michaelh/zephyr
1
0
Fork 0
Code Releases Activity

Revert "logging: Use spin lock"

Browse source 
This reverts commit dfe9561d9b.

Signed-off-by: Anas Nashif <anas.nashif@intel.com>
This commit is contained in:
Anas Nashif 2022-02-21 23:58:14 -05:00
commit 49d08c25fe
1 changed files with 23 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

54
subsys/logging/log_core.c
View file

@ -89,7 +89,6 @@ static atomic_t log_strdup_in_use;
static uint32_t log_strdup_max;
static uint32_t log_strdup_longest;
static struct k_timer log_process_thread_timer;
static struct k_spinlock list_lock;
static log_timestamp_t dummy_timestamp(void);
static log_timestamp_get_t timestamp_func = dummy_timestamp;
@ -229,50 +228,43 @@ static void detect_missed_strdup(struct log_msg *msg)
static void z_log_msg_post_finalize(void)
{
atomic_val_t cnt = atomic_inc(&buffered_cnt);
if (panic_mode) {
static struct k_spinlock process_lock;
k_spinlock_key_t key = k_spin_lock(&process_lock);
unsigned int key = irq_lock();
(void)log_process(false);
k_spin_unlock(&process_lock, key);
} else if (proc_tid != NULL) {
static atomic_t reentry_flag;
atomic_val_t cnt = atomic_inc(&buffered_cnt);
if (cnt == 0) {
if (atomic_inc(&reentry_flag) == 0) {
k_timer_start(&log_process_thread_timer,
K_MSEC(CONFIG_LOG_PROCESS_THREAD_SLEEP_MS),
K_NO_WAIT);
}
atomic_dec(&reentry_flag);
} else if (CONFIG_LOG_PROCESS_TRIGGER_THRESHOLD &&
cnt == CONFIG_LOG_PROCESS_TRIGGER_THRESHOLD) {
irq_unlock(key);
} else if (proc_tid != NULL && cnt == 0) {
k_timer_start(&log_process_thread_timer,
K_MSEC(CONFIG_LOG_PROCESS_THREAD_SLEEP_MS), K_NO_WAIT);
} else if (CONFIG_LOG_PROCESS_TRIGGER_THRESHOLD) {
if ((cnt == CONFIG_LOG_PROCESS_TRIGGER_THRESHOLD) &&
(proc_tid != NULL)) {
k_timer_stop(&log_process_thread_timer);
k_sem_give(&log_process_thread_sem);
} else {
/* No action needed. Message processing will be triggered by the
* timeout or when number of upcoming messages exceeds the
* threshold.
*/
}
} else {
/* No action needed. Message processing will be triggered by the
* timeout or when number of upcoming messages exceeds the
* threshold.
*/
;
}
}
static inline void msg_finalize(struct log_msg *msg,
struct log_msg_ids src_level)
{
k_spinlock_key_t key;
unsigned int key;
msg->hdr.ids = src_level;
msg->hdr.timestamp = timestamp_func();
key = k_spin_lock(&list_lock);
key = irq_lock();
log_list_add_tail(&list, msg);
k_spin_unlock(&list_lock, key);
irq_unlock(key);
z_log_msg_post_finalize();
}
@ -793,10 +785,10 @@ union log_msgs get_msg(void)
return msg;
}
k_spinlock_key_t key = k_spin_lock(&list_lock);
int key = irq_lock();
msg.msg = log_list_head_get(&list);
k_spin_unlock(&list_lock, key);
irq_unlock(key);
return msg;
}
@ -912,14 +904,14 @@ char *z_log_strdup(const char *str)
if (IS_ENABLED(CONFIG_LOG_STRDUP_POOL_PROFILING)) {
size_t slen = strlen(str);
static struct k_spinlock strdup_lock;
static struct k_spinlock lock;
k_spinlock_key_t key;
key = k_spin_lock(&strdup_lock);
key = k_spin_lock(&lock);
log_strdup_in_use++;
log_strdup_max = MAX(log_strdup_in_use, log_strdup_max);
log_strdup_longest = MAX(slen, log_strdup_longest);
k_spin_unlock(&strdup_lock, key);
k_spin_unlock(&lock, key);
}
/* Set 'allocated' flag. */