Merge tag 'nfs-for-6.6-2' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client fixes from Anna Schumaker:
 "Various O_DIRECT related fixes from Trond:
   - Error handling
   - Locking issues
   - Use the correct commit info for joining page groups
   - Fixes for rescheduling IO

  Sunrpc bad verifier fixes:
   - Report EINVAL errors from connect()
   - Revalidate creds that the server has rejected
   - Revert "SUNRPC: Fail faster on bad verifier"

  Misc:
   - Fix pNFS session trunking when MDS=DS
   - Fix zero-value filehandles for post-open getattr operations
   - Fix compiler warning about tautological comparisons
   - Revert 'SUNRPC: clean up integer overflow check' before Trond's fix"

* tag 'nfs-for-6.6-2' of git://git.linux-nfs.org/projects/anna/linux-nfs:
  SUNRPC: Silence compiler complaints about tautological comparisons
  Revert "SUNRPC: clean up integer overflow check"
  NFSv4.1: fix zero value filehandle in post open getattr
  NFSv4.1: fix pnfs MDS=DS session trunking
  Revert "SUNRPC: Fail faster on bad verifier"
  SUNRPC: Mark the cred for revalidation if the server rejects it
  NFS/pNFS: Report EINVAL errors from connect() to the server
  NFS: More fixes for nfs_direct_write_reschedule_io()
  NFS: Use the correct commit info in nfs_join_page_group()
  NFS: More O_DIRECT accounting fixes for error paths
  NFS: Fix O_DIRECT locking issues
  NFS: Fix error handling for O_DIRECT write scheduling

9 files changed, 132 insertions, 61 deletions

diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 47d892a1d363d9..f6c74f4246917f 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -93,12 +93,10 @@ nfs_direct_handle_truncated(struct nfs_direct_req *dreq,
 		dreq->max_count = dreq_len;
 		if (dreq->count > dreq_len)
 			dreq->count = dreq_len;
-
-		if (test_bit(NFS_IOHDR_ERROR, &hdr->flags))
-			dreq->error = hdr->error;
-		else /* Clear outstanding error if this is EOF */
-			dreq->error = 0;
 	}
+
+	if (test_bit(NFS_IOHDR_ERROR, &hdr->flags) && !dreq->error)
+		dreq->error = hdr->error;
 }
 
 static void
@@ -120,6 +118,18 @@ nfs_direct_count_bytes(struct nfs_direct_req *dreq,
 		dreq->count = dreq_len;
 }
 
+static void nfs_direct_truncate_request(struct nfs_direct_req *dreq,
+					struct nfs_page *req)
+{
+	loff_t offs = req_offset(req);
+	size_t req_start = (size_t)(offs - dreq->io_start);
+
+	if (req_start < dreq->max_count)
+		dreq->max_count = req_start;
+	if (req_start < dreq->count)
+		dreq->count = req_start;
+}
+
 /**
  * nfs_swap_rw - NFS address space operation for swap I/O
  * @iocb: target I/O control block
@@ -488,7 +498,9 @@ static void nfs_direct_add_page_head(struct list_head *list,
 	kref_get(&head->wb_kref);
 }
 
-static void nfs_direct_join_group(struct list_head *list, struct inode *inode)
+static void nfs_direct_join_group(struct list_head *list,
+				  struct nfs_commit_info *cinfo,
+				  struct inode *inode)
 {
 	struct nfs_page *req, *subreq;
 
@@ -510,7 +522,7 @@ static void nfs_direct_join_group(struct list_head *list, struct inode *inode)
 				nfs_release_request(subreq);
 			}
 		} while ((subreq = subreq->wb_this_page) != req);
-		nfs_join_page_group(req, inode);
+		nfs_join_page_group(req, cinfo, inode);
 	}
 }
 
