[RFC][PATCH 7/7] kprobes: Replace rp->free_instance with freelist

[Peter Zijlstra]

Gets rid of rp->lock, and as a result kretprobes are now fully
lockless.

Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
---
 include/linux/kprobes.h |   11 ++++++--
 kernel/kprobes.c        |   63 +++++++++++++++++++-----------------------------
 2 files changed, 34 insertions(+), 40 deletions(-)

--- a/include/linux/kprobes.h
+++ b/include/linux/kprobes.h
@@ -27,6 +27,7 @@
 #include <linux/rcupdate.h>
 #include <linux/mutex.h>
 #include <linux/ftrace.h>
+#include <linux/freelist.h>
 #include <asm/kprobes.h>
 
 #ifdef CONFIG_KPROBES
@@ -151,14 +152,18 @@ struct kretprobe {
 	int maxactive;
 	int nmissed;
 	size_t data_size;
-	struct hlist_head free_instances;
-	raw_spinlock_t lock;
+	struct freelist_head freelist;
 };
 
 struct kretprobe_instance {
 	union {
+		/*
+		 * Dodgy as heck, this relies on not clobbering freelist::refs.
+		 * llist: only clobbers freelist::next.
+		 * rcu: clobbers both, but only after rp::freelist is gone.
+		 */
+		struct freelist_node freelist;
 		struct llist_node llist;
-		struct hlist_node hlist;
 		struct rcu_head rcu;
 	};
 	struct kretprobe *rp;
--- a/kernel/kprobes.c
+++ b/kernel/kprobes.c
@@ -1219,12 +1219,8 @@ static void recycle_rp_inst(struct kretp
 	struct kretprobe *rp = ri->rp;
 
 	/* remove rp inst off the rprobe_inst_table */
-	hlist_del(&ri->hlist);
-	INIT_HLIST_NODE(&ri->hlist);
 	if (likely(rp)) {
-		raw_spin_lock(&rp->lock);
-		hlist_add_head(&ri->hlist, &rp->free_instances);
-		raw_spin_unlock(&rp->lock);
+		freelist_add(&ri->freelist, &rp->freelist);
 	} else {
 		kfree_rcu(ri, rcu);
 	}
@@ -1286,10 +1282,13 @@ NOKPROBE_SYMBOL(kprobe_flush_task);
 static inline void free_rp_inst(struct kretprobe *rp)
 {
 	struct kretprobe_instance *ri;
-	struct hlist_node *next;
+	struct freelist_node *node;
+
+	node = rp->freelist.head;
+	while (node) {
+		ri = container_of(node, struct kretprobe_instance, freelist);
+		node = node->next;
 
-	hlist_for_each_entry_safe(ri, next, &rp->free_instances, hlist) {
-		hlist_del(&ri->hlist);
 		kfree(ri);
 	}
 }
@@ -1986,36 +1985,28 @@ NOKPROBE_SYMBOL(__kretprobe_trampoline_h
 static int pre_handler_kretprobe(struct kprobe *p, struct pt_regs *regs)
 {
 	struct kretprobe *rp = container_of(p, struct kretprobe, kp);
-	unsigned long hash, flags = 0;
 	struct kretprobe_instance *ri;
+	struct freelist_node *fn;
 
-	/* TODO: consider to only swap the RA after the last pre_handler fired */
-	hash = hash_ptr(current, KPROBE_HASH_BITS);
-	raw_spin_lock_irqsave(&rp->lock, flags);
-	if (!hlist_empty(&rp->free_instances)) {
-		ri = hlist_entry(rp->free_instances.first,
-				struct kretprobe_instance, hlist);
-		hlist_del(&ri->hlist);
-		raw_spin_unlock_irqrestore(&rp->lock, flags);
-
-		ri->rp = rp;
-		ri->task = current;
-
-		if (rp->entry_handler && rp->entry_handler(ri, regs)) {
-			raw_spin_lock_irqsave(&rp->lock, flags);
-			hlist_add_head(&ri->hlist, &rp->free_instances);
-			raw_spin_unlock_irqrestore(&rp->lock, flags);
-			return 0;
-		}
-
-		arch_prepare_kretprobe(ri, regs);
+	fn = freelist_try_get(&rp->freelist);
+	if (!fn) {
+		rp->nmissed++;
+		return 0;
+	}
 
-		__llist_add(&ri->llist, &current->kretprobe_instances);
+	ri = container_of(fn, struct kretprobe_instance, freelist);
+	ri->rp = rp;
+	ri->task = current;
 
-	} else {
-		rp->nmissed++;
-		raw_spin_unlock_irqrestore(&rp->lock, flags);
+	if (rp->entry_handler && rp->entry_handler(ri, regs)) {
+		freelist_add(&ri->freelist, &rp->freelist);
+		return 0;
 	}
+
+	arch_prepare_kretprobe(ri, regs);
+
+	__llist_add(&ri->llist, &current->kretprobe_instances);
+
 	return 0;
 }
 NOKPROBE_SYMBOL(pre_handler_kretprobe);
@@ -2072,8 +2063,7 @@ int register_kretprobe(struct kretprobe
 		rp->maxactive = num_possible_cpus();
 #endif
 	}
-	raw_spin_lock_init(&rp->lock);
-	INIT_HLIST_HEAD(&rp->free_instances);
+	rp->freelist.head = NULL;
 	for (i = 0; i < rp->maxactive; i++) {
 		inst = kmalloc(sizeof(struct kretprobe_instance) +
 			       rp->data_size, GFP_KERNEL);
@@ -2081,8 +2071,7 @@ int register_kretprobe(struct kretprobe
 			free_rp_inst(rp);
 			return -ENOMEM;
 		}
-		INIT_HLIST_NODE(&inst->hlist);
-		hlist_add_head(&inst->hlist, &rp->free_instances);
+		freelist_add(&inst->freelist, &rp->freelist);
 	}
 
 	rp->nmissed = 0;