aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorJames Bottomley <jejb@mulgrave.il.steeleye.com>2006-03-14 15:18:01 -0500
committerJames Bottomley <jejb@mulgrave.il.steeleye.com>2006-03-14 15:18:01 -0500
commitf33b5d783b4f56be5ace6a1c98fb5f76b2d2d07d (patch)
treeb027b5f3429d416b3da5b9195024007dab062a5e /fs
parente935d5da8e5d12fabe5b632736c50eae0427e8c8 (diff)
parent67963132638e67ad3c5aa16765e6f3f2f3cdd85c (diff)
Merge ../linux-2.6
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/9p.c1
-rw-r--r--fs/9p/fid.c145
-rw-r--r--fs/9p/fid.h6
-rw-r--r--fs/9p/trans_fd.c1
-rw-r--r--fs/9p/v9fs.c1
-rw-r--r--fs/9p/v9fs_vfs.h1
-rw-r--r--fs/9p/vfs_dentry.c45
-rw-r--r--fs/9p/vfs_file.c106
-rw-r--r--fs/9p/vfs_inode.c476
-rw-r--r--fs/9p/vfs_super.c10
-rw-r--r--fs/cifs/cifsproto.h2
-rw-r--r--fs/cifs/connect.c2
-rw-r--r--fs/cifs/misc.c4
-rw-r--r--fs/compat_ioctl.c2
-rw-r--r--fs/cramfs/inode.c60
-rw-r--r--fs/dcache.c2
-rw-r--r--fs/exec.c6
-rw-r--r--fs/ext3/inode.c17
-rw-r--r--fs/ext3/namei.c3
-rw-r--r--fs/fifo.c7
-rw-r--r--fs/file_table.c87
-rw-r--r--fs/fuse/dir.c10
-rw-r--r--fs/jffs2/nodelist.c3
-rw-r--r--fs/jffs2/readinode.c2
-rw-r--r--fs/jffs2/scan.c2
-rw-r--r--fs/lockd/clntproc.c9
-rw-r--r--fs/namei.c13
-rw-r--r--fs/nfs/direct.c10
-rw-r--r--fs/nfs/nfs4proc.c2
-rw-r--r--fs/ocfs2/cluster/masklog.c1
-rw-r--r--fs/ocfs2/cluster/masklog.h2
-rw-r--r--fs/ocfs2/cluster/nodemanager.c4
-rw-r--r--fs/ocfs2/cluster/tcp.c14
-rw-r--r--fs/ocfs2/cluster/tcp.h5
-rw-r--r--fs/ocfs2/dlm/dlmcommon.h8
-rw-r--r--fs/ocfs2/dlm/dlmdebug.c12
-rw-r--r--fs/ocfs2/dlm/dlmdomain.c39
-rw-r--r--fs/ocfs2/dlm/dlmmaster.c4
-rw-r--r--fs/ocfs2/dlm/dlmrecovery.c23
-rw-r--r--fs/ocfs2/extent_map.c38
-rw-r--r--fs/ocfs2/file.c51
-rw-r--r--fs/ocfs2/heartbeat.c1
-rw-r--r--fs/ocfs2/inode.c46
-rw-r--r--fs/ocfs2/journal.c124
-rw-r--r--fs/ocfs2/ocfs2.h7
-rw-r--r--fs/ocfs2/ocfs2_fs.h1
-rw-r--r--fs/ocfs2/super.c11
-rw-r--r--fs/partitions/ibm.c16
-rw-r--r--fs/pipe.c6
-rw-r--r--fs/proc/task_mmu.c11
-rw-r--r--fs/ramfs/inode.c1
-rw-r--r--fs/reiserfs/file.c14
-rw-r--r--fs/reiserfs/inode.c8
-rw-r--r--fs/reiserfs/journal.c3
-rw-r--r--fs/reiserfs/namei.c8
-rw-r--r--fs/udf/inode.c16
-rw-r--r--fs/udf/super.c18
-rw-r--r--fs/udf/udf_sb.h4
-rw-r--r--fs/xfs/linux-2.6/xfs_aops.c2
-rw-r--r--fs/xfs/quota/xfs_qm.c11
-rw-r--r--fs/xfs/xfs_rtalloc.c29
61 files changed, 865 insertions, 708 deletions
diff --git a/fs/9p/9p.c b/fs/9p/9p.c
index 1a6d08761f39..f86a28d1d6a6 100644
--- a/fs/9p/9p.c
+++ b/fs/9p/9p.c
@@ -111,7 +111,6 @@ static void v9fs_t_clunk_cb(void *a, struct v9fs_fcall *tc,
111 if (!rc) 111 if (!rc)
112 return; 112 return;
113 113
114 dprintk(DEBUG_9P, "tcall id %d rcall id %d\n", tc->id, rc->id);
115 v9ses = a; 114 v9ses = a;
116 if (rc->id == RCLUNK) 115 if (rc->id == RCLUNK)
117 v9fs_put_idpool(fid, &v9ses->fidpool); 116 v9fs_put_idpool(fid, &v9ses->fidpool);
diff --git a/fs/9p/fid.c b/fs/9p/fid.c
index eda449778fa5..c4d13bf904d2 100644
--- a/fs/9p/fid.c
+++ b/fs/9p/fid.c
@@ -1,7 +1,7 @@
1/* 1/*
2 * V9FS FID Management 2 * V9FS FID Management
3 * 3 *
4 * Copyright (C) 2005 by Eric Van Hensbergen <ericvh@gmail.com> 4 * Copyright (C) 2005, 2006 by Eric Van Hensbergen <ericvh@gmail.com>
5 * 5 *
6 * This program is free software; you can redistribute it and/or modify 6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by 7 * it under the terms of the GNU General Public License as published by
@@ -40,7 +40,7 @@
40 * 40 *
41 */ 41 */
42 42
43static int v9fs_fid_insert(struct v9fs_fid *fid, struct dentry *dentry) 43int v9fs_fid_insert(struct v9fs_fid *fid, struct dentry *dentry)
44{ 44{
45 struct list_head *fid_list = (struct list_head *)dentry->d_fsdata; 45 struct list_head *fid_list = (struct list_head *)dentry->d_fsdata;
46 dprintk(DEBUG_9P, "fid %d (%p) dentry %s (%p)\n", fid->fid, fid, 46 dprintk(DEBUG_9P, "fid %d (%p) dentry %s (%p)\n", fid->fid, fid,
@@ -57,7 +57,6 @@ static int v9fs_fid_insert(struct v9fs_fid *fid, struct dentry *dentry)
57 } 57 }
58 58
59 fid->uid = current->uid; 59 fid->uid = current->uid;
60 fid->pid = current->pid;
61 list_add(&fid->list, fid_list); 60 list_add(&fid->list, fid_list);
62 return 0; 61 return 0;
63} 62}
@@ -68,14 +67,11 @@ static int v9fs_fid_insert(struct v9fs_fid *fid, struct dentry *dentry)
68 * 67 *
69 */ 68 */
70 69
71struct v9fs_fid *v9fs_fid_create(struct dentry *dentry, 70struct v9fs_fid *v9fs_fid_create(struct v9fs_session_info *v9ses, int fid)
72 struct v9fs_session_info *v9ses, int fid, int create)
73{ 71{
74 struct v9fs_fid *new; 72 struct v9fs_fid *new;
75 73
76 dprintk(DEBUG_9P, "fid create dentry %p, fid %d, create %d\n", 74 dprintk(DEBUG_9P, "fid create fid %d\n", fid);
77 dentry, fid, create);
78
79 new = kmalloc(sizeof(struct v9fs_fid), GFP_KERNEL); 75 new = kmalloc(sizeof(struct v9fs_fid), GFP_KERNEL);
80 if (new == NULL) { 76 if (new == NULL) {
81 dprintk(DEBUG_ERROR, "Out of Memory\n"); 77 dprintk(DEBUG_ERROR, "Out of Memory\n");
@@ -85,19 +81,13 @@ struct v9fs_fid *v9fs_fid_create(struct dentry *dentry,
85 new->fid = fid; 81 new->fid = fid;
86 new->v9ses = v9ses; 82 new->v9ses = v9ses;
87 new->fidopen = 0; 83 new->fidopen = 0;
88 new->fidcreate = create;
89 new->fidclunked = 0; 84 new->fidclunked = 0;
90 new->iounit = 0; 85 new->iounit = 0;
91 new->rdir_pos = 0; 86 new->rdir_pos = 0;
92 new->rdir_fcall = NULL; 87 new->rdir_fcall = NULL;
88 INIT_LIST_HEAD(&new->list);
93 89
94 if (v9fs_fid_insert(new, dentry) == 0) 90 return new;
95 return new;
96 else {
97 dprintk(DEBUG_ERROR, "Problems inserting to dentry\n");
98 kfree(new);
99 return NULL;
100 }
101} 91}
102 92
103/** 93/**
@@ -113,140 +103,29 @@ void v9fs_fid_destroy(struct v9fs_fid *fid)
113} 103}
114 104
115/** 105/**
116 * v9fs_fid_walk_up - walks from the process current directory
117 * up to the specified dentry.
118 */
119static struct v9fs_fid *v9fs_fid_walk_up(struct dentry *dentry)
120{
121 int fidnum, cfidnum, err;
122 struct v9fs_fid *cfid;
123 struct dentry *cde;
124 struct v9fs_session_info *v9ses;
125
126 v9ses = v9fs_inode2v9ses(current->fs->pwd->d_inode);
127 cfid = v9fs_fid_lookup(current->fs->pwd);
128 if (cfid == NULL) {
129 dprintk(DEBUG_ERROR, "process cwd doesn't have a fid\n");
130 return ERR_PTR(-ENOENT);
131 }
132
133 cfidnum = cfid->fid;
134 cde = current->fs->pwd;
135 /* TODO: take advantage of multiwalk */
136
137 fidnum = v9fs_get_idpool(&v9ses->fidpool);
138 if (fidnum < 0) {
139 dprintk(DEBUG_ERROR, "could not get a new fid num\n");
140 err = -ENOENT;
141 goto clunk_fid;
142 }
143
144 while (cde != dentry) {
145 if (cde == cde->d_parent) {
146 dprintk(DEBUG_ERROR, "can't find dentry\n");
147 err = -ENOENT;
148 goto clunk_fid;
149 }
150
151 err = v9fs_t_walk(v9ses, cfidnum, fidnum, "..", NULL);
152 if (err < 0) {
153 dprintk(DEBUG_ERROR, "problem walking to parent\n");
154 goto clunk_fid;
155 }
156
157 cfidnum = fidnum;
158 cde = cde->d_parent;
159 }
160
161 return v9fs_fid_create(dentry, v9ses, fidnum, 0);
162
163clunk_fid:
164 v9fs_t_clunk(v9ses, fidnum);
165 return ERR_PTR(err);
166}
167
168/**
169 * v9fs_fid_lookup - retrieve the right fid from a particular dentry 106 * v9fs_fid_lookup - retrieve the right fid from a particular dentry
170 * @dentry: dentry to look for fid in 107 * @dentry: dentry to look for fid in
171 * @type: intent of lookup (operation or traversal) 108 * @type: intent of lookup (operation or traversal)
172 * 109 *
173 * search list of fids associated with a dentry for a fid with a matching 110 * find a fid in the dentry
174 * thread id or uid. If that fails, look up the dentry's parents to see if you 111 *
175 * can find a matching fid. 112 * TODO: only match fids that have the same uid as current user
176 * 113 *
177 */ 114 */
178 115
179struct v9fs_fid *v9fs_fid_lookup(struct dentry *dentry) 116struct v9fs_fid *v9fs_fid_lookup(struct dentry *dentry)
180{ 117{
181 struct list_head *fid_list = (struct list_head *)dentry->d_fsdata; 118 struct list_head *fid_list = (struct list_head *)dentry->d_fsdata;
182 struct v9fs_fid *current_fid = NULL;
183 struct v9fs_fid *temp = NULL;
184 struct v9fs_fid *return_fid = NULL; 119 struct v9fs_fid *return_fid = NULL;
185 120
186 dprintk(DEBUG_9P, " dentry: %s (%p)\n", dentry->d_iname, dentry); 121 dprintk(DEBUG_9P, " dentry: %s (%p)\n", dentry->d_iname, dentry);
187 122
188 if (fid_list) { 123 if (fid_list)
189 list_for_each_entry_safe(current_fid, temp, fid_list, list) { 124 return_fid = list_entry(fid_list->next, struct v9fs_fid, list);
190 if (!current_fid->fidcreate) {
191 return_fid = current_fid;
192 break;
193 }
194 }
195
196 if (!return_fid)
197 return_fid = current_fid;
198 }
199
200 /* we are at the root but didn't match */
201 if ((!return_fid) && (dentry->d_parent == dentry)) {
202 /* TODO: clone attach with new uid */
203 return_fid = current_fid;
204 }
205 125
206 if (!return_fid) { 126 if (!return_fid) {
207 struct dentry *par = current->fs->pwd->d_parent; 127 dprintk(DEBUG_ERROR, "Couldn't find a fid in dentry\n");
208 int count = 1;
209 while (par != NULL) {
210 if (par == dentry)
211 break;
212 count++;
213 if (par == par->d_parent) {
214 dprintk(DEBUG_ERROR,
215 "got to root without finding dentry\n");
216 break;
217 }
218 par = par->d_parent;
219 }
220
221/* XXX - there may be some duplication we can get rid of */
222 if (par == dentry) {
223 return_fid = v9fs_fid_walk_up(dentry);
224 if (IS_ERR(return_fid))
225 return_fid = NULL;
226 }
227 } 128 }
228 129
229 return return_fid; 130 return return_fid;
230} 131}
231
232struct v9fs_fid *v9fs_fid_get_created(struct dentry *dentry)
233{
234 struct list_head *fid_list;
235 struct v9fs_fid *fid, *ftmp, *ret;
236
237 dprintk(DEBUG_9P, " dentry: %s (%p)\n", dentry->d_iname, dentry);
238 fid_list = (struct list_head *)dentry->d_fsdata;
239 ret = NULL;
240 if (fid_list) {
241 list_for_each_entry_safe(fid, ftmp, fid_list, list) {
242 if (fid->fidcreate && fid->pid == current->pid) {
243 list_del(&fid->list);
244 ret = fid;
245 break;
246 }
247 }
248 }
249
250 dprintk(DEBUG_9P, "return %p\n", ret);
251 return ret;
252}
diff --git a/fs/9p/fid.h b/fs/9p/fid.h
index 84c673a44c83..1fc2dd08d75a 100644
--- a/fs/9p/fid.h
+++ b/fs/9p/fid.h
@@ -33,7 +33,6 @@ struct v9fs_fid {
33 33
34 u32 fid; 34 u32 fid;
35 unsigned char fidopen; /* set when fid is opened */ 35 unsigned char fidopen; /* set when fid is opened */
36 unsigned char fidcreate; /* set when fid was just created */
37 unsigned char fidclunked; /* set when fid has already been clunked */ 36 unsigned char fidclunked; /* set when fid has already been clunked */
38 37
39 struct v9fs_qid qid; 38 struct v9fs_qid qid;
@@ -45,7 +44,6 @@ struct v9fs_fid {
45 struct v9fs_fcall *rdir_fcall; 44 struct v9fs_fcall *rdir_fcall;
46 45
47 /* management stuff */ 46 /* management stuff */
48 pid_t pid; /* thread associated with this fid */
49 uid_t uid; /* user associated with this fid */ 47 uid_t uid; /* user associated with this fid */
50 48
51 /* private data */ 49 /* private data */
@@ -56,5 +54,5 @@ struct v9fs_fid {
56struct v9fs_fid *v9fs_fid_lookup(struct dentry *dentry); 54struct v9fs_fid *v9fs_fid_lookup(struct dentry *dentry);
57struct v9fs_fid *v9fs_fid_get_created(struct dentry *); 55struct v9fs_fid *v9fs_fid_get_created(struct dentry *);
58void v9fs_fid_destroy(struct v9fs_fid *fid); 56void v9fs_fid_destroy(struct v9fs_fid *fid);
59struct v9fs_fid *v9fs_fid_create(struct dentry *, 57struct v9fs_fid *v9fs_fid_create(struct v9fs_session_info *, int fid);
60 struct v9fs_session_info *v9ses, int fid, int create); 58int v9fs_fid_insert(struct v9fs_fid *fid, struct dentry *dentry);
diff --git a/fs/9p/trans_fd.c b/fs/9p/trans_fd.c
index 1a28ef97a3d1..5b2ce21b10fa 100644
--- a/fs/9p/trans_fd.c
+++ b/fs/9p/trans_fd.c
@@ -80,6 +80,7 @@ static int v9fs_fd_send(struct v9fs_transport *trans, void *v, int len)
80 if (!trans || trans->status != Connected || !ts) 80 if (!trans || trans->status != Connected || !ts)
81 return -EIO; 81 return -EIO;
82 82
83 oldfs = get_fs();
83 set_fs(get_ds()); 84 set_fs(get_ds());
84 /* The cast to a user pointer is valid due to the set_fs() */ 85 /* The cast to a user pointer is valid due to the set_fs() */
85 ret = vfs_write(ts->out_file, (void __user *)v, len, &ts->out_file->f_pos); 86 ret = vfs_write(ts->out_file, (void __user *)v, len, &ts->out_file->f_pos);
diff --git a/fs/9p/v9fs.c b/fs/9p/v9fs.c
index ef3386549140..61352491ba36 100644
--- a/fs/9p/v9fs.c
+++ b/fs/9p/v9fs.c
@@ -397,6 +397,7 @@ v9fs_session_init(struct v9fs_session_info *v9ses,
397 } 397 }
398 398
399 if (v9ses->afid != ~0) { 399 if (v9ses->afid != ~0) {
400 dprintk(DEBUG_ERROR, "afid not equal to ~0\n");
400 if (v9fs_t_clunk(v9ses, v9ses->afid)) 401 if (v9fs_t_clunk(v9ses, v9ses->afid))
401 dprintk(DEBUG_ERROR, "clunk failed\n"); 402 dprintk(DEBUG_ERROR, "clunk failed\n");
402 } 403 }
diff --git a/fs/9p/v9fs_vfs.h b/fs/9p/v9fs_vfs.h
index 69cf2905dc90..a759278acaae 100644
--- a/fs/9p/v9fs_vfs.h
+++ b/fs/9p/v9fs_vfs.h
@@ -51,3 +51,4 @@ int v9fs_dir_release(struct inode *inode, struct file *filp);
51int v9fs_file_open(struct inode *inode, struct file *file); 51int v9fs_file_open(struct inode *inode, struct file *file);
52void v9fs_inode2stat(struct inode *inode, struct v9fs_stat *stat); 52void v9fs_inode2stat(struct inode *inode, struct v9fs_stat *stat);
53void v9fs_dentry_release(struct dentry *); 53void v9fs_dentry_release(struct dentry *);
54int v9fs_uflags2omode(int uflags);
diff --git a/fs/9p/vfs_dentry.c b/fs/9p/vfs_dentry.c
index 2dd806dac9f1..12c9cc926b71 100644
--- a/fs/9p/vfs_dentry.c
+++ b/fs/9p/vfs_dentry.c
@@ -43,47 +43,18 @@
43#include "fid.h" 43#include "fid.h"
44 44
45/** 45/**
46 * v9fs_dentry_validate - VFS dcache hook to validate cache 46 * v9fs_dentry_delete - called when dentry refcount equals 0
47 * @dentry: dentry that is being validated 47 * @dentry: dentry in question
48 * @nd: path data
49 * 48 *
50 * dcache really shouldn't be used for 9P2000 as at all due to 49 * By returning 1 here we should remove cacheing of unused
51 * potential attached semantics to directory traversal (walk). 50 * dentry components.
52 *
53 * FUTURE: look into how to use dcache to allow multi-stage
54 * walks in Plan 9 & potential for better dcache operation which
55 * would remain valid for Plan 9 semantics. Older versions
56 * had validation via stat for those interested. However, since
57 * stat has the same approximate overhead as walk there really
58 * is no difference. The only improvement would be from a
59 * time-decay cache like NFS has and that undermines the
60 * synchronous nature of 9P2000.
