diff options
author | 2017-02-28 15:36:09 -0800 | |
---|---|---|
committer | 2017-02-28 15:36:09 -0800 | |
commit | b2deee2dc06db7cdf99b84346e69bdb9db9baa85 (patch) | |
tree | ceb073fa12c1a9804761ec8ce8911a517b007ed6 /fs/ceph/file.c | |
parent | Merge branch 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm (diff) | |
parent | libceph, rbd, ceph: WRITE | ONDISK -> WRITE (diff) | |
download | linux-b2deee2dc06db7cdf99b84346e69bdb9db9baa85.tar.gz linux-b2deee2dc06db7cdf99b84346e69bdb9db9baa85.tar.bz2 linux-b2deee2dc06db7cdf99b84346e69bdb9db9baa85.zip |
Merge tag 'ceph-for-4.11-rc1' of git://github.com/ceph/ceph-client
Pull ceph updates from Ilya Dryomov:
"This time around we have:
- support for rbd data-pool feature, which enables rbd images on
erasure-coded pools (myself). CEPH_PG_MAX_SIZE has been bumped to
allow erasure-coded profiles with k+m up to 32.
- a patch for ceph_d_revalidate() performance regression introduced
in 4.9, along with some cleanups in the area (Jeff Layton)
- a set of fixes for unsafe ->d_parent accesses in CephFS (Jeff
Layton)
- buffered reads are now processed in rsize windows instead of rasize
windows (Andreas Gerstmayr). The new default for rsize mount option
is 64M.
- ack vs commit distinction is gone, greatly simplifying ->fsync()
and MOSDOpReply handling code (myself)
... also a few filesystem bug fixes from Zheng, a CRUSH sync up (CRUSH
computations are still serialized though) and several minor fixes and
cleanups all over"
* tag 'ceph-for-4.11-rc1' of git://github.com/ceph/ceph-client: (52 commits)
libceph, rbd, ceph: WRITE | ONDISK -> WRITE
libceph: get rid of ack vs commit
ceph: remove special ack vs commit behavior
ceph: tidy some white space in get_nonsnap_parent()
crush: fix dprintk compilation
crush: do is_out test only if we do not collide
ceph: remove req from unsafe list when unregistering it
rbd: constify device_type structure
rbd: kill obj_request->object_name and rbd_segment_name_cache
rbd: store and use obj_request->object_no
rbd: RBD_V{1,2}_DATA_FORMAT macros
rbd: factor out __rbd_osd_req_create()
rbd: set offset and length outside of rbd_obj_request_create()
rbd: support for data-pool feature
rbd: introduce rbd_init_layout()
rbd: use rbd_obj_bytes() more
rbd: remove now unused rbd_obj_request_wait() and helpers
rbd: switch rbd_obj_method_sync() to ceph_osdc_call()
libceph: pass reply buffer length through ceph_osdc_call()
rbd: do away with obj_request in rbd_obj_read_sync()
...
