sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
4b6d4de972
nuttx/mm/iob/iob.h

148 lines
5.1 KiB
C
Raw Normal View History

NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
/****************************************************************************
IOBs: Move from driver/iob to a better location in mm/iob
2017-05-09 15:34:59 +02:00
* mm/iob/iob.h
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
*
mm: Author Gregory Nutt: update licenses to Apache Gregory Nutt has submitted the SGA and we can mograte the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-02-08 13:50:10 +01:00
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
*
mm: Author Gregory Nutt: update licenses to Apache Gregory Nutt has submitted the SGA and we can mograte the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-02-08 13:50:10 +01:00
* http://www.apache.org/licenses/LICENSE-2.0
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
*
mm: Author Gregory Nutt: update licenses to Apache Gregory Nutt has submitted the SGA and we can mograte the licenses to Apache. Signed-off-by: Alin Jerpelea <alin.jerpelea@sony.com>
2021-02-08 13:50:10 +01:00
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
*
****************************************************************************/
IOBs: Move from driver/iob to a better location in mm/iob
2017-05-09 15:34:59 +02:00
#ifndef __MM_IOB_IOB_H
Refine the preprocessor conditional guard style (#190)
2020-01-31 19:07:39 +01:00
#define __MM_IOB_IOB_H
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
Move net/iob to drivers/iob so that the I/O buffering feature can be available to other drivers when networking is disabled.
2017-04-21 00:08:49 +02:00
#include <debug.h>
First cut at conversion of write-buffering to use I/O buffer chaings (IOBs)
2014-06-22 19:27:57 +02:00
IOBs: Move from driver/iob to a better location in mm/iob
2017-05-09 15:34:59 +02:00
#include <nuttx/mm/iob.h>
Replace #include <semaphore.h> to #include <nuttx/semaphore.h> Since the kernel side should call nxsem_xxx instead and remove the unused inclusion
2020-02-01 08:17:32 +01:00
#include <nuttx/semaphore.h>
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
IOBs: Move from driver/iob to a better location in mm/iob
2017-05-09 15:34:59 +02:00
#ifdef CONFIG_MM_IOB
Eliminate some warnings
2015-09-08 17:20:18 +02:00
Move net/iob to drivers/iob so that the I/O buffering feature can be available to other drivers when networking is disabled.
2017-04-21 00:08:49 +02:00
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
#if defined(CONFIG_DEBUG_FEATURES) && defined(CONFIG_IOB_DEBUG)
# define ioberr _err
# define iobwarn _warn
# define iobinfo _info
#else
debug: Reduce CONFIG_CPP_HAVE_VARARGS usage Signed-off-by: Xiang Xiao <xiaoxiang@xiaomi.com>
2020-05-04 12:55:11 +02:00
# define ioberr _none
# define iobwarn _none
# define iobinfo _none
Move net/iob to drivers/iob so that the I/O buffering feature can be available to other drivers when networking is disabled.
2017-04-21 00:08:49 +02:00
#endif /* CONFIG_DEBUG_FEATURES && CONFIG_IOB_DEBUG */
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
/****************************************************************************
* Public Data
****************************************************************************/
IOB: Add queue handling interfaces; improve lists
2014-06-06 17:35:31 +02:00
/* A list of all free, unallocated I/O buffers */
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
IOB: Add queue handling interfaces; improve lists
2014-06-06 17:35:31 +02:00
extern FAR struct iob_s *g_iob_freelist;
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
There can be a failure in IOB allocation to some asynchronous behavior caused by the use of sem_post(). Consider this scenario: Task A holds an IOB.  There are no further IOBs.  The value of semcount is zero. Task B calls iob_alloc().  Since there are not IOBs, it calls sem_wait().  The v alue of semcount is now -1. Task A frees the IOB.  iob_free() adds the IOB to the free list and calls sem_post() this makes Task B ready to run and sets semcount to zero NOT 1.  There is one IOB in the free list and semcount is zero.  When Task B wakes up it would increment the sem_count back to the correct value. But an interrupt or another task runs occurs before Task B executes.  The interrupt or other tak takes the IOB off of the free list and decrements the semcount.  But since semcount is then < 0, this causes the assertion because that is an invalid state in the interrupt handler. So I think that the root cause is that there the asynchrony between incrementing the semcount. This change separates the list of IOBs: Currently there is only a free list of IOBs. The problem, I believe, is because of asynchronies due sem_post() post cause the semcount and the list content to become out of sync. This change adds a new 'committed' list: When there is a task waiting for an IOB, it will go into the committed list rather than the free list before the semaphore is posted. On the waiting side, when awakened from the semaphore wait, it will expect to find its IOB in the committed list, rather than free list. In this way, the content of the free list and the value of the semaphore count always remain in sync.
