Merge tag 'ceph-for-4.16-rc1' of git://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "Things have been very quiet on the rbd side, as work continues on the
  big ticket items slated for the next merge window.

  On the CephFS side we have a large number of cap handling
  improvements, a fix for our long-standing abuse of ->journal_info in
  ceph_readpages() and yet another dentry pointer management patch"

* tag 'ceph-for-4.16-rc1' of git://github.com/ceph/ceph-client:
  ceph: improving efficiency of syncfs
  libceph: check kstrndup() return value
  ceph: try to allocate enough memory for reserved caps
  ceph: fix race of queuing delayed caps
  ceph: delete unreachable code in ceph_check_caps()
  ceph: limit rate of cap import/export error messages
  ceph: fix incorrect snaprealm when adding caps
  ceph: fix un-balanced fsc->writeback_count update
  ceph: track read contexts in ceph_file_info
  ceph: avoid dereferencing invalid pointer during cached readdir
  ceph: use atomic_t for ceph_inode_info::i_shared_gen
  ceph: cleanup traceless reply handling for rename
  ceph: voluntarily drop Fx cap for readdir request
  ceph: properly drop caps for setattr request
  ceph: voluntarily drop Lx cap for link/rename requests
  ceph: voluntarily drop Ax cap for requests that create new inode
  rbd: whitelist RBD_FEATURE_OPERATIONS feature bit
  rbd: don't NULL out ->obj_request in rbd_img_obj_parent_read_full()
  rbd: use kmem_cache_zalloc() in rbd_img_request_create()
  rbd: obj_request->completion is unused

Author: torvalds

drivers/block/rbd.c

@@ -124,11 +124,13 @@ static int atomic_dec_return_safe(atomic_t *v)
 #define RBD_FEATURE_STRIPINGV2		(1ULL<<1)
 #define RBD_FEATURE_EXCLUSIVE_LOCK	(1ULL<<2)
 #define RBD_FEATURE_DATA_POOL		(1ULL<<7)
+#define RBD_FEATURE_OPERATIONS		(1ULL<<8)
 
 #define RBD_FEATURES_ALL	(RBD_FEATURE_LAYERING |		\
 				 RBD_FEATURE_STRIPINGV2 |	\
 				 RBD_FEATURE_EXCLUSIVE_LOCK |	\
-				 RBD_FEATURE_DATA_POOL)
+				 RBD_FEATURE_DATA_POOL |	\
+				 RBD_FEATURE_OPERATIONS)
 
 /* Features supported by this (client software) implementation. */
 
@@ -281,7 +283,6 @@ struct rbd_obj_request {
 	int			result;
 
 	rbd_obj_callback_t	callback;
-	struct completion	completion;
 
 	struct kref		kref;
 };
@@ -1734,10 +1735,7 @@ static void rbd_obj_request_complete(struct rbd_obj_request *obj_request)
 {
 	dout("%s: obj %p cb %p\n", __func__, obj_request,
 		obj_request->callback);
-	if (obj_request->callback)
-		obj_request->callback(obj_request);
-	else
-		complete_all(&obj_request->completion);
+	obj_request->callback(obj_request);
 }
 
 static void rbd_obj_request_error(struct rbd_obj_request *obj_request, int err)
@@ -2013,7 +2011,6 @@ rbd_obj_request_create(enum obj_request_type type)
 	obj_request->which = BAD_WHICH;
 	obj_request->type = type;
 	INIT_LIST_HEAD(&obj_request->links);
-	init_completion(&obj_request->completion);
 	kref_init(&obj_request->kref);
 
 	dout("%s %p\n", __func__, obj_request);
@@ -2129,15 +2126,13 @@ static struct rbd_img_request *rbd_img_request_create(
 {
 	struct rbd_img_request *img_request;
 
-	img_request = kmem_cache_alloc(rbd_img_request_cache, GFP_NOIO);
+	img_request = kmem_cache_zalloc(rbd_img_request_cache, GFP_NOIO);
 	if (!img_request)
 		return NULL;
 
-	img_request->rq = NULL;
 	img_request->rbd_dev = rbd_dev;
 	img_request->offset = offset;
 	img_request->length = length;
-	img_request->flags = 0;
 	if (op_type == OBJ_OP_DISCARD) {
 		img_request_discard_set(img_request);
 		img_request->snapc = snapc;
@@ -2149,11 +2144,8 @@ static struct rbd_img_request *rbd_img_request_create(
 	}
 	if (rbd_dev_parent_get(rbd_dev))
 		img_request_layered_set(img_request);
+
 	spin_lock_init(&img_request->completion_lock);
-	img_request->next_completion = 0;
-	img_request->callback = NULL;
-	img_request->result = 0;
-	img_request->obj_request_count = 0;
 	INIT_LIST_HEAD(&img_request->obj_requests);
 	kref_init(&img_request->kref);
 
