wait_queue_head_t wait_q;
struct rb_root msg_tree;
+ struct posix_msg_tree_node *node_cache;
struct mq_attr attr;
struct sigevent notify;
else
p = &(*p)->rb_right;
}
- leaf = kzalloc(sizeof(*leaf), GFP_ATOMIC);
- if (!leaf)
- return -ENOMEM;
- rb_init_node(&leaf->rb_node);
- INIT_LIST_HEAD(&leaf->msg_list);
+ if (info->node_cache) {
+ leaf = info->node_cache;
+ info->node_cache = NULL;
+ } else {
+ leaf = kmalloc(sizeof(*leaf), GFP_ATOMIC);
+ if (!leaf)
+ return -ENOMEM;
+ rb_init_node(&leaf->rb_node);
+ INIT_LIST_HEAD(&leaf->msg_list);
+ info->qsize += sizeof(*leaf);
+ }
leaf->priority = msg->m_type;
rb_link_node(&leaf->rb_node, parent, p);
rb_insert_color(&leaf->rb_node, &info->msg_tree);
- info->qsize += sizeof(struct posix_msg_tree_node);
insert_msg:
info->attr.mq_curmsgs++;
info->qsize += msg->m_ts;
return NULL;
}
leaf = rb_entry(parent, struct posix_msg_tree_node, rb_node);
- if (list_empty(&leaf->msg_list)) {
+ if (unlikely(list_empty(&leaf->msg_list))) {
pr_warn_once("Inconsistency in POSIX message queue, "
"empty leaf node but we haven't implemented "
"lazy leaf delete!\n");
rb_erase(&leaf->rb_node, &info->msg_tree);
- info->qsize -= sizeof(struct posix_msg_tree_node);
- kfree(leaf);
+ if (info->node_cache) {
+ info->qsize -= sizeof(*leaf);
+ kfree(leaf);
+ } else {
+ info->node_cache = leaf;
+ }
goto try_again;
} else {
msg = list_first_entry(&leaf->msg_list,
list_del(&msg->m_list);
if (list_empty(&leaf->msg_list)) {
rb_erase(&leaf->rb_node, &info->msg_tree);
- info->qsize -= sizeof(struct posix_msg_tree_node);
- kfree(leaf);
+ if (info->node_cache) {
+ info->qsize -= sizeof(*leaf);
+ kfree(leaf);
+ } else {
+ info->node_cache = leaf;
+ }
}
}
info->attr.mq_curmsgs--;
info->qsize = 0;
info->user = NULL; /* set when all is ok */
info->msg_tree = RB_ROOT;
+ info->node_cache = NULL;
memset(&info->attr, 0, sizeof(info->attr));
info->attr.mq_maxmsg = min(ipc_ns->mq_msg_max,
ipc_ns->mq_msg_default);
spin_lock(&info->lock);
while ((msg = msg_get(info)) != NULL)
free_msg(msg);
+ kfree(info->node_cache);
spin_unlock(&info->lock);
/* Total amount of bytes accounted for the mqueue */
}
static int mqueue_create(struct inode *dir, struct dentry *dentry,
- umode_t mode, struct nameidata *nd)
+ umode_t mode, bool excl)
{
struct inode *inode;
struct mq_attr *attr = dentry->d_fsdata;
unsigned long total_size;
if (attr->mq_maxmsg <= 0 || attr->mq_msgsize <= 0)
- return 0;
+ return -EINVAL;
if (capable(CAP_SYS_RESOURCE)) {
if (attr->mq_maxmsg > HARD_MSGMAX ||
attr->mq_msgsize > HARD_MSGSIZEMAX)
- return 0;
+ return -EINVAL;
} else {
if (attr->mq_maxmsg > ipc_ns->mq_msg_max ||
attr->mq_msgsize > ipc_ns->mq_msgsize_max)
- return 0;
+ return -EINVAL;
}
/* check for overflow */
if (attr->mq_msgsize > ULONG_MAX/attr->mq_maxmsg)
- return 0;
+ return -EOVERFLOW;
mq_treesize = attr->mq_maxmsg * sizeof(struct msg_msg) +
min_t(unsigned int, attr->mq_maxmsg, MQ_PRIO_MAX) *
sizeof(struct posix_msg_tree_node);
total_size = attr->mq_maxmsg * attr->mq_msgsize;
if (total_size + mq_treesize < total_size)
- return 0;
- return 1;
+ return -EOVERFLOW;
+ return 0;
}
/*
* Invoked when creating a new queue via sys_mq_open
*/
-static struct file *do_create(struct ipc_namespace *ipc_ns, struct dentry *dir,
- struct dentry *dentry, int oflag, umode_t mode,
+static struct file *do_create(struct ipc_namespace *ipc_ns, struct inode *dir,
+ struct path *path, int oflag, umode_t mode,
struct mq_attr *attr)
{
const struct cred *cred = current_cred();
int ret;
if (attr) {
