Spdk/lib/env_ocf/ocf_env.h

938 lines
17 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: BSD-3-Clause
* Copyright (C) 2018 Intel Corporation.
* All rights reserved.
*/
#ifndef __LIBOCF_ENV_H__
#define __LIBOCF_ENV_H__
#ifndef _GNU_SOURCE
#define _GNU_SOURCE
#endif
#ifndef __USE_GNU
#define __USE_GNU
#endif
#include <linux/limits.h>
#include <linux/stddef.h>
#include "spdk/stdinc.h"
#include "spdk/likely.h"
#include "spdk/env.h"
#include "spdk/util.h"
#include "spdk/log.h"
#include "ocf_env_list.h"
#include "ocf/ocf_err.h"
#include "mpool.h"
typedef uint8_t u8;
typedef uint16_t u16;
typedef uint32_t u32;
typedef uint64_t u64;
typedef uint64_t sector_t;
#define __packed __attribute__((packed))
#define __aligned(x) __attribute__((aligned(x)))
/* linux sector 512-bytes */
#define ENV_SECTOR_SHIFT 9
#define ENV_SECTOR_SIZE (1<<ENV_SECTOR_SHIFT)
#define BYTES_TO_SECTOR(x) ((x) >> ENV_SECTOR_SHIFT)
/* *** MEMORY MANAGEMENT *** */
#define ENV_MEM_NORMAL 0
#define ENV_MEM_NOIO 0
#define ENV_MEM_ATOMIC 0
#define likely spdk_likely
#define unlikely spdk_unlikely
#define min(x, y) MIN(x, y)
#ifndef MIN
#define MIN(x, y) spdk_min(x, y)
#endif
#define ARRAY_SIZE(x) SPDK_COUNTOF(x)
/* LOGGING */
#define ENV_PRIu64 PRIu64
#define ENV_WARN(cond, fmt, args...) ({ \
if (spdk_unlikely((uintptr_t)(cond))) \
SPDK_WARNLOG(fmt, ##args); \
})
#define ENV_WARN_ON(cond) ({ \
if (spdk_unlikely((uintptr_t)(cond))) \
SPDK_WARNLOG("\n"); \
})
#define ENV_BUG() ({ \
SPDK_ERRLOG("BUG\n"); \
assert(0); \
abort(); \
})
#define ENV_BUG_ON(cond) ({ \
if (spdk_unlikely((uintptr_t)(cond))) { \
SPDK_ERRLOG("BUG\n"); \
assert(0); \
abort(); \
} \
})
#define ENV_BUILD_BUG_ON(cond) _Static_assert(!(cond), "static "\
"assertion failure")
#define container_of(ptr, type, member) SPDK_CONTAINEROF(ptr, type, member)
static inline void *
env_malloc(size_t size, int flags)
{
return spdk_malloc(size, 0, NULL, SPDK_ENV_LCORE_ID_ANY,
SPDK_MALLOC_DMA);
}
static inline void *
env_zalloc(size_t size, int flags)
{
return spdk_zmalloc(size, 0, NULL, SPDK_ENV_LCORE_ID_ANY,
SPDK_MALLOC_DMA);
}
static inline void
env_free(const void *ptr)
{
return spdk_free((void *)ptr);
}
static inline void *
env_vmalloc(size_t size)
{
return spdk_malloc(size, 0, NULL, SPDK_ENV_LCORE_ID_ANY,
SPDK_MALLOC_DMA);
}
static inline void *
env_vzalloc(size_t size)
{
/* TODO: raw_ram init can request huge amount of memory to store
* hashtable in it. need to ensure that allocation succeeds */
return spdk_zmalloc(size, 0, NULL, SPDK_ENV_LCORE_ID_ANY,
SPDK_MALLOC_DMA);
}
static inline void *
env_vzalloc_flags(size_t size, int flags)
{
return env_vzalloc(size);
}
static inline void *
env_secure_alloc(size_t size)
{
return spdk_zmalloc(size, 0, NULL, SPDK_ENV_LCORE_ID_ANY,
SPDK_MALLOC_DMA);
}
static inline void
env_secure_free(const void *ptr, size_t size)
{
return spdk_free((void *)ptr);
}
static inline void
env_vfree(const void *ptr)
