/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @datasync: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*/
int vfs_fsync(struct file *file, int datasync)
{
if (!fsync_enabled)
return 0;
return vfs_fsync_range(file, 0, LLONG_MAX, datasync);
}
static int wrapfs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
int err;
struct file *lower_file;
struct path lower_path;
struct dentry *dentry = file->f_path.dentry;
#ifdef EXTRA_CREDIT
if(wrapfs_get_debug(file->f_dentry->d_sb) & DEBUG_FILE)
DEBUG_MESG("Enter");
#endif
err = generic_file_fsync(file, start, end, datasync);
if (err)
goto out;
lower_file = wrapfs_lower_file(file);
wrapfs_get_lower_path(dentry, &lower_path);
err = vfs_fsync_range(lower_file, start, end, datasync);
wrapfs_put_lower_path(dentry, &lower_path);
out:
#ifdef EXTRA_CREDIT
if(wrapfs_get_debug(file->f_dentry->d_sb) & DEBUG_FILE)
DEBUG_RETURN("Exit", err);
#endif
return err;
}
/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @datasync: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*/
int vfs_fsync(struct file *file, int datasync)
{
#ifdef CONFIG_FSYNC_OFF
return 0;
#endif
return vfs_fsync_range(file, 0, LLONG_MAX, datasync);
}
/**
* generic_write_sync - perform syncing after a write if file / inode is sync
* @file: file to which the write happened
* @pos: offset where the write started
* @count: length of the write
*
* This is just a simple wrapper about our general syncing function.
*/
int generic_write_sync(struct file *file, loff_t pos, loff_t count)
{
if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
return 0;
return vfs_fsync_range(file, pos, pos + count - 1,
(file->f_flags & __O_SYNC) ? 0 : 1);
}
/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @datasync: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*/
int vfs_fsync(struct file *file, int datasync)
{
#ifdef CONFIG_FSYNC_CONTROL
if (!fsynccontrol_fsync_enabled())
return 0;
#endif
return vfs_fsync_range(file, 0, LLONG_MAX, datasync);
}
/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @datasync: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*/
int vfs_fsync(struct file *file, int datasync)
{
//conditional fsync disable
#ifdef CONFIG_FSYNC_OFF
return 0;
#endif
return vfs_fsync_range(file, 0, LLONG_MAX, datasync);
}
/**
* generic_write_sync - perform syncing after a write if file / inode is sync
* @file: file to which the write happened
* @pos: offset where the write started
* @count: length of the write
*
* This is just a simple wrapper about our general syncing function.
*/
int generic_write_sync(struct file *file, loff_t pos, loff_t count)
{
#ifdef CONFIG_FSYNC_CONTROL
if (!fsynccontrol_fsync_enabled())
return 0;
#endif
if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
return 0;
return vfs_fsync_range(file, pos, pos + count - 1,
(file->f_flags & __O_SYNC) ? 0 : 1);
}
/**
* generic_write_sync - perform syncing after a write if file / inode is sync
* @file: file to which the write happened
* @pos: offset where the write started
* @count: length of the write
*
* This is just a simple wrapper about our general syncing function.
*/
int generic_write_sync(struct file *file, loff_t pos, loff_t count)
{
//conditional fsync disable
#ifdef CONFIG_FSYNC_OFF
return 0;
#endif
if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
return 0;
return vfs_fsync_range(file, pos, pos + count - 1,
(file->f_flags & __O_SYNC) ? 0 : 1);
}
/**
* generic_write_sync - perform syncing after a write if file / inode is sync
* @file: file to which the write happened
* @pos: offset where the write started
* @count: length of the write
*
* This is just a simple wrapper about our general syncing function.