2017-05-16 19:03:35 +02:00
/* A list of I/O buffers that are committed for allocation */
extern FAR struct iob_s *g_iob_committed;
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
Add throttle support to the I/O buffer logic
2014-06-24 19:53:19 +02:00
#if CONFIG_IOB_NCHAINS > 0
There can be a failure in IOB allocation to some asynchronous behavior caused by the use of sem_post(). Consider this scenario: Task A holds an IOB.  There are no further IOBs.  The value of semcount is zero. Task B calls iob_alloc().  Since there are not IOBs, it calls sem_wait().  The v alue of semcount is now -1. Task A frees the IOB.  iob_free() adds the IOB to the free list and calls sem_post() this makes Task B ready to run and sets semcount to zero NOT 1.  There is one IOB in the free list and semcount is zero.  When Task B wakes up it would increment the sem_count back to the correct value. But an interrupt or another task runs occurs before Task B executes.  The interrupt or other tak takes the IOB off of the free list and decrements the semcount.  But since semcount is then < 0, this causes the assertion because that is an invalid state in the interrupt handler. So I think that the root cause is that there the asynchrony between incrementing the semcount. This change separates the list of IOBs: Currently there is only a free list of IOBs. The problem, I believe, is because of asynchronies due sem_post() post cause the semcount and the list content to become out of sync. This change adds a new 'committed' list: When there is a task waiting for an IOB, it will go into the committed list rather than the free list before the semaphore is posted. On the waiting side, when awakened from the semaphore wait, it will expect to find its IOB in the committed list, rather than free list. In this way, the content of the free list and the value of the semaphore count always remain in sync.
2017-05-16 19:03:35 +02:00
/* A list of all free, unallocated I/O buffer queue containers */
IOB: Add queue handling interfaces; improve lists
2014-06-06 17:35:31 +02:00
extern FAR struct iob_qentry_s *g_iob_freeqlist;
There can be a failure in IOB allocation to some asynchronous behavior caused by the use of sem_post(). Consider this scenario: Task A holds an IOB.  There are no further IOBs.  The value of semcount is zero. Task B calls iob_alloc().  Since there are not IOBs, it calls sem_wait().  The v alue of semcount is now -1. Task A frees the IOB.  iob_free() adds the IOB to the free list and calls sem_post() this makes Task B ready to run and sets semcount to zero NOT 1.  There is one IOB in the free list and semcount is zero.  When Task B wakes up it would increment the sem_count back to the correct value. But an interrupt or another task runs occurs before Task B executes.  The interrupt or other tak takes the IOB off of the free list and decrements the semcount.  But since semcount is then < 0, this causes the assertion because that is an invalid state in the interrupt handler. So I think that the root cause is that there the asynchrony between incrementing the semcount. This change separates the list of IOBs: Currently there is only a free list of IOBs. The problem, I believe, is because of asynchronies due sem_post() post cause the semcount and the list content to become out of sync. This change adds a new 'committed' list: When there is a task waiting for an IOB, it will go into the committed list rather than the free list before the semaphore is posted. On the waiting side, when awakened from the semaphore wait, it will expect to find its IOB in the committed list, rather than free list. In this way, the content of the free list and the value of the semaphore count always remain in sync.
2017-05-16 19:03:35 +02:00
/* A list of I/O buffer queue containers that are committed for allocation */
IOBs: Fix a typing error mm/iob/iob.h mm/iob/iob_initialize.c
2017-05-27 16:03:00 +02:00
extern FAR struct iob_qentry_s *g_iob_qcommitted;
Add throttle support to the I/O buffer logic
2014-06-24 19:53:19 +02:00
#endif
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
First cut at conversion of write-buffering to use I/O buffer chaings (IOBs)
2014-06-22 19:27:57 +02:00
/* Counting semaphores that tracks the number of free IOBs/qentries */
Add throttle support to the I/O buffer logic
2014-06-24 19:53:19 +02:00
extern sem_t g_iob_sem; /* Counts free I/O buffers */
#if CONFIG_IOB_THROTTLE > 0
extern sem_t g_throttle_sem; /* Counts available I/O buffers when throttled */
#endif
#if CONFIG_IOB_NCHAINS > 0
extern sem_t g_qentry_sem; /* Counts free I/O buffer queue containers */
#endif
First cut at conversion of write-buffering to use I/O buffer chaings (IOBs)
2014-06-22 19:27:57 +02:00
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
/****************************************************************************
* Public Function Prototypes
****************************************************************************/
IOB: Add queue handling interfaces; improve lists
2014-06-06 17:35:31 +02:00
/****************************************************************************
* Name: iob_alloc_qentry
*
* Description:
* Allocate an I/O buffer chain container by taking the buffer at the head
* of the free list. This function is intended only for internal use by
* the IOB module.
