2006-01-16 17:50:04 +01:00
|
|
|
/*
|
|
|
|
* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
|
2006-05-18 21:09:15 +02:00
|
|
|
* Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
|
2006-01-16 17:50:04 +01:00
|
|
|
*
|
|
|
|
* This copyrighted material is made available to anyone wishing to use,
|
|
|
|
* modify, copy, or redistribute it subject to the terms and conditions
|
2006-09-01 17:05:15 +02:00
|
|
|
* of the GNU General Public License version 2.
|
2006-01-16 17:50:04 +01:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/spinlock.h>
|
|
|
|
#include <linux/completion.h>
|
|
|
|
#include <linux/buffer_head.h>
|
|
|
|
#include <linux/pagemap.h>
|
|
|
|
#include <linux/uio.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/mm.h>
|
2008-07-02 21:12:01 +02:00
|
|
|
#include <linux/mount.h>
|
2006-02-08 12:50:51 +01:00
|
|
|
#include <linux/fs.h>
|
2006-02-27 23:23:27 +01:00
|
|
|
#include <linux/gfs2_ondisk.h>
|
2011-01-14 13:07:43 +01:00
|
|
|
#include <linux/falloc.h>
|
|
|
|
#include <linux/swap.h>
|
2006-03-28 21:14:04 +02:00
|
|
|
#include <linux/crc32.h>
|
2006-11-30 16:14:32 +01:00
|
|
|
#include <linux/writeback.h>
|
2016-12-24 20:46:01 +01:00
|
|
|
#include <linux/uaccess.h>
|
2009-01-12 11:43:39 +01:00
|
|
|
#include <linux/dlm.h>
|
|
|
|
#include <linux/dlm_plock.h>
|
2014-08-20 18:44:45 +02:00
|
|
|
#include <linux/delay.h>
|
2018-06-24 16:04:04 +02:00
|
|
|
#include <linux/backing-dev.h>
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
#include "gfs2.h"
|
2006-02-27 23:23:27 +01:00
|
|
|
#include "incore.h"
|
2006-01-16 17:50:04 +01:00
|
|
|
#include "bmap.h"
|
2018-06-24 16:04:04 +02:00
|
|
|
#include "aops.h"
|
2006-01-16 17:50:04 +01:00
|
|
|
#include "dir.h"
|
|
|
|
#include "glock.h"
|
|
|
|
#include "glops.h"
|
|
|
|
#include "inode.h"
|
|
|
|
#include "log.h"
|
|
|
|
#include "meta_io.h"
|
|
|
|
#include "quota.h"
|
|
|
|
#include "rgrp.h"
|
|
|
|
#include "trans.h"
|
2006-02-27 23:23:27 +01:00
|
|
|
#include "util.h"
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
/**
|
|
|
|
* gfs2_llseek - seek to a location in a file
|
|
|
|
* @file: the file
|
|
|
|
* @offset: the offset
|
2012-12-18 00:59:39 +01:00
|
|
|
* @whence: Where to seek from (SEEK_SET, SEEK_CUR, or SEEK_END)
|
2006-01-16 17:50:04 +01:00
|
|
|
*
|
|
|
|
* SEEK_END requires the glock for the file because it references the
|
|
|
|
* file's size.
|
|
|
|
*
|
|
|
|
* Returns: The new offset, or errno
|
|
|
|
*/
|
|
|
|
|
2012-12-18 00:59:39 +01:00
|
|
|
static loff_t gfs2_llseek(struct file *file, loff_t offset, int whence)
|
2006-01-16 17:50:04 +01:00
|
|
|
{
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
|
2006-01-16 17:50:04 +01:00
|
|
|
struct gfs2_holder i_gh;
|
|
|
|
loff_t error;
|
|
|
|
|
2012-12-18 00:59:39 +01:00
|
|
|
switch (whence) {
|
2017-03-15 19:12:59 +01:00
|
|
|
case SEEK_END:
|
2006-01-16 17:50:04 +01:00
|
|
|
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
|
|
|
|
&i_gh);
|
|
|
|
if (!error) {
|
2012-12-18 00:59:39 +01:00
|
|
|
error = generic_file_llseek(file, offset, whence);
|
2006-01-16 17:50:04 +01:00
|
|
|
gfs2_glock_dq_uninit(&i_gh);
|
|
|
|
}
|
2011-08-23 11:19:25 +02:00
|
|
|
break;
|
2017-03-15 19:12:59 +01:00
|
|
|
|
|
|
|
case SEEK_DATA:
|
|
|
|
error = gfs2_seek_data(file, offset);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SEEK_HOLE:
|
|
|
|
error = gfs2_seek_hole(file, offset);
|
|
|
|
break;
|
|
|
|
|
2011-08-23 11:19:25 +02:00
|
|
|
case SEEK_CUR:
|
|
|
|
case SEEK_SET:
|
2017-03-15 19:12:59 +01:00
|
|
|
/*
|
|
|
|
* These don't reference inode->i_size and don't depend on the
|
|
|
|
* block mapping, so we don't need the glock.
|
|
|
|
*/
|
2012-12-18 00:59:39 +01:00
|
|
|
error = generic_file_llseek(file, offset, whence);
|
2011-08-23 11:19:25 +02:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
error = -EINVAL;
|
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2013-05-16 20:14:48 +02:00
|
|
|
* gfs2_readdir - Iterator for a directory
|
2006-01-16 17:50:04 +01:00
|
|
|
* @file: The directory to read from
|
2013-05-16 20:14:48 +02:00
|
|
|
* @ctx: What to feed directory entries to
|
2006-01-16 17:50:04 +01:00
|
|
|
*
|
|
|
|
* Returns: errno
|
|
|
|
*/
|
|
|
|
|
2013-05-16 20:14:48 +02:00
|
|
|
static int gfs2_readdir(struct file *file, struct dir_context *ctx)
|
2006-01-16 17:50:04 +01:00
|
|
|
{
|
2006-03-28 21:14:04 +02:00
|
|
|
struct inode *dir = file->f_mapping->host;
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *dip = GFS2_I(dir);
|
2006-01-16 17:50:04 +01:00
|
|
|
struct gfs2_holder d_gh;
|
|
|
|
int error;
|
|
|
|
|
2013-05-16 20:14:48 +02:00
|
|
|
error = gfs2_glock_nq_init(dip->i_gl, LM_ST_SHARED, 0, &d_gh);
|
|
|
|
if (error)
|
2006-01-16 17:50:04 +01:00
|
|
|
return error;
|
|
|
|
|
2013-05-16 20:14:48 +02:00
|
|
|
error = gfs2_dir_read(dir, ctx, &file->f_ra);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
gfs2_glock_dq_uninit(&d_gh);
|
|
|
|
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2006-10-02 17:24:43 +02:00
|
|
|
/**
|
2017-10-09 16:15:30 +02:00
|
|
|
* fsflag_gfs2flag
|
2006-10-02 17:24:43 +02:00
|
|
|
*
|
2017-10-09 16:15:30 +02:00
|
|
|
* The FS_JOURNAL_DATA_FL flag maps to GFS2_DIF_INHERIT_JDATA for directories,
|
|
|
|
* and to GFS2_DIF_JDATA for non-directories.