*/
int generic_write_sync(struct file *file, loff_t pos, loff_t count)
{
#ifdef CONFIG_DYNAMIC_FSYNC
if (!early_suspend_active)
return 0;
#endif
if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
return 0;
return vfs_fsync_range(file, pos, pos + count - 1,
(file->f_flags & __O_SYNC) ? 0 : 1);
}
static int
ecryptfs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
int rc = 0;
rc = generic_file_fsync(file, start, end, datasync);
if (rc)
goto out;
rc = vfs_fsync_range(ecryptfs_file_to_lower(file), start, end,
datasync);
out:
return rc;
}
static int
sdcardfs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
int err;
struct file *lower_file;
struct path lower_path;
struct dentry *dentry = file->f_path.dentry;
lower_file = sdcardfs_lower_file(file);
sdcardfs_get_lower_path(dentry, &lower_path);
err = vfs_fsync_range(lower_file, start, end, datasync);
sdcardfs_put_lower_path(dentry, &lower_path);
return err;
}
/*
* WRITE Force Unit Access (FUA) emulation on a per struct se_task
* LBA range basis..
*/
static void fd_emulate_write_fua(struct se_cmd *cmd, struct se_task *task)
{
struct se_device *dev = cmd->se_dev;
struct fd_dev *fd_dev = dev->dev_ptr;
loff_t start = task->task_lba * dev->se_sub_dev->se_dev_attrib.block_size;
loff_t end = start + task->task_size;
int ret;
pr_debug("FILEIO: FUA WRITE LBA: %llu, bytes: %u\n",
task->task_lba, task->task_size);
ret = vfs_fsync_range(fd_dev->fd_file, start, end, 1);
if (ret != 0)
pr_err("FILEIO: vfs_fsync_range() failed: %d\n", ret);
}
static int wrapfs_fsync(struct file *file, loff_t start, loff_t end,
int datasync)
{
int err;
struct file *lower_file;
struct path lower_path;
struct dentry *dentry = file->f_path.dentry;
err = generic_file_fsync(file, start, end, datasync);
if (err)
goto out;
lower_file = wrapfs_lower_file(file);
wrapfs_get_lower_path(dentry, &lower_path);
err = vfs_fsync_range(lower_file, start, end, datasync);
wrapfs_put_lower_path(dentry, &lower_path);
out:
return err;
}
static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
struct fd real;
const struct cred *old_cred;
int ret;
ret = ovl_real_fdget_meta(file, &real, !datasync);
if (ret)
return ret;
/* Don't sync lower file for fear of receiving EROFS error */
if (file_inode(real.file) == ovl_inode_upper(file_inode(file))) {
old_cred = ovl_override_creds(file_inode(file)->i_sb);
ret = vfs_fsync_range(real.file, start, end, datasync);
revert_creds(old_cred);
}
fdput(real);
return ret;
}
static void fd_emulate_sync_cache(struct se_task *task)
{
struct se_cmd *cmd = task->task_se_cmd;
struct se_device *dev = cmd->se_dev;
struct fd_dev *fd_dev = dev->dev_ptr;
int immed = (cmd->t_task_cdb[1] & 0x2);
loff_t start, end;
int ret;
/*
* If the Immediate bit is set, queue up the GOOD response
* for this SYNCHRONIZE_CACHE op
*/
if (immed)
transport_complete_sync_cache(cmd, 1);
/*
* Determine if we will be flushing the entire device.
*/
if (cmd->t_task_lba == 0 && cmd->data_length == 0) {
start = 0;
end = LLONG_MAX;
} else {
start = cmd->t_task_lba * dev->se_sub_dev->se_dev_attrib.block_size;
if (cmd->data_length)
end = start + cmd->data_length;
else
end = LLONG_MAX;
}
ret = vfs_fsync_range(fd_dev->fd_file, start, end, 1);
if (ret != 0)
pr_err("FILEIO: vfs_fsync_range() failed: %d\n", ret);
if (!immed)
transport_complete_sync_cache(cmd, ret == 0);
}
/**
* vfs_fsync - perform a fsync or fdatasync on a file
* @file: file to sync
* @datasync: only perform a fdatasync operation
*
* Write back data and metadata for @file to disk. If @datasync is
* set only metadata needed to access modified file data is written.