{
return spdk_free((void *)ptr);
}
static inline uint64_t
env_get_free_memory(void)
{
return -1;
}
/* *** ALLOCATOR *** */
#define OCF_ALLOCATOR_NAME_MAX 24
typedef struct {
struct spdk_mempool *mempool;
size_t element_size;
size_t element_count;
bool zero;
} env_allocator;
env_allocator *env_allocator_create_extended(uint32_t size, const char *name, int limit, bool zero);
env_allocator *env_allocator_create(uint32_t size, const char *name, bool zero);
void env_allocator_destroy(env_allocator *allocator);
void *env_allocator_new(env_allocator *allocator);
void env_allocator_del(env_allocator *allocator, void *item);
uint32_t env_allocator_item_count(env_allocator *allocator);
/* *** MUTEX *** */
typedef struct {
pthread_mutex_t m;
} env_mutex;
static inline int
env_mutex_init(env_mutex *mutex)
{
return !!pthread_mutex_init(&mutex->m, NULL);
}
static inline void
env_mutex_lock(env_mutex *mutex)
{
ENV_BUG_ON(pthread_mutex_lock(&mutex->m));
}
static inline int
env_mutex_lock_interruptible(env_mutex *mutex)
{
env_mutex_lock(mutex);
return 0;
}
static inline int
env_mutex_trylock(env_mutex *mutex)
{
return pthread_mutex_trylock(&mutex->m) ? -OCF_ERR_NO_LOCK : 0;
}
static inline void
env_mutex_unlock(env_mutex *mutex)
{
ENV_BUG_ON(pthread_mutex_unlock(&mutex->m));
}
static inline int
env_mutex_is_locked(env_mutex *mutex)
{
if (env_mutex_trylock(mutex) == 0) {
env_mutex_unlock(mutex);
return 0;
}
return 1;
}
static inline int
env_mutex_destroy(env_mutex *mutex)
{
if (pthread_mutex_destroy(&mutex->m)) {
return 1;
}
return 0;
}
/* *** RECURSIVE MUTEX *** */
typedef env_mutex env_rmutex;
static inline int
env_rmutex_init(env_rmutex *rmutex)
{
pthread_mutexattr_t attr;
pthread_mutexattr_init(&attr);
pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_RECURSIVE);
pthread_mutex_init(&rmutex->m, &attr);
return 0;
}
static inline void
env_rmutex_lock(env_rmutex *rmutex)
{
env_mutex_lock(rmutex);
}
static inline int
env_rmutex_lock_interruptible(env_rmutex *rmutex)
{
return env_mutex_lock_interruptible(rmutex);
}
static inline int
env_rmutex_trylock(env_rmutex *rmutex)
{
return env_mutex_trylock(rmutex);
}
static inline void
env_rmutex_unlock(env_rmutex *rmutex)
{
env_mutex_unlock(rmutex);
}
static inline int
env_rmutex_is_locked(env_rmutex *rmutex)
{
return env_mutex_is_locked(rmutex);
}
static inline int
env_rmutex_destroy(env_rmutex *rmutex)
{
return env_mutex_destroy(rmutex);
}
/* *** RW SEMAPHORE *** */
typedef struct {
pthread_rwlock_t lock;
} env_rwsem;
static inline int
env_rwsem_init(env_rwsem *s)
{
return !!pthread_rwlock_init(&s->lock, NULL);
}
static inline void
env_rwsem_up_read(env_rwsem *s)
{
ENV_BUG_ON(pthread_rwlock_unlock(&s->lock));
}
static inline void
env_rwsem_down_read(env_rwsem *s)
{
ENV_BUG_ON(pthread_rwlock_rdlock(&s->lock));
}
static inline int
env_rwsem_down_read_trylock(env_rwsem *s)
{
