summaryrefslogtreecommitdiffstats
path: root/mm/mmu_notifier.c
diff options
context:
space:
mode:
Diffstat (limited to 'mm/mmu_notifier.c')
-rw-r--r--mm/mmu_notifier.c103
1 files changed, 60 insertions, 43 deletions
diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c
index 862b60822d9f..479a1e751a73 100644
--- a/mm/mmu_notifier.c
+++ b/mm/mmu_notifier.c
@@ -14,10 +14,14 @@
#include <linux/export.h>
#include <linux/mm.h>
#include <linux/err.h>
+#include <linux/srcu.h>
#include <linux/rcupdate.h>
#include <linux/sched.h>
#include <linux/slab.h>
+/* global SRCU for all MMs */
+static struct srcu_struct srcu;
+
/*
* This function can't run concurrently against mmu_notifier_register
* because mm->mm_users > 0 during mmu_notifier_register and exit_mmap
@@ -25,8 +29,8 @@
* in parallel despite there being no task using this mm any more,
* through the vmas outside of the exit_mmap context, such as with
* vmtruncate. This serializes against mmu_notifier_unregister with
- * the mmu_notifier_mm->lock in addition to RCU and it serializes
- * against the other mmu notifiers with RCU. struct mmu_notifier_mm
+ * the mmu_notifier_mm->lock in addition to SRCU and it serializes
+ * against the other mmu notifiers with SRCU. struct mmu_notifier_mm
* can't go away from under us as exit_mmap holds an mm_count pin
* itself.
*/
@@ -34,12 +38,13 @@ void __mmu_notifier_release(struct mm_struct *mm)
{
struct mmu_notifier *mn;
struct hlist_node *n;
+ int id;
/*
- * RCU here will block mmu_notifier_unregister until
+ * SRCU here will block mmu_notifier_unregister until
* ->release returns.
*/
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist)
/*
* if ->release runs before mmu_notifier_unregister it
@@ -50,7 +55,7 @@ void __mmu_notifier_release(struct mm_struct *mm)
*/
if (mn->ops->release)
mn->ops->release(mn, mm);
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
spin_lock(&mm->mmu_notifier_mm->lock);
while (unlikely(!hlist_empty(&mm->mmu_notifier_mm->list))) {
@@ -68,7 +73,7 @@ void __mmu_notifier_release(struct mm_struct *mm)
spin_unlock(&mm->mmu_notifier_mm->lock);
/*
- * synchronize_rcu here prevents mmu_notifier_release to
+ * synchronize_srcu here prevents mmu_notifier_release to
* return to exit_mmap (which would proceed freeing all pages
* in the mm) until the ->release method returns, if it was
* invoked by mmu_notifier_unregister.
@@ -76,7 +81,7 @@ void __mmu_notifier_release(struct mm_struct *mm)
* The mmu_notifier_mm can't go away from under us because one
* mm_count is hold by exit_mmap.
*/
- synchronize_rcu();
+ synchronize_srcu(&srcu);
}
/*
@@ -89,14 +94,14 @@ int __mmu_notifier_clear_flush_young(struct mm_struct *mm,
{
struct mmu_notifier *mn;
struct hlist_node *n;
- int young = 0;
+ int young = 0, id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->clear_flush_young)
young |= mn->ops->clear_flush_young(mn, mm, address);
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
return young;
}
@@ -106,9 +111,9 @@ int __mmu_notifier_test_young(struct mm_struct *mm,
{
struct mmu_notifier *mn;
struct hlist_node *n;
- int young = 0;
+ int young = 0, id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->test_young) {
young = mn->ops->test_young(mn, mm, address);
@@ -116,7 +121,7 @@ int __mmu_notifier_test_young(struct mm_struct *mm,
break;
}
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
return young;
}
@@ -126,19 +131,14 @@ void __mmu_notifier_change_pte(struct mm_struct *mm, unsigned long address,
{
struct mmu_notifier *mn;
struct hlist_node *n;
+ int id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->change_pte)
mn->ops->change_pte(mn, mm, address, pte);
- /*
- * Some drivers don't have change_pte,
- * so we must call invalidate_page in that case.
