aboutsummaryrefslogtreecommitdiffstats
path: root/fs/nfs
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2015-04-26 20:33:59 -0400
committerLinus Torvalds <torvalds@linux-foundation.org>2015-04-26 20:33:59 -0400
commit59953fba87e5e535657403cc6439d24187929559 (patch)
tree4f92cc3bcacf052cb3fb895512af5a7d3dad86cb /fs/nfs
parent9ec3a646fe09970f801ab15e0f1694060b9f19af (diff)
parentf139b6c676c7e49b66016b28bf3f8ec5c54be891 (diff)
Merge tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client updates from Trond Myklebust: "Another set of mainly bugfixes and a couple of cleanups. No new functionality in this round. Highlights include: Stable patches: - Fix a regression in /proc/self/mountstats - Fix the pNFS flexfiles O_DIRECT support - Fix high load average due to callback thread sleeping Bugfixes: - Various patches to fix the pNFS layoutcommit support - Do not cache pNFS deviceids unless server notifications are enabled - Fix a SUNRPC transport reconnection regression - make debugfs file creation failure non-fatal in SUNRPC - Another fix for circular directory warnings on NFSv4 "junctioned" mountpoints - Fix locking around NFSv4.2 fallocate() support - Truncating NFSv4 file opens should also sync O_DIRECT writes - Prevent infinite loop in rpcrdma_ep_create() Features: - Various improvements to the RDMA transport code's handling of memory registration - Various code cleanups" * tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (55 commits) fs/nfs: fix new compiler warning about boolean in switch nfs: Remove unneeded casts in nfs NFS: Don't attempt to decode missing directory entries Revert "nfs: replace nfs_add_stats with nfs_inc_stats when add one" NFS: Rename idmap.c to nfs4idmap.c NFS: Move nfs_idmap.h into fs/nfs/ NFS: Remove CONFIG_NFS_V4 checks from nfs_idmap.h NFS: Add a stub for GETDEVICELIST nfs: remove WARN_ON_ONCE from nfs_direct_good_bytes nfs: fix DIO good bytes calculation nfs: Fetch MOUNTED_ON_FILEID when updating an inode sunrpc: make debugfs file creation failure non-fatal nfs: fix high load average due to callback thread sleeping NFS: Reduce time spent holding the i_mutex during fallocate() NFS: Don't zap caches on fallocate() xprtrdma: Make rpcrdma_{un}map_one() into inline functions xprtrdma: Handle non-SEND completions via a callout xprtrdma: Add "open" memreg op xprtrdma: Add "destroy MRs" memreg op xprtrdma: Add "reset MRs" memreg op ...
Diffstat (limited to 'fs/nfs')
-rw-r--r--fs/nfs/Makefile2
-rw-r--r--fs/nfs/blocklayout/blocklayout.c1
-rw-r--r--fs/nfs/blocklayout/dev.c2
-rw-r--r--fs/nfs/callback.c6
-rw-r--r--fs/nfs/client.c1
-rw-r--r--fs/nfs/delegation.c4
-rw-r--r--fs/nfs/dir.c4
-rw-r--r--fs/nfs/direct.c39
-rw-r--r--fs/nfs/file.c3
-rw-r--r--fs/nfs/filelayout/filelayout.c10
-rw-r--r--fs/nfs/filelayout/filelayoutdev.c2
-rw-r--r--fs/nfs/flexfilelayout/flexfilelayout.c12
-rw-r--r--fs/nfs/flexfilelayout/flexfilelayoutdev.c2
-rw-r--r--fs/nfs/inode.c36
-rw-r--r--fs/nfs/nfs42proc.c31
-rw-r--r--fs/nfs/nfs42xdr.c20
-rw-r--r--fs/nfs/nfs4client.c2
-rw-r--r--fs/nfs/nfs4file.c20
-rw-r--r--fs/nfs/nfs4idmap.c (renamed from fs/nfs/idmap.c)2
-rw-r--r--fs/nfs/nfs4idmap.h68
-rw-r--r--fs/nfs/nfs4proc.c23
-rw-r--r--fs/nfs/nfs4state.c2
-rw-r--r--fs/nfs/nfs4super.c7
-rw-r--r--fs/nfs/nfs4sysctl.c2
-rw-r--r--fs/nfs/nfs4xdr.c22
-rw-r--r--fs/nfs/nfstrace.c3
-rw-r--r--fs/nfs/objlayout/objio_osd.c4
-rw-r--r--fs/nfs/pnfs.c68
-rw-r--r--fs/nfs/pnfs.h24
-rw-r--r--fs/nfs/pnfs_dev.c21
-rw-r--r--fs/nfs/pnfs_nfs.c12
-rw-r--r--fs/nfs/read.c2
-rw-r--r--fs/nfs/super.c4
-rw-r--r--fs/nfs/write.c15
34 files changed, 313 insertions, 163 deletions
diff --git a/fs/nfs/Makefile b/fs/nfs/Makefile
index 1e987acf20c9..8664417955a2 100644
--- a/fs/nfs/Makefile
+++ b/fs/nfs/Makefile
@@ -22,7 +22,7 @@ nfsv3-$(CONFIG_NFS_V3_ACL) += nfs3acl.o
22obj-$(CONFIG_NFS_V4) += nfsv4.o 22obj-$(CONFIG_NFS_V4) += nfsv4.o
23CFLAGS_nfs4trace.o += -I$(src) 23CFLAGS_nfs4trace.o += -I$(src)
24nfsv4-y := nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o nfs4super.o nfs4file.o \ 24nfsv4-y := nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o nfs4super.o nfs4file.o \
25 delegation.o idmap.o callback.o callback_xdr.o callback_proc.o \ 25 delegation.o nfs4idmap.o callback.o callback_xdr.o callback_proc.o \
26 nfs4namespace.o nfs4getroot.o nfs4client.o nfs4session.o \ 26 nfs4namespace.o nfs4getroot.o nfs4client.o nfs4session.o \
27 dns_resolve.o nfs4trace.o 27 dns_resolve.o nfs4trace.o
28nfsv4-$(CONFIG_NFS_USE_LEGACY_DNS) += cache_lib.o 28nfsv4-$(CONFIG_NFS_USE_LEGACY_DNS) += cache_lib.o
diff --git a/fs/nfs/blocklayout/blocklayout.c b/fs/nfs/blocklayout/blocklayout.c
index 1cac3c175d18..d2554fe140a3 100644
--- a/fs/nfs/blocklayout/blocklayout.c
+++ b/fs/nfs/blocklayout/blocklayout.c
@@ -890,6 +890,7 @@ static struct pnfs_layoutdriver_type blocklayout_type = {
890 .free_deviceid_node = bl_free_deviceid_node, 890 .free_deviceid_node = bl_free_deviceid_node,
891 .pg_read_ops = &bl_pg_read_ops, 891 .pg_read_ops = &bl_pg_read_ops,
892 .pg_write_ops = &bl_pg_write_ops, 892 .pg_write_ops = &bl_pg_write_ops,
893 .sync = pnfs_generic_sync,
893}; 894};
894 895
895static int __init nfs4blocklayout_init(void) 896static int __init nfs4blocklayout_init(void)
diff --git a/fs/nfs/blocklayout/dev.c b/fs/nfs/blocklayout/dev.c
index 5aed4f98df41..e535599a0719 100644
--- a/fs/nfs/blocklayout/dev.c
+++ b/fs/nfs/blocklayout/dev.c
@@ -33,7 +33,7 @@ bl_free_deviceid_node(struct nfs4_deviceid_node *d)
33 container_of(d, struct pnfs_block_dev, node); 33 container_of(d, struct pnfs_block_dev, node);
34 34
35 bl_free_device(dev); 35 bl_free_device(dev);
36 kfree(dev); 36 kfree_rcu(dev, node.rcu);
37} 37}
38 38
39static int 39static int
diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c
index 351be9205bf8..8d129bb7355a 100644
--- a/fs/nfs/callback.c
+++ b/fs/nfs/callback.c
@@ -128,7 +128,7 @@ nfs41_callback_svc(void *vrqstp)
128 if (try_to_freeze()) 128 if (try_to_freeze())
129 continue; 129 continue;
130 130
131 prepare_to_wait(&serv->sv_cb_waitq, &wq, TASK_UNINTERRUPTIBLE); 131 prepare_to_wait(&serv->sv_cb_waitq, &wq, TASK_INTERRUPTIBLE);
132 spin_lock_bh(&serv->sv_cb_lock); 132 spin_lock_bh(&serv->sv_cb_lock);
133 if (!list_empty(&serv->sv_cb_list)) { 133 if (!list_empty(&serv->sv_cb_list)) {
134 req = list_first_entry(&serv->sv_cb_list, 134 req = list_first_entry(&serv->sv_cb_list,
@@ -142,10 +142,10 @@ nfs41_callback_svc(void *vrqstp)
142 error); 142 error);
143 } else { 143 } else {
144 spin_unlock_bh(&serv->sv_cb_lock); 144 spin_unlock_bh(&serv->sv_cb_lock);
145 /* schedule_timeout to game the hung task watchdog */ 145 schedule();
146 schedule_timeout(60 * HZ);
147 finish_wait(&serv->sv_cb_waitq, &wq); 146 finish_wait(&serv->sv_cb_waitq, &wq);
148 } 147 }
148 flush_signals(current);
149 } 149 }
150 return 0; 150 return 0;
151} 151}
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 19874151e95c..892aefff3630 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -31,7 +31,6 @@
31#include <linux/lockd/bind.h> 31#include <linux/lockd/bind.h>
32#include <linux/seq_file.h> 32#include <linux/seq_file.h>
33#include <linux/mount.h> 33#include <linux/mount.h>
34#include <linux/nfs_idmap.h>
35#include <linux/vfs.h> 34#include <linux/vfs.h>
36#include <linux/inet.h> 35#include <linux/inet.h>
37#include <linux/in6.h> 36#include <linux/in6.h>
diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c
index a6ad68865880..029d688a969f 100644
--- a/fs/nfs/delegation.c
+++ b/fs/nfs/delegation.c
@@ -378,7 +378,7 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
378 if (freeme == NULL) 378 if (freeme == NULL)
379 goto out; 379 goto out;
380 } 380 }
381 list_add_rcu(&delegation->super_list, &server->delegations); 381 list_add_tail_rcu(&delegation->super_list, &server->delegations);
382 rcu_assign_pointer(nfsi->delegation, delegation); 382 rcu_assign_pointer(nfsi->delegation, delegation);
383 delegation = NULL; 383 delegation = NULL;
384 384
@@ -514,7 +514,7 @@ void nfs_inode_return_delegation_noreclaim(struct inode *inode)
514 514
515 delegation = nfs_inode_detach_delegation(inode); 515 delegation = nfs_inode_detach_delegation(inode);
516 if (delegation != NULL) 516 if (delegation != NULL)
517 nfs_do_return_delegation(inode, delegation, 0); 517 nfs_do_return_delegation(inode, delegation, 1);
518} 518}
519 519
520/** 520/**
diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 1e51ecd61854..b2c8b31b2be7 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -543,6 +543,9 @@ int nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *en
543 if (scratch == NULL) 543 if (scratch == NULL)
544 return -ENOMEM; 544 return -ENOMEM;
545 545
546 if (buflen == 0)
547 goto out_nopages;
548
546 xdr_init_decode_pages(&stream, &buf, xdr_pages, buflen); 549 xdr_init_decode_pages(&stream, &buf, xdr_pages, buflen);
547 xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE); 550 xdr_set_scratch_buffer(&stream, page_address(scratch), PAGE_SIZE);
