
*
|
|
|
|
|
* This is a noop on an already closed handle.
|
|
|
|
|
*/
|
2019-01-18 14:10:19 +01:00
|
|
|
DDS_EXPORT void
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
dds_handle_close(
|
|
|
|
|
struct dds_handle_link *link);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* This will remove the handle related information from the server administration
|
|
|
|
|
* to free up space.
|
|
|
|
|
*
|
|
|
|
|
* This is an implicit close().
|
|
|
|
|
*
|
|
|
|
|
* It will delete the information when there are no more active claims. It'll
|
|
|
|
|
* block when necessary to wait for all possible claims to be released.
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
DDS_EXPORT int32_t
|
|
|
|
|
dds_handle_delete(
|
|
|
|
|
struct dds_handle_link *link,
|
|
|
|
|
dds_time_t timeout);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If the a valid handle is given, which matches the kind and it is not closed,
|
|
|
|
|
* then the related arg will be provided and the claims count is increased.
|
|
|
|
|
*
|
|
|
|
|
* Returns OK when succeeded.
|
|
|
|
|
*/
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
DDS_EXPORT int32_t
|
Address locking order for entity locks
There were inconsistencies in the order in which entity locks were taken
when multiple entities needed to be locked at the same time. In most
cases, the order was first locking entity X, then locking the parent
entity of X. However, in some cases the order was reversed, a likely
cause of deadlocks.
This commit sorts these problems, and in particular propagating
operations into children. The entity refcount is now part of the handle
administration so that it is no longer necessary to lock an entity to
determine whether it is still allowed to be used (previously it had to
check the CLOSED flag afterward). This allows recursing into the
children while holding handles and the underlying objects alive, but
without violating lock order.
Attendant changes that would warrant there own commits but are too hard
to split off:
* Children are now no longer in a singly linked list, but in an AVL
tree; this was necessary at some intermediate stage to allow unlocking
an entity and restarting iteration over all children at the "next"
child (all thanks to the eternally unique instance handle);
* Waitsets shifted to using arrays of attached entities instead of
linked lists; this was a consequence of dealing with some locking
issues in reading triggers and considering which operations on the
"triggered" and "observed" sets are actually needed.
* Entity status flags and waitset/condition trigger counts are now
handled using atomic operations. Entities are now classified as
having a "status" with a corresponding mask, or as having a "trigger
count" (conditions). As there are fewer than 16 status bits, the
status and its mask can squeeze into the same 32-bits as the trigger
count. These atomic updates avoid the need for a separate lock just
for the trigger/status values and results in a significant speedup
with waitsets.
* Create topic now has a more rational behaviour when multiple
participants attempt to create the same topic: each participant now
gets its own topic definition, but the underlying type representation
is shared.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-06-21 07:13:57 +02:00
|
|
|
dds_handle_pin(
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
dds_handle_t hdl,
|
|
|
|
|
struct dds_handle_link **entity);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DDS_EXPORT void
|
Address locking order for entity locks
There were inconsistencies in the order in which entity locks were taken
when multiple entities needed to be locked at the same time. In most
cases, the order was first locking entity X, then locking the parent
entity of X. However, in some cases the order was reversed, a likely
cause of deadlocks.
This commit sorts these problems, and in particular propagating
operations into children. The entity refcount is now part of the handle
administration so that it is no longer necessary to lock an entity to
determine whether it is still allowed to be used (previously it had to
check the CLOSED flag afterward). This allows recursing into the
children while holding handles and the underlying objects alive, but
without violating lock order.
Attendant changes that would warrant there own commits but are too hard
to split off:
* Children are now no longer in a singly linked list, but in an AVL
tree; this was necessary at some intermediate stage to allow unlocking
an entity and restarting iteration over all children at the "next"
child (all thanks to the eternally unique instance handle);
* Waitsets shifted to using arrays of attached entities instead of
linked lists; this was a consequence of dealing with some locking
issues in reading triggers and considering which operations on the
"triggered" and "observed" sets are actually needed.
* Entity status flags and waitset/condition trigger counts are now
handled using atomic operations. Entities are now classified as
having a "status" with a corresponding mask, or as having a "trigger
count" (conditions). As there are fewer than 16 status bits, the
status and its mask can squeeze into the same 32-bits as the trigger
count. These atomic updates avoid the need for a separate lock just
for the trigger/status values and results in a significant speedup
with waitsets.
* Create topic now has a more rational behaviour when multiple
participants attempt to create the same topic: each participant now
gets its own topic definition, but the underlying type representation
is shared.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-06-21 07:13:57 +02:00
|
|
|
dds_handle_repin(
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
struct dds_handle_link *link);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* The active claims count is decreased.
|
|
|
|
|
*/
|
2019-01-18 14:10:19 +01:00
|
|
|
DDS_EXPORT void
|
Address locking order for entity locks
There were inconsistencies in the order in which entity locks were taken
when multiple entities needed to be locked at the same time. In most
cases, the order was first locking entity X, then locking the parent
entity of X. However, in some cases the order was reversed, a likely
cause of deadlocks.
This commit sorts these problems, and in particular propagating
operations into children. The entity refcount is now part of the handle
administration so that it is no longer necessary to lock an entity to
determine whether it is still allowed to be used (previously it had to
check the CLOSED flag afterward). This allows recursing into the
children while holding handles and the underlying objects alive, but
without violating lock order.
Attendant changes that would warrant there own commits but are too hard
to split off:
* Children are now no longer in a singly linked list, but in an AVL
tree; this was necessary at some intermediate stage to allow unlocking
an entity and restarting iteration over all children at the "next"
child (all thanks to the eternally unique instance handle);
* Waitsets shifted to using arrays of attached entities instead of
linked lists; this was a consequence of dealing with some locking
issues in reading triggers and considering which operations on the
"triggered" and "observed" sets are actually needed.