|
2006-10-02 17:24:43 +02:00
|
|
|
*/
|
2017-10-09 16:15:30 +02:00
|
|
|
static struct {
|
|
|
|
u32 fsflag;
|
|
|
|
u32 gfsflag;
|
|
|
|
} fsflag_gfs2flag[] = {
|
|
|
|
{FS_SYNC_FL, GFS2_DIF_SYNC},
|
|
|
|
{FS_IMMUTABLE_FL, GFS2_DIF_IMMUTABLE},
|
|
|
|
{FS_APPEND_FL, GFS2_DIF_APPENDONLY},
|
|
|
|
{FS_NOATIME_FL, GFS2_DIF_NOATIME},
|
|
|
|
{FS_INDEX_FL, GFS2_DIF_EXHASH},
|
|
|
|
{FS_TOPDIR_FL, GFS2_DIF_TOPDIR},
|
|
|
|
{FS_JOURNAL_DATA_FL, GFS2_DIF_JDATA | GFS2_DIF_INHERIT_JDATA},
|
2006-03-31 22:01:28 +02:00
|
|
|
};
|
2006-03-28 21:14:04 +02:00
|
|
|
|
2006-04-07 17:17:32 +02:00
|
|
|
static int gfs2_get_flags(struct file *filp, u32 __user *ptr)
|
2006-03-28 21:14:04 +02:00
|
|
|
{
|
2013-01-23 23:07:38 +01:00
|
|
|
struct inode *inode = file_inode(filp);
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2006-03-28 21:14:04 +02:00
|
|
|
struct gfs2_holder gh;
|
2017-10-09 16:15:30 +02:00
|
|
|
int i, error;
|
|
|
|
u32 gfsflags, fsflags = 0;
|
2006-03-28 21:14:04 +02:00
|
|
|
|
2008-09-18 14:53:59 +02:00
|
|
|
gfs2_holder_init(ip->i_gl, LM_ST_SHARED, 0, &gh);
|
|
|
|
error = gfs2_glock_nq(&gh);
|
2006-03-28 21:14:04 +02:00
|
|
|
if (error)
|
2016-04-20 01:57:45 +02:00
|
|
|
goto out_uninit;
|
2006-09-25 15:26:04 +02:00
|
|
|
|
2017-10-09 16:15:30 +02:00
|
|
|
gfsflags = ip->i_diskflags;
|
|
|
|
if (S_ISDIR(inode->i_mode))
|
|
|
|
gfsflags &= ~GFS2_DIF_JDATA;
|
|
|
|
else
|
|
|
|
gfsflags &= ~GFS2_DIF_INHERIT_JDATA;
|
|
|
|
for (i = 0; i < ARRAY_SIZE(fsflag_gfs2flag); i++)
|
|
|
|
if (gfsflags & fsflag_gfs2flag[i].gfsflag)
|
|
|
|
fsflags |= fsflag_gfs2flag[i].fsflag;
|
|
|
|
|
2006-10-02 17:24:43 +02:00
|
|
|
if (put_user(fsflags, ptr))
|
2006-03-28 21:14:04 +02:00
|
|
|
error = -EFAULT;
|
|
|
|
|
2007-10-15 16:40:33 +02:00
|
|
|
gfs2_glock_dq(&gh);
|
2016-04-20 01:57:45 +02:00
|
|
|
out_uninit:
|
2006-03-28 21:14:04 +02:00
|
|
|
gfs2_holder_uninit(&gh);
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2006-11-08 18:51:06 +01:00
|
|
|
void gfs2_set_inode_flags(struct inode *inode)
|
|
|
|
{
|
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
unsigned int flags = inode->i_flags;
|
|
|
|
|
2011-06-16 15:06:55 +02:00
|
|
|
flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC|S_NOSEC);
|
|
|
|
if ((ip->i_eattr == 0) && !is_sxid(inode->i_mode))
|
2015-05-05 19:25:48 +02:00
|
|
|
flags |= S_NOSEC;
|
2008-11-04 11:05:22 +01:00
|
|
|
if (ip->i_diskflags & GFS2_DIF_IMMUTABLE)
|
2006-11-08 18:51:06 +01:00
|
|
|
flags |= S_IMMUTABLE;
|
2008-11-04 11:05:22 +01:00
|
|
|
if (ip->i_diskflags & GFS2_DIF_APPENDONLY)
|
2006-11-08 18:51:06 +01:00
|
|
|
flags |= S_APPEND;
|
2008-11-04 11:05:22 +01:00
|
|
|
if (ip->i_diskflags & GFS2_DIF_NOATIME)
|
2006-11-08 18:51:06 +01:00
|
|
|
flags |= S_NOATIME;
|
2008-11-04 11:05:22 +01:00
|
|
|
if (ip->i_diskflags & GFS2_DIF_SYNC)
|
2006-11-08 18:51:06 +01:00
|
|
|
flags |= S_SYNC;
|
|
|
|
inode->i_flags = flags;
|
|
|
|
}
|
|
|
|
|
2006-03-28 21:14:04 +02:00
|
|
|
/* Flags that can be set by user space */
|
|
|
|
#define GFS2_FLAGS_USER_SET (GFS2_DIF_JDATA| \
|
|
|
|
GFS2_DIF_IMMUTABLE| \
|
|
|
|
GFS2_DIF_APPENDONLY| \
|
|
|
|
GFS2_DIF_NOATIME| \
|
|
|
|
GFS2_DIF_SYNC| \
|
2012-05-28 16:26:56 +02:00
|
|
|
GFS2_DIF_TOPDIR| \
|
2006-03-28 21:14:04 +02:00
|
|
|
GFS2_DIF_INHERIT_JDATA)
|
|
|
|
|
|
|
|
/**
|
2014-05-15 18:57:08 +02:00
|
|
|
* do_gfs2_set_flags - set flags on an inode
|
|
|
|
* @filp: file pointer
|
|
|
|
* @reqflags: The flags to set
|
2006-03-28 21:14:04 +02:00
|
|
|
* @mask: Indicates which flags are valid
|
|
|
|
*
|
|
|
|
*/
|
2006-04-07 17:17:32 +02:00
|
|
|
static int do_gfs2_set_flags(struct file *filp, u32 reqflags, u32 mask)
|
2006-03-28 21:14:04 +02:00
|
|
|
{
|
2013-01-23 23:07:38 +01:00
|
|
|
struct inode *inode = file_inode(filp);
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
2006-03-28 21:14:04 +02:00
|
|
|
struct buffer_head *bh;
|
|
|
|
struct gfs2_holder gh;
|
|
|
|
int error;
|
2006-04-04 20:29:30 +02:00
|
|
|
u32 new_flags, flags;
|
2006-03-28 21:14:04 +02:00
|
|
|
|
2011-11-23 17:57:51 +01:00
|
|
|
error = mnt_want_write_file(filp);
|
2006-07-21 08:03:21 +02:00
|
|
|
if (error)
|
2006-03-28 21:14:04 +02:00
|
|
|
return error;
|
|
|
|
|
2008-07-02 21:12:01 +02:00
|
|
|
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
|
|
|
|
if (error)
|
|
|
|
goto out_drop_write;
|
|
|
|
|
2010-05-24 15:36:48 +02:00
|
|
|
error = -EACCES;
|
2011-03-24 00:43:26 +01:00
|
|
|
if (!inode_owner_or_capable(inode))
|
2010-05-24 15:36:48 +02:00
|
|
|
goto out;
|
|
|
|
|
|
|
|
error = 0;
|
2008-11-04 11:05:22 +01:00
|
|
|
flags = ip->i_diskflags;
|
2006-04-04 20:29:30 +02:00
|
|
|
new_flags = (flags & ~mask) | (reqflags & mask);
|
2006-03-28 21:14:04 +02:00
|
|
|
if ((new_flags ^ flags) == 0)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
error = -EPERM;
|
|
|
|
if (IS_IMMUTABLE(inode) && (new_flags & GFS2_DIF_IMMUTABLE))
|
|
|
|
goto out;
|
|
|
|
if (IS_APPEND(inode) && (new_flags & GFS2_DIF_APPENDONLY))
|
|
|
|
goto out;
|
2006-09-25 15:26:04 +02:00
|
|
|
if (((new_flags ^ flags) & GFS2_DIF_IMMUTABLE) &&
|
2006-05-12 23:07:56 +02:00
|
|
|
!capable(CAP_LINUX_IMMUTABLE))
|
2006-03-28 21:14:04 +02:00
|
|
|
goto out;
|
2006-05-12 23:07:56 +02:00
|
|
|
if (!IS_IMMUTABLE(inode)) {
|
2011-06-21 01:28:19 +02:00
|
|
|
error = gfs2_permission(inode, MAY_WRITE);
|
2006-05-12 23:07:56 +02:00
|
|
|
if (error)
|
|
|
|
goto out;
|
|
|
|
}
|
2007-10-17 09:47:38 +02:00
|
|
|
if ((flags ^ new_flags) & GFS2_DIF_JDATA) {
|
2017-09-20 15:30:04 +02:00
|
|
|
if (new_flags & GFS2_DIF_JDATA)
|
2018-01-17 00:01:33 +01:00
|
|
|
gfs2_log_flush(sdp, ip->i_gl,
|
2018-01-08 16:34:17 +01:00
|
|
|
GFS2_LOG_HEAD_FLUSH_NORMAL |
|
|
|
|
GFS2_LFC_SET_FLAGS);
|
2007-10-17 09:47:38 +02:00
|
|
|
error = filemap_fdatawrite(inode->i_mapping);
|
|
|
|
if (error)
|
|
|
|
goto out;
|
|
|
|
error = filemap_fdatawait(inode->i_mapping);
|
|
|
|
if (error)
|
|
|
|
goto out;
|
2017-09-20 15:30:04 +02:00
|
|
|
if (new_flags & GFS2_DIF_JDATA)
|
|
|
|
gfs2_ordered_del_inode(ip);
|
2007-10-17 09:47:38 +02:00
|
|
|
}
|
2006-04-04 20:29:30 +02:00
|
|
|
error = gfs2_trans_begin(sdp, RES_DINODE, 0);
|
