]> git.kernelconcepts.de Git - karo-tx-linux.git/commitdiff
inode: convert inode lru list to generic lru list code.
authorDave Chinner <dchinner@redhat.com>
Wed, 3 Jul 2013 00:19:52 +0000 (10:19 +1000)
committerStephen Rothwell <sfr@canb.auug.org.au>
Wed, 17 Jul 2013 02:34:53 +0000 (12:34 +1000)
[glommer@openvz.org: adapted for new LRU return codes]
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Glauber Costa <glommer@openvz.org>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Cc: Arve Hjønnevåg <arve@android.com>
Cc: Carlos Maiolino <cmaiolino@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: David Rientjes <rientjes@google.com>
Cc: Gleb Natapov <gleb@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: J. Bruce Fields <bfields@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Kent Overstreet <koverstreet@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Thomas Hellstrom <thellstrom@vmware.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
fs/inode.c
fs/super.c
include/linux/fs.h

index 7244ef98fdecec14e98d81e18b63cdb45beada65..b91067bdcc3f6a6af22cf626a6af19d37314b299 100644 (file)
@@ -17,6 +17,7 @@
 #include <linux/prefetch.h>
 #include <linux/buffer_head.h> /* for inode_has_buffers */
 #include <linux/ratelimit.h>
+#include <linux/list_lru.h>
 #include "internal.h"
 
 /*
@@ -24,7 +25,7 @@
  *
  * inode->i_lock protects:
  *   inode->i_state, inode->i_hash, __iget()
- * inode->i_sb->s_inode_lru_lock protects:
+ * Inode LRU list locks protect:
  *   inode->i_sb->s_inode_lru, inode->i_lru
  * inode_sb_list_lock protects:
  *   sb->s_inodes, inode->i_sb_list
@@ -37,7 +38,7 @@
  *
  * inode_sb_list_lock
  *   inode->i_lock
- *     inode->i_sb->s_inode_lru_lock
+ *     Inode LRU list locks
  *
  * bdi->wb.list_lock
  *   inode->i_lock
@@ -401,13 +402,8 @@ EXPORT_SYMBOL(ihold);
 
 static void inode_lru_list_add(struct inode *inode)
 {
-       spin_lock(&inode->i_sb->s_inode_lru_lock);
-       if (list_empty(&inode->i_lru)) {
-               list_add(&inode->i_lru, &inode->i_sb->s_inode_lru);
-               inode->i_sb->s_nr_inodes_unused++;
+       if (list_lru_add(&inode->i_sb->s_inode_lru, &inode->i_lru))
                this_cpu_inc(nr_unused);
-       }
-       spin_unlock(&inode->i_sb->s_inode_lru_lock);
 }
 
 /*
@@ -425,13 +421,9 @@ void inode_add_lru(struct inode *inode)
 
 static void inode_lru_list_del(struct inode *inode)
 {
-       spin_lock(&inode->i_sb->s_inode_lru_lock);
-       if (!list_empty(&inode->i_lru)) {
-               list_del_init(&inode->i_lru);
-               inode->i_sb->s_nr_inodes_unused--;
+
+       if (list_lru_del(&inode->i_sb->s_inode_lru, &inode->i_lru))
                this_cpu_dec(nr_unused);
-       }
-       spin_unlock(&inode->i_sb->s_inode_lru_lock);
 }
 
 /**
@@ -675,24 +667,8 @@ int invalidate_inodes(struct super_block *sb, bool kill_dirty)
        return busy;
 }
 
-static int can_unuse(struct inode *inode)
-{
-       if (inode->i_state & ~I_REFERENCED)
-               return 0;
-       if (inode_has_buffers(inode))
-               return 0;
-       if (atomic_read(&inode->i_count))
-               return 0;
-       if (inode->i_data.nrpages)
-               return 0;
-       return 1;
-}
-
 /*
- * Walk the superblock inode LRU for freeable inodes and attempt to free them.
- * This is called from the superblock shrinker function with a number of inodes
- * to trim from the LRU. Inodes to be freed are moved to a temporary list and
- * then are freed outside inode_lock by dispose_list().
+ * Isolate the inode from the LRU in preparation for freeing it.
  *
  * Any inodes which are pinned purely because of attached pagecache have their
  * pagecache removed.  If the inode has metadata buffers attached to
@@ -706,90 +682,79 @@ static int can_unuse(struct inode *inode)
  * LRU does not have strict ordering. Hence we don't want to reclaim inodes
  * with this flag set because they are the inodes that are out of order.
  */
-long prune_icache_sb(struct super_block *sb, unsigned long nr_to_scan)
+static enum lru_status
+inode_lru_isolate(struct list_head *item, spinlock_t *lru_lock, void *arg)
 {
-       LIST_HEAD(freeable);
-       long nr_scanned;
-       long freed = 0;
-       unsigned long reap = 0;
+       struct list_head *freeable = arg;
+       struct inode    *inode = container_of(item, struct inode, i_lru);
 
