Merge tag 'ceph-for-5.4-rc1' of git://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "The highlights are:

   - automatic recovery of a blacklisted filesystem session (Zheng Yan).
     This is disabled by default and can be enabled by mounting with the
     new "recover_session=clean" option.

   - serialize buffered reads and O_DIRECT writes (Jeff Layton). Care is
     taken to avoid serializing O_DIRECT reads and writes with each
     other, this is based on the exclusion scheme from NFS.

   - handle large osdmaps better in the face of fragmented memory
     (myself)

   - don't limit what security.* xattrs can be get or set (Jeff Layton).
     We were overly restrictive here, unnecessarily preventing things
     like file capability sets stored in security.capability from
     working.

   - allow copy_file_range() within the same inode and across different
     filesystems within the same cluster (Luis Henriques)"

* tag 'ceph-for-5.4-rc1' of git://github.com/ceph/ceph-client: (41 commits)
  ceph: call ceph_mdsc_destroy from destroy_fs_client
  libceph: use ceph_kvmalloc() for osdmap arrays
  libceph: avoid a __vmalloc() deadlock in ceph_kvmalloc()
  ceph: allow object copies across different filesystems in the same cluster
  ceph: include ceph_debug.h in cache.c
  ceph: move static keyword to the front of declarations
  rbd: pull rbd_img_request_create() dout out into the callers
  ceph: reconnect connection if session hang in opening state
  libceph: drop unused con parameter of calc_target()
  ceph: use release_pages() directly
  rbd: fix response length parameter for encoded strings
  ceph: allow arbitrary security.* xattrs
  ceph: only set CEPH_I_SEC_INITED if we got a MAC label
  ceph: turn ceph_security_invalidate_secctx into static inline
  ceph: add buffered/direct exclusionary locking for reads and writes
  libceph: handle OSD op ceph_pagelist_append() errors
  ceph: don't return a value from void function
  ceph: don't freeze during write page faults
  ceph: update the mtime when truncating up
  ceph: fix indentation in __get_snap_name()
  ...

Documentation/filesystems/ceph.txt

@@ -158,6 +158,20 @@ Mount Options
         copies.  Currently, it's only used in copy_file_range, which will revert
         to the default VFS implementation if this option is used.
 
+  recover_session=<no|clean>
+	Set auto reconnect mode in the case where the client is blacklisted. The
+	available modes are "no" and "clean". The default is "no".
+
+	* no: never attempt to reconnect when client detects that it has been
+	blacklisted. Operations will generally fail after being blacklisted.
+
+	* clean: client reconnects to the ceph cluster automatically when it
+	detects that it has been blacklisted. During reconnect, client drops
+	dirty data/metadata, invalidates page caches and writable file handles.
+	After reconnect, file locks become stale because the MDS loses track
+	of them. If an inode contains any stale file locks, read/write on the
+	inode is not allowed until applications release all stale file locks.
+
 More Information
 ================
 

drivers/block/rbd.c

@@ -1754,8 +1754,6 @@ static struct rbd_img_request *rbd_img_request_create(
 	mutex_init(&img_request->state_mutex);
 	kref_init(&img_request->kref);
 
-	dout("%s: rbd_dev %p %s -> img %p\n", __func__, rbd_dev,
-	     obj_op_name(op_type), img_request);
 	return img_request;
 }
 
@@ -2944,6 +2942,9 @@ static int rbd_obj_read_from_parent(struct rbd_obj_request *obj_req)
 	__set_bit(IMG_REQ_CHILD, &child_img_req->flags);
 	child_img_req->obj_request = obj_req;
 
+	dout("%s child_img_req %p for obj_req %p\n", __func__, child_img_req,
+	     obj_req);
+
 	if (!rbd_img_is_write(img_req)) {
 		switch (img_req->data_type) {
 		case OBJ_REQUEST_BIO:
@@ -4877,6 +4878,9 @@ static void rbd_queue_workfn(struct work_struct *work)
 	img_request->rq = rq;
 	snapc = NULL; /* img_request consumes a ref */
 
+	dout("%s rbd_dev %p img_req %p %s %llu~%llu\n", __func__, rbd_dev,
+	     img_request, obj_op_name(op_type), offset, length);
+
 	if (op_type == OBJ_OP_DISCARD || op_type == OBJ_OP_ZEROOUT)
 		result = rbd_img_fill_nodata(img_request, offset, length);
 	else
@@ -5669,17 +5673,20 @@ static int rbd_dev_v2_image_size(struct rbd_device *rbd_dev)
 
 static int rbd_dev_v2_object_prefix(struct rbd_device *rbd_dev)
 {
+	size_t size;
 	void *reply_buf;
 	int ret;
 	void *p;
 
-	reply_buf = kzalloc(RBD_OBJ_PREFIX_LEN_MAX, GFP_KERNEL);
+	/* Response will be an encoded string, which includes a length */
+	size = sizeof(__le32) + RBD_OBJ_PREFIX_LEN_MAX;
+	reply_buf = kzalloc(size, GFP_KERNEL);
 	if (!reply_buf)
 		return -ENOMEM;
 
 	ret = rbd_obj_method_sync(rbd_dev, &rbd_dev->header_oid,
 				  &rbd_dev->header_oloc, "get_object_prefix",
-				  NULL, 0, reply_buf, RBD_OBJ_PREFIX_LEN_MAX);
+				  NULL, 0, reply_buf, size);
 	dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret);
 	if (ret < 0)
 		goto out;
@@ -6696,7 +6703,6 @@ static int rbd_dev_image_id(struct rbd_device *rbd_dev)
 	dout("rbd id object name is %s\n", oid.name);
 
 	/* Response will be an encoded string, which includes a length */
-
 	size = sizeof (__le32) + RBD_IMAGE_ID_LEN_MAX;
 	response = kzalloc(size, GFP_NOIO);
 	if (!response) {
@@ -6708,7 +6714,7 @@ static int rbd_dev_image_id(struct rbd_device *rbd_dev)
 
 	ret = rbd_obj_method_sync(rbd_dev, &oid, &rbd_dev->header_oloc,
 				  "get_id", NULL, 0,
-				  response, RBD_IMAGE_ID_LEN_MAX);
+				  response, size);
 	dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret);
 	if (ret == -ENOENT) {
 		image_id = kstrdup("", GFP_KERNEL);

fs/ceph/Makefile

@@ -6,7 +6,7 @@
 obj-$(CONFIG_CEPH_FS) += ceph.o
 
 ceph-y := super.o inode.o dir.o file.o locks.o addr.o ioctl.o \
-	export.o caps.o snap.o xattr.o quota.o \
+	export.o caps.o snap.o xattr.o quota.o io.o \
 	mds_client.o mdsmap.o strings.o ceph_frag.o \
 	debugfs.o
 

fs/ceph/addr.c

@@ -189,8 +189,7 @@ static int ceph_do_readpage(struct file *filp, struct page *page)
 {
 	struct inode *inode = file_inode(filp);
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_osd_client *osdc =
-		&ceph_inode_to_client(inode)->client->osdc;
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 	int err = 0;
 	u64 off = page_offset(page);
 	u64 len = PAGE_SIZE;
@@ -219,8 +218,8 @@ static int ceph_do_readpage(struct file *filp, struct page *page)
 
 	dout("readpage inode %p file %p page %p index %lu\n",
 	     inode, filp, page, page->index);
-	err = ceph_osdc_readpages(osdc, ceph_vino(inode), &ci->i_layout,
-				  off, &len,
+	err = ceph_osdc_readpages(&fsc->client->osdc, ceph_vino(inode),
+				  &ci->i_layout, off, &len,
 				  ci->i_truncate_seq, ci->i_truncate_size,
 				  &page, 1, 0);
 	if (err == -ENOENT)
@@ -228,6 +227,8 @@ static int ceph_do_readpage(struct file *filp, struct page *page)
 	if (err < 0) {
 		SetPageError(page);
 		ceph_fscache_readpage_cancel(inode, page);
+		if (err == -EBLACKLISTED)
+			fsc->blacklisted = true;
 		goto out;
 	}
 	if (err < PAGE_SIZE)
@@ -266,6 +267,8 @@ static void finish_read(struct ceph_osd_request *req)
 	int i;
 
 	dout("finish_read %p req %p rc %d bytes %d\n", inode, req, rc, bytes);
+	if (rc == -EBLACKLISTED)
+		ceph_inode_to_client(inode)->blacklisted = true;
 
 	/* unlock all pages, zeroing any data we didn't read */
 	osd_data = osd_req_op_extent_osd_data(req, 0);
@@ -323,7 +326,8 @@ static int start_read(struct inode *inode, struct ceph_rw_context *rw_ctx,
 		/* caller of readpages does not hold buffer and read caps
 		 * (fadvise, madvise and readahead cases) */
 		int want = CEPH_CAP_FILE_CACHE;
-		ret = ceph_try_get_caps(ci, CEPH_CAP_FILE_RD, want, true, &got);
+		ret = ceph_try_get_caps(inode, CEPH_CAP_FILE_RD, want,
+					true, &got);
 		if (ret < 0) {
 			dout("start_read %p, error getting cap\n", inode);
 		} else if (!(got & want)) {
@@ -569,7 +573,7 @@ static u64 get_writepages_data_length(struct inode *inode,
 /*
  * Write a single page, but leave the page locked.
  *
- * If we get a write error, set the page error bit, but still adjust the
+ * If we get a write error, mark the mapping for error, but still adjust the
  * dirty page accounting (i.e., page is no longer dirty).
  */
 static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
@@ -640,9 +644,10 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 			end_page_writeback(page);
 			return err;
 		}
+		if (err == -EBLACKLISTED)
+			fsc->blacklisted = true;
 		dout("writepage setting page/mapping error %d %p\n",
 		     err, page);
-		SetPageError(page);
 		mapping_set_error(&inode->i_data, err);
 		wbc->pages_skipped++;
 	} else {
@@ -679,23 +684,6 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
 	return err;
 }
 