@@ -528,20 +540,15 @@ nfs_direct_write_scan_commit_list(struct inode *inode,
 static void nfs_direct_write_reschedule(struct nfs_direct_req *dreq)
 {
 	struct nfs_pageio_descriptor desc;
-	struct nfs_page *req, *tmp;
+	struct nfs_page *req;
 	LIST_HEAD(reqs);
 	struct nfs_commit_info cinfo;
-	LIST_HEAD(failed);
 
 	nfs_init_cinfo_from_dreq(&cinfo, dreq);
 	nfs_direct_write_scan_commit_list(dreq->inode, &reqs, &cinfo);
 
-	nfs_direct_join_group(&reqs, dreq->inode);
+	nfs_direct_join_group(&reqs, &cinfo, dreq->inode);
 
-	dreq->count = 0;
-	dreq->max_count = 0;
-	list_for_each_entry(req, &reqs, wb_list)
-		dreq->max_count += req->wb_bytes;
 	nfs_clear_pnfs_ds_commit_verifiers(&dreq->ds_cinfo);
 	get_dreq(dreq);
 
@@ -549,27 +556,40 @@ static void nfs_direct_write_reschedule(struct nfs_direct_req *dreq)
 			      &nfs_direct_write_completion_ops);
 	desc.pg_dreq = dreq;
 
-	list_for_each_entry_safe(req, tmp, &reqs, wb_list) {
+	while (!list_empty(&reqs)) {
+		req = nfs_list_entry(reqs.next);
 		/* Bump the transmission count */
 		req->wb_nio++;
 		if (!nfs_pageio_add_request(&desc, req)) {
-			nfs_list_move_request(req, &failed);
-			spin_lock(&cinfo.inode->i_lock);
-			dreq->flags = 0;
-			if (desc.pg_error < 0)
+			spin_lock(&dreq->lock);
+			if (dreq->error < 0) {
+				desc.pg_error = dreq->error;
+			} else if (desc.pg_error != -EAGAIN) {
+				dreq->flags = 0;
+				if (!desc.pg_error)
+					desc.pg_error = -EIO;
 				dreq->error = desc.pg_error;
-			else
-				dreq->error = -EIO;
-			spin_unlock(&cinfo.inode->i_lock);
+			} else
+				dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
+			spin_unlock(&dreq->lock);
+			break;
 		}
 		nfs_release_request(req);
 	}
 	nfs_pageio_complete(&desc);
 
-	while (!list_empty(&failed)) {
-		req = nfs_list_entry(failed.next);
+	while (!list_empty(&reqs)) {
+		req = nfs_list_entry(reqs.next);
 		nfs_list_remove_request(req);
 		nfs_unlock_and_release_request(req);
+		if (desc.pg_error == -EAGAIN) {
+			nfs_mark_request_commit(req, NULL, &cinfo, 0);
+		} else {
+			spin_lock(&dreq->lock);
+			nfs_direct_truncate_request(dreq, req);
+			spin_unlock(&dreq->lock);
+			nfs_release_request(req);
+		}
 	}
 