*/
int vfs_fsync(struct file *file, int datasync)
{
return vfs_fsync_range(file, 0, LLONG_MAX, datasync);
}
static int aio_submit(struct aio_output *output, struct aio_mref_aspect *mref_a, bool use_fdsync)
{
struct mref_object *mref = mref_a->object;
mm_segment_t oldfs;
int res;
struct iocb iocb = {
.aio_data = (__u64)mref_a,
.aio_lio_opcode = use_fdsync ? IOCB_CMD_FDSYNC : (mref->ref_rw != 0 ? IOCB_CMD_PWRITE : IOCB_CMD_PREAD),
.aio_fildes = output->fd,
.aio_buf = (unsigned long)mref->ref_data,
.aio_nbytes = mref->ref_len,
.aio_offset = mref->ref_pos,
// .aio_reqprio = something(mref->ref_prio) field exists, but not yet implemented in kernelspace :(
};
struct iocb *iocbp = &iocb;
unsigned long long latency;
mars_trace(mref, "aio_submit");
if (unlikely(output->fd < 0)) {
MARS_ERR("bad fd = %d\n", output->fd);
res = -EBADF;
goto done;
}
oldfs = get_fs();
set_fs(get_ds());
latency = TIME_STATS(&timings[mref->ref_rw & 1], res = sys_io_submit(output->ctxp, 1, &iocbp));
set_fs(oldfs);
threshold_check(&aio_submit_threshold, latency);
atomic_inc(&output->total_submit_count);
if (likely(res >= 0)) {
atomic_inc(&output->submit_count);
} else if (likely(res == -EAGAIN)) {
atomic_inc(&output->total_again_count);
} else {
MARS_ERR("error = %d\n", res);
}
done:
return res;
}
static int aio_submit_dummy(struct aio_output *output)
{
mm_segment_t oldfs;
int res;
int dummy;
struct iocb iocb = {
.aio_buf = (__u64)&dummy,
};
struct iocb *iocbp = &iocb;
oldfs = get_fs();
set_fs(get_ds());
res = sys_io_submit(output->ctxp, 1, &iocbp);
set_fs(oldfs);
if (likely(res >= 0)) {
atomic_inc(&output->submit_count);
}
return res;
}
static
int aio_start_thread(
struct aio_output *output,
struct aio_threadinfo *tinfo,
int(*fn)(void*),
char class)
{
int j;
for (j = 0; j < MARS_PRIO_NR; j++) {
INIT_LIST_HEAD(&tinfo->mref_list[j]);
}
tinfo->output = output;
spin_lock_init(&tinfo->lock);
init_waitqueue_head(&tinfo->event);
init_waitqueue_head(&tinfo->terminate_event);
tinfo->terminated = false;
tinfo->thread = brick_thread_create(fn, tinfo, "mars_aio_%c%d", class, output->index);
if (unlikely(!tinfo->thread)) {
MARS_ERR("cannot create thread\n");
return -ENOENT;
}
return 0;
}
static
void aio_stop_thread(struct aio_output *output, int i, bool do_submit_dummy)
{
struct aio_threadinfo *tinfo = &output->tinfo[i];
if (tinfo->thread) {
MARS_DBG("stopping thread %d ...\n", i);
brick_thread_stop_nowait(tinfo->thread);
// workaround for waking up the receiver thread. TODO: check whether signal handlong could do better.
