1 files changed, 121 insertions, 38 deletions

diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
index 6f8192f4cfc..be93a622872 100644
--- a/fs/nfs/nfs4filelayout.c
+++ b/fs/nfs/nfs4filelayout.c
@@ -30,6 +30,7 @@
  */
 
 #include <linux/nfs_fs.h>
+#include <linux/nfs_page.h>
 
 #include "internal.h"
 #include "nfs4filelayout.h"
@@ -117,6 +118,8 @@ static int filelayout_async_handle_error(struct rpc_task *task,
 	case -EKEYEXPIRED:
 		rpc_delay(task, FILELAYOUT_POLL_RETRY_MAX);
 		break;
+	case -NFS4ERR_RETRY_UNCACHED_REP:
+		break;
 	default:
 		dprintk("%s DS error. Retry through MDS %d\n", __func__,
 			task->tk_status);
@@ -331,6 +334,9 @@ filelayout_read_pagelist(struct nfs_read_data *data)
 		__func__, data->inode->i_ino,
 		data->args.pgbase, (size_t)data->args.count, offset);
 
+	if (test_bit(NFS_DEVICEID_INVALID, &FILELAYOUT_DEVID_NODE(lseg)->flags))
+		return PNFS_NOT_ATTEMPTED;
+
 	/* Retrieve the correct rpc_client for the byte range */
 	j = nfs4_fl_calc_j_index(lseg, offset);
 	idx = nfs4_fl_calc_ds_index(lseg, j);
@@ -341,8 +347,7 @@ filelayout_read_pagelist(struct nfs_read_data *data)
 		set_bit(lo_fail_bit(IOMODE_READ), &lseg->pls_layout->plh_flags);
 		return PNFS_NOT_ATTEMPTED;
 	}
-	dprintk("%s USE DS:ip %x %hu\n", __func__,
-		ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
+	dprintk("%s USE DS: %s\n", __func__, ds->ds_remotestr);
 
 	/* No multipath support. Use first DS */
 	data->ds_clp = ds->ds_clp;
@@ -371,6 +376,9 @@ filelayout_write_pagelist(struct nfs_write_data *data, int sync)
 	struct nfs_fh *fh;
 	int status;
 
+	if (test_bit(NFS_DEVICEID_INVALID, &FILELAYOUT_DEVID_NODE(lseg)->flags))
+		return PNFS_NOT_ATTEMPTED;
+
 	/* Retrieve the correct rpc_client for the byte range */
 	j = nfs4_fl_calc_j_index(lseg, offset);
 	idx = nfs4_fl_calc_ds_index(lseg, j);
@@ -381,9 +389,9 @@ filelayout_write_pagelist(struct nfs_write_data *data, int sync)
 		set_bit(lo_fail_bit(IOMODE_READ), &lseg->pls_layout->plh_flags);
 		return PNFS_NOT_ATTEMPTED;
 	}
-	dprintk("%s ino %lu sync %d req %Zu@%llu DS:%x:%hu\n", __func__,
+	dprintk("%s ino %lu sync %d req %Zu@%llu DS: %s\n", __func__,
 		data->inode->i_ino, sync, (size_t) data->args.count, offset,
-		ntohl(ds->ds_ip_addr), ntohs(ds->ds_port));
+		ds->ds_remotestr);
 
 	data->write_done_cb = filelayout_write_done_cb;
 	data->ds_clp = ds->ds_clp;
@@ -395,7 +403,6 @@ filelayout_write_pagelist(struct nfs_write_data *data, int sync)
 	 * this offset and save the original offset.
 	 */
 	data->args.offset = filelayout_get_dserver_offset(lseg, offset);
-	data->mds_offset = offset;
 