- */
- else if (mn->ops->invalidate_page)
- mn->ops->invalidate_page(mn, mm, address);
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
}
void __mmu_notifier_invalidate_page(struct mm_struct *mm,
@@ -146,13 +146,14 @@ void __mmu_notifier_invalidate_page(struct mm_struct *mm,
{
struct mmu_notifier *mn;
struct hlist_node *n;
+ int id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->invalidate_page)
mn->ops->invalidate_page(mn, mm, address);
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
}
void __mmu_notifier_invalidate_range_start(struct mm_struct *mm,
@@ -160,13 +161,14 @@ void __mmu_notifier_invalidate_range_start(struct mm_struct *mm,
{
struct mmu_notifier *mn;
struct hlist_node *n;
+ int id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->invalidate_range_start)
mn->ops->invalidate_range_start(mn, mm, start, end);
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
}
void __mmu_notifier_invalidate_range_end(struct mm_struct *mm,
@@ -174,13 +176,14 @@ void __mmu_notifier_invalidate_range_end(struct mm_struct *mm,
{
struct mmu_notifier *mn;
struct hlist_node *n;
+ int id;
- rcu_read_lock();
+ id = srcu_read_lock(&srcu);
hlist_for_each_entry_rcu(mn, n, &mm->mmu_notifier_mm->list, hlist) {
if (mn->ops->invalidate_range_end)
mn->ops->invalidate_range_end(mn, mm, start, end);
}
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
}
static int do_mmu_notifier_register(struct mmu_notifier *mn,
@@ -192,22 +195,29 @@ static int do_mmu_notifier_register(struct mmu_notifier *mn,
BUG_ON(atomic_read(&mm->mm_users) <= 0);
- ret = -ENOMEM;
- mmu_notifier_mm = kmalloc(sizeof(struct mmu_notifier_mm), GFP_KERNEL);
- if (unlikely(!mmu_notifier_mm))
- goto out;
+ /*
+ * Verify that mmu_notifier_init() already run and the global srcu is
+ * initialized.
+ */
+ BUG_ON(!srcu.per_cpu_ref);
if (take_mmap_sem)
down_write(&mm->mmap_sem);
ret = mm_take_all_locks(mm);
if (unlikely(ret))
- goto out_cleanup;
+ goto out;
if (!mm_has_notifiers(mm)) {
+ mmu_notifier_mm = kmalloc(sizeof(struct mmu_notifier_mm),
+ GFP_KERNEL);
+ if (unlikely(!mmu_notifier_mm)) {
+ ret = -ENOMEM;
+ goto out_of_mem;
+ }
INIT_HLIST_HEAD(&mmu_notifier_mm->list);
spin_lock_init(&mmu_notifier_mm->lock);
+
mm->mmu_notifier_mm = mmu_notifier_mm;
- mmu_notifier_mm = NULL;
}
atomic_inc(&mm->mm_count);
@@ -223,13 +233,12 @@ static int do_mmu_notifier_register(struct mmu_notifier *mn,
hlist_add_head(&mn->hlist, &mm->mmu_notifier_mm->list);
spin_unlock(&mm->mmu_notifier_mm->lock);
+out_of_mem:
mm_drop_all_locks(mm);
-out_cleanup:
+out:
if (take_mmap_sem)
up_write(&mm->mmap_sem);
- /* kfree() does nothing if mmu_notifier_mm is NULL */
- kfree(mmu_notifier_mm);
-out:
+
BUG_ON(atomic_read(&mm->mm_users) <= 0);
return ret;
}
@@ -274,8 +283,8 @@ void __mmu_notifier_mm_destroy(struct mm_struct *mm)
/*
* This releases the mm_count pin automatically and frees the mm
* structure if it was the last user of it. It serializes against
- * running mmu notifiers with RCU and against mmu_notifier_unregister
- * with the unregister lock + RCU. All sptes must be dropped before
+ * running mmu notifiers with SRCU and against mmu_notifier_unregister
+ * with the unregister lock + SRCU. All sptes must be dropped before
* calling mmu_notifier_unregister. ->release or any other notifier
* method may be invoked concurrently with mmu_notifier_unregister,
* and only after mmu_notifier_unregister returned we're guaranteed
@@ -287,11 +296,12 @@ void mmu_notifier_unregister(struct mmu_notifier *mn, struct mm_struct *mm)
if (!hlist_unhashed(&mn->hlist)) {
/*
- * RCU here will force exit_mmap to wait ->release to finish
+ * SRCU here will force exit_mmap to wait ->release to finish
* before freeing the pages.
*/
- rcu_read_lock();
+ int id;
+ id = srcu_read_lock(&srcu);
/*
* exit_mmap will block in mmu_notifier_release to
* guarantee ->release is called before freeing the
@@ -299,7 +309,7 @@ void mmu_notifier_unregister(struct mmu_notifier *mn, struct mm_struct *mm)
*/
if (mn->ops->release)
mn->ops->release(mn, mm);
- rcu_read_unlock();
+ srcu_read_unlock(&srcu, id);
spin_lock(&mm->mmu_notifier_mm->lock);
hlist_del_rcu(&mn->hlist);
@@ -310,10 +320,17 @@ void mmu_notifier_unregister(struct mmu_notifier *mn, struct mm_struct *mm)
* Wait any running method to finish, of course including
* ->release if it was run by mmu_notifier_relase instead of us.
*/
- synchronize_rcu();
+ synchronize_srcu(&srcu);
BUG_ON(atomic_read(&mm->mm_count) <= 0);
mmdrop(mm);
}
EXPORT_SYMBOL_GPL(mmu_notifier_unregister);
+
+static int __init mmu_notifier_init(void)
+{
+ return init_srcu_struct(&srcu);
+}
+
+module_init(mmu_notifier_init);