summaryrefslogtreecommitdiffstats
path: root/kernel/mm/frontswap.c
diff options
context:
space:
mode:
authorJosé Pekkarinen <jose.pekkarinen@nokia.com>2016-04-11 10:41:07 +0300
committerJosé Pekkarinen <jose.pekkarinen@nokia.com>2016-04-13 08:17:18 +0300
commite09b41010ba33a20a87472ee821fa407a5b8da36 (patch)
treed10dc367189862e7ca5c592f033dc3726e1df4e3 /kernel/mm/frontswap.c
parentf93b97fd65072de626c074dbe099a1fff05ce060 (diff)
These changes are the raw update to linux-4.4.6-rt14. Kernel sources
are taken from kernel.org, and rt patch from the rt wiki download page. During the rebasing, the following patch collided: Force tick interrupt and get rid of softirq magic(I70131fb85). Collisions have been removed because its logic was found on the source already. Change-Id: I7f57a4081d9deaa0d9ccfc41a6c8daccdee3b769 Signed-off-by: José Pekkarinen <jose.pekkarinen@nokia.com>
Diffstat (limited to 'kernel/mm/frontswap.c')
-rw-r--r--kernel/mm/frontswap.c215
1 files changed, 131 insertions, 84 deletions
diff --git a/kernel/mm/frontswap.c b/kernel/mm/frontswap.c
index 8d82809eb..27a9924ca 100644
--- a/kernel/mm/frontswap.c
+++ b/kernel/mm/frontswap.c
@@ -21,11 +21,16 @@
#include <linux/swapfile.h>
/*
- * frontswap_ops is set by frontswap_register_ops to contain the pointers
- * to the frontswap "backend" implementation functions.
+ * frontswap_ops are added by frontswap_register_ops, and provide the
+ * frontswap "backend" implementation functions. Multiple implementations
+ * may be registered, but implementations can never deregister. This
+ * is a simple singly-linked list of all registered implementations.
*/
static struct frontswap_ops *frontswap_ops __read_mostly;
+#define for_each_frontswap_ops(ops) \
+ for ((ops) = frontswap_ops; (ops); (ops) = (ops)->next)
+
/*
* If enabled, frontswap_store will return failure even on success. As
* a result, the swap subsystem will always write the page to swap, in
@@ -79,15 +84,6 @@ static inline void inc_frontswap_invalidates(void) { }
* on all frontswap functions to not call the backend until the backend
* has registered.
*
- * Specifically when no backend is registered (nobody called
- * frontswap_register_ops) all calls to frontswap_init (which is done via
- * swapon -> enable_swap_info -> frontswap_init) are registered and remembered
- * (via the setting of need_init bitmap) but fail to create tmem_pools. When a
- * backend registers with frontswap at some later point the previous
- * calls to frontswap_init are executed (by iterating over the need_init
- * bitmap) to create tmem_pools and set the respective poolids. All of that is
- * guarded by us using atomic bit operations on the 'need_init' bitmap.
- *
* This would not guards us against the user deciding to call swapoff right as
* we are calling the backend to initialize (so swapon is in action).
* Fortunatly for us, the swapon_mutex has been taked by the callee so we are
@@ -106,37 +102,64 @@ static inline void inc_frontswap_invalidates(void) { }
*
* Obviously the opposite (unloading the backend) must be done after all
* the frontswap_[store|load|invalidate_area|invalidate_page] start
- * ignorning or failing the requests - at which point frontswap_ops
- * would have to be made in some fashion atomic.
+ * ignoring or failing the requests. However, there is currently no way
+ * to unload a backend once it is registered.
*/
-static DECLARE_BITMAP(need_init, MAX_SWAPFILES);
/*
- * Register operations for frontswap, returning previous thus allowing
- * detection of multiple backends and possible nesting.