 	/* Perform an asynchronous write */
 	status = nfs_initiate_write(data, ds->ds_clp->cl_rpcclient,
@@ -416,16 +423,26 @@ static int
 filelayout_check_layout(struct pnfs_layout_hdr *lo,
 			struct nfs4_filelayout_segment *fl,
 			struct nfs4_layoutget_res *lgr,
-			struct nfs4_deviceid *id)
+			struct nfs4_deviceid *id,
+			gfp_t gfp_flags)
 {
+	struct nfs4_deviceid_node *d;
 	struct nfs4_file_layout_dsaddr *dsaddr;
 	int status = -EINVAL;
 	struct nfs_server *nfss = NFS_SERVER(lo->plh_inode);
 
 	dprintk("--> %s\n", __func__);
 
+	/* FIXME: remove this check when layout segment support is added */
+	if (lgr->range.offset != 0 ||
+	    lgr->range.length != NFS4_MAX_UINT64) {
+		dprintk("%s Only whole file layouts supported. Use MDS i/o\n",
+			__func__);
+		goto out;
+	}
+
 	if (fl->pattern_offset > lgr->range.offset) {
-		dprintk("%s pattern_offset %lld to large\n",
+		dprintk("%s pattern_offset %lld too large\n",
 				__func__, fl->pattern_offset);
 		goto out;
 	}
@@ -437,12 +454,18 @@ filelayout_check_layout(struct pnfs_layout_hdr *lo,
 	}
 
 	/* find and reference the deviceid */
-	dsaddr = nfs4_fl_find_get_deviceid(id);
-	if (dsaddr == NULL) {
-		dsaddr = get_device_info(lo->plh_inode, id);
+	d = nfs4_find_get_deviceid(NFS_SERVER(lo->plh_inode)->pnfs_curr_ld,
+				   NFS_SERVER(lo->plh_inode)->nfs_client, id);
+	if (d == NULL) {
+		dsaddr = get_device_info(lo->plh_inode, id, gfp_flags);
 		if (dsaddr == NULL)
 			goto out;
-	}
+	} else
+		dsaddr = container_of(d, struct nfs4_file_layout_dsaddr, id_node);
+	/* Found deviceid is being reaped */
+	if (test_bit(NFS_DEVICEID_INVALID, &dsaddr->id_node.flags))
+			goto out_put;
+
 	fl->dsaddr = dsaddr;
 
 	if (fl->first_stripe_index < 0 ||
@@ -500,15 +523,11 @@ static int
 filelayout_decode_layout(struct pnfs_layout_hdr *flo,
 			 struct nfs4_filelayout_segment *fl,
 			 struct nfs4_layoutget_res *lgr,
-			 struct nfs4_deviceid *id)
+			 struct nfs4_deviceid *id,
+			 gfp_t gfp_flags)
 {
 	struct xdr_stream stream;
-	struct xdr_buf buf = {
-		.pages =  lgr->layoutp->pages,
-		.page_len =  lgr->layoutp->len,
-		.buflen =  lgr->layoutp->len,
-		.len = lgr->layoutp->len,
-	};
+	struct xdr_buf buf;
 	struct page *scratch;
 	__be32 *p;
 	uint32_t nfl_util;
@@ -516,11 +535,11 @@ filelayout_decode_layout(struct pnfs_layout_hdr *flo,
 
 	dprintk("%s: set_layout_map Begin\n", __func__);
 
-	scratch = alloc_page(GFP_KERNEL);
+	scratch = alloc_page(gfp_flags);
 	if (!scratch)
 		return -ENOMEM;
 
-	xdr_init_decode(&stream, &buf, NULL);
+	xdr_init_decode_pages(&stream, &buf, lgr->layoutp->pages, lgr->layoutp->len);
 	xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
 
 	/* 20 = ufl_util (4), first_stripe_index (4), pattern_offset (8),
@@ -531,7 +550,7 @@ filelayout_decode_layout(struct pnfs_layout_hdr *flo,
 
 	memcpy(id, p, sizeof(*id));
 	p += XDR_QUADLEN(NFS4_DEVICEID4_SIZE);
-	print_deviceid(id);
+	nfs4_print_deviceid(id);
 
 	nfl_util = be32_to_cpup(p++);
 	if (nfl_util & NFL4_UFLG_COMMIT_THRU_MDS)
@@ -550,17 +569,22 @@ filelayout_decode_layout(struct pnfs_layout_hdr *flo,
 		__func__, nfl_util, fl->num_fh, fl->first_stripe_index,
 		fl->pattern_offset);
 