548 551
@@ -564,6 +567,7 @@ int nfs_readdir_page_filler(nfs_readdir_descriptor_t *desc, struct nfs_entry *en
564 break; 567 break;
565 } while (!entry->eof); 568 } while (!entry->eof);
566 569
570out_nopages:
567 if (count == 0 || (status == -EBADCOOKIE && entry->eof != 0)) { 571 if (count == 0 || (status == -EBADCOOKIE && entry->eof != 0)) {
568 array = nfs_readdir_get_array(page); 572 array = nfs_readdir_get_array(page);
569 if (!IS_ERR(array)) { 573 if (!IS_ERR(array)) {
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index b2cbc3a6cdd9..38678d9a5cc4 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -129,22 +129,25 @@ nfs_direct_good_bytes(struct nfs_direct_req *dreq, struct nfs_pgio_header *hdr)
129 int i; 129 int i;
130 ssize_t count; 130 ssize_t count;
131 131
132 WARN_ON_ONCE(hdr->pgio_mirror_idx >= dreq->mirror_count); 132 if (dreq->mirror_count == 1) {
133 133 dreq->mirrors[hdr->pgio_mirror_idx].count += hdr->good_bytes;
134 count = dreq->mirrors[hdr->pgio_mirror_idx].count; 134 dreq->count += hdr->good_bytes;
135 if (count + dreq->io_start < hdr->io_start + hdr->good_bytes) { 135 } else {
136 count = hdr->io_start + hdr->good_bytes - dreq->io_start; 136 /* mirrored writes */
137 dreq->mirrors[hdr->pgio_mirror_idx].count = count; 137 count = dreq->mirrors[hdr->pgio_mirror_idx].count;
138 } 138 if (count + dreq->io_start < hdr->io_start + hdr->good_bytes) {
139 139 count = hdr->io_start + hdr->good_bytes - dreq->io_start;
140 /* update the dreq->count by finding the minimum agreed count from all 140 dreq->mirrors[hdr->pgio_mirror_idx].count = count;
141 * mirrors */ 141 }
142 count = dreq->mirrors[0].count; 142 /* update the dreq->count by finding the minimum agreed count from all
143 * mirrors */
144 count = dreq->mirrors[0].count;
143 145
144 for (i = 1; i < dreq->mirror_count; i++) 146 for (i = 1; i < dreq->mirror_count; i++)
145 count = min(count, dreq->mirrors[i].count); 147 count = min(count, dreq->mirrors[i].count);
146 148
147 dreq->count = count; 149 dreq->count = count;
150 }
148} 151}
149 152
150/* 153/*
@@ -258,18 +261,11 @@ ssize_t nfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter, loff_t pos)
258 if (!IS_SWAPFILE(inode)) 261 if (!IS_SWAPFILE(inode))
259 return 0; 262 return 0;
260 263
261#ifndef CONFIG_NFS_SWAP
262 dprintk("NFS: nfs_direct_IO (%pD) off/no(%Ld/%lu) EINVAL\n",
263 iocb->ki_filp, (long long) pos, iter->nr_segs);
264
265 return -EINVAL;
266#else
267 VM_BUG_ON(iov_iter_count(iter) != PAGE_SIZE); 264 VM_BUG_ON(iov_iter_count(iter) != PAGE_SIZE);
268 265
269 if (iov_iter_rw(iter) == READ) 266 if (iov_iter_rw(iter) == READ)
270 return nfs_file_direct_read(iocb, iter, pos); 267 return nfs_file_direct_read(iocb, iter, pos);
271 return nfs_file_direct_write(iocb, iter); 268 return nfs_file_direct_write(iocb, iter);
272#endif /* CONFIG_NFS_SWAP */
273} 269}
274 270
275static void nfs_direct_release_pages(struct page **pages, unsigned int npages) 271static void nfs_direct_release_pages(struct page **pages, unsigned int npages)
@@ -1030,6 +1026,7 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, struct iov_iter *iter)
1030 if (i_size_read(inode) < iocb->ki_pos) 1026 if (i_size_read(inode) < iocb->ki_pos)
1031 i_size_write(inode, iocb->ki_pos); 1027 i_size_write(inode, iocb->ki_pos);
1032 spin_unlock(&inode->i_lock); 1028 spin_unlock(&inode->i_lock);
1029 generic_write_sync(file, pos, result);
1033 } 1030 }
1034 } 1031 }
1035 nfs_direct_req_release(dreq); 1032 nfs_direct_req_release(dreq);
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index c40e4363e746..8b8d83a526ce 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -280,6 +280,7 @@ nfs_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
280 280
281 trace_nfs_fsync_enter(inode); 281 trace_nfs_fsync_enter(inode);
282 282
283 nfs_inode_dio_wait(inode);
283 do { 284 do {
284 ret = filemap_write_and_wait_range(inode->i_mapping, start, end); 285 ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
285 if (ret != 0) 286 if (ret != 0)
@@ -782,7 +783,7 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
782 * Flush all pending writes before doing anything 783 * Flush all pending writes before doing anything
783 * with locks.. 784 * with locks..
784 */ 785 */
785 nfs_sync_mapping(filp->f_mapping); 786 vfs_fsync(filp, 0);
786 787
787 l_ctx = nfs_get_lock_context(nfs_file_open_context(filp)); 788 l_ctx = nfs_get_lock_context(nfs_file_open_context(filp));
788 if (!IS_ERR(l_ctx)) { 789 if (!IS_ERR(l_ctx)) {
diff --git a/fs/nfs/filelayout/filelayout.c b/fs/nfs/filelayout/filelayout.c
index 91e88a7ecef0..a46bf6de9ce4 100644
--- a/fs/nfs/filelayout/filelayout.c
+++ b/fs/nfs/filelayout/filelayout.c
@@ -258,7 +258,8 @@ filelayout_set_layoutcommit(struct nfs_pgio_header *hdr)
258 hdr->res.verf->committed != NFS_DATA_SYNC) 258 hdr->res.verf->committed != NFS_DATA_SYNC)
259 return; 259 return;
260 260
261 pnfs_set_layoutcommit(hdr); 261 pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
262 hdr->mds_offset + hdr->res.count);
262 dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino, 263 dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
263 (unsigned long) NFS_I(hdr->inode)->layout->plh_lwb); 264 (unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
264} 265}
@@ -373,7 +374,7 @@ static int filelayout_commit_done_cb(struct rpc_task *task,
373 } 374 }
374 375
375 if (data->verf.committed == NFS_UNSTABLE) 376 if (data->verf.committed == NFS_UNSTABLE)
376 pnfs_commit_set_layoutcommit(data); 377 pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
377 378
378 return 0; 379 return 0;
379} 380}
@@ -1086,7 +1087,7 @@ filelayout_alloc_deviceid_node(struct nfs_server *server,
1086} 1087}
1087 1088
1088static void 1089static void
1089filelayout_free_deveiceid_node(struct nfs4_deviceid_node *d) 1090filelayout_free_deviceid_node(struct nfs4_deviceid_node *d)
1090{ 1091{
1091 nfs4_fl_free_deviceid(container_of(d, struct nfs4_file_layout_dsaddr, id_node)); 1092 nfs4_fl_free_deviceid(container_of(d, struct nfs4_file_layout_dsaddr, id_node));
1092} 1093}
@@ -1137,7 +1138,8 @@ static struct pnfs_layoutdriver_type filelayout_type = {
1137 .read_pagelist = filelayout_read_pagelist, 1138 .read_pagelist = filelayout_read_pagelist,
1138 .write_pagelist = filelayout_write_pagelist, 1139 .write_pagelist = filelayout_write_pagelist,
1139 .alloc_deviceid_node = filelayout_alloc_deviceid_node, 1140 .alloc_deviceid_node = filelayout_alloc_deviceid_node,
1140 .free_deviceid_node = filelayout_free_deveiceid_node, 1141 .free_deviceid_node = filelayout_free_deviceid_node,
1142 .sync = pnfs_nfs_generic_sync,
1141}; 1143};
1142 1144
1143static int __init nfs4filelayout_init(void) 1145static int __init nfs4filelayout_init(void)
diff --git a/fs/nfs/filelayout/filelayoutdev.c b/fs/nfs/filelayout/filelayoutdev.c
index 4f372e224603..4946ef40ba87 100644
--- a/fs/nfs/filelayout/filelayoutdev.c
+++ b/fs/nfs/filelayout/filelayoutdev.c
@@ -55,7 +55,7 @@ nfs4_fl_free_deviceid(struct nfs4_file_layout_dsaddr *dsaddr)
55 nfs4_pnfs_ds_put(ds); 55 nfs4_pnfs_ds_put(ds);
56 } 56 }
57 kfree(dsaddr->stripe_indices); 57 kfree(dsaddr->stripe_indices);
58 kfree(dsaddr); 58 kfree_rcu(dsaddr, id_node.rcu);
59} 59}
60 60
61/* Decode opaque device data and return the result */ 61/* Decode opaque device data and return the result */
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 315cc68945b9..7d05089e52d6 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -11,10 +11,10 @@
11#include <linux/module.h> 11#include <linux/module.h>
12 12
13#include <linux/sunrpc/metrics.h> 13#include <linux/sunrpc/metrics.h>
14#include <linux/nfs_idmap.h>
15 14
16#include "flexfilelayout.h" 15#include "flexfilelayout.h"
17#include "../nfs4session.h" 16#include "../nfs4session.h"
17#include "../nfs4idmap.h"
18#include "../internal.h" 18#include "../internal.h"
19#include "../delegation.h" 19#include "../delegation.h"
20#include "../nfs4trace.h" 20#include "../nfs4trace.h"
@@ -891,7 +891,8 @@ static int ff_layout_read_done_cb(struct rpc_task *task,
891static void 891static void
892ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr) 892ff_layout_set_layoutcommit(struct nfs_pgio_header *hdr)
893{ 893{
894 pnfs_set_layoutcommit(hdr); 894 pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
895 hdr->mds_offset + hdr->res.count);
895 dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino, 896 dprintk("%s inode %lu pls_end_pos %lu\n", __func__, hdr->inode->i_ino,
896 (unsigned long) NFS_I(hdr->inode)->layout->plh_lwb); 897 (unsigned long) NFS_I(hdr->inode)->layout->plh_lwb);
897} 898}
@@ -1074,7 +1075,7 @@ static int ff_layout_commit_done_cb(struct rpc_task *task,
1074 } 1075 }
1075 1076
1076 if (data->verf.committed == NFS_UNSTABLE) 1077 if (data->verf.committed == NFS_UNSTABLE)
1077 pnfs_commit_set_layoutcommit(data); 1078 pnfs_set_layoutcommit(data->inode, data->lseg, data->lwb);