61 * 51 *
62 */ 52 */
63 53
64static int v9fs_dentry_validate(struct dentry *dentry, struct nameidata *nd) 54int v9fs_dentry_delete(struct dentry *dentry)
65{ 55{
66 struct dentry *dc = current->fs->pwd; 56 dprintk(DEBUG_VFS, " dentry: %s (%p)\n", dentry->d_iname, dentry);
67 57 return 1;
68 dprintk(DEBUG_VFS, "dentry: %s (%p)\n", dentry->d_iname, dentry);
69 if (v9fs_fid_lookup(dentry)) {
70 dprintk(DEBUG_VFS, "VALID\n");
71 return 1;
72 }
73
74 while (dc != NULL) {
75 if (dc == dentry) {
76 dprintk(DEBUG_VFS, "VALID\n");
77 return 1;
78 }
79 if (dc == dc->d_parent)
80 break;
81
82 dc = dc->d_parent;
83 }
84
85 dprintk(DEBUG_VFS, "INVALID\n");
86 return 0;
87} 58}
88 59
89/** 60/**
@@ -118,6 +89,6 @@ void v9fs_dentry_release(struct dentry *dentry)
118} 89}
119 90
120struct dentry_operations v9fs_dentry_operations = { 91struct dentry_operations v9fs_dentry_operations = {
121 .d_revalidate = v9fs_dentry_validate, 92 .d_delete = v9fs_dentry_delete,
122 .d_release = v9fs_dentry_release, 93 .d_release = v9fs_dentry_release,
123}; 94};
diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
index c7e14d917215..de3a129698da 100644
--- a/fs/9p/vfs_file.c
+++ b/fs/9p/vfs_file.c
@@ -53,94 +53,70 @@
53int v9fs_file_open(struct inode *inode, struct file *file) 53int v9fs_file_open(struct inode *inode, struct file *file)
54{ 54{
55 struct v9fs_session_info *v9ses = v9fs_inode2v9ses(inode); 55 struct v9fs_session_info *v9ses = v9fs_inode2v9ses(inode);
56 struct v9fs_fid *v9fid, *fid; 56 struct v9fs_fid *vfid;
57 struct v9fs_fcall *fcall = NULL; 57 struct v9fs_fcall *fcall = NULL;
58 int open_mode = 0; 58 int omode;
59 unsigned int iounit = 0; 59 int fid = V9FS_NOFID;
60 int newfid = -1; 60 int err;
61 long result = -1;
62 61
63 dprintk(DEBUG_VFS, "inode: %p file: %p \n", inode, file); 62 dprintk(DEBUG_VFS, "inode: %p file: %p \n", inode, file);
64 63
65 v9fid = v9fs_fid_get_created(file->f_dentry); 64 vfid = v9fs_fid_lookup(file->f_dentry);
66 if (!v9fid) 65 if (!vfid) {
67 v9fid = v9fs_fid_lookup(file->f_dentry);
68
69 if (!v9fid) {
70 dprintk(DEBUG_ERROR, "Couldn't resolve fid from dentry\n"); 66 dprintk(DEBUG_ERROR, "Couldn't resolve fid from dentry\n");
71 return -EBADF; 67 return -EBADF;
72 } 68 }
73 69
74 if (!v9fid->fidcreate) { 70 fid = v9fs_get_idpool(&v9ses->fidpool);
75 fid = kmalloc(sizeof(struct v9fs_fid), GFP_KERNEL); 71 if (fid < 0) {
76 if (fid == NULL) {
77 dprintk(DEBUG_ERROR, "Out of Memory\n");
78 return -ENOMEM;
79 }
80
81 fid->fidopen = 0;
82 fid->fidcreate = 0;
83 fid->fidclunked = 0;
84 fid->iounit = 0;
85 fid->v9ses = v9ses;
86
87 newfid = v9fs_get_idpool(&v9ses->fidpool);
88 if (newfid < 0) {
89 eprintk(KERN_WARNING, "newfid fails!\n"); 72 eprintk(KERN_WARNING, "newfid fails!\n");
90 return -ENOSPC; 73 return -ENOSPC;
91 } 74 }
92 75
93 result = 76 err = v9fs_t_walk(v9ses, vfid->fid, fid, NULL, NULL);
94 v9fs_t_walk(v9ses, v9fid->fid, newfid, NULL, NULL); 77 if (err < 0) {
95
96 if (result < 0) {
97 v9fs_put_idpool(newfid, &v9ses->fidpool);
98 dprintk(DEBUG_ERROR, "rewalk didn't work\n"); 78 dprintk(DEBUG_ERROR, "rewalk didn't work\n");
99 return -EBADF; 79 goto put_fid;
80 }
81
82 vfid = kmalloc(sizeof(struct v9fs_fid), GFP_KERNEL);
83 if (vfid == NULL) {
84 dprintk(DEBUG_ERROR, "out of memory\n");
85 goto clunk_fid;
100 } 86 }
101 87
102 fid->fid = newfid;
103 v9fid = fid;
104 /* TODO: do special things for O_EXCL, O_NOFOLLOW, O_SYNC */ 88 /* TODO: do special things for O_EXCL, O_NOFOLLOW, O_SYNC */
105 /* translate open mode appropriately */ 89 /* translate open mode appropriately */
106 open_mode = file->f_flags & 0x3; 90 omode = v9fs_uflags2omode(file->f_flags);
91 err = v9fs_t_open(v9ses, fid, omode, &fcall);
92 if (err < 0) {
93 PRINT_FCALL_ERROR("open failed", fcall);
94 goto destroy_vfid;
95 }
107 96
108 if (file->f_flags & O_EXCL) 97 file->private_data = vfid;
109 open_mode |= V9FS_OEXCL; 98 vfid->fid = fid;
99 vfid->fidopen = 1;
100 vfid->fidclunked = 0;
101 vfid->iounit = fcall->params.ropen.iounit;
102 vfid->rdir_pos = 0;
103 vfid->rdir_fcall = NULL;
104 vfid->filp = file;
105 kfree(fcall);
110 106
111 if (v9ses->extended) { 107 return 0;
112 if (file->f_flags & O_TRUNC)
113 open_mode |= V9FS_OTRUNC;
114 108
115 if (file->f_flags & O_APPEND) 109destroy_vfid:
116 open_mode |= V9FS_OAPPEND; 110 v9fs_fid_destroy(vfid);
117 }
118 111
119 result = v9fs_t_open(v9ses, newfid, open_mode, &fcall); 112clunk_fid:
120 if (result < 0) { 113 v9fs_t_clunk(v9ses, fid);
121 PRINT_FCALL_ERROR("open failed", fcall);
122 kfree(fcall);
123 return result;
124 }
125 114
126 iounit = fcall->params.ropen.iounit; 115put_fid:
116 v9fs_put_idpool(fid, &v9ses->fidpool);
127 kfree(fcall); 117 kfree(fcall);
128 } else {
129 /* create case */
130 newfid = v9fid->fid;
131 iounit = v9fid->iounit;
132 v9fid->fidcreate = 0;
133 }
134
135 file->private_data = v9fid;
136
137 v9fid->rdir_pos = 0;
138 v9fid->rdir_fcall = NULL;
139 v9fid->fidopen = 1;
140 v9fid->filp = file;
141 v9fid->iounit = iounit;
142 118
143 return 0; 119 return err;
144} 120}
145 121
146/** 122/**
@@ -289,9 +265,7 @@ v9fs_file_write(struct file *filp, const char __user * data,
289 total += result; 265 total += result;
290 } while (count); 266 } while (count);
291 267
292 if(inode->i_mapping->nrpages)
293 invalidate_inode_pages2(inode->i_mapping); 268 invalidate_inode_pages2(inode->i_mapping);
294
295 return total; 269 return total;
296} 270}
297 271
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 63e5b0398e8b..3ad8455f8577 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -125,6 +125,38 @@ static int p9mode2unixmode(struct v9fs_session_info *v9ses, int mode)
125 return res; 125 return res;
126} 126}
127 127
128int v9fs_uflags2omode(int uflags)
129{
130 int ret;
131
132 ret = 0;
133 switch (uflags&3) {
134 default:
135 case O_RDONLY:
136 ret = V9FS_OREAD;
137 break;
138
139 case O_WRONLY:
140 ret = V9FS_OWRITE;
141 break;
142
143 case O_RDWR:
144 ret = V9FS_ORDWR;
145 break;
146 }
147
148 if (uflags & O_EXCL)
149 ret |= V9FS_OEXCL;
150
151 if (uflags & O_TRUNC)
152 ret |= V9FS_OTRUNC;
153
154 if (uflags & O_APPEND)
155 ret |= V9FS_OAPPEND;
156
157 return ret;
158}
159
128/** 160/**
129 * v9fs_blank_wstat - helper function to setup a 9P stat structure 161 * v9fs_blank_wstat - helper function to setup a 9P stat structure
130 * @v9ses: 9P session info (for determining extended mode) 162 * @v9ses: 9P session info (for determining extended mode)
@@ -163,7 +195,7 @@ v9fs_blank_wstat(struct v9fs_wstat *wstat)
163 195
164struct inode *v9fs_get_inode(struct super_block *sb, int mode) 196struct inode *v9fs_get_inode(struct super_block *sb, int mode)
165{ 197{
166 struct inode *inode = NULL; 198 struct inode *inode;
167 struct v9fs_session_info *v9ses = sb->s_fs_info; 199 struct v9fs_session_info *v9ses = sb->s_fs_info;
168 200
169 dprintk(DEBUG_VFS, "super block: %p mode: %o\n", sb, mode); 201 dprintk(DEBUG_VFS, "super block: %p mode: %o\n", sb, mode);
@@ -222,171 +254,133 @@ struct inode *v9fs_get_inode(struct super_block *sb, int mode)
222 return inode; 254 return inode;
223} 255}
224 256
225/**
226 * v9fs_create - helper function to create files and directories
227 * @dir: directory inode file is being created in
228 * @file_dentry: dentry file is being created in
229 * @perm: permissions file is being created with
230 * @open_mode: resulting open mode for file
231 *
232 */
233
234static int 257static int
235v9fs_create(struct inode *dir, 258v9fs_create(struct v9fs_session_info *v9ses, u32 pfid, char *name,
236 struct dentry *file_dentry, 259 u32 perm, u8 mode, u32 *fidp, struct v9fs_qid *qid, u32 *iounit)
237 unsigned int perm, unsigned int open_mode)
238{ 260{
239 struct v9fs_session_info *v9ses = v9fs_inode2v9ses(dir); 261 u32 fid;
240 struct super_block *sb = dir->i_sb;
241 struct v9fs_fid *dirfid =
242 v9fs_fid_lookup(file_dentry->d_parent);
243 struct v9fs_fid *fid = NULL;
244 struct inode *file_inode = NULL;
245 struct v9fs_fcall *fcall = NULL;
246 struct v9fs_qid qid;
247 int dirfidnum = -1;
248 long newfid = -1;
249 int result = 0;
250 unsigned int iounit = 0;
251 int wfidno = -1;
252 int err; 262 int err;
263 struct v9fs_fcall *fcall;
253 264
254 perm = unixmode2p9mode(v9ses, perm); 265 fid = v9fs_get_idpool(&v9ses->fidpool);
255 266 if (fid < 0) {
256 dprintk(DEBUG_VFS, "dir: %p dentry: %p perm: %o mode: %o\n", dir,
257 file_dentry, perm, open_mode);
258
259 if (!dirfid)
260 return -EBADF;
261
262 dirfidnum = dirfid->fid;
263 if (dirfidnum < 0) {
264 dprintk(DEBUG_ERROR, "No fid for the directory #%lu\n",
265 dir->i_ino);
266 return -EBADF;
267 }
268
269 if (file_dentry->d_inode) {
270 dprintk(DEBUG_ERROR,
271 "Odd. There is an inode for dir %lu, name :%s:\n",
272 dir->i_ino, file_dentry->d_name.name);
273 return -EEXIST;
274 }
275
276 newfid = v9fs_get_idpool(&v9ses->fidpool);
277 if (newfid < 0) {
278 eprintk(KERN_WARNING, "no free fids available\n"); 267 eprintk(KERN_WARNING, "no free fids available\n");
279 return -ENOSPC; 268 return -ENOSPC;
280 } 269 }
281 270
282 result = v9fs_t_walk(v9ses, dirfidnum, newfid, NULL, &fcall); 271 err = v9fs_t_walk(v9ses, pfid, fid, NULL, &fcall);
283 if (result < 0) { 272 if (err < 0) {
284 PRINT_FCALL_ERROR("clone error", fcall); 273 PRINT_FCALL_ERROR("clone error", fcall);
285 v9fs_put_idpool(newfid, &v9ses->fidpool); 274 goto error;
286 newfid = -1;
287 goto CleanUpFid;
288 } 275 }
289
290 kfree(fcall); 276 kfree(fcall);
291 fcall = NULL;
292 277
293 result = v9fs_t_create(v9ses, newfid, (char *)file_dentry->d_name.name, 278 err = v9fs_t_create(v9ses, fid, name, perm, mode, &fcall);
294 perm, open_mode, &fcall); 279 if (err < 0) {
295 if (result < 0) {
296 PRINT_FCALL_ERROR("create fails", fcall); 280 PRINT_FCALL_ERROR("create fails", fcall);
297 goto CleanUpFid; 281 goto error;
298 } 282 }
299 283
300 iounit = fcall->params.rcreate.iounit; 284 if (iounit)
301 qid = fcall->params.rcreate.qid; 285 *iounit = fcall->params.rcreate.iounit;
286
287 if (qid)
288 *qid = fcall->params.rcreate.qid;
289
290 if (fidp)
291 *fidp = fid;
292
302 kfree(fcall); 293 kfree(fcall);
303 fcall = NULL; 294 return 0;
304 295
305 if (!(perm&V9FS_DMDIR)) { 296error:
306 fid = v9fs_fid_create(file_dentry, v9ses, newfid, 1); 297 if (fid >= 0)
307 dprintk(DEBUG_VFS, "fid %p %d\n", fid, fid->fidcreate); 298 v9fs_put_idpool(fid, &v9ses->fidpool);
308 if (!fid) {
309 result = -ENOMEM;
310 goto CleanUpFid;
311 }
312 299
313 fid->qid = qid; 300 kfree(fcall);
314 fid->iounit = iounit; 301 return err;
315 } else { 302}
316 err = v9fs_t_clunk(v9ses, newfid); 303
317 newfid = -1; 304static struct v9fs_fid*
318 if (err < 0) 305v9fs_clone_walk(struct v9fs_session_info *v9ses, u32 fid, struct dentry *dentry)
319 dprintk(DEBUG_ERROR, "clunk for mkdir failed: %d\n", err); 306{
320 } 307 int err;
308 u32 nfid;
309 struct v9fs_fid *ret;
310 struct v9fs_fcall *fcall;
321 311
322 /* walk to the newly created file and put the fid in the dentry */ 312 nfid = v9fs_get_idpool(&v9ses->fidpool);
323 wfidno = v9fs_get_idpool(&v9ses->fidpool); 313 if (nfid < 0) {
324 if (wfidno < 0) {
325 eprintk(KERN_WARNING, "no free fids available\n"); 314 eprintk(KERN_WARNING, "no free fids available\n");
326 return -ENOSPC; 315 return ERR_PTR(-ENOSPC);
327 } 316 }
328 317
329 result = v9fs_t_walk(v9ses, dirfidnum, wfidno, 318 err = v9fs_t_walk(v9ses, fid, nfid, (char *) dentry->d_name.name,
330 (char *) file_dentry->d_name.name, &fcall); 319 &fcall);
331 if (result < 0) { 320
332 PRINT_FCALL_ERROR("clone error", fcall); 321 if (err < 0) {
333 v9fs_put_idpool(wfidno, &v9ses->fidpool); 322 PRINT_FCALL_ERROR("walk error", fcall);
334 wfidno = -1; 323 v9fs_put_idpool(nfid, &v9ses->fidpool);
335 goto CleanUpFid; 324 goto error;
336 } 325 }
326
337 kfree(fcall); 327 kfree(fcall);
338 fcall = NULL; 328 fcall = NULL;
329 ret = v9fs_fid_create(v9ses, nfid);
330 if (!ret) {
331 err = -ENOMEM;
332 goto clunk_fid;
333 }
339 334
340 if (!v9fs_fid_create(file_dentry, v9ses, wfidno, 0)) { 335 err = v9fs_fid_insert(ret, dentry);
341 v9fs_put_idpool(wfidno, &v9ses->fidpool); 336 if (err < 0) {
342 337 v9fs_fid_destroy(ret);
343 goto CleanUpFid; 338 goto clunk_fid;
344 } 339 }
345 340
346 if ((perm & V9FS_DMSYMLINK) || (perm & V9FS_DMLINK) || 341 return ret;
347 (perm & V9FS_DMNAMEDPIPE) || (perm & V9FS_DMSOCKET) ||
348 (perm & V9FS_DMDEVICE))
349 return 0;
350 342
351 result = v9fs_t_stat(v9ses, wfidno, &fcall); 343clunk_fid:
352 if (result < 0) { 344 v9fs_t_clunk(v9ses, nfid);
353 PRINT_FCALL_ERROR("stat error", fcall);
354 goto CleanUpFid;
355 }
356 345
346error:
347 kfree(fcall);
348 return ERR_PTR(err);
349}
357 350
358 file_inode = v9fs_get_inode(sb, 351struct inode *
359 p9mode2unixmode(v9ses, fcall->params.rstat.stat.mode)); 352v9fs_inode_from_fid(struct v9fs_session_info *v9ses, u32 fid,
353 struct super_block *sb)
354{
355 int err, umode;
356 struct inode *ret;
357 struct v9fs_fcall *fcall;
360 358
361 if ((!file_inode) || IS_ERR(file_inode)) { 359 ret = NULL;
362 dprintk(DEBUG_ERROR, "create inode failed\n"); 360 err = v9fs_t_stat(v9ses, fid, &fcall);
363 result = -EBADF; 361 if (err) {
364 goto CleanUpFid; 362 PRINT_FCALL_ERROR("stat error", fcall);
363 goto error;
365 } 364 }
366 365
367 v9fs_stat2inode(&fcall->params.rstat.stat, file_inode, sb); 366 umode = p9mode2unixmode(v9ses, fcall->params.rstat.stat.mode);
368 kfree(fcall); 367 ret = v9fs_get_inode(sb, umode);
369 fcall = NULL; 368 if (IS_ERR(ret)) {
370 file_dentry->d_op = &v9fs_dentry_operations; 369 err = PTR_ERR(ret);
371 d_instantiate(file_dentry, file_inode); 370 ret = NULL;
371 goto error;
372 }
372 373
373 return 0; 374 v9fs_stat2inode(&fcall->params.rstat.stat, ret, sb);
375 kfree(fcall);
376 return ret;
374 377
375 CleanUpFid: 378error:
376 kfree(fcall); 379 kfree(fcall);
377 fcall = NULL; 380 if (ret)
381 iput(ret);
378 382
379 if (newfid >= 0) { 383 return ERR_PTR(err);
380 err = v9fs_t_clunk(v9ses, newfid);
381 if (err < 0)
382 dprintk(DEBUG_ERROR, "clunk failed: %d\n", err);
383 }
384 if (wfidno >= 0) {
385 err = v9fs_t_clunk(v9ses, wfidno);
386 if (err < 0)
387 dprintk(DEBUG_ERROR, "clunk failed: %d\n", err);
388 }
389 return result;
390} 384}
391 385
392/** 386/**
@@ -440,20 +434,97 @@ static int v9fs_remove(struct inode *dir, struct dentry *file, int rmdir)
440 return result; 434 return result;
441} 435}
442 436
437static int
438v9fs_open_created(struct inode *inode, struct file *file)
439{
440 return 0;
441}
442
443/** 443/**
444 * v9fs_vfs_create - VFS hook to create files 444 * v9fs_vfs_create - VFS hook to create files
445 * @inode: directory inode that is being deleted 445 * @inode: directory inode that is being deleted
446 * @dentry: dentry that is being deleted 446 * @dentry: dentry that is being deleted
447 * @perm: create permissions 447 * @mode: create permissions
448 * @nd: path information 448 * @nd: path information
449 * 449 *
450 */ 450 */
451 451
452static int 452static int
453v9fs_vfs_create(struct inode *inode, struct dentry *dentry, int perm, 453v9fs_vfs_create(struct inode *dir, struct dentry *dentry, int mode,
454 struct nameidata *nd) 454 struct nameidata *nd)
455{ 455{
456 return v9fs_create(inode, dentry, perm, O_RDWR); 456 int err;
457 u32 fid, perm, iounit;
458 int flags;
459 struct v9fs_session_info *v9ses;
460 struct v9fs_fid *dfid, *vfid, *ffid;
461 struct inode *inode;
462 struct v9fs_qid qid;
463 struct file *filp;
464
465 inode = NULL;
466 vfid = NULL;
467 v9ses = v9fs_inode2v9ses(dir);
468 dfid = v9fs_fid_lookup(dentry->d_parent);
469 perm = unixmode2p9mode(v9ses, mode);
470
471 if (nd && nd->flags & LOOKUP_OPEN)
472 flags = nd->intent.open.flags - 1;
473 else
474 flags = O_RDWR;
475
476 err = v9fs_create(v9ses, dfid->fid, (char *) dentry->d_name.name,
477 perm, v9fs_uflags2omode(flags), &fid, &qid, &iounit);
478
479 if (err)
480 goto error;
481
482 vfid = v9fs_clone_walk(v9ses, dfid->fid, dentry);
483 if (IS_ERR(vfid)) {
484 err = PTR_ERR(vfid);
485 vfid = NULL;
486 goto error;
487 }
488
489 inode = v9fs_inode_from_fid(v9ses, vfid->fid, dir->i_sb);
490 if (IS_ERR(inode)) {
491 err = PTR_ERR(inode);
492 inode = NULL;
493 goto error;
494 }
495
496 dentry->d_op = &v9fs_dentry_operations;
497 d_instantiate(dentry, inode);
498
499 if (nd && nd->flags & LOOKUP_OPEN) {
500 ffid = v9fs_fid_create(v9ses, fid);
501 if (!ffid)
502 return -ENOMEM;
503
504 filp = lookup_instantiate_filp(nd, dentry, v9fs_open_created);
505 if (IS_ERR(filp)) {
506 v9fs_fid_destroy(ffid);
507 return PTR_ERR(filp);
508 }
509
510 ffid->rdir_pos = 0;
511 ffid->rdir_fcall = NULL;
512 ffid->fidopen = 1;
513 ffid->iounit = iounit;
514 ffid->filp = filp;
515 filp->private_data = ffid;
516 }
517
518 return 0;
519
520error:
521 if (vfid)
522 v9fs_fid_destroy(vfid);
523
524 if (inode)
525 iput(inode);
526
527 return err;
457} 528}
458 529
459/** 530/**
@@ -464,9 +535,57 @@ v9fs_vfs_create(struct inode *inode, struct dentry *dentry, int perm,
464 * 535 *
465 */ 536 */
466 537
467static int v9fs_vfs_mkdir(struct inode *inode, struct dentry *dentry, int mode) 538static int v9fs_vfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
468{ 539{
469 return v9fs_create(inode, dentry, mode | S_IFDIR, O_RDONLY); 540 int err;
541 u32 fid, perm;
542 struct v9fs_session_info *v9ses;
543 struct v9fs_fid *dfid, *vfid;
544 struct inode *inode;
545
546 inode = NULL;
547 vfid = NULL;
548 v9ses = v9fs_inode2v9ses(dir);
549 dfid = v9fs_fid_lookup(dentry->d_parent);
550 perm = unixmode2p9mode(v9ses, mode | S_IFDIR);
551
552 err = v9fs_create(v9ses, dfid->fid, (char *) dentry->d_name.name,
553 perm, V9FS_OREAD, &fid, NULL, NULL);
554
555 if (err) {
556 dprintk(DEBUG_ERROR, "create error %d\n", err);
557 goto error;
558 }
559
560 err = v9fs_t_clunk(v9ses, fid);
561 if (err) {
562 dprintk(DEBUG_ERROR, "clunk error %d\n", err);
563 goto error;
564 }
565
566 vfid = v9fs_clone_walk(v9ses, dfid->fid, dentry);
567 if (IS_ERR(vfid)) {
568 err = PTR_ERR(vfid);