if (do_submit_dummy) {
MARS_DBG("submitting dummy for wakeup %d...\n", i);
use_fake_mm();
aio_submit_dummy(output);
if (likely(current->mm)) {
unuse_fake_mm();
}
}
// wait for termination
MARS_DBG("waiting for thread %d ...\n", i);
wait_event_interruptible_timeout(
tinfo->terminate_event,
tinfo->terminated,
(60 - i * 2) * HZ);
if (likely(tinfo->terminated)) {
brick_thread_stop(tinfo->thread);
} else {
MARS_ERR("thread %d did not terminate - leaving a zombie\n", i);
}
}
}
static
int aio_sync(struct file *file)
{
int err;
switch (aio_sync_mode) {
case 1:
#if defined(S_BIAS) || (defined(RHEL_MAJOR) && (RHEL_MAJOR < 7))
err = vfs_fsync_range(file, file->f_path.dentry, 0, LLONG_MAX, 1);
#else
err = vfs_fsync_range(file, 0, LLONG_MAX, 1);
#endif
break;
case 2:
#if defined(S_BIAS) || (defined(RHEL_MAJOR) && (RHEL_MAJOR < 7))
err = vfs_fsync_range(file, file->f_path.dentry, 0, LLONG_MAX, 0);
#else
err = vfs_fsync_range(file, 0, LLONG_MAX, 0);
#endif
break;
default:
err = filemap_write_and_wait_range(file->f_mapping, 0, LLONG_MAX);
}
return err;
}
static int gfs2_page_mkwrite(struct vm_fault *vmf)
{
struct page *page = vmf->page;
struct inode *inode = file_inode(vmf->vma->vm_file);
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_sbd *sdp = GFS2_SB(inode);
struct gfs2_alloc_parms ap = { .aflags = 0, };
unsigned long last_index;
u64 pos = page->index << PAGE_SHIFT;
unsigned int data_blocks, ind_blocks, rblocks;
struct gfs2_holder gh;
loff_t size;
int ret;
sb_start_pagefault(inode->i_sb);
ret = gfs2_rsqa_alloc(ip);
if (ret)
goto out;
gfs2_size_hint(vmf->vma->vm_file, pos, PAGE_SIZE);
gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
ret = gfs2_glock_nq(&gh);
if (ret)
goto out_uninit;
/* Update file times before taking page lock */
file_update_time(vmf->vma->vm_file);
set_bit(GLF_DIRTY, &ip->i_gl->gl_flags);
set_bit(GIF_SW_PAGED, &ip->i_flags);
if (!gfs2_write_alloc_required(ip, pos, PAGE_SIZE)) {
lock_page(page);
if (!PageUptodate(page) || page->mapping != inode->i_mapping) {
ret = -EAGAIN;
unlock_page(page);
}
goto out_unlock;
}
ret = gfs2_rindex_update(sdp);
if (ret)
goto out_unlock;
gfs2_write_calc_reserv(ip, PAGE_SIZE, &data_blocks, &ind_blocks);
ap.target = data_blocks + ind_blocks;
ret = gfs2_quota_lock_check(ip, &ap);
if (ret)
goto out_unlock;
ret = gfs2_inplace_reserve(ip, &ap);
if (ret)
goto out_quota_unlock;
rblocks = RES_DINODE + ind_blocks;
if (gfs2_is_jdata(ip))
rblocks += data_blocks ? data_blocks : 1;
if (ind_blocks || data_blocks) {
rblocks += RES_STATFS + RES_QUOTA;
rblocks += gfs2_rg_blocks(ip, data_blocks + ind_blocks);
}
ret = gfs2_trans_begin(sdp, rblocks, 0);
if (ret)
goto out_trans_fail;
lock_page(page);
ret = -EINVAL;
size = i_size_read(inode);
last_index = (size - 1) >> PAGE_SHIFT;
/* Check page index against inode size */
if (size == 0 || (page->index > last_index))
goto out_trans_end;
ret = -EAGAIN;
/* If truncated, we must retry the operation, we may have raced
* with the glock demotion code.
*/
if (!PageUptodate(page) || page->mapping != inode->i_mapping)
goto out_trans_end;
/* Unstuff, if required, and allocate backing blocks for page */
ret = 0;
if (gfs2_is_stuffed(ip))
ret = gfs2_unstuff_dinode(ip, page);
if (ret == 0)
ret = gfs2_allocate_page_backing(page);
out_trans_end:
if (ret)
unlock_page(page);
gfs2_trans_end(sdp);
out_trans_fail:
gfs2_inplace_release(ip);
out_quota_unlock:
gfs2_quota_unlock(ip);
out_unlock:
gfs2_glock_dq(&gh);
out_uninit:
gfs2_holder_uninit(&gh);
if (ret == 0) {
set_page_dirty(page);
wait_for_stable_page(page);
}
out:
sb_end_pagefault(inode->i_sb);
return block_page_mkwrite_return(ret);
}
static const struct vm_operations_struct gfs2_vm_ops = {
.fault = filemap_fault,
.map_pages = filemap_map_pages,
.page_mkwrite = gfs2_page_mkwrite,
};
/**
* gfs2_mmap -
* @file: The file to map
* @vma: The VMA which described the mapping
*
* There is no need to get a lock here unless we should be updating
* atime. We ignore any locking errors since the only consequence is
* a missed atime update (which will just be deferred until later).