2006-03-28 21:14:04 +02:00
|
|
|
if (error)
|
|
|
|
goto out;
|
2006-04-04 20:29:30 +02:00
|
|
|
error = gfs2_meta_inode_buffer(ip, &bh);
|
|
|
|
if (error)
|
|
|
|
goto out_trans_end;
|
2017-09-22 15:29:19 +02:00
|
|
|
inode->i_ctime = current_time(inode);
|
2012-12-14 13:36:02 +01:00
|
|
|
gfs2_trans_add_meta(ip->i_gl, bh);
|
2008-11-04 11:05:22 +01:00
|
|
|
ip->i_diskflags = new_flags;
|
2006-10-31 21:07:05 +01:00
|
|
|
gfs2_dinode_out(ip, bh->b_data);
|
2006-03-28 21:14:04 +02:00
|
|
|
brelse(bh);
|
2006-11-08 18:51:06 +01:00
|
|
|
gfs2_set_inode_flags(inode);
|
2007-10-17 09:47:38 +02:00
|
|
|
gfs2_set_aops(inode);
|
2006-04-04 20:29:30 +02:00
|
|
|
out_trans_end:
|
|
|
|
gfs2_trans_end(sdp);
|
2006-03-28 21:14:04 +02:00
|
|
|
out:
|
|
|
|
gfs2_glock_dq_uninit(&gh);
|
2008-07-02 21:12:01 +02:00
|
|
|
out_drop_write:
|
2011-12-09 14:06:57 +01:00
|
|
|
mnt_drop_write_file(filp);
|
2006-03-28 21:14:04 +02:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2006-04-07 17:17:32 +02:00
|
|
|
static int gfs2_set_flags(struct file *filp, u32 __user *ptr)
|
2006-03-28 21:14:04 +02:00
|
|
|
{
|
2013-01-23 23:07:38 +01:00
|
|
|
struct inode *inode = file_inode(filp);
|
2017-10-09 16:15:30 +02:00
|
|
|
u32 fsflags, gfsflags = 0;
|
|
|
|
u32 mask;
|
|
|
|
int i;
|
2010-05-24 15:36:48 +02:00
|
|
|
|
2006-10-02 17:24:43 +02:00
|
|
|
if (get_user(fsflags, ptr))
|
2006-03-28 21:14:04 +02:00
|
|
|
return -EFAULT;
|
2010-05-24 15:36:48 +02:00
|
|
|
|
2017-10-09 16:15:30 +02:00
|
|
|
for (i = 0; i < ARRAY_SIZE(fsflag_gfs2flag); i++) {
|
|
|
|
if (fsflags & fsflag_gfs2flag[i].fsflag) {
|
|
|
|
fsflags &= ~fsflag_gfs2flag[i].fsflag;
|
|
|
|
gfsflags |= fsflag_gfs2flag[i].gfsflag;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (fsflags || gfsflags & ~GFS2_FLAGS_USER_SET)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
mask = GFS2_FLAGS_USER_SET;
|
|
|
|
if (S_ISDIR(inode->i_mode)) {
|
|
|
|
mask &= ~GFS2_DIF_JDATA;
|
|
|
|
} else {
|
|
|
|
/* The GFS2_DIF_TOPDIR flag is only valid for directories. */
|
|
|
|
if (gfsflags & GFS2_DIF_TOPDIR)
|
|
|
|
return -EINVAL;
|
|
|
|
mask &= ~(GFS2_DIF_TOPDIR | GFS2_DIF_INHERIT_JDATA);
|
2007-07-18 12:40:06 +02:00
|
|
|
}
|
2017-10-09 16:15:30 +02:00
|
|
|
|
|
|
|
return do_gfs2_set_flags(filp, gfsflags, mask);
|
2006-03-28 21:14:04 +02:00
|
|
|
}
|
|
|
|
|
2018-10-08 21:32:35 +02:00
|
|
|
static int gfs2_getlabel(struct file *filp, char __user *label)
|
|
|
|
{
|
|
|
|
struct inode *inode = file_inode(filp);
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
|
|
|
|
|
|
|
if (copy_to_user(label, sdp->sd_sb.sb_locktable, GFS2_LOCKNAME_LEN))
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-04-07 17:17:32 +02:00
|
|
|
static long gfs2_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
|
2006-03-28 21:14:04 +02:00
|
|
|
{
|
|
|
|
switch(cmd) {
|
2006-10-02 17:24:43 +02:00
|
|
|
case FS_IOC_GETFLAGS:
|
2006-04-07 17:17:32 +02:00
|
|
|
return gfs2_get_flags(filp, (u32 __user *)arg);
|
2006-10-02 17:24:43 +02:00
|
|
|
case FS_IOC_SETFLAGS:
|
2006-04-07 17:17:32 +02:00
|
|
|
return gfs2_set_flags(filp, (u32 __user *)arg);
|
2012-02-08 13:58:32 +01:00
|
|
|
case FITRIM:
|
|
|
|
return gfs2_fitrim(filp, (void __user *)arg);
|
2018-10-08 21:32:35 +02:00
|
|
|
case FS_IOC_GETFSLABEL:
|
|
|
|
return gfs2_getlabel(filp, (char __user *)arg);
|
2006-03-28 21:14:04 +02:00
|
|
|
}
|
2018-10-08 21:32:35 +02:00
|
|
|
|
2006-03-28 21:14:04 +02:00
|
|
|
return -ENOTTY;
|
|
|
|
}
|
|
|
|
|
2012-07-26 12:30:54 +02:00
|
|
|
/**
|
|
|
|
* gfs2_size_hint - Give a hint to the size of a write request
|
2014-05-15 18:57:08 +02:00
|
|
|
* @filep: The struct file
|
2012-07-26 12:30:54 +02:00
|
|
|
* @offset: The file offset of the write
|
|
|
|
* @size: The length of the write
|
|
|
|
*
|
|
|
|
* When we are about to do a write, this function records the total
|
|
|
|
* write size in order to provide a suitable hint to the lower layers
|
|
|
|
* about how many blocks will be required.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void gfs2_size_hint(struct file *filep, loff_t offset, size_t size)
|
|
|
|
{
|
2013-01-23 23:07:38 +01:00
|
|
|
struct inode *inode = file_inode(filep);
|
2012-07-26 12:30:54 +02:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
size_t blks = (size + sdp->sd_sb.sb_bsize - 1) >> sdp->sd_sb.sb_bsize_shift;
|
|
|
|
int hint = min_t(size_t, INT_MAX, blks);
|
|
|
|
|
2018-08-30 17:01:50 +02:00
|
|
|
if (hint > atomic_read(&ip->i_sizehint))
|
|
|
|
atomic_set(&ip->i_sizehint, hint);
|
2012-07-26 12:30:54 +02:00
|
|
|
}
|
|
|
|
|
2007-10-15 16:40:33 +02:00
|
|
|
/**
|
|
|
|
* gfs2_allocate_page_backing - Use bmap to allocate blocks
|
|
|
|
* @page: The (locked) page to allocate backing for
|
|
|
|
*
|
|
|
|
* We try to allocate all the blocks required for the page in
|
|
|
|
* one go. This might fail for various reasons, so we keep
|
|
|
|
* trying until all the blocks to back this page are allocated.
|
|
|
|
* If some of the blocks are already allocated, thats ok too.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static int gfs2_allocate_page_backing(struct page *page)
|
|
|
|
{
|
|
|
|
struct inode *inode = page->mapping->host;
|
|
|
|
struct buffer_head bh;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
unsigned long size = PAGE_SIZE;
|
|
|
|
u64 lblock = page->index << (PAGE_SHIFT - inode->i_blkbits);
|
2007-10-15 16:40:33 +02:00
|
|
|
|
|
|
|
do {
|
|
|
|
bh.b_state = 0;
|
|
|
|
bh.b_size = size;
|
2007-12-10 21:13:27 +01:00
|
|
|
gfs2_block_map(inode, lblock, &bh, 1);
|
2007-10-15 16:40:33 +02:00
|
|
|
if (!buffer_mapped(&bh))
|
|
|
|
return -EIO;
|
|
|
|
size -= bh.b_size;
|
|
|
|
lblock += (bh.b_size >> inode->i_blkbits);
|
|
|
|
} while(size > 0);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* gfs2_page_mkwrite - Make a shared, mmap()ed, page writable
|
|
|
|
* @vma: The virtual memory area
|
2014-05-15 18:57:08 +02:00
|
|
|
* @vmf: The virtual memory fault containing the page to become writable
|
2007-10-15 16:40:33 +02:00
|
|
|
*
|
|
|
|
* When the page becomes writable, we need to ensure that we have
|
|
|
|
* blocks allocated on disk to back that page.