569 vfid = NULL;
570 goto error;
571 }
572
573 inode = v9fs_inode_from_fid(v9ses, vfid->fid, dir->i_sb);
574 if (IS_ERR(inode)) {
575 err = PTR_ERR(inode);
576 inode = NULL;
577 goto error;
578 }
579
580 dentry->d_op = &v9fs_dentry_operations;
581 d_instantiate(dentry, inode);
582 return 0;
583
584error:
585 if (vfid)
586 v9fs_fid_destroy(vfid);
587
588 return err;
470} 589}
471 590
472/** 591/**
@@ -491,7 +610,7 @@ static struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
491 int result = 0; 610 int result = 0;
492 611
493 dprintk(DEBUG_VFS, "dir: %p dentry: (%s) %p nameidata: %p\n", 612 dprintk(DEBUG_VFS, "dir: %p dentry: (%s) %p nameidata: %p\n",
494 dir, dentry->d_iname, dentry, nameidata); 613 dir, dentry->d_name.name, dentry, nameidata);
495 614
496 sb = dir->i_sb; 615 sb = dir->i_sb;
497 v9ses = v9fs_inode2v9ses(dir); 616 v9ses = v9fs_inode2v9ses(dir);
@@ -516,9 +635,8 @@ static struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
516 return ERR_PTR(-ENOSPC); 635 return ERR_PTR(-ENOSPC);
517 } 636 }
518 637
519 result = 638 result = v9fs_t_walk(v9ses, dirfidnum, newfid,
520 v9fs_t_walk(v9ses, dirfidnum, newfid, (char *)dentry->d_name.name, 639 (char *)dentry->d_name.name, NULL);
521 NULL);
522 if (result < 0) { 640 if (result < 0) {
523 v9fs_put_idpool(newfid, &v9ses->fidpool); 641 v9fs_put_idpool(newfid, &v9ses->fidpool);
524 if (result == -ENOENT) { 642 if (result == -ENOENT) {
@@ -551,13 +669,17 @@ static struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
551 669
552 inode->i_ino = v9fs_qid2ino(&fcall->params.rstat.stat.qid); 670 inode->i_ino = v9fs_qid2ino(&fcall->params.rstat.stat.qid);
553 671
554 fid = v9fs_fid_create(dentry, v9ses, newfid, 0); 672 fid = v9fs_fid_create(v9ses, newfid);
555 if (fid == NULL) { 673 if (fid == NULL) {
556 dprintk(DEBUG_ERROR, "couldn't insert\n"); 674 dprintk(DEBUG_ERROR, "couldn't insert\n");
557 result = -ENOMEM; 675 result = -ENOMEM;
558 goto FreeFcall; 676 goto FreeFcall;
559 } 677 }
560 678
679 result = v9fs_fid_insert(fid, dentry);
680 if (result < 0)
681 goto FreeFcall;
682
561 fid->qid = fcall->params.rstat.stat.qid; 683 fid->qid = fcall->params.rstat.stat.qid;
562 684
563 dentry->d_op = &v9fs_dentry_operations; 685 dentry->d_op = &v9fs_dentry_operations;
@@ -886,8 +1008,8 @@ static int v9fs_readlink(struct dentry *dentry, char *buffer, int buflen)
886 } 1008 }
887 1009
888 /* copy extension buffer into buffer */ 1010 /* copy extension buffer into buffer */
889 if (fcall->params.rstat.stat.extension.len+1 < buflen) 1011 if (fcall->params.rstat.stat.extension.len < buflen)
890 buflen = fcall->params.rstat.stat.extension.len + 1; 1012 buflen = fcall->params.rstat.stat.extension.len;
891 1013
892 memcpy(buffer, fcall->params.rstat.stat.extension.str, buflen - 1); 1014 memcpy(buffer, fcall->params.rstat.stat.extension.str, buflen - 1);
893 buffer[buflen-1] = 0; 1015 buffer[buflen-1] = 0;
@@ -951,7 +1073,7 @@ static void *v9fs_vfs_follow_link(struct dentry *dentry, struct nameidata *nd)
951 if (!link) 1073 if (!link)
952 link = ERR_PTR(-ENOMEM); 1074 link = ERR_PTR(-ENOMEM);
953 else { 1075 else {
954 len = v9fs_readlink(dentry, link, PATH_MAX); 1076 len = v9fs_readlink(dentry, link, strlen(link));
955 1077
956 if (len < 0) { 1078 if (len < 0) {
957 __putname(link); 1079 __putname(link);
@@ -983,53 +1105,75 @@ static void v9fs_vfs_put_link(struct dentry *dentry, struct nameidata *nd, void
983static int v9fs_vfs_mkspecial(struct inode *dir, struct dentry *dentry, 1105static int v9fs_vfs_mkspecial(struct inode *dir, struct dentry *dentry,
984 int mode, const char *extension) 1106 int mode, const char *extension)
985{ 1107{
986 int err, retval; 1108 int err;
1109 u32 fid, perm;
987 struct v9fs_session_info *v9ses; 1110 struct v9fs_session_info *v9ses;
1111 struct v9fs_fid *dfid, *vfid;
1112 struct inode *inode;
988 struct v9fs_fcall *fcall; 1113 struct v9fs_fcall *fcall;
989 struct v9fs_fid *fid;
990 struct v9fs_wstat wstat; 1114 struct v9fs_wstat wstat;
991 1115
992 v9ses = v9fs_inode2v9ses(dir);
993 retval = -EPERM;
994 fcall = NULL; 1116 fcall = NULL;
1117 inode = NULL;
1118 vfid = NULL;
1119 v9ses = v9fs_inode2v9ses(dir);
1120 dfid = v9fs_fid_lookup(dentry->d_parent);
1121 perm = unixmode2p9mode(v9ses, mode);
995 1122
996 if (!v9ses->extended) { 1123 if (!v9ses->extended) {
997 dprintk(DEBUG_ERROR, "not extended\n"); 1124 dprintk(DEBUG_ERROR, "not extended\n");
998 goto free_mem; 1125 return -EPERM;
999 } 1126 }
1000 1127
1001 /* issue a create */ 1128 err = v9fs_create(v9ses, dfid->fid, (char *) dentry->d_name.name,
1002 retval = v9fs_create(dir, dentry, mode, 0); 1129 perm, V9FS_OREAD, &fid, NULL, NULL);
1003 if (retval != 0)
1004 goto free_mem;
1005 1130
1006 fid = v9fs_fid_get_created(dentry); 1131 if (err)
1007 if (!fid) { 1132 goto error;
1008 dprintk(DEBUG_ERROR, "couldn't resolve fid from dentry\n"); 1133
1009 goto free_mem; 1134 err = v9fs_t_clunk(v9ses, fid);
1135 if (err)
1136 goto error;
1137
1138 vfid = v9fs_clone_walk(v9ses, dfid->fid, dentry);
1139 if (IS_ERR(vfid)) {
1140 err = PTR_ERR(vfid);
1141 vfid = NULL;
1142 goto error;
1143 }
1144
1145 inode = v9fs_inode_from_fid(v9ses, vfid->fid, dir->i_sb);
1146 if (IS_ERR(inode)) {
1147 err = PTR_ERR(inode);
1148 inode = NULL;
1149 goto error;
1010 } 1150 }
1011 1151
1012 /* issue a Twstat */ 1152 /* issue a Twstat */
1013 v9fs_blank_wstat(&wstat); 1153 v9fs_blank_wstat(&wstat);
1014 wstat.muid = v9ses->name; 1154 wstat.muid = v9ses->name;
1015 wstat.extension = (char *) extension; 1155 wstat.extension = (char *) extension;
1016 retval = v9fs_t_wstat(v9ses, fid->fid, &wstat, &fcall); 1156 err = v9fs_t_wstat(v9ses, vfid->fid, &wstat, &fcall);
1017 if (retval < 0) {
1018 PRINT_FCALL_ERROR("wstat error", fcall);
1019 goto free_mem;
1020 }
1021
1022 err = v9fs_t_clunk(v9ses, fid->fid);
1023 if (err < 0) { 1157 if (err < 0) {
1024 dprintk(DEBUG_ERROR, "clunk failed: %d\n", err); 1158 PRINT_FCALL_ERROR("wstat error", fcall);
1025 goto free_mem; 1159 goto error;
1026 } 1160 }
1027 1161
1028 d_drop(dentry); /* FID - will this also clunk? */ 1162 kfree(fcall);
1163 dentry->d_op = &v9fs_dentry_operations;
1164 d_instantiate(dentry, inode);
1165 return 0;
1029 1166
1030free_mem: 1167error:
1031 kfree(fcall); 1168 kfree(fcall);
1032 return retval; 1169 if (vfid)
1170 v9fs_fid_destroy(vfid);
1171
1172 if (inode)
1173 iput(inode);
1174
1175 return err;
1176
1033} 1177}
1034 1178
1035/** 1179/**
diff --git a/fs/9p/vfs_super.c b/fs/9p/vfs_super.c
index 2c4fa75be025..d05318fa684e 100644
--- a/fs/9p/vfs_super.c
+++ b/fs/9p/vfs_super.c
@@ -146,7 +146,6 @@ static struct super_block *v9fs_get_sb(struct file_system_type
146 inode->i_gid = gid; 146 inode->i_gid = gid;
147 147
148 root = d_alloc_root(inode); 148 root = d_alloc_root(inode);
149
150 if (!root) { 149 if (!root) {
151 retval = -ENOMEM; 150 retval = -ENOMEM;
152 goto put_back_sb; 151 goto put_back_sb;
@@ -158,15 +157,20 @@ static struct super_block *v9fs_get_sb(struct file_system_type
158 if (stat_result < 0) { 157 if (stat_result < 0) {
159 dprintk(DEBUG_ERROR, "stat error\n"); 158 dprintk(DEBUG_ERROR, "stat error\n");
160 v9fs_t_clunk(v9ses, newfid); 159 v9fs_t_clunk(v9ses, newfid);
161 v9fs_put_idpool(newfid, &v9ses->fidpool);
162 } else { 160 } else {
163 /* Setup the Root Inode */ 161 /* Setup the Root Inode */
164 root_fid = v9fs_fid_create(root, v9ses, newfid, 0); 162 root_fid = v9fs_fid_create(v9ses, newfid);
165 if (root_fid == NULL) { 163 if (root_fid == NULL) {
166 retval = -ENOMEM; 164 retval = -ENOMEM;
167 goto put_back_sb; 165 goto put_back_sb;
168 } 166 }
169 167
168 retval = v9fs_fid_insert(root_fid, root);
169 if (retval < 0) {
170 kfree(fcall);
171 goto put_back_sb;
172 }
173
170 root_fid->qid = fcall->params.rstat.stat.qid; 174 root_fid->qid = fcall->params.rstat.stat.qid;
171 root->d_inode->i_ino = 175 root->d_inode->i_ino =
172 v9fs_qid2ino(&fcall->params.rstat.stat.qid); 176 v9fs_qid2ino(&fcall->params.rstat.stat.qid);
diff --git a/fs/cifs/cifsproto.h b/fs/cifs/cifsproto.h
index 3c03aadaff0c..7b25463d3c14 100644
--- a/fs/cifs/cifsproto.h
+++ b/fs/cifs/cifsproto.h
@@ -52,7 +52,7 @@ extern int SendReceive2(const unsigned int /* xid */ , struct cifsSesInfo *,
52 int * /* type of buf returned */ , const int long_op); 52 int * /* type of buf returned */ , const int long_op);
53extern int checkSMBhdr(struct smb_hdr *smb, __u16 mid); 53extern int checkSMBhdr(struct smb_hdr *smb, __u16 mid);
54extern int checkSMB(struct smb_hdr *smb, __u16 mid, int length); 54extern int checkSMB(struct smb_hdr *smb, __u16 mid, int length);
55extern int is_valid_oplock_break(struct smb_hdr *smb); 55extern int is_valid_oplock_break(struct smb_hdr *smb, struct TCP_Server_Info *);
56extern int is_size_safe_to_change(struct cifsInodeInfo *); 56extern int is_size_safe_to_change(struct cifsInodeInfo *);
57extern struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *); 57extern struct cifsFileInfo *find_writable_file(struct cifsInodeInfo *);
58extern unsigned int smbCalcSize(struct smb_hdr *ptr); 58extern unsigned int smbCalcSize(struct smb_hdr *ptr);
diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c
index ef5ae6f93c75..2a0c1f4ca0ae 100644
--- a/fs/cifs/connect.c
+++ b/fs/cifs/connect.c
@@ -630,7 +630,7 @@ multi_t2_fnd:
630 smallbuf = NULL; 630 smallbuf = NULL;
631 } 631 }
632 wake_up_process(task_to_wake); 632 wake_up_process(task_to_wake);
633 } else if ((is_valid_oplock_break(smb_buffer) == FALSE) 633 } else if ((is_valid_oplock_break(smb_buffer, server) == FALSE)
634 && (isMultiRsp == FALSE)) { 634 && (isMultiRsp == FALSE)) {
635 cERROR(1, ("No task to wake, unknown frame rcvd!")); 635 cERROR(1, ("No task to wake, unknown frame rcvd!"));
636 cifs_dump_mem("Received Data is: ",(char *)smb_buffer, 636 cifs_dump_mem("Received Data is: ",(char *)smb_buffer,
diff --git a/fs/cifs/misc.c b/fs/cifs/misc.c
index 812c6bb0fe38..432ba15e2c2d 100644
--- a/fs/cifs/misc.c
+++ b/fs/cifs/misc.c
@@ -475,7 +475,7 @@ checkSMB(struct smb_hdr *smb, __u16 mid, int length)
475 return 0; 475 return 0;
476} 476}
477int 477int
478is_valid_oplock_break(struct smb_hdr *buf) 478is_valid_oplock_break(struct smb_hdr *buf, struct TCP_Server_Info *srv)
479{ 479{
480 struct smb_com_lock_req * pSMB = (struct smb_com_lock_req *)buf; 480 struct smb_com_lock_req * pSMB = (struct smb_com_lock_req *)buf;
481 struct list_head *tmp; 481 struct list_head *tmp;
@@ -535,7 +535,7 @@ is_valid_oplock_break(struct smb_hdr *buf)
535 read_lock(&GlobalSMBSeslock); 535 read_lock(&GlobalSMBSeslock);
536 list_for_each(tmp, &GlobalTreeConnectionList) { 536 list_for_each(tmp, &GlobalTreeConnectionList) {
537 tcon = list_entry(tmp, struct cifsTconInfo, cifsConnectionList); 537 tcon = list_entry(tmp, struct cifsTconInfo, cifsConnectionList);
538 if (tcon->tid == buf->Tid) { 538 if ((tcon->tid == buf->Tid) && (srv == tcon->ses->server)) {
539 cifs_stats_inc(&tcon->num_oplock_brks); 539 cifs_stats_inc(&tcon->num_oplock_brks);
540 list_for_each(tmp1,&tcon->openFileList){ 540 list_for_each(tmp1,&tcon->openFileList){
541 netfile = list_entry(tmp1,struct cifsFileInfo, 541 netfile = list_entry(tmp1,struct cifsFileInfo,
diff --git a/fs/compat_ioctl.c b/fs/compat_ioctl.c
index 537ac70edfe5..c666769a875d 100644
--- a/fs/compat_ioctl.c
+++ b/fs/compat_ioctl.c
@@ -446,7 +446,7 @@ static int dev_ifconf(unsigned int fd, unsigned int cmd, unsigned long arg)
446 ifr = ifc.ifc_req; 446 ifr = ifc.ifc_req;
447 ifr32 = compat_ptr(ifc32.ifcbuf); 447 ifr32 = compat_ptr(ifc32.ifcbuf);
448 for (i = 0, j = 0; 448 for (i = 0, j = 0;
449 i + sizeof (struct ifreq32) < ifc32.ifc_len && j < ifc.ifc_len; 449 i + sizeof (struct ifreq32) <= ifc32.ifc_len && j < ifc.ifc_len;
450 i += sizeof (struct ifreq32), j += sizeof (struct ifreq)) { 450 i += sizeof (struct ifreq32), j += sizeof (struct ifreq)) {
451 if (copy_in_user(ifr32, ifr, sizeof (struct ifreq32))) 451 if (copy_in_user(ifr32, ifr, sizeof (struct ifreq32)))
452 return -EFAULT; 452 return -EFAULT;
diff --git a/fs/cramfs/inode.c b/fs/cramfs/inode.c
index 7fe85415ae7c..8ad52f5bf255 100644
--- a/fs/cramfs/inode.c
+++ b/fs/cramfs/inode.c
@@ -36,7 +36,7 @@ static DECLARE_MUTEX(read_mutex);
36 36
37/* These two macros may change in future, to provide better st_ino 37/* These two macros may change in future, to provide better st_ino
38 semantics. */ 38 semantics. */
39#define CRAMINO(x) ((x)->offset?(x)->offset<<2:1) 39#define CRAMINO(x) (((x)->offset && (x)->size)?(x)->offset<<2:1)
40#define OFFSET(x) ((x)->i_ino) 40#define OFFSET(x) ((x)->i_ino)
41 41
42 42
@@ -66,8 +66,36 @@ static int cramfs_iget5_test(struct inode *inode, void *opaque)
66 66
67static int cramfs_iget5_set(struct inode *inode, void *opaque) 67static int cramfs_iget5_set(struct inode *inode, void *opaque)
68{ 68{
69 static struct timespec zerotime;
69 struct cramfs_inode *cramfs_inode = opaque; 70 struct cramfs_inode *cramfs_inode = opaque;
71 inode->i_mode = cramfs_inode->mode;
72 inode->i_uid = cramfs_inode->uid;
73 inode->i_size = cramfs_inode->size;
74 inode->i_blocks = (cramfs_inode->size - 1) / 512 + 1;
75 inode->i_blksize = PAGE_CACHE_SIZE;
76 inode->i_gid = cramfs_inode->gid;
77 /* Struct copy intentional */
78 inode->i_mtime = inode->i_atime = inode->i_ctime = zerotime;
70 inode->i_ino = CRAMINO(cramfs_inode); 79 inode->i_ino = CRAMINO(cramfs_inode);
80 /* inode->i_nlink is left 1 - arguably wrong for directories,
81 but it's the best we can do without reading the directory
82 contents. 1 yields the right result in GNU find, even
83 without -noleaf option. */
84 if (S_ISREG(inode->i_mode)) {
85 inode->i_fop = &generic_ro_fops;
86 inode->i_data.a_ops = &cramfs_aops;
87 } else if (S_ISDIR(inode->i_mode)) {
88 inode->i_op = &cramfs_dir_inode_operations;
89 inode->i_fop = &cramfs_directory_operations;
90 } else if (S_ISLNK(inode->i_mode)) {
91 inode->i_op = &page_symlink_inode_operations;
92 inode->i_data.a_ops = &cramfs_aops;
93 } else {
94 inode->i_size = 0;
95 inode->i_blocks = 0;
96 init_special_inode(inode, inode->i_mode,
97 old_decode_dev(cramfs_inode->size));
98 }
71 return 0; 99 return 0;
72} 100}
73 101
@@ -77,37 +105,7 @@ static struct inode *get_cramfs_inode(struct super_block *sb,
77 struct inode *inode = iget5_locked(sb, CRAMINO(cramfs_inode), 105 struct inode *inode = iget5_locked(sb, CRAMINO(cramfs_inode),
78 cramfs_iget5_test, cramfs_iget5_set, 106 cramfs_iget5_test, cramfs_iget5_set,
79 cramfs_inode); 107 cramfs_inode);
80 static struct timespec zerotime;
81
82 if (inode && (inode->i_state & I_NEW)) { 108 if (inode && (inode->i_state & I_NEW)) {
83 inode->i_mode = cramfs_inode->mode;
84 inode->i_uid = cramfs_inode->uid;
85 inode->i_size = cramfs_inode->size;
86 inode->i_blocks = (cramfs_inode->size - 1) / 512 + 1;
87 inode->i_blksize = PAGE_CACHE_SIZE;
88 inode->i_gid = cramfs_inode->gid;
89 /* Struct copy intentional */
90 inode->i_mtime = inode->i_atime = inode->i_ctime = zerotime;
91 inode->i_ino = CRAMINO(cramfs_inode);
92 /* inode->i_nlink is left 1 - arguably wrong for directories,
93 but it's the best we can do without reading the directory
94 contents. 1 yields the right result in GNU find, even
95 without -noleaf option. */
96 if (S_ISREG(inode->i_mode)) {
97 inode->i_fop = &generic_ro_fops;
98 inode->i_data.a_ops = &cramfs_aops;
99 } else if (S_ISDIR(inode->i_mode)) {
100 inode->i_op = &cramfs_dir_inode_operations;
101 inode->i_fop = &cramfs_directory_operations;
102 } else if (S_ISLNK(inode->i_mode)) {
103 inode->i_op = &page_symlink_inode_operations;
104 inode->i_data.a_ops = &cramfs_aops;
105 } else {
106 inode->i_size = 0;
107 inode->i_blocks = 0;
108 init_special_inode(inode, inode->i_mode,
109 old_decode_dev(cramfs_inode->size));
110 }
111 unlock_new_inode(inode); 109 unlock_new_inode(inode);
112 } 110 }
113 return inode; 111 return inode;
diff --git a/fs/dcache.c b/fs/dcache.c
index a173bba32666..11dc83092d4a 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -1736,7 +1736,7 @@ void __init vfs_caches_init(unsigned long mempages)
1736 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL); 1736 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
1737 1737
1738 filp_cachep = kmem_cache_create("filp", sizeof(struct file), 0, 1738 filp_cachep = kmem_cache_create("filp", sizeof(struct file), 0,
1739 SLAB_HWCACHE_ALIGN|SLAB_PANIC, filp_ctor, filp_dtor); 1739 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
1740 1740
1741 dcache_init(mempages); 1741 dcache_init(mempages);
1742 inode_init(mempages); 1742 inode_init(mempages);
diff --git a/fs/exec.c b/fs/exec.c
index 0e1c95074d42..0b515ac53134 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -885,6 +885,12 @@ int flush_old_exec(struct linux_binprm * bprm)
885 current->flags &= ~PF_RANDOMIZE; 885 current->flags &= ~PF_RANDOMIZE;
886 flush_thread(); 886 flush_thread();
887 887
888 /* Set the new mm task size. We have to do that late because it may
889 * depend on TIF_32BIT which is only updated in flush_thread() on
890 * some architectures like powerpc
891 */
892 current->mm->task_size = TASK_SIZE;
893
888 if (bprm->e_uid != current->euid || bprm->e_gid != current->egid || 894 if (bprm->e_uid != current->euid || bprm->e_gid != current->egid ||
889 file_permission(bprm->file, MAY_READ) || 895 file_permission(bprm->file, MAY_READ) ||
890 (bprm->interp_flags & BINPRM_FLAGS_ENFORCE_NONDUMP)) { 896 (bprm->interp_flags & BINPRM_FLAGS_ENFORCE_NONDUMP)) {
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index 3fc4238e9703..0384e539b88f 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -1624,15 +1624,14 @@ static int ext3_block_truncate_page(handle_t *handle, struct page *page,
1624 * For "nobh" option, we can only work if we don't need to 1624 * For "nobh" option, we can only work if we don't need to
1625 * read-in the page - otherwise we create buffers to do the IO. 1625 * read-in the page - otherwise we create buffers to do the IO.