+ * Register operations for frontswap
*/
-struct frontswap_ops *frontswap_register_ops(struct frontswap_ops *ops)
+void frontswap_register_ops(struct frontswap_ops *ops)
{
- struct frontswap_ops *old = frontswap_ops;
- int i;
-
- for (i = 0; i < MAX_SWAPFILES; i++) {
- if (test_and_clear_bit(i, need_init)) {
- struct swap_info_struct *sis = swap_info[i];
- /* __frontswap_init _should_ have set it! */
- if (!sis->frontswap_map)
- return ERR_PTR(-EINVAL);
- ops->init(i);
- }
+ DECLARE_BITMAP(a, MAX_SWAPFILES);
+ DECLARE_BITMAP(b, MAX_SWAPFILES);
+ struct swap_info_struct *si;
+ unsigned int i;
+
+ bitmap_zero(a, MAX_SWAPFILES);
+ bitmap_zero(b, MAX_SWAPFILES);
+
+ spin_lock(&swap_lock);
+ plist_for_each_entry(si, &swap_active_head, list) {
+ if (!WARN_ON(!si->frontswap_map))
+ set_bit(si->type, a);
}
+ spin_unlock(&swap_lock);
+
+ /* the new ops needs to know the currently active swap devices */
+ for_each_set_bit(i, a, MAX_SWAPFILES)
+ ops->init(i);
+
/*
- * We MUST have frontswap_ops set _after_ the frontswap_init's
- * have been called. Otherwise __frontswap_store might fail. Hence
- * the barrier to make sure compiler does not re-order us.
+ * Setting frontswap_ops must happen after the ops->init() calls
+ * above; cmpxchg implies smp_mb() which will ensure the init is
+ * complete at this point.
*/
- barrier();
- frontswap_ops = ops;
- return old;
+ do {
+ ops->next = frontswap_ops;
+ } while (cmpxchg(&frontswap_ops, ops->next, ops) != ops->next);
+
+ spin_lock(&swap_lock);
+ plist_for_each_entry(si, &swap_active_head, list) {
+ if (si->frontswap_map)
+ set_bit(si->type, b);
+ }
+ spin_unlock(&swap_lock);
+
+ /*
+ * On the very unlikely chance that a swap device was added or
+ * removed between setting the "a" list bits and the ops init
+ * calls, we re-check and do init or invalidate for any changed
+ * bits.
+ */
+ if (unlikely(!bitmap_equal(a, b, MAX_SWAPFILES))) {
+ for (i = 0; i < MAX_SWAPFILES; i++) {
+ if (!test_bit(i, a) && test_bit(i, b))
+ ops->init(i);
+ else if (test_bit(i, a) && !test_bit(i, b))
+ ops->invalidate_area(i);
+ }
+ }
}
EXPORT_SYMBOL(frontswap_register_ops);
@@ -164,6 +187,7 @@ EXPORT_SYMBOL(frontswap_tmem_exclusive_gets);
void __frontswap_init(unsigned type, unsigned long *map)
{
struct swap_info_struct *sis = swap_info[type];
+ struct frontswap_ops *ops;
BUG_ON(sis == NULL);
@@ -179,28 +203,30 @@ void __frontswap_init(unsigned type, unsigned long *map)
* p->frontswap set to something valid to work properly.
*/
frontswap_map_set(sis, map);
- if (frontswap_ops)
- frontswap_ops->init(type);
- else {
- BUG_ON(type >= MAX_SWAPFILES);
- set_bit(type, need_init);
- }
+
+ for_each_frontswap_ops(ops)
+ ops->init(type);
}
EXPORT_SYMBOL(__frontswap_init);
bool __frontswap_test(struct swap_info_struct *sis,
pgoff_t offset)
{
- bool ret = false;
-
- if (frontswap_ops && sis->frontswap_map)
- ret = test_bit(offset, sis->frontswap_map);
- return ret;
+ if (sis->frontswap_map)
+ return test_bit(offset, sis->frontswap_map);
+ return false;
}
EXPORT_SYMBOL(__frontswap_test);
+static inline void __frontswap_set(struct swap_info_struct *sis,
+ pgoff_t offset)
+{
+ set_bit(offset, sis->frontswap_map);
+ atomic_inc(&sis->frontswap_pages);
+}
+
static inline void __frontswap_clear(struct swap_info_struct *sis,
- pgoff_t offset)
+ pgoff_t offset)
{
clear_bit(offset, sis->frontswap_map);
atomic_dec(&sis->frontswap_pages);
@@ -215,39 +241,46 @@ static inline void __frontswap_clear(struct swap_info_struct *sis,
*/
int __frontswap_store(struct page *page)
{
- int ret = -1, dup = 0;
+ int ret = -1;
swp_entry_t entry = { .val = page_private(page), };
int type = swp_type(entry);
struct swap_info_struct *sis = swap_info[type];
pgoff_t offset = swp_offset(entry);
+ struct frontswap_ops *ops;
/*
* Return if no backend registed.
* Don't need to inc frontswap_failed_stores here.