-/*
- * lame release_pages helper.  release_pages() isn't exported to
- * modules.
- */
-static void ceph_release_pages(struct page **pages, int num)
-{
-	struct pagevec pvec;
-	int i;
-
-	pagevec_init(&pvec);
-	for (i = 0; i < num; i++) {
-		if (pagevec_add(&pvec, pages[i]) == 0)
-			pagevec_release(&pvec);
-	}
-	pagevec_release(&pvec);
-}
-
 /*
  * async writeback completion handler.
  *
@@ -720,6 +708,8 @@ static void writepages_finish(struct ceph_osd_request *req)
 	if (rc < 0) {
 		mapping_set_error(mapping, rc);
 		ceph_set_error_write(ci);
+		if (rc == -EBLACKLISTED)
+			fsc->blacklisted = true;
 	} else {
 		ceph_clear_error_write(ci);
 	}
@@ -769,7 +759,7 @@ static void writepages_finish(struct ceph_osd_request *req)
 		dout("writepages_finish %p wrote %llu bytes cleaned %d pages\n",
 		     inode, osd_data->length, rc >= 0 ? num_pages : 0);
 
-		ceph_release_pages(osd_data->pages, num_pages);
+		release_pages(osd_data->pages, num_pages);
 	}
 
 	ceph_put_wrbuffer_cap_refs(ci, total_pages, snapc);
@@ -1452,7 +1442,8 @@ static vm_fault_t ceph_filemap_fault(struct vm_fault *vmf)
 		want = CEPH_CAP_FILE_CACHE;
 
 	got = 0;
-	err = ceph_get_caps(ci, CEPH_CAP_FILE_RD, want, -1, &got, &pinned_page);
+	err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_RD, want, -1,
+			    &got, &pinned_page);
 	if (err < 0)
 		goto out_restore;
 
@@ -1540,6 +1531,7 @@ static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf)
 	if (!prealloc_cf)
 		return VM_FAULT_OOM;
 
+	sb_start_pagefault(inode->i_sb);
 	ceph_block_sigs(&oldset);
 
 	if (ci->i_inline_version != CEPH_INLINE_NONE) {
@@ -1568,7 +1560,7 @@ static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf)
 		want = CEPH_CAP_FILE_BUFFER;
 
 	got = 0;
-	err = ceph_get_caps(ci, CEPH_CAP_FILE_WR, want, off + len,
+	err = ceph_get_caps(vma->vm_file, CEPH_CAP_FILE_WR, want, off + len,
 			    &got, NULL);
 	if (err < 0)
 		goto out_free;
@@ -1614,6 +1606,7 @@ static vm_fault_t ceph_page_mkwrite(struct vm_fault *vmf)
 	ceph_put_cap_refs(ci, got);
 out_free:
 	ceph_restore_sigs(&oldset);
+	sb_end_pagefault(inode->i_sb);
 	ceph_free_cap_flush(prealloc_cf);
 	if (err < 0)
 		ret = vmf_error(err);
@@ -1946,12 +1939,17 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
 
 	if (err >= 0 || err == -ENOENT)
 		have |= POOL_READ;
-	else if (err != -EPERM)
+	else if (err != -EPERM) {
+		if (err == -EBLACKLISTED)
+			fsc->blacklisted = true;
 		goto out_unlock;
+	}
 
 	if (err2 == 0 || err2 == -EEXIST)
 		have |= POOL_WRITE;
 	else if (err2 != -EPERM) {
+		if (err2 == -EBLACKLISTED)
+			fsc->blacklisted = true;
 		err = err2;
 		goto out_unlock;
 	}
@@ -1989,10 +1987,11 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
 	return err;
 }
 
-int ceph_pool_perm_check(struct ceph_inode_info *ci, int need)
+int ceph_pool_perm_check(struct inode *inode, int need)
 {
-	s64 pool;
+	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_string *pool_ns;
+	s64 pool;
 	int ret, flags;
 
 	if (ci->i_vino.snap != CEPH_NOSNAP) {
@@ -2004,7 +2003,7 @@ int ceph_pool_perm_check(struct ceph_inode_info *ci, int need)
 		return 0;
 	}
 
-	if (ceph_test_mount_opt(ceph_inode_to_client(&ci->vfs_inode),
+	if (ceph_test_mount_opt(ceph_inode_to_client(inode),
 				NOPOOLPERM))
 		return 0;
 

fs/ceph/cache.c

@@ -6,6 +6,8 @@
  *  Written by Milosz Tanski (milosz@adfin.com)
  */
 
+#include <linux/ceph/ceph_debug.h>
+
 #include "super.h"
 #include "cache.h"
 

fs/ceph/debugfs.c

@@ -294,7 +294,6 @@ void ceph_fs_debugfs_init(struct ceph_fs_client *fsc)
 
 void ceph_fs_debugfs_init(struct ceph_fs_client *fsc)
 {
-	return 0;
 }
 
 void ceph_fs_debugfs_cleanup(struct ceph_fs_client *fsc)

fs/ceph/export.c

@@ -35,7 +35,7 @@ struct ceph_nfs_snapfh {
 static int ceph_encode_snapfh(struct inode *inode, u32 *rawfh, int *max_len,
 			      struct inode *parent_inode)
 {
-	const static int snap_handle_length =
+	static const int snap_handle_length =
 		sizeof(struct ceph_nfs_snapfh) >> 2;
 	struct ceph_nfs_snapfh *sfh = (void *)rawfh;
 	u64 snapid = ceph_snap(inode);
@@ -85,9 +85,9 @@ static int ceph_encode_snapfh(struct inode *inode, u32 *rawfh, int *max_len,
 static int ceph_encode_fh(struct inode *inode, u32 *rawfh, int *max_len,
 			  struct inode *parent_inode)
 {
-	const static int handle_length =
+	static const int handle_length =
 		sizeof(struct ceph_nfs_fh) >> 2;
-	const static int connected_handle_length =
+	static const int connected_handle_length =
 		sizeof(struct ceph_nfs_confh) >> 2;
 	int type;
 
@@ -458,33 +458,33 @@ static int __get_snap_name(struct dentry *parent, char *name,
 		if (err < 0)
 			goto out;
 
-		 rinfo = &req->r_reply_info;
-		 for (i = 0; i < rinfo->dir_nr; i++) {
-			 rde = rinfo->dir_entries + i;
-			 BUG_ON(!rde->inode.in);
-			 if (ceph_snap(inode) ==
-			     le64_to_cpu(rde->inode.in->snapid)) {
-				 memcpy(name, rde->name, rde->name_len);
-				 name[rde->name_len] = '\0';
-				 err = 0;
-				 goto out;
-			 }
-		 }
-
-		 if (rinfo->dir_end)
-			 break;
-
-		 BUG_ON(rinfo->dir_nr <= 0);
-		 rde = rinfo->dir_entries + (rinfo->dir_nr - 1);
-		 next_offset += rinfo->dir_nr;
-		 last_name = kstrndup(rde->name, rde->name_len, GFP_KERNEL);
-		 if (!last_name) {
-			 err = -ENOMEM;
-			 goto out;
-		 }
-
-		 ceph_mdsc_put_request(req);
-		 req = NULL;
+		rinfo = &req->r_reply_info;
+		for (i = 0; i < rinfo->dir_nr; i++) {
+			rde = rinfo->dir_entries + i;
+			BUG_ON(!rde->inode.in);
+			if (ceph_snap(inode) ==
+			    le64_to_cpu(rde->inode.in->snapid)) {
+				memcpy(name, rde->name, rde->name_len);
+				name[rde->name_len] = '\0';
+				err = 0;
+				goto out;
+			}
+		}
+
+		if (rinfo->dir_end)
+			break;
+
+		BUG_ON(rinfo->dir_nr <= 0);
+		rde = rinfo->dir_entries + (rinfo->dir_nr - 1);
+		next_offset += rinfo->dir_nr;
+		last_name = kstrndup(rde->name, rde->name_len, GFP_KERNEL);
+		if (!last_name) {
+			err = -ENOMEM;
+			goto out;
+		}
+
+		ceph_mdsc_put_request(req);
+		req = NULL;
 	}
 	err = -ENOENT;
 out:

fs/ceph/inode.c

@@ -515,6 +515,8 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
 
 	ceph_fscache_inode_init(ci);
 
+	ci->i_meta_err = 0;
+
 	return &ci->vfs_inode;
 }
 
@@ -801,7 +803,12 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 
 	/* update inode */
 	inode->i_rdev = le32_to_cpu(info->rdev);
-	inode->i_blkbits = fls(le32_to_cpu(info->layout.fl_stripe_unit)) - 1;
+	/* directories have fl_stripe_unit set to zero */
+	if (le32_to_cpu(info->layout.fl_stripe_unit))
+		inode->i_blkbits =
+			fls(le32_to_cpu(info->layout.fl_stripe_unit)) - 1;
+	else
+		inode->i_blkbits = CEPH_BLOCK_SHIFT;
 
