summaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/btrfs/extent-tree.c6
-rw-r--r--fs/btrfs/extent_io.c10
-rw-r--r--fs/btrfs/inode.c13
-rw-r--r--fs/btrfs/ordered-data.c6
-rw-r--r--fs/btrfs/ordered-data.h2
-rw-r--r--fs/btrfs/reada.c2
-rw-r--r--fs/btrfs/scrub.c50
-rw-r--r--fs/btrfs/volumes.c32
-rw-r--r--fs/cachefiles/interface.c2
-rw-r--r--fs/ceph/addr.c6
-rw-r--r--fs/ceph/cache.c141
-rw-r--r--fs/ceph/cache.h44
-rw-r--r--fs/ceph/caps.c23
-rw-r--r--fs/ceph/file.c27
-rw-r--r--fs/ceph/super.h4
-rw-r--r--fs/fscache/page.c2
16 files changed, 208 insertions, 162 deletions
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index a400951e8678..689d25ac6a68 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2042,6 +2042,11 @@ int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
struct btrfs_bio *bbio = NULL;
+ /*
+ * Avoid races with device replace and make sure our bbio has devices
+ * associated to its stripes that don't go away while we are discarding.
+ */
+ btrfs_bio_counter_inc_blocked(root->fs_info);
/* Tell the block device(s) that the sectors can be discarded */
ret = btrfs_map_block(root->fs_info, REQ_DISCARD,
bytenr, &num_bytes, &bbio, 0);
@@ -2074,6 +2079,7 @@ int btrfs_discard_extent(struct btrfs_root *root, u64 bytenr,
}
btrfs_put_bbio(bbio);
}
+ btrfs_bio_counter_dec(root->fs_info);
if (actual_bytes)
*actual_bytes = discarded_bytes;
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 3cd57825c75f..6e953de83f08 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2025,9 +2025,16 @@ int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical,
bio->bi_iter.bi_size = 0;
map_length = length;
+ /*
+ * Avoid races with device replace and make sure our bbio has devices
+ * associated to its stripes that don't go away while we are doing the
+ * read repair operation.
+ */
+ btrfs_bio_counter_inc_blocked(fs_info);
ret = btrfs_map_block(fs_info, WRITE, logical,
&map_length, &bbio, mirror_num);
if (ret) {
+ btrfs_bio_counter_dec(fs_info);
bio_put(bio);
return -EIO;
}
@@ -2037,6 +2044,7 @@ int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical,
dev = bbio->stripes[mirror_num-1].dev;
btrfs_put_bbio(bbio);
if (!dev || !dev->bdev || !dev->writeable) {
+ btrfs_bio_counter_dec(fs_info);
bio_put(bio);
return -EIO;
}
@@ -2045,6 +2053,7 @@ int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical,
if (btrfsic_submit_bio_wait(WRITE_SYNC, bio)) {
/* try to remap that extent elsewhere? */
+ btrfs_bio_counter_dec(fs_info);
bio_put(bio);
btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
return -EIO;
@@ -2054,6 +2063,7 @@ int repair_io_failure(struct inode *inode, u64 start, u64 length, u64 logical,
"read error corrected: ino %llu off %llu (dev %s sector %llu)",
btrfs_ino(inode), start,
rcu_str_deref(dev->name), sector);
+ btrfs_bio_counter_dec(fs_info);
bio_put(bio);
return 0;
}
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 270499598ed4..8b1212e8f7a8 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6979,7 +6979,18 @@ insert:
* existing will always be non-NULL, since there must be
* extent causing the -EEXIST.