1626 */ 1626 */
1627 if (!page_has_buffers(page) && test_opt(inode->i_sb, NOBH)) { 1627 if (!page_has_buffers(page) && test_opt(inode->i_sb, NOBH) &&
1628 if (PageUptodate(page)) { 1628 ext3_should_writeback_data(inode) && PageUptodate(page)) {
1629 kaddr = kmap_atomic(page, KM_USER0); 1629 kaddr = kmap_atomic(page, KM_USER0);
1630 memset(kaddr + offset, 0, length); 1630 memset(kaddr + offset, 0, length);
1631 flush_dcache_page(page); 1631 flush_dcache_page(page);
1632 kunmap_atomic(kaddr, KM_USER0); 1632 kunmap_atomic(kaddr, KM_USER0);
1633 set_page_dirty(page); 1633 set_page_dirty(page);
1634 goto unlock; 1634 goto unlock;
1635 }
1636 } 1635 }
1637 1636
1638 if (!page_has_buffers(page)) 1637 if (!page_has_buffers(page))
diff --git a/fs/ext3/namei.c b/fs/ext3/namei.c
index 8bd8ac077704..b8f5cd1e540d 100644
--- a/fs/ext3/namei.c
+++ b/fs/ext3/namei.c
@@ -2141,7 +2141,8 @@ retry:
2141 * We have a transaction open. All is sweetness. It also sets 2141 * We have a transaction open. All is sweetness. It also sets
2142 * i_size in generic_commit_write(). 2142 * i_size in generic_commit_write().
2143 */ 2143 */
2144 err = page_symlink(inode, symname, l); 2144 err = __page_symlink(inode, symname, l,
2145 mapping_gfp_mask(inode->i_mapping) & ~__GFP_FS);
2145 if (err) { 2146 if (err) {
2146 ext3_dec_count(handle, inode); 2147 ext3_dec_count(handle, inode);
2147 ext3_mark_inode_dirty(handle, inode); 2148 ext3_mark_inode_dirty(handle, inode);
diff --git a/fs/fifo.c b/fs/fifo.c
index 923371b753ab..d13fcd3ec803 100644
--- a/fs/fifo.c
+++ b/fs/fifo.c
@@ -34,10 +34,7 @@ static int fifo_open(struct inode *inode, struct file *filp)
34{ 34{
35 int ret; 35 int ret;
36 36
37 ret = -ERESTARTSYS; 37 mutex_lock(PIPE_MUTEX(*inode));
38 if (mutex_lock_interruptible(PIPE_MUTEX(*inode)))
39 goto err_nolock_nocleanup;
40
41 if (!inode->i_pipe) { 38 if (!inode->i_pipe) {
42 ret = -ENOMEM; 39 ret = -ENOMEM;
43 if(!pipe_new(inode)) 40 if(!pipe_new(inode))
@@ -140,8 +137,6 @@ err:
140 137
141err_nocleanup: 138err_nocleanup:
142 mutex_unlock(PIPE_MUTEX(*inode)); 139 mutex_unlock(PIPE_MUTEX(*inode));
143
144err_nolock_nocleanup:
145 return ret; 140 return ret;
146} 141}
147 142
diff --git a/fs/file_table.c b/fs/file_table.c
index 768b58167543..44fabeaa9415 100644
--- a/fs/file_table.c
+++ b/fs/file_table.c
@@ -5,6 +5,7 @@
5 * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu) 5 * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
6 */ 6 */
7 7
8#include <linux/config.h>
8#include <linux/string.h> 9#include <linux/string.h>
9#include <linux/slab.h> 10#include <linux/slab.h>
10#include <linux/file.h> 11#include <linux/file.h>
@@ -19,52 +20,67 @@
19#include <linux/capability.h> 20#include <linux/capability.h>
20#include <linux/cdev.h> 21#include <linux/cdev.h>
21#include <linux/fsnotify.h> 22#include <linux/fsnotify.h>
23#include <linux/sysctl.h>
24#include <linux/percpu_counter.h>
25
26#include <asm/atomic.h>
22 27
23/* sysctl tunables... */ 28/* sysctl tunables... */
24struct files_stat_struct files_stat = { 29struct files_stat_struct files_stat = {
25 .max_files = NR_FILE 30 .max_files = NR_FILE
26}; 31};
27 32
28EXPORT_SYMBOL(files_stat); /* Needed by unix.o */
29
30/* public. Not pretty! */ 33/* public. Not pretty! */
31 __cacheline_aligned_in_smp DEFINE_SPINLOCK(files_lock); 34__cacheline_aligned_in_smp DEFINE_SPINLOCK(files_lock);
32 35
33static DEFINE_SPINLOCK(filp_count_lock); 36static struct percpu_counter nr_files __cacheline_aligned_in_smp;
34 37
35/* slab constructors and destructors are called from arbitrary 38static inline void file_free_rcu(struct rcu_head *head)
36 * context and must be fully threaded - use a local spinlock
37 * to protect files_stat.nr_files
38 */
39void filp_ctor(void *objp, struct kmem_cache *cachep, unsigned long cflags)
40{ 39{
41 if ((cflags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) == 40 struct file *f = container_of(head, struct file, f_u.fu_rcuhead);
42 SLAB_CTOR_CONSTRUCTOR) { 41 kmem_cache_free(filp_cachep, f);
43 unsigned long flags;
44 spin_lock_irqsave(&filp_count_lock, flags);
45 files_stat.nr_files++;
46 spin_unlock_irqrestore(&filp_count_lock, flags);
47 }
48} 42}
49 43
50void filp_dtor(void *objp, struct kmem_cache *cachep, unsigned long dflags) 44static inline void file_free(struct file *f)
51{ 45{
52 unsigned long flags; 46 percpu_counter_dec(&nr_files);
53 spin_lock_irqsave(&filp_count_lock, flags); 47 call_rcu(&f->f_u.fu_rcuhead, file_free_rcu);
54 files_stat.nr_files--;
55 spin_unlock_irqrestore(&filp_count_lock, flags);
56} 48}
57 49
58static inline void file_free_rcu(struct rcu_head *head) 50/*
51 * Return the total number of open files in the system
52 */
53static int get_nr_files(void)
59{ 54{
60 struct file *f = container_of(head, struct file, f_u.fu_rcuhead); 55 return percpu_counter_read_positive(&nr_files);
61 kmem_cache_free(filp_cachep, f);
62} 56}
63 57
64static inline void file_free(struct file *f) 58/*
59 * Return the maximum number of open files in the system
60 */
61int get_max_files(void)
65{ 62{
66 call_rcu(&f->f_u.fu_rcuhead, file_free_rcu); 63 return files_stat.max_files;
67} 64}
65EXPORT_SYMBOL_GPL(get_max_files);
66
67/*
68 * Handle nr_files sysctl
69 */
70#if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
71int proc_nr_files(ctl_table *table, int write, struct file *filp,
72 void __user *buffer, size_t *lenp, loff_t *ppos)
73{
74 files_stat.nr_files = get_nr_files();
75 return proc_dointvec(table, write, filp, buffer, lenp, ppos);
76}
77#else
78int proc_nr_files(ctl_table *table, int write, struct file *filp,
79 void __user *buffer, size_t *lenp, loff_t *ppos)
80{
81 return -ENOSYS;
82}
83#endif
68 84
69/* Find an unused file structure and return a pointer to it. 85/* Find an unused file structure and return a pointer to it.
70 * Returns NULL, if there are no more free file structures or 86 * Returns NULL, if there are no more free file structures or
@@ -78,14 +94,20 @@ struct file *get_empty_filp(void)
78 /* 94 /*
79 * Privileged users can go above max_files 95 * Privileged users can go above max_files
80 */ 96 */
81 if (files_stat.nr_files >= files_stat.max_files && 97 if (get_nr_files() >= files_stat.max_files && !capable(CAP_SYS_ADMIN)) {
82 !capable(CAP_SYS_ADMIN)) 98 /*
83 goto over; 99 * percpu_counters are inaccurate. Do an expensive check before
100 * we go and fail.
101 */
102 if (percpu_counter_sum(&nr_files) >= files_stat.max_files)
103 goto over;
104 }
84 105
85 f = kmem_cache_alloc(filp_cachep, GFP_KERNEL); 106 f = kmem_cache_alloc(filp_cachep, GFP_KERNEL);
86 if (f == NULL) 107 if (f == NULL)
87 goto fail; 108 goto fail;
88 109
110 percpu_counter_inc(&nr_files);
89 memset(f, 0, sizeof(*f)); 111 memset(f, 0, sizeof(*f));
90 if (security_file_alloc(f)) 112 if (security_file_alloc(f))
91 goto fail_sec; 113 goto fail_sec;
@@ -101,10 +123,10 @@ struct file *get_empty_filp(void)
101 123
102over: 124over:
103 /* Ran out of filps - report that */ 125 /* Ran out of filps - report that */
104 if (files_stat.nr_files > old_max) { 126 if (get_nr_files() > old_max) {
105 printk(KERN_INFO "VFS: file-max limit %d reached\n", 127 printk(KERN_INFO "VFS: file-max limit %d reached\n",
106 files_stat.max_files); 128 get_max_files());
107 old_max = files_stat.nr_files; 129 old_max = get_nr_files();
108 } 130 }
109 goto fail; 131 goto fail;
110 132
@@ -276,4 +298,5 @@ void __init files_init(unsigned long mempages)
276 if (files_stat.max_files < NR_FILE) 298 if (files_stat.max_files < NR_FILE)
277 files_stat.max_files = NR_FILE; 299 files_stat.max_files = NR_FILE;
278 files_defer_init(); 300 files_defer_init();
301 percpu_counter_init(&nr_files);
279} 302}
diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c
index 21fd59c7bc24..c72a8a97935c 100644
--- a/fs/fuse/dir.c
+++ b/fs/fuse/dir.c
@@ -111,6 +111,8 @@ static int fuse_dentry_revalidate(struct dentry *entry, struct nameidata *nd)
111 111
112 /* Doesn't hurt to "reset" the validity timeout */ 112 /* Doesn't hurt to "reset" the validity timeout */
113 fuse_invalidate_entry_cache(entry); 113 fuse_invalidate_entry_cache(entry);
114
115 /* For negative dentries, always do a fresh lookup */
114 if (!inode) 116 if (!inode)
115 return 0; 117 return 0;
116 118
@@ -122,6 +124,9 @@ static int fuse_dentry_revalidate(struct dentry *entry, struct nameidata *nd)
122 fuse_lookup_init(req, entry->d_parent->d_inode, entry, &outarg); 124 fuse_lookup_init(req, entry->d_parent->d_inode, entry, &outarg);
123 request_send(fc, req); 125 request_send(fc, req);
124 err = req->out.h.error; 126 err = req->out.h.error;
127 /* Zero nodeid is same as -ENOENT */
128 if (!err && !outarg.nodeid)
129 err = -ENOENT;
125 if (!err) { 130 if (!err) {
126 struct fuse_inode *fi = get_fuse_inode(inode); 131 struct fuse_inode *fi = get_fuse_inode(inode);
127 if (outarg.nodeid != get_node_id(inode)) { 132 if (outarg.nodeid != get_node_id(inode)) {
@@ -190,8 +195,9 @@ static struct dentry *fuse_lookup(struct inode *dir, struct dentry *entry,
190 fuse_lookup_init(req, dir, entry, &outarg); 195 fuse_lookup_init(req, dir, entry, &outarg);
191 request_send(fc, req); 196 request_send(fc, req);
192 err = req->out.h.error; 197 err = req->out.h.error;
193 if (!err && ((outarg.nodeid && invalid_nodeid(outarg.nodeid)) || 198 /* Zero nodeid is same as -ENOENT, but with valid timeout */
194 !valid_mode(outarg.attr.mode))) 199 if (!err && outarg.nodeid &&
200 (invalid_nodeid(outarg.nodeid) || !valid_mode(outarg.attr.mode)))
195 err = -EIO; 201 err = -EIO;
196 if (!err && outarg.nodeid) { 202 if (!err && outarg.nodeid) {
197 inode = fuse_iget(dir->i_sb, outarg.nodeid, outarg.generation, 203 inode = fuse_iget(dir->i_sb, outarg.nodeid, outarg.generation,
diff --git a/fs/jffs2/nodelist.c b/fs/jffs2/nodelist.c
index b635e167a3fa..d4d0c41490cd 100644
--- a/fs/jffs2/nodelist.c
+++ b/fs/jffs2/nodelist.c
@@ -406,7 +406,8 @@ static int check_node_data(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info
406 int err = 0, pointed = 0; 406 int err = 0, pointed = 0;
407 struct jffs2_eraseblock *jeb; 407 struct jffs2_eraseblock *jeb;
408 unsigned char *buffer; 408 unsigned char *buffer;
409 uint32_t crc, ofs, retlen, len; 409 uint32_t crc, ofs, len;
410 size_t retlen;
410 411
411 BUG_ON(tn->csize == 0); 412 BUG_ON(tn->csize == 0);
412 413
diff --git a/fs/jffs2/readinode.c b/fs/jffs2/readinode.c
index 5f0652df5d47..f1695642d0f7 100644
--- a/fs/jffs2/readinode.c
+++ b/fs/jffs2/readinode.c
@@ -112,7 +112,7 @@ static struct jffs2_raw_node_ref *jffs2_first_valid_node(struct jffs2_raw_node_r
112 * negative error code on failure. 112 * negative error code on failure.
113 */ 113 */
114static inline int read_direntry(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref, 114static inline int read_direntry(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref,
115 struct jffs2_raw_dirent *rd, uint32_t read, struct jffs2_full_dirent **fdp, 115 struct jffs2_raw_dirent *rd, size_t read, struct jffs2_full_dirent **fdp,
116 uint32_t *latest_mctime, uint32_t *mctime_ver) 116 uint32_t *latest_mctime, uint32_t *mctime_ver)
117{ 117{
118 struct jffs2_full_dirent *fd; 118 struct jffs2_full_dirent *fd;
diff --git a/fs/jffs2/scan.c b/fs/jffs2/scan.c
index 3e51dd1da8aa..cf55b221fc2b 100644
--- a/fs/jffs2/scan.c
+++ b/fs/jffs2/scan.c
@@ -233,7 +233,7 @@ int jffs2_scan_medium(struct jffs2_sb_info *c)
233 c->nextblock->dirty_size = 0; 233 c->nextblock->dirty_size = 0;
234 } 234 }
235#ifdef CONFIG_JFFS2_FS_WRITEBUFFER 235#ifdef CONFIG_JFFS2_FS_WRITEBUFFER
236 if (!jffs2_can_mark_obsolete(c) && c->nextblock && (c->nextblock->free_size % c->wbuf_pagesize)) { 236 if (!jffs2_can_mark_obsolete(c) && c->wbuf_pagesize && c->nextblock && (c->nextblock->free_size % c->wbuf_pagesize)) {
237 /* If we're going to start writing into a block which already 237 /* If we're going to start writing into a block which already
238 contains data, and the end of the data isn't page-aligned, 238 contains data, and the end of the data isn't page-aligned,
239 skip a little and align it. */ 239 skip a little and align it. */
diff --git a/fs/lockd/clntproc.c b/fs/lockd/clntproc.c
index 220058d8616d..970b6a6aa337 100644
--- a/fs/lockd/clntproc.c
+++ b/fs/lockd/clntproc.c
@@ -662,12 +662,18 @@ nlmclnt_unlock(struct nlm_rqst *req, struct file_lock *fl)
662 * reclaimed while we're stuck in the unlock call. */ 662 * reclaimed while we're stuck in the unlock call. */
663 fl->fl_u.nfs_fl.flags &= ~NFS_LCK_GRANTED; 663 fl->fl_u.nfs_fl.flags &= ~NFS_LCK_GRANTED;
664 664
665 /*
666 * Note: the server is supposed to either grant us the unlock
667 * request, or to deny it with NLM_LCK_DENIED_GRACE_PERIOD. In either
668 * case, we want to unlock.