1078 1079
1079 return 0; 1080 return 0;
1080} 1081}
@@ -1414,7 +1415,7 @@ ff_layout_get_ds_info(struct inode *inode)
1414} 1415}
1415 1416
1416static void 1417static void
1417ff_layout_free_deveiceid_node(struct nfs4_deviceid_node *d) 1418ff_layout_free_deviceid_node(struct nfs4_deviceid_node *d)
1418{ 1419{
1419 nfs4_ff_layout_free_deviceid(container_of(d, struct nfs4_ff_layout_ds, 1420 nfs4_ff_layout_free_deviceid(container_of(d, struct nfs4_ff_layout_ds,
1420 id_node)); 1421 id_node));
@@ -1498,7 +1499,7 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
1498 .pg_read_ops = &ff_layout_pg_read_ops, 1499 .pg_read_ops = &ff_layout_pg_read_ops,
1499 .pg_write_ops = &ff_layout_pg_write_ops, 1500 .pg_write_ops = &ff_layout_pg_write_ops,
1500 .get_ds_info = ff_layout_get_ds_info, 1501 .get_ds_info = ff_layout_get_ds_info,
1501 .free_deviceid_node = ff_layout_free_deveiceid_node, 1502 .free_deviceid_node = ff_layout_free_deviceid_node,
1502 .mark_request_commit = pnfs_layout_mark_request_commit, 1503 .mark_request_commit = pnfs_layout_mark_request_commit,
1503 .clear_request_commit = pnfs_generic_clear_request_commit, 1504 .clear_request_commit = pnfs_generic_clear_request_commit,
1504 .scan_commit_lists = pnfs_generic_scan_commit_lists, 1505 .scan_commit_lists = pnfs_generic_scan_commit_lists,
@@ -1508,6 +1509,7 @@ static struct pnfs_layoutdriver_type flexfilelayout_type = {
1508 .write_pagelist = ff_layout_write_pagelist, 1509 .write_pagelist = ff_layout_write_pagelist,
1509 .alloc_deviceid_node = ff_layout_alloc_deviceid_node, 1510 .alloc_deviceid_node = ff_layout_alloc_deviceid_node,
1510 .encode_layoutreturn = ff_layout_encode_layoutreturn, 1511 .encode_layoutreturn = ff_layout_encode_layoutreturn,
1512 .sync = pnfs_nfs_generic_sync,
1511}; 1513};
1512 1514
1513static int __init nfs4flexfilelayout_init(void) 1515static int __init nfs4flexfilelayout_init(void)
diff --git a/fs/nfs/flexfilelayout/flexfilelayoutdev.c b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
index e2c01f204a95..77a2d026aa12 100644
--- a/fs/nfs/flexfilelayout/flexfilelayoutdev.c
+++ b/fs/nfs/flexfilelayout/flexfilelayoutdev.c
@@ -30,7 +30,7 @@ void nfs4_ff_layout_free_deviceid(struct nfs4_ff_layout_ds *mirror_ds)
30{ 30{
31 nfs4_print_deviceid(&mirror_ds->id_node.deviceid); 31 nfs4_print_deviceid(&mirror_ds->id_node.deviceid);
32 nfs4_pnfs_ds_put(mirror_ds->ds); 32 nfs4_pnfs_ds_put(mirror_ds->ds);
33 kfree(mirror_ds); 33 kfree_rcu(mirror_ds, id_node.rcu);
34} 34}
35 35
36/* Decode opaque device data and construct new_ds using it */ 36/* Decode opaque device data and construct new_ds using it */
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 3689e95da79a..f734562c6d24 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -133,6 +133,13 @@ void nfs_evict_inode(struct inode *inode)
133 nfs_clear_inode(inode); 133 nfs_clear_inode(inode);
134} 134}
135 135
136int nfs_sync_inode(struct inode *inode)
137{
138 nfs_inode_dio_wait(inode);
139 return nfs_wb_all(inode);
140}
141EXPORT_SYMBOL_GPL(nfs_sync_inode);
142
136/** 143/**
137 * nfs_sync_mapping - helper to flush all mmapped dirty data to disk 144 * nfs_sync_mapping - helper to flush all mmapped dirty data to disk
138 */ 145 */
@@ -192,7 +199,6 @@ void nfs_zap_caches(struct inode *inode)
192 nfs_zap_caches_locked(inode); 199 nfs_zap_caches_locked(inode);
193 spin_unlock(&inode->i_lock); 200 spin_unlock(&inode->i_lock);
194} 201}
195EXPORT_SYMBOL_GPL(nfs_zap_caches);
196 202
197void nfs_zap_mapping(struct inode *inode, struct address_space *mapping) 203void nfs_zap_mapping(struct inode *inode, struct address_space *mapping)
198{ 204{
@@ -525,10 +531,8 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
525 trace_nfs_setattr_enter(inode); 531 trace_nfs_setattr_enter(inode);
526 532
527 /* Write all dirty data */ 533 /* Write all dirty data */
528 if (S_ISREG(inode->i_mode)) { 534 if (S_ISREG(inode->i_mode))
529 nfs_inode_dio_wait(inode); 535 nfs_sync_inode(inode);
530 nfs_wb_all(inode);
531 }
532 536
533 fattr = nfs_alloc_fattr(); 537 fattr = nfs_alloc_fattr();
534 if (fattr == NULL) 538 if (fattr == NULL)
@@ -644,8 +648,9 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
644 trace_nfs_getattr_enter(inode); 648 trace_nfs_getattr_enter(inode);
645 /* Flush out writes to the server in order to update c/mtime. */ 649 /* Flush out writes to the server in order to update c/mtime. */
646 if (S_ISREG(inode->i_mode)) { 650 if (S_ISREG(inode->i_mode)) {
647 nfs_inode_dio_wait(inode); 651 mutex_lock(&inode->i_mutex);
648 err = filemap_write_and_wait(inode->i_mapping); 652 err = nfs_sync_inode(inode);
653 mutex_unlock(&inode->i_mutex);
649 if (err) 654 if (err)
650 goto out; 655 goto out;
651 } 656 }
@@ -1588,6 +1593,19 @@ int nfs_post_op_update_inode_force_wcc(struct inode *inode, struct nfs_fattr *fa
1588} 1593}
1589EXPORT_SYMBOL_GPL(nfs_post_op_update_inode_force_wcc); 1594EXPORT_SYMBOL_GPL(nfs_post_op_update_inode_force_wcc);
1590 1595
1596
1597static inline bool nfs_fileid_valid(struct nfs_inode *nfsi,
1598 struct nfs_fattr *fattr)
1599{
1600 bool ret1 = true, ret2 = true;
1601
1602 if (fattr->valid & NFS_ATTR_FATTR_FILEID)
1603 ret1 = (nfsi->fileid == fattr->fileid);
1604 if (fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID)
1605 ret2 = (nfsi->fileid == fattr->mounted_on_fileid);
1606 return ret1 || ret2;
1607}
1608
1591/* 1609/*
1592 * Many nfs protocol calls return the new file attributes after 1610 * Many nfs protocol calls return the new file attributes after
1593 * an operation. Here we update the inode to reflect the state 1611 * an operation. Here we update the inode to reflect the state
@@ -1614,7 +1632,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1614 nfs_display_fhandle_hash(NFS_FH(inode)), 1632 nfs_display_fhandle_hash(NFS_FH(inode)),
1615 atomic_read(&inode->i_count), fattr->valid); 1633 atomic_read(&inode->i_count), fattr->valid);
1616 1634
1617 if ((fattr->valid & NFS_ATTR_FATTR_FILEID) && nfsi->fileid != fattr->fileid) { 1635 if (!nfs_fileid_valid(nfsi, fattr)) {
1618 printk(KERN_ERR "NFS: server %s error: fileid changed\n" 1636 printk(KERN_ERR "NFS: server %s error: fileid changed\n"
1619 "fsid %s: expected fileid 0x%Lx, got 0x%Lx\n", 1637 "fsid %s: expected fileid 0x%Lx, got 0x%Lx\n",
1620 NFS_SERVER(inode)->nfs_client->cl_hostname, 1638 NFS_SERVER(inode)->nfs_client->cl_hostname,
@@ -1819,7 +1837,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
1819struct inode *nfs_alloc_inode(struct super_block *sb) 1837struct inode *nfs_alloc_inode(struct super_block *sb)
1820{ 1838{
1821 struct nfs_inode *nfsi; 1839 struct nfs_inode *nfsi;
1822 nfsi = (struct nfs_inode *)kmem_cache_alloc(nfs_inode_cachep, GFP_KERNEL); 1840 nfsi = kmem_cache_alloc(nfs_inode_cachep, GFP_KERNEL);
1823 if (!nfsi) 1841 if (!nfsi)
1824 return NULL; 1842 return NULL;
1825 nfsi->flags = 0UL; 1843 nfsi->flags = 0UL;
diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
index cb170722769c..3a9e75235f30 100644
--- a/fs/nfs/nfs42proc.c
+++ b/fs/nfs/nfs42proc.c
@@ -36,13 +36,16 @@ static int _nfs42_proc_fallocate(struct rpc_message *msg, struct file *filep,
36 loff_t offset, loff_t len) 36 loff_t offset, loff_t len)
37{ 37{
38 struct inode *inode = file_inode(filep); 38 struct inode *inode = file_inode(filep);
39 struct nfs_server *server = NFS_SERVER(inode);
39 struct nfs42_falloc_args args = { 40 struct nfs42_falloc_args args = {
40 .falloc_fh = NFS_FH(inode), 41 .falloc_fh = NFS_FH(inode),
41 .falloc_offset = offset, 42 .falloc_offset = offset,
42 .falloc_length = len, 43 .falloc_length = len,
44 .falloc_bitmask = server->cache_consistency_bitmask,
45 };
46 struct nfs42_falloc_res res = {
47 .falloc_server = server,
43 }; 48 };
44 struct nfs42_falloc_res res;
45 struct nfs_server *server = NFS_SERVER(inode);
46 int status; 49 int status;
47 50
48 msg->rpc_argp = &args; 51 msg->rpc_argp = &args;
@@ -52,8 +55,17 @@ static int _nfs42_proc_fallocate(struct rpc_message *msg, struct file *filep,
52 if (status) 55 if (status)
53 return status; 56 return status;
54 57
55 return nfs4_call_sync(server->client, server, msg, 58 res.falloc_fattr = nfs_alloc_fattr();
56 &args.seq_args, &res.seq_res, 0); 59 if (!res.falloc_fattr)
60 return -ENOMEM;
61
62 status = nfs4_call_sync(server->client, server, msg,
63 &args.seq_args, &res.seq_res, 0);
64 if (status == 0)
65 status = nfs_post_op_update_inode(inode, res.falloc_fattr);
66
67 kfree(res.falloc_fattr);
68 return status;
57} 69}
58 70
59static int nfs42_proc_fallocate(struct rpc_message *msg, struct file *filep, 71static int nfs42_proc_fallocate(struct rpc_message *msg, struct file *filep,
@@ -84,9 +96,13 @@ int nfs42_proc_allocate(struct file *filep, loff_t offset, loff_t len)
84 if (!nfs_server_capable(inode, NFS_CAP_ALLOCATE)) 96 if (!nfs_server_capable(inode, NFS_CAP_ALLOCATE))
85 return -EOPNOTSUPP; 97 return -EOPNOTSUPP;