*
* Returns: 0
*/
static int gfs2_mmap(struct file *file, struct vm_area_struct *vma)
{
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
if (!(file->f_flags & O_NOATIME) &&
!IS_NOATIME(&ip->i_inode)) {
struct gfs2_holder i_gh;
int error;
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
&i_gh);
if (error)
return error;
/* grab lock to update inode */
gfs2_glock_dq_uninit(&i_gh);
file_accessed(file);
}
vma->vm_ops = &gfs2_vm_ops;
return 0;
}
/**
* gfs2_open_common - This is common to open and atomic_open
* @inode: The inode being opened
* @file: The file being opened
*
* This maybe called under a glock or not depending upon how it has
* been called. We must always be called under a glock for regular
* files, however. For other file types, it does not matter whether
* we hold the glock or not.
*
* Returns: Error code or 0 for success
*/
int gfs2_open_common(struct inode *inode, struct file *file)
{
struct gfs2_file *fp;
int ret;
if (S_ISREG(inode->i_mode)) {
ret = generic_file_open(inode, file);
if (ret)
return ret;
}
fp = kzalloc(sizeof(struct gfs2_file), GFP_NOFS);
if (!fp)
return -ENOMEM;
mutex_init(&fp->f_fl_mutex);
gfs2_assert_warn(GFS2_SB(inode), !file->private_data);
file->private_data = fp;
return 0;
}
/**
* gfs2_open - open a file
* @inode: the inode to open
* @file: the struct file for this opening
*
* After atomic_open, this function is only used for opening files
* which are already cached. We must still get the glock for regular
* files to ensure that we have the file size uptodate for the large
* file check which is in the common code. That is only an issue for
* regular files though.
*
* Returns: errno
*/
static int gfs2_open(struct inode *inode, struct file *file)
{
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_holder i_gh;
int error;
bool need_unlock = false;
if (S_ISREG(ip->i_inode.i_mode)) {
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
&i_gh);
if (error)
return error;
need_unlock = true;
}
error = gfs2_open_common(inode, file);
if (need_unlock)
gfs2_glock_dq_uninit(&i_gh);
return error;
}
/**
* gfs2_release - called to close a struct file
* @inode: the inode the struct file belongs to
* @file: the struct file being closed
*
* Returns: errno
*/
static int gfs2_release(struct inode *inode, struct file *file)
{
struct gfs2_inode *ip = GFS2_I(inode);
kfree(file->private_data);
file->private_data = NULL;
if (!(file->f_mode & FMODE_WRITE))
return 0;
gfs2_rsqa_delete(ip, &inode->i_writecount);
return 0;
}
/**
* gfs2_fsync - sync the dirty data for a file (across the cluster)
* @file: the file that points to the dentry
* @start: the start position in the file to sync
* @end: the end position in the file to sync
* @datasync: set if we can ignore timestamp changes
*
* We split the data flushing here so that we don't wait for the data
* until after we've also sent the metadata to disk. Note that for
* data=ordered, we will write & wait for the data at the log flush
* stage anyway, so this is unlikely to make much of a difference
* except in the data=writeback case.
*
* If the fdatawrite fails due to any reason except -EIO, we will
* continue the remainder of the fsync, although we'll still report
* the error at the end. This is to match filemap_write_and_wait_range()
* behaviour.