669 */
670 do_vfs_lock(fl);
671
665 if (req->a_flags & RPC_TASK_ASYNC) { 672 if (req->a_flags & RPC_TASK_ASYNC) {
666 status = nlmclnt_async_call(req, NLMPROC_UNLOCK, 673 status = nlmclnt_async_call(req, NLMPROC_UNLOCK,
667 &nlmclnt_unlock_ops); 674 &nlmclnt_unlock_ops);
668 /* Hrmf... Do the unlock early since locks_remove_posix() 675 /* Hrmf... Do the unlock early since locks_remove_posix()
669 * really expects us to free the lock synchronously */ 676 * really expects us to free the lock synchronously */
670 do_vfs_lock(fl);
671 if (status < 0) { 677 if (status < 0) {
672 nlmclnt_release_lockargs(req); 678 nlmclnt_release_lockargs(req);
673 kfree(req); 679 kfree(req);
@@ -680,7 +686,6 @@ nlmclnt_unlock(struct nlm_rqst *req, struct file_lock *fl)
680 if (status < 0) 686 if (status < 0)
681 return status; 687 return status;
682 688
683 do_vfs_lock(fl);
684 if (resp->status == NLM_LCK_GRANTED) 689 if (resp->status == NLM_LCK_GRANTED)
685 return 0; 690 return 0;
686 691
diff --git a/fs/namei.c b/fs/namei.c
index 557dcf395ca1..8dc2b038d5d9 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2613,13 +2613,15 @@ void page_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
2613 } 2613 }
2614} 2614}
2615 2615
2616int page_symlink(struct inode *inode, const char *symname, int len) 2616int __page_symlink(struct inode *inode, const char *symname, int len,
2617 gfp_t gfp_mask)
2617{ 2618{
2618 struct address_space *mapping = inode->i_mapping; 2619 struct address_space *mapping = inode->i_mapping;
2619 struct page *page = grab_cache_page(mapping, 0); 2620 struct page *page;
2620 int err = -ENOMEM; 2621 int err = -ENOMEM;
2621 char *kaddr; 2622 char *kaddr;
2622 2623
2624 page = find_or_create_page(mapping, 0, gfp_mask);
2623 if (!page) 2625 if (!page)
2624 goto fail; 2626 goto fail;
2625 err = mapping->a_ops->prepare_write(NULL, page, 0, len-1); 2627 err = mapping->a_ops->prepare_write(NULL, page, 0, len-1);
@@ -2654,6 +2656,12 @@ fail:
2654 return err; 2656 return err;
2655} 2657}
2656 2658
2659int page_symlink(struct inode *inode, const char *symname, int len)
2660{
2661 return __page_symlink(inode, symname, len,
2662 mapping_gfp_mask(inode->i_mapping));
2663}
2664
2657struct inode_operations page_symlink_inode_operations = { 2665struct inode_operations page_symlink_inode_operations = {
2658 .readlink = generic_readlink, 2666 .readlink = generic_readlink,
2659 .follow_link = page_follow_link_light, 2667 .follow_link = page_follow_link_light,
@@ -2672,6 +2680,7 @@ EXPORT_SYMBOL(lookup_one_len);
2672EXPORT_SYMBOL(page_follow_link_light); 2680EXPORT_SYMBOL(page_follow_link_light);
2673EXPORT_SYMBOL(page_put_link); 2681EXPORT_SYMBOL(page_put_link);
2674EXPORT_SYMBOL(page_readlink); 2682EXPORT_SYMBOL(page_readlink);
2683EXPORT_SYMBOL(__page_symlink);
2675EXPORT_SYMBOL(page_symlink); 2684EXPORT_SYMBOL(page_symlink);
2676EXPORT_SYMBOL(page_symlink_inode_operations); 2685EXPORT_SYMBOL(page_symlink_inode_operations);
2677EXPORT_SYMBOL(path_lookup); 2686EXPORT_SYMBOL(path_lookup);
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 04ab2fc360e7..4e9b3a1b36c5 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -57,6 +57,7 @@
57#define NFSDBG_FACILITY NFSDBG_VFS 57#define NFSDBG_FACILITY NFSDBG_VFS
58#define MAX_DIRECTIO_SIZE (4096UL << PAGE_SHIFT) 58#define MAX_DIRECTIO_SIZE (4096UL << PAGE_SHIFT)
59 59
60static void nfs_free_user_pages(struct page **pages, int npages, int do_dirty);
60static kmem_cache_t *nfs_direct_cachep; 61static kmem_cache_t *nfs_direct_cachep;
61 62
62/* 63/*
@@ -107,6 +108,15 @@ nfs_get_user_pages(int rw, unsigned long user_addr, size_t size,
107 page_count, (rw == READ), 0, 108 page_count, (rw == READ), 0,
108 *pages, NULL); 109 *pages, NULL);
109 up_read(&current->mm->mmap_sem); 110 up_read(&current->mm->mmap_sem);
111 /*
112 * If we got fewer pages than expected from get_user_pages(),
113 * the user buffer runs off the end of a mapping; return EFAULT.
114 */
115 if (result >= 0 && result < page_count) {
116 nfs_free_user_pages(*pages, result, 0);
117 *pages = NULL;
118 result = -EFAULT;
119 }
110 } 120 }
111 return result; 121 return result;
112} 122}
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 984ca3454d04..f8c0066e02e1 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1430,7 +1430,7 @@ static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
1430 if (status == 0) 1430 if (status == 0)
1431 status = nfs4_do_fsinfo(server, fhandle, info); 1431 status = nfs4_do_fsinfo(server, fhandle, info);
1432out: 1432out:
1433 return status; 1433 return nfs4_map_errors(status);
1434} 1434}
1435 1435
1436static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr) 1436static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
diff --git a/fs/ocfs2/cluster/masklog.c b/fs/ocfs2/cluster/masklog.c
index fd741cea5705..636593bf4d17 100644
--- a/fs/ocfs2/cluster/masklog.c
+++ b/fs/ocfs2/cluster/masklog.c
@@ -74,6 +74,7 @@ struct mlog_attribute {
74#define define_mask(_name) { \ 74#define define_mask(_name) { \
75 .attr = { \ 75 .attr = { \
76 .name = #_name, \ 76 .name = #_name, \
77 .owner = THIS_MODULE, \
77 .mode = S_IRUGO | S_IWUSR, \ 78 .mode = S_IRUGO | S_IWUSR, \
78 }, \ 79 }, \
79 .mask = ML_##_name, \ 80 .mask = ML_##_name, \
diff --git a/fs/ocfs2/cluster/masklog.h b/fs/ocfs2/cluster/masklog.h
index e8c56a3d9c64..2cadc3009c83 100644
--- a/fs/ocfs2/cluster/masklog.h
+++ b/fs/ocfs2/cluster/masklog.h
@@ -256,7 +256,7 @@ extern struct mlog_bits mlog_and_bits, mlog_not_bits;
256 } \ 256 } \
257} while (0) 257} while (0)
258 258
259#if (BITS_PER_LONG == 32) || defined(CONFIG_X86_64) 259#if (BITS_PER_LONG == 32) || defined(CONFIG_X86_64) || (defined(CONFIG_UML_X86) && defined(CONFIG_64BIT))
260#define MLFi64 "lld" 260#define MLFi64 "lld"
261#define MLFu64 "llu" 261#define MLFu64 "llu"
262#define MLFx64 "llx" 262#define MLFx64 "llx"
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index cf7828f23361..e1fceb8aa32d 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -756,7 +756,7 @@ static int __init init_o2nm(void)
756 if (!ocfs2_table_header) { 756 if (!ocfs2_table_header) {
757 printk(KERN_ERR "nodemanager: unable to register sysctl\n"); 757 printk(KERN_ERR "nodemanager: unable to register sysctl\n");
758 ret = -ENOMEM; /* or something. */ 758 ret = -ENOMEM; /* or something. */
759 goto out; 759 goto out_o2net;
760 } 760 }
761 761
762 ret = o2net_register_hb_callbacks(); 762 ret = o2net_register_hb_callbacks();
@@ -780,6 +780,8 @@ out_callbacks:
780 o2net_unregister_hb_callbacks(); 780 o2net_unregister_hb_callbacks();
781out_sysctl: 781out_sysctl:
782 unregister_sysctl_table(ocfs2_table_header); 782 unregister_sysctl_table(ocfs2_table_header);
783out_o2net:
784 o2net_exit();
783out: 785out:
784 return ret; 786 return ret;
785} 787}
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index d22d4cf08db1..0f60cc0d3985 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -1318,7 +1318,7 @@ static void o2net_start_connect(void *arg)
1318{ 1318{
1319 struct o2net_node *nn = arg; 1319 struct o2net_node *nn = arg;
1320 struct o2net_sock_container *sc = NULL; 1320 struct o2net_sock_container *sc = NULL;
1321 struct o2nm_node *node = NULL; 1321 struct o2nm_node *node = NULL, *mynode = NULL;
1322 struct socket *sock = NULL; 1322 struct socket *sock = NULL;
1323 struct sockaddr_in myaddr = {0, }, remoteaddr = {0, }; 1323 struct sockaddr_in myaddr = {0, }, remoteaddr = {0, };
1324 int ret = 0; 1324 int ret = 0;
@@ -1334,6 +1334,12 @@ static void o2net_start_connect(void *arg)
1334 goto out; 1334 goto out;
1335 } 1335 }
1336 1336
1337 mynode = o2nm_get_node_by_num(o2nm_this_node());
1338 if (mynode == NULL) {
1339 ret = 0;
1340 goto out;
1341 }
1342
1337 spin_lock(&nn->nn_lock); 1343 spin_lock(&nn->nn_lock);
1338 /* see if we already have one pending or have given up */ 1344 /* see if we already have one pending or have given up */
1339 if (nn->nn_sc || nn->nn_persistent_error) 1345 if (nn->nn_sc || nn->nn_persistent_error)
@@ -1361,12 +1367,14 @@ static void o2net_start_connect(void *arg)
1361 sock->sk->sk_allocation = GFP_ATOMIC; 1367 sock->sk->sk_allocation = GFP_ATOMIC;
1362 1368
1363 myaddr.sin_family = AF_INET; 1369 myaddr.sin_family = AF_INET;
1370 myaddr.sin_addr.s_addr = (__force u32)mynode->nd_ipv4_address;
1364 myaddr.sin_port = (__force u16)htons(0); /* any port */ 1371 myaddr.sin_port = (__force u16)htons(0); /* any port */
1365 1372
1366 ret = sock->ops->bind(sock, (struct sockaddr *)&myaddr, 1373 ret = sock->ops->bind(sock, (struct sockaddr *)&myaddr,
1367 sizeof(myaddr)); 1374 sizeof(myaddr));
1368 if (ret) { 1375 if (ret) {
1369 mlog(0, "bind failed: %d\n", ret); 1376 mlog(ML_ERROR, "bind failed with %d at address %u.%u.%u.%u\n",
1377 ret, NIPQUAD(mynode->nd_ipv4_address));
1370 goto out; 1378 goto out;
1371 } 1379 }
1372 1380
@@ -1407,6 +1415,8 @@ out:
1407 sc_put(sc); 1415 sc_put(sc);
1408 if (node) 1416 if (node)
1409 o2nm_node_put(node); 1417 o2nm_node_put(node);
1418 if (mynode)
1419 o2nm_node_put(mynode);
1410 1420
1411 return; 1421 return;
1412} 1422}
diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
index a6f4585501c8..616ff2b8434a 100644
--- a/fs/ocfs2/cluster/tcp.h
+++ b/fs/ocfs2/cluster/tcp.h
@@ -85,13 +85,10 @@ enum {
85 O2NET_DRIVER_READY, 85 O2NET_DRIVER_READY,
86}; 86};
87 87
88int o2net_init_tcp_sock(struct inode *inode);
89int o2net_send_message(u32 msg_type, u32 key, void *data, u32 len, 88int o2net_send_message(u32 msg_type, u32 key, void *data, u32 len,
90 u8 target_node, int *status); 89 u8 target_node, int *status);
91int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *vec, 90int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *vec,
92 size_t veclen, u8 target_node, int *status); 91 size_t veclen, u8 target_node, int *status);
93int o2net_broadcast_message(u32 msg_type, u32 key, void *data, u32 len,
94 struct inode *group);
95 92
96int o2net_register_handler(u32 msg_type, u32 key, u32 max_len, 93int o2net_register_handler(u32 msg_type, u32 key, u32 max_len,
97 o2net_msg_handler_func *func, void *data, 94 o2net_msg_handler_func *func, void *data,
@@ -107,7 +104,5 @@ void o2net_disconnect_node(struct o2nm_node *node);
107 104
108int o2net_init(void); 105int o2net_init(void);
109void o2net_exit(void); 106void o2net_exit(void);
110int o2net_proc_init(struct proc_dir_entry *parent);
111void o2net_proc_exit(struct proc_dir_entry *parent);
112 107
113#endif /* O2CLUSTER_TCP_H */ 108#endif /* O2CLUSTER_TCP_H */
diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h
index 23ceaa7127b4..9c772583744a 100644
--- a/fs/ocfs2/dlm/dlmcommon.h
+++ b/fs/ocfs2/dlm/dlmcommon.h
@@ -37,9 +37,7 @@
37#define DLM_THREAD_SHUFFLE_INTERVAL 5 // flush everything every 5 passes 37#define DLM_THREAD_SHUFFLE_INTERVAL 5 // flush everything every 5 passes
38#define DLM_THREAD_MS 200 // flush at least every 200 ms 38#define DLM_THREAD_MS 200 // flush at least every 200 ms
39 39
40#define DLM_HASH_BITS 7 40#define DLM_HASH_BUCKETS (PAGE_SIZE / sizeof(struct hlist_head))
41#define DLM_HASH_SIZE (1 << DLM_HASH_BITS)
42#define DLM_HASH_MASK (DLM_HASH_SIZE - 1)
43 41
44enum dlm_ast_type { 42enum dlm_ast_type {
45 DLM_AST = 0, 43 DLM_AST = 0,
@@ -87,7 +85,7 @@ enum dlm_ctxt_state {
87struct dlm_ctxt 85struct dlm_ctxt
88{ 86{
89 struct list_head list; 87 struct list_head list;
90 struct list_head *resources; 88 struct hlist_head *lockres_hash;
91 struct list_head dirty_list; 89 struct list_head dirty_list;
92 struct list_head purge_list; 90 struct list_head purge_list;
93 struct list_head pending_asts; 91 struct list_head pending_asts;
@@ -217,7 +215,7 @@ struct dlm_lock_resource
217{ 215{
218 /* WARNING: Please see the comment in dlm_init_lockres before 216 /* WARNING: Please see the comment in dlm_init_lockres before
219 * adding fields here. */ 217 * adding fields here. */
220 struct list_head list; 218 struct hlist_node hash_node;
221 struct kref refs; 219 struct kref refs;
222 220
223 /* please keep these next 3 in this order 221 /* please keep these next 3 in this order
diff --git a/fs/ocfs2/dlm/dlmdebug.c b/fs/ocfs2/dlm/dlmdebug.c
index f339fe27975a..54f61b76ab51 100644
--- a/fs/ocfs2/dlm/dlmdebug.c
+++ b/fs/ocfs2/dlm/dlmdebug.c
@@ -117,8 +117,8 @@ EXPORT_SYMBOL_GPL(dlm_print_one_lock);
117void dlm_dump_lock_resources(struct dlm_ctxt *dlm) 117void dlm_dump_lock_resources(struct dlm_ctxt *dlm)
118{ 118{
119 struct dlm_lock_resource *res; 119 struct dlm_lock_resource *res;
120 struct list_head *iter; 120 struct hlist_node *iter;
121 struct list_head *bucket; 121 struct hlist_head *bucket;
122 int i; 122 int i;
123 123
124 mlog(ML_NOTICE, "struct dlm_ctxt: %s, node=%u, key=%u\n", 124 mlog(ML_NOTICE, "struct dlm_ctxt: %s, node=%u, key=%u\n",
@@ -129,12 +129,10 @@ void dlm_dump_lock_resources(struct dlm_ctxt *dlm)
129 } 129 }
130 130
131 spin_lock(&dlm->spinlock); 131 spin_lock(&dlm->spinlock);
132 for (i=0; i<DLM_HASH_SIZE; i++) { 132 for (i=0; i<DLM_HASH_BUCKETS; i++) {
133 bucket = &(dlm->resources[i]); 133 bucket = &(dlm->lockres_hash[i]);
134 list_for_each(iter, bucket) { 134 hlist_for_each_entry(res, iter, bucket, hash_node)
135 res = list_entry(iter, struct dlm_lock_resource, list);
136 dlm_print_one_lock_resource(res); 135 dlm_print_one_lock_resource(res);
137 }
138 } 136 }
139 spin_unlock(&dlm->spinlock); 137 spin_unlock(&dlm->spinlock);
140} 138}
diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
index 6ee30837389c..8f3a9e3106fd 100644
--- a/fs/ocfs2/dlm/dlmdomain.c
+++ b/fs/ocfs2/dlm/dlmdomain.c
@@ -77,26 +77,26 @@ static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm);
77 77
78void __dlm_unhash_lockres(struct dlm_lock_resource *lockres) 78void __dlm_unhash_lockres(struct dlm_lock_resource *lockres)
79{ 79{
80 list_del_init(&lockres->list); 80 hlist_del_init(&lockres->hash_node);
81 dlm_lockres_put(lockres); 81 dlm_lockres_put(lockres);
82} 82}
83 83
84void __dlm_insert_lockres(struct dlm_ctxt *dlm, 84void __dlm_insert_lockres(struct dlm_ctxt *dlm,
85 struct dlm_lock_resource *res) 85 struct dlm_lock_resource *res)
86{ 86{
87 struct list_head *bucket; 87 struct hlist_head *bucket;
88 struct qstr *q; 88 struct qstr *q;
89 89
90 assert_spin_locked(&dlm->spinlock); 90 assert_spin_locked(&dlm->spinlock);
91 91
92 q = &res->lockname; 92 q = &res->lockname;
93 q->hash = full_name_hash(q->name, q->len); 93 q->hash = full_name_hash(q->name, q->len);
94 bucket = &(dlm->resources[q->hash & DLM_HASH_MASK]); 94 bucket = &(dlm->lockres_hash[q->hash % DLM_HASH_BUCKETS]);
95 95
96 /* get a reference for our hashtable */ 96 /* get a reference for our hashtable */
97 dlm_lockres_get(res); 97 dlm_lockres_get(res);
98 98
99 list_add_tail(&res->list, bucket); 99 hlist_add_head(&res->hash_node, bucket);
100} 100}
101 101
102struct dlm_lock_resource * __dlm_lookup_lockres(struct dlm_ctxt *dlm, 102struct dlm_lock_resource * __dlm_lookup_lockres(struct dlm_ctxt *dlm,
@@ -104,9 +104,9 @@ struct dlm_lock_resource * __dlm_lookup_lockres(struct dlm_ctxt *dlm,
104 unsigned int len) 104 unsigned int len)
105{ 105{
106 unsigned int hash; 106 unsigned int hash;
107 struct list_head *iter; 107 struct hlist_node *iter;
108 struct dlm_lock_resource *tmpres=NULL; 108 struct dlm_lock_resource *tmpres=NULL;
109 struct list_head *bucket; 109 struct hlist_head *bucket;
110 110
111 mlog_entry("%.*s\n", len, name); 111 mlog_entry("%.*s\n", len, name);
112 112
@@ -114,11 +114,11 @@ struct dlm_lock_resource * __dlm_lookup_lockres(struct dlm_ctxt *dlm,
114 114
115 hash = full_name_hash(name, len); 115 hash = full_name_hash(name, len);
116 116
117 bucket = &(dlm->resources[hash & DLM_HASH_MASK]); 117 bucket = &(dlm->lockres_hash[hash % DLM_HASH_BUCKETS]);
118 118
119 /* check for pre-existing lock */ 119 /* check for pre-existing lock */
120 list_for_each(iter, bucket) { 120 hlist_for_each(iter, bucket) {
121 tmpres = list_entry(iter, struct dlm_lock_resource, list); 121 tmpres = hlist_entry(iter, struct dlm_lock_resource, hash_node);
122 if (tmpres->lockname.len == len && 122 if (tmpres->lockname.len == len &&
123 memcmp(tmpres->lockname.name, name, len) == 0) { 123 memcmp(tmpres->lockname.name, name, len) == 0) {
124 dlm_lockres_get(tmpres); 124 dlm_lockres_get(tmpres);
@@ -193,8 +193,8 @@ static int dlm_wait_on_domain_helper(const char *domain)
193 193
194static void dlm_free_ctxt_mem(struct dlm_ctxt *dlm) 194static void dlm_free_ctxt_mem(struct dlm_ctxt *dlm)
195{ 195{
196 if (dlm->resources) 196 if (dlm->lockres_hash)
197 free_page((unsigned long) dlm->resources); 197 free_page((unsigned long) dlm->lockres_hash);
198 198
199 if (dlm->name) 199 if (dlm->name)
200 kfree(dlm->name); 200 kfree(dlm->name);
@@ -303,10 +303,10 @@ static void dlm_migrate_all_locks(struct dlm_ctxt *dlm)
303 mlog(0, "Migrating locks from domain %s\n", dlm->name); 303 mlog(0, "Migrating locks from domain %s\n", dlm->name);
304restart: 304restart:
305 spin_lock(&dlm->spinlock); 305 spin_lock(&dlm->spinlock);
306 for (i=0; i<DLM_HASH_SIZE; i++) { 306 for (i = 0; i < DLM_HASH_BUCKETS; i++) {
307 while (!list_empty(&dlm->resources[i])) { 307 while (!hlist_empty(&dlm->lockres_hash[i])) {
308 res = list_entry(dlm->resources[i].next, 308 res = hlist_entry(dlm->lockres_hash[i].first,
309 struct dlm_lock_resource, list); 309 struct dlm_lock_resource, hash_node);
310 /* need reference when manually grabbing lockres */ 310 /* need reference when manually grabbing lockres */
311 dlm_lockres_get(res); 311 dlm_lockres_get(res);
312 /* this should unhash the lockres 312 /* this should unhash the lockres
@@ -1191,18 +1191,17 @@ static struct dlm_ctxt *dlm_alloc_ctxt(const char *domain,
1191 goto leave; 1191 goto leave;
1192 } 1192 }
1193 1193
1194 dlm->resources = (struct list_head *) __get_free_page(GFP_KERNEL); 1194 dlm->lockres_hash = (struct hlist_head *) __get_free_page(GFP_KERNEL);
1195 if (!dlm->resources) { 1195 if (!dlm->lockres_hash) {
1196 mlog_errno(-ENOMEM); 1196 mlog_errno(-ENOMEM);
1197 kfree(dlm->name); 1197 kfree(dlm->name);
1198 kfree(dlm); 1198 kfree(dlm);
1199 dlm = NULL; 1199 dlm = NULL;
1200 goto leave; 1200 goto leave;
1201 } 1201 }
1202 memset(dlm->resources, 0, PAGE_SIZE);
1203 1202
1204 for (i=0; i<DLM_HASH_SIZE; i++) 1203 for (i=0; i<DLM_HASH_BUCKETS; i++)
1205 INIT_LIST_HEAD(&dlm->resources[i]); 1204 INIT_HLIST_HEAD(&dlm->lockres_hash[i]);
1206 1205
1207 strcpy(dlm->name, domain); 1206 strcpy(dlm->name, domain);
1208 dlm->key = key; 1207 dlm->key = key;
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index 2e2e95e69499..847dd3cc4cf5 100644
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -564,7 +564,7 @@ static void dlm_lockres_release(struct kref *kref)
564 564
565 /* By the time we're ready to blow this guy away, we shouldn't 565 /* By the time we're ready to blow this guy away, we shouldn't
566 * be on any lists. */ 566 * be on any lists. */
567 BUG_ON(!list_empty(&res->list)); 567 BUG_ON(!hlist_unhashed(&res->hash_node));
568 BUG_ON(!list_empty(&res->granted)); 568 BUG_ON(!list_empty(&res->granted));
569 BUG_ON(!list_empty(&res->converting)); 569 BUG_ON(!list_empty(&res->converting));
570 BUG_ON(!list_empty(&res->blocked)); 570 BUG_ON(!list_empty(&res->blocked));
@@ -605,7 +605,7 @@ static void dlm_init_lockres(struct dlm_ctxt *dlm,
605 605
606 init_waitqueue_head(&res->wq); 606 init_waitqueue_head(&res->wq);
607 spin_lock_init(&res->spinlock); 607 spin_lock_init(&res->spinlock);
608 INIT_LIST_HEAD(&res->list); 608 INIT_HLIST_NODE(&res->hash_node);
609 INIT_LIST_HEAD(&res->granted); 609 INIT_LIST_HEAD(&res->granted);
610 INIT_LIST_HEAD(&res->converting); 610 INIT_LIST_HEAD(&res->converting);
611 INIT_LIST_HEAD(&res->blocked); 611 INIT_LIST_HEAD(&res->blocked);
diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
index ed76bda1a534..1e232000f3f7 100644
--- a/fs/ocfs2/dlm/dlmrecovery.c
+++ b/fs/ocfs2/dlm/dlmrecovery.c
@@ -1693,7 +1693,10 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm,
1693 u8 dead_node, u8 new_master) 1693 u8 dead_node, u8 new_master)
1694{ 1694{
1695 int i; 1695 int i;
1696 struct list_head *iter, *iter2, *bucket; 1696 struct list_head *iter, *iter2;
1697 struct hlist_node *hash_iter;
1698 struct hlist_head *bucket;
1699
1697 struct dlm_lock_resource *res; 1700 struct dlm_lock_resource *res;
1698 1701
1699 mlog_entry_void(); 1702 mlog_entry_void();
@@ -1717,10 +1720,9 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm,
1717 * for now we need to run the whole hash, clear 1720 * for now we need to run the whole hash, clear
1718 * the RECOVERING state and set the owner 1721 * the RECOVERING state and set the owner
1719 * if necessary */ 1722 * if necessary */
1720 for (i=0; i<DLM_HASH_SIZE; i++) { 1723 for (i = 0; i < DLM_HASH_BUCKETS; i++) {
1721 bucket = &(dlm->resources[i]); 1724 bucket = &(dlm->lockres_hash[i]);
1722 list_for_each(iter, bucket) { 1725 hlist_for_each_entry(res, hash_iter, bucket, hash_node) {
1723 res = list_entry (iter, struct dlm_lock_resource, list);
1724 if (res->state & DLM_LOCK_RES_RECOVERING) { 1726 if (res->state & DLM_LOCK_RES_RECOVERING) {
1725 if (res->owner == dead_node) { 1727 if (res->owner == dead_node) {
1726 mlog(0, "(this=%u) res %.*s owner=%u " 1728 mlog(0, "(this=%u) res %.*s owner=%u "
@@ -1852,10 +1854,10 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm,
1852 1854
1853static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node) 1855static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
1854{ 1856{
1855 struct list_head *iter; 1857 struct hlist_node *iter;
1856 struct dlm_lock_resource *res; 1858 struct dlm_lock_resource *res;
1857 int i; 1859 int i;
1858 struct list_head *bucket; 1860 struct hlist_head *bucket;
1859 struct dlm_lock *lock; 1861 struct dlm_lock *lock;
1860 1862
1861 1863
@@ -1876,10 +1878,9 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
1876 * can be kicked again to see if any ASTs or BASTs 1878 * can be kicked again to see if any ASTs or BASTs
1877 * need to be fired as a result. 1879 * need to be fired as a result.
