Merge tag 'nfs-for-4.14-2' of git://git.linux-nfs.org/projects/trondm…

…y/linux-nfs

Pull more NFS client updates from Trond Myklebust:
 "Hightlights include:

  Bugfixes:
   - Various changes relating to reporting IO errors.
   - pnfs: Use the standard I/O stateid when calling LAYOUTGET

  Features:
   - Add static NFS I/O tracepoints for debugging"

* tag 'nfs-for-4.14-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFS: various changes relating to reporting IO errors.
  NFS: Add static NFS I/O tracepoints
  pNFS: Use the standard I/O stateid when calling LAYOUTGET

fs/nfs/file.c

@@ -208,21 +208,19 @@ EXPORT_SYMBOL_GPL(nfs_file_mmap);
  * fall back to doing a synchronous write.
  */
 static int
-nfs_file_fsync_commit(struct file *file, loff_t start, loff_t end, int datasync)
+nfs_file_fsync_commit(struct file *file, int datasync)
 {
 	struct nfs_open_context *ctx = nfs_file_open_context(file);
 	struct inode *inode = file_inode(file);
-	int have_error, do_resend, status;
+	int do_resend, status;
 	int ret = 0;
 
 	dprintk("NFS: fsync file(%pD2) datasync %d\n", file, datasync);
 
 	nfs_inc_stats(inode, NFSIOS_VFSFSYNC);
 	do_resend = test_and_clear_bit(NFS_CONTEXT_RESEND_WRITES, &ctx->flags);
-	have_error = test_and_clear_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
 	status = nfs_commit_inode(inode, FLUSH_SYNC);
-	have_error |= test_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
-	if (have_error) {
+	if (test_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags)) {
 		ret = xchg(&ctx->error, 0);
 		if (ret)
 			goto out;
@@ -247,10 +245,16 @@ nfs_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 	trace_nfs_fsync_enter(inode);
 
 	do {
+		struct nfs_open_context *ctx = nfs_file_open_context(file);
 		ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
+		if (test_and_clear_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags)) {
+			int ret2 = xchg(&ctx->error, 0);
+			if (ret2)
+				ret = ret2;
+		}
 		if (ret != 0)
 			break;
-		ret = nfs_file_fsync_commit(file, start, end, datasync);
+		ret = nfs_file_fsync_commit(file, datasync);
 		if (!ret)
 			ret = pnfs_sync_inode(inode, !!datasync);
 		/*

fs/nfs/internal.h

@@ -768,3 +768,10 @@ static inline bool nfs_error_is_fatal(int err)
 		return false;
 	}
 }
+
+static inline void nfs_context_set_write_error(struct nfs_open_context *ctx, int error)
+{
+	ctx->error = error;
+	smp_wmb();
+	set_bit(NFS_CONTEXT_ERROR_WRITE, &ctx->flags);
+}

fs/nfs/nfstrace.h

@@ -719,6 +719,254 @@ TRACE_EVENT(nfs_sillyrename_unlink,
 			__get_str(name)
 		)
 );
+
+TRACE_EVENT(nfs_initiate_read,
+		TP_PROTO(
+			const struct inode *inode,
+			loff_t offset, unsigned long count
+		),
+
+		TP_ARGS(inode, offset, count),
+
+		TP_STRUCT__entry(
+			__field(loff_t, offset)
+			__field(unsigned long, count)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->offset = offset;
+			__entry->count = count;
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld count=%lu",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->count
+		)
+);
+
+TRACE_EVENT(nfs_readpage_done,
+		TP_PROTO(
+			const struct inode *inode,
+			int status, loff_t offset, bool eof
+		),
+
+		TP_ARGS(inode, status, offset, eof),
+
+		TP_STRUCT__entry(
+			__field(int, status)
+			__field(loff_t, offset)
+			__field(bool, eof)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->status = status;
+			__entry->offset = offset;
+			__entry->eof = eof;
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld status=%d%s",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->status,
+			__entry->eof ? " eof" : ""
+		)
+);
+
+/*
+ * XXX: I tried using NFS_UNSTABLE and friends in this table, but they
+ * all evaluate to 0 for some reason, even if I include linux/nfs.h.
+ */
+#define nfs_show_stable(stable) \
+	__print_symbolic(stable, \
+			{ 0, " (UNSTABLE)" }, \
+			{ 1, " (DATA_SYNC)" }, \
+			{ 2, " (FILE_SYNC)" })
+
+TRACE_EVENT(nfs_initiate_write,
+		TP_PROTO(
+			const struct inode *inode,
+			loff_t offset, unsigned long count,
+			enum nfs3_stable_how stable
+		),
+
+		TP_ARGS(inode, offset, count, stable),
+
+		TP_STRUCT__entry(
+			__field(loff_t, offset)
+			__field(unsigned long, count)
+			__field(enum nfs3_stable_how, stable)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->offset = offset;
+			__entry->count = count;
+			__entry->stable = stable;
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld count=%lu stable=%d%s",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->count,
+			__entry->stable, nfs_show_stable(__entry->stable)
+		)
+);
+
+TRACE_EVENT(nfs_writeback_done,
+		TP_PROTO(
+			const struct inode *inode,
+			int status,
+			loff_t offset,
+			struct nfs_writeverf *writeverf
+		),
+
+		TP_ARGS(inode, status, offset, writeverf),
+
+		TP_STRUCT__entry(
+			__field(int, status)
+			__field(loff_t, offset)
+			__field(enum nfs3_stable_how, stable)
+			__field(unsigned long long, verifier)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->status = status;
+			__entry->offset = offset;
+			__entry->stable = writeverf->committed;
+			memcpy(&__entry->verifier, &writeverf->verifier,
+			       sizeof(__entry->verifier));
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld status=%d stable=%d%s "
+			"verifier 0x%016llx",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->status,
+			__entry->stable, nfs_show_stable(__entry->stable),
+			__entry->verifier
+		)
+);
+
+TRACE_EVENT(nfs_initiate_commit,
+		TP_PROTO(
+			const struct nfs_commit_data *data
+		),
+
+		TP_ARGS(data),
+
+		TP_STRUCT__entry(
+			__field(loff_t, offset)
+			__field(unsigned long, count)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct inode *inode = data->inode;
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->offset = data->args.offset;
+			__entry->count = data->args.count;
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld count=%lu",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->count
+		)
+);
+
+TRACE_EVENT(nfs_commit_done,
+		TP_PROTO(
+			const struct nfs_commit_data *data
+		),
+
+		TP_ARGS(data),
+
+		TP_STRUCT__entry(
+			__field(int, status)
+			__field(loff_t, offset)
+			__field(unsigned long long, verifier)
+			__field(dev_t, dev)
+			__field(u32, fhandle)
+			__field(u64, fileid)
+		),
+
+		TP_fast_assign(
+			const struct inode *inode = data->inode;
+			const struct nfs_inode *nfsi = NFS_I(inode);
+
+			__entry->status = data->res.op_status;
+			__entry->offset = data->args.offset;
+			memcpy(&__entry->verifier, &data->verf.verifier,
+			       sizeof(__entry->verifier));
+			__entry->dev = inode->i_sb->s_dev;
+			__entry->fileid = nfsi->fileid;
+			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
+		),
+
+		TP_printk(
+			"fileid=%02x:%02x:%llu fhandle=0x%08x "
+			"offset=%lld status=%d verifier 0x%016llx",
+			MAJOR(__entry->dev), MINOR(__entry->dev),
+			(unsigned long long)__entry->fileid,
+			__entry->fhandle,
+			__entry->offset, __entry->status,
+			__entry->verifier
+		)
+);
+
 #endif /* _TRACE_NFS_H */
 
