You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
suricata/src/util-ringbuffer.c

509 lines
13 KiB
C

/* Copyright (C) 2007-2010 Open Information Security Foundation
*
* You can copy, redistribute or modify this Program under the terms of
* the GNU General Public License version 2 as published by the Free
* Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* version 2 along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
* 02110-1301, USA.
*/
/**
* \file
*
* \author Victor Julien <victor@inliniac.net>
*
* Ringbuffer implementation that is lockless for the most part IF atomic
* operations are available.
*
* Two sizes are implemented currently: 256 and 65536. Those sizes are chosen
* for simplicity when working with the read and write indexes. Both can just
* wrap around.
*
* Implemented are:
* Single reader, single writer (lockless)
* Multi reader, single writer (lockless)
* Multi reader, multi writer (partly locked)
*/
#include "suricata-common.h"
#include "suricata.h"
#include "util-ringbuffer.h"
#include "util-atomic.h"
#define USLEEP_TIME 5
/* Multi Reader, Single Writer, 8 bits */
RingBuffer8 *RingBufferMrSw8Init(void) {
RingBuffer8 *rb = SCMalloc(sizeof(RingBuffer8));
if (rb == NULL) {
return NULL;
}
memset(rb, 0x00, sizeof(RingBuffer8));
SC_ATOMIC_INIT(rb->write);
SC_ATOMIC_INIT(rb->read);
return rb;
}
void RingBufferMrSw8Destroy(RingBuffer8 *rb) {
if (rb != NULL) {
SC_ATOMIC_DESTROY(rb->write);
SC_ATOMIC_DESTROY(rb->read);
SCFree(rb);
}
}
/**
* \brief get the next ptr from the ring buffer
*
* Because we allow for multiple readers we take great care in making sure
* that the threads don't interfere with one another.
*
*/
void *RingBufferMrSw8Get(RingBuffer8 *rb) {
void *ptr;
/** local pointer for data races. If SCAtomicCompareAndSwap (CAS)
* fails we increase our local array idx to try the next array member
* until we succeed. Or when the buffer is empty again we jump back
* to the waiting loop. */
unsigned char readp;
/* buffer is empty, wait... */
retry:
while (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return NULL;
usleep(USLEEP_TIME);
}
/* atomically update rb->read */
readp = SC_ATOMIC_GET(rb->read) - 1;
do {
/* with multiple readers we can get in the situation that we exitted
* from the wait loop but the rb is empty again once we get here. */
if (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write))
goto retry;
readp++;
ptr = rb->array[readp];
} while (!(SC_ATOMIC_CAS(&rb->read, readp, (readp + 1))));
SCLogDebug("ptr %p", ptr);
return ptr;
}
/**
* \brief put a ptr in the RingBuffer
*/
int RingBufferMrSw8Put(RingBuffer8 *rb, void *ptr) {
SCLogDebug("ptr %p", ptr);
/* buffer is full, wait... */
while ((unsigned char)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return -1;
usleep(USLEEP_TIME);
}
rb->array[SC_ATOMIC_GET(rb->write)] = ptr;
SC_ATOMIC_ADD(rb->write, 1);
return 0;
}
/* Multi Reader, Single Writer */
RingBuffer16 *RingBufferMrSwInit(void) {
RingBuffer16 *rb = SCMalloc(sizeof(RingBuffer16));
if (rb == NULL) {
return NULL;
}
memset(rb, 0x00, sizeof(RingBuffer16));
SC_ATOMIC_INIT(rb->write);
SC_ATOMIC_INIT(rb->read);
return rb;
}
void RingBufferMrSwDestroy(RingBuffer16 *rb) {
if (rb != NULL) {
SC_ATOMIC_DESTROY(rb->write);
SC_ATOMIC_DESTROY(rb->read);
SCFree(rb);
}
}
/**
* \brief get the next ptr from the ring buffer
*
* Because we allow for multiple readers we take great care in making sure
* that the threads don't interfere with one another.