return pthread_rwlock_tryrdlock(&s->lock) ? -OCF_ERR_NO_LOCK : 0;
}
static inline void
env_rwsem_up_write(env_rwsem *s)
{
ENV_BUG_ON(pthread_rwlock_unlock(&s->lock));
}
static inline void
env_rwsem_down_write(env_rwsem *s)
{
ENV_BUG_ON(pthread_rwlock_wrlock(&s->lock));
}
static inline int
env_rwsem_down_write_trylock(env_rwsem *s)
{
return pthread_rwlock_trywrlock(&s->lock) ? -OCF_ERR_NO_LOCK : 0;
}
static inline int
env_rwsem_is_locked(env_rwsem *s)
{
if (env_rwsem_down_read_trylock(s) == 0) {
env_rwsem_up_read(s);
return 0;
}
return 1;
}
static inline int
env_rwsem_down_read_interruptible(env_rwsem *s)
{
return pthread_rwlock_rdlock(&s->lock);
}
static inline int
env_rwsem_down_write_interruptible(env_rwsem *s)
{
return pthread_rwlock_wrlock(&s->lock);
}
static inline int
env_rwsem_destroy(env_rwsem *s)
{
return pthread_rwlock_destroy(&s->lock);
}
/* *** ATOMIC VARIABLES *** */
typedef int env_atomic;
typedef long env_atomic64;
#ifndef atomic_read
#define atomic_read(ptr) (*(__typeof__(*ptr) *volatile) (ptr))
#endif
#ifndef atomic_set
#define atomic_set(ptr, i) ((*(__typeof__(*ptr) *volatile) (ptr)) = (i))
#endif
#define atomic_inc(ptr) ((void) __sync_fetch_and_add(ptr, 1))
#define atomic_dec(ptr) ((void) __sync_fetch_and_add(ptr, -1))
#define atomic_add(ptr, n) ((void) __sync_fetch_and_add(ptr, n))
#define atomic_sub(ptr, n) ((void) __sync_fetch_and_sub(ptr, n))
#define atomic_cmpxchg __sync_val_compare_and_swap
static inline int
env_atomic_read(const env_atomic *a)
{
return atomic_read(a);
}
static inline void
env_atomic_set(env_atomic *a, int i)
{
atomic_set(a, i);
}
static inline void
env_atomic_add(int i, env_atomic *a)
{
atomic_add(a, i);
}
static inline void
env_atomic_sub(int i, env_atomic *a)
{
atomic_sub(a, i);
}
static inline bool
env_atomic_sub_and_test(int i, env_atomic *a)
{
return __sync_sub_and_fetch(a, i) == 0;
}
static inline void
env_atomic_inc(env_atomic *a)
{
atomic_inc(a);
}
static inline void
env_atomic_dec(env_atomic *a)
{
atomic_dec(a);
}
static inline bool
env_atomic_dec_and_test(env_atomic *a)
{
return __sync_sub_and_fetch(a, 1) == 0;
}
static inline bool
env_atomic_inc_and_test(env_atomic *a)
{
return __sync_add_and_fetch(a, 1) == 0;
}
static inline int
env_atomic_add_return(int i, env_atomic *a)
{
return __sync_add_and_fetch(a, i);
}
static inline int
env_atomic_sub_return(int i, env_atomic *a)
{
return __sync_sub_and_fetch(a, i);
}
static inline int
env_atomic_inc_return(env_atomic *a)
{
return env_atomic_add_return(1, a);
}
static inline int
env_atomic_dec_return(env_atomic *a)
{
return env_atomic_sub_return(1, a);
}
static inline int
env_atomic_cmpxchg(env_atomic *a, int old, int new_value)
{
return atomic_cmpxchg(a, old, new_value);
}
static inline int
env_atomic_add_unless(env_atomic *a, int i, int u)
{
int c, old;
c = env_atomic_read(a);
for (;;) {
if (spdk_unlikely(c == (u))) {
break;
}
old = env_atomic_cmpxchg((a), c, c + (i));