86 98
99 mutex_lock(&inode->i_mutex);
100
87 err = nfs42_proc_fallocate(&msg, filep, offset, len); 101 err = nfs42_proc_fallocate(&msg, filep, offset, len);
88 if (err == -EOPNOTSUPP) 102 if (err == -EOPNOTSUPP)
89 NFS_SERVER(inode)->caps &= ~NFS_CAP_ALLOCATE; 103 NFS_SERVER(inode)->caps &= ~NFS_CAP_ALLOCATE;
104
105 mutex_unlock(&inode->i_mutex);
90 return err; 106 return err;
91} 107}
92 108
@@ -101,9 +117,16 @@ int nfs42_proc_deallocate(struct file *filep, loff_t offset, loff_t len)
101 if (!nfs_server_capable(inode, NFS_CAP_DEALLOCATE)) 117 if (!nfs_server_capable(inode, NFS_CAP_DEALLOCATE))
102 return -EOPNOTSUPP; 118 return -EOPNOTSUPP;
103 119
120 nfs_wb_all(inode);
121 mutex_lock(&inode->i_mutex);
122
104 err = nfs42_proc_fallocate(&msg, filep, offset, len); 123 err = nfs42_proc_fallocate(&msg, filep, offset, len);
124 if (err == 0)
125 truncate_pagecache_range(inode, offset, (offset + len) -1);
105 if (err == -EOPNOTSUPP) 126 if (err == -EOPNOTSUPP)
106 NFS_SERVER(inode)->caps &= ~NFS_CAP_DEALLOCATE; 127 NFS_SERVER(inode)->caps &= ~NFS_CAP_DEALLOCATE;
128
129 mutex_unlock(&inode->i_mutex);
107 return err; 130 return err;
108} 131}
109 132
diff --git a/fs/nfs/nfs42xdr.c b/fs/nfs/nfs42xdr.c
index 038a7e1521fa..1a25b27248f2 100644
--- a/fs/nfs/nfs42xdr.c
+++ b/fs/nfs/nfs42xdr.c
@@ -25,16 +25,20 @@
25 25
26#define NFS4_enc_allocate_sz (compound_encode_hdr_maxsz + \ 26#define NFS4_enc_allocate_sz (compound_encode_hdr_maxsz + \
27 encode_putfh_maxsz + \ 27 encode_putfh_maxsz + \
28 encode_allocate_maxsz) 28 encode_allocate_maxsz + \
29 encode_getattr_maxsz)
29#define NFS4_dec_allocate_sz (compound_decode_hdr_maxsz + \ 30#define NFS4_dec_allocate_sz (compound_decode_hdr_maxsz + \
30 decode_putfh_maxsz + \ 31 decode_putfh_maxsz + \
31 decode_allocate_maxsz) 32 decode_allocate_maxsz + \
33 decode_getattr_maxsz)
32#define NFS4_enc_deallocate_sz (compound_encode_hdr_maxsz + \ 34#define NFS4_enc_deallocate_sz (compound_encode_hdr_maxsz + \
33 encode_putfh_maxsz + \ 35 encode_putfh_maxsz + \
34 encode_deallocate_maxsz) 36 encode_deallocate_maxsz + \
37 encode_getattr_maxsz)
35#define NFS4_dec_deallocate_sz (compound_decode_hdr_maxsz + \ 38#define NFS4_dec_deallocate_sz (compound_decode_hdr_maxsz + \
36 decode_putfh_maxsz + \ 39 decode_putfh_maxsz + \
37 decode_deallocate_maxsz) 40 decode_deallocate_maxsz + \
41 decode_getattr_maxsz)
38#define NFS4_enc_seek_sz (compound_encode_hdr_maxsz + \ 42#define NFS4_enc_seek_sz (compound_encode_hdr_maxsz + \
39 encode_putfh_maxsz + \ 43 encode_putfh_maxsz + \
40 encode_seek_maxsz) 44 encode_seek_maxsz)
@@ -92,6 +96,7 @@ static void nfs4_xdr_enc_allocate(struct rpc_rqst *req,
92 encode_sequence(xdr, &args->seq_args, &hdr); 96 encode_sequence(xdr, &args->seq_args, &hdr);
93 encode_putfh(xdr, args->falloc_fh, &hdr); 97 encode_putfh(xdr, args->falloc_fh, &hdr);
94 encode_allocate(xdr, args, &hdr); 98 encode_allocate(xdr, args, &hdr);
99 encode_getfattr(xdr, args->falloc_bitmask, &hdr);
95 encode_nops(&hdr); 100 encode_nops(&hdr);
96} 101}
97 102
@@ -110,6 +115,7 @@ static void nfs4_xdr_enc_deallocate(struct rpc_rqst *req,
110 encode_sequence(xdr, &args->seq_args, &hdr); 115 encode_sequence(xdr, &args->seq_args, &hdr);
111 encode_putfh(xdr, args->falloc_fh, &hdr); 116 encode_putfh(xdr, args->falloc_fh, &hdr);
112 encode_deallocate(xdr, args, &hdr); 117 encode_deallocate(xdr, args, &hdr);
118 encode_getfattr(xdr, args->falloc_bitmask, &hdr);
113 encode_nops(&hdr); 119 encode_nops(&hdr);
114} 120}
115 121
@@ -183,6 +189,9 @@ static int nfs4_xdr_dec_allocate(struct rpc_rqst *rqstp,
183 if (status) 189 if (status)
184 goto out; 190 goto out;
185 status = decode_allocate(xdr, res); 191 status = decode_allocate(xdr, res);
192 if (status)
193 goto out;
194 decode_getfattr(xdr, res->falloc_fattr, res->falloc_server);
186out: 195out:
187 return status; 196 return status;
188} 197}
@@ -207,6 +216,9 @@ static int nfs4_xdr_dec_deallocate(struct rpc_rqst *rqstp,
207 if (status) 216 if (status)
208 goto out; 217 goto out;
209 status = decode_deallocate(xdr, res); 218 status = decode_deallocate(xdr, res);
219 if (status)
220 goto out;
221 decode_getfattr(xdr, res->falloc_fattr, res->falloc_server);
210out: 222out:
211 return status; 223 return status;
212} 224}
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index 51c2dbd1e942..e42be52a8c18 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -4,7 +4,6 @@
4 */ 4 */
5#include <linux/module.h> 5#include <linux/module.h>
6#include <linux/nfs_fs.h> 6#include <linux/nfs_fs.h>
7#include <linux/nfs_idmap.h>
8#include <linux/nfs_mount.h> 7#include <linux/nfs_mount.h>
9#include <linux/sunrpc/addr.h> 8#include <linux/sunrpc/addr.h>
10#include <linux/sunrpc/auth.h> 9#include <linux/sunrpc/auth.h>
@@ -15,6 +14,7 @@
15#include "callback.h" 14#include "callback.h"
16#include "delegation.h" 15#include "delegation.h"
17#include "nfs4session.h" 16#include "nfs4session.h"
17#include "nfs4idmap.h"
18#include "pnfs.h" 18#include "pnfs.h"
19#include "netns.h" 19#include "netns.h"
20 20
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index 619eca34e70f..f58c17b3b480 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -10,6 +10,8 @@
10#include "fscache.h" 10#include "fscache.h"
11#include "pnfs.h" 11#include "pnfs.h"
12 12
13#include "nfstrace.h"
14
13#ifdef CONFIG_NFS_V4_2 15#ifdef CONFIG_NFS_V4_2
14#include "nfs42.h" 16#include "nfs42.h"
15#endif 17#endif
@@ -57,7 +59,7 @@ nfs4_file_open(struct inode *inode, struct file *filp)
57 if (openflags & O_TRUNC) { 59 if (openflags & O_TRUNC) {
58 attr.ia_valid |= ATTR_SIZE; 60 attr.ia_valid |= ATTR_SIZE;
59 attr.ia_size = 0; 61 attr.ia_size = 0;
60 nfs_wb_all(inode); 62 nfs_sync_inode(inode);
61 } 63 }
62 64
63 inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, &opened); 65 inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, &opened);
@@ -100,6 +102,9 @@ nfs4_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
100 int ret; 102 int ret;
101 struct inode *inode = file_inode(file); 103 struct inode *inode = file_inode(file);
102 104
105 trace_nfs_fsync_enter(inode);
106
107 nfs_inode_dio_wait(inode);
103 do { 108 do {
104 ret = filemap_write_and_wait_range(inode->i_mapping, start, end); 109 ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
105 if (ret != 0) 110 if (ret != 0)
@@ -107,7 +112,7 @@ nfs4_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
107 mutex_lock(&inode->i_mutex); 112 mutex_lock(&inode->i_mutex);
108 ret = nfs_file_fsync_commit(file, start, end, datasync); 113 ret = nfs_file_fsync_commit(file, start, end, datasync);
109 if (!ret) 114 if (!ret)
110 ret = pnfs_layoutcommit_inode(inode, true); 115 ret = pnfs_sync_inode(inode, !!datasync);
111 mutex_unlock(&inode->i_mutex); 116 mutex_unlock(&inode->i_mutex);
112 /* 117 /*
113 * If nfs_file_fsync_commit detected a server reboot, then 118 * If nfs_file_fsync_commit detected a server reboot, then
@@ -118,6 +123,7 @@ nfs4_file_fsync(struct file *file, loff_t start, loff_t end, int datasync)
118 end = LLONG_MAX; 123 end = LLONG_MAX;
119 } while (ret == -EAGAIN); 124 } while (ret == -EAGAIN);
120 125
126 trace_nfs_fsync_exit(inode, ret);
121 return ret; 127 return ret;
122} 128}
123 129
@@ -152,15 +158,9 @@ static long nfs42_fallocate(struct file *filep, int mode, loff_t offset, loff_t
152 if (ret < 0) 158 if (ret < 0)
153 return ret; 159 return ret;
154 160
155 mutex_lock(&inode->i_mutex);
156 if (mode & FALLOC_FL_PUNCH_HOLE) 161 if (mode & FALLOC_FL_PUNCH_HOLE)
157 ret = nfs42_proc_deallocate(filep, offset, len); 162 return nfs42_proc_deallocate(filep, offset, len);
158 else 163 return nfs42_proc_allocate(filep, offset, len);
159 ret = nfs42_proc_allocate(filep, offset, len);
160 mutex_unlock(&inode->i_mutex);
161
162 nfs_zap_caches(inode);
163 return ret;
164} 164}
165#endif /* CONFIG_NFS_V4_2 */ 165#endif /* CONFIG_NFS_V4_2 */
166 166
diff --git a/fs/nfs/idmap.c b/fs/nfs/nfs4idmap.c
index 857e2a99acc8..2e1737c40a29 100644
--- a/fs/nfs/idmap.c
+++ b/fs/nfs/nfs4idmap.c
@@ -36,7 +36,6 @@
36#include <linux/types.h> 36#include <linux/types.h>
37#include <linux/parser.h> 37#include <linux/parser.h>
38#include <linux/fs.h> 38#include <linux/fs.h>
39#include <linux/nfs_idmap.h>
40#include <net/net_namespace.h> 39#include <net/net_namespace.h>
41#include <linux/sunrpc/rpc_pipe_fs.h> 40#include <linux/sunrpc/rpc_pipe_fs.h>
42#include <linux/nfs_fs.h> 41#include <linux/nfs_fs.h>
@@ -49,6 +48,7 @@
49 48
50#include "internal.h" 49#include "internal.h"
51#include "netns.h" 50#include "netns.h"
51#include "nfs4idmap.h"
52#include "nfs4trace.h" 52#include "nfs4trace.h"
53 53
54#define NFS_UINT_MAXLEN 11 54#define NFS_UINT_MAXLEN 11
diff --git a/fs/nfs/nfs4idmap.h b/fs/nfs/nfs4idmap.h
new file mode 100644
index 000000000000..de44d7330ab3
--- /dev/null
+++ b/fs/nfs/nfs4idmap.h
@@ -0,0 +1,68 @@
1/*
2 * fs/nfs/nfs4idmap.h
3 *
4 * UID and GID to name mapping for clients.