- if (!mq_attr_ok(ipc_ns, attr)) {
- ret = -EINVAL;
- goto out;
- }
+ ret = mq_attr_ok(ipc_ns, attr);
+ if (ret)
+ return ERR_PTR(ret);
/* store for use during create */
- dentry->d_fsdata = attr;
+ path->dentry->d_fsdata = attr;
+ } else {
+ struct mq_attr def_attr;
+
+ def_attr.mq_maxmsg = min(ipc_ns->mq_msg_max,
+ ipc_ns->mq_msg_default);
+ def_attr.mq_msgsize = min(ipc_ns->mq_msgsize_max,
+ ipc_ns->mq_msgsize_default);
+ ret = mq_attr_ok(ipc_ns, &def_attr);
+ if (ret)
+ return ERR_PTR(ret);
}
mode &= ~current_umask();
- ret = mnt_want_write(ipc_ns->mq_mnt);
- if (ret)
- goto out;
- ret = vfs_create(dir->d_inode, dentry, mode, NULL);
- dentry->d_fsdata = NULL;
+ ret = mnt_want_write(path->mnt);
if (ret)
- goto out_drop_write;
-
- result = dentry_open(dentry, ipc_ns->mq_mnt, oflag, cred);
+ return ERR_PTR(ret);
+ ret = vfs_create(dir, path->dentry, mode, true);
+ path->dentry->d_fsdata = NULL;
+ if (!ret)
+ result = dentry_open(path, oflag, cred);
+ else
+ result = ERR_PTR(ret);
/*
* dentry_open() took a persistent mnt_want_write(),
* so we can now drop this one.
*/
- mnt_drop_write(ipc_ns->mq_mnt);
+ mnt_drop_write(path->mnt);
return result;
-
-out_drop_write:
- mnt_drop_write(ipc_ns->mq_mnt);
-out:
- dput(dentry);
- mntput(ipc_ns->mq_mnt);
- return ERR_PTR(ret);
}
/* Opens existing queue */
-static struct file *do_open(struct ipc_namespace *ipc_ns,
- struct dentry *dentry, int oflag)
+static struct file *do_open(struct path *path, int oflag)
{
- int ret;
- const struct cred *cred = current_cred();
-
static const int oflag2acc[O_ACCMODE] = { MAY_READ, MAY_WRITE,
MAY_READ | MAY_WRITE };
-
- if ((oflag & O_ACCMODE) == (O_RDWR | O_WRONLY)) {
- ret = -EINVAL;
- goto err;
- }
-
- if (inode_permission(dentry->d_inode, oflag2acc[oflag & O_ACCMODE])) {
- ret = -EACCES;
- goto err;
- }
-
- return dentry_open(dentry, ipc_ns->mq_mnt, oflag, cred);
-
-err:
- dput(dentry);
- mntput(ipc_ns->mq_mnt);
- return ERR_PTR(ret);
+ int acc;
+ if ((oflag & O_ACCMODE) == (O_RDWR | O_WRONLY))
+ return ERR_PTR(-EINVAL);
+ acc = oflag2acc[oflag & O_ACCMODE];
+ if (inode_permission(path->dentry->d_inode, acc))
+ return ERR_PTR(-EACCES);
+ return dentry_open(path, oflag, current_cred());
}
SYSCALL_DEFINE4(mq_open, const char __user *, u_name, int, oflag, umode_t, mode,
struct mq_attr __user *, u_attr)
{
- struct dentry *dentry;
+ struct path path;
struct file *filp;
char *name;
struct mq_attr attr;
int fd, error;
struct ipc_namespace *ipc_ns = current->nsproxy->ipc_ns;
+ struct dentry *root = ipc_ns->mq_mnt->mnt_root;
if (u_attr && copy_from_user(&attr, u_attr, sizeof(struct mq_attr)))
return -EFAULT;
if (fd < 0)
goto out_putname;
- mutex_lock(&ipc_ns->mq_mnt->mnt_root->d_inode->i_mutex);
- dentry = lookup_one_len(name, ipc_ns->mq_mnt->mnt_root, strlen(name));
- if (IS_ERR(dentry)) {
- error = PTR_ERR(dentry);
+ error = 0;
+ mutex_lock(&root->d_inode->i_mutex);
+ path.dentry = lookup_one_len(name, root, strlen(name));
+ if (IS_ERR(path.dentry)) {
+ error = PTR_ERR(path.dentry);
goto out_putfd;
}
- mntget(ipc_ns->mq_mnt);
+ path.mnt = mntget(ipc_ns->mq_mnt);
if (oflag & O_CREAT) {
- if (dentry->d_inode) { /* entry already exists */
- audit_inode(name, dentry);
+ if (path.dentry->d_inode) { /* entry already exists */
+ audit_inode(name, path.dentry);
if (oflag & O_EXCL) {
error = -EEXIST;
goto out;
}
- filp = do_open(ipc_ns, dentry, oflag);
+ filp = do_open(&path, oflag);
} else {
- filp = do_create(ipc_ns, ipc_ns->mq_mnt->mnt_root,
- dentry, oflag, mode,
+ filp = do_create(ipc_ns, root->d_inode,
