percpu_ref: add PERCPU_REF_INIT_* flags

With the recent addition of percpu_ref_reinit(), percpu_ref now can be
used as a persistent switch which can be turned on and off repeatedly
where turning off maps to killing the ref and waiting for it to drain;
however, there currently isn't a way to initialize a percpu_ref in its
off (killed and drained) state, which can be inconvenient for certain
persistent switch use cases.

Similarly, percpu_ref_switch_to_atomic/percpu() allow dynamic
selection of operation mode; however, currently a newly initialized
percpu_ref is always in percpu mode making it impossible to avoid the
latency overhead of switching to atomic mode.

This patch adds @flags to percpu_ref_init() and implements the
following flags.

* PERCPU_REF_INIT_ATOMIC	: start ref in atomic mode
* PERCPU_REF_INIT_DEAD		: start ref killed and drained

These flags should be able to serve the above two use cases.

v2: target_core_tpg.c conversion was missing.  Fixed.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Kent Overstreet <kmo@daterainc.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>

6 files changed, 43 insertions, 13 deletions

diff --git a/block/blk-mq.c b/block/blk-mq.c
index 44a78ae3f899..d85fe01c44ef 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -1796,7 +1796,7 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_tag_set *set)
                 goto err_hctxs;
 
         if (percpu_ref_init(&q->mq_usage_counter, blk_mq_usage_counter_release,
-                            GFP_KERNEL))
+                            0, GFP_KERNEL))
                 goto err_map;
 
         setup_timer(&q->timeout, blk_mq_rq_timer, (unsigned long) q);
diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index 4ab6da338585..be783f717f19 100644
--- a/drivers/target/target_core_tpg.c
+++ b/drivers/target/target_core_tpg.c
@@ -819,7 +819,7 @@ int core_tpg_add_lun(
 {
         int ret;
 
-        ret = percpu_ref_init(&lun->lun_ref, core_tpg_lun_ref_release,
+        ret = percpu_ref_init(&lun->lun_ref, core_tpg_lun_ref_release, 0,
                               GFP_KERNEL);
         if (ret < 0)
                 return ret;
diff --git a/fs/aio.c b/fs/aio.c
index 8d217ed04e6e..84a751005f5b 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -661,10 +661,10 @@ static struct kioctx *ioctx_alloc(unsigned nr_events)
 
         INIT_LIST_HEAD(&ctx->active_reqs);
 
-        if (percpu_ref_init(&ctx->users, free_ioctx_users, GFP_KERNEL))
+        if (percpu_ref_init(&ctx->users, free_ioctx_users, 0, GFP_KERNEL))
                 goto err;
 
-        if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, GFP_KERNEL))
+        if (percpu_ref_init(&ctx->reqs, free_ioctx_reqs, 0, GFP_KERNEL))
                 goto err;
 
         ctx->cpu = alloc_percpu(struct kioctx_cpu);
diff --git a/include/linux/percpu-refcount.h b/include/linux/percpu-refcount.h
index cd7e20f0fe47..b0293f268cd2 100644
--- a/include/linux/percpu-refcount.h
+++ b/include/linux/percpu-refcount.h
@@ -63,6 +63,21 @@ enum {
         __PERCPU_REF_FLAG_BITS  = 2,
 };
 
+/* @flags for percpu_ref_init() */
+enum {
+        /*
+         * Start w/ ref == 1 in atomic mode.  Can be switched to percpu
+         * operation using percpu_ref_switch_to_percpu().
+         */
+        PERCPU_REF_INIT_ATOMIC  = 1 << 0,
+
+        /*
+         * Start dead w/ ref == 0 in atomic mode.  Must be revived with
+         * percpu_ref_reinit() before used.  Implies INIT_ATOMIC.
+         */
+        PERCPU_REF_INIT_DEAD    = 1 << 1,
+};
+
 struct percpu_ref {
         atomic_long_t           count;
         /*
@@ -76,7 +91,8 @@ struct percpu_ref {
 };
 
 int __must_check percpu_ref_init(struct percpu_ref *ref,
-                                 percpu_ref_func_t *release, gfp_t gfp);
+                                 percpu_ref_func_t *release, unsigned int flags,
+                                 gfp_t gfp);
 void percpu_ref_exit(struct percpu_ref *ref);
 void percpu_ref_switch_to_atomic(struct percpu_ref *ref,
                                  percpu_ref_func_t *confirm_switch);
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index a99d504294de..753df01a9831 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -1634,7 +1634,8 @@ static int cgroup_setup_root(struct cgroup_root *root, unsigned int ss_mask)
                 goto out;
         root_cgrp->id = ret;
 