*/
- if (start >= extent_map_end(existing) ||
+ if (existing->start == em->start &&
+ extent_map_end(existing) == extent_map_end(em) &&
+ em->block_start == existing->block_start) {
+ /*
+ * these two extents are the same, it happens
+ * with inlines especially
+ */
+ free_extent_map(em);
+ em = existing;
+ err = 0;
+
+ } else if (start >= extent_map_end(existing) ||
start <= existing->start) {
/*
* The existing extent map is the one nearest to
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 559170464d7c..e96634a725c3 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -718,12 +718,13 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr,
return count;
}
-void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
+int btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
const u64 range_start, const u64 range_len)
{
struct btrfs_root *root;
struct list_head splice;
int done;
+ int total_done = 0;
INIT_LIST_HEAD(&splice);
@@ -742,6 +743,7 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
done = btrfs_wait_ordered_extents(root, nr,
range_start, range_len);
btrfs_put_fs_root(root);
+ total_done += done;
spin_lock(&fs_info->ordered_root_lock);
if (nr != -1) {
@@ -752,6 +754,8 @@ void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
list_splice_tail(&splice, &fs_info->ordered_roots);
spin_unlock(&fs_info->ordered_root_lock);
mutex_unlock(&fs_info->ordered_operations_mutex);
+
+ return total_done;
}
/*
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index 2049c9be85ee..451507776ff5 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -199,7 +199,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
u32 *sum, int len);
int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr,
const u64 range_start, const u64 range_len);
-void btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
+int btrfs_wait_ordered_roots(struct btrfs_fs_info *fs_info, int nr,
const u64 range_start, const u64 range_len);
void btrfs_get_logged_extents(struct inode *inode,
struct list_head *logged_list,
diff --git a/fs/btrfs/reada.c b/fs/btrfs/reada.c
index 298631eaee78..8428db7cd88f 100644
--- a/fs/btrfs/reada.c
+++ b/fs/btrfs/reada.c
@@ -761,12 +761,14 @@ static void __reada_start_machine(struct btrfs_fs_info *fs_info)
do {
enqueued = 0;
+ mutex_lock(&fs_devices->device_list_mutex);
list_for_each_entry(device, &fs_devices->devices, dev_list) {
if (atomic_read(&device->reada_in_flight) <
MAX_IN_FLIGHT)
enqueued += reada_start_machine_dev(fs_info,
device);
}
+ mutex_unlock(&fs_devices->device_list_mutex);
total += enqueued;
} while (enqueued && total < 10000);
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 46d847f66e4b..70427ef66b04 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -3582,6 +3582,46 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
*/
scrub_pause_on(fs_info);
ret = btrfs_inc_block_group_ro(root, cache);
+ if (!ret && is_dev_replace) {
+ /*
+ * If we are doing a device replace wait for any tasks
+ * that started dellaloc right before we set the block
+ * group to RO mode, as they might have just allocated
+ * an extent from it or decided they could do a nocow
+ * write. And if any such tasks did that, wait for their
+ * ordered extents to complete and then commit the
+ * current transaction, so that we can later see the new
+ * extent items in the extent tree - the ordered extents
+ * create delayed data references (for cow writes) when
+ * they complete, which will be run and insert the
+ * corresponding extent items into the extent tree when
+ * we commit the transaction they used when running
+ * inode.c:btrfs_finish_ordered_io(). We later use
+ * the commit root of the extent tree to find extents
+ * to copy from the srcdev into the tgtdev, and we don't
+ * want to miss any new extents.
