6.6-stable review patch. If anyone has any objections, please let me know.
------------------
From: Ilya Dryomov idryomov@gmail.com
[ Upstream commit 66e0c4f91461d17d48071695271c824620bed4ef ]
The bvecs array which is allocated in iter_get_bvecs_alloc() is leaked and pages remain pinned if ceph_alloc_sparse_ext_map() fails.
There is no need to delay the allocation of sparse_ext map until after the bvecs array is set up, so fix this by moving sparse_ext allocation a bit earlier. Also, make a similar adjustment in __ceph_sync_read() for consistency (a leak of the same kind in __ceph_sync_read() has been addressed differently).
Cc: stable@vger.kernel.org Fixes: 03bc06c7b0bd ("ceph: add new mount option to enable sparse reads") Signed-off-by: Ilya Dryomov idryomov@gmail.com Reviewed-by: Alex Markuze amarkuze@redhat.com Stable-dep-of: 18d44c5d062b ("ceph: allocate sparse_ext map only for sparse reads") Signed-off-by: Sasha Levin sashal@kernel.org --- fs/ceph/file.c | 43 ++++++++++++++++++++++--------------------- 1 file changed, 22 insertions(+), 21 deletions(-)
diff --git a/fs/ceph/file.c b/fs/ceph/file.c index 23dcfb916298..5233bbab8a76 100644 --- a/fs/ceph/file.c +++ b/fs/ceph/file.c @@ -1026,6 +1026,16 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos, len = read_off + read_len - off; more = len < iov_iter_count(to);
+ op = &req->r_ops[0]; + if (sparse) { + extent_cnt = __ceph_sparse_read_ext_count(inode, read_len); + ret = ceph_alloc_sparse_ext_map(op, extent_cnt); + if (ret) { + ceph_osdc_put_request(req); + break; + } + } + num_pages = calc_pages_for(read_off, read_len); page_off = offset_in_page(off); pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL); @@ -1039,16 +1049,6 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos, offset_in_page(read_off), false, true);
- op = &req->r_ops[0]; - if (sparse) { - extent_cnt = __ceph_sparse_read_ext_count(inode, read_len); - ret = ceph_alloc_sparse_ext_map(op, extent_cnt); - if (ret) { - ceph_osdc_put_request(req); - break; - } - } - ceph_osdc_start_request(osdc, req); ret = ceph_osdc_wait_request(osdc, req);
@@ -1454,6 +1454,16 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, break; }
+ op = &req->r_ops[0]; + if (sparse) { + extent_cnt = __ceph_sparse_read_ext_count(inode, size); + ret = ceph_alloc_sparse_ext_map(op, extent_cnt); + if (ret) { + ceph_osdc_put_request(req); + break; + } + } + len = iter_get_bvecs_alloc(iter, size, &bvecs, &num_pages); if (len < 0) { ceph_osdc_put_request(req); @@ -1463,6 +1473,8 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, if (len != size) osd_req_op_extent_update(req, 0, len);
+ osd_req_op_extent_osd_data_bvecs(req, 0, bvecs, num_pages, len); + /* * To simplify error handling, allow AIO when IO within i_size * or IO can be satisfied by single OSD request. @@ -1494,17 +1506,6 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter, req->r_mtime = mtime; }
- osd_req_op_extent_osd_data_bvecs(req, 0, bvecs, num_pages, len); - op = &req->r_ops[0]; - if (sparse) { - extent_cnt = __ceph_sparse_read_ext_count(inode, size); - ret = ceph_alloc_sparse_ext_map(op, extent_cnt); - if (ret) { - ceph_osdc_put_request(req); - break; - } - } - if (aio_req) { aio_req->total_len += len; aio_req->num_reqs++;