 	__ceph_update_quota(ci, iinfo->max_bytes, iinfo->max_files);
 
@@ -1982,7 +1989,7 @@ static const struct inode_operations ceph_symlink_iops = {
 int __ceph_setattr(struct inode *inode, struct iattr *attr)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	const unsigned int ia_valid = attr->ia_valid;
+	unsigned int ia_valid = attr->ia_valid;
 	struct ceph_mds_request *req;
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
 	struct ceph_cap_flush *prealloc_cf;
@@ -2087,6 +2094,26 @@ int __ceph_setattr(struct inode *inode, struct iattr *attr)
 				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
 		}
 	}
+	if (ia_valid & ATTR_SIZE) {
+		dout("setattr %p size %lld -> %lld\n", inode,
+		     inode->i_size, attr->ia_size);
+		if ((issued & CEPH_CAP_FILE_EXCL) &&
+		    attr->ia_size > inode->i_size) {
+			i_size_write(inode, attr->ia_size);
+			inode->i_blocks = calc_inode_blocks(attr->ia_size);
+			ci->i_reported_size = attr->ia_size;
+			dirtied |= CEPH_CAP_FILE_EXCL;
+			ia_valid |= ATTR_MTIME;
+		} else if ((issued & CEPH_CAP_FILE_SHARED) == 0 ||
+			   attr->ia_size != inode->i_size) {
+			req->r_args.setattr.size = cpu_to_le64(attr->ia_size);
+			req->r_args.setattr.old_size =
+				cpu_to_le64(inode->i_size);
+			mask |= CEPH_SETATTR_SIZE;
+			release |= CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL |
+				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
+		}
+	}
 	if (ia_valid & ATTR_MTIME) {
 		dout("setattr %p mtime %lld.%ld -> %lld.%ld\n", inode,
 		     inode->i_mtime.tv_sec, inode->i_mtime.tv_nsec,
@@ -2109,25 +2136,6 @@ int __ceph_setattr(struct inode *inode, struct iattr *attr)
 				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
 		}
 	}
-	if (ia_valid & ATTR_SIZE) {
-		dout("setattr %p size %lld -> %lld\n", inode,
-		     inode->i_size, attr->ia_size);
-		if ((issued & CEPH_CAP_FILE_EXCL) &&
-		    attr->ia_size > inode->i_size) {
-			i_size_write(inode, attr->ia_size);
-			inode->i_blocks = calc_inode_blocks(attr->ia_size);
-			ci->i_reported_size = attr->ia_size;
-			dirtied |= CEPH_CAP_FILE_EXCL;
-		} else if ((issued & CEPH_CAP_FILE_SHARED) == 0 ||
-			   attr->ia_size != inode->i_size) {
-			req->r_args.setattr.size = cpu_to_le64(attr->ia_size);
-			req->r_args.setattr.old_size =
-				cpu_to_le64(inode->i_size);
-			mask |= CEPH_SETATTR_SIZE;
-			release |= CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL |
-				   CEPH_CAP_FILE_RD | CEPH_CAP_FILE_WR;
-		}
-	}
 
 	/* these do nothing */
 	if (ia_valid & ATTR_CTIME) {

fs/ceph/io.c

+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2016 Trond Myklebust
+ * Copyright (c) 2019 Jeff Layton
+ *
+ * I/O and data path helper functionality.
+ *
+ * Heavily borrowed from equivalent code in fs/nfs/io.c
+ */
+
+#include <linux/ceph/ceph_debug.h>
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/rwsem.h>
+#include <linux/fs.h>
+
+#include "super.h"
+#include "io.h"
+
+/* Call with exclusively locked inode->i_rwsem */
+static void ceph_block_o_direct(struct ceph_inode_info *ci, struct inode *inode)
+{
+	lockdep_assert_held_write(&inode->i_rwsem);
+
+	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags &= ~CEPH_I_ODIRECT;
+		spin_unlock(&ci->i_ceph_lock);
+		inode_dio_wait(inode);
+	}
+}
+
+/**
+ * ceph_start_io_read - declare the file is being used for buffered reads
+ * @inode: file inode
+ *
+ * Declare that a buffered read operation is about to start, and ensure
+ * that we block all direct I/O.
+ * On exit, the function ensures that the CEPH_I_ODIRECT flag is unset,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that buffered read operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas direct I/O
+ * operations need to wait to grab an exclusive lock in order to set
+ * CEPH_I_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. the reads.
+ */
+void
+ceph_start_io_read(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT))
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	ceph_block_o_direct(ci, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * ceph_end_io_read - declare that the buffered read operation is done
+ * @inode: file inode
+ *
+ * Declare that a buffered read operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_read(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}
+
+/**
+ * ceph_start_io_write - declare the file is being used for buffered writes
+ * @inode: file inode
+ *
+ * Declare that a buffered write operation is about to start, and ensure
+ * that we block all direct I/O.
+ */
+void
+ceph_start_io_write(struct inode *inode)
+{
+	down_write(&inode->i_rwsem);
+	ceph_block_o_direct(ceph_inode(inode), inode);
+}
+
+/**
+ * ceph_end_io_write - declare that the buffered write operation is done
+ * @inode: file inode
+ *
+ * Declare that a buffered write operation is done, and release the
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_write(struct inode *inode)
+{
+	up_write(&inode->i_rwsem);
+}
+
+/* Call with exclusively locked inode->i_rwsem */
+static void ceph_block_buffered(struct ceph_inode_info *ci, struct inode *inode)
+{
+	lockdep_assert_held_write(&inode->i_rwsem);
+
+	if (!(READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT)) {
+		spin_lock(&ci->i_ceph_lock);
+		ci->i_ceph_flags |= CEPH_I_ODIRECT;
+		spin_unlock(&ci->i_ceph_lock);
+		/* FIXME: unmap_mapping_range? */
+		filemap_write_and_wait(inode->i_mapping);
+	}
+}
+
+/**
+ * ceph_end_io_direct - declare the file is being used for direct i/o
+ * @inode: file inode
+ *
+ * Declare that a direct I/O operation is about to start, and ensure
+ * that we block all buffered I/O.
+ * On exit, the function ensures that the CEPH_I_ODIRECT flag is set,
+ * and holds a shared lock on inode->i_rwsem to ensure that the flag
+ * cannot be changed.
+ * In practice, this means that direct I/O operations are allowed to
+ * execute in parallel, thanks to the shared lock, whereas buffered I/O
+ * operations need to wait to grab an exclusive lock in order to clear
+ * CEPH_I_ODIRECT.
+ * Note that buffered writes and truncates both take a write lock on
+ * inode->i_rwsem, meaning that those are serialised w.r.t. O_DIRECT.
+ */
+void
+ceph_start_io_direct(struct inode *inode)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	/* Be an optimist! */
+	down_read(&inode->i_rwsem);
+	if (READ_ONCE(ci->i_ceph_flags) & CEPH_I_ODIRECT)
+		return;
+	up_read(&inode->i_rwsem);
+	/* Slow path.... */
+	down_write(&inode->i_rwsem);
+	ceph_block_buffered(ci, inode);
+	downgrade_write(&inode->i_rwsem);
+}
+
+/**
+ * ceph_end_io_direct - declare that the direct i/o operation is done
+ * @inode: file inode
+ *
+ * Declare that a direct I/O operation is done, and release the shared
+ * lock on inode->i_rwsem.
+ */
+void
+ceph_end_io_direct(struct inode *inode)
+{
+	up_read(&inode->i_rwsem);
+}

fs/ceph/io.h

+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _FS_CEPH_IO_H
+#define _FS_CEPH_IO_H
+
+void ceph_start_io_read(struct inode *inode);
+void ceph_end_io_read(struct inode *inode);
+void ceph_start_io_write(struct inode *inode);
+void ceph_end_io_write(struct inode *inode);
+void ceph_start_io_direct(struct inode *inode);
+void ceph_end_io_direct(struct inode *inode);
+
+#endif /* FS_CEPH_IO_H */

fs/ceph/locks.c

@@ -32,14 +32,18 @@ void __init ceph_flock_init(void)
 
 static void ceph_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
 {
-	struct inode *inode = file_inode(src->fl_file);
+	struct ceph_file_info *fi = dst->fl_file->private_data;
+	struct inode *inode = file_inode(dst->fl_file);
 	atomic_inc(&ceph_inode(inode)->i_filelock_ref);
+	atomic_inc(&fi->num_locks);
 }
 
 static void ceph_fl_release_lock(struct file_lock *fl)
 {
+	struct ceph_file_info *fi = fl->fl_file->private_data;
 	struct inode *inode = file_inode(fl->fl_file);
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	atomic_dec(&fi->num_locks);
 	if (atomic_dec_and_test(&ci->i_filelock_ref)) {
 		/* clear error when all locks are released */
 		spin_lock(&ci->i_ceph_lock);
@@ -73,7 +77,7 @@ static int ceph_lock_message(u8 lock_type, u16 operation, struct inode *inode,
 		 * window. Caller function will decrease the counter.
 		 */
 		fl->fl_ops = &ceph_fl_lock_ops;
-		atomic_inc(&ceph_inode(inode)->i_filelock_ref);
+		fl->fl_ops->fl_copy_lock(fl, NULL);
 	}
 
 	if (operation != CEPH_MDS_OP_SETFILELOCK || cmd == CEPH_LOCK_UNLOCK)