1878 */ 1880 */
1879 for (i=0; i<DLM_HASH_SIZE; i++) { 1881 for (i = 0; i < DLM_HASH_BUCKETS; i++) {
1880 bucket = &(dlm->resources[i]); 1882 bucket = &(dlm->lockres_hash[i]);
1881 list_for_each(iter, bucket) { 1883 hlist_for_each_entry(res, iter, bucket, hash_node) {
1882 res = list_entry (iter, struct dlm_lock_resource, list);
1883 /* always prune any $RECOVERY entries for dead nodes, 1884 /* always prune any $RECOVERY entries for dead nodes,
1884 * otherwise hangs can occur during later recovery */ 1885 * otherwise hangs can occur during later recovery */
1885 if (dlm_is_recovery_lock(res->lockname.name, 1886 if (dlm_is_recovery_lock(res->lockname.name,
diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index b6ba292e9544..e6f207eebab4 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -181,6 +181,12 @@ static int ocfs2_extent_map_find_leaf(struct inode *inode,
181 ret = -EBADR; 181 ret = -EBADR;
182 if (rec_end > OCFS2_I(inode)->ip_clusters) { 182 if (rec_end > OCFS2_I(inode)->ip_clusters) {
183 mlog_errno(ret); 183 mlog_errno(ret);
184 ocfs2_error(inode->i_sb,
185 "Extent %d at e_blkno %"MLFu64" of inode %"MLFu64" goes past ip_clusters of %u\n",
186 i,
187 le64_to_cpu(rec->e_blkno),
188 OCFS2_I(inode)->ip_blkno,
189 OCFS2_I(inode)->ip_clusters);
184 goto out_free; 190 goto out_free;
185 } 191 }
186 192
@@ -226,6 +232,12 @@ static int ocfs2_extent_map_find_leaf(struct inode *inode,
226 ret = -EBADR; 232 ret = -EBADR;
227 if (blkno) { 233 if (blkno) {
228 mlog_errno(ret); 234 mlog_errno(ret);
235 ocfs2_error(inode->i_sb,
236 "Multiple extents for (cpos = %u, clusters = %u) on inode %"MLFu64"; e_blkno %"MLFu64" and rec %d at e_blkno %"MLFu64"\n",
237 cpos, clusters,
238 OCFS2_I(inode)->ip_blkno,
239 blkno, i,
240 le64_to_cpu(rec->e_blkno));
229 goto out_free; 241 goto out_free;
230 } 242 }
231 243
@@ -238,6 +250,10 @@ static int ocfs2_extent_map_find_leaf(struct inode *inode,
238 */ 250 */
239 ret = -EBADR; 251 ret = -EBADR;
240 if (!blkno) { 252 if (!blkno) {
253 ocfs2_error(inode->i_sb,
254 "No record found for (cpos = %u, clusters = %u) on inode %"MLFu64"\n",
255 cpos, clusters,
256 OCFS2_I(inode)->ip_blkno);
241 mlog_errno(ret); 257 mlog_errno(ret);
242 goto out_free; 258 goto out_free;
243 } 259 }
@@ -266,6 +282,20 @@ static int ocfs2_extent_map_find_leaf(struct inode *inode,
266 282
267 for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) { 283 for (i = 0; i < le16_to_cpu(el->l_next_free_rec); i++) {
268 rec = &el->l_recs[i]; 284 rec = &el->l_recs[i];
285
286 if ((le32_to_cpu(rec->e_cpos) + le32_to_cpu(rec->e_clusters)) >
287 OCFS2_I(inode)->ip_clusters) {
288 ret = -EBADR;
289 mlog_errno(ret);
290 ocfs2_error(inode->i_sb,
291 "Extent %d at e_blkno %"MLFu64" of inode %"MLFu64" goes past ip_clusters of %u\n",
292 i,
293 le64_to_cpu(rec->e_blkno),
294 OCFS2_I(inode)->ip_blkno,
295 OCFS2_I(inode)->ip_clusters);
296 return ret;
297 }
298
269 ret = ocfs2_extent_map_insert(inode, rec, 299 ret = ocfs2_extent_map_insert(inode, rec,
270 le16_to_cpu(el->l_tree_depth)); 300 le16_to_cpu(el->l_tree_depth));
271 if (ret) { 301 if (ret) {
@@ -526,6 +556,10 @@ static int ocfs2_extent_map_insert(struct inode *inode,
526 OCFS2_I(inode)->ip_map.em_clusters) { 556 OCFS2_I(inode)->ip_map.em_clusters) {
527 ret = -EBADR; 557 ret = -EBADR;
528 mlog_errno(ret); 558 mlog_errno(ret);
559 ocfs2_error(inode->i_sb,
560 "Zero e_clusters on non-tail extent record at e_blkno %"MLFu64" on inode %"MLFu64"\n",
561 le64_to_cpu(rec->e_blkno),
562 OCFS2_I(inode)->ip_blkno);
529 return ret; 563 return ret;
530 } 564 }
531 565
@@ -588,12 +622,12 @@ static int ocfs2_extent_map_insert(struct inode *inode,
588 * Existing record in the extent map: 622 * Existing record in the extent map:
589 * 623 *
590 * cpos = 10, len = 10 624 * cpos = 10, len = 10
591 * |---------| 625 * |---------|
592 * 626 *
593 * New Record: 627 * New Record:
594 * 628 *
595 * cpos = 10, len = 20 629 * cpos = 10, len = 20
596 * |------------------| 630 * |------------------|
597 * 631 *
598 * The passed record is the new on-disk record. The new_clusters value 632 * The passed record is the new on-disk record. The new_clusters value
599 * is how many clusters were added to the file. If the append is a 633 * is how many clusters were added to the file. If the append is a
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 1715bc90e705..8a4048b55fdc 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -933,9 +933,6 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
933 struct file *filp = iocb->ki_filp; 933 struct file *filp = iocb->ki_filp;
934 struct inode *inode = filp->f_dentry->d_inode; 934 struct inode *inode = filp->f_dentry->d_inode;
935 loff_t newsize, saved_pos; 935 loff_t newsize, saved_pos;
936#ifdef OCFS2_ORACORE_WORKAROUNDS
937 struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
938#endif
939 936
940 mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf, 937 mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf,
941 (unsigned int)count, 938 (unsigned int)count,
@@ -951,14 +948,6 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
951 return -EIO; 948 return -EIO;
952 } 949 }
953 950
954#ifdef OCFS2_ORACORE_WORKAROUNDS
955 /* ugh, work around some applications which open everything O_DIRECT +
956 * O_APPEND and really don't mean to use O_DIRECT. */
957 if (osb->s_mount_opt & OCFS2_MOUNT_COMPAT_OCFS &&
958 (filp->f_flags & O_APPEND) && (filp->f_flags & O_DIRECT))
959 filp->f_flags &= ~O_DIRECT;
960#endif
961
962 mutex_lock(&inode->i_mutex); 951 mutex_lock(&inode->i_mutex);
963 /* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */ 952 /* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */
964 if (filp->f_flags & O_DIRECT) { 953 if (filp->f_flags & O_DIRECT) {
@@ -1079,27 +1068,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
1079 /* communicate with ocfs2_dio_end_io */ 1068 /* communicate with ocfs2_dio_end_io */
1080 ocfs2_iocb_set_rw_locked(iocb); 1069 ocfs2_iocb_set_rw_locked(iocb);
1081 1070
1082#ifdef OCFS2_ORACORE_WORKAROUNDS 1071 ret = generic_file_aio_write_nolock(iocb, &local_iov, 1, &iocb->ki_pos);
1083 if (osb->s_mount_opt & OCFS2_MOUNT_COMPAT_OCFS &&
1084 filp->f_flags & O_DIRECT) {
1085 unsigned int saved_flags = filp->f_flags;
1086 int sector_size = 1 << osb->s_sectsize_bits;
1087
1088 if ((saved_pos & (sector_size - 1)) ||
1089 (count & (sector_size - 1)) ||
1090 ((unsigned long)buf & (sector_size - 1))) {
1091 filp->f_flags |= O_SYNC;
1092 filp->f_flags &= ~O_DIRECT;
1093 }
1094
1095 ret = generic_file_aio_write_nolock(iocb, &local_iov, 1,
1096 &iocb->ki_pos);
1097
1098 filp->f_flags = saved_flags;
1099 } else
1100#endif
1101 ret = generic_file_aio_write_nolock(iocb, &local_iov, 1,
1102 &iocb->ki_pos);
1103 1072
1104 /* buffered aio wouldn't have proper lock coverage today */ 1073 /* buffered aio wouldn't have proper lock coverage today */
1105 BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT)); 1074 BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT));
@@ -1140,9 +1109,6 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
1140 int ret = 0, rw_level = -1, have_alloc_sem = 0; 1109 int ret = 0, rw_level = -1, have_alloc_sem = 0;
1141 struct file *filp = iocb->ki_filp; 1110 struct file *filp = iocb->ki_filp;
1142 struct inode *inode = filp->f_dentry->d_inode; 1111 struct inode *inode = filp->f_dentry->d_inode;
1143#ifdef OCFS2_ORACORE_WORKAROUNDS
1144 struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
1145#endif
1146 1112
1147 mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf, 1113 mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", filp, buf,
1148 (unsigned int)count, 1114 (unsigned int)count,
@@ -1155,21 +1121,6 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
1155 goto bail; 1121 goto bail;
1156 } 1122 }
1157 1123
1158#ifdef OCFS2_ORACORE_WORKAROUNDS
1159 if (osb->s_mount_opt & OCFS2_MOUNT_COMPAT_OCFS) {
1160 if (filp->f_flags & O_DIRECT) {
1161 int sector_size = 1 << osb->s_sectsize_bits;
1162
1163 if ((pos & (sector_size - 1)) ||
1164 (count & (sector_size - 1)) ||
1165 ((unsigned long)buf & (sector_size - 1)) ||
1166 (i_size_read(inode) & (sector_size -1))) {
1167 filp->f_flags &= ~O_DIRECT;
1168 }
1169 }
1170 }
1171#endif
1172
1173 /* 1124 /*
1174 * buffered reads protect themselves in ->readpage(). O_DIRECT reads 1125 * buffered reads protect themselves in ->readpage(). O_DIRECT reads
1175 * need locks to protect pending reads from racing with truncate. 1126 * need locks to protect pending reads from racing with truncate.
diff --git a/fs/ocfs2/heartbeat.c b/fs/ocfs2/heartbeat.c
index 0bbd22f46c80..cbfd45a97a63 100644
--- a/fs/ocfs2/heartbeat.c
+++ b/fs/ocfs2/heartbeat.c
@@ -67,6 +67,7 @@ void ocfs2_init_node_maps(struct ocfs2_super *osb)
67 ocfs2_node_map_init(&osb->mounted_map); 67 ocfs2_node_map_init(&osb->mounted_map);
68 ocfs2_node_map_init(&osb->recovery_map); 68 ocfs2_node_map_init(&osb->recovery_map);
69 ocfs2_node_map_init(&osb->umount_map); 69 ocfs2_node_map_init(&osb->umount_map);
70 ocfs2_node_map_init(&osb->osb_recovering_orphan_dirs);
70} 71}
71 72
72static void ocfs2_do_node_down(int node_num, 73static void ocfs2_do_node_down(int node_num,
diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c
index 8122489c5762..315472a5c192 100644
--- a/fs/ocfs2/inode.c
+++ b/fs/ocfs2/inode.c
@@ -41,6 +41,7 @@
41#include "dlmglue.h" 41#include "dlmglue.h"
42#include "extent_map.h" 42#include "extent_map.h"
43#include "file.h" 43#include "file.h"
44#include "heartbeat.h"
44#include "inode.h" 45#include "inode.h"
45#include "journal.h" 46#include "journal.h"
46#include "namei.h" 47#include "namei.h"
@@ -544,6 +545,42 @@ bail:
544 return status; 545 return status;
545} 546}
546 547
548/*
549 * Serialize with orphan dir recovery. If the process doing
550 * recovery on this orphan dir does an iget() with the dir
551 * i_mutex held, we'll deadlock here. Instead we detect this
552 * and exit early - recovery will wipe this inode for us.
