int ocfs2_mmap(struct file *file, struct vm_area_struct *vma)
{
int ret = 0, lock_level = 0;
struct ocfs2_super *osb = OCFS2_SB(file->f_dentry->d_inode->i_sb);
/*
* Only support shared writeable mmap for local mounts which
* don't know about holes.
*/
if ((!ocfs2_mount_local(osb) || ocfs2_sparse_alloc(osb)) &&
((vma->vm_flags & VM_SHARED) || (vma->vm_flags & VM_MAYSHARE)) &&
((vma->vm_flags & VM_WRITE) || (vma->vm_flags & VM_MAYWRITE))) {
mlog(0, "disallow shared writable mmaps %lx\n", vma->vm_flags);
/* This is -EINVAL because generic_file_readonly_mmap
* returns it in a similar situation. */
return -EINVAL;
}
ret = ocfs2_meta_lock_atime(file->f_dentry->d_inode,
file->f_vfsmnt, &lock_level);
if (ret < 0) {
mlog_errno(ret);
goto out;
}
ocfs2_meta_unlock(file->f_dentry->d_inode, lock_level);
out:
vma->vm_ops = &ocfs2_file_vm_ops;
vma->vm_flags |= VM_CAN_NONLINEAR;
return 0;
}
static ssize_t ocfs2_file_aio_read(struct kiocb *iocb,
const struct iovec *iov,
unsigned long nr_segs,
loff_t pos)
{
int ret = 0, rw_level = -1, have_alloc_sem = 0, lock_level = 0;
struct file *filp = iocb->ki_filp;
struct inode *inode = filp->f_path.dentry->d_inode;
mlog_entry("(0x%p, %u, '%.*s')\n", filp,
(unsigned int)nr_segs,
filp->f_path.dentry->d_name.len,
filp->f_path.dentry->d_name.name);
if (!inode) {
ret = -EINVAL;
mlog_errno(ret);
goto bail;
}
/*
* buffered reads protect themselves in ->readpage(). O_DIRECT reads
* need locks to protect pending reads from racing with truncate.
*/
if (filp->f_flags & O_DIRECT) {
down_read(&inode->i_alloc_sem);
have_alloc_sem = 1;
ret = ocfs2_rw_lock(inode, 0);
if (ret < 0) {
mlog_errno(ret);
goto bail;
}
rw_level = 0;
/* communicate with ocfs2_dio_end_io */
ocfs2_iocb_set_rw_locked(iocb, rw_level);
}
/*
* 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_atime(inode, filp->f_vfsmnt, &lock_level);
if (ret < 0) {
mlog_errno(ret);
goto bail;
}
ocfs2_meta_unlock(inode, lock_level);
ret = generic_file_aio_read(iocb, iov, nr_segs, iocb->ki_pos);
if (ret == -EINVAL)
mlog(ML_ERROR, "generic_file_aio_read returned -EINVAL\n");
/* buffered aio wouldn't have proper lock coverage today */
BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT));
/* see ocfs2_file_aio_write */
if (ret == -EIOCBQUEUED || !ocfs2_iocb_is_rw_locked(iocb)) {
rw_level = -1;
have_alloc_sem = 0;
}
bail:
if (have_alloc_sem)
up_read(&inode->i_alloc_sem);
if (rw_level != -1)
ocfs2_rw_unlock(inode, rw_level);
mlog_exit(ret);
return ret;
}
/*
* ocfs2_readdir()
*
*/
int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir)
{
int error = 0;
unsigned long offset, blk, last_ra_blk = 0;
int i, stored;
struct buffer_head * bh, * tmp;
struct ocfs2_dir_entry * de;
int err;
struct inode *inode = filp->f_path.dentry->d_inode;
struct super_block * sb = inode->i_sb;
unsigned int ra_sectors = 16;
int lock_level = 0;
mlog_entry("dirino=%llu\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno);
stored = 0;
bh = NULL;
error = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level);
if (lock_level && error >= 0) {
/* We release EX lock which used to update atime
* and get PR lock again to reduce contention
* on commonly accessed directories. */
ocfs2_meta_unlock(inode, 1);
lock_level = 0;
error = ocfs2_meta_lock(inode, NULL, 0);
}
if (error < 0) {
if (error != -ENOENT)
mlog_errno(error);
/* we haven't got any yet, so propagate the error. */
stored = error;
goto bail_nolock;
}
offset = filp->f_pos & (sb->s_blocksize - 1);
while (!error && !stored && filp->f_pos < i_size_read(inode)) {
blk = (filp->f_pos) >> sb->s_blocksize_bits;
bh = ocfs2_bread(inode, blk, &err, 0);
if (!bh) {
mlog(ML_ERROR,
"directory #%llu contains a hole at offset %lld\n",
(unsigned long long)OCFS2_I(inode)->ip_blkno,
filp->f_pos);
filp->f_pos += sb->s_blocksize - offset;
continue;
}
/* The idea here is to begin with 8k read-ahead and to stay
* 4k ahead of our current position.