+ */
+ btrfs_wait_block_group_reservations(cache);
+ btrfs_wait_nocow_writers(cache);
+ ret = btrfs_wait_ordered_roots(fs_info, -1,
+ cache->key.objectid,
+ cache->key.offset);
+ if (ret > 0) {
+ struct btrfs_trans_handle *trans;
+
+ trans = btrfs_join_transaction(root);
+ if (IS_ERR(trans))
+ ret = PTR_ERR(trans);
+ else
+ ret = btrfs_commit_transaction(trans,
+ root);
+ if (ret) {
+ scrub_pause_off(fs_info);
+ btrfs_put_block_group(cache);
+ break;
+ }
+ }
+ }
scrub_pause_off(fs_info);
if (ret == 0) {
@@ -3602,9 +3642,11 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
break;
}
+ btrfs_dev_replace_lock(&fs_info->dev_replace, 1);
dev_replace->cursor_right = found_key.offset + length;
dev_replace->cursor_left = found_key.offset;
dev_replace->item_needs_writeback = 1;
+ btrfs_dev_replace_unlock(&fs_info->dev_replace, 1);
ret = scrub_chunk(sctx, scrub_dev, chunk_offset, length,
found_key.offset, cache, is_dev_replace);
@@ -3640,6 +3682,11 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
scrub_pause_off(fs_info);
+ btrfs_dev_replace_lock(&fs_info->dev_replace, 1);
+ dev_replace->cursor_left = dev_replace->cursor_right;
+ dev_replace->item_needs_writeback = 1;
+ btrfs_dev_replace_unlock(&fs_info->dev_replace, 1);
+
if (ro_set)
btrfs_dec_block_group_ro(root, cache);
@@ -3677,9 +3724,6 @@ int scrub_enumerate_chunks(struct scrub_ctx *sctx,
ret = -ENOMEM;
break;
}
-
- dev_replace->cursor_left = dev_replace->cursor_right;
- dev_replace->item_needs_writeback = 1;
skip:
key.offset = found_key.offset + length;
btrfs_release_path(path);
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index bdc62561ede8..da9e0036a864 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -2761,6 +2761,7 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
u64 dev_extent_len = 0;
u64 chunk_objectid = BTRFS_FIRST_CHUNK_TREE_OBJECTID;
int i, ret = 0;
+ struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
/* Just in case */
root = root->fs_info->chunk_root;
@@ -2787,12 +2788,19 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
check_system_chunk(trans, extent_root, map->type);
unlock_chunks(root->fs_info->chunk_root);
+ /*
+ * Take the device list mutex to prevent races with the final phase of
+ * a device replace operation that replaces the device object associated
+ * with map stripes (dev-replace.c:btrfs_dev_replace_finishing()).
+ */
+ mutex_lock(&fs_devices->device_list_mutex);
for (i = 0; i < map->num_stripes; i++) {
struct btrfs_device *device = map->stripes[i].dev;
ret = btrfs_free_dev_extent(trans, device,
map->stripes[i].physical,
&dev_extent_len);
if (ret) {
+ mutex_unlock(&fs_devices->device_list_mutex);
btrfs_abort_transaction(trans, root, ret);
goto out;
}
@@ -2811,11 +2819,14 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
if (map->stripes[i].dev) {
ret = btrfs_update_device(trans, map->stripes[i].dev);
if (ret) {
+ mutex_unlock(&fs_devices->device_list_mutex);
btrfs_abort_transaction(trans, root, ret);
goto out;
}
}
}
+ mutex_unlock(&fs_devices->device_list_mutex);
+
ret = btrfs_free_chunk(trans, root, chunk_objectid, chunk_offset);
if (ret) {
btrfs_abort_transaction(trans, root, ret);
@@ -5762,20 +5773,17 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
}
}
if (found) {
- if (physical_of_found + map->stripe_len <=
- dev_replace->cursor_left) {
- struct btrfs_bio_stripe *tgtdev_stripe =
- bbio->stripes + num_stripes;
+ struct btrfs_bio_stripe *tgtdev_stripe =