 	if (put_dreq(dreq))
@@ -589,8 +609,6 @@ static void nfs_direct_commit_complete(struct nfs_commit_data *data)
 	if (status < 0) {
 		/* Errors in commit are fatal */
 		dreq->error = status;
-		dreq->max_count = 0;
-		dreq->count = 0;
 		dreq->flags = NFS_ODIRECT_DONE;
 	} else {
 		status = dreq->error;
@@ -601,7 +619,12 @@ static void nfs_direct_commit_complete(struct nfs_commit_data *data)
 	while (!list_empty(&data->pages)) {
 		req = nfs_list_entry(data->pages.next);
 		nfs_list_remove_request(req);
-		if (status >= 0 && !nfs_write_match_verf(verf, req)) {
+		if (status < 0) {
+			spin_lock(&dreq->lock);
+			nfs_direct_truncate_request(dreq, req);
+			spin_unlock(&dreq->lock);
+			nfs_release_request(req);
+		} else if (!nfs_write_match_verf(verf, req)) {
 			dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
 			/*
 			 * Despite the reboot, the write was successful,
@@ -609,7 +632,7 @@ static void nfs_direct_commit_complete(struct nfs_commit_data *data)
 			 */
 			req->wb_nio = 0;
 			nfs_mark_request_commit(req, NULL, &cinfo, 0);
-		} else /* Error or match */
+		} else
 			nfs_release_request(req);
 		nfs_unlock_and_release_request(req);
 	}
@@ -662,6 +685,7 @@ static void nfs_direct_write_clear_reqs(struct nfs_direct_req *dreq)
 	while (!list_empty(&reqs)) {
 		req = nfs_list_entry(reqs.next);
 		nfs_list_remove_request(req);
+		nfs_direct_truncate_request(dreq, req);
 		nfs_release_request(req);
 		nfs_unlock_and_release_request(req);
 	}
@@ -711,7 +735,8 @@ static void nfs_direct_write_completion(struct nfs_pgio_header *hdr)
 	}
 
 	nfs_direct_count_bytes(dreq, hdr);
-	if (test_bit(NFS_IOHDR_UNSTABLE_WRITES, &hdr->flags)) {
+	if (test_bit(NFS_IOHDR_UNSTABLE_WRITES, &hdr->flags) &&
+	    !test_bit(NFS_IOHDR_ERROR, &hdr->flags)) {
 		if (!dreq->flags)
 			dreq->flags = NFS_ODIRECT_DO_COMMIT;
 		flags = dreq->flags;
@@ -755,18 +780,23 @@ static void nfs_write_sync_pgio_error(struct list_head *head, int error)
 static void nfs_direct_write_reschedule_io(struct nfs_pgio_header *hdr)
 {
 	struct nfs_direct_req *dreq = hdr->dreq;
+	struct nfs_page *req;
+	struct nfs_commit_info cinfo;
 
 	trace_nfs_direct_write_reschedule_io(dreq);
 
+	nfs_init_cinfo_from_dreq(&cinfo, dreq);
 	spin_lock(&dreq->lock);
-	if (dreq->error == 0) {
+	if (dreq->error == 0)
 		dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
-		/* fake unstable write to let common nfs resend pages */
-		hdr->verf.committed = NFS_UNSTABLE;
-		hdr->good_bytes = hdr->args.offset + hdr->args.count -
-			hdr->io_start;
-	}
+	set_bit(NFS_IOHDR_REDO, &hdr->flags);
 	spin_unlock(&dreq->lock);
+	while (!list_empty(&hdr->pages)) {
+		req = nfs_list_entry(hdr->pages.next);
+		nfs_list_remove_request(req);
+		nfs_unlock_request(req);
+		nfs_mark_request_commit(req, NULL, &cinfo, 0);
+	}
 }
 
 static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops = {
@@ -794,9 +824,11 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 {
 	struct nfs_pageio_descriptor desc;
 	struct inode *inode = dreq->inode;
+	struct nfs_commit_info cinfo;
 	ssize_t result = 0;
 	size_t requested_bytes = 0;
 	size_t wsize = max_t(size_t, NFS_SERVER(inode)->wsize, PAGE_SIZE);
+	bool defer = false;
 
 	trace_nfs_direct_write_schedule_iovec(dreq);
 
@@ -837,17 +869,37 @@ static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq,
 				break;
 			}
 
