#include <linux/module.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/backing-dev.h>
#include <linux/poll.h>
#include <linux/uaccess.h>
#include "../stackglue.h"
#include "userdlm.h"
#define MLOG_MASK_PREFIX ML_DLMFS
#include "../cluster/masklog.h"
static const struct super_operations dlmfs_ops;
static const struct file_operations dlmfs_file_operations;
static const struct inode_operations dlmfs_dir_inode_operations;
static const struct inode_operations dlmfs_root_inode_operations;
static const struct inode_operations dlmfs_file_inode_operations;
static struct kmem_cache *dlmfs_inode_cache;
struct workqueue_struct *user_dlm_worker;
#define DLMFS_CAPABILITIES "bast stackglue"
static int param_set_dlmfs_capabilities(const char *val,
const struct kernel_param *kp)
{
printk(KERN_ERR "%s: readonly parameter\n", kp->name);
return -EINVAL;
}
static int param_get_dlmfs_capabilities(char *buffer,
const struct kernel_param *kp)
{
return strlcpy(buffer, DLMFS_CAPABILITIES,
strlen(DLMFS_CAPABILITIES) + 1);
}
module_param_call(capabilities, param_set_dlmfs_capabilities,
param_get_dlmfs_capabilities, NULL, 0444);
MODULE_PARM_DESC(capabilities, DLMFS_CAPABILITIES);
static int dlmfs_decode_open_flags(int open_flags,
int *level,
int *flags)
{
if (open_flags & (O_WRONLY|O_RDWR))
*level = DLM_LOCK_EX;
else
*level = DLM_LOCK_PR;
*flags = 0;
if (open_flags & O_NONBLOCK)
*flags |= DLM_LKF_NOQUEUE;
return 0;
}
static int dlmfs_file_open(struct inode *inode,
struct file *file)
{
int status, level, flags;
struct dlmfs_filp_private *fp = NULL;
struct dlmfs_inode_private *ip;
if (S_ISDIR(inode->i_mode))
BUG();
mlog(0, "open called on inode %lu, flags 0x%x\n", inode->i_ino,
file->f_flags);
status = dlmfs_decode_open_flags(file->f_flags, &level, &flags);
if (status < 0)
goto bail;
file->f_flags &= ~O_APPEND;
fp = kmalloc(sizeof(*fp), GFP_NOFS);
if (!fp) {
status = -ENOMEM;
goto bail;
}
fp->fp_lock_level = level;
ip = DLMFS_I(inode);
status = user_dlm_cluster_lock(&ip->ip_lockres, level, flags);
if (status < 0) {
if (flags & DLM_LKF_NOQUEUE && status == -EAGAIN)
status = -ETXTBSY;
kfree(fp);
goto bail;
}
file->private_data = fp;
bail:
return status;
}
static int dlmfs_file_release(struct inode *inode,
struct file *file)
{
int level;
struct dlmfs_inode_private *ip = DLMFS_I(inode);
struct dlmfs_filp_private *fp = file->private_data;
if (S_ISDIR(inode->i_mode))
BUG();
mlog(0, "close called on inode %lu\n", inode->i_ino);
if (fp) {
level = fp->fp_lock_level;
if (level != DLM_LOCK_IV)
user_dlm_cluster_unlock(&ip->ip_lockres, level);
kfree(fp);
file->private_data = NULL;
}
return 0;
}
static int dlmfs_file_setattr(struct mnt_idmap *idmap,
struct dentry *dentry, struct iattr *attr)
{
int error;
struct inode *inode = d_inode(dentry);
attr->ia_valid &= ~ATTR_SIZE;
error = setattr_prepare(&nop_mnt_idmap, dentry, attr);
if (error)
return error;
setattr_copy(&nop_mnt_idmap, inode, attr);
mark_inode_dirty(inode);
return 0;
}
static __poll_t dlmfs_file_poll(struct file *file, poll_table *wait)
{
__poll_t event = 0;
struct inode *inode = file_inode(file);
struct dlmfs_inode_private *ip = DLMFS_I(inode);
poll_wait(file, &ip->ip_lockres.l_event, wait);
spin_lock(&ip->ip_lockres.l_lock);
if (ip->ip_lockres.l_flags & USER_LOCK_BLOCKED)
event = EPOLLIN | EPOLLRDNORM;
spin_unlock(&ip->ip_lockres.l_lock);
return event;
}
static ssize_t dlmfs_file_read(struct file *file,
char __user *buf,
size_t count,
loff_t *ppos)
{
char lvb[DLM_LVB_LEN];
if (!user_dlm_read_lvb(file_inode(file), lvb))
return 0;
return simple_read_from_buffer(buf, count, ppos, lvb, sizeof(lvb));
}
static ssize_t dlmfs_file_write(struct file *filp,
const char __user *buf,
size_t count,
loff_t *ppos)
{
char lvb_buf[DLM_LVB_LEN];
int bytes_left;
