summaryrefslogtreecommitdiff
path: root/fs/ceph/file.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ceph/file.c')
-rw-r--r--fs/ceph/file.c61
1 files changed, 45 insertions, 16 deletions
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index e55fe32c6224..8b79d87eaf46 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -89,13 +89,14 @@ static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
case S_IFDIR:
dout("init_file %p %p 0%o (regular)\n", inode, file,
inode->i_mode);
- cf = kmem_cache_alloc(ceph_file_cachep, GFP_NOFS | __GFP_ZERO);
+ cf = kmem_cache_alloc(ceph_file_cachep, GFP_KERNEL | __GFP_ZERO);
if (cf == NULL) {
ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
return -ENOMEM;
}
cf->fmode = fmode;
cf->next_offset = 2;
+ cf->readdir_cache_idx = -1;
file->private_data = cf;
BUG_ON(inode->i_fop->release != ceph_release);
break;
@@ -324,7 +325,6 @@ int ceph_release(struct inode *inode, struct file *file)
ceph_mdsc_put_request(cf->last_readdir);
kfree(cf->last_name);
kfree(cf->dir_info);
- dput(cf->dentry);
kmem_cache_free(ceph_file_cachep, cf);
/* wake up anyone waiting for caps on this inode */
@@ -483,7 +483,7 @@ static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *i,
}
} else {
num_pages = calc_pages_for(off, len);
- pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+ pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
if (IS_ERR(pages))
return PTR_ERR(pages);
ret = striped_read(inode, off, len, pages,
@@ -557,13 +557,13 @@ static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe)
* objects, rollback on failure, etc.)
*/
static ssize_t
-ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
+ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
+ struct ceph_snap_context *snapc)
{
struct file *file = iocb->ki_filp;
struct inode *inode = file_inode(file);
struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
- struct ceph_snap_context *snapc;
struct ceph_vino vino;
struct ceph_osd_request *req;
struct page **pages;
@@ -600,7 +600,6 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
size_t start;
ssize_t n;
- snapc = ci->i_snap_realm->cached_context;
vino = ceph_vino(inode);
req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
vino, pos, &len, 0,
@@ -614,7 +613,7 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
break;
}
- osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC);
+ osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC, 0);
n = iov_iter_get_pages_alloc(from, &pages, len, &start);
if (unlikely(n < 0)) {
@@ -674,13 +673,13 @@ ceph_sync_direct_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
* objects, rollback on failure, etc.)
*/
static ssize_t
-ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
+ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
+ struct ceph_snap_context *snapc)
{
struct file *file = iocb->ki_filp;
struct inode *inode = file_inode(file);
struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
- struct ceph_snap_context *snapc;
struct ceph_vino vino;
struct ceph_osd_request *req;
struct page **pages;
@@ -717,7 +716,6 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
size_t left;
int n;
- snapc = ci->i_snap_realm->cached_context;
vino = ceph_vino(inode);
req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
vino, pos, &len, 0, 1,
@@ -736,7 +734,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos)
*/
num_pages = (len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
- pages = ceph_alloc_page_vector(num_pages, GFP_NOFS);
+ pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
if (IS_ERR(pages)) {