-			nfs_lock_request(req);
-			if (!nfs_pageio_add_request(&desc, req)) {
-				result = desc.pg_error;
-				nfs_unlock_and_release_request(req);
-				break;
-			}
 			pgbase = 0;
 			bytes -= req_len;
 			requested_bytes += req_len;
 			pos += req_len;
 			dreq->bytes_left -= req_len;
+
+			if (defer) {
+				nfs_mark_request_commit(req, NULL, &cinfo, 0);
+				continue;
+			}
+
+			nfs_lock_request(req);
+			if (nfs_pageio_add_request(&desc, req))
+				continue;
+
+			/* Exit on hard errors */
+			if (desc.pg_error < 0 && desc.pg_error != -EAGAIN) {
+				result = desc.pg_error;
+				nfs_unlock_and_release_request(req);
+				break;
+			}
+
+			/* If the error is soft, defer remaining requests */
+			nfs_init_cinfo_from_dreq(&cinfo, dreq);
+			spin_lock(&dreq->lock);
+			dreq->flags = NFS_ODIRECT_RESCHED_WRITES;
+			spin_unlock(&dreq->lock);
+			nfs_unlock_request(req);
+			nfs_mark_request_commit(req, NULL, &cinfo, 0);
+			desc.pg_error = 0;
+			defer = true;
 		}
 		nfs_direct_release_pages(pagevec, npages);
 		kvfree(pagevec);
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 7deb3cd76abe4b..a1dc338649062d 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -1235,6 +1235,7 @@ static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
 		case -EPFNOSUPPORT:
 		case -EPROTONOSUPPORT:
 		case -EOPNOTSUPP:
+		case -EINVAL:
 		case -ECONNREFUSED:
 		case -ECONNRESET:
 		case -EHOSTDOWN:
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 27fb25567ce755..11e3a285594c23 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -417,6 +417,8 @@ static void nfs4_add_trunk(struct nfs_client *clp, struct nfs_client *old)
 		.net = old->cl_net,
 		.servername = old->cl_hostname,
 	};
+	int max_connect = test_bit(NFS_CS_PNFS, &clp->cl_flags) ?
+		clp->cl_max_connect : old->cl_max_connect;
 
 	if (clp->cl_proto != old->cl_proto)
 		return;
@@ -430,7 +432,7 @@ static void nfs4_add_trunk(struct nfs_client *clp, struct nfs_client *old)
 	xprt_args.addrlen = clp_salen;
 
 	rpc_clnt_add_xprt(old->cl_rpcclient, &xprt_args,
-			  rpc_clnt_test_and_add_xprt, NULL);
+			  rpc_clnt_test_and_add_xprt, &max_connect);
 }
 
 /**
@@ -1010,6 +1012,8 @@ struct nfs_client *nfs4_set_ds_client(struct nfs_server *mds_srv,
 		__set_bit(NFS_CS_NORESVPORT, &cl_init.init_flags);
 
 	__set_bit(NFS_CS_DS, &cl_init.init_flags);
+	__set_bit(NFS_CS_PNFS, &cl_init.init_flags);
+	cl_init.max_connect = NFS_MAX_TRANSPORTS;
 	/*
 	 * Set an authflavor equual to the MDS value. Use the MDS nfs_client
 	 * cl_ipaddr so as to use the same EXCHANGE_ID co_ownerid as the MDS
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 794343790ea8bb..3508d82388263e 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2703,8 +2703,12 @@ static int _nfs4_proc_open(struct nfs4_opendata *data,
 			return status;
 	}
 	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR)) {
+		struct nfs_fh *fh = &o_res->fh;
+
 		nfs4_sequence_free_slot(&o_res->seq_res);
-		nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr, NULL);
+		if (o_arg->claim == NFS4_OPEN_CLAIM_FH)
+			fh = NFS_FH(d_inode(data->dentry));
+		nfs4_proc_getattr(server, fh, o_res->f_attr, NULL);
 	}
 	return 0;
 }
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index f4cca8f00c0c20..8c1ee1a1a28f11 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -59,7 +59,8 @@ static const struct nfs_pgio_completion_ops nfs_async_write_completion_ops;
 static const struct nfs_commit_completion_ops nfs_commit_completion_ops;
 static const struct nfs_rw_ops nfs_rw_write_ops;
 static void nfs_inode_remove_request(struct nfs_page *req);
-static void nfs_clear_request_commit(struct nfs_page *req);
+static void nfs_clear_request_commit(struct nfs_commit_info *cinfo,
+				     struct nfs_page *req);
 static void nfs_init_cinfo_from_inode(struct nfs_commit_info *cinfo,
 				      struct inode *inode);
 static struct nfs_page *
@@ -502,8 +503,8 @@ nfs_destroy_unlinked_subrequests(struct nfs_page *destroy_list,
  * the (former) group.  All subrequests are removed from any write or commit
  * lists, unlinked from the group and destroyed.
  */
-void
-nfs_join_page_group(struct nfs_page *head, struct inode *inode)
+void nfs_join_page_group(struct nfs_page *head, struct nfs_commit_info *cinfo,
+			 struct inode *inode)
 {
 	struct nfs_page *subreq;
 	struct nfs_page *destroy_list = NULL;
@@ -533,7 +534,7 @@ nfs_join_page_group(struct nfs_page *head, struct inode *inode)
 	 * Commit list removal accounting is done after locks are dropped */
 	subreq = head;
 	do {
-		nfs_clear_request_commit(subreq);
+		nfs_clear_request_commit(cinfo, subreq);
 		subreq = subreq->wb_this_page;
 	} while (subreq != head);
 