5 *
6 * Copyright (c) 2002 The Regents of the University of Michigan.
7 * All rights reserved.
8 *
9 * Marius Aamodt Eriksen <marius@umich.edu>
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 *
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its
21 * contributors may be used to endorse or promote products derived
22 * from this software without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
25 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
26 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
27 * DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
31 * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
32 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
33 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
34 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
35 */
36#ifndef NFS_IDMAP_H
37#define NFS_IDMAP_H
38
39#include <linux/uidgid.h>
40#include <uapi/linux/nfs_idmap.h>
41
42
43/* Forward declaration to make this header independent of others */
44struct nfs_client;
45struct nfs_server;
46struct nfs_fattr;
47struct nfs4_string;
48
49int nfs_idmap_init(void);
50void nfs_idmap_quit(void);
51int nfs_idmap_new(struct nfs_client *);
52void nfs_idmap_delete(struct nfs_client *);
53
54void nfs_fattr_init_names(struct nfs_fattr *fattr,
55 struct nfs4_string *owner_name,
56 struct nfs4_string *group_name);
57void nfs_fattr_free_names(struct nfs_fattr *);
58void nfs_fattr_map_and_free_names(struct nfs_server *, struct nfs_fattr *);
59
60int nfs_map_name_to_uid(const struct nfs_server *, const char *, size_t, kuid_t *);
61int nfs_map_group_to_gid(const struct nfs_server *, const char *, size_t, kgid_t *);
62int nfs_map_uid_to_name(const struct nfs_server *, kuid_t, char *, size_t);
63int nfs_map_gid_to_group(const struct nfs_server *, kgid_t, char *, size_t);
64
65int nfs_map_string_to_numeric(const char *name, size_t namelen, __u32 *res);
66
67extern unsigned int nfs_idmap_cache_timeout;
68#endif /* NFS_IDMAP_H */
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 98e533f2c94a..45b35b9b1e36 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -51,7 +51,6 @@
51#include <linux/namei.h> 51#include <linux/namei.h>
52#include <linux/mount.h> 52#include <linux/mount.h>
53#include <linux/module.h> 53#include <linux/module.h>
54#include <linux/nfs_idmap.h>
55#include <linux/xattr.h> 54#include <linux/xattr.h>
56#include <linux/utsname.h> 55#include <linux/utsname.h>
57#include <linux/freezer.h> 56#include <linux/freezer.h>
@@ -63,6 +62,7 @@
63#include "callback.h" 62#include "callback.h"
64#include "pnfs.h" 63#include "pnfs.h"
65#include "netns.h" 64#include "netns.h"
65#include "nfs4idmap.h"
66#include "nfs4session.h" 66#include "nfs4session.h"
67#include "fscache.h" 67#include "fscache.h"
68 68
@@ -185,7 +185,8 @@ const u32 nfs4_fattr_bitmap[3] = {
185 | FATTR4_WORD1_SPACE_USED 185 | FATTR4_WORD1_SPACE_USED
186 | FATTR4_WORD1_TIME_ACCESS 186 | FATTR4_WORD1_TIME_ACCESS
187 | FATTR4_WORD1_TIME_METADATA 187 | FATTR4_WORD1_TIME_METADATA
188 | FATTR4_WORD1_TIME_MODIFY, 188 | FATTR4_WORD1_TIME_MODIFY
189 | FATTR4_WORD1_MOUNTED_ON_FILEID,
189#ifdef CONFIG_NFS_V4_SECURITY_LABEL 190#ifdef CONFIG_NFS_V4_SECURITY_LABEL
190 FATTR4_WORD2_SECURITY_LABEL 191 FATTR4_WORD2_SECURITY_LABEL
191#endif 192#endif
@@ -3095,16 +3096,13 @@ int nfs4_proc_get_rootfh(struct nfs_server *server, struct nfs_fh *fhandle,
3095 struct nfs_fsinfo *info, 3096 struct nfs_fsinfo *info,
3096 bool auth_probe) 3097 bool auth_probe)
3097{ 3098{
3098 int status; 3099 int status = 0;
3099 3100
3100 switch (auth_probe) { 3101 if (!auth_probe)
3101 case false:
3102 status = nfs4_lookup_root(server, fhandle, info); 3102 status = nfs4_lookup_root(server, fhandle, info);
3103 if (status != -NFS4ERR_WRONGSEC) 3103
3104 break; 3104 if (auth_probe || status == NFS4ERR_WRONGSEC)
3105 default:
3106 status = nfs4_do_find_root_sec(server, fhandle, info); 3105 status = nfs4_do_find_root_sec(server, fhandle, info);
3107 }
3108 3106
3109 if (status == 0) 3107 if (status == 0)
3110 status = nfs4_server_capabilities(server, fhandle); 3108 status = nfs4_server_capabilities(server, fhandle);
@@ -7944,6 +7942,8 @@ _nfs4_proc_getdeviceinfo(struct nfs_server *server,
7944{ 7942{
7945 struct nfs4_getdeviceinfo_args args = { 7943 struct nfs4_getdeviceinfo_args args = {
7946 .pdev = pdev, 7944 .pdev = pdev,
7945 .notify_types = NOTIFY_DEVICEID4_CHANGE |
7946 NOTIFY_DEVICEID4_DELETE,
7947 }; 7947 };
7948 struct nfs4_getdeviceinfo_res res = { 7948 struct nfs4_getdeviceinfo_res res = {
7949 .pdev = pdev, 7949 .pdev = pdev,
@@ -7958,6 +7958,11 @@ _nfs4_proc_getdeviceinfo(struct nfs_server *server,
7958 7958
7959 dprintk("--> %s\n", __func__); 7959 dprintk("--> %s\n", __func__);
7960 status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0); 7960 status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
7961 if (res.notification & ~args.notify_types)
7962 dprintk("%s: unsupported notification\n", __func__);
7963 if (res.notification != args.notify_types)
7964 pdev->nocache = 1;
7965
7961 dprintk("<-- %s status=%d\n", __func__, status); 7966 dprintk("<-- %s status=%d\n", __func__, status);
7962 7967
7963 return status; 7968 return status;
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 3b2b20534a3a..2782cfca2265 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -42,7 +42,6 @@
42#include <linux/slab.h> 42#include <linux/slab.h>
43#include <linux/fs.h> 43#include <linux/fs.h>
44#include <linux/nfs_fs.h> 44#include <linux/nfs_fs.h>
45#include <linux/nfs_idmap.h>
46#include <linux/kthread.h> 45#include <linux/kthread.h>
47#include <linux/module.h> 46#include <linux/module.h>
48#include <linux/random.h> 47#include <linux/random.h>
@@ -57,6 +56,7 @@
57#include "callback.h" 56#include "callback.h"
58#include "delegation.h" 57#include "delegation.h"
59#include "internal.h" 58#include "internal.h"
59#include "nfs4idmap.h"
60#include "nfs4session.h" 60#include "nfs4session.h"
61#include "pnfs.h" 61#include "pnfs.h"
62#include "netns.h" 62#include "netns.h"
diff --git a/fs/nfs/nfs4super.c b/fs/nfs/nfs4super.c
index 75090feeafad..6fb7cb6b3f4b 100644
--- a/fs/nfs/nfs4super.c
+++ b/fs/nfs/nfs4super.c
@@ -3,12 +3,12 @@
3 */ 3 */
4#include <linux/init.h> 4#include <linux/init.h>
5#include <linux/module.h> 5#include <linux/module.h>
6#include <linux/nfs_idmap.h>
7#include <linux/nfs4_mount.h> 6#include <linux/nfs4_mount.h>
8#include <linux/nfs_fs.h> 7#include <linux/nfs_fs.h>
9#include "delegation.h" 8#include "delegation.h"
10#include "internal.h" 9#include "internal.h"
11#include "nfs4_fs.h" 10#include "nfs4_fs.h"
11#include "nfs4idmap.h"
12#include "dns_resolve.h" 12#include "dns_resolve.h"
13#include "pnfs.h" 13#include "pnfs.h"
14#include "nfs.h" 14#include "nfs.h"
@@ -91,10 +91,11 @@ static void nfs4_evict_inode(struct inode *inode)
91{ 91{
92 truncate_inode_pages_final(&inode->i_data); 92 truncate_inode_pages_final(&inode->i_data);
93 clear_inode(inode); 93 clear_inode(inode);
94 pnfs_return_layout(inode);
95 pnfs_destroy_layout(NFS_I(inode));
96 /* If we are holding a delegation, return it! */ 94 /* If we are holding a delegation, return it! */
97 nfs_inode_return_delegation_noreclaim(inode); 95 nfs_inode_return_delegation_noreclaim(inode);
96 /* Note that above delegreturn would trigger pnfs return-on-close */
97 pnfs_return_layout(inode);
98 pnfs_destroy_layout(NFS_I(inode));
98 /* First call standard NFS clear_inode() code */ 99 /* First call standard NFS clear_inode() code */
99 nfs_clear_inode(inode); 100 nfs_clear_inode(inode);
100} 101}
diff --git a/fs/nfs/nfs4sysctl.c b/fs/nfs/nfs4sysctl.c
index b6ebe7e445f6..0fbd3ab1be22 100644
--- a/fs/nfs/nfs4sysctl.c
+++ b/fs/nfs/nfs4sysctl.c
@@ -6,10 +6,10 @@
6 * Copyright (c) 2006 Trond Myklebust <Trond.Myklebust@netapp.com> 6 * Copyright (c) 2006 Trond Myklebust <Trond.Myklebust@netapp.com>
7 */ 7 */
8#include <linux/sysctl.h> 8#include <linux/sysctl.h>
9#include <linux/nfs_idmap.h>
10#include <linux/nfs_fs.h> 9#include <linux/nfs_fs.h>
11 10
12#include "nfs4_fs.h" 11#include "nfs4_fs.h"
12#include "nfs4idmap.h"
13#include "callback.h" 13#include "callback.h"
14 14
15static const int nfs_set_port_min = 0; 15static const int nfs_set_port_min = 0;
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index 5c399ec41079..0aea97841d30 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -52,10 +52,10 @@
52#include <linux/nfs.h> 52#include <linux/nfs.h>
53#include <linux/nfs4.h> 53#include <linux/nfs4.h>
54#include <linux/nfs_fs.h> 54#include <linux/nfs_fs.h>
55#include <linux/nfs_idmap.h>