fs/ceph/mds_client.c

@@ -639,7 +639,6 @@ static struct ceph_mds_session *register_session(struct ceph_mds_client *mdsc,
 	s->s_renew_seq = 0;
 	INIT_LIST_HEAD(&s->s_caps);
 	s->s_nr_caps = 0;
-	s->s_trim_caps = 0;
 	refcount_set(&s->s_ref, 1);
 	INIT_LIST_HEAD(&s->s_waiting);
 	INIT_LIST_HEAD(&s->s_unsafe);
@@ -1270,6 +1269,7 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc,
 {
 	struct ceph_mds_request *req;
 	struct rb_node *p;
+	struct ceph_inode_info *ci;
 
 	dout("cleanup_session_requests mds%d\n", session->s_mds);
 	mutex_lock(&mdsc->mutex);
@@ -1278,6 +1278,16 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc,
 				       struct ceph_mds_request, r_unsafe_item);
 		pr_warn_ratelimited(" dropping unsafe request %llu\n",
 				    req->r_tid);
+		if (req->r_target_inode) {
+			/* dropping unsafe change of inode's attributes */
+			ci = ceph_inode(req->r_target_inode);
+			errseq_set(&ci->i_meta_err, -EIO);
+		}
+		if (req->r_unsafe_dir) {
+			/* dropping unsafe directory operation */
+			ci = ceph_inode(req->r_unsafe_dir);
+			errseq_set(&ci->i_meta_err, -EIO);
+		}
 		__unregister_request(mdsc, req);
 	}
 	/* zero r_attempts, so kick_requests() will re-send requests */
@@ -1370,7 +1380,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 	struct ceph_fs_client *fsc = (struct ceph_fs_client *)arg;
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	LIST_HEAD(to_remove);
-	bool drop = false;
+	bool dirty_dropped = false;
 	bool invalidate = false;
 
 	dout("removing cap %p, ci is %p, inode is %p\n",
@@ -1383,9 +1393,12 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 		struct ceph_cap_flush *cf;
 		struct ceph_mds_client *mdsc = fsc->mdsc;
 
-		if (ci->i_wrbuffer_ref > 0 &&
-		    READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
-			invalidate = true;
+		if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+			if (inode->i_data.nrpages > 0)
+				invalidate = true;
+			if (ci->i_wrbuffer_ref > 0)
+				mapping_set_error(&inode->i_data, -EIO);
+		}
 
 		while (!list_empty(&ci->i_cap_flush_list)) {
 			cf = list_first_entry(&ci->i_cap_flush_list,
@@ -1405,7 +1418,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 				inode, ceph_ino(inode));
 			ci->i_dirty_caps = 0;
 			list_del_init(&ci->i_dirty_item);
-			drop = true;
+			dirty_dropped = true;
 		}
 		if (!list_empty(&ci->i_flushing_item)) {
 			pr_warn_ratelimited(
@@ -1415,10 +1428,22 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 			ci->i_flushing_caps = 0;
 			list_del_init(&ci->i_flushing_item);
 			mdsc->num_cap_flushing--;
-			drop = true;
+			dirty_dropped = true;
 		}
 		spin_unlock(&mdsc->cap_dirty_lock);
 
+		if (dirty_dropped) {
+			errseq_set(&ci->i_meta_err, -EIO);
+
+			if (ci->i_wrbuffer_ref_head == 0 &&
+			    ci->i_wr_ref == 0 &&
+			    ci->i_dirty_caps == 0 &&
+			    ci->i_flushing_caps == 0) {
+				ceph_put_snap_context(ci->i_head_snapc);
+				ci->i_head_snapc = NULL;
+			}
+		}
+
 		if (atomic_read(&ci->i_filelock_ref) > 0) {
 			/* make further file lock syscall return -EIO */
 			ci->i_ceph_flags |= CEPH_I_ERROR_FILELOCK;
@@ -1430,15 +1455,6 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 			list_add(&ci->i_prealloc_cap_flush->i_list, &to_remove);
 			ci->i_prealloc_cap_flush = NULL;
 		}
-
-               if (drop &&
-                  ci->i_wrbuffer_ref_head == 0 &&
-                  ci->i_wr_ref == 0 &&
-                  ci->i_dirty_caps == 0 &&
-                  ci->i_flushing_caps == 0) {
-                      ceph_put_snap_context(ci->i_head_snapc);
-                      ci->i_head_snapc = NULL;
-               }
 	}
 	spin_unlock(&ci->i_ceph_lock);
 	while (!list_empty(&to_remove)) {
@@ -1452,7 +1468,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 	wake_up_all(&ci->i_cap_wq);
 	if (invalidate)
 		ceph_queue_invalidate(inode);
-	if (drop)
+	if (dirty_dropped)
 		iput(inode);
 	return 0;
 }
@@ -1705,11 +1721,11 @@ static bool drop_negative_children(struct dentry *dentry)
  */
 static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 {
-	struct ceph_mds_session *session = arg;
+	int *remaining = arg;
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int used, wanted, oissued, mine;
 
-	if (session->s_trim_caps <= 0)
+	if (*remaining <= 0)
 		return -1;
 
 	spin_lock(&ci->i_ceph_lock);
@@ -1746,7 +1762,7 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 	if (oissued) {
 		/* we aren't the only cap.. just remove us */
 		__ceph_remove_cap(cap, true);
-		session->s_trim_caps--;
+		(*remaining)--;
 	} else {
 		struct dentry *dentry;
 		/* try dropping referring dentries */
@@ -1758,7 +1774,7 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 			d_prune_aliases(inode);
 			count = atomic_read(&inode->i_count);
 			if (count == 1)
-				session->s_trim_caps--;
+				(*remaining)--;
 			dout("trim_caps_cb %p cap %p pruned, count now %d\n",
 			     inode, cap, count);
 		} else {
@@ -1784,12 +1800,12 @@ int ceph_trim_caps(struct ceph_mds_client *mdsc,
 	dout("trim_caps mds%d start: %d / %d, trim %d\n",
 	     session->s_mds, session->s_nr_caps, max_caps, trim_caps);
 	if (trim_caps > 0) {
-		session->s_trim_caps = trim_caps;
-		ceph_iterate_session_caps(session, trim_caps_cb, session);
+		int remaining = trim_caps;
+
+		ceph_iterate_session_caps(session, trim_caps_cb, &remaining);
 		dout("trim_caps mds%d done: %d / %d, trimmed %d\n",
 		     session->s_mds, session->s_nr_caps, max_caps,
-			trim_caps - session->s_trim_caps);
-		session->s_trim_caps = 0;
+			trim_caps - remaining);
 	}
 
 	ceph_flush_cap_releases(mdsc, session);
@@ -3015,18 +3031,23 @@ static void handle_forward(struct ceph_mds_client *mdsc,
 	pr_err("mdsc_handle_forward decode error err=%d\n", err);
 }
 
-static int __decode_and_drop_session_metadata(void **p, void *end)
+static int __decode_session_metadata(void **p, void *end,
+				     bool *blacklisted)
 {
 	/* map<string,string> */
 	u32 n;
+	bool err_str;
 	ceph_decode_32_safe(p, end, n, bad);
 	while (n-- > 0) {
 		u32 len;
 		ceph_decode_32_safe(p, end, len, bad);
 		ceph_decode_need(p, end, len, bad);
+		err_str = !strncmp(*p, "error_string", len);
 		*p += len;
 		ceph_decode_32_safe(p, end, len, bad);
 		ceph_decode_need(p, end, len, bad);
+		if (err_str && strnstr(*p, "blacklisted", len))
+			*blacklisted = true;
 		*p += len;
 	}
 	return 0;
@@ -3050,6 +3071,7 @@ static void handle_session(struct ceph_mds_session *session,
 	u64 seq;
 	unsigned long features = 0;
 	int wake = 0;
+	bool blacklisted = false;
 
 	/* decode */
 	ceph_decode_need(&p, end, sizeof(*h), bad);
@@ -3062,7 +3084,7 @@ static void handle_session(struct ceph_mds_session *session,
 	if (msg_version >= 3) {
 		u32 len;
 		/* version >= 2, metadata */
-		if (__decode_and_drop_session_metadata(&p, end) < 0)
+		if (__decode_session_metadata(&p, end, &blacklisted) < 0)
 			goto bad;
 		/* version >= 3, feature bits */
 		ceph_decode_32_safe(&p, end, len, bad);
@@ -3149,6 +3171,8 @@ static void handle_session(struct ceph_mds_session *session,
 		session->s_state = CEPH_MDS_SESSION_REJECTED;
 		cleanup_session_requests(mdsc, session);
 		remove_session_caps(session);
+		if (blacklisted)
+			mdsc->fsc->blacklisted = true;
 		wake = 2; /* for good measure */
 		break;
 
@@ -3998,7 +4022,27 @@ static void lock_unlock_sessions(struct ceph_mds_client *mdsc)
 	mutex_unlock(&mdsc->mutex);
 }
 
+static void maybe_recover_session(struct ceph_mds_client *mdsc)
+{
+	struct ceph_fs_client *fsc = mdsc->fsc;
+
+	if (!ceph_test_mount_opt(fsc, CLEANRECOVER))
+		return;
+
+	if (READ_ONCE(fsc->mount_state) != CEPH_MOUNT_MOUNTED)
+		return;
+
+	if (!READ_ONCE(fsc->blacklisted))
+		return;
+
+	if (fsc->last_auto_reconnect &&
+	    time_before(jiffies, fsc->last_auto_reconnect + HZ * 60 * 30))
+		return;
 
+	pr_info("auto reconnect after blacklisted\n");
+	fsc->last_auto_reconnect = jiffies;
+	ceph_force_reconnect(fsc->sb);
+}
 