*
*/
void *RingBufferMrSwGet(RingBuffer16 *rb) {
void *ptr;
/** local pointer for data races. If SCAtomicCompareAndSwap (CAS)
* fails we increase our local array idx to try the next array member
* until we succeed. Or when the buffer is empty again we jump back
* to the waiting loop. */
unsigned short readp;
/* buffer is empty, wait... */
retry:
while (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return NULL;
usleep(USLEEP_TIME);
}
/* atomically update rb->read */
readp = SC_ATOMIC_GET(rb->read) - 1;
do {
/* with multiple readers we can get in the situation that we exitted
* from the wait loop but the rb is empty again once we get here. */
if (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write))
goto retry;
readp++;
ptr = rb->array[readp];
} while (!(SC_ATOMIC_CAS(&rb->read, readp, (readp + 1))));
SCLogDebug("ptr %p", ptr);
return ptr;
}
/**
* \brief put a ptr in the RingBuffer
*/
int RingBufferMrSwPut(RingBuffer16 *rb, void *ptr) {
SCLogDebug("ptr %p", ptr);
/* buffer is full, wait... */
while ((unsigned short)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return -1;
usleep(USLEEP_TIME);
}
rb->array[SC_ATOMIC_GET(rb->write)] = ptr;
SC_ATOMIC_ADD(rb->write, 1);
return 0;
}
/* Single Reader, Single Writer */
RingBuffer16 *RingBufferSrSwInit(void) {
RingBuffer16 *rb = SCMalloc(sizeof(RingBuffer16));
if (rb == NULL) {
return NULL;
}
memset(rb, 0x00, sizeof(RingBuffer16));
SC_ATOMIC_INIT(rb->write);
SC_ATOMIC_INIT(rb->read);
return rb;
}
void RingBufferSrSwDestroy(RingBuffer16 *rb) {
if (rb != NULL) {
SC_ATOMIC_DESTROY(rb->write);
SC_ATOMIC_DESTROY(rb->read);
SCFree(rb);
}
}
void *RingBufferSrSwGet(RingBuffer16 *rb) {
void *ptr = NULL;
/* buffer is empty, wait... */
while (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return NULL;
usleep(USLEEP_TIME);
}
ptr = rb->array[SC_ATOMIC_GET(rb->read)];
SC_ATOMIC_ADD(rb->read, 1);
return ptr;
}
int RingBufferSrSwPut(RingBuffer16 *rb, void *ptr) {
/* buffer is full, wait... */
while ((unsigned short)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return -1;
usleep(USLEEP_TIME);
}
rb->array[SC_ATOMIC_GET(rb->write)] = ptr;
SC_ATOMIC_ADD(rb->write, 1);
return 0;
}
/* Multi Reader, Multi Writer, 8 bits */
RingBuffer8 *RingBufferMrMw8Init(void) {
RingBuffer8 *rb = SCMalloc(sizeof(RingBuffer8));
if (rb == NULL) {
return NULL;
}
memset(rb, 0x00, sizeof(RingBuffer8));
SC_ATOMIC_INIT(rb->write);
SC_ATOMIC_INIT(rb->read);
SCSpinInit(&rb->spin, 0);
return rb;
}
void RingBufferMrMw8Destroy(RingBuffer8 *rb) {
if (rb != NULL) {
SC_ATOMIC_DESTROY(rb->write);
SC_ATOMIC_DESTROY(rb->read);
SCSpinDestroy(&rb->spin);
SCFree(rb);
}
}
/**
* \brief get the next ptr from the ring buffer
*
* Because we allow for multiple readers we take great care in making sure
* that the threads don't interfere with one another.
*
*/
void *RingBufferMrMw8Get(RingBuffer8 *rb) {
void *ptr;
/** local pointer for data races. If SCAtomicCompareAndSwap (CAS)
* fails we increase our local array idx to try the next array member
* until we succeed. Or when the buffer is empty again we jump back
* to the waiting loop. */
unsigned char readp;
/* buffer is empty, wait... */
retry:
while (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return NULL;
usleep(USLEEP_TIME);
}
/* atomically update rb->read */
readp = SC_ATOMIC_GET(rb->read) - 1;
do {
/* with multiple readers we can get in the situation that we exitted
* from the wait loop but the rb is empty again once we get here. */
if (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write))
goto retry;
readp++;
ptr = rb->array[readp];
} while (!(SC_ATOMIC_CAS(&rb->read, readp, (readp + 1))));
SCLogDebug("ptr %p", ptr);
return ptr;
}
/**
* \brief put a ptr in the RingBuffer.
*
* As we support multiple writers we need to protect 2 things:
* 1. writing the ptr to the array
* 2. incrementing the rb->write idx
*
* We can't do both at the same time in one atomic operation, so
* we need to (spin) lock it. We do increment rb->write atomically
* after that, so that we don't need to use the lock in our *Get
* function.