56 55
57#include "nfs4_fs.h" 56#include "nfs4_fs.h"
58#include "internal.h" 57#include "internal.h"
58#include "nfs4idmap.h"
59#include "nfs4session.h" 59#include "nfs4session.h"
60#include "pnfs.h" 60#include "pnfs.h"
61#include "netns.h" 61#include "netns.h"
@@ -1920,7 +1920,7 @@ encode_getdeviceinfo(struct xdr_stream *xdr,
1920 1920
1921 p = reserve_space(xdr, 4 + 4); 1921 p = reserve_space(xdr, 4 + 4);
1922 *p++ = cpu_to_be32(1); /* bitmap length */ 1922 *p++ = cpu_to_be32(1); /* bitmap length */
1923 *p++ = cpu_to_be32(NOTIFY_DEVICEID4_CHANGE | NOTIFY_DEVICEID4_DELETE); 1923 *p++ = cpu_to_be32(args->notify_types);
1924} 1924}
1925 1925
1926static void 1926static void
@@ -5753,8 +5753,9 @@ out_overflow:
5753 5753
5754#if defined(CONFIG_NFS_V4_1) 5754#if defined(CONFIG_NFS_V4_1)
5755static int decode_getdeviceinfo(struct xdr_stream *xdr, 5755static int decode_getdeviceinfo(struct xdr_stream *xdr,
5756 struct pnfs_device *pdev) 5756 struct nfs4_getdeviceinfo_res *res)
5757{ 5757{
5758 struct pnfs_device *pdev = res->pdev;
5758 __be32 *p; 5759 __be32 *p;
5759 uint32_t len, type; 5760 uint32_t len, type;
5760 int status; 5761 int status;
@@ -5802,12 +5803,7 @@ static int decode_getdeviceinfo(struct xdr_stream *xdr,
5802 if (unlikely(!p)) 5803 if (unlikely(!p))
5803 goto out_overflow; 5804 goto out_overflow;
5804 5805
5805 if (be32_to_cpup(p++) & 5806 res->notification = be32_to_cpup(p++);
5806 ~(NOTIFY_DEVICEID4_CHANGE | NOTIFY_DEVICEID4_DELETE)) {
5807 dprintk("%s: unsupported notification\n",
5808 __func__);
5809 }
5810
5811 for (i = 1; i < len; i++) { 5807 for (i = 1; i < len; i++) {
5812 if (be32_to_cpup(p++)) { 5808 if (be32_to_cpup(p++)) {
5813 dprintk("%s: unsupported notification\n", 5809 dprintk("%s: unsupported notification\n",
@@ -7061,7 +7057,7 @@ static int nfs4_xdr_dec_getdeviceinfo(struct rpc_rqst *rqstp,
7061 status = decode_sequence(xdr, &res->seq_res, rqstp); 7057 status = decode_sequence(xdr, &res->seq_res, rqstp);
7062 if (status != 0) 7058 if (status != 0)
7063 goto out; 7059 goto out;
7064 status = decode_getdeviceinfo(xdr, res->pdev); 7060 status = decode_getdeviceinfo(xdr, res);
7065out: 7061out:
7066 return status; 7062 return status;
7067} 7063}
@@ -7365,6 +7361,11 @@ nfs4_stat_to_errno(int stat)
7365 .p_name = #proc, \ 7361 .p_name = #proc, \
7366} 7362}
7367 7363
7364#define STUB(proc) \
7365[NFSPROC4_CLNT_##proc] = { \
7366 .p_name = #proc, \
7367}
7368
7368struct rpc_procinfo nfs4_procedures[] = { 7369struct rpc_procinfo nfs4_procedures[] = {
7369 PROC(READ, enc_read, dec_read), 7370 PROC(READ, enc_read, dec_read),
7370 PROC(WRITE, enc_write, dec_write), 7371 PROC(WRITE, enc_write, dec_write),
@@ -7417,6 +7418,7 @@ struct rpc_procinfo nfs4_procedures[] = {
7417 PROC(SECINFO_NO_NAME, enc_secinfo_no_name, dec_secinfo_no_name), 7418 PROC(SECINFO_NO_NAME, enc_secinfo_no_name, dec_secinfo_no_name),
7418 PROC(TEST_STATEID, enc_test_stateid, dec_test_stateid), 7419 PROC(TEST_STATEID, enc_test_stateid, dec_test_stateid),
7419 PROC(FREE_STATEID, enc_free_stateid, dec_free_stateid), 7420 PROC(FREE_STATEID, enc_free_stateid, dec_free_stateid),
7421 STUB(GETDEVICELIST),
7420 PROC(BIND_CONN_TO_SESSION, 7422 PROC(BIND_CONN_TO_SESSION,
7421 enc_bind_conn_to_session, dec_bind_conn_to_session), 7423 enc_bind_conn_to_session, dec_bind_conn_to_session),
7422 PROC(DESTROY_CLIENTID, enc_destroy_clientid, dec_destroy_clientid), 7424 PROC(DESTROY_CLIENTID, enc_destroy_clientid, dec_destroy_clientid),
diff --git a/fs/nfs/nfstrace.c b/fs/nfs/nfstrace.c
index 4eb0aead69b6..c74f7af23d77 100644
--- a/fs/nfs/nfstrace.c
+++ b/fs/nfs/nfstrace.c
@@ -7,3 +7,6 @@
7 7
8#define CREATE_TRACE_POINTS 8#define CREATE_TRACE_POINTS
9#include "nfstrace.h" 9#include "nfstrace.h"
10
11EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_fsync_enter);
12EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_fsync_exit);
diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
index 24e1d7403c0b..5aaed363556a 100644
--- a/fs/nfs/objlayout/objio_osd.c
+++ b/fs/nfs/objlayout/objio_osd.c
@@ -57,7 +57,7 @@ objio_free_deviceid_node(struct nfs4_deviceid_node *d)
57 57
58 dprintk("%s: free od=%p\n", __func__, de->od.od); 58 dprintk("%s: free od=%p\n", __func__, de->od.od);
59 osduld_put_device(de->od.od); 59 osduld_put_device(de->od.od);
60 kfree(de); 60 kfree_rcu(d, rcu);
61} 61}
62 62
63struct objio_segment { 63struct objio_segment {
@@ -637,6 +637,8 @@ static struct pnfs_layoutdriver_type objlayout_type = {
637 .pg_read_ops = &objio_pg_read_ops, 637 .pg_read_ops = &objio_pg_read_ops,
638 .pg_write_ops = &objio_pg_write_ops, 638 .pg_write_ops = &objio_pg_write_ops,
639 639
640 .sync = pnfs_generic_sync,
641
640 .free_deviceid_node = objio_free_deviceid_node, 642 .free_deviceid_node = objio_free_deviceid_node,
641 643
642 .encode_layoutcommit = objlayout_encode_layoutcommit, 644 .encode_layoutcommit = objlayout_encode_layoutcommit,
diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
index 4f802b02fbb9..230606243be6 100644
--- a/fs/nfs/pnfs.c
+++ b/fs/nfs/pnfs.c
@@ -1090,6 +1090,7 @@ bool pnfs_roc(struct inode *ino)
1090 pnfs_get_layout_hdr(lo); /* matched in pnfs_roc_release */ 1090 pnfs_get_layout_hdr(lo); /* matched in pnfs_roc_release */
1091 spin_unlock(&ino->i_lock); 1091 spin_unlock(&ino->i_lock);
1092 pnfs_free_lseg_list(&tmp_list); 1092 pnfs_free_lseg_list(&tmp_list);
1093 pnfs_layoutcommit_inode(ino, true);
1093 return true; 1094 return true;
1094 1095
1095out_noroc: 1096out_noroc:
@@ -1104,8 +1105,10 @@ out_noroc:
1104 } 1105 }
1105 } 1106 }
1106 spin_unlock(&ino->i_lock); 1107 spin_unlock(&ino->i_lock);
1107 if (layoutreturn) 1108 if (layoutreturn) {
1109 pnfs_layoutcommit_inode(ino, true);
1108 pnfs_send_layoutreturn(lo, stateid, IOMODE_ANY, true); 1110 pnfs_send_layoutreturn(lo, stateid, IOMODE_ANY, true);
1111 }
1109 return false; 1112 return false;
1110} 1113}
1111 1114
@@ -1841,7 +1844,8 @@ void pnfs_ld_write_done(struct nfs_pgio_header *hdr)
1841{ 1844{
1842 trace_nfs4_pnfs_write(hdr, hdr->pnfs_error); 1845 trace_nfs4_pnfs_write(hdr, hdr->pnfs_error);
1843 if (!hdr->pnfs_error) { 1846 if (!hdr->pnfs_error) {
1844 pnfs_set_layoutcommit(hdr); 1847 pnfs_set_layoutcommit(hdr->inode, hdr->lseg,
1848 hdr->mds_offset + hdr->res.count);
1845 hdr->mds_ops->rpc_call_done(&hdr->task, hdr); 1849 hdr->mds_ops->rpc_call_done(&hdr->task, hdr);
1846 } else 1850 } else
1847 pnfs_ld_handle_write_error(hdr); 1851 pnfs_ld_handle_write_error(hdr);
@@ -1902,7 +1906,6 @@ static void pnfs_writehdr_free(struct nfs_pgio_header *hdr)
1902 pnfs_put_lseg(hdr->lseg); 1906 pnfs_put_lseg(hdr->lseg);
1903 nfs_pgio_header_free(hdr); 1907 nfs_pgio_header_free(hdr);
1904} 1908}
1905EXPORT_SYMBOL_GPL(pnfs_writehdr_free);
1906 1909
1907int 1910int
1908pnfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc) 1911pnfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc)
@@ -2032,7 +2035,6 @@ static void pnfs_readhdr_free(struct nfs_pgio_header *hdr)
2032 pnfs_put_lseg(hdr->lseg); 2035 pnfs_put_lseg(hdr->lseg);
2033 nfs_pgio_header_free(hdr); 2036 nfs_pgio_header_free(hdr);
2034} 2037}
2035EXPORT_SYMBOL_GPL(pnfs_readhdr_free);
2036 2038
2037int 2039int
2038pnfs_generic_pg_readpages(struct nfs_pageio_descriptor *desc) 2040pnfs_generic_pg_readpages(struct nfs_pageio_descriptor *desc)
@@ -2099,64 +2101,34 @@ void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg)
2099EXPORT_SYMBOL_GPL(pnfs_set_lo_fail); 2101EXPORT_SYMBOL_GPL(pnfs_set_lo_fail);
2100 2102
2101void 2103void
2102pnfs_set_layoutcommit(struct nfs_pgio_header *hdr) 2104pnfs_set_layoutcommit(struct inode *inode, struct pnfs_layout_segment *lseg,
2105 loff_t end_pos)
2103{ 2106{
2104 struct inode *inode = hdr->inode;
2105 struct nfs_inode *nfsi = NFS_I(inode); 2107 struct nfs_inode *nfsi = NFS_I(inode);
2106 loff_t end_pos = hdr->mds_offset + hdr->res.count;
2107 bool mark_as_dirty = false; 2108 bool mark_as_dirty = false;
2108 2109
2109 spin_lock(&inode->i_lock); 2110 spin_lock(&inode->i_lock);
2110 if (!test_and_set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) { 2111 if (!test_and_set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) {
2111 mark_as_dirty = true;
2112 dprintk("%s: Set layoutcommit for inode %lu ",
2113 __func__, inode->i_ino);
2114 }
2115 if (!test_and_set_bit(NFS_LSEG_LAYOUTCOMMIT, &hdr->lseg->pls_flags)) {
2116 /* references matched in nfs4_layoutcommit_release */
2117 pnfs_get_lseg(hdr->lseg);
2118 }
2119 if (end_pos > nfsi->layout->plh_lwb)
2120 nfsi->layout->plh_lwb = end_pos; 2112 nfsi->layout->plh_lwb = end_pos;
2121 spin_unlock(&inode->i_lock);