-	if (!fl->num_fh)
+	/* Note that a zero value for num_fh is legal for STRIPE_SPARSE.
+	 * Futher checking is done in filelayout_check_layout */
+	if (fl->num_fh < 0 || fl->num_fh >
+	    max(NFS4_PNFS_MAX_STRIPE_CNT, NFS4_PNFS_MAX_MULTI_CNT))
 		goto out_err;
 
-	fl->fh_array = kzalloc(fl->num_fh * sizeof(struct nfs_fh *),
-			       GFP_KERNEL);
-	if (!fl->fh_array)
-		goto out_err;
+	if (fl->num_fh > 0) {
+		fl->fh_array = kzalloc(fl->num_fh * sizeof(struct nfs_fh *),
+				       gfp_flags);
+		if (!fl->fh_array)
+			goto out_err;
+	}
 
 	for (i = 0; i < fl->num_fh; i++) {
 		/* Do we want to use a mempool here? */
-		fl->fh_array[i] = kmalloc(sizeof(struct nfs_fh), GFP_KERNEL);
+		fl->fh_array[i] = kmalloc(sizeof(struct nfs_fh), gfp_flags);
 		if (!fl->fh_array[i])
 			goto out_err_free;
 
@@ -605,19 +629,20 @@ filelayout_free_lseg(struct pnfs_layout_segment *lseg)
 
 static struct pnfs_layout_segment *
 filelayout_alloc_lseg(struct pnfs_layout_hdr *layoutid,
-		      struct nfs4_layoutget_res *lgr)
+		      struct nfs4_layoutget_res *lgr,
+		      gfp_t gfp_flags)
 {
 	struct nfs4_filelayout_segment *fl;
 	int rc;
 	struct nfs4_deviceid id;
 
 	dprintk("--> %s\n", __func__);
-	fl = kzalloc(sizeof(*fl), GFP_KERNEL);
+	fl = kzalloc(sizeof(*fl), gfp_flags);
 	if (!fl)
 		return NULL;
 
-	rc = filelayout_decode_layout(layoutid, fl, lgr, &id);
-	if (rc != 0 || filelayout_check_layout(layoutid, fl, lgr, &id)) {
+	rc = filelayout_decode_layout(layoutid, fl, lgr, &id, gfp_flags);
+	if (rc != 0 || filelayout_check_layout(layoutid, fl, lgr, &id, gfp_flags)) {
 		_filelayout_free_lseg(fl);
 		return NULL;
 	}
@@ -633,7 +658,7 @@ filelayout_alloc_lseg(struct pnfs_layout_hdr *layoutid,
 		int size = (fl->stripe_type == STRIPE_SPARSE) ?
 			fl->dsaddr->ds_num : fl->dsaddr->stripe_count;
 
-		fl->commit_buckets = kcalloc(size, sizeof(struct list_head), GFP_KERNEL);
+		fl->commit_buckets = kcalloc(size, sizeof(struct list_head), gfp_flags);
 		if (!fl->commit_buckets) {
 			filelayout_free_lseg(&fl->generic_hdr);
 			return NULL;
@@ -648,18 +673,20 @@ filelayout_alloc_lseg(struct pnfs_layout_hdr *layoutid,
 /*
  * filelayout_pg_test(). Called by nfs_can_coalesce_requests()
  *
- * return 1 :  coalesce page
- * return 0 :  don't coalesce page
+ * return true  : coalesce page
+ * return false : don't coalesce page
  */
-int
+static bool
 filelayout_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
 		   struct nfs_page *req)
 {
 	u64 p_stripe, r_stripe;
 	u32 stripe_unit;
 
