LU-1872 ptlrpc: cleanup jobstats in thread context

Since cfs_hash function isn't safe in interrupt context, we should
move the jobstats cleanup work into each stats logging function.

If there are any activities, cleanup will be triggerred on logging
time, and the hash table will be shrinked automatically; If there
isn't any activity, the hash table will stop growing, and it can
be cleared manually.

Signed-off-by: Niu Yawei <niu@whamcloud.com>
Change-Id: I1c60e123f67c0ad3b959bbb330282cd07c63a28c
Reviewed-on: http://review.whamcloud.com/3945
Tested-by: Hudson
Reviewed-by: Liang Zhen <liang@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Tested-by: Maloo <whamcloud.maloo@gmail.com>

diff --git a/lustre/include/lprocfs_status.h b/lustre/include/lprocfs_status.h
index c8f16b8..8f83201 100644 (file)
--- a/lustre/include/lprocfs_status.h
+++ b/lustre/include/lprocfs_status.h
@@ -366,10 +366,10 @@ struct obd_job_stats {
        cfs_hash_t        *ojs_hash;
        cfs_list_t         ojs_list;
        cfs_rwlock_t       ojs_lock; /* protect the obj_list */
-       int                ojs_cntr_num;
        cntr_init_callback ojs_cntr_init_fn;
-       cfs_timer_t        ojs_cleanup_timer;
+       int                ojs_cntr_num;
        int                ojs_cleanup_interval;
+       time_t             ojs_last_cleanup;
 };
 
 #ifdef LPROCFS

diff --git a/lustre/obdclass/lprocfs_jobstats.c b/lustre/obdclass/lprocfs_jobstats.c
index 0072feb..f369f97 100644 (file)
--- a/lustre/obdclass/lprocfs_jobstats.c
+++ b/lustre/obdclass/lprocfs_jobstats.c
@@ -153,6 +153,37 @@ static cfs_hash_ops_t job_stats_hash_ops = {
        .hs_exit       = job_stat_exit,
 };
 
+static int job_iter_callback(cfs_hash_t *hs, cfs_hash_bd_t *bd,
+                            cfs_hlist_node_t *hnode, void *data)
+{
+       time_t oldest = *((time_t *)data);
+       struct job_stat *job;
+
+       job = cfs_hlist_entry(hnode, struct job_stat, js_hash);
+       if (!oldest || job->js_timestamp < oldest)
+               cfs_hash_bd_del_locked(hs, bd, hnode);
+
+       return 0;
+}
+
+static void lprocfs_job_cleanup(struct obd_job_stats *stats, bool force)
+{
+       time_t oldest, now;
+
+       if (stats->ojs_cleanup_interval == 0)
+               return;
+
+       now = cfs_time_current_sec();
+       if (!force && now < stats->ojs_last_cleanup +
+                           stats->ojs_cleanup_interval)
+               return;
+
+       oldest = now - stats->ojs_cleanup_interval;
+       cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback,
+                              &oldest);
+       stats->ojs_last_cleanup = cfs_time_current_sec();
+}
+
 static struct job_stat *job_alloc(char *jobid, struct obd_job_stats *jobs)
 {
        struct job_stat *job;
@@ -190,6 +221,8 @@ int lprocfs_job_stats_log(struct obd_device *obd, char *jobid,
 
        LASSERT(stats && stats->ojs_hash);
 
+       lprocfs_job_cleanup(stats, false);
+
        if (!jobid || !strlen(jobid))
                RETURN(-EINVAL);
 
@@ -231,19 +264,6 @@ found:
 }
 EXPORT_SYMBOL(lprocfs_job_stats_log);
 
-static int job_iter_callback(cfs_hash_t *hs, cfs_hash_bd_t *bd,
-                            cfs_hlist_node_t *hnode, void *data)
-{
-       time_t oldest = *((time_t *)data);
-       struct job_stat *job;
-
-       job = cfs_hlist_entry(hnode, struct job_stat, js_hash);
-       if (!oldest || job->js_timestamp < oldest)
-               cfs_hash_bd_del_locked(hs, bd, hnode);
-
-       return 0;
-}
-
 void lprocfs_job_stats_fini(struct obd_device *obd)
 {
        struct obd_job_stats *stats = &obd->u.obt.obt_jobstats;
@@ -251,7 +271,6 @@ void lprocfs_job_stats_fini(struct obd_device *obd)
 
        if (stats->ojs_hash == NULL)
                return;
-       cfs_timer_disarm(&stats->ojs_cleanup_timer);
        cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback, &oldest);
        cfs_hash_putref(stats->ojs_hash);
        stats->ojs_hash = NULL;
@@ -463,20 +482,6 @@ struct file_operations lprocfs_jobstats_seq_fops = {
        .release = lprocfs_seq_release,
 };
 
-static void job_cleanup_callback(unsigned long data)
-{
-       struct obd_job_stats *stats = (struct obd_job_stats *)data;
-       time_t oldest;
-
-       if (stats->ojs_cleanup_interval) {
-               oldest = cfs_time_current_sec() - stats->ojs_cleanup_interval;
-               cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback,
-                                      &oldest);
-               cfs_timer_arm(&stats->ojs_cleanup_timer,
-                             cfs_time_shift(stats->ojs_cleanup_interval));
-       }
-}
-
 int lprocfs_job_stats_init(struct obd_device *obd, int cntr_num,
                           cntr_init_callback init_fn)
 {
@@ -510,10 +515,8 @@ int lprocfs_job_stats_init(struct obd_device *obd, int cntr_num,
        cfs_rwlock_init(&stats->ojs_lock);
        stats->ojs_cntr_num = cntr_num;
        stats->ojs_cntr_init_fn = init_fn;
-       cfs_timer_init(&stats->ojs_cleanup_timer, job_cleanup_callback, stats);
        stats->ojs_cleanup_interval = 600; /* 10 mins by default */
-       cfs_timer_arm(&stats->ojs_cleanup_timer,
-                     cfs_time_shift(stats->ojs_cleanup_interval));
+       stats->ojs_last_cleanup = cfs_time_current_sec();
 
        LPROCFS_WRITE_ENTRY();
        entry = create_proc_entry("job_stats", 0644, obd->obd_proc_entry);
@@ -557,11 +560,7 @@ int lprocfs_wr_job_interval(struct file *file, const char *buffer,
                return rc;
 
        stats->ojs_cleanup_interval = val;
-       if (!stats->ojs_cleanup_interval)
-               cfs_timer_disarm(&stats->ojs_cleanup_timer);
-       else
-               cfs_timer_arm(&stats->ojs_cleanup_timer,
-                             cfs_time_shift(stats->ojs_cleanup_interval));
+       lprocfs_job_cleanup(stats, true);
 
        return count;