2122 dprintk("%s: lseg %p end_pos %llu\n",
2123 __func__, hdr->lseg, nfsi->layout->plh_lwb);
2124
2125 /* if pnfs_layoutcommit_inode() runs between inode locks, the next one
2126 * will be a noop because NFS_INO_LAYOUTCOMMIT will not be set */
2127 if (mark_as_dirty)
2128 mark_inode_dirty_sync(inode);
2129}
2130EXPORT_SYMBOL_GPL(pnfs_set_layoutcommit);
2131
2132void pnfs_commit_set_layoutcommit(struct nfs_commit_data *data)
2133{
2134 struct inode *inode = data->inode;
2135 struct nfs_inode *nfsi = NFS_I(inode);
2136 bool mark_as_dirty = false;
2137
2138 spin_lock(&inode->i_lock);
2139 if (!test_and_set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) {
2140 mark_as_dirty = true; 2113 mark_as_dirty = true;
2141 dprintk("%s: Set layoutcommit for inode %lu ", 2114 dprintk("%s: Set layoutcommit for inode %lu ",
2142 __func__, inode->i_ino); 2115 __func__, inode->i_ino);
2143 } 2116 } else if (end_pos > nfsi->layout->plh_lwb)
2144 if (!test_and_set_bit(NFS_LSEG_LAYOUTCOMMIT, &data->lseg->pls_flags)) { 2117 nfsi->layout->plh_lwb = end_pos;
2118 if (!test_and_set_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) {
2145 /* references matched in nfs4_layoutcommit_release */ 2119 /* references matched in nfs4_layoutcommit_release */
2146 pnfs_get_lseg(data->lseg); 2120 pnfs_get_lseg(lseg);
2147 } 2121 }
2148 if (data->lwb > nfsi->layout->plh_lwb)
2149 nfsi->layout->plh_lwb = data->lwb;
2150 spin_unlock(&inode->i_lock); 2122 spin_unlock(&inode->i_lock);
2151 dprintk("%s: lseg %p end_pos %llu\n", 2123 dprintk("%s: lseg %p end_pos %llu\n",
2152 __func__, data->lseg, nfsi->layout->plh_lwb); 2124 __func__, lseg, nfsi->layout->plh_lwb);
2153 2125
2154 /* if pnfs_layoutcommit_inode() runs between inode locks, the next one 2126 /* if pnfs_layoutcommit_inode() runs between inode locks, the next one
2155 * will be a noop because NFS_INO_LAYOUTCOMMIT will not be set */ 2127 * will be a noop because NFS_INO_LAYOUTCOMMIT will not be set */
2156 if (mark_as_dirty) 2128 if (mark_as_dirty)
2157 mark_inode_dirty_sync(inode); 2129 mark_inode_dirty_sync(inode);
2158} 2130}
2159EXPORT_SYMBOL_GPL(pnfs_commit_set_layoutcommit); 2131EXPORT_SYMBOL_GPL(pnfs_set_layoutcommit);
2160 2132
2161void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data) 2133void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data)
2162{ 2134{
@@ -2216,7 +2188,6 @@ pnfs_layoutcommit_inode(struct inode *inode, bool sync)
2216 pnfs_list_write_lseg(inode, &data->lseg_list); 2188 pnfs_list_write_lseg(inode, &data->lseg_list);
2217 2189
2218 end_pos = nfsi->layout->plh_lwb; 2190 end_pos = nfsi->layout->plh_lwb;
2219 nfsi->layout->plh_lwb = 0;
2220 2191
2221 nfs4_stateid_copy(&data->args.stateid, &nfsi->layout->plh_stateid); 2192 nfs4_stateid_copy(&data->args.stateid, &nfsi->layout->plh_stateid);
2222 spin_unlock(&inode->i_lock); 2193 spin_unlock(&inode->i_lock);
@@ -2233,11 +2204,11 @@ pnfs_layoutcommit_inode(struct inode *inode, bool sync)
2233 status = ld->prepare_layoutcommit(&data->args); 2204 status = ld->prepare_layoutcommit(&data->args);
2234 if (status) { 2205 if (status) {
2235 spin_lock(&inode->i_lock); 2206 spin_lock(&inode->i_lock);
2236 if (end_pos < nfsi->layout->plh_lwb) 2207 set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags);
2208 if (end_pos > nfsi->layout->plh_lwb)
2237 nfsi->layout->plh_lwb = end_pos; 2209 nfsi->layout->plh_lwb = end_pos;
2238 spin_unlock(&inode->i_lock); 2210 spin_unlock(&inode->i_lock);
2239 put_rpccred(data->cred); 2211 put_rpccred(data->cred);
2240 set_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags);
2241 goto clear_layoutcommitting; 2212 goto clear_layoutcommitting;
2242 } 2213 }
2243 } 2214 }
@@ -2258,6 +2229,13 @@ clear_layoutcommitting:
2258} 2229}
2259EXPORT_SYMBOL_GPL(pnfs_layoutcommit_inode); 2230EXPORT_SYMBOL_GPL(pnfs_layoutcommit_inode);
2260 2231
2232int
2233pnfs_generic_sync(struct inode *inode, bool datasync)
2234{
2235 return pnfs_layoutcommit_inode(inode, true);
2236}
2237EXPORT_SYMBOL_GPL(pnfs_generic_sync);
2238
2261struct nfs4_threshold *pnfs_mdsthreshold_alloc(void) 2239struct nfs4_threshold *pnfs_mdsthreshold_alloc(void)
2262{ 2240{
2263 struct nfs4_threshold *thp; 2241 struct nfs4_threshold *thp;
diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h
index 084c9144f86d..1e6308f82fc3 100644
--- a/fs/nfs/pnfs.h
+++ b/fs/nfs/pnfs.h
@@ -155,6 +155,8 @@ struct pnfs_layoutdriver_type {
155 int how, 155 int how,
156 struct nfs_commit_info *cinfo); 156 struct nfs_commit_info *cinfo);
157 157
158 int (*sync)(struct inode *inode, bool datasync);
159
158 /* 160 /*
159 * Return PNFS_ATTEMPTED to indicate the layout code has attempted 161 * Return PNFS_ATTEMPTED to indicate the layout code has attempted
160 * I/O, else return PNFS_NOT_ATTEMPTED to fall back to normal NFS 162 * I/O, else return PNFS_NOT_ATTEMPTED to fall back to normal NFS
@@ -203,6 +205,7 @@ struct pnfs_device {
203 struct page **pages; 205 struct page **pages;
204 unsigned int pgbase; 206 unsigned int pgbase;
205 unsigned int pglen; /* reply buffer length */ 207 unsigned int pglen; /* reply buffer length */
208 unsigned char nocache : 1;/* May not be cached */
206}; 209};
207 210
208#define NFS4_PNFS_GETDEVLIST_MAXNUM 16 211#define NFS4_PNFS_GETDEVLIST_MAXNUM 16
@@ -263,10 +266,11 @@ bool pnfs_roc(struct inode *ino);
263void pnfs_roc_release(struct inode *ino); 266void pnfs_roc_release(struct inode *ino);
264void pnfs_roc_set_barrier(struct inode *ino, u32 barrier); 267void pnfs_roc_set_barrier(struct inode *ino, u32 barrier);
265bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task); 268bool pnfs_roc_drain(struct inode *ino, u32 *barrier, struct rpc_task *task);
266void pnfs_set_layoutcommit(struct nfs_pgio_header *); 269void pnfs_set_layoutcommit(struct inode *, struct pnfs_layout_segment *, loff_t);
267void pnfs_commit_set_layoutcommit(struct nfs_commit_data *data);
268void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data); 270void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data);
269int pnfs_layoutcommit_inode(struct inode *inode, bool sync); 271int pnfs_layoutcommit_inode(struct inode *inode, bool sync);
272int pnfs_generic_sync(struct inode *inode, bool datasync);
273int pnfs_nfs_generic_sync(struct inode *inode, bool datasync);
270int _pnfs_return_layout(struct inode *); 274int _pnfs_return_layout(struct inode *);
271int pnfs_commit_and_return_layout(struct inode *); 275int pnfs_commit_and_return_layout(struct inode *);
272void pnfs_ld_write_done(struct nfs_pgio_header *); 276void pnfs_ld_write_done(struct nfs_pgio_header *);
@@ -291,6 +295,7 @@ void pnfs_error_mark_layout_for_return(struct inode *inode,
291enum { 295enum {
292 NFS_DEVICEID_INVALID = 0, /* set when MDS clientid recalled */ 296 NFS_DEVICEID_INVALID = 0, /* set when MDS clientid recalled */
293 NFS_DEVICEID_UNAVAILABLE, /* device temporarily unavailable */ 297 NFS_DEVICEID_UNAVAILABLE, /* device temporarily unavailable */
298 NFS_DEVICEID_NOCACHE, /* device may not be cached */
294}; 299};
295 300
296/* pnfs_dev.c */ 301/* pnfs_dev.c */
@@ -302,6 +307,7 @@ struct nfs4_deviceid_node {
302 unsigned long flags; 307 unsigned long flags;
303 unsigned long timestamp_unavailable; 308 unsigned long timestamp_unavailable;
304 struct nfs4_deviceid deviceid; 309 struct nfs4_deviceid deviceid;
310 struct rcu_head rcu;
305 atomic_t ref; 311 atomic_t ref;
306}; 312};
307 313
@@ -486,6 +492,14 @@ pnfs_ld_read_whole_page(struct inode *inode)
486 return NFS_SERVER(inode)->pnfs_curr_ld->flags & PNFS_READ_WHOLE_PAGE; 492 return NFS_SERVER(inode)->pnfs_curr_ld->flags & PNFS_READ_WHOLE_PAGE;
487} 493}
488 494
495static inline int
496pnfs_sync_inode(struct inode *inode, bool datasync)
497{
498 if (!pnfs_enabled_sb(NFS_SERVER(inode)))
499 return 0;
500 return NFS_SERVER(inode)->pnfs_curr_ld->sync(inode, datasync);
501}
502
489static inline bool 503static inline bool
490pnfs_layoutcommit_outstanding(struct inode *inode) 504pnfs_layoutcommit_outstanding(struct inode *inode)
491{ 505{
@@ -568,6 +582,12 @@ pnfs_ld_read_whole_page(struct inode *inode)
568 return false; 582 return false;
569} 583}
570 584
585static inline int
586pnfs_sync_inode(struct inode *inode, bool datasync)
587{
588 return 0;
589}
590
571static inline bool 591static inline bool
572pnfs_roc(struct inode *ino) 592pnfs_roc(struct inode *ino)
573{ 593{
diff --git a/fs/nfs/pnfs_dev.c b/fs/nfs/pnfs_dev.c
index aa2ec0015183..2961fcd7a2df 100644
--- a/fs/nfs/pnfs_dev.c
+++ b/fs/nfs/pnfs_dev.c
@@ -149,6 +149,8 @@ nfs4_get_device_info(struct nfs_server *server,
149 */ 149 */