-	if (!pgio->pg_lseg)
-		return 1;
+	if (!pnfs_generic_pg_test(pgio, prev, req) ||
+	    !nfs_generic_pg_test(pgio, prev, req))
+		return false;
+
 	p_stripe = (u64)prev->wb_index << PAGE_CACHE_SHIFT;
 	r_stripe = (u64)req->wb_index << PAGE_CACHE_SHIFT;
 	stripe_unit = FILELAYOUT_LSEG(pgio->pg_lseg)->stripe_unit;
@@ -670,6 +697,52 @@ filelayout_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
 	return (p_stripe == r_stripe);
 }
 
+void
+filelayout_pg_init_read(struct nfs_pageio_descriptor *pgio,
+			struct nfs_page *req)
+{
+	BUG_ON(pgio->pg_lseg != NULL);
+
+	pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
+					   req->wb_context,
+					   0,
+					   NFS4_MAX_UINT64,
+					   IOMODE_READ,
+					   GFP_KERNEL);
+	/* If no lseg, fall back to read through mds */
+	if (pgio->pg_lseg == NULL)
+		nfs_pageio_reset_read_mds(pgio);
+}
+
+void
+filelayout_pg_init_write(struct nfs_pageio_descriptor *pgio,
+			 struct nfs_page *req)
+{
+	BUG_ON(pgio->pg_lseg != NULL);
+
+	pgio->pg_lseg = pnfs_update_layout(pgio->pg_inode,
+					   req->wb_context,
+					   0,
+					   NFS4_MAX_UINT64,
+					   IOMODE_RW,
+					   GFP_NOFS);
+	/* If no lseg, fall back to write through mds */
+	if (pgio->pg_lseg == NULL)
+		nfs_pageio_reset_write_mds(pgio);
+}
+
+static const struct nfs_pageio_ops filelayout_pg_read_ops = {
+	.pg_init = filelayout_pg_init_read,
+	.pg_test = filelayout_pg_test,
+	.pg_doio = pnfs_generic_pg_readpages,
+};
+
+static const struct nfs_pageio_ops filelayout_pg_write_ops = {
+	.pg_init = filelayout_pg_init_write,
+	.pg_test = filelayout_pg_test,
+	.pg_doio = pnfs_generic_pg_writepages,
+};
+
 static bool filelayout_mark_pnfs_commit(struct pnfs_layout_segment *lseg)
 {
 	return !FILELAYOUT_LSEG(lseg)->commit_through_mds;
@@ -855,18 +928,26 @@ filelayout_commit_pagelist(struct inode *inode, struct list_head *mds_pages,
 	return -ENOMEM;
 }
 
+static void
+filelayout_free_deveiceid_node(struct nfs4_deviceid_node *d)
+{
+	nfs4_fl_free_deviceid(container_of(d, struct nfs4_file_layout_dsaddr, id_node));
+}
+
 static struct pnfs_layoutdriver_type filelayout_type = {
 	.id			= LAYOUT_NFSV4_1_FILES,
 	.name			= "LAYOUT_NFSV4_1_FILES",
 	.owner			= THIS_MODULE,
 	.alloc_lseg		= filelayout_alloc_lseg,
 	.free_lseg		= filelayout_free_lseg,
-	.pg_test		= filelayout_pg_test,
+	.pg_read_ops		= &filelayout_pg_read_ops,
+	.pg_write_ops		= &filelayout_pg_write_ops,
 	.mark_pnfs_commit	= filelayout_mark_pnfs_commit,
 	.choose_commit_list	= filelayout_choose_commit_list,
 	.commit_pagelist	= filelayout_commit_pagelist,
 	.read_pagelist		= filelayout_read_pagelist,
 	.write_pagelist		= filelayout_write_pagelist,
+	.free_deviceid_node	= filelayout_free_deveiceid_node,
 };
 
 static int __init nfs4filelayout_init(void)
@@ -883,5 +964,7 @@ static void __exit nfs4filelayout_exit(void)
 	pnfs_unregister_layoutdriver(&filelayout_type);
 }
 
+MODULE_ALIAS("nfs-layouttype4-1");
+
 module_init(nfs4filelayout_init);
 module_exit(nfs4filelayout_exit);