 /*
  * delayed work -- periodically trim expired leases, renew caps with mds
@@ -4044,7 +4088,9 @@ static void delayed_work(struct work_struct *work)
 				pr_info("mds%d hung\n", s->s_mds);
 			}
 		}
-		if (s->s_state < CEPH_MDS_SESSION_OPEN) {
+		if (s->s_state == CEPH_MDS_SESSION_NEW ||
+		    s->s_state == CEPH_MDS_SESSION_RESTARTING ||
+		    s->s_state == CEPH_MDS_SESSION_REJECTED) {
 			/* this mds is failed or recovering, just wait */
 			ceph_put_mds_session(s);
 			continue;
@@ -4072,6 +4118,8 @@ static void delayed_work(struct work_struct *work)
 
 	ceph_trim_snapid_map(mdsc);
 
+	maybe_recover_session(mdsc);
+
 	schedule_delayed(mdsc);
 }
 
@@ -4355,7 +4403,12 @@ void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
 		session = __ceph_lookup_mds_session(mdsc, mds);
 		if (!session)
 			continue;
+
+		if (session->s_state == CEPH_MDS_SESSION_REJECTED)
+			__unregister_session(mdsc, session);
+		__wake_requests(mdsc, &session->s_waiting);
 		mutex_unlock(&mdsc->mutex);
+
 		mutex_lock(&session->s_mutex);
 		__close_session(mdsc, session);
 		if (session->s_state == CEPH_MDS_SESSION_CLOSING) {
@@ -4364,6 +4417,7 @@ void ceph_mdsc_force_umount(struct ceph_mds_client *mdsc)
 		}
 		mutex_unlock(&session->s_mutex);
 		ceph_put_mds_session(session);
+
 		mutex_lock(&mdsc->mutex);
 		kick_requests(mdsc, mds);
 	}

fs/ceph/mds_client.h

@@ -148,9 +148,9 @@ enum {
 	CEPH_MDS_SESSION_OPENING = 2,
 	CEPH_MDS_SESSION_OPEN = 3,
 	CEPH_MDS_SESSION_HUNG = 4,
-	CEPH_MDS_SESSION_CLOSING = 5,
-	CEPH_MDS_SESSION_RESTARTING = 6,
-	CEPH_MDS_SESSION_RECONNECTING = 7,
+	CEPH_MDS_SESSION_RESTARTING = 5,
+	CEPH_MDS_SESSION_RECONNECTING = 6,
+	CEPH_MDS_SESSION_CLOSING = 7,
 	CEPH_MDS_SESSION_REJECTED = 8,
 };
 
@@ -176,7 +176,7 @@ struct ceph_mds_session {
 	spinlock_t        s_cap_lock;
 	struct list_head  s_caps;     /* all caps issued by this session */
 	struct ceph_cap  *s_cap_iterator;
-	int               s_nr_caps, s_trim_caps;
+	int               s_nr_caps;
 	int               s_num_cap_releases;
 	int		  s_cap_reconnect;
 	int		  s_readonly;

fs/ceph/super.c

@@ -143,6 +143,7 @@ enum {
 	Opt_snapdirname,
 	Opt_mds_namespace,
 	Opt_fscache_uniq,
+	Opt_recover_session,
 	Opt_last_string,
 	/* string args above */
 	Opt_dirstat,
@@ -184,6 +185,7 @@ static match_table_t fsopt_tokens = {
 	/* int args above */
 	{Opt_snapdirname, "snapdirname=%s"},
 	{Opt_mds_namespace, "mds_namespace=%s"},
+	{Opt_recover_session, "recover_session=%s"},
 	{Opt_fscache_uniq, "fsc=%s"},
 	/* string args above */
 	{Opt_dirstat, "dirstat"},
@@ -254,6 +256,17 @@ static int parse_fsopt_token(char *c, void *private)
 		if (!fsopt->mds_namespace)
 			return -ENOMEM;
 		break;
+	case Opt_recover_session:
+		if (!strncmp(argstr[0].from, "no",
+			     argstr[0].to - argstr[0].from)) {
+			fsopt->flags &= ~CEPH_MOUNT_OPT_CLEANRECOVER;
+		} else if (!strncmp(argstr[0].from, "clean",
+				    argstr[0].to - argstr[0].from)) {
+			fsopt->flags |= CEPH_MOUNT_OPT_CLEANRECOVER;
+		} else {
+			return -EINVAL;
+		}
+		break;
 	case Opt_fscache_uniq:
 		kfree(fsopt->fscache_uniq);
 		fsopt->fscache_uniq = kstrndup(argstr[0].from,
@@ -576,6 +589,10 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
 
 	if (fsopt->mds_namespace)
 		seq_show_option(m, "mds_namespace", fsopt->mds_namespace);
+
+	if (fsopt->flags & CEPH_MOUNT_OPT_CLEANRECOVER)
+		seq_show_option(m, "recover_session", "clean");
+
 	if (fsopt->wsize != CEPH_MAX_WRITE_SIZE)
 		seq_printf(m, ",wsize=%d", fsopt->wsize);
 	if (fsopt->rsize != CEPH_MAX_READ_SIZE)
@@ -664,6 +681,7 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
 
 	fsc->sb = NULL;
 	fsc->mount_state = CEPH_MOUNT_MOUNTING;
+	fsc->filp_gen = 1;
 
 	atomic_long_set(&fsc->writeback_count, 0);
 
@@ -713,6 +731,7 @@ static void destroy_fs_client(struct ceph_fs_client *fsc)
 {
 	dout("destroy_fs_client %p\n", fsc);
 
+	ceph_mdsc_destroy(fsc);
 	destroy_workqueue(fsc->inode_wq);
 	destroy_workqueue(fsc->cap_wq);
 
@@ -829,7 +848,7 @@ static void ceph_umount_begin(struct super_block *sb)
 	fsc->mount_state = CEPH_MOUNT_SHUTDOWN;
 	ceph_osdc_abort_requests(&fsc->client->osdc, -EIO);
 	ceph_mdsc_force_umount(fsc->mdsc);
-	return;
+	fsc->filp_gen++; // invalidate open files
 }
 
 static int ceph_remount(struct super_block *sb, int *flags, char *data)
@@ -1089,7 +1108,6 @@ static struct dentry *ceph_mount(struct file_system_type *fs_type,
 	}
 
 	if (ceph_sb_to_client(sb) != fsc) {
-		ceph_mdsc_destroy(fsc);
 		destroy_fs_client(fsc);
 		fsc = ceph_sb_to_client(sb);
 		dout("get_sb got existing client %p\n", fsc);
@@ -1115,7 +1133,6 @@ static struct dentry *ceph_mount(struct file_system_type *fs_type,
 	goto out_final;
 
 out:
-	ceph_mdsc_destroy(fsc);
 	destroy_fs_client(fsc);
 out_final:
 	dout("ceph_mount fail %ld\n", PTR_ERR(res));
@@ -1139,8 +1156,6 @@ static void ceph_kill_sb(struct super_block *s)
 
 	ceph_fscache_unregister_fs(fsc);
 
-	ceph_mdsc_destroy(fsc);
-
 	destroy_fs_client(fsc);
 	free_anon_bdev(dev);
 }
@@ -1154,6 +1169,33 @@ static struct file_system_type ceph_fs_type = {
 };
 MODULE_ALIAS_FS("ceph");
 
+int ceph_force_reconnect(struct super_block *sb)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
+	int err = 0;
+
+	ceph_umount_begin(sb);
+
+	/* Make sure all page caches get invalidated.
+	 * see remove_session_caps_cb() */
+	flush_workqueue(fsc->inode_wq);
+
+	/* In case that we were blacklisted. This also reset
+	 * all mon/osd connections */
+	ceph_reset_client_addr(fsc->client);
+
+	ceph_osdc_clear_abort_err(&fsc->client->osdc);
+
+	fsc->blacklisted = false;
+	fsc->mount_state = CEPH_MOUNT_MOUNTED;
+
+	if (sb->s_root) {
+		err = __ceph_do_getattr(d_inode(sb->s_root), NULL,
+					CEPH_STAT_CAP_INODE, true);
+	}
+	return err;
+}
+
 static int __init init_ceph(void)
 {
 	int ret = init_caches();

fs/ceph/super.h

@@ -16,6 +16,7 @@
 #include <linux/slab.h>
 #include <linux/posix_acl.h>
 #include <linux/refcount.h>
+#include <linux/security.h>
 
 #include <linux/ceph/libceph.h>
 
@@ -31,6 +32,7 @@
 #define CEPH_BLOCK_SHIFT   22  /* 4 MB */
 #define CEPH_BLOCK         (1 << CEPH_BLOCK_SHIFT)
 
+#define CEPH_MOUNT_OPT_CLEANRECOVER    (1<<1) /* auto reonnect (clean mode) after blacklisted */
 #define CEPH_MOUNT_OPT_DIRSTAT         (1<<4) /* `cat dirname` for stats */
 #define CEPH_MOUNT_OPT_RBYTES          (1<<5) /* dir st_bytes = rbytes */
 #define CEPH_MOUNT_OPT_NOASYNCREADDIR  (1<<7) /* no dcache readdir */
@@ -101,6 +103,11 @@ struct ceph_fs_client {
 	struct ceph_client *client;
 
 	unsigned long mount_state;
+
+	unsigned long last_auto_reconnect;
+	bool blacklisted;
+
+	u32 filp_gen;
 	loff_t max_file_size;
 
 	struct ceph_mds_client *mdsc;
@@ -395,6 +402,8 @@ struct ceph_inode_info {
 	struct fscache_cookie *fscache;
 	u32 i_fscache_gen;
 #endif
+	errseq_t i_meta_err;
+
 	struct inode vfs_inode; /* at end */
 };
 
