summaryrefslogtreecommitdiff
path: root/fs
diff options
context:
space:
mode:
Diffstat (limited to 'fs')
-rw-r--r--fs/9p/fcall.c21
-rw-r--r--fs/9p/mux.c222
-rw-r--r--fs/9p/mux.h4
-rw-r--r--fs/9p/vfs_file.c13
-rw-r--r--fs/9p/vfs_inode.c19
-rw-r--r--fs/Makefile2
-rw-r--r--fs/autofs4/autofs_i.h5
-rw-r--r--fs/autofs4/root.c10
-rw-r--r--fs/autofs4/waitq.c77
-rw-r--r--fs/compat.c4
-rw-r--r--fs/configfs/dir.c137
-rw-r--r--fs/ext3/inode.c13
-rw-r--r--fs/jffs2/nodelist.c6
-rw-r--r--fs/locks.c21
-rw-r--r--fs/namespace.c7
-rw-r--r--fs/ocfs2/aops.c46
-rw-r--r--fs/ocfs2/aops.h4
-rw-r--r--fs/ocfs2/extent_map.c6
-rw-r--r--fs/ocfs2/file.c86
-rw-r--r--fs/ocfs2/journal.c8
-rw-r--r--fs/ocfs2/uptodate.c4
-rw-r--r--fs/ocfs2/vote.c6
-rw-r--r--fs/open.c1
-rw-r--r--fs/partitions/check.c3
-rw-r--r--fs/smbfs/dir.c5
-rw-r--r--fs/smbfs/request.c4
-rw-r--r--fs/splice.c74
-rw-r--r--fs/xfs/xfs_alloc.c5
-rw-r--r--fs/xfs/xfs_rename.c12
-rw-r--r--fs/xfs/xfs_vfsops.c27
-rw-r--r--fs/xfs/xfs_vnodeops.c2
31 files changed, 541 insertions, 313 deletions
diff --git a/fs/9p/fcall.c b/fs/9p/fcall.c
index 71742ba150c4..6f2617820a4e 100644
--- a/fs/9p/fcall.c
+++ b/fs/9p/fcall.c
@@ -98,23 +98,20 @@ v9fs_t_attach(struct v9fs_session_info *v9ses, char *uname, char *aname,
static void v9fs_t_clunk_cb(void *a, struct v9fs_fcall *tc,
struct v9fs_fcall *rc, int err)
{
- int fid;
+ int fid, id;
struct v9fs_session_info *v9ses;
- if (err)
- return;
-
+ id = 0;
fid = tc->params.tclunk.fid;
- kfree(tc);
-
- if (!rc)
- return;
-
- v9ses = a;
- if (rc->id == RCLUNK)
- v9fs_put_idpool(fid, &v9ses->fidpool);
+ if (rc)
+ id = rc->id;
+ kfree(tc);
kfree(rc);
+ if (id == RCLUNK) {
+ v9ses = a;
+ v9fs_put_idpool(fid, &v9ses->fidpool);
+ }
}
/**
diff --git a/fs/9p/mux.c b/fs/9p/mux.c
index 3e5b124a7212..f4407eb276c7 100644
--- a/fs/9p/mux.c
+++ b/fs/9p/mux.c
@@ -50,15 +50,23 @@ enum {
Wpending = 8, /* can write */
};
+enum {
+ None,
+ Flushing,
+ Flushed,
+};
+
struct v9fs_mux_poll_task;
struct v9fs_req {
+ spinlock_t lock;
int tag;
struct v9fs_fcall *tcall;
struct v9fs_fcall *rcall;
int err;
v9fs_mux_req_callback cb;
void *cba;
+ int flush;
struct list_head req_list;
};
@@ -96,8 +104,8 @@ struct v9fs_mux_poll_task {
struct v9fs_mux_rpc {
struct v9fs_mux_data *m;
- struct v9fs_req *req;
int err;
+ struct v9fs_fcall *tcall;
struct v9fs_fcall *rcall;
wait_queue_head_t wqueue;
};
@@ -524,10 +532,9 @@ again:
static void process_request(struct v9fs_mux_data *m, struct v9fs_req *req)
{
- int ecode, tag;
+ int ecode;
struct v9fs_str *ename;
- tag = req->tag;
if (!req->err && req->rcall->id == RERROR) {
ecode = req->rcall->params.rerror.errno;
ename = &req->rcall->params.rerror.error;
@@ -553,23 +560,6 @@ static void process_request(struct v9fs_mux_data *m, struct v9fs_req *req)
if (!req->err)
req->err = -EIO;
}
-
- if (req->err == ERREQFLUSH)
- return;
-
- if (req->cb) {
- dprintk(DEBUG_MUX, "calling callback tcall %p rcall %p\n",
- req->tcall, req->rcall);
-
- (*req->cb) (req->cba, req->tcall, req->rcall, req->err);
- req->cb = NULL;
- } else
- kfree(req->rcall);
-
- v9fs_mux_put_tag(m, tag);
-
- wake_up(&m->equeue);
- kfree(req);
}
/**
@@ -669,17 +659,26 @@ static void v9fs_read_work(void *a)
list_for_each_entry_safe(rreq, rptr, &m->req_list, req_list) {
if (rreq->tag == rcall->tag) {
req = rreq;
- req->rcall = rcall;
- list_del(&req->req_list);
- spin_unlock(&m->lock);
- process_request(m, req);
+ if (req->flush != Flushing)
+ list_del(&req->req_list);
break;
}
-
}
+ spin_unlock(&m->lock);
- if (!req) {
- spin_unlock(&m->lock);
+ if (req) {
+ req->rcall = rcall;
+ process_request(m, req);
+
+ if (req->flush != Flushing) {
+ if (req->cb)
+ (*req->cb) (req, req->cba);
+ else
+ kfree(req->rcall);
+
+ wake_up(&m->equeue);
+ }
+ } else {
if (err >= 0 && rcall->id != RFLUSH)
dprintk(DEBUG_ERROR,
"unexpected response mux %p id %d tag %d\n",
@@ -746,7 +745,6 @@ static struct v9fs_req *v9fs_send_request(struct v9fs_mux_data *m,
return ERR_PTR(-ENOMEM);
v9fs_set_tag(tc, n);
-
if ((v9fs_debug_level&DEBUG_FCALL) == DEBUG_FCALL) {
char buf[150];
@@ -754,12 +752,14 @@ static struct v9fs_req *v9fs_send_request(struct v9fs_mux_data *m,
printk(KERN_NOTICE "<<< %p %s\n", m, buf);
}
+ spin_lock_init(&req->lock);
req->tag = n;
req->tcall = tc;
req->rcall = NULL;
req->err = 0;
req->cb = cb;
req->cba = cba;
+ req->flush = None;
spin_lock(&m->lock);
list_add_tail(&req->req_list, &m->unsent_req_list);
@@ -776,72 +776,108 @@ static struct v9fs_req *v9fs_send_request(struct v9fs_mux_data *m,
return req;
}
-static void v9fs_mux_flush_cb(void *a, struct v9fs_fcall *tc,
- struct v9fs_fcall *rc, int err)
+static void v9fs_mux_free_request(struct v9fs_mux_data *m, struct v9fs_req *req)
+{
+ v9fs_mux_put_tag(m, req->tag);
+ kfree(req);
+}
+
+static void v9fs_mux_flush_cb(struct v9fs_req *freq, void *a)
{
v9fs_mux_req_callback cb;
int tag;
struct v9fs_mux_data *m;
- struct v9fs_req *req, *rptr;
+ struct v9fs_req *req, *rreq, *rptr;
m = a;
- dprintk(DEBUG_MUX, "mux %p tc %p rc %p err %d oldtag %d\n", m, tc,
- rc, err, tc->params.tflush.oldtag);
+ dprintk(DEBUG_MUX, "mux %p tc %p rc %p err %d oldtag %d\n", m,
+ freq->tcall, freq->rcall, freq->err,
+ freq->tcall->params.tflush.oldtag);
spin_lock(&m->lock);
cb = NULL;
- tag = tc->params.tflush.oldtag;
- list_for_each_entry_safe(req, rptr, &m->req_list, req_list) {
- if (req->tag == tag) {
+ tag = freq->tcall->params.tflush.oldtag;
+ req = NULL;
+ list_for_each_entry_safe(rreq, rptr, &m->req_list, req_list) {
+ if (rreq->tag == tag) {
+ req = rreq;
list_del(&req->req_list);
- if (req->cb) {
- cb = req->cb;
- req->cb = NULL;
- spin_unlock(&m->lock);
- (*cb) (req->cba, req->tcall, req->rcall,
- req->err);
- }
- kfree(req);
- wake_up(&m->equeue);
break;
}
}
+ spin_unlock(&m->lock);
- if (!cb)
- spin_unlock(&m->lock);
+ if (req) {
+ spin_lock(&req->lock);
+ req->flush = Flushed;
+ spin_unlock(&req->lock);
+
+ if (req->cb)
+ (*req->cb) (req, req->cba);
+ else