-       spin_lock(&sb->s_inode_lru_lock);
-       for (nr_scanned = nr_to_scan; nr_scanned >= 0; nr_scanned--) {
-               struct inode *inode;
+       /*
+        * we are inverting the lru lock/inode->i_lock here, so use a trylock.
+        * If we fail to get the lock, just skip it.
+        */
+       if (!spin_trylock(&inode->i_lock))
+               return LRU_SKIP;
 
-               if (list_empty(&sb->s_inode_lru))
-                       break;
+       /*
+        * Referenced or dirty inodes are still in use. Give them another pass
+        * through the LRU as we canot reclaim them now.
+        */
+       if (atomic_read(&inode->i_count) ||
+           (inode->i_state & ~I_REFERENCED)) {
+               list_del_init(&inode->i_lru);
+               spin_unlock(&inode->i_lock);
+               this_cpu_dec(nr_unused);
+               return LRU_REMOVED;
+       }
 
-               inode = list_entry(sb->s_inode_lru.prev, struct inode, i_lru);
+       /* recently referenced inodes get one more pass */
+       if (inode->i_state & I_REFERENCED) {
+               inode->i_state &= ~I_REFERENCED;
+               spin_unlock(&inode->i_lock);
+               return LRU_ROTATE;
+       }
 
-               /*
-                * we are inverting the sb->s_inode_lru_lock/inode->i_lock here,
-                * so use a trylock. If we fail to get the lock, just move the
-                * inode to the back of the list so we don't spin on it.
-                */
-               if (!spin_trylock(&inode->i_lock)) {
-                       list_move(&inode->i_lru, &sb->s_inode_lru);
-                       continue;
+       if (inode_has_buffers(inode) || inode->i_data.nrpages) {
+               __iget(inode);
+               spin_unlock(&inode->i_lock);
+               spin_unlock(lru_lock);
+               if (remove_inode_buffers(inode)) {
+                       unsigned long reap;
+                       reap = invalidate_mapping_pages(&inode->i_data, 0, -1);
+                       if (current_is_kswapd())
+                               __count_vm_events(KSWAPD_INODESTEAL, reap);
+                       else
+                               __count_vm_events(PGINODESTEAL, reap);
+                       if (current->reclaim_state)
+                               current->reclaim_state->reclaimed_slab += reap;
                }
+               iput(inode);
+               spin_lock(lru_lock);
+               return LRU_RETRY;
+       }
 
-               /*
-                * Referenced or dirty inodes are still in use. Give them
-                * another pass through the LRU as we canot reclaim them now.
-                */
-               if (atomic_read(&inode->i_count) ||
-                   (inode->i_state & ~I_REFERENCED)) {
-                       list_del_init(&inode->i_lru);
-                       spin_unlock(&inode->i_lock);
-                       sb->s_nr_inodes_unused--;
-                       this_cpu_dec(nr_unused);
-                       continue;
-               }
+       WARN_ON(inode->i_state & I_NEW);
+       inode->i_state |= I_FREEING;
+       spin_unlock(&inode->i_lock);
 
-               /* recently referenced inodes get one more pass */
-               if (inode->i_state & I_REFERENCED) {
-                       inode->i_state &= ~I_REFERENCED;
-                       list_move(&inode->i_lru, &sb->s_inode_lru);
-                       spin_unlock(&inode->i_lock);
-                       continue;
-               }
-               if (inode_has_buffers(inode) || inode->i_data.nrpages) {
-                       __iget(inode);
-                       spin_unlock(&inode->i_lock);
-                       spin_unlock(&sb->s_inode_lru_lock);
-                       if (remove_inode_buffers(inode))
-                               reap += invalidate_mapping_pages(&inode->i_data,
-                                                               0, -1);
-                       iput(inode);
-                       spin_lock(&sb->s_inode_lru_lock);
-
-                       if (inode != list_entry(sb->s_inode_lru.next,
-                                               struct inode, i_lru))
-                               continue;       /* wrong inode or list_empty */
-                       /* avoid lock inversions with trylock */
-                       if (!spin_trylock(&inode->i_lock))
-                               continue;
-                       if (!can_unuse(inode)) {
-                               spin_unlock(&inode->i_lock);
-                               continue;
-                       }
-               }
-               WARN_ON(inode->i_state & I_NEW);
-               inode->i_state |= I_FREEING;
-               spin_unlock(&inode->i_lock);
+       list_move(&inode->i_lru, freeable);
+       this_cpu_dec(nr_unused);
+       return LRU_REMOVED;
+}
 