|
|
|
|
*/
|
|
|
|
|
2018-07-02 18:46:13 +02:00
|
|
|
static vm_fault_t gfs2_page_mkwrite(struct vm_fault *vmf)
|
2007-10-15 16:40:33 +02:00
|
|
|
{
|
2009-04-01 00:23:21 +02:00
|
|
|
struct page *page = vmf->page;
|
2017-02-24 23:56:41 +01:00
|
|
|
struct inode *inode = file_inode(vmf->vma->vm_file);
|
2007-10-15 16:40:33 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
2013-10-02 12:13:25 +02:00
|
|
|
struct gfs2_alloc_parms ap = { .aflags = 0, };
|
2007-10-15 16:40:33 +02:00
|
|
|
unsigned long last_index;
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
u64 pos = page->index << PAGE_SHIFT;
|
2007-10-15 16:40:33 +02:00
|
|
|
unsigned int data_blocks, ind_blocks, rblocks;
|
|
|
|
struct gfs2_holder gh;
|
2011-09-07 16:12:51 +02:00
|
|
|
loff_t size;
|
2007-10-15 16:40:33 +02:00
|
|
|
int ret;
|
|
|
|
|
2012-06-12 16:20:41 +02:00
|
|
|
sb_start_pagefault(inode->i_sb);
|
2011-09-07 16:12:51 +02:00
|
|
|
|
2015-10-26 16:40:28 +01:00
|
|
|
ret = gfs2_rsqa_alloc(ip);
|
2012-06-06 12:17:59 +02:00
|
|
|
if (ret)
|
2015-07-16 15:28:04 +02:00
|
|
|
goto out;
|
2012-06-06 12:17:59 +02:00
|
|
|
|
2017-02-24 23:56:41 +01:00
|
|
|
gfs2_size_hint(vmf->vma->vm_file, pos, PAGE_SIZE);
|
2012-07-19 14:12:40 +02:00
|
|
|
|
2008-09-18 14:53:59 +02:00
|
|
|
gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &gh);
|
|
|
|
ret = gfs2_glock_nq(&gh);
|
2007-10-15 16:40:33 +02:00
|
|
|
if (ret)
|
2013-05-28 16:04:44 +02:00
|
|
|
goto out_uninit;
|
2007-10-15 16:40:33 +02:00
|
|
|
|
2016-09-26 20:20:19 +02:00
|
|
|
/* Update file times before taking page lock */
|
2017-02-24 23:56:41 +01:00
|
|
|
file_update_time(vmf->vma->vm_file);
|
2016-09-26 20:20:19 +02:00
|
|
|
|
2009-03-19 14:15:44 +01:00
|
|
|
set_bit(GLF_DIRTY, &ip->i_gl->gl_flags);
|
|
|
|
set_bit(GIF_SW_PAGED, &ip->i_flags);
|
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
if (!gfs2_write_alloc_required(ip, pos, PAGE_SIZE)) {
|
2011-09-07 16:12:51 +02:00
|
|
|
lock_page(page);
|
|
|
|
if (!PageUptodate(page) || page->mapping != inode->i_mapping) {
|
|
|
|
ret = -EAGAIN;
|
|
|
|
unlock_page(page);
|
|
|
|
}
|
2007-10-15 16:40:33 +02:00
|
|
|
goto out_unlock;
|
2011-09-07 16:12:51 +02:00
|
|
|
}
|
|
|
|
|
2012-05-18 15:28:23 +02:00
|
|
|
ret = gfs2_rindex_update(sdp);
|
|
|
|
if (ret)
|
2008-01-10 16:18:55 +01:00
|
|
|
goto out_unlock;
|
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
gfs2_write_calc_reserv(ip, PAGE_SIZE, &data_blocks, &ind_blocks);
|
2013-10-02 12:13:25 +02:00
|
|
|
ap.target = data_blocks + ind_blocks;
|
2015-03-18 18:03:41 +01:00
|
|
|
ret = gfs2_quota_lock_check(ip, &ap);
|
|
|
|
if (ret)
|
|
|
|
goto out_unlock;
|
2013-10-02 12:13:25 +02:00
|
|
|
ret = gfs2_inplace_reserve(ip, &ap);
|
2007-10-15 16:40:33 +02:00
|
|
|
if (ret)
|
|
|
|
goto out_quota_unlock;
|
|
|
|
|
|
|
|
rblocks = RES_DINODE + ind_blocks;
|
|
|
|
if (gfs2_is_jdata(ip))
|
|
|
|
rblocks += data_blocks ? data_blocks : 1;
|
2010-09-27 23:00:04 +02:00
|
|
|
if (ind_blocks || data_blocks) {
|
2007-10-15 16:40:33 +02:00
|
|
|
rblocks += RES_STATFS + RES_QUOTA;
|
2012-07-30 15:53:19 +02:00
|
|
|
rblocks += gfs2_rg_blocks(ip, data_blocks + ind_blocks);
|
2010-09-27 23:00:04 +02:00
|
|
|
}
|
2007-10-15 16:40:33 +02:00
|
|
|
ret = gfs2_trans_begin(sdp, rblocks, 0);
|
|
|
|
if (ret)
|
|
|
|
goto out_trans_fail;
|
|
|
|
|
|
|
|
lock_page(page);
|
|
|
|
ret = -EINVAL;
|
2011-09-07 16:12:51 +02:00
|
|
|
size = i_size_read(inode);
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
last_index = (size - 1) >> PAGE_SHIFT;
|
2011-09-07 16:12:51 +02:00
|
|
|
/* 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 */
|
2008-01-17 16:12:03 +01:00
|
|
|
ret = 0;
|
2011-09-07 16:12:51 +02:00
|
|
|
if (gfs2_is_stuffed(ip))
|
2007-10-15 16:40:33 +02:00
|
|
|
ret = gfs2_unstuff_dinode(ip, page);
|
2011-09-07 16:12:51 +02:00
|
|
|
if (ret == 0)
|
|
|
|
ret = gfs2_allocate_page_backing(page);
|
2007-10-15 16:40:33 +02:00
|
|
|
|
2011-09-07 16:12:51 +02:00
|
|
|
out_trans_end:
|
|
|
|
if (ret)
|
|
|
|
unlock_page(page);
|
2007-10-15 16:40:33 +02:00
|
|
|
gfs2_trans_end(sdp);
|
|
|
|
out_trans_fail:
|
|
|
|
gfs2_inplace_release(ip);
|
|
|
|
out_quota_unlock:
|
|
|
|
gfs2_quota_unlock(ip);
|
|
|
|
out_unlock:
|
|
|
|
gfs2_glock_dq(&gh);
|
2013-05-28 16:04:44 +02:00
|
|
|
out_uninit:
|
2007-10-15 16:40:33 +02:00
|
|
|
gfs2_holder_uninit(&gh);
|
2011-09-07 16:12:51 +02:00
|
|
|
if (ret == 0) {
|
|
|
|
set_page_dirty(page);
|
mm: only enforce stable page writes if the backing device requires it
Create a helper function to check if a backing device requires stable
page writes and, if so, performs the necessary wait. Then, make it so
that all points in the memory manager that handle making pages writable
use the helper function. This should provide stable page write support
to most filesystems, while eliminating unnecessary waiting for devices
that don't require the feature.
Before this patchset, all filesystems would block, regardless of whether
or not it was necessary. ext3 would wait, but still generate occasional
checksum errors. The network filesystems were left to do their own
thing, so they'd wait too.
After this patchset, all the disk filesystems except ext3 and btrfs will
wait only if the hardware requires it. ext3 (if necessary) snapshots
pages instead of blocking, and btrfs provides its own bdi so the mm will
never wait. Network filesystems haven't been touched, so either they
provide their own stable page guarantees or they don't block at all.
The blocking behavior is back to what it was before 3.0 if you don't
have a disk requiring stable page writes.