struct inode *inode = file_inode(filp);
mlog(0, "inode %lu, count = %zu, *ppos = %llu\n",
inode->i_ino, count, *ppos);
if (*ppos >= DLM_LVB_LEN)
return -ENOSPC;
if (count > DLM_LVB_LEN - *ppos)
count = DLM_LVB_LEN - *ppos;
if (!count)
return 0;
bytes_left = copy_from_user(lvb_buf, buf, count);
count -= bytes_left;
if (count)
user_dlm_write_lvb(inode, lvb_buf, count);
*ppos = *ppos + count;
mlog(0, "wrote %zu bytes\n", count);
return count;
}
static void dlmfs_init_once(void *foo)
{
struct dlmfs_inode_private *ip =
(struct dlmfs_inode_private *) foo;
ip->ip_conn = NULL;
ip->ip_parent = NULL;
inode_init_once(&ip->ip_vfs_inode);
}
static struct inode *dlmfs_alloc_inode(struct super_block *sb)
{
struct dlmfs_inode_private *ip;
ip = alloc_inode_sb(sb, dlmfs_inode_cache, GFP_NOFS);
if (!ip)
return NULL;
return &ip->ip_vfs_inode;
}
static void dlmfs_free_inode(struct inode *inode)
{
kmem_cache_free(dlmfs_inode_cache, DLMFS_I(inode));
}
static void dlmfs_evict_inode(struct inode *inode)
{
int status;
struct dlmfs_inode_private *ip;
struct user_lock_res *lockres;
int teardown;
clear_inode(inode);
mlog(0, "inode %lu\n", inode->i_ino);
ip = DLMFS_I(inode);
lockres = &ip->ip_lockres;
if (S_ISREG(inode->i_mode)) {
spin_lock(&lockres->l_lock);
teardown = !!(lockres->l_flags & USER_LOCK_IN_TEARDOWN);
spin_unlock(&lockres->l_lock);
if (!teardown) {
status = user_dlm_destroy_lock(lockres);
if (status < 0)
mlog_errno(status);
}
iput(ip->ip_parent);
goto clear_fields;
}
mlog(0, "we're a directory, ip->ip_conn = 0x%p\n", ip->ip_conn);
if (ip->ip_conn)
user_dlm_unregister(ip->ip_conn);
clear_fields:
ip->ip_parent = NULL;
ip->ip_conn = NULL;
}
static struct inode *dlmfs_get_root_inode(struct super_block *sb)
{
struct inode *inode = new_inode(sb);
umode_t mode = S_IFDIR | 0755;
if (inode) {
inode->i_ino = get_next_ino();
inode_init_owner(&nop_mnt_idmap, inode, NULL, mode);
inode->i_atime = inode->i_mtime = inode_set_ctime_current(inode);
inc_nlink(inode);
inode->i_fop = &simple_dir_operations;
inode->i_op = &dlmfs_root_inode_operations;
}
return inode;
}
static struct inode *dlmfs_get_inode(struct inode *parent,
struct dentry *dentry,
umode_t mode)
{
struct super_block *sb = parent->i_sb;
struct inode * inode = new_inode(sb);
struct dlmfs_inode_private *ip;
if (!inode)
return NULL;
inode->i_ino = get_next_ino();
inode_init_owner(&nop_mnt_idmap, inode, parent, mode);
inode->i_atime = inode->i_mtime = inode_set_ctime_current(inode);
ip = DLMFS_I(inode);
ip->ip_conn = DLMFS_I(parent)->ip_conn;
switch (mode & S_IFMT) {
default:
BUG();
break;
case S_IFREG:
inode->i_op = &dlmfs_file_inode_operations;
inode->i_fop = &dlmfs_file_operations;
i_size_write(inode, DLM_LVB_LEN);
user_dlm_lock_res_init(&ip->ip_lockres, dentry);
ip->ip_parent = igrab(parent);
BUG_ON(!ip->ip_parent);
break;
case S_IFDIR:
inode->i_op = &dlmfs_dir_inode_operations;
inode->i_fop = &simple_dir_operations;
inc_nlink(inode);
break;
}
return inode;
}
static int dlmfs_mkdir(struct mnt_idmap * idmap,
struct inode * dir,
struct dentry * dentry,
umode_t mode)
{
int status;
struct inode *inode = NULL;
const struct qstr *domain = &dentry->d_name;
struct dlmfs_inode_private *ip;
struct ocfs2_cluster_connection *conn;
mlog(0, "mkdir %.*s\n", domain->len, domain->name);
if (domain->len >= GROUP_NAME_MAX) {
status = -EINVAL;
mlog(ML_ERROR, "invalid domain name for directory.\n");
goto bail;
}
inode = dlmfs_get_inode(dir, dentry, mode | S_IFDIR);
if (!inode) {
status = -ENOMEM;
mlog_errno(status);
goto bail;
}
ip = DLMFS_I(inode);
conn = user_dlm_register(domain);
if (IS_ERR(conn)) {
status = PTR_ERR(conn);
mlog(ML_ERROR, "Error %d could not register domain \"%.*s\"\n",
status, domain->len, domain->name);
goto bail;
}
ip->ip_conn = conn;
inc_nlink(dir);