+ bbio->stripes + num_stripes;
- tgtdev_stripe->physical = physical_of_found;
- tgtdev_stripe->length =
- bbio->stripes[index_srcdev].length;
- tgtdev_stripe->dev = dev_replace->tgtdev;
- bbio->tgtdev_map[index_srcdev] = num_stripes;
+ tgtdev_stripe->physical = physical_of_found;
+ tgtdev_stripe->length =
+ bbio->stripes[index_srcdev].length;
+ tgtdev_stripe->dev = dev_replace->tgtdev;
+ bbio->tgtdev_map[index_srcdev] = num_stripes;
- tgtdev_indexes++;
- num_stripes++;
- }
+ tgtdev_indexes++;
+ num_stripes++;
}
}
diff --git a/fs/cachefiles/interface.c b/fs/cachefiles/interface.c
index 861d611b8c05..ce5f345d70f5 100644
--- a/fs/cachefiles/interface.c
+++ b/fs/cachefiles/interface.c
@@ -380,7 +380,7 @@ static void cachefiles_sync_cache(struct fscache_cache *_cache)
* check if the backing cache is updated to FS-Cache
* - called by FS-Cache when evaluates if need to invalidate the cache
*/
-static bool cachefiles_check_consistency(struct fscache_operation *op)
+static int cachefiles_check_consistency(struct fscache_operation *op)
{
struct cachefiles_object *object;
struct cachefiles_cache *cache;
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index eeb71e5de27a..26a9d10d75e9 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -276,8 +276,10 @@ static void finish_read(struct ceph_osd_request *req)
for (i = 0; i < num_pages; i++) {
struct page *page = osd_data->pages[i];
- if (rc < 0 && rc != -ENOENT)
+ if (rc < 0 && rc != -ENOENT) {
+ ceph_fscache_readpage_cancel(inode, page);
goto unlock;
+ }
if (bytes < (int)PAGE_SIZE) {
/* zero (remainder of) page */
int s = bytes < 0 ? 0 : bytes;
@@ -535,8 +537,6 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
CONGESTION_ON_THRESH(fsc->mount_options->congestion_kb))
set_bdi_congested(&fsc->backing_dev_info, BLK_RW_ASYNC);
- ceph_readpage_to_fscache(inode, page);
-
set_page_writeback(page);
err = ceph_osdc_writepages(osdc, ceph_vino(inode),
&ci->i_layout, snapc,
diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c
index c052b5bf219b..238c55b01723 100644
--- a/fs/ceph/cache.c
+++ b/fs/ceph/cache.c
@@ -25,6 +25,7 @@
#include "cache.h"
struct ceph_aux_inode {
+ u64 version;
struct timespec mtime;
loff_t size;
};
@@ -69,15 +70,8 @@ int ceph_fscache_register_fs(struct ceph_fs_client* fsc)
fsc->fscache = fscache_acquire_cookie(ceph_cache_netfs.primary_index,
&ceph_fscache_fsid_object_def,
fsc, true);
-
- if (fsc->fscache == NULL) {
+ if (!fsc->fscache)
pr_err("Unable to resgister fsid: %p fscache cookie", fsc);
- return 0;
- }
-
- fsc->revalidate_wq = alloc_workqueue("ceph-revalidate", 0, 1);
- if (fsc->revalidate_wq == NULL)
- return -ENOMEM;
return 0;
}
@@ -105,6 +99,7 @@ static uint16_t ceph_fscache_inode_get_aux(const void *cookie_netfs_data,
const struct inode* inode = &ci->vfs_inode;
memset(&aux, 0, sizeof(aux));
+ aux.version = ci->i_version;
aux.mtime = inode->i_mtime;
aux.size = i_size_read(inode);
@@ -131,6 +126,7 @@ static enum fscache_checkaux ceph_fscache_inode_check_aux(
return FSCACHE_CHECKAUX_OBSOLETE;
memset(&aux, 0, sizeof(aux));
+ aux.version = ci->i_version;
aux.mtime = inode->i_mtime;
aux.size = i_size_read(inode);
@@ -181,32 +177,26 @@ static const struct fscache_cookie_def ceph_fscache_inode_object_def = {
.now_uncached = ceph_fscache_inode_now_uncached,
};
-void ceph_fscache_register_inode_cookie(struct ceph_fs_client* fsc,
- struct ceph_inode_info* ci)
+void ceph_fscache_register_inode_cookie(struct inode *inode)
{