Here's the result of using dbench to test latency on ext2:
3.8.0-rc3:
Operation Count AvgLat MaxLat
----------------------------------------
WriteX 109347 0.028 59.817
ReadX 347180 0.004 3.391
Flush 15514 29.828 287.283
Throughput 57.429 MB/sec 4 clients 4 procs max_latency=287.290 ms
3.8.0-rc3 + patches:
WriteX 105556 0.029 4.273
ReadX 335004 0.005 4.112
Flush 14982 30.540 298.634
Throughput 55.4496 MB/sec 4 clients 4 procs max_latency=298.650 ms
As you can see, the maximum write latency drops considerably with this
patch enabled. The other filesystems (ext3/ext4/xfs/btrfs) behave
similarly, but see the cover letter for those results.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Acked-by: Steven Whitehouse <swhiteho@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Eric Van Hensbergen <ericvh@gmail.com>
Cc: Ron Minnich <rminnich@sandia.gov>
Cc: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-22 01:42:51 +01:00
|
|
|
wait_for_stable_page(page);
|
2011-09-07 16:12:51 +02:00
|
|
|
}
|
2013-05-28 16:04:44 +02:00
|
|
|
out:
|
2012-06-12 16:20:41 +02:00
|
|
|
sb_end_pagefault(inode->i_sb);
|
2011-09-07 16:12:51 +02:00
|
|
|
return block_page_mkwrite_return(ret);
|
2007-10-15 16:40:33 +02:00
|
|
|
}
|
|
|
|
|
2009-09-27 20:29:37 +02:00
|
|
|
static const struct vm_operations_struct gfs2_vm_ops = {
|
2007-10-15 16:40:33 +02:00
|
|
|
.fault = filemap_fault,
|
2014-04-08 00:37:19 +02:00
|
|
|
.map_pages = filemap_map_pages,
|
2007-10-15 16:40:33 +02:00
|
|
|
.page_mkwrite = gfs2_page_mkwrite,
|
|
|
|
};
|
|
|
|
|
2006-01-16 17:50:04 +01:00
|
|
|
/**
|
|
|
|
* gfs2_mmap -
|
|
|
|
* @file: The file to map
|
|
|
|
* @vma: The VMA which described the mapping
|
|
|
|
*
|
2009-04-29 14:59:35 +02:00
|
|
|
* 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
|
2006-01-16 17:50:04 +01:00
|
|
|
*/
|
|
|
|
|
|
|
|
static int gfs2_mmap(struct file *file, struct vm_area_struct *vma)
|
|
|
|
{
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2011-02-02 15:48:10 +01:00
|
|
|
if (!(file->f_flags & O_NOATIME) &&
|
|
|
|
!IS_NOATIME(&ip->i_inode)) {
|
2009-04-29 14:59:35 +02:00
|
|
|
struct gfs2_holder i_gh;
|
|
|
|
int error;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2012-11-06 07:49:28 +01:00
|
|
|
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
|
|
|
|
&i_gh);
|
2011-02-02 15:48:10 +01:00
|
|
|
if (error)
|
|
|
|
return error;
|
2012-11-06 07:49:28 +01:00
|
|
|
/* grab lock to update inode */
|
|
|
|
gfs2_glock_dq_uninit(&i_gh);
|
|
|
|
file_accessed(file);
|
2009-04-29 14:59:35 +02:00
|
|
|
}
|
2007-10-15 16:40:33 +02:00
|
|
|
vma->vm_ops = &gfs2_vm_ops;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2009-04-29 14:59:35 +02:00
|
|
|
return 0;
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2013-06-14 12:17:15 +02:00
|
|
|
* gfs2_open_common - This is common to open and atomic_open
|
|
|
|
* @inode: The inode being opened
|
|
|
|
* @file: The file being opened
|
2006-01-16 17:50:04 +01:00
|
|
|
*
|
2013-06-14 12:17:15 +02:00
|
|
|
* 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
|
2006-01-16 17:50:04 +01:00
|
|
|
*/
|
|
|
|
|
2013-06-14 12:17:15 +02:00
|
|
|
int gfs2_open_common(struct inode *inode, struct file *file)
|
2006-01-16 17:50:04 +01:00
|
|
|
{
|
|
|
|
struct gfs2_file *fp;
|
2013-06-14 12:17:15 +02:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
if (S_ISREG(inode->i_mode)) {
|
|
|
|
ret = generic_file_open(inode, file);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2013-06-14 12:17:15 +02:00
|
|
|
fp = kzalloc(sizeof(struct gfs2_file), GFP_NOFS);
|
2006-01-16 17:50:04 +01:00
|
|
|
if (!fp)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2006-02-21 13:51:39 +01:00
|
|
|
mutex_init(&fp->f_fl_mutex);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2006-06-14 21:32:57 +02:00
|
|
|
gfs2_assert_warn(GFS2_SB(inode), !file->private_data);
|
2006-02-27 23:23:27 +01:00
|
|
|
file->private_data = fp;
|
2013-06-14 12:17:15 +02:00
|
|
|
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;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2006-11-01 18:22:46 +01:00
|
|
|
if (S_ISREG(ip->i_inode.i_mode)) {
|
2006-01-16 17:50:04 +01:00
|
|
|
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
|
|
|
|
&i_gh);
|
|
|
|
if (error)
|
2013-06-14 12:17:15 +02:00
|
|
|
return error;
|
|
|
|
need_unlock = true;
|
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2013-06-14 12:17:15 +02:00
|
|
|
error = gfs2_open_common(inode, file);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2013-06-14 12:17:15 +02:00
|
|
|
if (need_unlock)
|
2006-01-16 17:50:04 +01:00
|
|
|
gfs2_glock_dq_uninit(&i_gh);
|
|
|
|
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2012-04-11 18:56:41 +02:00
|
|
|
* gfs2_release - called to close a struct file
|
2006-01-16 17:50:04 +01:00
|
|
|
* @inode: the inode the struct file belongs to
|
|
|
|
* @file: the struct file being closed
|
|
|
|
*
|
|
|
|
* Returns: errno
|
|
|
|
*/
|
|
|
|
|
2012-04-11 18:56:41 +02:00
|
|
|
static int gfs2_release(struct inode *inode, struct file *file)
|
2006-01-16 17:50:04 +01:00
|
|
|
{
|
2012-06-06 12:17:59 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2012-07-19 14:12:40 +02:00
|
|
|
kfree(file->private_data);
|
2006-02-27 23:23:27 +01:00
|
|
|
file->private_data = NULL;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2013-05-28 16:04:44 +02:00
|
|
|
if (!(file->f_mode & FMODE_WRITE))
|
|
|
|
return 0;
|
2012-06-06 12:17:59 +02:00
|
|
|
|
2015-10-26 16:40:28 +01:00
|
|
|
gfs2_rsqa_delete(ip, &inode->i_writecount);
|
2006-01-16 17:50:04 +01:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* gfs2_fsync - sync the dirty data for a file (across the cluster)
|
2011-07-17 02:44:56 +02:00
|
|
|
* @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
|
2011-04-14 10:54:02 +02:00
|
|
|
* @datasync: set if we can ignore timestamp changes
|
2006-01-16 17:50:04 +01:00
|
|
|
*
|
2011-07-27 11:58:48 +02:00
|
|
|
* 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.
|
2006-12-07 15:13:14 +01:00
|
|
|
*
|
2006-01-16 17:50:04 +01:00
|
|
|
* Returns: errno
|
|
|
|
*/
|
|
|
|
|
2011-07-17 02:44:56 +02:00
|
|
|
static int gfs2_fsync(struct file *file, loff_t start, loff_t end,
|
|
|
|
int datasync)
|
2006-01-16 17:50:04 +01:00
|
|
|
{
|
2011-07-27 11:58:48 +02:00
|
|
|
struct address_space *mapping = file->f_mapping;
|
|
|
|
struct inode *inode = mapping->host;
|
2015-02-02 06:37:00 +01:00
|
|
|
int sync_state = inode->i_state & I_DIRTY_ALL;
|
2011-04-14 10:54:02 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2011-11-08 15:04:20 +01:00
|
|
|
int ret = 0, ret1 = 0;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2011-07-27 11:58:48 +02:00
|
|
|
if (mapping->nrpages) {
|
|
|
|
ret1 = filemap_fdatawrite_range(mapping, start, end);
|
|
|
|
if (ret1 == -EIO)
|
|
|
|
return ret1;
|
|
|
|
}
|
2011-07-17 02:44:56 +02:00
|
|
|
|
2013-09-03 23:59:42 +02:00
|
|
|
if (!gfs2_is_jdata(ip))
|
|
|
|
sync_state &= ~I_DIRTY_PAGES;
|
2011-04-14 10:54:02 +02:00
|
|
|
if (datasync)
|
2015-02-02 06:37:00 +01:00
|
|
|
sync_state &= ~(I_DIRTY_SYNC | I_DIRTY_TIME);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2011-04-14 10:54:02 +02:00
|
|
|
if (sync_state) {
|
|
|
|
ret = sync_inode_metadata(inode, 1);
|
2011-09-07 11:33:25 +02:00
|
|
|
if (ret)
|
2011-04-14 10:54:02 +02:00
|
|
|
return ret;
|
2011-08-05 11:12:47 +02:00
|
|
|
if (gfs2_is_jdata(ip))
|
2017-07-07 21:20:53 +02:00
|
|
|
ret = file_write_and_wait(file);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
2011-09-07 11:33:25 +02:00
|
|
|
gfs2_ail_flush(ip->i_gl, 1);
|
2006-11-30 16:14:32 +01:00
|
|
|
}
|
|
|
|
|
2011-07-27 11:58:48 +02:00
|
|
|
if (mapping->nrpages)
|
2017-07-07 21:20:53 +02:00
|
|
|
ret = file_fdatawait_range(file, start, end);
|
2011-07-27 11:58:48 +02:00
|
|
|
|
|
|
|
return ret ? ret : ret1;
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
2018-06-19 16:08:02 +02:00
|
|
|
static ssize_t gfs2_file_direct_read(struct kiocb *iocb, struct iov_iter *to)
|
|
|
|
{
|
|
|
|
struct file *file = iocb->ki_filp;
|
|
|
|
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
|
|
|
|
size_t count = iov_iter_count(to);
|
|
|
|
struct gfs2_holder gh;
|
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
if (!count)
|
|
|
|
return 0; /* skip atime */
|
|
|
|
|
|
|
|
gfs2_holder_init(ip->i_gl, LM_ST_DEFERRED, 0, &gh);
|
|
|
|
ret = gfs2_glock_nq(&gh);
|
|
|
|
if (ret)
|
|
|
|
goto out_uninit;
|
|
|
|
|
|
|
|
ret = iomap_dio_rw(iocb, to, &gfs2_iomap_ops, NULL);
|
|
|
|
|
|
|
|
gfs2_glock_dq(&gh);
|
|
|
|
out_uninit:
|
|
|
|
gfs2_holder_uninit(&gh);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t gfs2_file_direct_write(struct kiocb *iocb, struct iov_iter *from)
|
|
|
|
{
|
|
|
|
struct file *file = iocb->ki_filp;
|
|
|
|
struct inode *inode = file->f_mapping->host;
|
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
size_t len = iov_iter_count(from);
|
|
|
|
loff_t offset = iocb->ki_pos;
|
|
|
|
struct gfs2_holder gh;
|
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Deferred lock, even if its a write, since we do no allocation on
|
|
|
|
* this path. All we need to change is the atime, and this lock mode
|
|
|
|
* ensures that other nodes have flushed their buffered read caches
|
|
|
|
* (i.e. their page cache entries for this inode). We do not,
|
|
|
|
* unfortunately, have the option of only flushing a range like the
|
|
|
|
* VFS does.