d_instantiate(dentry, inode);
dget(dentry);
status = 0;
bail:
if (status < 0)
iput(inode);
return status;
}
static int dlmfs_create(struct mnt_idmap *idmap,
struct inode *dir,
struct dentry *dentry,
umode_t mode,
bool excl)
{
int status = 0;
struct inode *inode;
const struct qstr *name = &dentry->d_name;
mlog(0, "create %.*s\n", name->len, name->name);
if (name->len >= USER_DLM_LOCK_ID_MAX_LEN ||
name->name[0] == '$') {
status = -EINVAL;
mlog(ML_ERROR, "invalid lock name, %.*s\n", name->len,
name->name);
goto bail;
}
inode = dlmfs_get_inode(dir, dentry, mode | S_IFREG);
if (!inode) {
status = -ENOMEM;
mlog_errno(status);
goto bail;
}
d_instantiate(dentry, inode);
dget(dentry);
bail:
return status;
}
static int dlmfs_unlink(struct inode *dir,
struct dentry *dentry)
{
int status;
struct inode *inode = d_inode(dentry);
mlog(0, "unlink inode %lu\n", inode->i_ino);
status = user_dlm_destroy_lock(&DLMFS_I(inode)->ip_lockres);
if (status < 0) {
mlog(ML_ERROR, "unlink %pd, error %d from destroy\n",
dentry, status);
goto bail;
}
status = simple_unlink(dir, dentry);
bail:
return status;
}
static int dlmfs_fill_super(struct super_block * sb,
void * data,
int silent)
{
sb->s_maxbytes = MAX_LFS_FILESIZE;
sb->s_blocksize = PAGE_SIZE;
sb->s_blocksize_bits = PAGE_SHIFT;
sb->s_magic = DLMFS_MAGIC;
sb->s_op = &dlmfs_ops;
sb->s_root = d_make_root(dlmfs_get_root_inode(sb));
if (!sb->s_root)
return -ENOMEM;
return 0;
}
static const struct file_operations dlmfs_file_operations = {
.open = dlmfs_file_open,
.release = dlmfs_file_release,
.poll = dlmfs_file_poll,
.read = dlmfs_file_read,
.write = dlmfs_file_write,
.llseek = default_llseek,
};
static const struct inode_operations dlmfs_dir_inode_operations = {
.create = dlmfs_create,
.lookup = simple_lookup,
.unlink = dlmfs_unlink,
};
static const struct inode_operations dlmfs_root_inode_operations = {
.lookup = simple_lookup,
.mkdir = dlmfs_mkdir,
.rmdir = simple_rmdir,
};
static const struct super_operations dlmfs_ops = {
.statfs = simple_statfs,
.alloc_inode = dlmfs_alloc_inode,
.free_inode = dlmfs_free_inode,
.evict_inode = dlmfs_evict_inode,
.drop_inode = generic_delete_inode,
};
static const struct inode_operations dlmfs_file_inode_operations = {
.getattr = simple_getattr,
.setattr = dlmfs_file_setattr,
};
static struct dentry *dlmfs_mount(struct file_system_type *fs_type,
int flags, const char *dev_name, void *data)
{
return mount_nodev(fs_type, flags, data, dlmfs_fill_super);
}
static struct file_system_type dlmfs_fs_type = {
.owner = THIS_MODULE,
.name = "ocfs2_dlmfs",
.mount = dlmfs_mount,
.kill_sb = kill_litter_super,
};
MODULE_ALIAS_FS("ocfs2_dlmfs");
static int __init init_dlmfs_fs(void)
{
int status;
int cleanup_inode = 0, cleanup_worker = 0;
dlmfs_inode_cache = kmem_cache_create("dlmfs_inode_cache",
sizeof(struct dlmfs_inode_private),
0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
SLAB_MEM_SPREAD|SLAB_ACCOUNT),
dlmfs_init_once);
if (!dlmfs_inode_cache) {
status = -ENOMEM;
goto bail;
}
cleanup_inode = 1;
user_dlm_worker = alloc_workqueue("user_dlm", WQ_MEM_RECLAIM, 0);
if (!user_dlm_worker) {
status = -ENOMEM;
goto bail;
}
cleanup_worker = 1;
user_dlm_set_locking_protocol();
status = register_filesystem(&dlmfs_fs_type);
bail:
if (status) {
if (cleanup_inode)
kmem_cache_destroy(dlmfs_inode_cache);
if (cleanup_worker)
destroy_workqueue(user_dlm_worker);
} else
printk("OCFS2 User DLM kernel interface loaded\n");
return status;
}
static void __exit exit_dlmfs_fs(void)
{
unregister_filesystem(&dlmfs_fs_type);
destroy_workqueue(user_dlm_worker);
rcu_barrier();
kmem_cache_destroy(dlmfs_inode_cache);
}
MODULE_AUTHOR("Oracle");
MODULE_LICENSE("GPL");
MODULE_DESCRIPTION("OCFS2 DLM-Filesystem");
module_init(init_dlmfs_fs)
module_exit