*
* Returns: errno
*/
static int gfs2_fsync(struct file *file, loff_t start, loff_t end,
int datasync)
{
struct address_space *mapping = file->f_mapping;
struct inode *inode = mapping->host;
int sync_state = inode->i_state & I_DIRTY_ALL;
struct gfs2_inode *ip = GFS2_I(inode);
int ret = 0, ret1 = 0;
if (mapping->nrpages) {
ret1 = filemap_fdatawrite_range(mapping, start, end);
if (ret1 == -EIO)
return ret1;
}
if (!gfs2_is_jdata(ip))
sync_state &= ~I_DIRTY_PAGES;
if (datasync)
sync_state &= ~(I_DIRTY_SYNC | I_DIRTY_TIME);
if (sync_state) {
ret = sync_inode_metadata(inode, 1);
if (ret)
return ret;
if (gfs2_is_jdata(ip))
ret = file_write_and_wait(file);
if (ret)
return ret;
gfs2_ail_flush(ip->i_gl, 1);
}
if (mapping->nrpages)
ret = file_fdatawait_range(file, start, end);
return ret ? ret : ret1;
}
/**
* gfs2_file_write_iter - Perform a write to a file
* @iocb: The io context
* @iov: The data to write
* @nr_segs: Number of @iov segments
* @pos: The file position
*
* We have to do a lock/unlock here to refresh the inode size for
* O_APPEND writes, otherwise we can land up writing at the wrong
* offset. There is still a race, but provided the app is using its
* own file locking, this will make O_APPEND work as expected.
*
*/
static ssize_t gfs2_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
{
struct file *file = iocb->ki_filp;
struct gfs2_inode *ip = GFS2_I(file_inode(file));
int ret;
ret = gfs2_rsqa_alloc(ip);
if (ret)
return ret;
gfs2_size_hint(file, iocb->ki_pos, iov_iter_count(from));
if (iocb->ki_flags & IOCB_APPEND) {
struct gfs2_holder gh;
ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, 0, &gh);
if (ret)
return ret;
gfs2_glock_dq_uninit(&gh);
}
return generic_file_write_iter(iocb, from);
}
static int fallocate_chunk(struct inode *inode, loff_t offset, loff_t len,
int mode)
{
struct gfs2_inode *ip = GFS2_I(inode);
struct buffer_head *dibh;
int error;
unsigned int nr_blks;
sector_t lblock = offset >> inode->i_blkbits;
error = gfs2_meta_inode_buffer(ip, &dibh);
if (unlikely(error))
return error;
gfs2_trans_add_meta(ip->i_gl, dibh);
if (gfs2_is_stuffed(ip)) {
error = gfs2_unstuff_dinode(ip, NULL);
if (unlikely(error))
goto out;
}
while (len) {
struct buffer_head bh_map = { .b_state = 0, .b_blocknr = 0 };
bh_map.b_size = len;
set_buffer_zeronew(&bh_map);
error = gfs2_block_map(inode, lblock, &bh_map, 1);
if (unlikely(error))
goto out;
len -= bh_map.b_size;
nr_blks = bh_map.b_size >> inode->i_blkbits;
lblock += nr_blks;
if (!buffer_new(&bh_map))
continue;
if (unlikely(!buffer_zeronew(&bh_map))) {
error = -EIO;
goto out;
}
}
out:
brelse(dibh);
return error;
}
/**
* calc_max_reserv() - Reverse of write_calc_reserv. Given a number of
* blocks, determine how many bytes can be written.
* @ip: The inode in question.
* @len: Max cap of bytes. What we return in *len must be <= this.
* @data_blocks: Compute and return the number of data blocks needed
* @ind_blocks: Compute and return the number of indirect blocks needed
* @max_blocks: The total blocks available to work with.
*
* Returns: void, but @len, @data_blocks and @ind_blocks are filled in.