*
* \param rb the ringbuffer
* \param ptr ptr to store
*
* \retval 0 ok
* \retval -1 wait loop interrupted because of engine flags
*/
int RingBufferMrMw8Put(RingBuffer8 *rb, void *ptr) {
SCLogDebug("ptr %p", ptr);
/* buffer is full, wait... */
retry:
while ((unsigned char)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return -1;
usleep(USLEEP_TIME);
}
/* get our lock */
SCSpinLock(&rb->spin);
/* if while we got our lock the buffer changed, we need to retry */
if ((unsigned char)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
SCSpinUnlock(&rb->spin);
goto retry;
}
SCLogDebug("rb->write %u, ptr %p", SC_ATOMIC_GET(rb->write), ptr);
/* update the ring buffer */
rb->array[SC_ATOMIC_GET(rb->write)] = ptr;
SC_ATOMIC_ADD(rb->write, 1);
SCSpinUnlock(&rb->spin);
SCLogDebug("ptr %p, done", ptr);
return 0;
}
/* Multi Reader, Multi Writer, 16 bits */
RingBuffer16 *RingBufferMrMwInit(void) {
RingBuffer16 *rb = SCMalloc(sizeof(RingBuffer16));
if (rb == NULL) {
return NULL;
}
memset(rb, 0x00, sizeof(RingBuffer16));
SC_ATOMIC_INIT(rb->write);
SC_ATOMIC_INIT(rb->read);
SCSpinInit(&rb->spin, 0);
return rb;
}
void RingBufferMrMwDestroy(RingBuffer16 *rb) {
if (rb != NULL) {
SC_ATOMIC_DESTROY(rb->write);
SC_ATOMIC_DESTROY(rb->read);
SCSpinDestroy(&rb->spin);
SCFree(rb);
}
}
/**
* \brief get the next ptr from the ring buffer
*
* Because we allow for multiple readers we take great care in making sure
* that the threads don't interfere with one another.
*
*/
void *RingBufferMrMwGet(RingBuffer16 *rb) {
void *ptr;
/** local pointer for data races. If SCAtomicCompareAndSwap (CAS)
* fails we increase our local array idx to try the next array member
* until we succeed. Or when the buffer is empty again we jump back
* to the waiting loop. */
unsigned short readp;
/* buffer is empty, wait... */
retry:
while (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return NULL;
usleep(USLEEP_TIME);
}
/* atomically update rb->read */
readp = SC_ATOMIC_GET(rb->read) - 1;
do {
/* with multiple readers we can get in the situation that we exitted
* from the wait loop but the rb is empty again once we get here. */
if (SC_ATOMIC_GET(rb->read) == SC_ATOMIC_GET(rb->write))
goto retry;
readp++;
ptr = rb->array[readp];
} while (!(SC_ATOMIC_CAS(&rb->read, readp, (readp + 1))));
SCLogDebug("ptr %p", ptr);
return ptr;
}
/**
* \brief put a ptr in the RingBuffer.
*
* As we support multiple writers we need to protect 2 things:
* 1. writing the ptr to the array
* 2. incrementing the rb->write idx
*
* We can't do both at the same time in one atomic operation, so
* we need to (spin) lock it. We do increment rb->write atomically
* after that, so that we don't need to use the lock in our *Get
* function.
*
* \param rb the ringbuffer
* \param ptr ptr to store
*
* \retval 0 ok
* \retval -1 wait loop interrupted because of engine flags
*/
int RingBufferMrMwPut(RingBuffer16 *rb, void *ptr) {
SCLogDebug("ptr %p", ptr);
/* buffer is full, wait... */
retry:
while ((unsigned short)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
/* break out if the engine wants to shutdown */
if (rb->shutdown != 0)
return -1;
usleep(USLEEP_TIME);
}
/* get our lock */
SCSpinLock(&rb->spin);
/* if while we got our lock the buffer changed, we need to retry */
if ((unsigned short)(SC_ATOMIC_GET(rb->write) + 1) == SC_ATOMIC_GET(rb->read)) {
SCSpinUnlock(&rb->spin);
goto retry;
}
SCLogDebug("rb->write %u, ptr %p", SC_ATOMIC_GET(rb->write), ptr);
/* update the ring buffer */
rb->array[SC_ATOMIC_GET(rb->write)] = ptr;
SC_ATOMIC_ADD(rb->write, 1);
SCSpinUnlock(&rb->spin);
SCLogDebug("ptr %p, done", ptr);
return 0;
}