+ kfree(req->rcall);
+
+ wake_up(&m->equeue);
+ }
- v9fs_mux_put_tag(m, tag);
- kfree(tc);
- kfree(rc);
+ kfree(freq->tcall);
+ kfree(freq->rcall);
+ v9fs_mux_free_request(m, freq);
}
-static void
+static int
v9fs_mux_flush_request(struct v9fs_mux_data *m, struct v9fs_req *req)
{
struct v9fs_fcall *fc;
+ struct v9fs_req *rreq, *rptr;
dprintk(DEBUG_MUX, "mux %p req %p tag %d\n", m, req, req->tag);
+ /* if a response was received for a request, do nothing */
+ spin_lock(&req->lock);
+ if (req->rcall || req->err) {
+ spin_unlock(&req->lock);
+ dprintk(DEBUG_MUX, "mux %p req %p response already received\n", m, req);
+ return 0;
+ }
+
+ req->flush = Flushing;
+ spin_unlock(&req->lock);
+
+ spin_lock(&m->lock);
+ /* if the request is not sent yet, just remove it from the list */
+ list_for_each_entry_safe(rreq, rptr, &m->unsent_req_list, req_list) {
+ if (rreq->tag == req->tag) {
+ dprintk(DEBUG_MUX, "mux %p req %p request is not sent yet\n", m, req);
+ list_del(&rreq->req_list);
+ req->flush = Flushed;
+ spin_unlock(&m->lock);
+ if (req->cb)
+ (*req->cb) (req, req->cba);
+ return 0;
+ }
+ }
+ spin_unlock(&m->lock);
+
+ clear_thread_flag(TIF_SIGPENDING);
fc = v9fs_create_tflush(req->tag);
v9fs_send_request(m, fc, v9fs_mux_flush_cb, m);
+ return 1;
}
static void
-v9fs_mux_rpc_cb(void *a, struct v9fs_fcall *tc, struct v9fs_fcall *rc, int err)
+v9fs_mux_rpc_cb(struct v9fs_req *req, void *a)
{
struct v9fs_mux_rpc *r;
- if (err == ERREQFLUSH) {
- kfree(rc);
- dprintk(DEBUG_MUX, "err req flush\n");
- return;
- }
-
+ dprintk(DEBUG_MUX, "req %p r %p\n", req, a);
r = a;
- dprintk(DEBUG_MUX, "mux %p req %p tc %p rc %p err %d\n", r->m, r->req,
- tc, rc, err);
- r->rcall = rc;
- r->err = err;
+ r->rcall = req->rcall;
+ r->err = req->err;
+
+ if (req->flush!=None && !req->err)
+ r->err = -ERESTARTSYS;
+
wake_up(&r->wqueue);
}
@@ -856,12 +892,13 @@ int
v9fs_mux_rpc(struct v9fs_mux_data *m, struct v9fs_fcall *tc,
struct v9fs_fcall **rc)
{
- int err;
+ int err, sigpending;
unsigned long flags;
struct v9fs_req *req;
struct v9fs_mux_rpc r;
r.err = 0;
+ r.tcall = tc;
r.rcall = NULL;
r.m = m;
init_waitqueue_head(&r.wqueue);
@@ -869,48 +906,50 @@ v9fs_mux_rpc(struct v9fs_mux_data *m, struct v9fs_fcall *tc,
if (rc)
*rc = NULL;
+ sigpending = 0;
+ if (signal_pending(current)) {
+ sigpending = 1;
+ clear_thread_flag(TIF_SIGPENDING);
+ }
+
req = v9fs_send_request(m, tc, v9fs_mux_rpc_cb, &r);
if (IS_ERR(req)) {
err = PTR_ERR(req);
dprintk(DEBUG_MUX, "error %d\n", err);
- return PTR_ERR(req);
+ return err;
}
- r.req = req;
- dprintk(DEBUG_MUX, "mux %p tc %p tag %d rpc %p req %p\n", m, tc,
- req->tag, &r, req);
err = wait_event_interruptible(r.wqueue, r.rcall != NULL || r.err < 0);
if (r.err < 0)
err = r.err;
if (err == -ERESTARTSYS && m->trans->status == Connected && m->err == 0) {
- spin_lock(&m->lock);
- req->tcall = NULL;
- req->err = ERREQFLUSH;
- spin_unlock(&m->lock);
+ if (v9fs_mux_flush_request(m, req)) {
+ /* wait until we get response of the flush message */
+ do {
+ clear_thread_flag(TIF_SIGPENDING);
+ err = wait_event_interruptible(r.wqueue,
+ r.rcall || r.err);
+ } while (!r.rcall && !r.err && err==-ERESTARTSYS &&
+ m->trans->status==Connected && !m->err);
+ }
+ sigpending = 1;
+ }
- clear_thread_flag(TIF_SIGPENDING);
- v9fs_mux_flush_request(m, req);
+ if (sigpending) {
spin_lock_irqsave(&current->sighand->siglock, flags);
recalc_sigpending();
spin_unlock_irqrestore(&current->sighand->siglock, flags);
}
- if (!err) {
- if (r.rcall)
- dprintk(DEBUG_MUX, "got response id %d tag %d\n",
- r.rcall->id, r.rcall->tag);
-
- if (rc)
- *rc = r.rcall;
- else
- kfree(r.rcall);
- } else {
+ if (rc)
+ *rc = r.rcall;
+ else
kfree(r.rcall);
- dprintk(DEBUG_MUX, "got error %d\n", err);
- if (err > 0)
- err = -EIO;
- }
+
+ v9fs_mux_free_request(m, req);
+ if (err > 0)
+ err = -EIO;
return err;
}
@@ -951,12 +990,15 @@ void v9fs_mux_cancel(struct v9fs_mux_data *m, int err)
struct v9fs_req *req, *rtmp;
LIST_HEAD(cancel_list);
- dprintk(DEBUG_MUX, "mux %p err %d\n", m, err);
+ dprintk(DEBUG_ERROR, "mux %p err %d\n", m, err);
m->err = err;
spin_lock(&m->lock);
list_for_each_entry_safe(req, rtmp, &m->req_list, req_list) {
list_move(&req->req_list, &cancel_list);
}
+ list_for_each_entry_safe(req, rtmp, &m->unsent_req_list, req_list) {
+ list_move(&req->req_list, &cancel_list);
+ }
spin_unlock(&m->lock);
list_for_each_entry_safe(req, rtmp, &cancel_list, req_list) {
@@ -965,11 +1007,9 @@ void v9fs_mux_cancel(struct v9fs_mux_data *m, int err)
req->err = err;
if (req->cb)
- (*req->cb) (req->cba, req->tcall, req->rcall, req->err);
+ (*req->cb) (req, req->cba);
else
kfree(req->rcall);
-
- kfree(req);
}
wake_up(&m->equeue);
diff --git a/fs/9p/mux.h b/fs/9p/mux.h
index e90bfd32ea42..fb10c50186a1 100644
--- a/fs/9p/mux.h
+++ b/fs/9p/mux.h
@@ -24,6 +24,7 @@
*/
struct v9fs_mux_data;
+struct v9fs_req;
/**
* v9fs_mux_req_callback - callback function that is called when the
@@ -36,8 +37,7 @@ struct v9fs_mux_data;
* @rc - response call
* @err - error code (non-zero if error occured)
*/
-typedef void (*v9fs_mux_req_callback)(void *a, struct v9fs_fcall *tc,
- struct v9fs_fcall *rc, int err);
+typedef void (*v9fs_mux_req_callback)(struct v9fs_req *req, void *a);
int v9fs_mux_global_init(void);
void v9fs_mux_global_exit(void);
diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
index 083dcfcd158e..1a8e46084f0e 100644
--- a/fs/9p/vfs_file.c
+++ b/fs/9p/vfs_file.c
@@ -72,11 +72,17 @@ int v9fs_file_open(struct inode *inode, struct file *file)
return -ENOSPC;
}
- err = v9fs_t_walk(v9ses, vfid->fid, fid, NULL, NULL);
+ err = v9fs_t_walk(v9ses, vfid->fid, fid, NULL, &fcall);
if (err < 0) {
dprintk(DEBUG_ERROR, "rewalk didn't work\n");
- goto put_fid;
+ if (fcall && fcall->id == RWALK)
+ goto clunk_fid;
+ else {
+ v9fs_put_idpool(fid, &v9ses->fidpool);
+ goto free_fcall;
+ }
}
+ kfree(fcall);
/* TODO: do special things for O_EXCL, O_NOFOLLOW, O_SYNC */