@@ -566,8 +567,10 @@ static struct nfs_page *nfs_lock_and_join_requests(struct folio *folio)
 {
 	struct inode *inode = folio_file_mapping(folio)->host;
 	struct nfs_page *head;
+	struct nfs_commit_info cinfo;
 	int ret;
 
+	nfs_init_cinfo_from_inode(&cinfo, inode);
 	/*
 	 * A reference is taken only on the head request which acts as a
 	 * reference to the whole page group - the group will not be destroyed
@@ -584,7 +587,7 @@ static struct nfs_page *nfs_lock_and_join_requests(struct folio *folio)
 		return ERR_PTR(ret);
 	}
 
-	nfs_join_page_group(head, inode);
+	nfs_join_page_group(head, &cinfo, inode);
 
 	return head;
 }
@@ -955,18 +958,16 @@ static void nfs_folio_clear_commit(struct folio *folio)
 }
 
 /* Called holding the request lock on @req */
-static void
-nfs_clear_request_commit(struct nfs_page *req)
+static void nfs_clear_request_commit(struct nfs_commit_info *cinfo,
+				     struct nfs_page *req)
 {
 	if (test_bit(PG_CLEAN, &req->wb_flags)) {
 		struct nfs_open_context *ctx = nfs_req_openctx(req);
 		struct inode *inode = d_inode(ctx->dentry);
-		struct nfs_commit_info cinfo;
 