* Entity status flags and waitset/condition trigger counts are now
handled using atomic operations. Entities are now classified as
having a "status" with a corresponding mask, or as having a "trigger
count" (conditions). As there are fewer than 16 status bits, the
status and its mask can squeeze into the same 32-bits as the trigger
count. These atomic updates avoid the need for a separate lock just
for the trigger/status values and results in a significant speedup
with waitsets.
* Create topic now has a more rational behaviour when multiple
participants attempt to create the same topic: each participant now
gets its own topic definition, but the underlying type representation
is shared.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-06-21 07:13:57 +02:00
|
|
|
dds_handle_unpin(
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
struct dds_handle_link *link);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Check if the handle is closed.
|
|
|
|
|
*
|
|
|
|
|
* This is only useful when you have already claimed a handle and it is
|
|
|
|
|
* possible that another thread is trying to delete the handle while you
|
|
|
|
|
* were (for instance) sleeping or waiting for something. Now you can
|
|
|
|
|
* break of your process and release the handle, making the deletion
|
|
|
|
|
* possible.
|
|
|
|
|
*/
|
2019-01-18 14:10:19 +01:00
|
|
|
DDS_EXPORT bool
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
dds_handle_is_closed(
|
|
|
|
|
struct dds_handle_link *link);
|
2018-04-10 17:03:59 +02:00
|
|
|
|
|
|
|
|
|
Address locking order for entity locks
There were inconsistencies in the order in which entity locks were taken
when multiple entities needed to be locked at the same time. In most
cases, the order was first locking entity X, then locking the parent
entity of X. However, in some cases the order was reversed, a likely
cause of deadlocks.
This commit sorts these problems, and in particular propagating
operations into children. The entity refcount is now part of the handle
administration so that it is no longer necessary to lock an entity to
determine whether it is still allowed to be used (previously it had to
check the CLOSED flag afterward). This allows recursing into the
children while holding handles and the underlying objects alive, but
without violating lock order.
Attendant changes that would warrant there own commits but are too hard
to split off:
* Children are now no longer in a singly linked list, but in an AVL
tree; this was necessary at some intermediate stage to allow unlocking
an entity and restarting iteration over all children at the "next"
child (all thanks to the eternally unique instance handle);
* Waitsets shifted to using arrays of attached entities instead of
linked lists; this was a consequence of dealing with some locking
issues in reading triggers and considering which operations on the
"triggered" and "observed" sets are actually needed.
* Entity status flags and waitset/condition trigger counts are now
handled using atomic operations. Entities are now classified as
having a "status" with a corresponding mask, or as having a "trigger
count" (conditions). As there are fewer than 16 status bits, the
status and its mask can squeeze into the same 32-bits as the trigger
count. These atomic updates avoid the need for a separate lock just
for the trigger/status values and results in a significant speedup
with waitsets.
* Create topic now has a more rational behaviour when multiple
participants attempt to create the same topic: each participant now
gets its own topic definition, but the underlying type representation
is shared.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-06-21 07:13:57 +02:00
|
|
|
DDS_EXPORT void dds_handle_add_ref (struct dds_handle_link *link);
|
|
|
|
|
DDS_EXPORT bool dds_handle_drop_ref (struct dds_handle_link *link);
|
|
|
|
|
|
2018-04-10 17:03:59 +02:00
|
|
|
#if defined (__cplusplus)
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
lift limits on handle allocation and reuse (#95)
The old entity handle mechanism suffered from a number of problems, the
most terrible one being that it would only ever allocate 1000 handles
(not even have at most 1000 in use at the same time). Secondarily, it
was protected by a single mutex that actually does show up as a limiting
factor in, say, a polling-based throughput test with small messages.
Thirdly, it tried to provide for various use cases that don't exist in
practice but add complexity and overhead.
This commit totally rewrites the mechanism, by replacing the old array
with a hash table and allowing a near-arbitrary number of handles as
well as reuse of handles. It also removes the entity "kind" bits in the
most significant bits of the handles, because they only resulted in
incorrect checking of argument validity. All that is taken out, but
there is still more cleaning up to be done. It furthermore removes an
indirection in the handle-to-entity lookup by embedding the
"dds_handle_link" structure in the entity.
Handle allocation is randomized to avoid the have a high probability of
quickly finding an available handle (the total number of handles is
limited to a number much smaller than the domain from which they are
allocated). The likelihood of handle reuse is still dependent on the
number of allocated handles -- the fewer handles there are, the longer
the expected time to reuse. Non-randomized handles would give a few
guarantees more, though.
It moreover moves the code from the "util" to the "core/ddsc" component,
because it really is only used for entities, and besides the new
implementation relies on the deferred freeing (a.k.a. garbage collection
mechanism) implemented in the core.
The actual handle management has two variants, selectable with a macro:
the preferred embodiment uses a concurrent hash table, the actually used
one performs all operations inside a single mutex and uses a
non-concurrent version of the hash table. The reason the
less-predeferred embodiment is used is that the concurrent version
requires the freeing of entity objects to be deferred (much like the
GUID-to-entity hash tables in DDSI function, or indeed the key value to
instance handle mapping). That is a fair bit of work, and the
non-concurrent version is a reasonable intermediate step.
Signed-off-by: Erik Boasson <eb@ilities.com>
2019-02-19 10:57:21 +01:00
|
|
|
#endif /* DDS__HANDLES_H */
|