|
|
|
|
*/
|
|
|
|
gfs2_holder_init(ip->i_gl, LM_ST_DEFERRED, 0, &gh);
|
|
|
|
ret = gfs2_glock_nq(&gh);
|
|
|
|
if (ret)
|
|
|
|
goto out_uninit;
|
|
|
|
|
|
|
|
/* Silently fall back to buffered I/O when writing beyond EOF */
|
|
|
|
if (offset + len > i_size_read(&ip->i_inode))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
ret = iomap_dio_rw(iocb, from, &gfs2_iomap_ops, NULL);
|
|
|
|
|
|
|
|
out:
|
|
|
|
gfs2_glock_dq(&gh);
|
|
|
|
out_uninit:
|
|
|
|
gfs2_holder_uninit(&gh);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t gfs2_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
|
|
|
|
{
|
|
|
|
ssize_t ret;
|
|
|
|
|
|
|
|
if (iocb->ki_flags & IOCB_DIRECT) {
|
|
|
|
ret = gfs2_file_direct_read(iocb, to);
|
|
|
|
if (likely(ret != -ENOTBLK))
|
|
|
|
return ret;
|
|
|
|
iocb->ki_flags &= ~IOCB_DIRECT;
|
|
|
|
}
|
|
|
|
return generic_file_read_iter(iocb, to);
|
|
|
|
}
|
|
|
|
|
2009-12-08 11:25:33 +01:00
|
|
|
/**
|
2014-04-03 20:11:01 +02:00
|
|
|
* gfs2_file_write_iter - Perform a write to a file
|
2009-12-08 11:25:33 +01:00
|
|
|
* @iocb: The io context
|
2018-06-24 16:04:04 +02:00
|
|
|
* @from: The data to write
|
2009-12-08 11:25:33 +01:00
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2014-04-03 20:11:01 +02:00
|
|
|
static ssize_t gfs2_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
|
2009-12-08 11:25:33 +01:00
|
|
|
{
|
|
|
|
struct file *file = iocb->ki_filp;
|
2018-06-24 16:04:04 +02:00
|
|
|
struct inode *inode = file_inode(file);
|
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2018-06-19 16:08:02 +02:00
|
|
|
ssize_t written = 0, ret;
|
2012-06-06 12:17:59 +02:00
|
|
|
|
2015-10-26 16:40:28 +01:00
|
|
|
ret = gfs2_rsqa_alloc(ip);
|
2012-06-06 12:17:59 +02:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
2009-12-08 11:25:33 +01:00
|
|
|
|
2014-04-03 20:11:01 +02:00
|
|
|
gfs2_size_hint(file, iocb->ki_pos, iov_iter_count(from));
|
2012-07-26 12:30:54 +02:00
|
|
|
|
2015-04-09 19:52:01 +02:00
|
|
|
if (iocb->ki_flags & IOCB_APPEND) {
|
2009-12-08 11:25:33 +01:00
|
|
|
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);
|
|
|
|
}
|
|
|
|
|
2018-06-24 16:04:04 +02:00
|
|
|
inode_lock(inode);
|
|
|
|
ret = generic_write_checks(iocb, from);
|
|
|
|
if (ret <= 0)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* We can write back this queue in page reclaim */
|
|
|
|
current->backing_dev_info = inode_to_bdi(inode);
|
|
|
|
|
|
|
|
ret = file_remove_privs(file);
|
|
|
|
if (ret)
|
|
|
|
goto out2;
|
|
|
|
|
|
|
|
ret = file_update_time(file);
|
|
|
|
if (ret)
|
|
|
|
goto out2;
|
|
|
|
|
2018-06-19 16:08:02 +02:00
|
|
|
if (iocb->ki_flags & IOCB_DIRECT) {
|
|
|
|
struct address_space *mapping = file->f_mapping;
|
|
|
|
loff_t pos, endbyte;
|
|
|
|
ssize_t buffered;
|
|
|
|
|
|
|
|
written = gfs2_file_direct_write(iocb, from);
|
|
|
|
if (written < 0 || !iov_iter_count(from))
|
|
|
|
goto out2;
|
|
|
|
|
|
|
|
ret = iomap_file_buffered_write(iocb, from, &gfs2_iomap_ops);
|
|
|
|
if (unlikely(ret < 0))
|
|
|
|
goto out2;
|
|
|
|
buffered = ret;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to ensure that the page cache pages are written to
|
|
|
|
* disk and invalidated to preserve the expected O_DIRECT
|
|
|
|
* semantics.
|
|
|
|
*/
|
|
|
|
pos = iocb->ki_pos;
|
|
|
|
endbyte = pos + buffered - 1;
|
|
|
|
ret = filemap_write_and_wait_range(mapping, pos, endbyte);
|
|
|
|
if (!ret) {
|
|
|
|
iocb->ki_pos += buffered;
|
|
|
|
written += buffered;
|
|
|
|
invalidate_mapping_pages(mapping,
|
|
|
|
pos >> PAGE_SHIFT,
|
|
|
|
endbyte >> PAGE_SHIFT);
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* We don't know how much we wrote, so just return
|
|
|
|
* the number of bytes which were direct-written
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
ret = iomap_file_buffered_write(iocb, from, &gfs2_iomap_ops);
|
|
|
|
if (likely(ret > 0))
|
|
|
|
iocb->ki_pos += ret;
|
|
|
|
}
|
2018-06-24 16:04:04 +02:00
|
|
|
|
|
|
|
out2:
|
|
|
|
current->backing_dev_info = NULL;
|
|
|
|
out:
|
|
|
|
inode_unlock(inode);
|
|
|
|
if (likely(ret > 0)) {
|
|
|
|
/* Handle various SYNC-type writes */
|
|
|
|
ret = generic_write_sync(iocb, ret);
|
|
|
|
}
|
2018-06-19 16:08:02 +02:00
|
|
|
return written ? written : ret;
|
2009-12-08 11:25:33 +01:00
|
|
|
}
|
|
|
|
|
2011-01-14 13:07:43 +01:00
|
|
|
static int fallocate_chunk(struct inode *inode, loff_t offset, loff_t len,
|
|
|
|
int mode)
|
|
|
|
{
|
2018-03-29 15:50:32 +02:00
|
|
|
struct super_block *sb = inode->i_sb;
|
2011-01-14 13:07:43 +01:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2018-03-29 15:50:32 +02:00
|
|
|
loff_t end = offset + len;
|
2011-01-14 13:07:43 +01:00
|
|
|
struct buffer_head *dibh;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
error = gfs2_meta_inode_buffer(ip, &dibh);
|
|
|
|
if (unlikely(error))
|
2011-09-13 01:15:24 +02:00
|
|
|
return error;
|
2011-01-14 13:07:43 +01:00
|
|
|
|
2012-12-14 13:36:02 +01:00
|
|
|
gfs2_trans_add_meta(ip->i_gl, dibh);
|
2011-01-14 13:07:43 +01:00
|
|
|
|
|
|
|
if (gfs2_is_stuffed(ip)) {
|
|
|
|
error = gfs2_unstuff_dinode(ip, NULL);
|
|
|
|
if (unlikely(error))
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2018-03-29 15:50:32 +02:00
|
|
|
while (offset < end) {
|
2018-07-07 00:05:41 +02:00
|
|
|
struct iomap iomap = { };
|
|
|
|
|
2018-06-04 14:56:51 +02:00
|
|
|
error = gfs2_iomap_get_alloc(inode, offset, end - offset,
|
|
|
|
&iomap);
|
2018-03-29 15:50:32 +02:00
|
|
|
if (error)
|
2011-09-13 01:15:24 +02:00
|
|
|
goto out;
|
2018-03-29 15:50:32 +02:00
|
|
|
offset = iomap.offset + iomap.length;
|
2018-06-24 11:43:49 +02:00
|
|
|
if (!(iomap.flags & IOMAP_F_NEW))
|
2011-09-13 01:15:24 +02:00
|
|
|
continue;
|
2018-03-29 15:50:32 +02:00
|
|
|
error = sb_issue_zeroout(sb, iomap.addr >> inode->i_blkbits,
|
|
|
|
iomap.length >> inode->i_blkbits,
|
|
|
|
GFP_NOFS);
|
|