-		nfs_init_cinfo_from_inode(&cinfo, inode);
 		mutex_lock(&NFS_I(inode)->commit_mutex);
-		if (!pnfs_clear_request_commit(req, &cinfo)) {
-			nfs_request_remove_commit_list(req, &cinfo);
+		if (!pnfs_clear_request_commit(req, cinfo)) {
+			nfs_request_remove_commit_list(req, cinfo);
 		}
 		mutex_unlock(&NFS_I(inode)->commit_mutex);
 		nfs_folio_clear_commit(nfs_page_to_folio(req));
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 20eeba8b009df1..cd628c4b011e54 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -48,6 +48,7 @@ struct nfs_client {
 #define NFS_CS_NOPING		6		/* - don't ping on connect */
 #define NFS_CS_DS		7		/* - Server is a DS */
 #define NFS_CS_REUSEPORT	8		/* - reuse src port on reconnect */
+#define NFS_CS_PNFS		9		/* - Server used for pnfs */
 	struct sockaddr_storage	cl_addr;	/* server identifier */
 	size_t			cl_addrlen;
 	char *			cl_hostname;	/* hostname of server */
diff --git a/include/linux/nfs_page.h b/include/linux/nfs_page.h
index aa9f4c6ebe261d..1c315f854ea801 100644
--- a/include/linux/nfs_page.h
+++ b/include/linux/nfs_page.h
@@ -157,7 +157,9 @@ extern	void nfs_unlock_request(struct nfs_page *req);
 extern	void nfs_unlock_and_release_request(struct nfs_page *);
 extern	struct nfs_page *nfs_page_group_lock_head(struct nfs_page *req);
 extern	int nfs_page_group_lock_subrequests(struct nfs_page *head);
-extern	void nfs_join_page_group(struct nfs_page *head, struct inode *inode);
+extern void nfs_join_page_group(struct nfs_page *head,
+				struct nfs_commit_info *cinfo,
+				struct inode *inode);
 extern int nfs_page_group_lock(struct nfs_page *);
 extern void nfs_page_group_unlock(struct nfs_page *);
 extern bool nfs_page_group_sync_on_bit(struct nfs_page *, unsigned int);
diff --git a/include/linux/sunrpc/xdr.h b/include/linux/sunrpc/xdr.h
index 5b4fb3c791bc26..2f8dc47f1eb075 100644
--- a/include/linux/sunrpc/xdr.h
+++ b/include/linux/sunrpc/xdr.h
@@ -779,7 +779,9 @@ xdr_stream_decode_uint32_array(struct xdr_stream *xdr,
 
 	if (unlikely(xdr_stream_decode_u32(xdr, &len) < 0))
 		return -EBADMSG;
-	p = xdr_inline_decode(xdr, size_mul(len, sizeof(*p)));
+	if (U32_MAX >= SIZE_MAX / sizeof(*p) && len > SIZE_MAX / sizeof(*p))
+		return -EBADMSG;
+	p = xdr_inline_decode(xdr, len * sizeof(*p));
 	if (unlikely(!p))
 		return -EBADMSG;
 	if (array == NULL)
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 8d75290f1a31d2..37b0b212b93475 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -2725,7 +2725,7 @@ out_unparsable:
 
 out_verifier:
 	trace_rpc_bad_verifier(task);
-	goto out_err;
+	goto out_garbage;
 
 out_msg_denied:
 	error = -EACCES;
@@ -2751,6 +2751,7 @@ out_msg_denied:
 	case rpc_autherr_rejectedverf:
 	case rpcsec_gsserr_credproblem:
 	case rpcsec_gsserr_ctxproblem:
+		rpcauth_invalcred(task);
 		if (!task->tk_cred_retry)
 			break;
 		task->tk_cred_retry--;
@@ -2907,19 +2908,22 @@ static const struct rpc_call_ops rpc_cb_add_xprt_call_ops = {
  * @clnt: pointer to struct rpc_clnt
  * @xps: pointer to struct rpc_xprt_switch,
  * @xprt: pointer struct rpc_xprt
- * @dummy: unused
+ * @in_max_connect: pointer to the max_connect value for the passed in xprt transport
  */
 int rpc_clnt_test_and_add_xprt(struct rpc_clnt *clnt,
 		struct rpc_xprt_switch *xps, struct rpc_xprt *xprt,
-		void *dummy)
+		void *in_max_connect)
 {
 	struct rpc_cb_add_xprt_calldata *data;
 	struct rpc_task *task;
+	int max_connect = clnt->cl_max_connect;
 
-	if (xps->xps_nunique_destaddr_xprts + 1 > clnt->cl_max_connect) {
+	if (in_max_connect)
+		max_connect = *(int *)in_max_connect;
+	if (xps->xps_nunique_destaddr_xprts + 1 > max_connect) {
 		rcu_read_lock();
 		pr_warn("SUNRPC: reached max allowed number (%d) did not add "
-			"transport to server: %s\n", clnt->cl_max_connect,
+			"transport to server: %s\n", max_connect,
 			rpc_peeraddr2str(clnt, RPC_DISPLAY_ADDR));
 		rcu_read_unlock();
 		return -EINVAL;