/* translate open mode appropriately */
@@ -109,8 +115,7 @@ int v9fs_file_open(struct inode *inode, struct file *file)
clunk_fid:
v9fs_t_clunk(v9ses, fid);
-put_fid:
- v9fs_put_idpool(fid, &v9ses->fidpool);
+free_fcall:
kfree(fcall);
return err;
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 133db366d306..2cb87ba4b1c1 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -270,7 +270,10 @@ v9fs_create(struct v9fs_session_info *v9ses, u32 pfid, char *name, u32 perm,
err = v9fs_t_walk(v9ses, pfid, fid, NULL, &fcall);
if (err < 0) {
PRINT_FCALL_ERROR("clone error", fcall);
- goto put_fid;
+ if (fcall && fcall->id == RWALK)
+ goto clunk_fid;
+ else
+ goto put_fid;
}
kfree(fcall);
@@ -322,6 +325,9 @@ v9fs_clone_walk(struct v9fs_session_info *v9ses, u32 fid, struct dentry *dentry)
&fcall);
if (err < 0) {
+ if (fcall && fcall->id == RWALK)
+ goto clunk_fid;
+
PRINT_FCALL_ERROR("walk error", fcall);
v9fs_put_idpool(nfid, &v9ses->fidpool);
goto error;
@@ -640,19 +646,26 @@ static struct dentry *v9fs_vfs_lookup(struct inode *dir, struct dentry *dentry,
}
result = v9fs_t_walk(v9ses, dirfidnum, newfid,
- (char *)dentry->d_name.name, NULL);
+ (char *)dentry->d_name.name, &fcall);
+
if (result < 0) {
- v9fs_put_idpool(newfid, &v9ses->fidpool);
+ if (fcall && fcall->id == RWALK)
+ v9fs_t_clunk(v9ses, newfid);
+ else
+ v9fs_put_idpool(newfid, &v9ses->fidpool);
+
if (result == -ENOENT) {
d_add(dentry, NULL);
dprintk(DEBUG_VFS,
"Return negative dentry %p count %d\n",
dentry, atomic_read(&dentry->d_count));
+ kfree(fcall);
return NULL;
}
dprintk(DEBUG_ERROR, "walk error:%d\n", result);
goto FreeFcall;
}
+ kfree(fcall);
result = v9fs_t_stat(v9ses, newfid, &fcall);
if (result < 0) {
diff --git a/fs/Makefile b/fs/Makefile
index 83bf478e786b..078d3d1191a5 100644
--- a/fs/Makefile
+++ b/fs/Makefile
@@ -45,6 +45,7 @@ obj-$(CONFIG_DNOTIFY) += dnotify.o
obj-$(CONFIG_PROC_FS) += proc/
obj-y += partitions/
obj-$(CONFIG_SYSFS) += sysfs/
+obj-$(CONFIG_CONFIGFS_FS) += configfs/
obj-y += devpts/
obj-$(CONFIG_PROFILING) += dcookies.o
@@ -100,5 +101,4 @@ obj-$(CONFIG_BEFS_FS) += befs/
obj-$(CONFIG_HOSTFS) += hostfs/
obj-$(CONFIG_HPPFS) += hppfs/
obj-$(CONFIG_DEBUG_FS) += debugfs/
-obj-$(CONFIG_CONFIGFS_FS) += configfs/
obj-$(CONFIG_OCFS2_FS) += ocfs2/
diff --git a/fs/autofs4/autofs_i.h b/fs/autofs4/autofs_i.h
index 57c4903614e5..d6603d02304c 100644
--- a/fs/autofs4/autofs_i.h
+++ b/fs/autofs4/autofs_i.h
@@ -74,8 +74,8 @@ struct autofs_wait_queue {
struct autofs_wait_queue *next;
autofs_wqt_t wait_queue_token;
/* We use the following to see what we are waiting for */
- int hash;
- int len;
+ unsigned int hash;
+ unsigned int len;
char *name;
u32 dev;
u64 ino;
@@ -85,7 +85,6 @@ struct autofs_wait_queue {
pid_t tgid;
/* This is for status reporting upon return */
int status;
- atomic_t notify;
atomic_t wait_ctr;
};
diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
index 84e030c8ddd0..5100f984783f 100644
--- a/fs/autofs4/root.c
+++ b/fs/autofs4/root.c
@@ -327,6 +327,7 @@ static int try_to_fill_dentry(struct dentry *dentry, int flags)
static void *autofs4_follow_link(struct dentry *dentry, struct nameidata *nd)
{
struct autofs_sb_info *sbi = autofs4_sbi(dentry->d_sb);
+ struct autofs_info *ino = autofs4_dentry_ino(dentry);
int oz_mode = autofs4_oz_mode(sbi);
unsigned int lookup_type;
int status;
@@ -340,13 +341,8 @@ static void *autofs4_follow_link(struct dentry *dentry, struct nameidata *nd)
if (oz_mode || !lookup_type)
goto done;
- /*
- * If a request is pending wait for it.
- * If it's a mount then it won't be expired till at least
- * a liitle later and if it's an expire then we might need
- * to mount it again.
- */
- if (autofs4_ispending(dentry)) {
+ /* If an expire request is pending wait for it. */
+ if (ino && (ino->flags & AUTOFS_INF_EXPIRING)) {
DPRINTK("waiting for active request %p name=%.*s",
dentry, dentry->d_name.len, dentry->d_name.name);
diff --git a/fs/autofs4/waitq.c b/fs/autofs4/waitq.c
index 142ab6aa2aa1..ce103e7b0bc3 100644
--- a/fs/autofs4/waitq.c
+++ b/fs/autofs4/waitq.c
@@ -189,14 +189,30 @@ static int autofs4_getpath(struct autofs_sb_info *sbi,
return len;
}
+static struct autofs_wait_queue *
+autofs4_find_wait(struct autofs_sb_info *sbi,
+ char *name, unsigned int hash, unsigned int len)
+{
+ struct autofs_wait_queue *wq;
+
+ for (wq = sbi->queues; wq; wq = wq->next) {
+ if (wq->hash == hash &&
+ wq->len == len &&
+ wq->name && !memcmp(wq->name, name, len))
+ break;
+ }
+ return wq;
+}
+
int autofs4_wait(struct autofs_sb_info *sbi, struct dentry *dentry,
enum autofs_notify notify)
{
+ struct autofs_info *ino;
struct autofs_wait_queue *wq;
char *name;
unsigned int len = 0;
unsigned int hash = 0;
- int status;
+ int status, type;
/* In catatonic mode, we don't wait for nobody */
if (sbi->catatonic)
@@ -223,21 +239,41 @@ int autofs4_wait(struct autofs_sb_info *sbi, struct dentry *dentry,
return -EINTR;
}
- for (wq = sbi->queues ; wq ; wq = wq->next) {
- if (wq->hash == dentry->d_name.hash &&
- wq->len == len &&
- wq->name && !memcmp(wq->name, name, len))
- break;
- }
+ wq = autofs4_find_wait(sbi, name, hash, len);
+ ino = autofs4_dentry_ino(dentry);
+ if (!wq && ino && notify == NFY_NONE) {
+ /*
+ * Either we've betean the pending expire to post it's
+ * wait or it finished while we waited on the mutex.
+ * So we need to wait till either, the wait appears
+ * or the expire finishes.
+ */
+
+ while (ino->flags & AUTOFS_INF_EXPIRING) {
+ mutex_unlock(&sbi->wq_mutex);
+ schedule_timeout_interruptible(HZ/10);
+ if (mutex_lock_interruptible(&sbi->wq_mutex)) {
+ kfree(name);
+ return -EINTR;
+ }
+ wq = autofs4_find_wait(sbi, name, hash, len);
+ if (wq)
+ break;
+ }
- if (!wq) {
- /* Can't wait for an expire if there's no mount */
- if (notify == NFY_NONE && !d_mountpoint(dentry)) {
+ /*
+ * Not ideal but the status has already gone. Of the two
+ * cases where we wait on NFY_NONE neither depend on the
+ * return status of the wait.