*/
if (!frontswap_ops)
- return ret;
+ return -1;
BUG_ON(!PageLocked(page));
BUG_ON(sis == NULL);
- if (__frontswap_test(sis, offset))
- dup = 1;
- ret = frontswap_ops->store(type, offset, page);
+
+ /*
+ * If a dup, we must remove the old page first; we can't leave the
+ * old page no matter if the store of the new page succeeds or fails,
+ * and we can't rely on the new page replacing the old page as we may
+ * not store to the same implementation that contains the old page.
+ */
+ if (__frontswap_test(sis, offset)) {
+ __frontswap_clear(sis, offset);
+ for_each_frontswap_ops(ops)
+ ops->invalidate_page(type, offset);
+ }
+
+ /* Try to store in each implementation, until one succeeds. */
+ for_each_frontswap_ops(ops) {
+ ret = ops->store(type, offset, page);
+ if (!ret) /* successful store */
+ break;
+ }
if (ret == 0) {
- set_bit(offset, sis->frontswap_map);
+ __frontswap_set(sis, offset);
inc_frontswap_succ_stores();
- if (!dup)
- atomic_inc(&sis->frontswap_pages);
} else {
- /*
- failed dup always results in automatic invalidate of
- the (older) page from frontswap
- */
inc_frontswap_failed_stores();
- if (dup) {
- __frontswap_clear(sis, offset);
- frontswap_ops->invalidate_page(type, offset);
- }
}
if (frontswap_writethrough_enabled)
/* report failure so swap also writes to swap device */
@@ -268,14 +301,22 @@ int __frontswap_load(struct page *page)
int type = swp_type(entry);
struct swap_info_struct *sis = swap_info[type];
pgoff_t offset = swp_offset(entry);
+ struct frontswap_ops *ops;
+
+ if (!frontswap_ops)
+ return -1;
BUG_ON(!PageLocked(page));
BUG_ON(sis == NULL);
- /*
- * __frontswap_test() will check whether there is backend registered
- */
- if (__frontswap_test(sis, offset))
- ret = frontswap_ops->load(type, offset, page);
+ if (!__frontswap_test(sis, offset))
+ return -1;
+
+ /* Try loading from each implementation, until one succeeds. */
+ for_each_frontswap_ops(ops) {
+ ret = ops->load(type, offset, page);
+ if (!ret) /* successful load */
+ break;
+ }
if (ret == 0) {
inc_frontswap_loads();
if (frontswap_tmem_exclusive_gets_enabled) {
@@ -294,16 +335,19 @@ EXPORT_SYMBOL(__frontswap_load);
void __frontswap_invalidate_page(unsigned type, pgoff_t offset)
{
struct swap_info_struct *sis = swap_info[type];
+ struct frontswap_ops *ops;
+
+ if (!frontswap_ops)
+ return;
BUG_ON(sis == NULL);
- /*
- * __frontswap_test() will check whether there is backend registered
- */
- if (__frontswap_test(sis, offset)) {
- frontswap_ops->invalidate_page(type, offset);
- __frontswap_clear(sis, offset);
- inc_frontswap_invalidates();
- }
+ if (!__frontswap_test(sis, offset))
+ return;
+
+ for_each_frontswap_ops(ops)
+ ops->invalidate_page(type, offset);
+ __frontswap_clear(sis, offset);
+ inc_frontswap_invalidates();
}
EXPORT_SYMBOL(__frontswap_invalidate_page);
@@ -314,16 +358,19 @@ EXPORT_SYMBOL(__frontswap_invalidate_page);
void __frontswap_invalidate_area(unsigned type)
{
struct swap_info_struct *sis = swap_info[type];
+ struct frontswap_ops *ops;
- if (frontswap_ops) {
- BUG_ON(sis == NULL);
- if (sis->frontswap_map == NULL)
- return;
- frontswap_ops->invalidate_area(type);
- atomic_set(&sis->frontswap_pages, 0);
- bitmap_zero(sis->frontswap_map, sis->max);
- }
- clear_bit(type, need_init);
+ if (!frontswap_ops)
+ return;
+
+ BUG_ON(sis == NULL);
+ if (sis->frontswap_map == NULL)
+ return;
+
+ for_each_frontswap_ops(ops)
+ ops->invalidate_area(type);
+ atomic_set(&sis->frontswap_pages, 0);
+ bitmap_zero(sis->frontswap_map, sis->max);
}
EXPORT_SYMBOL(__frontswap_invalidate_area);