ret = PTR_ERR(pages);
goto out;
@@ -860,7 +858,7 @@ again:
struct page *page = NULL;
loff_t i_size;
if (retry_op == READ_INLINE) {
- page = __page_cache_alloc(GFP_NOFS);
+ page = __page_cache_alloc(GFP_KERNEL);
if (!page)
return -ENOMEM;
}
@@ -941,6 +939,7 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_osd_client *osdc =
&ceph_sb_to_client(inode->i_sb)->client->osdc;
+ struct ceph_cap_flush *prealloc_cf;
ssize_t count, written = 0;
int err, want, got;
loff_t pos;
@@ -948,6 +947,10 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
if (ceph_snap(inode) != CEPH_NOSNAP)
return -EROFS;
+ prealloc_cf = ceph_alloc_cap_flush();
+ if (!prealloc_cf)
+ return -ENOMEM;
+
mutex_lock(&inode->i_mutex);
/* We can write back this queue in page reclaim */
@@ -996,14 +999,30 @@ retry_snap:
if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
(iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC)) {
+ struct ceph_snap_context *snapc;
struct iov_iter data;
mutex_unlock(&inode->i_mutex);
+
+ spin_lock(&ci->i_ceph_lock);
+ if (__ceph_have_pending_cap_snap(ci)) {
+ struct ceph_cap_snap *capsnap =
+ list_last_entry(&ci->i_cap_snaps,
+ struct ceph_cap_snap,
+ ci_item);
+ snapc = ceph_get_snap_context(capsnap->context);
+ } else {
+ BUG_ON(!ci->i_head_snapc);
+ snapc = ceph_get_snap_context(ci->i_head_snapc);
+ }
+ spin_unlock(&ci->i_ceph_lock);
+
/* we might need to revert back to that point */
data = *from;
if (iocb->ki_flags & IOCB_DIRECT)
- written = ceph_sync_direct_write(iocb, &data, pos);
+ written = ceph_sync_direct_write(iocb, &data, pos,
+ snapc);
else
- written = ceph_sync_write(iocb, &data, pos);
+ written = ceph_sync_write(iocb, &data, pos, snapc);
if (written == -EOLDSNAPC) {
dout("aio_write %p %llx.%llx %llu~%u"
"got EOLDSNAPC, retrying\n",
@@ -1014,6 +1033,7 @@ retry_snap:
}
if (written > 0)
iov_iter_advance(from, written);
+ ceph_put_snap_context(snapc);
} else {
loff_t old_size = inode->i_size;
/*
@@ -1035,7 +1055,8 @@ retry_snap:
int dirty;
spin_lock(&ci->i_ceph_lock);
ci->i_inline_version = CEPH_INLINE_NONE;
- dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
+ dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+ &prealloc_cf);
spin_unlock(&ci->i_ceph_lock);
if (dirty)
__mark_inode_dirty(inode, dirty);
@@ -1059,6 +1080,7 @@ retry_snap:
out:
mutex_unlock(&inode->i_mutex);
out_unlocked:
+ ceph_free_cap_flush(prealloc_cf);
current->backing_dev_info = NULL;
return written ? written : err;
}
@@ -1255,6 +1277,7 @@ static long ceph_fallocate(struct file *file, int mode,
struct ceph_inode_info *ci = ceph_inode(inode);
struct ceph_osd_client *osdc =
&ceph_inode_to_client(inode)->client->osdc;
+ struct ceph_cap_flush *prealloc_cf;
int want, got = 0;
int dirty;
int ret = 0;
@@ -1267,6 +1290,10 @@ static long ceph_fallocate(struct file *file, int mode,
if (!S_ISREG(inode->i_mode))
return -EOPNOTSUPP;
+ prealloc_cf = ceph_alloc_cap_flush();
+ if (!prealloc_cf)
+ return -ENOMEM;
+
mutex_lock(&inode->i_mutex);
if (ceph_snap(inode) != CEPH_NOSNAP) {
@@ -1313,7 +1340,8 @@ static long ceph_fallocate(struct file *file, int mode,
if (!ret) {
spin_lock(&ci->i_ceph_lock);
ci->i_inline_version = CEPH_INLINE_NONE;
- dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
+ dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
+ &prealloc_cf);
spin_unlock(&ci->i_ceph_lock);
if (dirty)
__mark_inode_dirty(inode, dirty);
@@ -1322,6 +1350,7 @@ static long ceph_fallocate(struct file *file, int mode,
ceph_put_cap_refs(ci, got);
unlock:
mutex_unlock(&inode->i_mutex);
+ ceph_free_cap_flush(prealloc_cf);
return ret;
}