*/
static void calc_max_reserv(struct gfs2_inode *ip, loff_t *len,
unsigned int *data_blocks, unsigned int *ind_blocks,
unsigned int max_blocks)
{
loff_t max = *len;
const struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
unsigned int tmp, max_data = max_blocks - 3 * (sdp->sd_max_height - 1);
for (tmp = max_data; tmp > sdp->sd_diptrs;) {
tmp = DIV_ROUND_UP(tmp, sdp->sd_inptrs);
max_data -= tmp;
}
*data_blocks = max_data;
*ind_blocks = max_blocks - max_data;
*len = ((loff_t)max_data - 3) << sdp->sd_sb.sb_bsize_shift;
if (*len > max) {
*len = max;
gfs2_write_calc_reserv(ip, max, data_blocks, ind_blocks);
}
}
static long __gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
{
struct inode *inode = file_inode(file);
struct gfs2_sbd *sdp = GFS2_SB(inode);
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_alloc_parms ap = { .aflags = 0, };
unsigned int data_blocks = 0, ind_blocks = 0, rblocks;
loff_t bytes, max_bytes, max_blks = UINT_MAX;
int error;
const loff_t pos = offset;
const loff_t count = len;
loff_t bsize_mask = ~((loff_t)sdp->sd_sb.sb_bsize - 1);
loff_t next = (offset + len - 1) >> sdp->sd_sb.sb_bsize_shift;
loff_t max_chunk_size = UINT_MAX & bsize_mask;
next = (next + 1) << sdp->sd_sb.sb_bsize_shift;
offset &= bsize_mask;
len = next - offset;
bytes = sdp->sd_max_rg_data * sdp->sd_sb.sb_bsize / 2;
if (!bytes)
bytes = UINT_MAX;
bytes &= bsize_mask;
if (bytes == 0)
bytes = sdp->sd_sb.sb_bsize;
gfs2_size_hint(file, offset, len);
gfs2_write_calc_reserv(ip, PAGE_SIZE, &data_blocks, &ind_blocks);
ap.min_target = data_blocks + ind_blocks;
while (len > 0) {
if (len < bytes)
bytes = len;
if (!gfs2_write_alloc_required(ip, offset, bytes)) {
len -= bytes;
offset += bytes;
continue;
}
/* We need to determine how many bytes we can actually
* fallocate without exceeding quota or going over the
* end of the fs. We start off optimistically by assuming
* we can write max_bytes */
max_bytes = (len > max_chunk_size) ? max_chunk_size : len;
/* Since max_bytes is most likely a theoretical max, we
* calculate a more realistic 'bytes' to serve as a good
* starting point for the number of bytes we may be able
* to write */
gfs2_write_calc_reserv(ip, bytes, &data_blocks, &ind_blocks);
ap.target = data_blocks + ind_blocks;
error = gfs2_quota_lock_check(ip, &ap);
if (error)
return error;
/* ap.allowed tells us how many blocks quota will allow
* us to write. Check if this reduces max_blks */
if (ap.allowed && ap.allowed < max_blks)
max_blks = ap.allowed;
error = gfs2_inplace_reserve(ip, &ap);
if (error)
goto out_qunlock;
/* check if the selected rgrp limits our max_blks further */
if (ap.allowed && ap.allowed < max_blks)
max_blks = ap.allowed;
/* Almost done. Calculate bytes that can be written using
* max_blks. We also recompute max_bytes, data_blocks and
* ind_blocks */
calc_max_reserv(ip, &max_bytes, &data_blocks,
&ind_blocks, max_blks);
rblocks = RES_DINODE + ind_blocks + RES_STATFS + RES_QUOTA +
RES_RG_HDR + gfs2_rg_blocks(ip, data_blocks + ind_blocks);
if (gfs2_is_jdata(ip))
rblocks += data_blocks ? data_blocks : 1;
error = gfs2_trans_begin(sdp, rblocks,
PAGE_SIZE/sdp->sd_sb.sb_bsize);
if (error)
goto out_trans_fail;
error = fallocate_chunk(inode, offset, max_bytes, mode);
gfs2_trans_end(sdp);
if (error)
goto out_trans_fail;
len -= max_bytes;
offset += max_bytes;
gfs2_inplace_release(ip);
gfs2_quota_unlock(ip);
}
if (!(mode & FALLOC_FL_KEEP_SIZE) && (pos + count) > inode->i_size) {