+ */
+ if (!wq) {
kfree(name);
mutex_unlock(&sbi->wq_mutex);
- return -ENOENT;
+ return 0;
}
+ }
+ if (!wq) {
/* Create a new wait queue */
wq = kmalloc(sizeof(struct autofs_wait_queue),GFP_KERNEL);
if (!wq) {
@@ -263,20 +299,7 @@ int autofs4_wait(struct autofs_sb_info *sbi, struct dentry *dentry,
wq->tgid = current->tgid;
wq->status = -EINTR; /* Status return if interrupted */
atomic_set(&wq->wait_ctr, 2);
- atomic_set(&wq->notify, 1);
- mutex_unlock(&sbi->wq_mutex);
- } else {
- atomic_inc(&wq->wait_ctr);
mutex_unlock(&sbi->wq_mutex);
- kfree(name);
- DPRINTK("existing wait id = 0x%08lx, name = %.*s, nfy=%d",
- (unsigned long) wq->wait_queue_token, wq->len, wq->name, notify);
- }
-
- if (notify != NFY_NONE && atomic_read(&wq->notify)) {
- int type;
-
- atomic_dec(&wq->notify);
if (sbi->version < 5) {
if (notify == NFY_MOUNT)
@@ -299,6 +322,12 @@ int autofs4_wait(struct autofs_sb_info *sbi, struct dentry *dentry,
/* autofs4_notify_daemon() may block */
autofs4_notify_daemon(sbi, wq, type);
+ } else {
+ atomic_inc(&wq->wait_ctr);
+ mutex_unlock(&sbi->wq_mutex);
+ kfree(name);
+ DPRINTK("existing wait id = 0x%08lx, name = %.*s, nfy=%d",
+ (unsigned long) wq->wait_queue_token, wq->len, wq->name, notify);
}
/* wq->name is NULL if and only if the lock is already released */
diff --git a/fs/compat.c b/fs/compat.c
index 3f3e8f4d43d6..01f39f87f372 100644
--- a/fs/compat.c
+++ b/fs/compat.c
@@ -1323,7 +1323,7 @@ compat_sys_vmsplice(int fd, const struct compat_iovec __user *iov32,
{
unsigned i;
struct iovec *iov;
- if (nr_segs >= UIO_MAXIOV)
+ if (nr_segs > UIO_MAXIOV)
return -EINVAL;
iov = compat_alloc_user_space(nr_segs * sizeof(struct iovec));
for (i = 0; i < nr_segs; i++) {
@@ -1913,7 +1913,7 @@ asmlinkage long compat_sys_ppoll(struct pollfd __user *ufds,
}
if (sigmask) {
- if (sigsetsize |= sizeof(compat_sigset_t))
+ if (sigsetsize != sizeof(compat_sigset_t))
return -EINVAL;
if (copy_from_user(&ss32, sigmask, sizeof(ss32)))
return -EFAULT;
diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c
index 5638c8f9362f..5f952187fc53 100644
--- a/fs/configfs/dir.c
+++ b/fs/configfs/dir.c
@@ -505,13 +505,15 @@ static int populate_groups(struct config_group *group)
int i;
if (group->default_groups) {
- /* FYI, we're faking mkdir here
+ /*
+ * FYI, we're faking mkdir here
* I'm not sure we need this semaphore, as we're called
* from our parent's mkdir. That holds our parent's
* i_mutex, so afaik lookup cannot continue through our
* parent to find us, let alone mess with our tree.
* That said, taking our i_mutex is closer to mkdir
- * emulation, and shouldn't hurt. */
+ * emulation, and shouldn't hurt.
+ */
mutex_lock(&dentry->d_inode->i_mutex);
for (i = 0; group->default_groups[i]; i++) {
@@ -546,20 +548,34 @@ static void unlink_obj(struct config_item *item)
item->ci_group = NULL;
item->ci_parent = NULL;
+
+ /* Drop the reference for ci_entry */
config_item_put(item);
+ /* Drop the reference for ci_parent */
config_group_put(group);
}
}
static void link_obj(struct config_item *parent_item, struct config_item *item)
{
- /* Parent seems redundant with group, but it makes certain
- * traversals much nicer. */
+ /*
+ * Parent seems redundant with group, but it makes certain
+ * traversals much nicer.
+ */
item->ci_parent = parent_item;
+
+ /*
+ * We hold a reference on the parent for the child's ci_parent
+ * link.
+ */
item->ci_group = config_group_get(to_config_group(parent_item));
list_add_tail(&item->ci_entry, &item->ci_group->cg_children);
+ /*
+ * We hold a reference on the child for ci_entry on the parent's
+ * cg_children
+ */
config_item_get(item);
}
@@ -684,6 +700,10 @@ static void client_drop_item(struct config_item *parent_item,
type = parent_item->ci_type;
BUG_ON(!type);
+ /*
+ * If ->drop_item() exists, it is responsible for the
+ * config_item_put().
+ */
if (type->ct_group_ops && type->ct_group_ops->drop_item)
type->ct_group_ops->drop_item(to_config_group(parent_item),
item);
@@ -694,23 +714,28 @@ static void client_drop_item(struct config_item *parent_item,
static int configfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
- int ret;
+ int ret, module_got = 0;
struct config_group *group;
struct config_item *item;
struct config_item *parent_item;
struct configfs_subsystem *subsys;
struct configfs_dirent *sd;
struct config_item_type *type;
- struct module *owner;
+ struct module *owner = NULL;
char *name;
- if (dentry->d_parent == configfs_sb->s_root)
- return -EPERM;
+ if (dentry->d_parent == configfs_sb->s_root) {
+ ret = -EPERM;
+ goto out;
+ }
sd = dentry->d_parent->d_fsdata;
- if (!(sd->s_type & CONFIGFS_USET_DIR))
- return -EPERM;
+ if (!(sd->s_type & CONFIGFS_USET_DIR)) {
+ ret = -EPERM;
+ goto out;
+ }
+ /* Get a working ref for the duration of this function */
parent_item = configfs_get_config_item(dentry->d_parent);
type = parent_item->ci_type;
subsys = to_config_group(parent_item)->cg_subsys;
@@ -719,15 +744,16 @@ static int configfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
if (!type || !type->ct_group_ops ||
(!type->ct_group_ops->make_group &&
!type->ct_group_ops->make_item)) {
- config_item_put(parent_item);
- return -EPERM; /* What lack-of-mkdir returns */
+ ret = -EPERM; /* Lack-of-mkdir returns -EPERM */
+ goto out_put;
}
name = kmalloc(dentry->d_name.len + 1, GFP_KERNEL);
if (!name) {
- config_item_put(parent_item);
- return -ENOMEM;
+ ret = -ENOMEM;
+ goto out_put;
}
+
snprintf(name, dentry->d_name.len + 1, "%s", dentry->d_name.name);
down(&subsys->su_sem);
@@ -748,40 +774,67 @@ static int configfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
kfree(name);
if (!item) {
- config_item_put(parent_item);
- return -ENOMEM;
+ /*
+ * If item == NULL, then link_obj() was never called.
+ * There are no extra references to clean up.
+ */
+ ret = -ENOMEM;
+ goto out_put;
}
- ret = -EINVAL;
+ /*
+ * link_obj() has been called (via link_group() for groups).
+ * From here on out, errors must clean that up.
+ */
+
type = item->ci_type;
- if (type) {
- owner = type->ct_owner;
- if (try_module_get(owner)) {
- if (group) {
- ret = configfs_attach_group(parent_item,
- item,
- dentry);
- } else {
- ret = configfs_attach_item(parent_item,
- item,
- dentry);
- }
+ if (!type) {
+ ret = -EINVAL;
+ goto out_unlink;
+ }
- if (ret) {
- down(&subsys->su_sem);
- if (group)
- unlink_group(group);
- else
- unlink_obj(item);
- client_drop_item(parent_item, item);
- up(&subsys->su_sem);
+ owner = type->ct_owner;
+ if (!try_module_get(owner)) {
+ ret = -EINVAL;
+ goto out_unlink;
+ }
- config_item_put(parent_item);
- module_put(owner);
- }
- }
+ /*
+ * I hate doing it this way, but if there is
+ * an error, module_put() probably should
+ * happen after any cleanup.
+ */
+ module_got = 1;
+
+ if (group)
+ ret = configfs_attach_group(parent_item, item, dentry);
+ else
+ ret = configfs_attach_item(parent_item, item, dentry);
+
+out_unlink:
+ if (ret) {
+ /* Tear down everything we built up */
+ down(&subsys->su_sem);
+ if (group)
+ unlink_group(group);
+ else
+ unlink_obj(item);
+ client_drop_item(parent_item, item);
+ up(&subsys->su_sem);
+
+ if (module_got)
+ module_put(owner);
}
+out_put:
+ /*
+ * link_obj()/link_group() took a reference from child->parent,
+ * so the parent is safely pinned. We can drop our working
+ * reference.