@@ -2692,8 +2684,6 @@ static int rbd_img_obj_parent_read_full(struct rbd_obj_request *obj_request)
 
 	parent_request->copyup_pages = NULL;
 	parent_request->copyup_page_count = 0;
-	parent_request->obj_request = NULL;
-	rbd_obj_request_put(obj_request);
 out_err:
 	if (pages)
 		ceph_release_page_vector(pages, page_count);

fs/ceph/addr.c

@@ -299,7 +299,8 @@ static void finish_read(struct ceph_osd_request *req)
  * start an async read(ahead) operation.  return nr_pages we submitted
  * a read for on success, or negative error code.
  */
-static int start_read(struct inode *inode, struct list_head *page_list, int max)
+static int start_read(struct inode *inode, struct ceph_rw_context *rw_ctx,
+		      struct list_head *page_list, int max)
 {
 	struct ceph_osd_client *osdc =
 		&ceph_inode_to_client(inode)->client->osdc;
@@ -316,7 +317,7 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max)
 	int got = 0;
 	int ret = 0;
 
-	if (!current->journal_info) {
+	if (!rw_ctx) {
 		/* caller of readpages does not hold buffer and read caps
 		 * (fadvise, madvise and readahead cases) */
 		int want = CEPH_CAP_FILE_CACHE;
@@ -437,6 +438,8 @@ static int ceph_readpages(struct file *file, struct address_space *mapping,
 {
 	struct inode *inode = file_inode(file);
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
+	struct ceph_file_info *ci = file->private_data;
+	struct ceph_rw_context *rw_ctx;
 	int rc = 0;
 	int max = 0;
 
@@ -449,11 +452,12 @@ static int ceph_readpages(struct file *file, struct address_space *mapping,
 	if (rc == 0)
 		goto out;
 
+	rw_ctx = ceph_find_rw_context(ci);
 	max = fsc->mount_options->rsize >> PAGE_SHIFT;
-	dout("readpages %p file %p nr_pages %d max %d\n",
-	     inode, file, nr_pages, max);
+	dout("readpages %p file %p ctx %p nr_pages %d max %d\n",
+	     inode, file, rw_ctx, nr_pages, max);
 	while (!list_empty(page_list)) {
-		rc = start_read(inode, page_list, max);
+		rc = start_read(inode, rw_ctx, page_list, max);
 		if (rc < 0)
 			goto out;
 	}
@@ -574,7 +578,6 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 	struct ceph_fs_client *fsc;
 	struct ceph_snap_context *snapc, *oldest;
 	loff_t page_off = page_offset(page);
-	long writeback_stat;
 	int err, len = PAGE_SIZE;
 	struct ceph_writeback_ctl ceph_wbc;
 
@@ -615,8 +618,7 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 	dout("writepage %p page %p index %lu on %llu~%u snapc %p seq %lld\n",
 	     inode, page, page->index, page_off, len, snapc, snapc->seq);
 
-	writeback_stat = atomic_long_inc_return(&fsc->writeback_count);
-	if (writeback_stat >
+	if (atomic_long_inc_return(&fsc->writeback_count) >
 	    CONGESTION_ON_THRESH(fsc->mount_options->congestion_kb))
 		set_bdi_congested(inode_to_bdi(inode), BLK_RW_ASYNC);
 
@@ -651,6 +653,11 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 	end_page_writeback(page);
 	ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
 	ceph_put_snap_context(snapc);  /* page's reference */
+
+	if (atomic_long_dec_return(&fsc->writeback_count) <
+	    CONGESTION_OFF_THRESH(fsc->mount_options->congestion_kb))
+		clear_bdi_congested(inode_to_bdi(inode), BLK_RW_ASYNC);
+
 	return err;
 }
 
@@ -1450,9 +1457,10 @@ static int ceph_filemap_fault(struct vm_fault *vmf)
 
 	if ((got & (CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO)) ||
 	    ci->i_inline_version == CEPH_INLINE_NONE) {
-		current->journal_info = vma->vm_file;
+		CEPH_DEFINE_RW_CONTEXT(rw_ctx, got);
+		ceph_add_rw_context(fi, &rw_ctx);
 		ret = filemap_fault(vmf);
-		current->journal_info = NULL;
+		ceph_del_rw_context(fi, &rw_ctx);
 	} else
 		ret = -EAGAIN;