- struct inode* inode = &ci->vfs_inode;
+ struct ceph_inode_info *ci = ceph_inode(inode);
+ struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
/* No caching for filesystem */
if (fsc->fscache == NULL)
return;
/* Only cache for regular files that are read only */
- if ((ci->vfs_inode.i_mode & S_IFREG) == 0)
+ if (!S_ISREG(inode->i_mode))
return;
- /* Avoid multiple racing open requests */
- inode_lock(inode);
-
- if (ci->fscache)
- goto done;
-
- ci->fscache = fscache_acquire_cookie(fsc->fscache,
- &ceph_fscache_inode_object_def,
- ci, true);
- fscache_check_consistency(ci->fscache);
-done:
+ inode_lock_nested(inode, I_MUTEX_CHILD);
+ if (!ci->fscache) {
+ ci->fscache = fscache_acquire_cookie(fsc->fscache,
+ &ceph_fscache_inode_object_def,
+ ci, false);
+ }
inode_unlock(inode);
-
}
void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci)
@@ -222,6 +212,34 @@ void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci)
fscache_relinquish_cookie(cookie, 0);
}
+static bool ceph_fscache_can_enable(void *data)
+{
+ struct inode *inode = data;
+ return !inode_is_open_for_write(inode);
+}
+
+void ceph_fscache_file_set_cookie(struct inode *inode, struct file *filp)
+{
+ struct ceph_inode_info *ci = ceph_inode(inode);
+
+ if (!fscache_cookie_valid(ci->fscache))
+ return;
+
+ if (inode_is_open_for_write(inode)) {
+ dout("fscache_file_set_cookie %p %p disabling cache\n",
+ inode, filp);
+ fscache_disable_cookie(ci->fscache, false);
+ fscache_uncache_all_inode_pages(ci->fscache, inode);
+ } else {
+ fscache_enable_cookie(ci->fscache, ceph_fscache_can_enable,
+ inode);
+ if (fscache_cookie_enabled(ci->fscache)) {
+ dout("fscache_file_set_cookie %p %p enabing cache\n",
+ inode, filp);
+ }
+ }
+}
+
static void ceph_vfs_readpage_complete(struct page *page, void *data, int error)
{
if (!error)
@@ -238,8 +256,7 @@ static void ceph_vfs_readpage_complete_unlock(struct page *page, void *data, int
static inline bool cache_valid(struct ceph_inode_info *ci)
{
- return ((ceph_caps_issued(ci) & CEPH_CAP_FILE_CACHE) &&
- (ci->i_fscache_gen == ci->i_rdcache_gen));
+ return ci->i_fscache_gen == ci->i_rdcache_gen;
}
@@ -332,69 +349,27 @@ void ceph_invalidate_fscache_page(struct inode* inode, struct page *page)
void ceph_fscache_unregister_fs(struct ceph_fs_client* fsc)
{
- if (fsc->revalidate_wq)
- destroy_workqueue(fsc->revalidate_wq);
-
fscache_relinquish_cookie(fsc->fscache, 0);
fsc->fscache = NULL;
}
-static void ceph_revalidate_work(struct work_struct *work)
-{
- int issued;
- u32 orig_gen;
- struct ceph_inode_info *ci = container_of(work, struct ceph_inode_info,
- i_revalidate_work);
- struct inode *inode = &ci->vfs_inode;
-
- spin_lock(&ci->i_ceph_lock);
- issued = __ceph_caps_issued(ci, NULL);
- orig_gen = ci->i_rdcache_gen;
- spin_unlock(&ci->i_ceph_lock);
-
- if (!(issued & CEPH_CAP_FILE_CACHE)) {
- dout("revalidate_work lost cache before validation %p\n",
- inode);
- goto out;
- }
-
- if (!fscache_check_consistency(ci->fscache))
- fscache_invalidate(ci->fscache);
-
- spin_lock(&ci->i_ceph_lock);
- /* Update the new valid generation (backwards sanity check too) */
- if (orig_gen > ci->i_fscache_gen) {
- ci->i_fscache_gen = orig_gen;
- }
- spin_unlock(&ci->i_ceph_lock);
-
-out:
- iput(&ci->vfs_inode);
-}
-
-void ceph_queue_revalidate(struct inode *inode)
+/*
+ * caller should hold CEPH_CAP_FILE_{RD,CACHE}
+ */
+void ceph_fscache_revalidate_cookie(struct ceph_inode_info *ci)
{
- struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
- struct ceph_inode_info *ci = ceph_inode(inode);
-
- if (fsc->revalidate_wq == NULL || ci->fscache == NULL)
+ if (cache_valid(ci))
return;
- ihold(inode);
-
- if (queue_work(ceph_sb_to_client(inode->i_sb)->revalidate_wq,
- &ci->i_revalidate_work)) {
- dout("ceph_queue_revalidate %p\n", inode);
- } else {
- dout("ceph_queue_revalidate %p failed\n)", inode);
- iput(inode);
+ /* resue i_truncate_mutex. There should be no pending
+ * truncate while the caller holds CEPH_CAP_FILE_RD */
+ mutex_lock(&ci->i_truncate_mutex);
+ if (!cache_valid(ci)) {
+ if (fscache_check_consistency(ci->fscache))
+ fscache_invalidate(ci->fscache);
+ spin_lock(&ci->i_ceph_lock);
+ ci->i_fscache_gen = ci->i_rdcache_gen;
+ spin_unlock(&ci->i_ceph_lock);
}
-}
-
-void ceph_fscache_inode_init(struct ceph_inode_info *ci)
-{
- ci->fscache = NULL;
- /* The first load is verifed cookie open time */
- ci->i_fscache_gen = 1;
- INIT_WORK(&ci->i_revalidate_work, ceph_revalidate_work);
+ mutex_unlock(&ci->i_truncate_mutex);
}
diff --git a/fs/ceph/cache.h b/fs/ceph/cache.h
index 5ac591bd012b..7e72c7594f0c 100644
--- a/fs/ceph/cache.h
+++ b/fs/ceph/cache.h
@@ -34,10 +34,10 @@ void ceph_fscache_unregister(void);
int ceph_fscache_register_fs(struct ceph_fs_client* fsc);
void ceph_fscache_unregister_fs(struct ceph_fs_client* fsc);
-void ceph_fscache_inode_init(struct ceph_inode_info *ci);
-void ceph_fscache_register_inode_cookie(struct ceph_fs_client* fsc,
- struct ceph_inode_info* ci);
+void ceph_fscache_register_inode_cookie(struct inode *inode);
void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci);
+void ceph_fscache_file_set_cookie(struct inode *inode, struct file *filp);
+void ceph_fscache_revalidate_cookie(struct ceph_inode_info *ci);
int ceph_readpage_from_fscache(struct inode *inode, struct page *page);
int ceph_readpages_from_fscache(struct inode *inode,
@@ -46,12 +46,11 @@ int ceph_readpages_from_fscache(struct inode *inode,
unsigned *nr_pages);
void ceph_readpage_to_fscache(struct inode *inode, struct page *page);
void ceph_invalidate_fscache_page(struct inode* inode, struct page *page);
-void ceph_queue_revalidate(struct inode *inode);
-static inline void ceph_fscache_update_objectsize(struct inode *inode)
+static inline void ceph_fscache_inode_init(struct ceph_inode_info *ci)
{
- struct ceph_inode_info *ci = ceph_inode(inode);
- fscache_attr_changed(ci->fscache);
+ ci->fscache = NULL;
+ ci->i_fscache_gen = 0;
}
static inline void ceph_fscache_invalidate(struct inode *inode)
@@ -88,6 +87,11 @@ static inline void ceph_fscache_readpages_cancel(struct inode *inode,
return fscache_readpages_cancel(ci->fscache, pages);
}
+static inline void ceph_disable_fscache_readpage(struct ceph_inode_info *ci)
+{
+ ci->i_fscache_gen = ci->i_rdcache_gen - 1;
+}
+
#else
static inline int ceph_fscache_register(void)
@@ -112,8 +116,20 @@ static inline void ceph_fscache_inode_init(struct ceph_inode_info *ci)
{
}
-static inline void ceph_fscache_register_inode_cookie(struct ceph_fs_client* parent_fsc,
- struct ceph_inode_info* ci)
+static inline void ceph_fscache_register_inode_cookie(struct inode *inode)
+{
+}
+
+static inline void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci)
+{
+}
+
+static inline void ceph_fscache_file_set_cookie(struct inode *inode,
+ struct file *filp)
+{
+}
+
+static inline void ceph_fscache_revalidate_cookie(struct ceph_inode_info *ci)
{
}