+ */
+ config_item_put(parent_item);
+
+out:
return ret;
}
@@ -801,6 +854,7 @@ static int configfs_rmdir(struct inode *dir, struct dentry *dentry)
if (sd->s_type & CONFIGFS_USET_DEFAULT)
return -EPERM;
+ /* Get a working ref until we have the child */
parent_item = configfs_get_config_item(dentry->d_parent);
subsys = to_config_group(parent_item)->cg_subsys;
BUG_ON(!subsys);
@@ -817,6 +871,7 @@ static int configfs_rmdir(struct inode *dir, struct dentry *dentry)
return ret;
}
+ /* Get a working ref for the duration of this function */
item = configfs_get_config_item(dentry);
/* Drop reference from above, item already holds one. */
diff --git a/fs/ext3/inode.c b/fs/ext3/inode.c
index 48ae0339af17..2edd7eec88fd 100644
--- a/fs/ext3/inode.c
+++ b/fs/ext3/inode.c
@@ -711,7 +711,7 @@ static int ext3_splice_branch(handle_t *handle, struct inode *inode,
* direct blocks blocks
*/
if (num == 0 && blks > 1) {
- current_block = le32_to_cpu(where->key + 1);
+ current_block = le32_to_cpu(where->key) + 1;
for (i = 1; i < blks; i++)
*(where->p + i ) = cpu_to_le32(current_block++);
}
@@ -724,7 +724,7 @@ static int ext3_splice_branch(handle_t *handle, struct inode *inode,
if (block_i) {
block_i->last_alloc_logical_block = block + blks - 1;
block_i->last_alloc_physical_block =
- le32_to_cpu(where[num].key + blks - 1);
+ le32_to_cpu(where[num].key) + blks - 1;
}
/* We are done with atomic stuff, now do the rest of housekeeping */
@@ -814,11 +814,13 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
/* Simplest case - block found, no allocation needed */
if (!partial) {
- first_block = chain[depth - 1].key;
+ first_block = le32_to_cpu(chain[depth - 1].key);
clear_buffer_new(bh_result);
count++;
/*map more blocks*/
while (count < maxblocks && count <= blocks_to_boundary) {
+ unsigned long blk;
+
if (!verify_chain(chain, partial)) {
/*
* Indirect block might be removed by
@@ -831,8 +833,9 @@ int ext3_get_blocks_handle(handle_t *handle, struct inode *inode,
count = 0;
break;
}
- if (le32_to_cpu(*(chain[depth-1].p+count) ==
- (first_block + count)))
+ blk = le32_to_cpu(*(chain[depth-1].p + count));
+
+ if (blk == first_block + count)
count++;
else
break;
diff --git a/fs/jffs2/nodelist.c b/fs/jffs2/nodelist.c
index d4d0c41490cd..1d46677afd17 100644
--- a/fs/jffs2/nodelist.c
+++ b/fs/jffs2/nodelist.c
@@ -438,7 +438,8 @@ static int check_node_data(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info
if (c->mtd->point) {
err = c->mtd->point(c->mtd, ofs, len, &retlen, &buffer);
if (!err && retlen < tn->csize) {
- JFFS2_WARNING("MTD point returned len too short: %u instead of %u.\n", retlen, tn->csize);
+ JFFS2_WARNING("MTD point returned len too short: %zu "
+ "instead of %u.\n", retlen, tn->csize);
c->mtd->unpoint(c->mtd, buffer, ofs, len);
} else if (err)
JFFS2_WARNING("MTD point failed: error code %d.\n", err);
@@ -461,7 +462,8 @@ static int check_node_data(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info
}
if (retlen != len) {
- JFFS2_ERROR("short read at %#08x: %d instead of %d.\n", ofs, retlen, len);
+ JFFS2_ERROR("short read at %#08x: %zd instead of %d.\n",
+ ofs, retlen, len);
err = -EIO;
goto free_out;
}
diff --git a/fs/locks.c b/fs/locks.c
index efad798824dc..6f99c0a6f836 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -446,15 +446,14 @@ static struct lock_manager_operations lease_manager_ops = {
*/
static int lease_init(struct file *filp, int type, struct file_lock *fl)
{
+ if (assign_type(fl, type) != 0)
+ return -EINVAL;
+
fl->fl_owner = current->files;
fl->fl_pid = current->tgid;
fl->fl_file = filp;
fl->fl_flags = FL_LEASE;
- if (assign_type(fl, type) != 0) {
- locks_free_lock(fl);
- return -EINVAL;
- }
fl->fl_start = 0;
fl->fl_end = OFFSET_MAX;
fl->fl_ops = NULL;
@@ -466,16 +465,19 @@ static int lease_init(struct file *filp, int type, struct file_lock *fl)
static int lease_alloc(struct file *filp, int type, struct file_lock **flp)
{
struct file_lock *fl = locks_alloc_lock();
- int error;
+ int error = -ENOMEM;
if (fl == NULL)
- return -ENOMEM;
+ goto out;
error = lease_init(filp, type, fl);
- if (error)
- return error;
+ if (error) {
+ locks_free_lock(fl);
+ fl = NULL;
+ }
+out:
*flp = fl;
- return 0;
+ return error;
}
/* Check if two locks overlap each other.
@@ -1372,6 +1374,7 @@ static int __setlease(struct file *filp, long arg, struct file_lock **flp)
goto out;
if (my_before != NULL) {
+ *flp = *my_before;
error = lease->fl_lmops->fl_change(my_before, arg);
goto out;
}
diff --git a/fs/namespace.c b/fs/namespace.c
index 2c5f1f80bdc2..bf478addb852 100644
--- a/fs/namespace.c
+++ b/fs/namespace.c
@@ -899,13 +899,11 @@ static int do_change_type(struct nameidata *nd, int flag)
/*
* do loopback mount.
*/
-static int do_loopback(struct nameidata *nd, char *old_name, unsigned long flags, int mnt_flags)
+static int do_loopback(struct nameidata *nd, char *old_name, int recurse)
{
struct nameidata old_nd;
struct vfsmount *mnt = NULL;
- int recurse = flags & MS_REC;
int err = mount_is_safe(nd);
-
if (err)
return err;
if (!old_name || !*old_name)
@@ -939,7 +937,6 @@ static int do_loopback(struct nameidata *nd, char *old_name, unsigned long flags
spin_unlock(&vfsmount_lock);
release_mounts(&umount_list);
}
- mnt->mnt_flags = mnt_flags;
out:
up_write(&namespace_sem);
@@ -1353,7 +1350,7 @@ long do_mount(char *dev_name, char *dir_name, char *type_page,
retval = do_remount(&nd, flags & ~MS_REMOUNT, mnt_flags,
data_page);
else if (flags & MS_BIND)
- retval = do_loopback(&nd, dev_name, flags, mnt_flags);
+ retval = do_loopback(&nd, dev_name, flags & MS_REC);
else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
retval = do_change_type(&nd, flags);
else if (flags & MS_MOVE)
diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 0d858d0b25be..47152bf9a7f2 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -276,13 +276,29 @@ static int ocfs2_writepage(struct page *page, struct writeback_control *wbc)
return ret;
}
+/* This can also be called from ocfs2_write_zero_page() which has done
+ * it's own cluster locking. */
+int ocfs2_prepare_write_nolock(struct inode *inode, struct page *page,
+ unsigned from, unsigned to)
+{
+ int ret;
+
+ down_read(&OCFS2_I(inode)->ip_alloc_sem);
+
+ ret = block_prepare_write(page, from, to, ocfs2_get_block);
+
+ up_read(&OCFS2_I(inode)->ip_alloc_sem);
+
+ return ret;
+}
+
/*
* ocfs2_prepare_write() can be an outer-most ocfs2 call when it is called
* from loopback. It must be able to perform its own locking around
* ocfs2_get_block().
*/
-int ocfs2_prepare_write(struct file *file, struct page *page,
- unsigned from, unsigned to)
+static int ocfs2_prepare_write(struct file *file, struct page *page,
+ unsigned from, unsigned to)
{
struct inode *inode = page->mapping->host;
int ret;
@@ -295,11 +311,7 @@ int ocfs2_prepare_write(struct file *file, struct page *page,
goto out;
}
- down_read(&OCFS2_I(inode)->ip_alloc_sem);
-
- ret = block_prepare_write(page, from, to, ocfs2_get_block);
-
- up_read(&OCFS2_I(inode)->ip_alloc_sem);
+ ret = ocfs2_prepare_write_nolock(inode, page, from, to);
ocfs2_meta_unlock(inode, 0);
out:
@@ -625,11 +637,31 @@ static ssize_t ocfs2_direct_IO(int rw,
int ret;
mlog_entry_void();
+
+ /*
+ * We get PR data locks even for O_DIRECT. This allows
+ * concurrent O_DIRECT I/O but doesn't let O_DIRECT with
+ * extending and buffered zeroing writes race. If they did
+ * race then the buffered zeroing could be written back after
+ * the O_DIRECT I/O. It's one thing to tell people not to mix
+ * buffered and O_DIRECT writes, but expecting them to
+ * understand that file extension is also an implicit buffered
+ * write is too much. By getting the PR we force writeback of
+ * the buffered zeroing before proceeding.