553 */
554static int ocfs2_check_orphan_recovery_state(struct ocfs2_super *osb,
555 int slot)
556{
557 int ret = 0;
558
559 spin_lock(&osb->osb_lock);
560 if (ocfs2_node_map_test_bit(osb, &osb->osb_recovering_orphan_dirs, slot)) {
561 mlog(0, "Recovery is happening on orphan dir %d, will skip "
562 "this inode\n", slot);
563 ret = -EDEADLK;
564 goto out;
565 }
566 /* This signals to the orphan recovery process that it should
567 * wait for us to handle the wipe. */
568 osb->osb_orphan_wipes[slot]++;
569out:
570 spin_unlock(&osb->osb_lock);
571 return ret;
572}
573
574static void ocfs2_signal_wipe_completion(struct ocfs2_super *osb,
575 int slot)
576{
577 spin_lock(&osb->osb_lock);
578 osb->osb_orphan_wipes[slot]--;
579 spin_unlock(&osb->osb_lock);
580
581 wake_up(&osb->osb_wipe_event);
582}
583
547static int ocfs2_wipe_inode(struct inode *inode, 584static int ocfs2_wipe_inode(struct inode *inode,
548 struct buffer_head *di_bh) 585 struct buffer_head *di_bh)
549{ 586{
@@ -555,6 +592,11 @@ static int ocfs2_wipe_inode(struct inode *inode,
555 /* We've already voted on this so it should be readonly - no 592 /* We've already voted on this so it should be readonly - no
556 * spinlock needed. */ 593 * spinlock needed. */
557 orphaned_slot = OCFS2_I(inode)->ip_orphaned_slot; 594 orphaned_slot = OCFS2_I(inode)->ip_orphaned_slot;
595
596 status = ocfs2_check_orphan_recovery_state(osb, orphaned_slot);
597 if (status)
598 return status;
599
558 orphan_dir_inode = ocfs2_get_system_file_inode(osb, 600 orphan_dir_inode = ocfs2_get_system_file_inode(osb,
559 ORPHAN_DIR_SYSTEM_INODE, 601 ORPHAN_DIR_SYSTEM_INODE,
560 orphaned_slot); 602 orphaned_slot);
@@ -597,6 +639,7 @@ bail_unlock_dir:
597 brelse(orphan_dir_bh); 639 brelse(orphan_dir_bh);
598bail: 640bail:
599 iput(orphan_dir_inode); 641 iput(orphan_dir_inode);
642 ocfs2_signal_wipe_completion(osb, orphaned_slot);
600 643
601 return status; 644 return status;
602} 645}
@@ -822,7 +865,8 @@ void ocfs2_delete_inode(struct inode *inode)
822 865
823 status = ocfs2_wipe_inode(inode, di_bh); 866 status = ocfs2_wipe_inode(inode, di_bh);
824 if (status < 0) { 867 if (status < 0) {
825 mlog_errno(status); 868 if (status != -EDEADLK)
869 mlog_errno(status);
826 goto bail_unlock_inode; 870 goto bail_unlock_inode;
827 } 871 }
828 872
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index d329c9df90ae..4be801f4559b 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -1408,21 +1408,17 @@ bail:
1408 return status; 1408 return status;
1409} 1409}
1410 1410
1411static int ocfs2_recover_orphans(struct ocfs2_super *osb, 1411static int ocfs2_queue_orphans(struct ocfs2_super *osb,
1412 int slot) 1412 int slot,
1413 struct inode **head)
1413{ 1414{
1414 int status = 0; 1415 int status;
1415 int have_disk_lock = 0;
1416 struct inode *inode = NULL;
1417 struct inode *iter;
1418 struct inode *orphan_dir_inode = NULL; 1416 struct inode *orphan_dir_inode = NULL;
1417 struct inode *iter;
1419 unsigned long offset, blk, local; 1418 unsigned long offset, blk, local;
1420 struct buffer_head *bh = NULL; 1419 struct buffer_head *bh = NULL;
1421 struct ocfs2_dir_entry *de; 1420 struct ocfs2_dir_entry *de;
1422 struct super_block *sb = osb->sb; 1421 struct super_block *sb = osb->sb;
1423 struct ocfs2_inode_info *oi;
1424
1425 mlog(0, "Recover inodes from orphan dir in slot %d\n", slot);
1426 1422
1427 orphan_dir_inode = ocfs2_get_system_file_inode(osb, 1423 orphan_dir_inode = ocfs2_get_system_file_inode(osb,
1428 ORPHAN_DIR_SYSTEM_INODE, 1424 ORPHAN_DIR_SYSTEM_INODE,
@@ -1430,17 +1426,15 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1430 if (!orphan_dir_inode) { 1426 if (!orphan_dir_inode) {
1431 status = -ENOENT; 1427 status = -ENOENT;
1432 mlog_errno(status); 1428 mlog_errno(status);
1433 goto out; 1429 return status;
1434 } 1430 }
1435 1431
1436 mutex_lock(&orphan_dir_inode->i_mutex); 1432 mutex_lock(&orphan_dir_inode->i_mutex);
1437 status = ocfs2_meta_lock(orphan_dir_inode, NULL, NULL, 0); 1433 status = ocfs2_meta_lock(orphan_dir_inode, NULL, NULL, 0);
1438 if (status < 0) { 1434 if (status < 0) {
1439 mutex_unlock(&orphan_dir_inode->i_mutex);
1440 mlog_errno(status); 1435 mlog_errno(status);
1441 goto out; 1436 goto out;
1442 } 1437 }
1443 have_disk_lock = 1;
1444 1438
1445 offset = 0; 1439 offset = 0;
1446 iter = NULL; 1440 iter = NULL;
@@ -1451,11 +1445,10 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1451 if (!bh) 1445 if (!bh)
1452 status = -EINVAL; 1446 status = -EINVAL;
1453 if (status < 0) { 1447 if (status < 0) {
1454 mutex_unlock(&orphan_dir_inode->i_mutex);
1455 if (bh) 1448 if (bh)
1456 brelse(bh); 1449 brelse(bh);
1457 mlog_errno(status); 1450 mlog_errno(status);
1458 goto out; 1451 goto out_unlock;
1459 } 1452 }
1460 1453
1461 local = 0; 1454 local = 0;
@@ -1465,11 +1458,10 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1465 1458
1466 if (!ocfs2_check_dir_entry(orphan_dir_inode, 1459 if (!ocfs2_check_dir_entry(orphan_dir_inode,
1467 de, bh, local)) { 1460 de, bh, local)) {
1468 mutex_unlock(&orphan_dir_inode->i_mutex);
1469 status = -EINVAL; 1461 status = -EINVAL;
1470 mlog_errno(status); 1462 mlog_errno(status);
1471 brelse(bh); 1463 brelse(bh);
1472 goto out; 1464 goto out_unlock;
1473 } 1465 }
1474 1466
1475 local += le16_to_cpu(de->rec_len); 1467 local += le16_to_cpu(de->rec_len);
@@ -1504,18 +1496,95 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1504 1496
1505 mlog(0, "queue orphan %"MLFu64"\n", 1497 mlog(0, "queue orphan %"MLFu64"\n",
1506 OCFS2_I(iter)->ip_blkno); 1498 OCFS2_I(iter)->ip_blkno);
1507 OCFS2_I(iter)->ip_next_orphan = inode; 1499 /* No locking is required for the next_orphan
1508 inode = iter; 1500 * queue as there is only ever a single
1501 * process doing orphan recovery. */
1502 OCFS2_I(iter)->ip_next_orphan = *head;
1503 *head = iter;
1509 } 1504 }
1510 brelse(bh); 1505 brelse(bh);
1511 } 1506 }
1512 mutex_unlock(&orphan_dir_inode->i_mutex);
1513 1507
1508out_unlock:
1514 ocfs2_meta_unlock(orphan_dir_inode, 0); 1509 ocfs2_meta_unlock(orphan_dir_inode, 0);
1515 have_disk_lock = 0; 1510out:
1516 1511 mutex_unlock(&orphan_dir_inode->i_mutex);
1517 iput(orphan_dir_inode); 1512 iput(orphan_dir_inode);
1518 orphan_dir_inode = NULL; 1513 return status;
1514}
1515
1516static int ocfs2_orphan_recovery_can_continue(struct ocfs2_super *osb,
1517 int slot)
1518{
1519 int ret;
1520
1521 spin_lock(&osb->osb_lock);
1522 ret = !osb->osb_orphan_wipes[slot];
1523 spin_unlock(&osb->osb_lock);
1524 return ret;
1525}
1526
1527static void ocfs2_mark_recovering_orphan_dir(struct ocfs2_super *osb,
1528 int slot)
1529{
1530 spin_lock(&osb->osb_lock);
1531 /* Mark ourselves such that new processes in delete_inode()
1532 * know to quit early. */
1533 ocfs2_node_map_set_bit(osb, &osb->osb_recovering_orphan_dirs, slot);
1534 while (osb->osb_orphan_wipes[slot]) {
1535 /* If any processes are already in the middle of an
1536 * orphan wipe on this dir, then we need to wait for
1537 * them. */
1538 spin_unlock(&osb->osb_lock);
1539 wait_event_interruptible(osb->osb_wipe_event,
1540 ocfs2_orphan_recovery_can_continue(osb, slot));
1541 spin_lock(&osb->osb_lock);
1542 }
1543 spin_unlock(&osb->osb_lock);
1544}
1545
1546static void ocfs2_clear_recovering_orphan_dir(struct ocfs2_super *osb,
1547 int slot)
1548{
1549 ocfs2_node_map_clear_bit(osb, &osb->osb_recovering_orphan_dirs, slot);
1550}
1551
1552/*
1553 * Orphan recovery. Each mounted node has it's own orphan dir which we
1554 * must run during recovery. Our strategy here is to build a list of
1555 * the inodes in the orphan dir and iget/iput them. The VFS does
1556 * (most) of the rest of the work.
1557 *
1558 * Orphan recovery can happen at any time, not just mount so we have a
1559 * couple of extra considerations.
1560 *
1561 * - We grab as many inodes as we can under the orphan dir lock -
1562 * doing iget() outside the orphan dir risks getting a reference on
1563 * an invalid inode.
1564 * - We must be sure not to deadlock with other processes on the
1565 * system wanting to run delete_inode(). This can happen when they go
1566 * to lock the orphan dir and the orphan recovery process attempts to
1567 * iget() inside the orphan dir lock. This can be avoided by
1568 * advertising our state to ocfs2_delete_inode().
1569 */
1570static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1571 int slot)
1572{
1573 int ret = 0;
1574 struct inode *inode = NULL;
1575 struct inode *iter;
1576 struct ocfs2_inode_info *oi;
1577
1578 mlog(0, "Recover inodes from orphan dir in slot %d\n", slot);
1579
1580 ocfs2_mark_recovering_orphan_dir(osb, slot);
1581 ret = ocfs2_queue_orphans(osb, slot, &inode);
1582 ocfs2_clear_recovering_orphan_dir(osb, slot);
1583
1584 /* Error here should be noted, but we want to continue with as
1585 * many queued inodes as we've got. */
1586 if (ret)
1587 mlog_errno(ret);
1519 1588
1520 while (inode) { 1589 while (inode) {
1521 oi = OCFS2_I(inode); 1590 oi = OCFS2_I(inode);
@@ -1541,14 +1610,7 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
1541 inode = iter; 1610 inode = iter;
1542 } 1611 }
1543 1612
1544out: 1613 return ret;
1545 if (have_disk_lock)
1546 ocfs2_meta_unlock(orphan_dir_inode, 0);
1547
1548 if (orphan_dir_inode)
1549 iput(orphan_dir_inode);
1550
1551 return status;
1552} 1614}
1553 1615
1554static int ocfs2_wait_on_mount(struct ocfs2_super *osb) 1616static int ocfs2_wait_on_mount(struct ocfs2_super *osb)
diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h
index 8d8e4779df92..e89de9b6e491 100644
--- a/fs/ocfs2/ocfs2.h
+++ b/fs/ocfs2/ocfs2.h
@@ -174,9 +174,6 @@ enum ocfs2_mount_options
174 OCFS2_MOUNT_NOINTR = 1 << 2, /* Don't catch signals */ 174 OCFS2_MOUNT_NOINTR = 1 << 2, /* Don't catch signals */
175 OCFS2_MOUNT_ERRORS_PANIC = 1 << 3, /* Panic on errors */ 175 OCFS2_MOUNT_ERRORS_PANIC = 1 << 3, /* Panic on errors */
176 OCFS2_MOUNT_DATA_WRITEBACK = 1 << 4, /* No data ordering */ 176 OCFS2_MOUNT_DATA_WRITEBACK = 1 << 4, /* No data ordering */
177#ifdef OCFS2_ORACORE_WORKAROUNDS
178 OCFS2_MOUNT_COMPAT_OCFS = 1 << 30, /* ocfs1 compatibility mode */
179#endif
180}; 177};
181 178
182#define OCFS2_OSB_SOFT_RO 0x0001 179#define OCFS2_OSB_SOFT_RO 0x0001
@@ -290,6 +287,10 @@ struct ocfs2_super
290 struct inode *osb_tl_inode; 287 struct inode *osb_tl_inode;
291 struct buffer_head *osb_tl_bh; 288 struct buffer_head *osb_tl_bh;
292 struct work_struct osb_truncate_log_wq; 289 struct work_struct osb_truncate_log_wq;
290
291 struct ocfs2_node_map osb_recovering_orphan_dirs;
292 unsigned int *osb_orphan_wipes;
293 wait_queue_head_t osb_wipe_event;
293}; 294};
294 295
295#define OCFS2_SB(sb) ((struct ocfs2_super *)(sb)->s_fs_info) 296#define OCFS2_SB(sb) ((struct ocfs2_super *)(sb)->s_fs_info)
diff --git a/fs/ocfs2/ocfs2_fs.h b/fs/ocfs2/ocfs2_fs.h
index dfb8a5bedfc8..c5b1ac547c15 100644
--- a/fs/ocfs2/ocfs2_fs.h
+++ b/fs/ocfs2/ocfs2_fs.h
@@ -138,7 +138,6 @@
138 138
139/* Journal limits (in bytes) */ 139/* Journal limits (in bytes) */
140#define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024) 140#define OCFS2_MIN_JOURNAL_SIZE (4 * 1024 * 1024)
141#define OCFS2_MAX_JOURNAL_SIZE (500 * 1024 * 1024)
142 141
143struct ocfs2_system_inode_info { 142struct ocfs2_system_inode_info {
144 char *si_name; 143 char *si_name;
diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c
index 046824b6b625..8dd3aafec499 100644
--- a/fs/ocfs2/super.c
+++ b/fs/ocfs2/super.c
@@ -1325,6 +1325,16 @@ static int ocfs2_initialize_super(struct super_block *sb,
1325 } 1325 }
1326 mlog(ML_NOTICE, "max_slots for this device: %u\n", osb->max_slots); 1326 mlog(ML_NOTICE, "max_slots for this device: %u\n", osb->max_slots);
1327 1327
1328 init_waitqueue_head(&osb->osb_wipe_event);
1329 osb->osb_orphan_wipes = kcalloc(osb->max_slots,
1330 sizeof(*osb->osb_orphan_wipes),
1331 GFP_KERNEL);
1332 if (!osb->osb_orphan_wipes) {
1333 status = -ENOMEM;
1334 mlog_errno(status);
1335 goto bail;
1336 }
1337
1328 osb->s_feature_compat = 1338 osb->s_feature_compat =
1329 le32_to_cpu(OCFS2_RAW_SB(di)->s_feature_compat); 1339 le32_to_cpu(OCFS2_RAW_SB(di)->s_feature_compat);
1330 osb->s_feature_ro_compat = 1340 osb->s_feature_ro_compat =
@@ -1638,6 +1648,7 @@ static void ocfs2_delete_osb(struct ocfs2_super *osb)
1638 if (osb->slot_info) 1648 if (osb->slot_info)
1639 ocfs2_free_slot_info(osb->slot_info); 1649 ocfs2_free_slot_info(osb->slot_info);
1640 1650
1651 kfree(osb->osb_orphan_wipes);
1641 /* FIXME 1652 /* FIXME
1642 * This belongs in journal shutdown, but because we have to 1653 * This belongs in journal shutdown, but because we have to
1643 * allocate osb->journal at the start of ocfs2_initalize_osb(), 1654 * allocate osb->journal at the start of ocfs2_initalize_osb(),
diff --git a/fs/partitions/ibm.c b/fs/partitions/ibm.c
index 78010ad60e47..1e4a93835fed 100644
--- a/fs/partitions/ibm.c
+++ b/fs/partitions/ibm.c
@@ -52,6 +52,7 @@ int
52ibm_partition(struct parsed_partitions *state, struct block_device *bdev) 52ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
53{ 53{
54 int blocksize, offset, size; 54 int blocksize, offset, size;
55 loff_t i_size;
55 dasd_information_t *info; 56 dasd_information_t *info;
56 struct hd_geometry *geo; 57 struct hd_geometry *geo;
57 char type[5] = {0,}; 58 char type[5] = {0,};
@@ -63,6 +64,13 @@ ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
63 unsigned char *data; 64 unsigned char *data;
64 Sector sect; 65 Sector sect;
65 66
67 blocksize = bdev_hardsect_size(bdev);
68 if (blocksize <= 0)
69 return 0;
70 i_size = i_size_read(bdev->bd_inode);
71 if (i_size == 0)
72 return 0;
73
66 if ((info = kmalloc(sizeof(dasd_information_t), GFP_KERNEL)) == NULL) 74 if ((info = kmalloc(sizeof(dasd_information_t), GFP_KERNEL)) == NULL)
67 goto out_noinfo; 75 goto out_noinfo;
68 if ((geo = kmalloc(sizeof(struct hd_geometry), GFP_KERNEL)) == NULL) 76 if ((geo = kmalloc(sizeof(struct hd_geometry), GFP_KERNEL)) == NULL)
@@ -73,9 +81,6 @@ ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
73 if (ioctl_by_bdev(bdev, BIODASDINFO, (unsigned long)info) != 0 || 81 if (ioctl_by_bdev(bdev, BIODASDINFO, (unsigned long)info) != 0 ||
74 ioctl_by_bdev(bdev, HDIO_GETGEO, (unsigned long)geo) != 0) 82 ioctl_by_bdev(bdev, HDIO_GETGEO, (unsigned long)geo) != 0)
75 goto out_noioctl; 83 goto out_noioctl;
76
77 if ((blocksize = bdev_hardsect_size(bdev)) <= 0)
78 goto out_badsect;
79 84
80 /* 85 /*
81 * Get volume label, extract name and type. 86 * Get volume label, extract name and type.
@@ -111,7 +116,7 @@ ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
111 } else { 116 } else {
112 printk("CMS1/%8s:", name); 117 printk("CMS1/%8s:", name);
113 offset = (info->label_block + 1); 118 offset = (info->label_block + 1);
114 size = bdev->bd_inode->i_size >> 9; 119 size = i_size >> 9;
115 } 120 }
116 put_partition(state, 1, offset*(blocksize >> 9), 121 put_partition(state, 1, offset*(blocksize >> 9),
117 size-offset*(blocksize >> 9)); 122 size-offset*(blocksize >> 9));
@@ -168,7 +173,7 @@ ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
168 else 173 else
169 printk("(nonl)/%8s:", name); 174 printk("(nonl)/%8s:", name);
170 offset = (info->label_block + 1); 175 offset = (info->label_block + 1);
171 size = (bdev->bd_inode->i_size >> 9); 176 size = i_size >> 9;
172 put_partition(state, 1, offset*(blocksize >> 9), 177 put_partition(state, 1, offset*(blocksize >> 9),
173 size-offset*(blocksize >> 9)); 178 size-offset*(blocksize >> 9));
174 } 179 }
@@ -180,7 +185,6 @@ ibm_partition(struct parsed_partitions *state, struct block_device *bdev)
180 return 1; 185 return 1;
181 186
182out_readerr: 187out_readerr:
183out_badsect:
184out_noioctl: 188out_noioctl:
185 kfree(label); 189 kfree(label);
186out_nolab: 190out_nolab:
diff --git a/fs/pipe.c b/fs/pipe.c
index d722579df79a..8aada8e426f4 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -605,7 +605,7 @@ struct file_operations rdwr_fifo_fops = {
605 .fasync = pipe_rdwr_fasync, 605 .fasync = pipe_rdwr_fasync,
606}; 606};
607 607
608struct file_operations read_pipe_fops = { 608static struct file_operations read_pipe_fops = {
609 .llseek = no_llseek, 609 .llseek = no_llseek,
610 .read = pipe_read, 610 .read = pipe_read,
611 .readv = pipe_readv, 611 .readv = pipe_readv,
@@ -617,7 +617,7 @@ struct file_operations read_pipe_fops = {
617 .fasync = pipe_read_fasync, 617 .fasync = pipe_read_fasync,
618}; 618};
619 619
620struct file_operations write_pipe_fops = { 620static struct file_operations write_pipe_fops = {
621 .llseek = no_llseek, 621 .llseek = no_llseek,
622 .read = bad_pipe_r, 622 .read = bad_pipe_r,
623 .write = pipe_write, 623 .write = pipe_write,
@@ -629,7 +629,7 @@ struct file_operations write_pipe_fops = {
629 .fasync = pipe_write_fasync, 629 .fasync = pipe_write_fasync,
630}; 630};
631 631
632struct file_operations rdwr_pipe_fops = { 632static struct file_operations rdwr_pipe_fops = {
633 .llseek = no_llseek, 633 .llseek = no_llseek,
634 .read = pipe_read, 634 .read = pipe_read,
635 .readv = pipe_readv, 635 .readv = pipe_readv,
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 0eaad41f4658..91b7c15ab373 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -204,7 +204,6 @@ static void smaps_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
204{ 204{
205 pte_t *pte, ptent; 205 pte_t *pte, ptent;
206 spinlock_t *ptl; 206 spinlock_t *ptl;
207 unsigned long pfn;
208 struct page *page; 207 struct page *page;
209 208
210 pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); 209 pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
@@ -214,12 +213,12 @@ static void smaps_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
214 continue; 213 continue;
215 214
216 mss->resident += PAGE_SIZE; 215 mss->resident += PAGE_SIZE;
217 pfn = pte_pfn(ptent); 216
218 if (!pfn_valid(pfn)) 217 page = vm_normal_page(vma, addr, ptent);
218 if (!page)
219 continue; 219 continue;
220 220
221 page = pfn_to_page(pfn); 221 if (page_mapcount(page) >= 2) {
222 if (page_count(page) >= 2) {
223 if (pte_dirty(ptent)) 222 if (pte_dirty(ptent))
224 mss->shared_dirty += PAGE_SIZE; 223 mss->shared_dirty += PAGE_SIZE;
225 else 224 else
@@ -289,7 +288,7 @@ static int show_smap(struct seq_file *m, void *v)
289 struct mem_size_stats mss; 288 struct mem_size_stats mss;
290 289
291 memset(&mss, 0, sizeof mss); 290 memset(&mss, 0, sizeof mss);
292 if (vma->vm_mm) 291 if (vma->vm_mm && !is_vm_hugetlb_page(vma))
293 smaps_pgd_range(vma, vma->vm_start, vma->vm_end, &mss); 292 smaps_pgd_range(vma, vma->vm_start, vma->vm_end, &mss);
294 return show_map_internal(m, v, &mss); 293 return show_map_internal(m, v, &mss);
295} 294}
diff --git a/fs/ramfs/inode.c b/fs/ramfs/inode.c
index cde5d48994ae..14bd2246fb6d 100644
--- a/fs/ramfs/inode.c
+++ b/fs/ramfs/inode.c
@@ -137,6 +137,7 @@ static int ramfs_symlink(struct inode * dir, struct dentry *dentry, const char *
137 inode->i_gid = dir->i_gid; 137 inode->i_gid = dir->i_gid;
138 d_instantiate(dentry, inode); 138 d_instantiate(dentry, inode);
139 dget(dentry); 139 dget(dentry);
140 dir->i_mtime = dir->i_ctime = CURRENT_TIME;
140 } else 141 } else
141 iput(inode); 142 iput(inode);
142 } 143 }
diff --git a/fs/reiserfs/file.c b/fs/reiserfs/file.c
index f3473176c83a..be12879bb179 100644
--- a/fs/reiserfs/file.c
+++ b/fs/reiserfs/file.c
@@ -1464,13 +1464,11 @@ static ssize_t reiserfs_file_write(struct file *file, /* the file we are going t
1464 partially overwritten pages, if needed. And lock the pages, 1464 partially overwritten pages, if needed. And lock the pages,
1465 so that nobody else can access these until we are done. 1465 so that nobody else can access these until we are done.