if (spdk_likely(old == c)) {
break;
}
c = old;
}
return c != (u);
}
static inline long
env_atomic64_read(const env_atomic64 *a)
{
return atomic_read(a);
}
static inline void
env_atomic64_set(env_atomic64 *a, long i)
{
atomic_set(a, i);
}
static inline void
env_atomic64_add(long i, env_atomic64 *a)
{
atomic_add(a, i);
}
static inline void
env_atomic64_sub(long i, env_atomic64 *a)
{
atomic_sub(a, i);
}
static inline void
env_atomic64_inc(env_atomic64 *a)
{
atomic_inc(a);
}
static inline void
env_atomic64_dec(env_atomic64 *a)
{
atomic_dec(a);
}
static inline long
env_atomic64_add_return(long i, env_atomic64 *a)
{
return __sync_add_and_fetch(a, i);
}
static inline long
env_atomic64_sub_return(long i, env_atomic64 *a)
{
return __sync_sub_and_fetch(a, i);
}
static inline long
env_atomic64_inc_return(env_atomic64 *a)
{
return env_atomic64_add_return(1, a);
}
static inline long
env_atomic64_dec_return(env_atomic64 *a)
{
return env_atomic64_sub_return(1, a);
}
static inline long
env_atomic64_cmpxchg(env_atomic64 *a, long old, long new)
{
return atomic_cmpxchg(a, old, new);
}
/* *** COMPLETION *** */
typedef struct completion {
sem_t sem;
} env_completion;
static inline void
env_completion_init(env_completion *completion)
{
sem_init(&completion->sem, 0, 0);
}
static inline void
env_completion_wait(env_completion *completion)
{
sem_wait(&completion->sem);
}
static inline void
env_completion_complete(env_completion *completion)
{
sem_post(&completion->sem);
}
static inline void
env_completion_destroy(env_completion *completion)
{
sem_destroy(&completion->sem);
}
/* *** SPIN LOCKS *** */
typedef struct {
pthread_spinlock_t lock;
} env_spinlock;
static inline int
env_spinlock_init(env_spinlock *l)
{
return pthread_spin_init(&l->lock, 0);
}
static inline int
env_spinlock_trylock(env_spinlock *l)
{
return pthread_spin_trylock(&l->lock) ? -OCF_ERR_NO_LOCK : 0;
}
static inline void
env_spinlock_lock(env_spinlock *l)
{
ENV_BUG_ON(pthread_spin_lock(&l->lock));
}
static inline void
env_spinlock_unlock(env_spinlock *l)
{
ENV_BUG_ON(pthread_spin_unlock(&l->lock));
}
#define env_spinlock_lock_irqsave(l, flags) \
(void)flags; \
env_spinlock_lock(l)
#define env_spinlock_unlock_irqrestore(l, flags) \
(void)flags; \
env_spinlock_unlock(l)
static inline void
env_spinlock_destroy(env_spinlock *l)
{
ENV_BUG_ON(pthread_spin_destroy(&l->lock));
}
/* *** RW LOCKS *** */
typedef struct {
pthread_rwlock_t lock;
} env_rwlock;
static inline void
env_rwlock_init(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_init(&l->lock, NULL));
}
static inline void
env_rwlock_read_lock(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_rdlock(&l->lock));
}
static inline void
env_rwlock_read_unlock(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_unlock(&l->lock));
}
static inline void
env_rwlock_write_lock(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_wrlock(&l->lock));
}