-               list_move(&inode->i_lru, &freeable);
-               sb->s_nr_inodes_unused--;
-               this_cpu_dec(nr_unused);
-               freed++;
-       }
-       if (current_is_kswapd())
-               __count_vm_events(KSWAPD_INODESTEAL, reap);
-       else
-               __count_vm_events(PGINODESTEAL, reap);
-       spin_unlock(&sb->s_inode_lru_lock);
-       if (current->reclaim_state)
-               current->reclaim_state->reclaimed_slab += reap;
+/*
+ * Walk the superblock inode LRU for freeable inodes and attempt to free them.
+ * This is called from the superblock shrinker function with a number of inodes
+ * to trim from the LRU. Inodes to be freed are moved to a temporary list and
+ * then are freed outside inode_lock by dispose_list().
+ */
+long prune_icache_sb(struct super_block *sb, unsigned long nr_to_scan)
+{
+       LIST_HEAD(freeable);
+       long freed;
 
+       freed = list_lru_walk(&sb->s_inode_lru, inode_lru_isolate,
+                                               &freeable, nr_to_scan);
        dispose_list(&freeable);
        return freed;
 }
index 00375e8de4ef65dc17737783aecd96bed252fa8d..3784de973f4438eac57f889aa6328a46ea52651b 100644 (file)
@@ -78,14 +78,13 @@ static unsigned long super_cache_scan(struct shrinker *shrink,
        if (sb->s_op && sb->s_op->nr_cached_objects)
                fs_objects = sb->s_op->nr_cached_objects(sb);
 
-       total_objects = sb->s_nr_dentry_unused +
-                       sb->s_nr_inodes_unused + fs_objects + 1;
+       inodes = list_lru_count(&sb->s_inode_lru);
+       total_objects = sb->s_nr_dentry_unused + inodes + fs_objects + 1;
 
        /* proportion the scan between the caches */
        dentries = mult_frac(sc->nr_to_scan, sb->s_nr_dentry_unused,
                                                                total_objects);
-       inodes = mult_frac(sc->nr_to_scan, sb->s_nr_inodes_unused,
-                                                               total_objects);
+       inodes = mult_frac(sc->nr_to_scan, inodes, total_objects);
 
        /*
         * prune the dcache first as the icache is pinned by it, then
@@ -119,7 +118,7 @@ static unsigned long super_cache_count(struct shrinker *shrink,
                total_objects = sb->s_op->nr_cached_objects(sb);
 
        total_objects += sb->s_nr_dentry_unused;
-       total_objects += sb->s_nr_inodes_unused;
+       total_objects += list_lru_count(&sb->s_inode_lru);
 
        total_objects = vfs_pressure_ratio(total_objects);
        drop_super(sb);
@@ -200,8 +199,7 @@ static struct super_block *alloc_super(struct file_system_type *type, int flags)
                INIT_LIST_HEAD(&s->s_inodes);
                INIT_LIST_HEAD(&s->s_dentry_lru);
                spin_lock_init(&s->s_dentry_lru_lock);
-               INIT_LIST_HEAD(&s->s_inode_lru);
-               spin_lock_init(&s->s_inode_lru_lock);
+               list_lru_init(&s->s_inode_lru);
                INIT_LIST_HEAD(&s->s_mounts);
                init_rwsem(&s->s_umount);
                lockdep_set_class(&s->s_umount, &type->s_umount_key);
index 5cbd1735e387dfe1fa31dc533fbfd38bb0065f7e..eab83e232644b5505bac5a213d44b8f8e6429972 100644 (file)
@@ -10,6 +10,7 @@
 #include <linux/stat.h>
 #include <linux/cache.h>
 #include <linux/list.h>
+#include <linux/list_lru.h>
 #include <linux/llist.h>
 #include <linux/radix-tree.h>
 #include <linux/rbtree.h>
@@ -1275,10 +1276,7 @@ struct super_block {
        struct list_head        s_dentry_lru;   /* unused dentry lru */
        long                    s_nr_dentry_unused;     /* # of dentry on lru */
 
-       /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
-       spinlock_t              s_inode_lru_lock ____cacheline_aligned_in_smp;
-       struct list_head        s_inode_lru;            /* unused inode lru */
-       long                    s_nr_inodes_unused;     /* # of inodes on lru */
+       struct list_lru         s_inode_lru ____cacheline_aligned_in_smp;
 
        struct block_device     *s_bdev;
        struct backing_dev_info *s_bdi;