|
|
if (error) {
|
|
|
|
fs_err(GFS2_SB(inode), "Failed to zero data buffers\n");
|
2011-01-14 13:07:43 +01:00
|
|
|
goto out;
|
2011-09-13 01:15:24 +02:00
|
|
|
}
|
2011-01-14 13:07:43 +01:00
|
|
|
}
|
|
|
|
out:
|
2011-09-13 01:15:24 +02:00
|
|
|
brelse(dibh);
|
2011-01-14 13:07:43 +01:00
|
|
|
return error;
|
|
|
|
}
|
2015-03-18 18:05:15 +01:00
|
|
|
/**
|
|
|
|
* 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)
|
2011-01-14 13:07:43 +01:00
|
|
|
{
|
2015-03-18 18:05:15 +01:00
|
|
|
loff_t max = *len;
|
2011-01-14 13:07:43 +01:00
|
|
|
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;
|
|
|
|
}
|
2015-03-18 18:05:15 +01:00
|
|
|
|
2011-01-14 13:07:43 +01:00
|
|
|
*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);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-11-12 18:24:03 +01:00
|
|
|
static long __gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
|
2011-01-14 13:07:43 +01:00
|
|
|
{
|
2013-01-23 23:07:38 +01:00
|
|
|
struct inode *inode = file_inode(file);
|
2011-01-14 13:07:43 +01:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
2013-10-02 12:13:25 +02:00
|
|
|
struct gfs2_alloc_parms ap = { .aflags = 0, };
|
2011-01-14 13:07:43 +01:00
|
|
|
unsigned int data_blocks = 0, ind_blocks = 0, rblocks;
|
2018-02-20 16:03:24 +01:00
|
|
|
loff_t bytes, max_bytes, max_blks;
|
2011-01-14 13:07:43 +01:00
|
|
|
int error;
|
2011-11-21 11:01:25 +01:00
|
|
|
const loff_t pos = offset;
|
|
|
|
const loff_t count = len;
|
2011-04-26 08:13:24 +02:00
|
|
|
loff_t bsize_mask = ~((loff_t)sdp->sd_sb.sb_bsize - 1);
|
2011-01-14 13:07:43 +01:00
|
|
|
loff_t next = (offset + len - 1) >> sdp->sd_sb.sb_bsize_shift;
|
2011-09-13 01:15:24 +02:00
|
|
|
loff_t max_chunk_size = UINT_MAX & bsize_mask;
|
2014-02-06 16:43:50 +01:00
|
|
|
|
2011-01-14 13:07:43 +01:00
|
|
|
next = (next + 1) << sdp->sd_sb.sb_bsize_shift;
|
|
|
|
|
2011-04-26 08:13:24 +02:00
|
|
|
offset &= bsize_mask;
|
2011-01-14 13:07:43 +01:00
|
|
|
|
|
|
|
len = next - offset;
|
|
|
|
bytes = sdp->sd_max_rg_data * sdp->sd_sb.sb_bsize / 2;
|
|
|
|
if (!bytes)
|
|
|
|
bytes = UINT_MAX;
|
2011-04-26 08:13:24 +02:00
|
|
|
bytes &= bsize_mask;
|
|
|
|
if (bytes == 0)
|
|
|
|
bytes = sdp->sd_sb.sb_bsize;
|
2011-01-14 13:07:43 +01:00
|
|
|
|
2012-07-26 12:30:54 +02:00
|
|
|
gfs2_size_hint(file, offset, len);
|
2012-07-19 14:12:40 +02:00
|
|
|
|
2015-03-18 18:05:15 +01:00
|
|
|
gfs2_write_calc_reserv(ip, PAGE_SIZE, &data_blocks, &ind_blocks);
|
|
|
|
ap.min_target = data_blocks + ind_blocks;
|
|
|
|
|
2011-01-14 13:07:43 +01:00
|
|
|
while (len > 0) {
|
|
|
|
if (len < bytes)
|
|
|
|
bytes = len;
|
2012-03-08 20:16:32 +01:00
|
|
|
if (!gfs2_write_alloc_required(ip, offset, bytes)) {
|
|
|
|
len -= bytes;
|
|
|
|
offset += bytes;
|
|
|
|
continue;
|
|
|
|
}
|
2015-03-18 18:05:15 +01:00
|
|
|
|
|
|
|
/* 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 */
|
2011-01-14 13:07:43 +01:00
|
|
|
gfs2_write_calc_reserv(ip, bytes, &data_blocks, &ind_blocks);
|
2013-10-02 12:13:25 +02:00
|
|
|
ap.target = data_blocks + ind_blocks;
|
2015-03-18 18:03:41 +01:00
|
|
|
|
|
|
|
error = gfs2_quota_lock_check(ip, &ap);
|
2011-01-14 13:07:43 +01:00
|
|
|
if (error)
|
2014-11-12 18:24:03 +01:00
|
|
|
return error;
|
2015-03-18 18:05:15 +01:00
|
|
|
/* ap.allowed tells us how many blocks quota will allow
|
|
|
|
* us to write. Check if this reduces max_blks */
|
2018-02-20 16:03:24 +01:00
|
|
|
max_blks = UINT_MAX;
|
|
|
|
if (ap.allowed)
|
2015-03-18 18:05:15 +01:00
|
|
|
max_blks = ap.allowed;
|
2011-01-14 13:07:43 +01:00
|
|
|
|
2013-10-02 12:13:25 +02:00
|
|
|
error = gfs2_inplace_reserve(ip, &ap);
|
2015-03-18 18:05:15 +01:00
|
|
|
if (error)
|
2011-01-14 13:07:43 +01:00
|
|
|
goto out_qunlock;
|
2015-03-18 18:05:15 +01:00
|
|
|
|
|
|
|
/* 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);
|
2011-01-14 13:07:43 +01:00
|
|
|
|
|
|
|
rblocks = RES_DINODE + ind_blocks + RES_STATFS + RES_QUOTA +
|
2012-07-30 15:53:19 +02:00
|
|
|
RES_RG_HDR + gfs2_rg_blocks(ip, data_blocks + ind_blocks);
|
2011-01-14 13:07:43 +01:00
|
|
|
if (gfs2_is_jdata(ip))
|
|
|
|
rblocks += data_blocks ? data_blocks : 1;
|
|
|
|
|
|
|
|
error = gfs2_trans_begin(sdp, rblocks,
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 14:29:47 +02:00
|
|
|
PAGE_SIZE/sdp->sd_sb.sb_bsize);
|
2011-01-14 13:07:43 +01:00
|
|
|
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);
|
|
|
|
}
|
2011-11-21 11:01:25 +01:00
|
|
|
|
2014-11-12 18:24:04 +01:00
|
|
|
if (!(mode & FALLOC_FL_KEEP_SIZE) && (pos + count) > inode->i_size) {
|
|
|
|
i_size_write(inode, pos + count);
|
2014-11-12 18:24:05 +01:00
|
|
|
file_update_time(file);
|
2015-09-22 14:41:21 +02:00
|
|
|
mark_inode_dirty(inode);
|
2014-11-12 18:24:04 +01:00
|
|
|
}
|
|
|
|
|
2016-04-07 17:52:00 +02:00
|
|
|
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;
|
2011-01-14 13:07:43 +01:00
|
|
|
|
|
|
|
out_trans_fail:
|
|
|
|
gfs2_inplace_release(ip);
|
|
|
|
out_qunlock:
|
|
|
|
gfs2_quota_unlock(ip);
|
2014-11-12 18:24:03 +01:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
static long gfs2_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
|
|
|
|
{
|
|
|
|
struct inode *inode = file_inode(file);
|
2017-04-05 17:45:26 +02:00
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(inode);
|
2014-11-12 18:24:03 +01:00
|
|
|
struct gfs2_inode *ip = GFS2_I(inode);
|
|
|
|
struct gfs2_holder gh;
|
|
|
|
int ret;
|
|
|
|
|
2017-12-14 17:11:03 +01:00
|
|
|
if (mode & ~(FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE))
|
2017-04-05 17:45:26 +02:00
|
|
|
return -EOPNOTSUPP;
|
|
|
|
/* fallocate is needed by gfs2_grow to reserve space in the rindex */
|
|
|
|
if (gfs2_is_jdata(ip) && inode != sdp->sd_rindex)
|
2014-11-12 18:24:03 +01:00
|
|
|
return -EOPNOTSUPP;
|
|
|
|
|
2016-01-22 21:40:57 +01:00
|
|
|
inode_lock(inode);
|
2014-11-12 18:24:03 +01:00
|
|
|
|
|