static inline void
env_rwlock_write_unlock(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_unlock(&l->lock));
}
static inline void
env_rwlock_destroy(env_rwlock *l)
{
ENV_BUG_ON(pthread_rwlock_destroy(&l->lock));
}
static inline void
env_bit_set(int nr, volatile void *addr)
{
char *byte = (char *)addr + (nr >> 3);
char mask = 1 << (nr & 7);
__sync_or_and_fetch(byte, mask);
}
static inline void
env_bit_clear(int nr, volatile void *addr)
{
char *byte = (char *)addr + (nr >> 3);
char mask = 1 << (nr & 7);
__sync_and_and_fetch(byte, ~mask);
}
static inline bool
env_bit_test(int nr, const volatile unsigned long *addr)
{
const char *byte = (char *)addr + (nr >> 3);
char mask = 1 << (nr & 7);
return !!(*byte & mask);
}
/* *** WAITQUEUE *** */
typedef struct {
sem_t sem;
} env_waitqueue;
static inline void
env_waitqueue_init(env_waitqueue *w)
{
sem_init(&w->sem, 0, 0);
}
static inline void
env_waitqueue_wake_up(env_waitqueue *w)
{
sem_post(&w->sem);
}
#define env_waitqueue_wait(w, condition) \
({ \
int __ret = 0; \
if (!(condition)) \
sem_wait(&w.sem); \
__ret = __ret; \
})
/* *** SCHEDULING *** */
/* CAS does not need this while in user-space */
static inline void
env_schedule(void)
{
}
#define env_cond_resched env_schedule
static inline int
env_in_interrupt(void)
{
return 0;
}
static inline uint64_t
env_get_tick_count(void)
{
return spdk_get_ticks();
}
static inline uint64_t
env_ticks_to_secs(uint64_t j)
{
return j / spdk_get_ticks_hz();
}
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
/**
* @brief Dividing first tick_hz by 1000 is better than multiply j by 1000
* because if we would multiply j by 1000 we could only handle j
* up to 54b (*1000 is 10b).
* with this implementation we can handle all 64b in j.
* we only assume that ticks_hz is perfectly divisible by 1000
* which is probably good assumption because CPU frequency is in GHz/MHz scale.
*
* @param[in] j ticks count
*/
static inline uint64_t
env_ticks_to_msecs(uint64_t j)
{
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
return j / (spdk_get_ticks_hz() / 1000);
}
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
/**
* @brief Same as in msec case
* we divide ticks_hz by 1000 * 1000.
* so we use all 64b in j here as well.
* we assume that ticks_hz is perfectly divisible by 1000 * 1000
* i.e. CPU frequency is divisible by 1MHz.
*
* @param[in] j ticks count
*/
static inline uint64_t
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
env_ticks_to_usecs(uint64_t j)
{
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
return j / (spdk_get_ticks_hz() / (1000 * 1000));
}
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
/**
* @brief We can't divide ticks_hz by 10^9
* because we can't assume that CPU frequency is prefectly divisible by 10^9.
* for example there are CPUs with 2.8GHz or 3.3GHz.
* so in here we multiply j by 1000
* which means we can only handle 54b of j correctly.