+ &path, oflag, mode,
u_attr ? &attr : NULL);
}
} else {
- if (!dentry->d_inode) {
+ if (!path.dentry->d_inode) {
error = -ENOENT;
goto out;
}
- audit_inode(name, dentry);
- filp = do_open(ipc_ns, dentry, oflag);
+ audit_inode(name, path.dentry);
+ filp = do_open(&path, oflag);
}
- if (IS_ERR(filp)) {
+ if (!IS_ERR(filp))
+ fd_install(fd, filp);
+ else
error = PTR_ERR(filp);
- goto out_putfd;
- }
-
- fd_install(fd, filp);
- goto out_upsem;
-
out:
- dput(dentry);
- mntput(ipc_ns->mq_mnt);
+ path_put(&path);
out_putfd:
- put_unused_fd(fd);
- fd = error;
-out_upsem:
- mutex_unlock(&ipc_ns->mq_mnt->mnt_root->d_inode->i_mutex);
+ if (error) {
+ put_unused_fd(fd);
+ fd = error;
+ }
+ mutex_unlock(&root->d_inode->i_mutex);
out_putname:
putname(name);
return fd;
struct mqueue_inode_info *info;
ktime_t expires, *timeout = NULL;
struct timespec ts;
- int ret;
+ struct posix_msg_tree_node *new_leaf = NULL;
+ int ret = 0;
if (u_abs_timeout) {
int res = prepare_timeout(u_abs_timeout, &expires, &ts);
msg_ptr->m_ts = msg_len;
msg_ptr->m_type = msg_prio;
+ /*
+ * msg_insert really wants us to have a valid, spare node struct so
+ * it doesn't have to kmalloc a GFP_ATOMIC allocation, but it will
+ * fall back to that if necessary.
+ */
+ if (!info->node_cache)
+ new_leaf = kmalloc(sizeof(*new_leaf), GFP_KERNEL);
+
spin_lock(&info->lock);
+ if (!info->node_cache && new_leaf) {
+ /* Save our speculative allocation into the cache */
+ rb_init_node(&new_leaf->rb_node);
+ INIT_LIST_HEAD(&new_leaf->msg_list);
+ info->node_cache = new_leaf;
+ info->qsize += sizeof(*new_leaf);
+ new_leaf = NULL;
+ } else {
+ kfree(new_leaf);
+ }
+
if (info->attr.mq_curmsgs == info->attr.mq_maxmsg) {
if (filp->f_flags & O_NONBLOCK) {
- spin_unlock(&info->lock);
ret = -EAGAIN;
} else {
wait.task = current;
wait.msg = (void *) msg_ptr;
wait.state = STATE_NONE;
ret = wq_sleep(info, SEND, timeout, &wait);
+ /*
+ * wq_sleep must be called with info->lock held, and
+ * returns with the lock released
+ */
+ goto out_free;
}
- if (ret < 0)
- free_msg(msg_ptr);
} else {
receiver = wq_get_first_waiter(info, RECV);
if (receiver) {
pipelined_send(info, msg_ptr, receiver);
} else {
/* adds message to the queue */
- if (msg_insert(msg_ptr, info)) {
- free_msg(msg_ptr);
- ret = -ENOMEM;
- spin_unlock(&info->lock);
- goto out_fput;
- }
+ ret = msg_insert(msg_ptr, info);
+ if (ret)
+ goto out_unlock;
__do_notify(info);
}
inode->i_atime = inode->i_mtime = inode->i_ctime =
CURRENT_TIME;
- spin_unlock(&info->lock);
- ret = 0;
}
+out_unlock:
+ spin_unlock(&info->lock);
+out_free:
+ if (ret)
+ free_msg(msg_ptr);
out_fput:
fput(filp);
out:
struct ext_wait_queue wait;
ktime_t expires, *timeout = NULL;
struct timespec ts;
+ struct posix_msg_tree_node *new_leaf = NULL;
if (u_abs_timeout) {
int res = prepare_timeout(u_abs_timeout, &expires, &ts);
goto out_fput;
}
+ /*
+ * msg_insert really wants us to have a valid, spare node struct so
+ * it doesn't have to kmalloc a GFP_ATOMIC allocation, but it will
+ * fall back to that if necessary.
+ */
+ if (!info->node_cache)
+ new_leaf = kmalloc(sizeof(*new_leaf), GFP_KERNEL);
+
spin_lock(&info->lock);
+
+ if (!info->node_cache && new_leaf) {
+ /* Save our speculative allocation into the cache */
+ rb_init_node(&new_leaf->rb_node);
+ INIT_LIST_HEAD(&new_leaf->msg_list);
+ info->node_cache = new_leaf;
+ info->qsize += sizeof(*new_leaf);
+ } else {
+ kfree(new_leaf);
+ }
+
if (info->attr.mq_curmsgs == 0) {
if (filp->f_flags & O_NONBLOCK) {
spin_unlock(&info->lock);