@@ -141,10 +157,6 @@ static inline void ceph_readpage_to_fscache(struct inode *inode,
{
}
-static inline void ceph_fscache_update_objectsize(struct inode *inode)
-{
-}
-
static inline void ceph_fscache_invalidate(struct inode *inode)
{
}
@@ -154,10 +166,6 @@ static inline void ceph_invalidate_fscache_page(struct inode *inode,
{
}
-static inline void ceph_fscache_unregister_inode_cookie(struct ceph_inode_info* ci)
-{
-}
-
static inline int ceph_release_fscache_page(struct page *page, gfp_t gfp)
{
return 1;
@@ -173,7 +181,7 @@ static inline void ceph_fscache_readpages_cancel(struct inode *inode,
{
}
-static inline void ceph_queue_revalidate(struct inode *inode)
+static inline void ceph_disable_fscache_readpage(struct ceph_inode_info *ci)
{
}
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index c17b5d76d75e..6f60d0a3d0f9 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2393,6 +2393,9 @@ again:
snap_rwsem_locked = true;
}
*got = need | (have & want);
+ if ((need & CEPH_CAP_FILE_RD) &&
+ !(*got & CEPH_CAP_FILE_CACHE))
+ ceph_disable_fscache_readpage(ci);
__take_cap_refs(ci, *got, true);
ret = 1;
}
@@ -2554,6 +2557,9 @@ int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
break;
}
+ if ((_got & CEPH_CAP_FILE_RD) && (_got & CEPH_CAP_FILE_CACHE))
+ ceph_fscache_revalidate_cookie(ci);
+
*got = _got;
return 0;
}
@@ -2795,7 +2801,6 @@ static void handle_cap_grant(struct ceph_mds_client *mdsc,
bool writeback = false;
bool queue_trunc = false;
bool queue_invalidate = false;
- bool queue_revalidate = false;
bool deleted_inode = false;
bool fill_inline = false;
@@ -2837,8 +2842,6 @@ static void handle_cap_grant(struct ceph_mds_client *mdsc,
ci->i_rdcache_revoking = ci->i_rdcache_gen;
}
}
-
- ceph_fscache_invalidate(inode);
}
/* side effects now are allowed */
@@ -2880,11 +2883,6 @@ static void handle_cap_grant(struct ceph_mds_client *mdsc,
}
}
- /* Do we need to revalidate our fscache cookie. Don't bother on the
- * first cache cap as we already validate at cookie creation time. */
- if ((issued & CEPH_CAP_FILE_CACHE) && ci->i_rdcache_gen > 1)
- queue_revalidate = true;
-
if (newcaps & CEPH_CAP_ANY_RD) {
/* ctime/mtime/atime? */
ceph_decode_timespec(&mtime, &grant->mtime);
@@ -2993,11 +2991,8 @@ static void handle_cap_grant(struct ceph_mds_client *mdsc,
if (fill_inline)
ceph_fill_inline_data(inode, NULL, inline_data, inline_len);
- if (queue_trunc) {
+ if (queue_trunc)
ceph_queue_vmtruncate(inode);
- ceph_queue_revalidate(inode);
- } else if (queue_revalidate)
- ceph_queue_revalidate(inode);
if (writeback)
/*
@@ -3199,10 +3194,8 @@ static void handle_cap_trunc(struct inode *inode,
truncate_seq, truncate_size, size);
spin_unlock(&ci->i_ceph_lock);
- if (queue_trunc) {
+ if (queue_trunc)
ceph_queue_vmtruncate(inode);
- ceph_fscache_invalidate(inode);
- }
}
/*
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index a888df6f2d71..ce2f5795e44b 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -137,23 +137,11 @@ static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
{
struct ceph_file_info *cf;
int ret = 0;
- struct ceph_inode_info *ci = ceph_inode(inode);
- struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
- struct ceph_mds_client *mdsc = fsc->mdsc;
switch (inode->i_mode & S_IFMT) {
case S_IFREG:
- /* First file open request creates the cookie, we want to keep
- * this cookie around for the filetime of the inode as not to
- * have to worry about fscache register / revoke / operation
- * races.