*
****************************************************************************/
FAR struct iob_qentry_s *iob_alloc_qentry(void);
Networking: The are issues with the TCP write-ahead buffering if CONFIG_NET_NOINTS is enabled: There is a possibility of deadlocks in certain timing conditions. I have not seen this with the Tiva driver that I have been users but other people claim to see the issue on other platforms. Certainly it is a logic error: The network should never wait for TCP read-ahead buffering space to be available. It should drop the packets immediately. This was fixed by duplicating most of the IOB interfaces: The versions that waited are still present (like iob_alloc()), but now there are non-waiting verisons of the same interfaces (like iob_tryalloc()). The TCP read-ahead logic now uses only these non-waiting interfaces.
2015-01-28 04:23:42 +01:00
/****************************************************************************
* Name: iob_tryalloc_qentry
*
* Description:
* Try to allocate an I/O buffer chain container by taking the buffer at
* the head of the free list without waiting for the container to become
* free. This function is intended only for internal use by the IOB module.
*
****************************************************************************/
FAR struct iob_qentry_s *iob_tryalloc_qentry(void);
IOB: Add queue handling interfaces; improve lists
2014-06-06 17:35:31 +02:00
/****************************************************************************
* Name: iob_free_qentry
*
* Description:
* Free the I/O buffer chain container by returning it to the free list.
* The link to the next I/O buffer in the chain is return.
*
****************************************************************************/
FAR struct iob_qentry_s *iob_free_qentry(FAR struct iob_qentry_s *iobq);
NET: Add generic I/O buffering logic
2014-06-03 20:41:34 +02:00
mm/iob: Improve the IOB notifier data structures. Replace the fixed array with a singly linked list. For small value of CONFIG_IOB_NWAITERS, either solution is fine. But the fixed array does not scale well when CONFIG_IOB_NWAITERS is large.
2018-09-09 01:46:57 +02:00
/****************************************************************************
sched/signal: Add a generic signal notification facility. Modify the custom IOB available notifier so that it is now just a wrapper around this generic signal notification. This generic signal notification faility will, eventually, be used to support network polling. Squashed commit of the following: mm/iob: The IOB available notifier is now just a wrapper around the common signal notifier. sched/signal: Add a generic signal notification facility. sched/signal/sig_evthread.c: More trivial naming changes. sched/signal: Rename nxsig_notification() to nxsig_evthread() to make forthcoming naming additions more consistent.
2018-09-09 16:32:37 +02:00
* Name: iob_notifier_signal
mm/iob: Add an IOB notifier that will send a signal to any registered threads that want to be notified when an IOB has been freed. This is an untested work-in-progress and is intended to be a part of a larger solution to correctly handling network poll operations.
2018-09-08 19:21:18 +02:00
*
* Description:
* An IOB has become available. Signal all threads waiting for an IOB
* that an IOB is available.
*
* When an IOB becomes available, *all* of the waiters in this thread will
* be signaled. If there are multiple waiters then only the highest
* priority thread will get the IOB. Lower priority threads will need to
sched/signal: Add a generic signal notification facility. Modify the custom IOB available notifier so that it is now just a wrapper around this generic signal notification. This generic signal notification faility will, eventually, be used to support network polling. Squashed commit of the following: mm/iob: The IOB available notifier is now just a wrapper around the common signal notifier. sched/signal: Add a generic signal notification facility. sched/signal/sig_evthread.c: More trivial naming changes. sched/signal: Rename nxsig_notification() to nxsig_evthread() to make forthcoming naming additions more consistent.
2018-09-09 16:32:37 +02:00
* call iob_notifier_setup() once again.
mm/iob: Add an IOB notifier that will send a signal to any registered threads that want to be notified when an IOB has been freed. This is an untested work-in-progress and is intended to be a part of a larger solution to correctly handling network poll operations.
2018-09-08 19:21:18 +02:00
*
* Input Parameters:
* None.
*
* Returned Value:
* None.
*
****************************************************************************/
#ifdef CONFIG_IOB_NOTIFIER
sched/signal: Add a generic signal notification facility. Modify the custom IOB available notifier so that it is now just a wrapper around this generic signal notification. This generic signal notification faility will, eventually, be used to support network polling. Squashed commit of the following: mm/iob: The IOB available notifier is now just a wrapper around the common signal notifier. sched/signal: Add a generic signal notification facility. sched/signal/sig_evthread.c: More trivial naming changes. sched/signal: Rename nxsig_notification() to nxsig_evthread() to make forthcoming naming additions more consistent.
2018-09-09 16:32:37 +02:00
void iob_notifier_signal(void);
mm/iob: Add an IOB notifier that will send a signal to any registered threads that want to be notified when an IOB has been freed. This is an untested work-in-progress and is intended to be a part of a larger solution to correctly handling network poll operations.
2018-09-08 19:21:18 +02:00
#endif
IOBs: Move from driver/iob to a better location in mm/iob
2017-05-09 15:34:59 +02:00
#endif /* CONFIG_MM_IOB */
#endif /* __MM_IOB_IOB_H */
Reference in New Issue Copy Permalink