1cc363b8ce
Introduce head/tail sync mode for MT ring synchronization. In that mode enqueue/dequeue operation is fully serialized: only one thread at a time is allowed to perform given op. Suppose to reduce stall times in case when ring is used on overcommitted cpus (multiple active threads on the same cpu). Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com> Acked-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
165 lines
4.2 KiB
C
165 lines
4.2 KiB
C
/* SPDX-License-Identifier: BSD-3-Clause
|
|
*
|
|
* Copyright (c) 2010-2020 Intel Corporation
|
|
* Copyright (c) 2007-2009 Kip Macy kmacy@freebsd.org
|
|
* All rights reserved.
|
|
* Derived from FreeBSD's bufring.h
|
|
* Used as BSD-3 Licensed with permission from Kip Macy.
|
|
*/
|
|
|
|
#ifndef _RTE_RING_HTS_C11_MEM_H_
|
|
#define _RTE_RING_HTS_C11_MEM_H_
|
|
|
|
/**
|
|
* @file rte_ring_hts_c11_mem.h
|
|
* It is not recommended to include this file directly,
|
|
* include <rte_ring.h> instead.
|
|
* Contains internal helper functions for head/tail sync (HTS) ring mode.
|
|
* For more information please refer to <rte_ring_hts.h>.
|
|
*/
|
|
|
|
/**
|
|
* @internal update tail with new value.
|
|
*/
|
|
static __rte_always_inline void
|
|
__rte_ring_hts_update_tail(struct rte_ring_hts_headtail *ht, uint32_t old_tail,
|
|
uint32_t num, uint32_t enqueue)
|
|
{
|
|
uint32_t tail;
|
|
|
|
RTE_SET_USED(enqueue);
|
|
|
|
tail = old_tail + num;
|
|
__atomic_store_n(&ht->ht.pos.tail, tail, __ATOMIC_RELEASE);
|
|
}
|
|
|
|
/**
|
|
* @internal waits till tail will become equal to head.
|
|
* Means no writer/reader is active for that ring.
|
|
* Suppose to work as serialization point.
|
|
*/
|
|
static __rte_always_inline void
|
|
__rte_ring_hts_head_wait(const struct rte_ring_hts_headtail *ht,
|
|
union __rte_ring_hts_pos *p)
|
|
{
|
|
while (p->pos.head != p->pos.tail) {
|
|
rte_pause();
|
|
p->raw = __atomic_load_n(&ht->ht.raw, __ATOMIC_ACQUIRE);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* @internal This function updates the producer head for enqueue
|
|
*/
|
|
static __rte_always_inline unsigned int
|
|
__rte_ring_hts_move_prod_head(struct rte_ring *r, unsigned int num,
|
|
enum rte_ring_queue_behavior behavior, uint32_t *old_head,
|
|
uint32_t *free_entries)
|
|
{
|
|
uint32_t n;
|
|
union __rte_ring_hts_pos np, op;
|
|
|
|
const uint32_t capacity = r->capacity;
|
|
|
|
op.raw = __atomic_load_n(&r->hts_prod.ht.raw, __ATOMIC_ACQUIRE);
|
|
|
|
do {
|
|
/* Reset n to the initial burst count */
|
|
n = num;
|
|
|
|
/*
|
|
* wait for tail to be equal to head,
|
|
* make sure that we read prod head/tail *before*
|
|
* reading cons tail.
|
|
*/
|
|
__rte_ring_hts_head_wait(&r->hts_prod, &op);
|
|
|
|
/*
|
|
* The subtraction is done between two unsigned 32bits value
|
|
* (the result is always modulo 32 bits even if we have
|
|
* *old_head > cons_tail). So 'free_entries' is always between 0
|
|
* and capacity (which is < size).
|
|
*/
|
|
*free_entries = capacity + r->cons.tail - op.pos.head;
|
|
|
|
/* check that we have enough room in ring */
|
|
if (unlikely(n > *free_entries))
|
|
n = (behavior == RTE_RING_QUEUE_FIXED) ?
|
|
0 : *free_entries;
|
|
|
|
if (n == 0)
|
|
break;
|
|
|
|
np.pos.tail = op.pos.tail;
|
|
np.pos.head = op.pos.head + n;
|
|
|
|
/*
|
|
* this CAS(ACQUIRE, ACQUIRE) serves as a hoist barrier to prevent:
|
|
* - OOO reads of cons tail value
|
|
* - OOO copy of elems from the ring
|
|
*/
|
|
} while (__atomic_compare_exchange_n(&r->hts_prod.ht.raw,
|
|
&op.raw, np.raw,
|
|
0, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE) == 0);
|
|
|
|
*old_head = op.pos.head;
|
|
return n;
|
|
}
|
|
|
|
/**
|
|
* @internal This function updates the consumer head for dequeue
|
|
*/
|
|
static __rte_always_inline unsigned int
|
|
__rte_ring_hts_move_cons_head(struct rte_ring *r, unsigned int num,
|
|
enum rte_ring_queue_behavior behavior, uint32_t *old_head,
|
|
uint32_t *entries)
|
|
{
|
|
uint32_t n;
|
|
union __rte_ring_hts_pos np, op;
|
|
|
|
op.raw = __atomic_load_n(&r->hts_cons.ht.raw, __ATOMIC_ACQUIRE);
|
|
|
|
/* move cons.head atomically */
|
|
do {
|
|
/* Restore n as it may change every loop */
|
|
n = num;
|
|
|
|
/*
|
|
* wait for tail to be equal to head,
|
|
* make sure that we read cons head/tail *before*
|
|
* reading prod tail.
|
|
*/
|
|
__rte_ring_hts_head_wait(&r->hts_cons, &op);
|
|
|
|
/* The subtraction is done between two unsigned 32bits value
|
|
* (the result is always modulo 32 bits even if we have
|
|
* cons_head > prod_tail). So 'entries' is always between 0
|
|
* and size(ring)-1.
|
|
*/
|
|
*entries = r->prod.tail - op.pos.head;
|
|
|
|
/* Set the actual entries for dequeue */
|
|
if (n > *entries)
|
|
n = (behavior == RTE_RING_QUEUE_FIXED) ? 0 : *entries;
|
|
|
|
if (unlikely(n == 0))
|
|
break;
|
|
|
|
np.pos.tail = op.pos.tail;
|
|
np.pos.head = op.pos.head + n;
|
|
|
|
/*
|
|
* this CAS(ACQUIRE, ACQUIRE) serves as a hoist barrier to prevent:
|
|
* - OOO reads of prod tail value
|
|
* - OOO copy of elems from the ring
|
|
*/
|
|
} while (__atomic_compare_exchange_n(&r->hts_cons.ht.raw,
|
|
&op.raw, np.raw,
|
|
0, __ATOMIC_ACQUIRE, __ATOMIC_ACQUIRE) == 0);
|
|
|
|
*old_head = op.pos.head;
|
|
return n;
|
|
}
|
|
|
|
#endif /* _RTE_RING_HTS_C11_MEM_H_ */
|