i_size_write(inode, pos + count);
file_update_time(file);
mark_inode_dirty(inode);
}
if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
return vfs_fsync_range(file, pos, pos + count - 1,
(file->f_flags & __O_SYNC) ? 0 : 1);
return 0;
out_trans_fail:
gfs2_inplace_release(ip);
out_qunlock:
gfs2_quota_unlock(ip);
return error;
}
static long gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
{
struct inode *inode = file_inode(file);
struct gfs2_sbd *sdp = GFS2_SB(inode);
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_holder gh;
int ret;
if (mode & ~FALLOC_FL_KEEP_SIZE)
return -EOPNOTSUPP;
/* fallocate is needed by gfs2_grow to reserve space in the rindex */
if (gfs2_is_jdata(ip) && inode != sdp->sd_rindex)
return -EOPNOTSUPP;
inode_lock(inode);
gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
ret = gfs2_glock_nq(&gh);
if (ret)
goto out_uninit;
if (!(mode & FALLOC_FL_KEEP_SIZE) &&
(offset + len) > inode->i_size) {
ret = inode_newsize_ok(inode, offset + len);
if (ret)
goto out_unlock;
}
ret = get_write_access(inode);
if (ret)
goto out_unlock;
ret = gfs2_rsqa_alloc(ip);
if (ret)
goto out_putw;
ret = __gfs2_fallocate(file, mode, offset, len);
if (ret)
gfs2_rs_deltree(&ip->i_res);
out_putw:
put_write_access(inode);
out_unlock:
gfs2_glock_dq(&gh);
out_uninit:
gfs2_holder_uninit(&gh);
inode_unlock(inode);
return ret;
}
static ssize_t gfs2_file_splice_write(struct pipe_inode_info *pipe,
struct file *out, loff_t *ppos,
size_t len, unsigned int flags)
{
int error;
struct gfs2_inode *ip = GFS2_I(out->f_mapping->host);
error = gfs2_rsqa_alloc(ip);
if (error)
return (ssize_t)error;
gfs2_size_hint(out, *ppos, len);
return iter_file_splice_write(pipe, out, ppos, len, flags);
}
#ifdef CONFIG_GFS2_FS_LOCKING_DLM
/**
* gfs2_lock - acquire/release a posix lock on a file
* @file: the file pointer
* @cmd: either modify or retrieve lock state, possibly wait
* @fl: type and range of lock
*
* Returns: errno
*/
static int gfs2_lock(struct file *file, int cmd, struct file_lock *fl)
{
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
struct gfs2_sbd *sdp = GFS2_SB(file->f_mapping->host);
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
if (!(fl->fl_flags & FL_POSIX))
return -ENOLCK;
if (__mandatory_lock(&ip->i_inode) && fl->fl_type != F_UNLCK)
return -ENOLCK;
if (cmd == F_CANCELLK) {
/* Hack: */
cmd = F_SETLK;
fl->fl_type = F_UNLCK;
}
if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) {
if (fl->fl_type == F_UNLCK)
locks_lock_file_wait(file, fl);
return -EIO;
}
if (IS_GETLK(cmd))
return dlm_posix_get(ls->ls_dlm, ip->i_no_addr, file, fl);
else if (fl->fl_type == F_UNLCK)
return dlm_posix_unlock(ls->ls_dlm, ip->i_no_addr, file, fl);
else
return dlm_posix_lock(ls->ls_dlm, ip->i_no_addr, file, cmd, fl);
}
static int do_flock(struct file *file, int cmd, struct file_lock *fl)
{
struct gfs2_file *fp = file->private_data;
struct gfs2_holder *fl_gh = &fp->f_fl_gh;
struct gfs2_inode *ip = GFS2_I(file_inode(file));
struct gfs2_glock *gl;
unsigned int state;
u16 flags;
int error = 0;
int sleeptime;
state = (fl->fl_type == F_WRLCK) ? LM_ST_EXCLUSIVE : LM_ST_SHARED;
flags = (IS_SETLKW(cmd) ? 0 : LM_FLAG_TRY_1CB) | GL_EXACT;
mutex_lock(&fp->f_fl_mutex);
if (gfs2_holder_initialized(fl_gh)) {
if (fl_gh->gh_state == state)
goto out;
locks_lock_file_wait(file,
&(struct file_lock) {
.fl_type = F_UNLCK,
.fl_flags = FL_FLOCK
});
gfs2_glock_dq(fl_gh);
gfs2_holder_reinit(state, flags, fl_gh);
} else {