@@ -499,17 +508,16 @@ static inline struct inode *ceph_find_inode(struct super_block *sb,
 #define CEPH_I_DIR_ORDERED	(1 << 0)  /* dentries in dir are ordered */
 #define CEPH_I_NODELAY		(1 << 1)  /* do not delay cap release */
 #define CEPH_I_FLUSH		(1 << 2)  /* do not delay flush of dirty metadata */
-#define CEPH_I_NOFLUSH		(1 << 3)  /* do not flush dirty caps */
-#define CEPH_I_POOL_PERM	(1 << 4)  /* pool rd/wr bits are valid */
-#define CEPH_I_POOL_RD		(1 << 5)  /* can read from pool */
-#define CEPH_I_POOL_WR		(1 << 6)  /* can write to pool */
-#define CEPH_I_SEC_INITED	(1 << 7)  /* security initialized */
-#define CEPH_I_CAP_DROPPED	(1 << 8)  /* caps were forcibly dropped */
-#define CEPH_I_KICK_FLUSH	(1 << 9)  /* kick flushing caps */
-#define CEPH_I_FLUSH_SNAPS	(1 << 10) /* need flush snapss */
-#define CEPH_I_ERROR_WRITE	(1 << 11) /* have seen write errors */
-#define CEPH_I_ERROR_FILELOCK	(1 << 12) /* have seen file lock errors */
-
+#define CEPH_I_POOL_PERM	(1 << 3)  /* pool rd/wr bits are valid */
+#define CEPH_I_POOL_RD		(1 << 4)  /* can read from pool */
+#define CEPH_I_POOL_WR		(1 << 5)  /* can write to pool */
+#define CEPH_I_SEC_INITED	(1 << 6)  /* security initialized */
+#define CEPH_I_CAP_DROPPED	(1 << 7)  /* caps were forcibly dropped */
+#define CEPH_I_KICK_FLUSH	(1 << 8)  /* kick flushing caps */
+#define CEPH_I_FLUSH_SNAPS	(1 << 9)  /* need flush snapss */
+#define CEPH_I_ERROR_WRITE	(1 << 10) /* have seen write errors */
+#define CEPH_I_ERROR_FILELOCK	(1 << 11) /* have seen file lock errors */
+#define CEPH_I_ODIRECT		(1 << 12) /* inode in direct I/O mode */
 
 /*
  * Masks of ceph inode work.
@@ -703,6 +711,10 @@ struct ceph_file_info {
 
 	spinlock_t rw_contexts_lock;
 	struct list_head rw_contexts;
+
+	errseq_t meta_err;
+	u32 filp_gen;
+	atomic_t num_locks;
 };
 
 struct ceph_dir_file_info {
@@ -842,7 +854,8 @@ static inline int default_congestion_kb(void)
 }
 
 
-
+/* super.c */
+extern int ceph_force_reconnect(struct super_block *sb);
 /* snap.c */
 struct ceph_snap_realm *ceph_lookup_snap_realm(struct ceph_mds_client *mdsc,
 					       u64 ino);
@@ -959,7 +972,10 @@ static inline bool ceph_security_xattr_wanted(struct inode *in)
 #ifdef CONFIG_CEPH_FS_SECURITY_LABEL
 extern int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
 				     struct ceph_acl_sec_ctx *ctx);
-extern void ceph_security_invalidate_secctx(struct inode *inode);
+static inline void ceph_security_invalidate_secctx(struct inode *inode)
+{
+	security_inode_invalidate_secctx(inode);
+}
 #else
 static inline int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
 					    struct ceph_acl_sec_ctx *ctx)
@@ -1039,7 +1055,6 @@ extern void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 				    struct ceph_mds_session *session);
 extern struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci,
 					     int mds);
-extern int ceph_get_cap_mds(struct inode *inode);
 extern void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps);
 extern void ceph_put_cap_refs(struct ceph_inode_info *ci, int had);
 extern void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
@@ -1058,9 +1073,9 @@ extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
 				      struct inode *dir,
 				      int mds, int drop, int unless);
 
-extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
+extern int ceph_get_caps(struct file *filp, int need, int want,
 			 loff_t endoff, int *got, struct page **pinned_page);
-extern int ceph_try_get_caps(struct ceph_inode_info *ci,
+extern int ceph_try_get_caps(struct inode *inode,
 			     int need, int want, bool nonblock, int *got);
 
 /* for counting open files by mode */
@@ -1071,7 +1086,7 @@ extern void ceph_put_fmode(struct ceph_inode_info *ci, int mode);
 extern const struct address_space_operations ceph_aops;
 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
 extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
-extern int ceph_pool_perm_check(struct ceph_inode_info *ci, int need);
+extern int ceph_pool_perm_check(struct inode *inode, int need);
 extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
 
 /* file.c */

fs/ceph/xattr.c

@@ -20,7 +20,8 @@ static int __remove_xattr(struct ceph_inode_info *ci,
 
 static bool ceph_is_valid_xattr(const char *name)
 {
-	return !strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN) ||
+	return !strncmp(name, XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN) ||
+	       !strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN) ||
 	       !strncmp(name, XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN) ||
 	       !strncmp(name, XATTR_USER_PREFIX, XATTR_USER_PREFIX_LEN);
 }
@@ -892,7 +893,8 @@ ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
 	memcpy(value, xattr->val, xattr->val_len);
 
 	if (current->journal_info &&
-	    !strncmp(name, XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN))
+	    !strncmp(name, XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN) &&
+	    security_ismaclabel(name + XATTR_SECURITY_PREFIX_LEN))
 		ci->i_ceph_flags |= CEPH_I_SEC_INITED;
 out:
 	spin_unlock(&ci->i_ceph_lock);
@@ -903,11 +905,9 @@ ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
 {
 	struct inode *inode = d_inode(dentry);
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct ceph_vxattr *vxattrs = ceph_inode_vxattrs(inode);
 	bool len_only = (size == 0);
 	u32 namelen;
 	int err;
-	int i;
 
 	spin_lock(&ci->i_ceph_lock);
 	dout("listxattr %p ver=%lld index_ver=%lld\n", inode,
@@ -936,33 +936,6 @@ ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
 		names = __copy_xattr_names(ci, names);
 		size -= namelen;
 	}
-
-
-	/* virtual xattr names, too */
-	if (vxattrs) {
-		for (i = 0; vxattrs[i].name; i++) {
-			size_t this_len;
-
-			if (vxattrs[i].flags & VXATTR_FLAG_HIDDEN)
-				continue;
-			if (vxattrs[i].exists_cb && !vxattrs[i].exists_cb(ci))
-				continue;
-
-			this_len = strlen(vxattrs[i].name) + 1;
-			namelen += this_len;
-			if (len_only)
-				continue;
-
-			if (this_len > size) {
-				err = -ERANGE;
-				goto out;
-			}
-
-			memcpy(names, vxattrs[i].name, this_len);
-			names += this_len;
-			size -= this_len;
-		}
-	}
 	err = namelen;
 out:
 	spin_unlock(&ci->i_ceph_lock);
@@ -1293,42 +1266,8 @@ int ceph_security_init_secctx(struct dentry *dentry, umode_t mode,
 		ceph_pagelist_release(pagelist);
 	return err;
 }
-
-void ceph_security_invalidate_secctx(struct inode *inode)
-{
-	security_inode_invalidate_secctx(inode);
-}
-
-static int ceph_xattr_set_security_label(const struct xattr_handler *handler,
-				    struct dentry *unused, struct inode *inode,
-				    const char *key, const void *buf,
-				    size_t buflen, int flags)
-{
-	if (security_ismaclabel(key)) {
-		const char *name = xattr_full_name(handler, key);
-		return __ceph_setxattr(inode, name, buf, buflen, flags);
-	}
-	return  -EOPNOTSUPP;
-}
-
-static int ceph_xattr_get_security_label(const struct xattr_handler *handler,
-				    struct dentry *unused, struct inode *inode,
-				    const char *key, void *buf, size_t buflen)
-{
-	if (security_ismaclabel(key)) {
-		const char *name = xattr_full_name(handler, key);
-		return __ceph_getxattr(inode, name, buf, buflen);
-	}
-	return  -EOPNOTSUPP;
-}
-
-static const struct xattr_handler ceph_security_label_handler = {
-	.prefix = XATTR_SECURITY_PREFIX,
-	.get    = ceph_xattr_get_security_label,
-	.set    = ceph_xattr_set_security_label,
-};
-#endif
-#endif
+#endif /* CONFIG_CEPH_FS_SECURITY_LABEL */
+#endif /* CONFIG_SECURITY */
 
 void ceph_release_acl_sec_ctx(struct ceph_acl_sec_ctx *as_ctx)
 {
@@ -1351,9 +1290,6 @@ const struct xattr_handler *ceph_xattr_handlers[] = {
 #ifdef CONFIG_CEPH_FS_POSIX_ACL
 	&posix_acl_access_xattr_handler,
 	&posix_acl_default_xattr_handler,
-#endif
-#ifdef CONFIG_CEPH_FS_SECURITY_LABEL
-	&ceph_security_label_handler,
 #endif
 	&ceph_other_xattr_handler,
 	NULL,