+ */
+ ret = ocfs2_data_lock(inode, 0);
+ if (ret < 0) {
+ mlog_errno(ret);
+ goto out;
+ }
+ ocfs2_data_unlock(inode, 0);
+
ret = blockdev_direct_IO_no_locking(rw, iocb, inode,
inode->i_sb->s_bdev, iov, offset,
nr_segs,
ocfs2_direct_IO_get_blocks,
ocfs2_dio_end_io);
+out:
mlog_exit(ret);
return ret;
}
diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h
index d40456d509a0..e88c3f0b8fa9 100644
--- a/fs/ocfs2/aops.h
+++ b/fs/ocfs2/aops.h
@@ -22,8 +22,8 @@
#ifndef OCFS2_AOPS_H
#define OCFS2_AOPS_H
-int ocfs2_prepare_write(struct file *file, struct page *page,
- unsigned from, unsigned to);
+int ocfs2_prepare_write_nolock(struct inode *inode, struct page *page,
+ unsigned from, unsigned to);
struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
struct page *page,
diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c
index 4601fc256f11..1a5c69071df6 100644
--- a/fs/ocfs2/extent_map.c
+++ b/fs/ocfs2/extent_map.c
@@ -569,7 +569,7 @@ static int ocfs2_extent_map_insert(struct inode *inode,
ret = -ENOMEM;
ctxt.new_ent = kmem_cache_alloc(ocfs2_em_ent_cachep,
- GFP_KERNEL);
+ GFP_NOFS);
if (!ctxt.new_ent) {
mlog_errno(ret);
return ret;
@@ -583,14 +583,14 @@ static int ocfs2_extent_map_insert(struct inode *inode,
if (ctxt.need_left && !ctxt.left_ent) {
ctxt.left_ent =
kmem_cache_alloc(ocfs2_em_ent_cachep,
- GFP_KERNEL);
+ GFP_NOFS);
if (!ctxt.left_ent)
break;
}
if (ctxt.need_right && !ctxt.right_ent) {
ctxt.right_ent =
kmem_cache_alloc(ocfs2_em_ent_cachep,
- GFP_KERNEL);
+ GFP_NOFS);
if (!ctxt.right_ent)
break;
}
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index 581eb451a41a..a9559c874530 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -613,7 +613,8 @@ leave:
/* Some parts of this taken from generic_cont_expand, which turned out
* to be too fragile to do exactly what we need without us having to
- * worry about recursive locking in ->commit_write(). */
+ * worry about recursive locking in ->prepare_write() and
+ * ->commit_write(). */
static int ocfs2_write_zero_page(struct inode *inode,
u64 size)
{
@@ -641,7 +642,7 @@ static int ocfs2_write_zero_page(struct inode *inode,
goto out;
}
- ret = ocfs2_prepare_write(NULL, page, offset, offset);
+ ret = ocfs2_prepare_write_nolock(inode, page, offset, offset);
if (ret < 0) {
mlog_errno(ret);
goto out_unlock;
@@ -695,13 +696,26 @@ out:
return ret;
}
+/*
+ * A tail_to_skip value > 0 indicates that we're being called from
+ * ocfs2_file_aio_write(). This has the following implications:
+ *
+ * - we don't want to update i_size
+ * - di_bh will be NULL, which is fine because it's only used in the
+ * case where we want to update i_size.
+ * - ocfs2_zero_extend() will then only be filling the hole created
+ * between i_size and the start of the write.
+ */
static int ocfs2_extend_file(struct inode *inode,
struct buffer_head *di_bh,
- u64 new_i_size)
+ u64 new_i_size,
+ size_t tail_to_skip)
{
int ret = 0;
u32 clusters_to_add;
+ BUG_ON(!tail_to_skip && !di_bh);
+
/* setattr sometimes calls us like this. */
if (new_i_size == 0)
goto out;
@@ -714,27 +728,44 @@ static int ocfs2_extend_file(struct inode *inode,
OCFS2_I(inode)->ip_clusters;
if (clusters_to_add) {
- ret = ocfs2_extend_allocation(inode, clusters_to_add);
+ /*
+ * protect the pages that ocfs2_zero_extend is going to
+ * be pulling into the page cache.. we do this before the
+ * metadata extend so that we don't get into the situation
+ * where we've extended the metadata but can't get the data
+ * lock to zero.
+ */
+ ret = ocfs2_data_lock(inode, 1);
if (ret < 0) {
mlog_errno(ret);
goto out;
}
- ret = ocfs2_zero_extend(inode, new_i_size);
+ ret = ocfs2_extend_allocation(inode, clusters_to_add);
if (ret < 0) {
mlog_errno(ret);
- goto out;
+ goto out_unlock;
}
- }
- /* No allocation required, we just use this helper to
- * do a trivial update of i_size. */
- ret = ocfs2_simple_size_update(inode, di_bh, new_i_size);
- if (ret < 0) {
- mlog_errno(ret);
- goto out;
+ ret = ocfs2_zero_extend(inode, (u64)new_i_size - tail_to_skip);
+ if (ret < 0) {
+ mlog_errno(ret);
+ goto out_unlock;
+ }
+ }
+
+ if (!tail_to_skip) {
+ /* We're being called from ocfs2_setattr() which wants
+ * us to update i_size */
+ ret = ocfs2_simple_size_update(inode, di_bh, new_i_size);
+ if (ret < 0)
+ mlog_errno(ret);
}
+out_unlock:
+ if (clusters_to_add) /* this is the only case in which we lock */
+ ocfs2_data_unlock(inode, 1);
+
out:
return ret;
}
@@ -793,7 +824,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr)
if (i_size_read(inode) > attr->ia_size)
status = ocfs2_truncate_file(inode, bh, attr->ia_size);
else
- status = ocfs2_extend_file(inode, bh, attr->ia_size);
+ status = ocfs2_extend_file(inode, bh, attr->ia_size, 0);
if (status < 0) {
if (status != -ENOSPC)
mlog_errno(status);
@@ -1049,21 +1080,12 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb,
if (!clusters)
break;
- ret = ocfs2_extend_allocation(inode, clusters);
+ ret = ocfs2_extend_file(inode, NULL, newsize, count);
if (ret < 0) {
if (ret != -ENOSPC)
mlog_errno(ret);
goto out;
}
-
- /* Fill any holes which would've been created by this
- * write. If we're O_APPEND, this will wind up
- * (correctly) being a noop. */
- ret = ocfs2_zero_extend(inode, (u64) newsize - count);
- if (ret < 0) {
- mlog_errno(ret);
- goto out;
- }
break;
}
@@ -1146,6 +1168,22 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
ocfs2_iocb_set_rw_locked(iocb);
}
+ /*
+ * We're fine letting folks race truncates and extending
+ * writes with read across the cluster, just like they can
+ * locally. Hence no rw_lock during read.
+ *
+ * Take and drop the meta data lock to update inode fields
+ * like i_size. This allows the checks down below
+ * generic_file_aio_read() a chance of actually working.