*
* TODO: Use the pagecache for this. We just need to
* make sure it's cluster-safe... */
if (!last_ra_blk
|| (((last_ra_blk - blk) << 9) <= (ra_sectors / 2))) {
for (i = ra_sectors >> (sb->s_blocksize_bits - 9);
i > 0; i--) {
tmp = ocfs2_bread(inode, ++blk, &err, 1);
if (tmp)
brelse(tmp);
}
last_ra_blk = blk;
ra_sectors = 8;
}
revalidate:
/* If the dir block has changed since the last call to
* readdir(2), then we might be pointing to an invalid
* dirent right now. Scan from the start of the block
* to make sure. */
if (filp->f_version != inode->i_version) {
for (i = 0; i < sb->s_blocksize && i < offset; ) {
de = (struct ocfs2_dir_entry *) (bh->b_data + i);
/* It's too expensive to do a full
* dirent test each time round this
* loop, but we do have to test at
* least that it is non-zero. A
* failure will be detected in the
* dirent test below. */
if (le16_to_cpu(de->rec_len) <
OCFS2_DIR_REC_LEN(1))
break;
i += le16_to_cpu(de->rec_len);
}
offset = i;
filp->f_pos = (filp->f_pos & ~(sb->s_blocksize - 1))
| offset;
filp->f_version = inode->i_version;
}
while (!error && filp->f_pos < i_size_read(inode)
&& offset < sb->s_blocksize) {
de = (struct ocfs2_dir_entry *) (bh->b_data + offset);
if (!ocfs2_check_dir_entry(inode, de, bh, offset)) {
/* On error, skip the f_pos to the
next block. */
filp->f_pos = (filp->f_pos |
(sb->s_blocksize - 1)) + 1;
brelse(bh);
goto bail;
}
offset += le16_to_cpu(de->rec_len);
if (le64_to_cpu(de->inode)) {
/* We might block in the next section
* if the data destination is
* currently swapped out. So, use a
* version stamp to detect whether or
* not the directory has been modified
* during the copy operation.
*/
unsigned long version = filp->f_version;
unsigned char d_type = DT_UNKNOWN;
if (de->file_type < OCFS2_FT_MAX)
d_type = ocfs2_filetype_table[de->file_type];
error = filldir(dirent, de->name,
de->name_len,
filp->f_pos,
ino_from_blkno(sb, le64_to_cpu(de->inode)),
d_type);
if (error)
break;
if (version != filp->f_version)
goto revalidate;
stored ++;
}
filp->f_pos += le16_to_cpu(de->rec_len);
}
offset = 0;
brelse(bh);
}
stored = 0;
bail:
ocfs2_meta_unlock(inode, lock_level);
bail_nolock:
mlog_exit(stored);
return stored;
}