include/linux/ceph/libceph.h

@@ -293,6 +293,7 @@ struct ceph_client *ceph_create_client(struct ceph_options *opt, void *private);
 struct ceph_entity_addr *ceph_client_addr(struct ceph_client *client);
 u64 ceph_client_gid(struct ceph_client *client);
 extern void ceph_destroy_client(struct ceph_client *client);
+extern void ceph_reset_client_addr(struct ceph_client *client);
 extern int __ceph_open_session(struct ceph_client *client,
 			       unsigned long started);
 extern int ceph_open_session(struct ceph_client *client);

include/linux/ceph/messenger.h

@@ -337,6 +337,7 @@ extern void ceph_msgr_flush(void);
 extern void ceph_messenger_init(struct ceph_messenger *msgr,
 				struct ceph_entity_addr *myaddr);
 extern void ceph_messenger_fini(struct ceph_messenger *msgr);
+extern void ceph_messenger_reset_nonce(struct ceph_messenger *msgr);
 
 extern void ceph_con_init(struct ceph_connection *con, void *private,
 			const struct ceph_connection_operations *ops,

include/linux/ceph/mon_client.h

@@ -109,6 +109,7 @@ extern int ceph_monmap_contains(struct ceph_monmap *m,
 
 extern int ceph_monc_init(struct ceph_mon_client *monc, struct ceph_client *cl);
 extern void ceph_monc_stop(struct ceph_mon_client *monc);
+extern void ceph_monc_reopen_session(struct ceph_mon_client *monc);
 
 enum {
 	CEPH_SUB_MONMAP = 0,

include/linux/ceph/osd_client.h

@@ -381,6 +381,7 @@ extern void ceph_osdc_cleanup(void);
 extern int ceph_osdc_init(struct ceph_osd_client *osdc,
 			  struct ceph_client *client);
 extern void ceph_osdc_stop(struct ceph_osd_client *osdc);
+extern void ceph_osdc_reopen_osds(struct ceph_osd_client *osdc);
 
 extern void ceph_osdc_handle_reply(struct ceph_osd_client *osdc,
 				   struct ceph_msg *msg);
@@ -388,6 +389,7 @@ extern void ceph_osdc_handle_map(struct ceph_osd_client *osdc,
 				 struct ceph_msg *msg);
 void ceph_osdc_update_epoch_barrier(struct ceph_osd_client *osdc, u32 eb);
 void ceph_osdc_abort_requests(struct ceph_osd_client *osdc, int err);
+void ceph_osdc_clear_abort_err(struct ceph_osd_client *osdc);
 
 #define osd_req_op_data(oreq, whch, typ, fld)				\
 ({									\

net/ceph/ceph_common.c

@@ -13,6 +13,7 @@
 #include <linux/nsproxy.h>
 #include <linux/parser.h>
 #include <linux/sched.h>
+#include <linux/sched/mm.h>
 #include <linux/seq_file.h>
 #include <linux/slab.h>
 #include <linux/statfs.h>
@@ -185,18 +186,34 @@ int ceph_compare_options(struct ceph_options *new_opt,
 }
 EXPORT_SYMBOL(ceph_compare_options);
 
+/*
+ * kvmalloc() doesn't fall back to the vmalloc allocator unless flags are
+ * compatible with (a superset of) GFP_KERNEL.  This is because while the
+ * actual pages are allocated with the specified flags, the page table pages
+ * are always allocated with GFP_KERNEL.  map_vm_area() doesn't even take
+ * flags because GFP_KERNEL is hard-coded in {p4d,pud,pmd,pte}_alloc().
+ *
+ * ceph_kvmalloc() may be called with GFP_KERNEL, GFP_NOFS or GFP_NOIO.
+ */
 void *ceph_kvmalloc(size_t size, gfp_t flags)
 {
-	if (size <= (PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER)) {
-		void *ptr = kmalloc(size, flags | __GFP_NOWARN);
-		if (ptr)
-			return ptr;
+	void *p;
+
+	if ((flags & (__GFP_IO | __GFP_FS)) == (__GFP_IO | __GFP_FS)) {
+		p = kvmalloc(size, flags);
+	} else if ((flags & (__GFP_IO | __GFP_FS)) == __GFP_IO) {
+		unsigned int nofs_flag = memalloc_nofs_save();
+		p = kvmalloc(size, GFP_KERNEL);
+		memalloc_nofs_restore(nofs_flag);
+	} else {
+		unsigned int noio_flag = memalloc_noio_save();
+		p = kvmalloc(size, GFP_KERNEL);
+		memalloc_noio_restore(noio_flag);
 	}
 
-	return __vmalloc(size, flags, PAGE_KERNEL);
+	return p;
 }
 
-
 static int parse_fsid(const char *str, struct ceph_fsid *fsid)
 {
 	int i = 0;
@@ -694,6 +711,14 @@ void ceph_destroy_client(struct ceph_client *client)
 }
 EXPORT_SYMBOL(ceph_destroy_client);
 
+void ceph_reset_client_addr(struct ceph_client *client)
+{
+	ceph_messenger_reset_nonce(&client->msgr);
+	ceph_monc_reopen_session(&client->monc);
+	ceph_osdc_reopen_osds(&client->osdc);
+}
+EXPORT_SYMBOL(ceph_reset_client_addr);
+
 /*
  * true if we have the mon map (and have thus joined the cluster)
  */

net/ceph/messenger.c

@@ -3031,6 +3031,12 @@ static void con_fault(struct ceph_connection *con)
 }
 
 
+void ceph_messenger_reset_nonce(struct ceph_messenger *msgr)
+{
+	u32 nonce = le32_to_cpu(msgr->inst.addr.nonce) + 1000000;
+	msgr->inst.addr.nonce = cpu_to_le32(nonce);
+	encode_my_addr(msgr);
+}
 
 /*
  * initialize a new messenger instance

net/ceph/mon_client.c

@@ -213,6 +213,13 @@ static void reopen_session(struct ceph_mon_client *monc)
 	__open_session(monc);
 }
 
+void ceph_monc_reopen_session(struct ceph_mon_client *monc)
+{
+	mutex_lock(&monc->mutex);
+	reopen_session(monc);
+	mutex_unlock(&monc->mutex);
+}
+
 static void un_backoff(struct ceph_mon_client *monc)
 {
 	monc->hunt_mult /= 2; /* reduce by 50% */

net/ceph/osd_client.c

@@ -841,6 +841,7 @@ int osd_req_op_cls_init(struct ceph_osd_request *osd_req, unsigned int which,
 	struct ceph_pagelist *pagelist;
 	size_t payload_len = 0;
 	size_t size;
+	int ret;
 
 	op = _osd_req_op_init(osd_req, which, CEPH_OSD_OP_CALL, 0);
 
@@ -852,20 +853,27 @@ int osd_req_op_cls_init(struct ceph_osd_request *osd_req, unsigned int which,
 	size = strlen(class);
 	BUG_ON(size > (size_t) U8_MAX);
 	op->cls.class_len = size;
-	ceph_pagelist_append(pagelist, class, size);
+	ret = ceph_pagelist_append(pagelist, class, size);
+	if (ret)
+		goto err_pagelist_free;
 	payload_len += size;
 
 	op->cls.method_name = method;
 	size = strlen(method);
 	BUG_ON(size > (size_t) U8_MAX);
 	op->cls.method_len = size;
-	ceph_pagelist_append(pagelist, method, size);
+	ret = ceph_pagelist_append(pagelist, method, size);
+	if (ret)
+		goto err_pagelist_free;
 	payload_len += size;
 
 	osd_req_op_cls_request_info_pagelist(osd_req, which, pagelist);
-
 	op->indata_len = payload_len;
 	return 0;
+
+err_pagelist_free:
+	ceph_pagelist_release(pagelist);
+	return ret;
 }
 EXPORT_SYMBOL(osd_req_op_cls_init);
 
@@ -877,6 +885,7 @@ int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which,
 						      opcode, 0);
 	struct ceph_pagelist *pagelist;
 	size_t payload_len;
+	int ret;
 
 	BUG_ON(opcode != CEPH_OSD_OP_SETXATTR && opcode != CEPH_OSD_OP_CMPXATTR);
 
@@ -886,10 +895,14 @@ int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which,
 
 	payload_len = strlen(name);
 	op->xattr.name_len = payload_len;
-	ceph_pagelist_append(pagelist, name, payload_len);
+	ret = ceph_pagelist_append(pagelist, name, payload_len);
+	if (ret)
+		goto err_pagelist_free;
 
 	op->xattr.value_len = size;
-	ceph_pagelist_append(pagelist, value, size);
+	ret = ceph_pagelist_append(pagelist, value, size);
+	if (ret)
+		goto err_pagelist_free;
 	payload_len += size;
 
 	op->xattr.cmp_op = cmp_op;
@@ -898,6 +911,10 @@ int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which,
 	ceph_osd_data_pagelist_init(&op->xattr.osd_data, pagelist);
 	op->indata_len = payload_len;
 	return 0;
+
+err_pagelist_free:
+	ceph_pagelist_release(pagelist);
+	return ret;
 }
 EXPORT_SYMBOL(osd_req_op_xattr_init);
 
@@ -1488,7 +1505,6 @@ enum calc_target_result {
 
 static enum calc_target_result calc_target(struct ceph_osd_client *osdc,
 					   struct ceph_osd_request_target *t,
-					   struct ceph_connection *con,
 					   bool any_change)
 {
 	struct ceph_pg_pool_info *pi;
@@ -2272,7 +2288,7 @@ static void __submit_request(struct ceph_osd_request *req, bool wrlocked)
 	dout("%s req %p wrlocked %d\n", __func__, req, wrlocked);
 
 again:
-	ct_res = calc_target(osdc, &req->r_t, NULL, false);
+	ct_res = calc_target(osdc, &req->r_t, false);
 	if (ct_res == CALC_TARGET_POOL_DNE && !wrlocked)
 		goto promote;
 