*
* @param[in] j ticks count
*/
static inline uint64_t
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
env_ticks_to_nsecs(uint64_t j)
{
ocf: fix env_ticks_to_{msec,usec,nsec} precision & accuracy - fix precision when one convert to seconds and then multiply we can have precision errors for example if one have 77ms, it will go to 0 when converted to seconds and then multiply that 0 by 1000 will return 0 instead of 77ms. - fix mismatch nsec/usec nsec was multiplied by 1000*1000 while usec by 1000*1000*1000 it should be the opposite. anyway the implementation had changed. - implementation description * env_ticks_to_msec: j / (tick_hz / 1000) this is exactly the same as (j * 1000) / tick_hz (eq #2). but this implementation (eq #2) can only handle 54b in j (before overflowing) because of the multiplication by 1000 (10b). with the correct implementation we use all 64b in j. we assume that tick_hz will be prefectly divisible by 1000 so we are ok. * env_ticks_to_usec: j / (tick_hz / (1000 * 1000)) same as in msec case, we use all 64b in j. here we assume that tick_hz is perfectly divisible by (1000 * 1000) i.e. we assume that CPU frequency is some multiple of 1MHz. * env_ticks_to_nsec: (j * 1000) / (tick_hz / (1000 * 1000)) in this case we can't assume that tick_hz is divisible by 10^9 because there are many CPUs with 2.8GHz or 3.3GHz for example. so we multiply j by 1000 this means that we can only handle correctly j up to 54b. (64b - 10b, 10b for the *1000 operation) Signed-off-by: Amir Haroush <amir.haroush@huawei.com> Signed-off-by: Shai Fultheim <shai.fultheim@huawei.com> Change-Id: Ia8ea7f88b718df206fa0731e3f39f419ee922aa7 Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/17078 Community-CI: Mellanox Build Bot Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com>
2023-03-07 09:41:19 +00:00
return (j * 1000) / (spdk_get_ticks_hz() / (1000 * 1000));
}
static inline uint64_t
env_secs_to_ticks(uint64_t j)
{
return j * spdk_get_ticks_hz();
}
/* *** STRING OPERATIONS *** */
/* 512 KB is sufficient amount of memory for OCF operations */
#define ENV_MAX_MEM (512 * 1024)
static inline int
env_memset(void *dest, size_t len, uint8_t value)
{
if (dest == NULL || len == 0) {
return 1;
}
memset(dest, value, len);
return 0;
}
static inline int
env_memcpy(void *dest, size_t dmax, const void *src, size_t len)
{
if (dest == NULL || src == NULL) {
return 1;
}
if (dmax == 0 || dmax > ENV_MAX_MEM) {
return 1;
}
if (len == 0 || len > dmax) {
return 1;
}
memcpy(dest, src, len);
return 0;
}
static inline int
env_memcmp(const void *aptr, size_t dmax, const void *bptr, size_t len,
int *diff)
{
if (diff == NULL || aptr == NULL || bptr == NULL) {
return 1;
}
if (dmax == 0 || dmax > ENV_MAX_MEM) {
return 1;
}
if (len == 0 || len > dmax) {
return 1;
}
*diff = memcmp(aptr, bptr, len);
return 0;
}
/* 4096 is sufficient max length for any OCF operation on string */
#define ENV_MAX_STR (4 * 1024)
static inline size_t
env_strnlen(const char *src, size_t dmax)
{
return strnlen(src, dmax);
}
static inline int
env_strncpy(char *dest, size_t dmax, const char *src, size_t len)
{
if (dest == NULL || src == NULL) {
return 1;
}
if (dmax == 0 || dmax > ENV_MAX_STR) {
return 1;
}
if (len == 0) {
return 1;
}
/* Just copy as many characters as we can instead of return failure */
len = min(len, dmax);
strncpy(dest, src, len);
return 0;
}
#define env_strncmp(s1, slen1, s2, slen2) strncmp(s1, s2, min(slen1, slen2))
static inline char *
env_strdup(const char *src, int flags)
{
int len;
char *ret;
if (src == NULL) {
return NULL;
}
len = env_strnlen(src, ENV_MAX_STR) + 1;
ret = env_malloc(len, flags);
if (env_strncpy(ret, ENV_MAX_STR, src, len)) {
return NULL;
} else {
return ret;
}
}
/* *** SORTING *** */
static inline void
env_sort(void *base, size_t num, size_t size,
int (*cmp_fn)(const void *, const void *),
void (*swap_fn)(void *, void *, int size))
{
qsort(base, num, size, cmp_fn);
}
static inline void
env_msleep(uint64_t n)
{
usleep(n * 1000);
}
static inline void
env_touch_softlockup_wd(void)
{
}
/* *** CRC *** */
uint32_t env_crc32(uint32_t crc, uint8_t const *data, size_t len);
/* EXECUTION CONTEXTS */
unsigned env_get_execution_context(void);
void env_put_execution_context(unsigned ctx);
unsigned env_get_execution_context_count(void);
#endif /* __OCF_ENV_H__ */