- *
- * Also, if we know the operation is going to invalidate data
- * (non readonly) just nuke the cache right away.
- */
- ceph_fscache_register_inode_cookie(mdsc->fsc, ci);
- if ((fmode & CEPH_FILE_MODE_WR))
- ceph_fscache_invalidate(inode);
+ ceph_fscache_register_inode_cookie(inode);
+ ceph_fscache_file_set_cookie(inode, file);
case S_IFDIR:
dout("init_file %p %p 0%o (regular)\n", inode, file,
inode->i_mode);
@@ -1349,7 +1337,7 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
}
retry_snap:
- if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL)) {
+ if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL)) {
err = -ENOSPC;
goto out;
}
@@ -1407,7 +1395,6 @@ retry_snap:
iov_iter_advance(from, written);
ceph_put_snap_context(snapc);
} else {
- loff_t old_size = i_size_read(inode);
/*
* No need to acquire the i_truncate_mutex. Because
* the MDS revokes Fwb caps before sending truncate
@@ -1418,8 +1405,6 @@ retry_snap:
written = generic_perform_write(file, from, pos);
if (likely(written >= 0))
iocb->ki_pos = pos + written;
- if (i_size_read(inode) > old_size)
- ceph_fscache_update_objectsize(inode);
inode_unlock(inode);
}
@@ -1440,7 +1425,7 @@ retry_snap:
ceph_put_cap_refs(ci, got);
if (written >= 0) {
- if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_NEARFULL))
+ if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_NEARFULL))
iocb->ki_flags |= IOCB_DSYNC;
written = generic_write_sync(iocb, written);
@@ -1672,8 +1657,8 @@ static long ceph_fallocate(struct file *file, int mode,
goto unlock;
}
- if (ceph_osdmap_flag(osdc->osdmap, CEPH_OSDMAP_FULL) &&
- !(mode & FALLOC_FL_PUNCH_HOLE)) {
+ if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) &&
+ !(mode & FALLOC_FL_PUNCH_HOLE)) {
ret = -ENOSPC;
goto unlock;
}
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 0130a8592191..0168b49fb6ad 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -103,7 +103,6 @@ struct ceph_fs_client {
#ifdef CONFIG_CEPH_FSCACHE
struct fscache_cookie *fscache;
- struct workqueue_struct *revalidate_wq;
#endif
};
@@ -360,8 +359,7 @@ struct ceph_inode_info {
#ifdef CONFIG_CEPH_FSCACHE
struct fscache_cookie *fscache;
- u32 i_fscache_gen; /* sequence, for delayed fscache validate */
- struct work_struct i_revalidate_work;
+ u32 i_fscache_gen;
#endif
struct inode vfs_inode; /* at end */
};
diff --git a/fs/fscache/page.c b/fs/fscache/page.c
index 3078b679fcd1..c8c4f79c7ce1 100644
--- a/fs/fscache/page.c
+++ b/fs/fscache/page.c
@@ -887,6 +887,8 @@ void fscache_invalidate_writes(struct fscache_cookie *cookie)
put_page(results[i]);
}
+ wake_up_bit(&cookie->flags, 0);
+
_leave("");
}
generated by cgit v1.2.3 (git 2.39.1) at 2025-01-24 04:26:40 +0100