Diffstat (limited to 'fs/ceph/file.c')
-rw-r--r-- | fs/ceph/file.c | 106 |
1 files changed, 7 insertions, 99 deletions
diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 045d30d26624..26cc95421cca 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -283,7 +283,7 @@ int ceph_open(struct inode *inode, struct file *file) spin_lock(&ci->i_ceph_lock); if (__ceph_is_any_real_caps(ci) && (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) { - int mds_wanted = __ceph_caps_mds_wanted(ci); + int mds_wanted = __ceph_caps_mds_wanted(ci, true); int issued = __ceph_caps_issued(ci, NULL); dout("open %p fmode %d want %s issued %s using existing\n", @@ -379,7 +379,8 @@ int ceph_atomic_open(struct inode *dir, struct dentry *dentry, mask |= CEPH_CAP_XATTR_SHARED; req->r_args.open.mask = cpu_to_le32(mask); - req->r_locked_dir = dir; /* caller holds dir->i_mutex */ + req->r_parent = dir; + set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags); err = ceph_mdsc_do_request(mdsc, (flags & (O_CREAT|O_TRUNC)) ? dir : NULL, req); @@ -758,9 +759,7 @@ static void ceph_aio_retry_work(struct work_struct *work) goto out; } - req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE; + req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc); ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid); @@ -794,89 +793,6 @@ out: kfree(aio_work); } -/* - * Write commit request unsafe callback, called to tell us when a - * request is unsafe (that is, in flight--has been handed to the - * messenger to send to its target osd). It is called again when - * we've received a response message indicating the request is - * "safe" (its CEPH_OSD_FLAG_ONDISK flag is set), or when a request - * is completed early (and unsuccessfully) due to a timeout or - * interrupt. - * - * This is used if we requested both an ACK and ONDISK commit reply - * from the OSD. - */ -static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe) -{ - struct ceph_inode_info *ci = ceph_inode(req->r_inode); - - dout("%s %p tid %llu %ssafe\n", __func__, req, req->r_tid, - unsafe ? "un" : ""); - if (unsafe) { - ceph_get_cap_refs(ci, CEPH_CAP_FILE_WR); - spin_lock(&ci->i_unsafe_lock); - list_add_tail(&req->r_unsafe_item, - &ci->i_unsafe_writes); - spin_unlock(&ci->i_unsafe_lock); - - complete_all(&req->r_completion); - } else { - spin_lock(&ci->i_unsafe_lock); - list_del_init(&req->r_unsafe_item); - spin_unlock(&ci->i_unsafe_lock); - ceph_put_cap_refs(ci, CEPH_CAP_FILE_WR); - } -} - -/* - * Wait on any unsafe replies for the given inode. First wait on the - * newest request, and make that the upper bound. Then, if there are - * more requests, keep waiting on the oldest as long as it is still older - * than the original request. - */ -void ceph_sync_write_wait(struct inode *inode) -{ - struct ceph_inode_info *ci = ceph_inode(inode); - struct list_head *head = &ci->i_unsafe_writes; - struct ceph_osd_request *req; - u64 last_tid; - - if (!S_ISREG(inode->i_mode)) - return; - - spin_lock(&ci->i_unsafe_lock); - if (list_empty(head)) - goto out; - - /* set upper bound as _last_ entry in chain */ - - req = list_last_entry(head, struct ceph_osd_request, - r_unsafe_item); - last_tid = req->r_tid; - - do { - ceph_osdc_get_request(req); - spin_unlock(&ci->i_unsafe_lock); - - dout("sync_write_wait on tid %llu (until %llu)\n", - req->r_tid, last_tid); - wait_for_completion(&req->r_done_completion); - ceph_osdc_put_request(req); - - spin_lock(&ci->i_unsafe_lock); - /* - * from here on look at first entry in chain, since we - * only want to wait for anything older than last_tid - */ - if (list_empty(head)) - break; - req = list_first_entry(head, struct ceph_osd_request, - r_unsafe_item); - } while (req->r_tid < last_tid); -out: - spin_unlock(&ci->i_unsafe_lock); -} - static ssize_t ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, struct ceph_snap_context *snapc, @@ -915,9 +831,7 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, if (ret2 < 0) dout("invalidate_inode_pages2_range returned %d\n", ret2); - flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE; + flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; } else { flags = CEPH_OSD_FLAG_READ; } @@ -1116,10 +1030,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos, if (ret < 0) dout("invalidate_inode_pages2_range returned %d\n", ret); - flags = CEPH_OSD_FLAG_ORDERSNAP | - CEPH_OSD_FLAG_ONDISK | - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ACK; + flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE; while ((len = iov_iter_count(from)) > 0) { size_t left; @@ -1165,8 +1076,6 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos, goto out; } - /* get a second commit callback */ - req->r_unsafe_callback = ceph_sync_write_unsafe; req->r_inode = inode; osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0, @@ -1616,8 +1525,7 @@ static int ceph_zero_partial_object(struct inode *inode, ceph_vino(inode), offset, length, 0, 1, op, - CEPH_OSD_FLAG_WRITE | - CEPH_OSD_FLAG_ONDISK, + CEPH_OSD_FLAG_WRITE, NULL, 0, 0, false); if (IS_ERR(req)) { ret = PTR_ERR(req); |