@@ -2476,6 +2492,14 @@ void ceph_osdc_abort_requests(struct ceph_osd_client *osdc, int err)
 }
 EXPORT_SYMBOL(ceph_osdc_abort_requests);
 
+void ceph_osdc_clear_abort_err(struct ceph_osd_client *osdc)
+{
+	down_write(&osdc->lock);
+	osdc->abort_err = 0;
+	up_write(&osdc->lock);
+}
+EXPORT_SYMBOL(ceph_osdc_clear_abort_err);
+
 static void update_epoch_barrier(struct ceph_osd_client *osdc, u32 eb)
 {
 	if (likely(eb > osdc->epoch_barrier)) {
@@ -3087,7 +3111,7 @@ static void linger_submit(struct ceph_osd_linger_request *lreq)
 		lreq->reg_req->r_ops[0].notify.cookie = lreq->linger_id;
 	}
 
-	calc_target(osdc, &lreq->t, NULL, false);
+	calc_target(osdc, &lreq->t, false);
 	osd = lookup_create_osd(osdc, lreq->t.osd, true);
 	link_linger(osd, lreq);
 
@@ -3704,7 +3728,7 @@ recalc_linger_target(struct ceph_osd_linger_request *lreq)
 	struct ceph_osd_client *osdc = lreq->osdc;
 	enum calc_target_result ct_res;
 
-	ct_res = calc_target(osdc, &lreq->t, NULL, true);
+	ct_res = calc_target(osdc, &lreq->t, true);
 	if (ct_res == CALC_TARGET_NEED_RESEND) {
 		struct ceph_osd *osd;
 
@@ -3776,8 +3800,7 @@ static void scan_requests(struct ceph_osd *osd,
 		n = rb_next(n); /* unlink_request(), check_pool_dne() */
 
 		dout("%s req %p tid %llu\n", __func__, req, req->r_tid);
-		ct_res = calc_target(osdc, &req->r_t, &req->r_osd->o_con,
-				     false);
+		ct_res = calc_target(osdc, &req->r_t, false);
 		switch (ct_res) {
 		case CALC_TARGET_NO_ACTION:
 			force_resend_writes = cleared_full ||
@@ -3886,7 +3909,7 @@ static void kick_requests(struct ceph_osd_client *osdc,
 		n = rb_next(n);
 
 		if (req->r_t.epoch < osdc->osdmap->epoch) {
-			ct_res = calc_target(osdc, &req->r_t, NULL, false);
+			ct_res = calc_target(osdc, &req->r_t, false);
 			if (ct_res == CALC_TARGET_POOL_DNE) {
 				erase_request(need_resend, req);
 				check_pool_dne(req);
@@ -5086,6 +5109,24 @@ int ceph_osdc_call(struct ceph_osd_client *osdc,
 }
 EXPORT_SYMBOL(ceph_osdc_call);
 
+/*
+ * reset all osd connections
+ */
+void ceph_osdc_reopen_osds(struct ceph_osd_client *osdc)
+{
+	struct rb_node *n;
+
+	down_write(&osdc->lock);
+	for (n = rb_first(&osdc->osds); n; ) {
+		struct ceph_osd *osd = rb_entry(n, struct ceph_osd, o_node);
+
+		n = rb_next(n);
+		if (!reopen_osd(osd))
+			kick_osd_requests(osd);
+	}
+	up_write(&osdc->lock);
+}
+
 /*
  * init, shutdown
  */

net/ceph/osdmap.c

@@ -973,11 +973,11 @@ void ceph_osdmap_destroy(struct ceph_osdmap *map)
 				 struct ceph_pg_pool_info, node);
 		__remove_pg_pool(&map->pg_pools, pi);
 	}
-	kfree(map->osd_state);
-	kfree(map->osd_weight);
-	kfree(map->osd_addr);
-	kfree(map->osd_primary_affinity);
-	kfree(map->crush_workspace);
+	kvfree(map->osd_state);
+	kvfree(map->osd_weight);
+	kvfree(map->osd_addr);
+	kvfree(map->osd_primary_affinity);
+	kvfree(map->crush_workspace);
 	kfree(map);
 }
 
@@ -986,28 +986,41 @@ void ceph_osdmap_destroy(struct ceph_osdmap *map)
  *
  * The new elements are properly initialized.
  */
-static int osdmap_set_max_osd(struct ceph_osdmap *map, int max)
+static int osdmap_set_max_osd(struct ceph_osdmap *map, u32 max)
 {
 	u32 *state;
 	u32 *weight;
 	struct ceph_entity_addr *addr;
+	u32 to_copy;
 	int i;
 
-	state = krealloc(map->osd_state, max*sizeof(*state), GFP_NOFS);
-	if (!state)
-		return -ENOMEM;
-	map->osd_state = state;
+	dout("%s old %u new %u\n", __func__, map->max_osd, max);
+	if (max == map->max_osd)
+		return 0;
 
-	weight = krealloc(map->osd_weight, max*sizeof(*weight), GFP_NOFS);
-	if (!weight)
+	state = ceph_kvmalloc(array_size(max, sizeof(*state)), GFP_NOFS);
+	weight = ceph_kvmalloc(array_size(max, sizeof(*weight)), GFP_NOFS);
+	addr = ceph_kvmalloc(array_size(max, sizeof(*addr)), GFP_NOFS);
+	if (!state || !weight || !addr) {
+		kvfree(state);
+		kvfree(weight);
+		kvfree(addr);
 		return -ENOMEM;
-	map->osd_weight = weight;
+	}
 
-	addr = krealloc(map->osd_addr, max*sizeof(*addr), GFP_NOFS);
-	if (!addr)
-		return -ENOMEM;
-	map->osd_addr = addr;
+	to_copy = min(map->max_osd, max);
+	if (map->osd_state) {
+		memcpy(state, map->osd_state, to_copy * sizeof(*state));
+		memcpy(weight, map->osd_weight, to_copy * sizeof(*weight));
+		memcpy(addr, map->osd_addr, to_copy * sizeof(*addr));
+		kvfree(map->osd_state);
+		kvfree(map->osd_weight);
+		kvfree(map->osd_addr);
+	}
 
+	map->osd_state = state;
+	map->osd_weight = weight;
+	map->osd_addr = addr;
 	for (i = map->max_osd; i < max; i++) {
 		map->osd_state[i] = 0;
 		map->osd_weight[i] = CEPH_OSD_OUT;
@@ -1017,12 +1030,16 @@ static int osdmap_set_max_osd(struct ceph_osdmap *map, int max)
 	if (map->osd_primary_affinity) {
 		u32 *affinity;
 
-		affinity = krealloc(map->osd_primary_affinity,
-				    max*sizeof(*affinity), GFP_NOFS);
+		affinity = ceph_kvmalloc(array_size(max, sizeof(*affinity)),
+					 GFP_NOFS);
 		if (!affinity)
 			return -ENOMEM;
-		map->osd_primary_affinity = affinity;
 
+		memcpy(affinity, map->osd_primary_affinity,
+		       to_copy * sizeof(*affinity));
+		kvfree(map->osd_primary_affinity);
+
+		map->osd_primary_affinity = affinity;
 		for (i = map->max_osd; i < max; i++)
 			map->osd_primary_affinity[i] =
 			    CEPH_OSD_DEFAULT_PRIMARY_AFFINITY;
@@ -1043,7 +1060,7 @@ static int osdmap_set_crush(struct ceph_osdmap *map, struct crush_map *crush)
 
 	work_size = crush_work_size(crush, CEPH_PG_MAX_SIZE);
 	dout("%s work_size %zu bytes\n", __func__, work_size);
-	workspace = kmalloc(work_size, GFP_NOIO);
+	workspace = ceph_kvmalloc(work_size, GFP_NOIO);
 	if (!workspace) {
 		crush_destroy(crush);
 		return -ENOMEM;
@@ -1052,7 +1069,7 @@ static int osdmap_set_crush(struct ceph_osdmap *map, struct crush_map *crush)
 
 	if (map->crush)
 		crush_destroy(map->crush);
-	kfree(map->crush_workspace);
+	kvfree(map->crush_workspace);
 	map->crush = crush;
 	map->crush_workspace = workspace;
 	return 0;
@@ -1298,9 +1315,9 @@ static int set_primary_affinity(struct ceph_osdmap *map, int osd, u32 aff)
 	if (!map->osd_primary_affinity) {
 		int i;
 
-		map->osd_primary_affinity = kmalloc_array(map->max_osd,
-							  sizeof(u32),
-							  GFP_NOFS);
+		map->osd_primary_affinity = ceph_kvmalloc(
+		    array_size(map->max_osd, sizeof(*map->osd_primary_affinity)),
+		    GFP_NOFS);
 		if (!map->osd_primary_affinity)
 			return -ENOMEM;
 
@@ -1321,7 +1338,7 @@ static int decode_primary_affinity(void **p, void *end,
 
 	ceph_decode_32_safe(p, end, len, e_inval);
 	if (len == 0) {
-		kfree(map->osd_primary_affinity);
+		kvfree(map->osd_primary_affinity);
 		map->osd_primary_affinity = NULL;
 		return 0;
 	}