+ */
+ ret = ocfs2_meta_lock(inode, NULL, NULL, 0);
+ if (ret < 0) {
+ mlog_errno(ret);
+ goto bail;
+ }
+ ocfs2_meta_unlock(inode, 0);
+
ret = generic_file_aio_read(iocb, buf, count, iocb->ki_pos);
if (ret == -EINVAL)
mlog(ML_ERROR, "generic_file_aio_read returned -EINVAL\n");
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index 6a610ae53583..eebc3cfa6be8 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -117,7 +117,7 @@ struct ocfs2_journal_handle *ocfs2_alloc_handle(struct ocfs2_super *osb)
{
struct ocfs2_journal_handle *retval = NULL;
- retval = kcalloc(1, sizeof(*retval), GFP_KERNEL);
+ retval = kcalloc(1, sizeof(*retval), GFP_NOFS);
if (!retval) {
mlog(ML_ERROR, "Failed to allocate memory for journal "
"handle!\n");
@@ -870,9 +870,11 @@ static int ocfs2_force_read_journal(struct inode *inode)
if (p_blocks > CONCURRENT_JOURNAL_FILL)
p_blocks = CONCURRENT_JOURNAL_FILL;
+ /* We are reading journal data which should not
+ * be put in the uptodate cache */
status = ocfs2_read_blocks(OCFS2_SB(inode->i_sb),
p_blkno, p_blocks, bhs, 0,
- inode);
+ NULL);
if (status < 0) {
mlog_errno(status);
goto bail;
@@ -982,7 +984,7 @@ static void ocfs2_queue_recovery_completion(struct ocfs2_journal *journal,
{
struct ocfs2_la_recovery_item *item;
- item = kmalloc(sizeof(struct ocfs2_la_recovery_item), GFP_KERNEL);
+ item = kmalloc(sizeof(struct ocfs2_la_recovery_item), GFP_NOFS);
if (!item) {
/* Though we wish to avoid it, we are in fact safe in
* skipping local alloc cleanup as fsck.ocfs2 is more
diff --git a/fs/ocfs2/uptodate.c b/fs/ocfs2/uptodate.c
index 04a684dfdd96..b8a00a793326 100644
--- a/fs/ocfs2/uptodate.c
+++ b/fs/ocfs2/uptodate.c
@@ -337,7 +337,7 @@ static void __ocfs2_set_buffer_uptodate(struct ocfs2_inode_info *oi,
(unsigned long long)oi->ip_blkno,
(unsigned long long)block, expand_tree);
- new = kmem_cache_alloc(ocfs2_uptodate_cachep, GFP_KERNEL);
+ new = kmem_cache_alloc(ocfs2_uptodate_cachep, GFP_NOFS);
if (!new) {
mlog_errno(-ENOMEM);
return;
@@ -349,7 +349,7 @@ static void __ocfs2_set_buffer_uptodate(struct ocfs2_inode_info *oi,
* has no way of tracking that. */
for(i = 0; i < OCFS2_INODE_MAX_CACHE_ARRAY; i++) {
tree[i] = kmem_cache_alloc(ocfs2_uptodate_cachep,
- GFP_KERNEL);
+ GFP_NOFS);
if (!tree[i]) {
mlog_errno(-ENOMEM);
goto out_free;
diff --git a/fs/ocfs2/vote.c b/fs/ocfs2/vote.c
index 53049a204197..ee42765a8553 100644
--- a/fs/ocfs2/vote.c
+++ b/fs/ocfs2/vote.c
@@ -586,7 +586,7 @@ static struct ocfs2_net_wait_ctxt *ocfs2_new_net_wait_ctxt(unsigned int response
{
struct ocfs2_net_wait_ctxt *w;
- w = kcalloc(1, sizeof(*w), GFP_KERNEL);
+ w = kcalloc(1, sizeof(*w), GFP_NOFS);
if (!w) {
mlog_errno(-ENOMEM);
goto bail;
@@ -749,7 +749,7 @@ static struct ocfs2_vote_msg * ocfs2_new_vote_request(struct ocfs2_super *osb,
BUG_ON(!ocfs2_is_valid_vote_request(type));
- request = kcalloc(1, sizeof(*request), GFP_KERNEL);
+ request = kcalloc(1, sizeof(*request), GFP_NOFS);
if (!request) {
mlog_errno(-ENOMEM);
} else {
@@ -1129,7 +1129,7 @@ static int ocfs2_handle_vote_message(struct o2net_msg *msg,
struct ocfs2_super *osb = data;
struct ocfs2_vote_work *work;
- work = kmalloc(sizeof(struct ocfs2_vote_work), GFP_KERNEL);
+ work = kmalloc(sizeof(struct ocfs2_vote_work), GFP_NOFS);
if (!work) {
status = -ENOMEM;
mlog_errno(status);
diff --git a/fs/open.c b/fs/open.c
index 53ec28c36777..317b7c7f38a7 100644
--- a/fs/open.c
+++ b/fs/open.c
@@ -1124,7 +1124,6 @@ asmlinkage long sys_openat(int dfd, const char __user *filename, int flags,
prevent_tail_call(ret);
return ret;
}
-EXPORT_SYMBOL_GPL(sys_openat);
#ifndef __alpha__
diff --git a/fs/partitions/check.c b/fs/partitions/check.c
index 45ae7dd3c650..7ef1f094de91 100644
--- a/fs/partitions/check.c
+++ b/fs/partitions/check.c
@@ -533,6 +533,7 @@ void del_gendisk(struct gendisk *disk)
devfs_remove_disk(disk);
+ kobject_uevent(&disk->kobj, KOBJ_REMOVE);
if (disk->holder_dir)
kobject_unregister(disk->holder_dir);
if (disk->slave_dir)
@@ -545,7 +546,7 @@ void del_gendisk(struct gendisk *disk)
kfree(disk_name);
}
put_device(disk->driverfs_dev);
+ disk->driverfs_dev = NULL;
}
- kobject_uevent(&disk->kobj, KOBJ_REMOVE);
kobject_del(&disk->kobj);
}
diff --git a/fs/smbfs/dir.c b/fs/smbfs/dir.c
index 34c7a11d91f0..70d9c5a37f5a 100644
--- a/fs/smbfs/dir.c
+++ b/fs/smbfs/dir.c
@@ -434,6 +434,11 @@ smb_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
if (dentry->d_name.len > SMB_MAXNAMELEN)
goto out;
+ /* Do not allow lookup of names with backslashes in */
+ error = -EINVAL;
+ if (memchr(dentry->d_name.name, '\\', dentry->d_name.len))
+ goto out;
+
lock_kernel();
error = smb_proc_getattr(dentry, &finfo);
#ifdef SMBFS_PARANOIA
diff --git a/fs/smbfs/request.c b/fs/smbfs/request.c
index c71c375863cc..c71dd2760d32 100644
--- a/fs/smbfs/request.c
+++ b/fs/smbfs/request.c
@@ -339,9 +339,11 @@ int smb_add_request(struct smb_request *req)
/*
* On timeout or on interrupt we want to try and remove the
* request from the recvq/xmitq.
+ * First check if the request is still part of a queue. (May
+ * have been removed by some error condition)
*/
smb_lock_server(server);
- if (!(req->rq_flags & SMB_REQ_RECEIVED)) {
+ if (!list_empty(&req->rq_queue)) {
list_del_init(&req->rq_queue);
smb_rput(req);
}
diff --git a/fs/splice.c b/fs/splice.c
index 7fb04970c72d..a285fd746dc0 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -51,7 +51,7 @@ struct splice_pipe_desc {
* addition of remove_mapping(). If success is returned, the caller may
* attempt to reuse this page for another destination.
*/
-static int page_cache_pipe_buf_steal(struct pipe_inode_info *info,
+static int page_cache_pipe_buf_steal(struct pipe_inode_info *pipe,
struct pipe_buffer *buf)
{
struct page *page = buf->page;
@@ -78,16 +78,18 @@ static int page_cache_pipe_buf_steal(struct pipe_inode_info *info,
return 1;
}
+ buf->flags |= PIPE_BUF_FLAG_LRU;
return 0;
}
-static void page_cache_pipe_buf_release(struct pipe_inode_info *info,
+static void page_cache_pipe_buf_release(struct pipe_inode_info *pipe,
struct pipe_buffer *buf)
{
page_cache_release(buf->page);
+ buf->flags &= ~PIPE_BUF_FLAG_LRU;
}
-static int page_cache_pipe_buf_pin(struct pipe_inode_info *info,
+static int page_cache_pipe_buf_pin(struct pipe_inode_info *pipe,
struct pipe_buffer *buf)
{
struct page *page = buf->page;
@@ -141,6 +143,7 @@ static int user_page_pipe_buf_steal(struct pipe_inode_info *pipe,
if (!(buf->flags & PIPE_BUF_FLAG_GIFT))
return 1;
+ buf->flags |= PIPE_BUF_FLAG_LRU;
return generic_pipe_buf_steal(pipe, buf);
}
@@ -321,6 +324,8 @@ __generic_file_splice_read(struct file *in, loff_t *ppos,
mapping_gfp_mask(mapping));
if (unlikely(error)) {
page_cache_release(page);
+ if (error == -EEXIST)
+ continue;
break;
}
/*
@@ -497,14 +502,14 @@ EXPORT_SYMBOL(generic_file_splice_read);
* Send 'sd->len' bytes to socket from 'sd->file' at position 'sd->pos'
* using sendpage(). Return the number of bytes sent.