-        ret = percpu_ref_init(&root_cgrp->self.refcnt, css_release, GFP_KERNEL);
+        ret = percpu_ref_init(&root_cgrp->self.refcnt, css_release, 0,
+                              GFP_KERNEL);
         if (ret)
                 goto out;
 
@@ -4510,7 +4511,7 @@ static int create_css(struct cgroup *cgrp, struct cgroup_subsys *ss,
 
         init_and_link_css(css, ss, cgrp);
 
-        err = percpu_ref_init(&css->refcnt, css_release, GFP_KERNEL);
+        err = percpu_ref_init(&css->refcnt, css_release, 0, GFP_KERNEL);
         if (err)
                 goto err_free_css;
 
@@ -4583,7 +4584,7 @@ static int cgroup_mkdir(struct kernfs_node *parent_kn, const char *name,
                 goto out_unlock;
         }
 
-        ret = percpu_ref_init(&cgrp->self.refcnt, css_release, GFP_KERNEL);
+        ret = percpu_ref_init(&cgrp->self.refcnt, css_release, 0, GFP_KERNEL);
         if (ret)
                 goto out_free_cgrp;
 
diff --git a/lib/percpu-refcount.c b/lib/percpu-refcount.c
index 5a6d43baccc5..ed280fb1e5b5 100644
--- a/lib/percpu-refcount.c
+++ b/lib/percpu-refcount.c
@@ -45,27 +45,40 @@ static unsigned long __percpu *percpu_count_ptr(struct percpu_ref *ref)
  * percpu_ref_init - initialize a percpu refcount
  * @ref: percpu_ref to initialize
  * @release: function which will be called when refcount hits 0
+ * @flags: PERCPU_REF_INIT_* flags
  * @gfp: allocation mask to use
  *
- * Initializes the refcount in single atomic counter mode with a refcount of 1;
- * analagous to atomic_long_set(ref, 1).
+ * Initializes @ref.  If @flags is zero, @ref starts in percpu mode with a
+ * refcount of 1; analagous to atomic_long_set(ref, 1).  See the
+ * definitions of PERCPU_REF_INIT_* flags for flag behaviors.
  *
  * Note that @release must not sleep - it may potentially be called from RCU
  * callback context by percpu_ref_kill().
  */
 int percpu_ref_init(struct percpu_ref *ref, percpu_ref_func_t *release,
-                    gfp_t gfp)
+                    unsigned int flags, gfp_t gfp)
 {
         size_t align = max_t(size_t, 1 << __PERCPU_REF_FLAG_BITS,
                              __alignof__(unsigned long));
-
-        atomic_long_set(&ref->count, 1 + PERCPU_COUNT_BIAS);
+        unsigned long start_count = 0;
 
         ref->percpu_count_ptr = (unsigned long)
                 __alloc_percpu_gfp(sizeof(unsigned long), align, gfp);
         if (!ref->percpu_count_ptr)
                 return -ENOMEM;
 
+        if (flags & (PERCPU_REF_INIT_ATOMIC | PERCPU_REF_INIT_DEAD))
+                ref->percpu_count_ptr |= __PERCPU_REF_ATOMIC;
+        else
+                start_count += PERCPU_COUNT_BIAS;
+
+        if (flags & PERCPU_REF_INIT_DEAD)
+                ref->percpu_count_ptr |= __PERCPU_REF_DEAD;
+        else
+                start_count++;
+
+        atomic_long_set(&ref->count, start_count);
+
         ref->release = release;
         return 0;
 }