1466 We get number of actual blocks needed as a result. */ 1466 We get number of actual blocks needed as a result. */
1467 blocks_to_allocate = 1467 res = reiserfs_prepare_file_region_for_write(inode, pos,
1468 reiserfs_prepare_file_region_for_write(inode, pos, 1468 num_pages,
1469 num_pages, 1469 write_bytes,
1470 write_bytes, 1470 prepared_pages);
1471 prepared_pages); 1471 if (res < 0) {
1472 if (blocks_to_allocate < 0) {
1473 res = blocks_to_allocate;
1474 reiserfs_release_claimed_blocks(inode->i_sb, 1472 reiserfs_release_claimed_blocks(inode->i_sb,
1475 num_pages << 1473 num_pages <<
1476 (PAGE_CACHE_SHIFT - 1474 (PAGE_CACHE_SHIFT -
@@ -1478,6 +1476,8 @@ static ssize_t reiserfs_file_write(struct file *file, /* the file we are going t
1478 break; 1476 break;
1479 } 1477 }
1480 1478
1479 blocks_to_allocate = res;
1480
1481 /* First we correct our estimate of how many blocks we need */ 1481 /* First we correct our estimate of how many blocks we need */
1482 reiserfs_release_claimed_blocks(inode->i_sb, 1482 reiserfs_release_claimed_blocks(inode->i_sb,
1483 (num_pages << 1483 (num_pages <<
diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c
index b33d67bba2fd..d60f6238c66a 100644
--- a/fs/reiserfs/inode.c
+++ b/fs/reiserfs/inode.c
@@ -627,11 +627,6 @@ int reiserfs_get_block(struct inode *inode, sector_t block,
627 reiserfs_write_lock(inode->i_sb); 627 reiserfs_write_lock(inode->i_sb);
628 version = get_inode_item_key_version(inode); 628 version = get_inode_item_key_version(inode);
629 629
630 if (block < 0) {
631 reiserfs_write_unlock(inode->i_sb);
632 return -EIO;
633 }
634
635 if (!file_capable(inode, block)) { 630 if (!file_capable(inode, block)) {
636 reiserfs_write_unlock(inode->i_sb); 631 reiserfs_write_unlock(inode->i_sb);
637 return -EFBIG; 632 return -EFBIG;
@@ -934,12 +929,13 @@ int reiserfs_get_block(struct inode *inode, sector_t block,
934 //pos_in_item * inode->i_sb->s_blocksize, 929 //pos_in_item * inode->i_sb->s_blocksize,
935 TYPE_INDIRECT, 3); // key type is unimportant 930 TYPE_INDIRECT, 3); // key type is unimportant
936 931
932 RFALSE(cpu_key_k_offset(&tmp_key) > cpu_key_k_offset(&key),
933 "green-805: invalid offset");
937 blocks_needed = 934 blocks_needed =
938 1 + 935 1 +
939 ((cpu_key_k_offset(&key) - 936 ((cpu_key_k_offset(&key) -
940 cpu_key_k_offset(&tmp_key)) >> inode->i_sb-> 937 cpu_key_k_offset(&tmp_key)) >> inode->i_sb->
941 s_blocksize_bits); 938 s_blocksize_bits);
942 RFALSE(blocks_needed < 0, "green-805: invalid offset");
943 939
944 if (blocks_needed == 1) { 940 if (blocks_needed == 1) {
945 un = &unf_single; 941 un = &unf_single;
diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c
index b7a179560ab4..5a9d2722fa0a 100644
--- a/fs/reiserfs/journal.c
+++ b/fs/reiserfs/journal.c
@@ -2319,8 +2319,7 @@ static int journal_read(struct super_block *p_s_sb)
2319 return 1; 2319 return 1;
2320 } 2320 }
2321 jh = (struct reiserfs_journal_header *)(journal->j_header_bh->b_data); 2321 jh = (struct reiserfs_journal_header *)(journal->j_header_bh->b_data);
2322 if (le32_to_cpu(jh->j_first_unflushed_offset) >= 0 && 2322 if (le32_to_cpu(jh->j_first_unflushed_offset) <
2323 le32_to_cpu(jh->j_first_unflushed_offset) <
2324 SB_ONDISK_JOURNAL_SIZE(p_s_sb) 2323 SB_ONDISK_JOURNAL_SIZE(p_s_sb)
2325 && le32_to_cpu(jh->j_last_flush_trans_id) > 0) { 2324 && le32_to_cpu(jh->j_last_flush_trans_id) > 0) {
2326 oldest_start = 2325 oldest_start =
diff --git a/fs/reiserfs/namei.c b/fs/reiserfs/namei.c
index c8123308e060..284f7852de8b 100644
--- a/fs/reiserfs/namei.c
+++ b/fs/reiserfs/namei.c
@@ -247,7 +247,7 @@ static int linear_search_in_dir_item(struct cpu_key *key,
247 /* mark, that this generation number is used */ 247 /* mark, that this generation number is used */
248 if (de->de_gen_number_bit_string) 248 if (de->de_gen_number_bit_string)
249 set_bit(GET_GENERATION_NUMBER(deh_offset(deh)), 249 set_bit(GET_GENERATION_NUMBER(deh_offset(deh)),
250 (unsigned long *)de->de_gen_number_bit_string); 250 de->de_gen_number_bit_string);
251 251
252 // calculate pointer to name and namelen 252 // calculate pointer to name and namelen
253 de->de_entry_num = i; 253 de->de_entry_num = i;
@@ -431,7 +431,7 @@ static int reiserfs_add_entry(struct reiserfs_transaction_handle *th,
431 struct reiserfs_de_head *deh; 431 struct reiserfs_de_head *deh;
432 INITIALIZE_PATH(path); 432 INITIALIZE_PATH(path);
433 struct reiserfs_dir_entry de; 433 struct reiserfs_dir_entry de;
434 int bit_string[MAX_GENERATION_NUMBER / (sizeof(int) * 8) + 1]; 434 DECLARE_BITMAP(bit_string, MAX_GENERATION_NUMBER + 1);
435 int gen_number; 435 int gen_number;
436 char small_buf[32 + DEH_SIZE]; /* 48 bytes now and we avoid kmalloc 436 char small_buf[32 + DEH_SIZE]; /* 48 bytes now and we avoid kmalloc
437 if we create file with short name */ 437 if we create file with short name */
@@ -486,7 +486,7 @@ static int reiserfs_add_entry(struct reiserfs_transaction_handle *th,
486 486
487 /* find the proper place for the new entry */ 487 /* find the proper place for the new entry */
488 memset(bit_string, 0, sizeof(bit_string)); 488 memset(bit_string, 0, sizeof(bit_string));
489 de.de_gen_number_bit_string = (char *)bit_string; 489 de.de_gen_number_bit_string = bit_string;
490 retval = reiserfs_find_entry(dir, name, namelen, &path, &de); 490 retval = reiserfs_find_entry(dir, name, namelen, &path, &de);
491 if (retval != NAME_NOT_FOUND) { 491 if (retval != NAME_NOT_FOUND) {
492 if (buffer != small_buf) 492 if (buffer != small_buf)
@@ -508,7 +508,7 @@ static int reiserfs_add_entry(struct reiserfs_transaction_handle *th,
508 } 508 }
509 509
510 gen_number = 510 gen_number =
511 find_first_zero_bit((unsigned long *)bit_string, 511 find_first_zero_bit(bit_string,
512 MAX_GENERATION_NUMBER + 1); 512 MAX_GENERATION_NUMBER + 1);
513 if (gen_number > MAX_GENERATION_NUMBER) { 513 if (gen_number > MAX_GENERATION_NUMBER) {
514 /* there is no free generation number */ 514 /* there is no free generation number */
diff --git a/fs/udf/inode.c b/fs/udf/inode.c
index 395e582ee542..d04cff2273b6 100644
--- a/fs/udf/inode.c
+++ b/fs/udf/inode.c
@@ -1045,10 +1045,14 @@ static void udf_fill_inode(struct inode *inode, struct buffer_head *bh)
1045 } 1045 }
1046 1046
1047 inode->i_uid = le32_to_cpu(fe->uid); 1047 inode->i_uid = le32_to_cpu(fe->uid);
1048 if ( inode->i_uid == -1 ) inode->i_uid = UDF_SB(inode->i_sb)->s_uid; 1048 if (inode->i_uid == -1 || UDF_QUERY_FLAG(inode->i_sb,
1049 UDF_FLAG_UID_IGNORE))
1050 inode->i_uid = UDF_SB(inode->i_sb)->s_uid;
1049 1051
1050 inode->i_gid = le32_to_cpu(fe->gid); 1052 inode->i_gid = le32_to_cpu(fe->gid);
1051 if ( inode->i_gid == -1 ) inode->i_gid = UDF_SB(inode->i_sb)->s_gid; 1053 if (inode->i_gid == -1 || UDF_QUERY_FLAG(inode->i_sb,
1054 UDF_FLAG_GID_IGNORE))
1055 inode->i_gid = UDF_SB(inode->i_sb)->s_gid;
1052 1056
1053 inode->i_nlink = le16_to_cpu(fe->fileLinkCount); 1057 inode->i_nlink = le16_to_cpu(fe->fileLinkCount);
1054 if (!inode->i_nlink) 1058 if (!inode->i_nlink)
@@ -1335,10 +1339,14 @@ udf_update_inode(struct inode *inode, int do_sync)
1335 return err; 1339 return err;
1336 } 1340 }
1337 1341
1338 if (inode->i_uid != UDF_SB(inode->i_sb)->s_uid) 1342 if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_UID_FORGET))
1343 fe->uid = cpu_to_le32(-1);
1344 else if (inode->i_uid != UDF_SB(inode->i_sb)->s_uid)
1339 fe->uid = cpu_to_le32(inode->i_uid); 1345 fe->uid = cpu_to_le32(inode->i_uid);
1340 1346
1341 if (inode->i_gid != UDF_SB(inode->i_sb)->s_gid) 1347 if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_GID_FORGET))
1348 fe->gid = cpu_to_le32(-1);
1349 else if (inode->i_gid != UDF_SB(inode->i_sb)->s_gid)
1342 fe->gid = cpu_to_le32(inode->i_gid); 1350 fe->gid = cpu_to_le32(inode->i_gid);
1343 1351
1344 udfperms = ((inode->i_mode & S_IRWXO) ) | 1352 udfperms = ((inode->i_mode & S_IRWXO) ) |
diff --git a/fs/udf/super.c b/fs/udf/super.c
index 4a6f49adc609..368d8f81fe54 100644
--- a/fs/udf/super.c
+++ b/fs/udf/super.c
@@ -269,7 +269,7 @@ enum {
269 Opt_gid, Opt_uid, Opt_umask, Opt_session, Opt_lastblock, 269 Opt_gid, Opt_uid, Opt_umask, Opt_session, Opt_lastblock,
270 Opt_anchor, Opt_volume, Opt_partition, Opt_fileset, 270 Opt_anchor, Opt_volume, Opt_partition, Opt_fileset,
271 Opt_rootdir, Opt_utf8, Opt_iocharset, 271 Opt_rootdir, Opt_utf8, Opt_iocharset,
272 Opt_err 272 Opt_err, Opt_uforget, Opt_uignore, Opt_gforget, Opt_gignore
273}; 273};
274 274
275static match_table_t tokens = { 275static match_table_t tokens = {
@@ -282,6 +282,10 @@ static match_table_t tokens = {
282 {Opt_adinicb, "adinicb"}, 282 {Opt_adinicb, "adinicb"},
283 {Opt_shortad, "shortad"}, 283 {Opt_shortad, "shortad"},
284 {Opt_longad, "longad"}, 284 {Opt_longad, "longad"},
285 {Opt_uforget, "uid=forget"},
286 {Opt_uignore, "uid=ignore"},
287 {Opt_gforget, "gid=forget"},
288 {Opt_gignore, "gid=ignore"},
285 {Opt_gid, "gid=%u"}, 289 {Opt_gid, "gid=%u"},
286 {Opt_uid, "uid=%u"}, 290 {Opt_uid, "uid=%u"},
287 {Opt_umask, "umask=%o"}, 291 {Opt_umask, "umask=%o"},
@@ -414,6 +418,18 @@ udf_parse_options(char *options, struct udf_options *uopt)
414 uopt->flags |= (1 << UDF_FLAG_NLS_MAP); 418 uopt->flags |= (1 << UDF_FLAG_NLS_MAP);
415 break; 419 break;
416#endif 420#endif
421 case Opt_uignore:
422 uopt->flags |= (1 << UDF_FLAG_UID_IGNORE);
423 break;
424 case Opt_uforget:
425 uopt->flags |= (1 << UDF_FLAG_UID_FORGET);
426 break;
427 case Opt_gignore:
428 uopt->flags |= (1 << UDF_FLAG_GID_IGNORE);
429 break;
430 case Opt_gforget:
431 uopt->flags |= (1 << UDF_FLAG_GID_FORGET);
432 break;
417 default: 433 default:
418 printk(KERN_ERR "udf: bad mount option \"%s\" " 434 printk(KERN_ERR "udf: bad mount option \"%s\" "
419 "or missing value\n", p); 435 "or missing value\n", p);
diff --git a/fs/udf/udf_sb.h b/fs/udf/udf_sb.h
index 663669810be6..110f8d62616f 100644
--- a/fs/udf/udf_sb.h
+++ b/fs/udf/udf_sb.h
@@ -20,6 +20,10 @@
20#define UDF_FLAG_VARCONV 8 20#define UDF_FLAG_VARCONV 8
21#define UDF_FLAG_NLS_MAP 9 21#define UDF_FLAG_NLS_MAP 9
22#define UDF_FLAG_UTF8 10 22#define UDF_FLAG_UTF8 10
23#define UDF_FLAG_UID_FORGET 11 /* save -1 for uid to disk */
24#define UDF_FLAG_UID_IGNORE 12 /* use sb uid instead of on disk uid */
25#define UDF_FLAG_GID_FORGET 13
26#define UDF_FLAG_GID_IGNORE 14
23 27
24#define UDF_PART_FLAG_UNALLOC_BITMAP 0x0001 28#define UDF_PART_FLAG_UNALLOC_BITMAP 0x0001
25#define UDF_PART_FLAG_UNALLOC_TABLE 0x0002 29#define UDF_PART_FLAG_UNALLOC_TABLE 0x0002
diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c
index 8f2beec526cf..74d8be87f983 100644
--- a/fs/xfs/linux-2.6/xfs_aops.c
+++ b/fs/xfs/linux-2.6/xfs_aops.c
@@ -540,7 +540,7 @@ xfs_probe_cluster(
540 540
541 /* First sum forwards in this page */ 541 /* First sum forwards in this page */
542 do { 542 do {
543 if (mapped != buffer_mapped(bh)) 543 if (!buffer_uptodate(bh) || (mapped != buffer_mapped(bh)))
544 return total; 544 return total;
545 total += bh->b_size; 545 total += bh->b_size;
546 } while ((bh = bh->b_this_page) != head); 546 } while ((bh = bh->b_this_page) != head);
diff --git a/fs/xfs/quota/xfs_qm.c b/fs/xfs/quota/xfs_qm.c
index 53a00fb217fa..7c0e39dc6189 100644
--- a/fs/xfs/quota/xfs_qm.c
+++ b/fs/xfs/quota/xfs_qm.c
@@ -68,6 +68,9 @@ kmem_zone_t *qm_dqzone;
68kmem_zone_t *qm_dqtrxzone; 68kmem_zone_t *qm_dqtrxzone;
69STATIC kmem_shaker_t xfs_qm_shaker; 69STATIC kmem_shaker_t xfs_qm_shaker;
70 70
71STATIC cred_t xfs_zerocr;
72STATIC xfs_inode_t xfs_zeroino;
73
71STATIC void xfs_qm_list_init(xfs_dqlist_t *, char *, int); 74STATIC void xfs_qm_list_init(xfs_dqlist_t *, char *, int);
72STATIC void xfs_qm_list_destroy(xfs_dqlist_t *); 75STATIC void xfs_qm_list_destroy(xfs_dqlist_t *);
73 76
@@ -1393,8 +1396,6 @@ xfs_qm_qino_alloc(
1393 xfs_trans_t *tp; 1396 xfs_trans_t *tp;
1394 int error; 1397 int error;
1395 unsigned long s; 1398 unsigned long s;
1396 cred_t zerocr;
1397 xfs_inode_t zeroino;
1398 int committed; 1399 int committed;
1399 1400
1400 tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QINOCREATE); 1401 tp = xfs_trans_alloc(mp, XFS_TRANS_QM_QINOCREATE);
@@ -1406,11 +1407,9 @@ xfs_qm_qino_alloc(
1406 xfs_trans_cancel(tp, 0); 1407 xfs_trans_cancel(tp, 0);
1407 return error; 1408 return error;
1408 } 1409 }
1409 memset(&zerocr, 0, sizeof(zerocr));
1410 memset(&zeroino, 0, sizeof(zeroino));
1411 1410
1412 if ((error = xfs_dir_ialloc(&tp, &zeroino, S_IFREG, 1, 0, 1411 if ((error = xfs_dir_ialloc(&tp, &xfs_zeroino, S_IFREG, 1, 0,
1413 &zerocr, 0, 1, ip, &committed))) { 1412 &xfs_zerocr, 0, 1, ip, &committed))) {
1414 xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES | 1413 xfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES |
1415 XFS_TRANS_ABORT); 1414 XFS_TRANS_ABORT);
1416 return error; 1415 return error;
diff --git a/fs/xfs/xfs_rtalloc.c b/fs/xfs/xfs_rtalloc.c
index 06fc061c50fc..5b413946b1c5 100644
--- a/fs/xfs/xfs_rtalloc.c
+++ b/fs/xfs/xfs_rtalloc.c
@@ -130,7 +130,8 @@ xfs_growfs_rt_alloc(
130 /* 130 /*
131 * Lock the inode. 131 * Lock the inode.
132 */ 132 */
133 if ((error = xfs_trans_iget(mp, tp, ino, 0, XFS_ILOCK_EXCL, &ip))) 133 if ((error = xfs_trans_iget(mp, tp, ino, 0,
134 XFS_ILOCK_EXCL, &ip)))
134 goto error_exit; 135 goto error_exit;
135 XFS_BMAP_INIT(&flist, &firstblock); 136 XFS_BMAP_INIT(&flist, &firstblock);
136 /* 137 /*
@@ -170,8 +171,8 @@ xfs_growfs_rt_alloc(
170 /* 171 /*
171 * Lock the bitmap inode. 172 * Lock the bitmap inode.
172 */ 173 */
173 if ((error = xfs_trans_iget(mp, tp, ino, 0, XFS_ILOCK_EXCL, 174 if ((error = xfs_trans_iget(mp, tp, ino, 0,
174 &ip))) 175 XFS_ILOCK_EXCL, &ip)))
175 goto error_exit; 176 goto error_exit;
176 /* 177 /*
177 * Get a buffer for the block. 178 * Get a buffer for the block.
@@ -2023,8 +2024,8 @@ xfs_growfs_rt(
2023 /* 2024 /*
2024 * Lock out other callers by grabbing the bitmap inode lock. 2025 * Lock out other callers by grabbing the bitmap inode lock.
2025 */ 2026 */
2026 if ((error = xfs_trans_iget(mp, tp, 0, mp->m_sb.sb_rbmino, 2027 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
2027 XFS_ILOCK_EXCL, &ip))) 2028 XFS_ILOCK_EXCL, &ip)))
2028 goto error_exit; 2029 goto error_exit;
2029 ASSERT(ip == mp->m_rbmip); 2030 ASSERT(ip == mp->m_rbmip);
2030 /* 2031 /*
@@ -2037,8 +2038,8 @@ xfs_growfs_rt(
2037 /* 2038 /*
2038 * Get the summary inode into the transaction. 2039 * Get the summary inode into the transaction.
2039 */ 2040 */
2040 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rsumino, 2041 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rsumino, 0,
2041 0, XFS_ILOCK_EXCL, &ip))) 2042 XFS_ILOCK_EXCL, &ip)))
2042 goto error_exit; 2043 goto error_exit;
2043 ASSERT(ip == mp->m_rsumip); 2044 ASSERT(ip == mp->m_rsumip);
2044 /* 2045 /*
@@ -2158,10 +2159,9 @@ xfs_rtallocate_extent(
2158 /* 2159 /*
2159 * Lock out other callers by grabbing the bitmap inode lock. 2160 * Lock out other callers by grabbing the bitmap inode lock.
2160 */ 2161 */
2161 error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0, XFS_ILOCK_EXCL, &ip); 2162 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
2162 if (error) { 2163 XFS_ILOCK_EXCL, &ip)))
2163 return error; 2164 return error;
2164 }
2165 sumbp = NULL; 2165 sumbp = NULL;
2166 /* 2166 /*
2167 * Allocate by size, or near another block, or exactly at some block. 2167 * Allocate by size, or near another block, or exactly at some block.
@@ -2221,10 +2221,9 @@ xfs_rtfree_extent(
2221 /* 2221 /*
2222 * Synchronize by locking the bitmap inode. 2222 * Synchronize by locking the bitmap inode.
2223 */ 2223 */
2224 error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0, XFS_ILOCK_EXCL, &ip); 2224 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
2225 if (error) { 2225 XFS_ILOCK_EXCL, &ip)))
2226 return error; 2226 return error;
2227 }
2228#if defined(__KERNEL__) && defined(DEBUG) 2227#if defined(__KERNEL__) && defined(DEBUG)
2229 /* 2228 /*
2230 * Check to see that this whole range is currently allocated. 2229 * Check to see that this whole range is currently allocated.
@@ -2365,8 +2364,8 @@ xfs_rtpick_extent(
2365 __uint64_t seq; /* sequence number of file creation */ 2364 __uint64_t seq; /* sequence number of file creation */
2366 __uint64_t *seqp; /* pointer to seqno in inode */ 2365 __uint64_t *seqp; /* pointer to seqno in inode */
2367 2366
2368 error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0, XFS_ILOCK_EXCL, &ip); 2367 if ((error = xfs_trans_iget(mp, tp, mp->m_sb.sb_rbmino, 0,
2369 if (error) 2368 XFS_ILOCK_EXCL, &ip)))
2370 return error; 2369 return error;
2371 ASSERT(ip == mp->m_rbmip); 2370 ASSERT(ip == mp->m_rbmip);
2372 seqp = (__uint64_t *)&ip->i_d.di_atime; 2371 seqp = (__uint64_t *)&ip->i_d.di_atime;