*/
-static int pipe_to_sendpage(struct pipe_inode_info *info,
+static int pipe_to_sendpage(struct pipe_inode_info *pipe,
struct pipe_buffer *buf, struct splice_desc *sd)
{
struct file *file = sd->file;
loff_t pos = sd->pos;
int ret, more;
- ret = buf->ops->pin(info, buf);
+ ret = buf->ops->pin(pipe, buf);
if (!ret) {
more = (sd->flags & SPLICE_F_MORE) || sd->len < sd->total_len;
@@ -535,7 +540,7 @@ static int pipe_to_sendpage(struct pipe_inode_info *info,
* SPLICE_F_MOVE isn't set, or we cannot move the page, we simply create
* a new page in the output file page cache and fill/dirty that.
*/
-static int pipe_to_file(struct pipe_inode_info *info, struct pipe_buffer *buf,
+static int pipe_to_file(struct pipe_inode_info *pipe, struct pipe_buffer *buf,
struct splice_desc *sd)
{
struct file *file = sd->file;
@@ -549,7 +554,7 @@ static int pipe_to_file(struct pipe_inode_info *info, struct pipe_buffer *buf,
/*
* make sure the data in this buffer is uptodate
*/
- ret = buf->ops->pin(info, buf);
+ ret = buf->ops->pin(pipe, buf);
if (unlikely(ret))
return ret;
@@ -566,37 +571,23 @@ static int pipe_to_file(struct pipe_inode_info *info, struct pipe_buffer *buf,
*/
if ((sd->flags & SPLICE_F_MOVE) && this_len == PAGE_CACHE_SIZE) {
/*
- * If steal succeeds, buf->page is now pruned from the vm
- * side (page cache) and we can reuse it. The page will also
- * be locked on successful return.
+ * If steal succeeds, buf->page is now pruned from the
+ * pagecache and we can reuse it. The page will also be
+ * locked on successful return.
*/
- if (buf->ops->steal(info, buf))
+ if (buf->ops->steal(pipe, buf))
goto find_page;
page = buf->page;
- page_cache_get(page);
-
- /*
- * page must be on the LRU for adding to the pagecache.
- * Check this without grabbing the zone lock, if it isn't
- * the do grab the zone lock, recheck, and add if necessary.
- */
- if (!PageLRU(page)) {
- struct zone *zone = page_zone(page);
-
- spin_lock_irq(&zone->lru_lock);
- if (!PageLRU(page)) {
- SetPageLRU(page);
- add_page_to_inactive_list(zone, page);
- }
- spin_unlock_irq(&zone->lru_lock);
- }
-
if (add_to_page_cache(page, mapping, index, gfp_mask)) {
- page_cache_release(page);
unlock_page(page);
goto find_page;
}
+
+ page_cache_get(page);
+
+ if (!(buf->flags & PIPE_BUF_FLAG_LRU))
+ lru_cache_add(page);
} else {
find_page:
page = find_lock_page(mapping, index);
@@ -647,23 +638,36 @@ find_page:
}
ret = mapping->a_ops->prepare_write(file, page, offset, offset+this_len);
- if (ret == AOP_TRUNCATED_PAGE) {
+ if (unlikely(ret)) {
+ loff_t isize = i_size_read(mapping->host);
+
+ if (ret != AOP_TRUNCATED_PAGE)
+ unlock_page(page);
page_cache_release(page);
- goto find_page;
- } else if (ret)
+ if (ret == AOP_TRUNCATED_PAGE)
+ goto find_page;
+
+ /*
+ * prepare_write() may have instantiated a few blocks
+ * outside i_size. Trim these off again.
+ */
+ if (sd->pos + this_len > isize)
+ vmtruncate(mapping->host, isize);
+
goto out;
+ }
if (buf->page != page) {
/*
* Careful, ->map() uses KM_USER0!
*/
- char *src = buf->ops->map(info, buf, 1);
+ char *src = buf->ops->map(pipe, buf, 1);
char *dst = kmap_atomic(page, KM_USER1);
memcpy(dst + offset, src + buf->offset, this_len);
flush_dcache_page(page);
kunmap_atomic(dst, KM_USER1);
- buf->ops->unmap(info, buf, src);
+ buf->ops->unmap(pipe, buf, src);
}
ret = mapping->a_ops->commit_write(file, page, offset, offset+this_len);
diff --git a/fs/xfs/xfs_alloc.c b/fs/xfs/xfs_alloc.c
index 64ee07db0d5e..8558226281c4 100644
--- a/fs/xfs/xfs_alloc.c
+++ b/fs/xfs/xfs_alloc.c
@@ -1942,8 +1942,10 @@ xfs_alloc_fix_freelist(
/*
* Allocate as many blocks as possible at once.
*/
- if ((error = xfs_alloc_ag_vextent(&targs)))
+ if ((error = xfs_alloc_ag_vextent(&targs))) {
+ xfs_trans_brelse(tp, agflbp);
return error;
+ }
/*
* Stop if we run out. Won't happen if callers are obeying
* the restrictions correctly. Can happen for free calls
@@ -1960,6 +1962,7 @@ xfs_alloc_fix_freelist(
return error;
}
}
+ xfs_trans_brelse(tp, agflbp);
args->agbp = agbp;
return 0;
}
diff --git a/fs/xfs/xfs_rename.c b/fs/xfs/xfs_rename.c
index 81a05cfd77d2..1f148762eb28 100644
--- a/fs/xfs/xfs_rename.c
+++ b/fs/xfs/xfs_rename.c
@@ -316,6 +316,18 @@ xfs_rename(
}
}
+ /*
+ * If we are using project inheritance, we only allow renames
+ * into our tree when the project IDs are the same; else the
+ * tree quota mechanism would be circumvented.
+ */
+ if (unlikely((target_dp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
+ (target_dp->i_d.di_projid != src_ip->i_d.di_projid))) {
+ error = XFS_ERROR(EXDEV);
+ xfs_rename_unlock4(inodes, XFS_ILOCK_SHARED);
+ goto rele_return;
+ }
+
new_parent = (src_dp != target_dp);
src_is_directory = ((src_ip->i_d.di_mode & S_IFMT) == S_IFDIR);
diff --git a/fs/xfs/xfs_vfsops.c b/fs/xfs/xfs_vfsops.c
index f0e09ca14139..36ea1b2094f2 100644
--- a/fs/xfs/xfs_vfsops.c
+++ b/fs/xfs/xfs_vfsops.c
@@ -669,31 +669,22 @@ xfs_mntupdate(
xfs_mount_t *mp = XFS_BHVTOM(bdp);
int error;
- if (args->flags & XFSMNT_BARRIER)
- mp->m_flags |= XFS_MOUNT_BARRIER;
- else
- mp->m_flags &= ~XFS_MOUNT_BARRIER;
-
- if ((vfsp->vfs_flag & VFS_RDONLY) &&
- !(*flags & MS_RDONLY)) {
- vfsp->vfs_flag &= ~VFS_RDONLY;
-
- if (args->flags & XFSMNT_BARRIER)
+ if (!(*flags & MS_RDONLY)) { /* rw/ro -> rw */
+ if (vfsp->vfs_flag & VFS_RDONLY)
+ vfsp->vfs_flag &= ~VFS_RDONLY;
+ if (args->flags & XFSMNT_BARRIER) {
+ mp->m_flags |= XFS_MOUNT_BARRIER;
xfs_mountfs_check_barriers(mp);
- }
-
- if (!(vfsp->vfs_flag & VFS_RDONLY) &&
- (*flags & MS_RDONLY)) {
+ } else {
+ mp->m_flags &= ~XFS_MOUNT_BARRIER;
+ }
+ } else if (!(vfsp->vfs_flag & VFS_RDONLY)) { /* rw -> ro */
VFS_SYNC(vfsp, SYNC_FSDATA|SYNC_BDFLUSH|SYNC_ATTR, NULL, error);
-
xfs_quiesce_fs(mp);
-
- /* Ok now write out an unmount record */
xfs_log_unmount_write(mp);
xfs_unmountfs_writesb(mp);
vfsp->vfs_flag |= VFS_RDONLY;
}
-
return 0;
}
diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c
index fa71b305ba5c..7027ae68ee38 100644
--- a/fs/xfs/xfs_vnodeops.c
+++ b/fs/xfs/xfs_vnodeops.c
@@ -2663,7 +2663,7 @@ xfs_link(
*/
if (unlikely((tdp->i_d.di_flags & XFS_DIFLAG_PROJINHERIT) &&
(tdp->i_d.di_projid != sip->i_d.di_projid))) {
- error = XFS_ERROR(EPERM);
+ error = XFS_ERROR(EXDEV);
goto error_return;
}
generated by cgit v1.2.3-58-ga151 (git 2.39.0) at 2024-12-21 17:47:45 +0000