 #undef TRACE_INCLUDE_PATH

fs/nfs/pagelist.c

@@ -1170,8 +1170,8 @@ int nfs_pageio_add_request(struct nfs_pageio_descriptor *desc,
 
 		/* remember fatal errors */
 		if (nfs_error_is_fatal(desc->pg_error))
-			mapping_set_error(desc->pg_inode->i_mapping,
-					  desc->pg_error);
+			nfs_context_set_write_error(req->wb_context,
+						    desc->pg_error);
 
 		func = desc->pg_completion_ops->error_cleanup;
 		for (midx = 0; midx < desc->pg_mirror_count; midx++) {

fs/nfs/pnfs.c

@@ -1664,7 +1664,7 @@ pnfs_update_layout(struct inode *ino,
 		.offset = pos,
 		.length = count,
 	};
-	unsigned pg_offset, seq;
+	unsigned pg_offset;
 	struct nfs_server *server = NFS_SERVER(ino);
 	struct nfs_client *clp = server->nfs_client;
 	struct pnfs_layout_hdr *lo = NULL;
@@ -1754,10 +1754,14 @@ pnfs_update_layout(struct inode *ino,
 		}
 
 		first = true;
-		do {
-			seq = read_seqbegin(&ctx->state->seqlock);
-			nfs4_stateid_copy(&stateid, &ctx->state->stateid);
-		} while (read_seqretry(&ctx->state->seqlock, seq));
+		if (nfs4_select_rw_stateid(ctx->state,
+					iomode == IOMODE_RW ? FMODE_WRITE : FMODE_READ,
+					NULL, &stateid, NULL) != 0) {
+			trace_pnfs_update_layout(ino, pos, count,
+					iomode, lo, lseg,
+					PNFS_UPDATE_LAYOUT_INVALID_OPEN);
+			goto out_unlock;
+		}
 	} else {
 		nfs4_stateid_copy(&stateid, &lo->plh_stateid);
 	}

fs/nfs/read.c

@@ -25,6 +25,7 @@
 #include "iostat.h"
 #include "fscache.h"
 #include "pnfs.h"
+#include "nfstrace.h"
 
 #define NFSDBG_FACILITY		NFSDBG_PAGECACHE
 
@@ -200,6 +201,7 @@ static void nfs_initiate_read(struct nfs_pgio_header *hdr,
 
 	task_setup_data->flags |= swap_flags;
 	rpc_ops->read_setup(hdr, msg);
+	trace_nfs_initiate_read(inode, hdr->io_start, hdr->good_bytes);
 }
 
 static void
@@ -232,6 +234,8 @@ static int nfs_readpage_done(struct rpc_task *task,
 		return status;
 
 	nfs_add_stats(inode, NFSIOS_SERVERREADBYTES, hdr->res.count);
+	trace_nfs_readpage_done(inode, task->tk_status,
+				hdr->args.offset, hdr->res.eof);
 
 	if (task->tk_status == -ESTALE) {
 		set_bit(NFS_INO_STALE, &NFS_I(inode)->flags);