150 d = server->pnfs_curr_ld->alloc_deviceid_node(server, pdev, 150 d = server->pnfs_curr_ld->alloc_deviceid_node(server, pdev,
151 gfp_flags); 151 gfp_flags);
152 if (d && pdev->nocache)
153 set_bit(NFS_DEVICEID_NOCACHE, &d->flags);
152 154
153out_free_pages: 155out_free_pages:
154 for (i = 0; i < max_pages; i++) 156 for (i = 0; i < max_pages; i++)
@@ -175,8 +177,8 @@ __nfs4_find_get_deviceid(struct nfs_server *server,
175 rcu_read_lock(); 177 rcu_read_lock();
176 d = _lookup_deviceid(server->pnfs_curr_ld, server->nfs_client, id, 178 d = _lookup_deviceid(server->pnfs_curr_ld, server->nfs_client, id,
177 hash); 179 hash);
178 if (d != NULL) 180 if (d != NULL && !atomic_inc_not_zero(&d->ref))
179 atomic_inc(&d->ref); 181 d = NULL;
180 rcu_read_unlock(); 182 rcu_read_unlock();
181 return d; 183 return d;
182} 184}
@@ -235,12 +237,11 @@ nfs4_delete_deviceid(const struct pnfs_layoutdriver_type *ld,
235 return; 237 return;
236 } 238 }
237 hlist_del_init_rcu(&d->node); 239 hlist_del_init_rcu(&d->node);
240 clear_bit(NFS_DEVICEID_NOCACHE, &d->flags);
238 spin_unlock(&nfs4_deviceid_lock); 241 spin_unlock(&nfs4_deviceid_lock);
239 synchronize_rcu();
240 242
241 /* balance the initial ref set in pnfs_insert_deviceid */ 243 /* balance the initial ref set in pnfs_insert_deviceid */
242 if (atomic_dec_and_test(&d->ref)) 244 nfs4_put_deviceid_node(d);
243 d->ld->free_deviceid_node(d);
244} 245}
245EXPORT_SYMBOL_GPL(nfs4_delete_deviceid); 246EXPORT_SYMBOL_GPL(nfs4_delete_deviceid);
246 247
@@ -271,6 +272,11 @@ EXPORT_SYMBOL_GPL(nfs4_init_deviceid_node);
271bool 272bool
272nfs4_put_deviceid_node(struct nfs4_deviceid_node *d) 273nfs4_put_deviceid_node(struct nfs4_deviceid_node *d)
273{ 274{
275 if (test_bit(NFS_DEVICEID_NOCACHE, &d->flags)) {
276 if (atomic_add_unless(&d->ref, -1, 2))
277 return false;
278 nfs4_delete_deviceid(d->ld, d->nfs_client, &d->deviceid);
279 }
274 if (!atomic_dec_and_test(&d->ref)) 280 if (!atomic_dec_and_test(&d->ref))
275 return false; 281 return false;
276 d->ld->free_deviceid_node(d); 282 d->ld->free_deviceid_node(d);
@@ -314,6 +320,7 @@ _deviceid_purge_client(const struct nfs_client *clp, long hash)
314 if (d->nfs_client == clp && atomic_read(&d->ref)) { 320 if (d->nfs_client == clp && atomic_read(&d->ref)) {
315 hlist_del_init_rcu(&d->node); 321 hlist_del_init_rcu(&d->node);
316 hlist_add_head(&d->tmpnode, &tmp); 322 hlist_add_head(&d->tmpnode, &tmp);
323 clear_bit(NFS_DEVICEID_NOCACHE, &d->flags);
317 } 324 }
318 rcu_read_unlock(); 325 rcu_read_unlock();
319 spin_unlock(&nfs4_deviceid_lock); 326 spin_unlock(&nfs4_deviceid_lock);
@@ -321,12 +328,10 @@ _deviceid_purge_client(const struct nfs_client *clp, long hash)
321 if (hlist_empty(&tmp)) 328 if (hlist_empty(&tmp))
322 return; 329 return;
323 330
324 synchronize_rcu();
325 while (!hlist_empty(&tmp)) { 331 while (!hlist_empty(&tmp)) {
326 d = hlist_entry(tmp.first, struct nfs4_deviceid_node, tmpnode); 332 d = hlist_entry(tmp.first, struct nfs4_deviceid_node, tmpnode);
327 hlist_del(&d->tmpnode); 333 hlist_del(&d->tmpnode);
328 if (atomic_dec_and_test(&d->ref)) 334 nfs4_put_deviceid_node(d);
329 d->ld->free_deviceid_node(d);
330 } 335 }
331} 336}
332 337
diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index 54e36b38fb5f..f37e25b6311c 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -561,7 +561,7 @@ static bool load_v3_ds_connect(void)
561 return(get_v3_ds_connect != NULL); 561 return(get_v3_ds_connect != NULL);
562} 562}
563 563
564void __exit nfs4_pnfs_v3_ds_connect_unload(void) 564void nfs4_pnfs_v3_ds_connect_unload(void)
565{ 565{
566 if (get_v3_ds_connect) { 566 if (get_v3_ds_connect) {
567 symbol_put(nfs3_set_ds_client); 567 symbol_put(nfs3_set_ds_client);
@@ -868,3 +868,13 @@ pnfs_layout_mark_request_commit(struct nfs_page *req,
868 nfs_request_add_commit_list(req, list, cinfo); 868 nfs_request_add_commit_list(req, list, cinfo);
869} 869}
870EXPORT_SYMBOL_GPL(pnfs_layout_mark_request_commit); 870EXPORT_SYMBOL_GPL(pnfs_layout_mark_request_commit);
871
872int
873pnfs_nfs_generic_sync(struct inode *inode, bool datasync)
874{
875 if (datasync)
876 return 0;
877 return pnfs_layoutcommit_inode(inode, true);
878}
879EXPORT_SYMBOL_GPL(pnfs_nfs_generic_sync);
880
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index a5b7427c3754..ae0ff7a11b40 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -284,7 +284,7 @@ int nfs_readpage(struct file *file, struct page *page)
284 dprintk("NFS: nfs_readpage (%p %ld@%lu)\n", 284 dprintk("NFS: nfs_readpage (%p %ld@%lu)\n",
285 page, PAGE_CACHE_SIZE, page_file_index(page)); 285 page, PAGE_CACHE_SIZE, page_file_index(page));
286 nfs_inc_stats(inode, NFSIOS_VFSREADPAGE); 286 nfs_inc_stats(inode, NFSIOS_VFSREADPAGE);
287 nfs_inc_stats(inode, NFSIOS_READPAGES); 287 nfs_add_stats(inode, NFSIOS_READPAGES, 1);
288 288
289 /* 289 /*
290 * Try to flush any pending writes to the file.. 290 * Try to flush any pending writes to the file..
diff --git a/fs/nfs/super.c b/fs/nfs/super.c
index 21f8f52bf37d..f175b833b6ba 100644
--- a/fs/nfs/super.c
+++ b/fs/nfs/super.c
@@ -43,7 +43,6 @@
43#include <linux/seq_file.h> 43#include <linux/seq_file.h>
44#include <linux/mount.h> 44#include <linux/mount.h>
45#include <linux/namei.h> 45#include <linux/namei.h>
46#include <linux/nfs_idmap.h>
47#include <linux/vfs.h> 46#include <linux/vfs.h>
48#include <linux/inet.h> 47#include <linux/inet.h>
49#include <linux/in6.h> 48#include <linux/in6.h>
@@ -2193,7 +2192,7 @@ nfs_compare_remount_data(struct nfs_server *nfss,
2193 data->version != nfss->nfs_client->rpc_ops->version || 2192 data->version != nfss->nfs_client->rpc_ops->version ||
2194 data->minorversion != nfss->nfs_client->cl_minorversion || 2193 data->minorversion != nfss->nfs_client->cl_minorversion ||
2195 data->retrans != nfss->client->cl_timeout->to_retries || 2194 data->retrans != nfss->client->cl_timeout->to_retries ||
2196 data->selected_flavor != nfss->client->cl_auth->au_flavor || 2195 !nfs_auth_info_match(&data->auth_info, nfss->client->cl_auth->au_flavor) ||
2197 data->acregmin != nfss->acregmin / HZ || 2196 data->acregmin != nfss->acregmin / HZ ||
2198 data->acregmax != nfss->acregmax / HZ || 2197 data->acregmax != nfss->acregmax / HZ ||
2199 data->acdirmin != nfss->acdirmin / HZ || 2198 data->acdirmin != nfss->acdirmin / HZ ||
@@ -2241,7 +2240,6 @@ nfs_remount(struct super_block *sb, int *flags, char *raw_data)
2241 data->wsize = nfss->wsize; 2240 data->wsize = nfss->wsize;
2242 data->retrans = nfss->client->cl_timeout->to_retries; 2241 data->retrans = nfss->client->cl_timeout->to_retries;
2243 data->selected_flavor = nfss->client->cl_auth->au_flavor; 2242 data->selected_flavor = nfss->client->cl_auth->au_flavor;
2244 data->auth_info = nfss->auth_info;
2245 data->acregmin = nfss->acregmin / HZ; 2243 data->acregmin = nfss->acregmin / HZ;
2246 data->acregmax = nfss->acregmax / HZ; 2244 data->acregmax = nfss->acregmax / HZ;
2247 data->acdirmin = nfss->acdirmin / HZ; 2245 data->acdirmin = nfss->acdirmin / HZ;
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 3612b4622337..d12a4be613a5 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -580,7 +580,7 @@ static int nfs_do_writepage(struct page *page, struct writeback_control *wbc, st
580 int ret; 580 int ret;
581 581
582 nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE); 582 nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
583 nfs_inc_stats(inode, NFSIOS_WRITEPAGES); 583 nfs_add_stats(inode, NFSIOS_WRITEPAGES, 1);
584 584
585 nfs_pageio_cond_complete(pgio, page_file_index(page)); 585 nfs_pageio_cond_complete(pgio, page_file_index(page));
586 ret = nfs_page_async_flush(pgio, page, wbc->sync_mode == WB_SYNC_NONE); 586 ret = nfs_page_async_flush(pgio, page, wbc->sync_mode == WB_SYNC_NONE);
@@ -1840,17 +1840,16 @@ EXPORT_SYMBOL_GPL(nfs_write_inode);
1840 */ 1840 */
1841int nfs_wb_all(struct inode *inode) 1841int nfs_wb_all(struct inode *inode)
1842{ 1842{
1843 struct writeback_control wbc = {
1844 .sync_mode = WB_SYNC_ALL,
1845 .nr_to_write = LONG_MAX,
1846 .range_start = 0,
1847 .range_end = LLONG_MAX,
1848 };
1849 int ret; 1843 int ret;
1850 1844
1851 trace_nfs_writeback_inode_enter(inode); 1845 trace_nfs_writeback_inode_enter(inode);
1852 1846
1853 ret = sync_inode(inode, &wbc); 1847 ret = filemap_write_and_wait(inode->i_mapping);
1848 if (!ret) {
1849 ret = nfs_commit_inode(inode, FLUSH_SYNC);
1850 if (!ret)
1851 pnfs_sync_inode(inode, true);
1852 }
1854 1853
1855 trace_nfs_writeback_inode_exit(inode, ret); 1854 trace_nfs_writeback_inode_exit(inode, ret);
1856 return ret; 1855 return ret;