|
|
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;
|
|
|
|
|
2017-12-14 17:11:03 +01:00
|
|
|
if (mode & FALLOC_FL_PUNCH_HOLE) {
|
|
|
|
ret = __gfs2_punch_hole(file, offset, len);
|
|
|
|
} else {
|
|
|
|
ret = gfs2_rsqa_alloc(ip);
|
|
|
|
if (ret)
|
|
|
|
goto out_putw;
|
2014-11-12 18:24:03 +01:00
|
|
|
|
2017-12-14 17:11:03 +01:00
|
|
|
ret = __gfs2_fallocate(file, mode, offset, len);
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
gfs2_rs_deltree(&ip->i_res);
|
|
|
|
}
|
2015-07-16 15:28:04 +02:00
|
|
|
|
2014-11-12 18:24:03 +01:00
|
|
|
out_putw:
|
|
|
|
put_write_access(inode);
|
2011-01-14 13:07:43 +01:00
|
|
|
out_unlock:
|
2014-02-06 16:43:50 +01:00
|
|
|
gfs2_glock_dq(&gh);
|
2011-01-14 13:07:43 +01:00
|
|
|
out_uninit:
|
2014-02-06 16:43:50 +01:00
|
|
|
gfs2_holder_uninit(&gh);
|
2016-01-22 21:40:57 +01:00
|
|
|
inode_unlock(inode);
|
2014-11-12 18:24:03 +01:00
|
|
|
return ret;
|
2011-01-14 13:07:43 +01:00
|
|
|
}
|
|
|
|
|
2015-02-24 14:22:28 +01:00
|
|
|
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);
|
|
|
|
|
2015-10-26 16:40:28 +01:00
|
|
|
error = gfs2_rsqa_alloc(ip);
|
2015-02-24 14:22:28 +01:00
|
|
|
if (error)
|
|
|
|
return (ssize_t)error;
|
|
|
|
|
|
|
|
gfs2_size_hint(out, *ppos, len);
|
|
|
|
|
|
|
|
return iter_file_splice_write(pipe, out, ppos, len, flags);
|
|
|
|
}
|
|
|
|
|
2009-01-12 11:43:39 +01:00
|
|
|
#ifdef CONFIG_GFS2_FS_LOCKING_DLM
|
|
|
|
|
2006-01-16 17:50:04 +01:00
|
|
|
/**
|
|
|
|
* 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)
|
|
|
|
{
|
2006-06-14 21:32:57 +02:00
|
|
|
struct gfs2_inode *ip = GFS2_I(file->f_mapping->host);
|
|
|
|
struct gfs2_sbd *sdp = GFS2_SB(file->f_mapping->host);
|
2009-01-12 11:43:39 +01:00
|
|
|
struct lm_lockstruct *ls = &sdp->sd_lockstruct;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
if (!(fl->fl_flags & FL_POSIX))
|
|
|
|
return -ENOLCK;
|
2010-03-11 18:24:45 +01:00
|
|
|
if (__mandatory_lock(&ip->i_inode) && fl->fl_type != F_UNLCK)
|
2006-01-16 17:50:04 +01:00
|
|
|
return -ENOLCK;
|
|
|
|
|
2006-11-14 22:37:25 +01:00
|
|
|
if (cmd == F_CANCELLK) {
|
|
|
|
/* Hack: */
|
|
|
|
cmd = F_SETLK;
|
|
|
|
fl->fl_type = F_UNLCK;
|
|
|
|
}
|
2013-03-14 16:49:59 +01:00
|
|
|
if (unlikely(test_bit(SDF_SHUTDOWN, &sdp->sd_flags))) {
|
|
|
|
if (fl->fl_type == F_UNLCK)
|
2015-10-22 19:38:14 +02:00
|
|
|
locks_lock_file_wait(file, fl);
|
2009-01-12 11:43:39 +01:00
|
|
|
return -EIO;
|
2013-03-14 16:49:59 +01:00
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
if (IS_GETLK(cmd))
|
2009-01-12 11:43:39 +01:00
|
|
|
return dlm_posix_get(ls->ls_dlm, ip->i_no_addr, file, fl);
|
2006-01-16 17:50:04 +01:00
|
|
|
else if (fl->fl_type == F_UNLCK)
|
2009-01-12 11:43:39 +01:00
|
|
|
return dlm_posix_unlock(ls->ls_dlm, ip->i_no_addr, file, fl);
|
2006-01-16 17:50:04 +01:00
|
|
|
else
|
2009-01-12 11:43:39 +01:00
|
|
|
return dlm_posix_lock(ls->ls_dlm, ip->i_no_addr, file, cmd, fl);
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static int do_flock(struct file *file, int cmd, struct file_lock *fl)
|
|
|
|
{
|
2006-02-27 23:23:27 +01:00
|
|
|
struct gfs2_file *fp = file->private_data;
|
2006-01-16 17:50:04 +01:00
|
|
|
struct gfs2_holder *fl_gh = &fp->f_fl_gh;
|
2013-01-23 23:07:38 +01:00
|
|
|
struct gfs2_inode *ip = GFS2_I(file_inode(file));
|
2006-01-16 17:50:04 +01:00
|
|
|
struct gfs2_glock *gl;
|
|
|
|
unsigned int state;
|
2015-07-24 16:45:43 +02:00
|
|
|
u16 flags;
|
2006-01-16 17:50:04 +01:00
|
|
|
int error = 0;
|
2014-08-20 18:44:45 +02:00
|
|
|
int sleeptime;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
state = (fl->fl_type == F_WRLCK) ? LM_ST_EXCLUSIVE : LM_ST_SHARED;
|
2014-08-20 18:44:45 +02:00
|
|
|
flags = (IS_SETLKW(cmd) ? 0 : LM_FLAG_TRY_1CB) | GL_EXACT;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2006-02-21 13:51:39 +01:00
|
|
|
mutex_lock(&fp->f_fl_mutex);
|
2006-01-16 17:50:04 +01:00
|
|
|
|
2017-07-17 20:39:15 +02:00
|
|
|
if (gfs2_holder_initialized(fl_gh)) {
|
2006-01-16 17:50:04 +01:00
|
|
|
if (fl_gh->gh_state == state)
|
|
|
|
goto out;
|
2015-10-22 19:38:14 +02:00
|
|
|
locks_lock_file_wait(file,
|
2015-12-22 15:06:08 +01:00
|
|
|
&(struct file_lock) {
|
|
|
|
.fl_type = F_UNLCK,
|
|
|
|
.fl_flags = FL_FLOCK
|
|
|
|
});
|
2014-06-26 16:46:25 +02:00
|
|
|
gfs2_glock_dq(fl_gh);
|
2007-09-14 06:35:27 +02:00
|
|
|
gfs2_holder_reinit(state, flags, fl_gh);
|
2006-01-16 17:50:04 +01:00
|
|
|
} else {
|
2008-05-21 18:03:22 +02:00
|
|
|
error = gfs2_glock_get(GFS2_SB(&ip->i_inode), ip->i_no_addr,
|
|
|
|
&gfs2_flock_glops, CREATE, &gl);
|
2006-01-16 17:50:04 +01:00
|
|
|
if (error)
|
|
|
|
goto out;
|
2007-09-14 06:35:27 +02:00
|
|
|
gfs2_holder_init(gl, state, flags, fl_gh);
|
|
|
|
gfs2_glock_put(gl);
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
2014-08-20 18:44:45 +02:00
|
|
|
for (sleeptime = 1; sleeptime <= 4; sleeptime <<= 1) {
|
|
|
|
error = gfs2_glock_nq(fl_gh);
|
|
|
|
if (error != GLR_TRYFAILED)
|
|
|
|
break;
|
|
|
|
fl_gh->gh_flags = LM_FLAG_TRY | GL_EXACT;
|
|
|
|
fl_gh->gh_error = 0;
|
|
|
|
msleep(sleeptime);
|
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
if (error) {
|
|
|
|
gfs2_holder_uninit(fl_gh);
|
|
|
|
if (error == GLR_TRYFAILED)
|
|
|
|
error = -EAGAIN;
|
|
|
|
} else {
|
2015-10-22 19:38:14 +02:00
|
|
|
error = locks_lock_file_wait(file, fl);
|
2006-06-14 21:32:57 +02:00
|
|
|
gfs2_assert_warn(GFS2_SB(&ip->i_inode), !error);
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
2006-07-31 21:42:17 +02:00
|
|
|
out:
|
2006-02-21 13:51:39 +01:00
|
|
|
mutex_unlock(&fp->f_fl_mutex);
|
2006-01-16 17:50:04 +01:00
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void do_unflock(struct file *file, struct file_lock *fl)
|
|
|
|
{
|
2006-02-27 23:23:27 +01:00
|
|
|
struct gfs2_file *fp = file->private_data;
|
2006-01-16 17:50:04 +01:00
|
|
|
struct gfs2_holder *fl_gh = &fp->f_fl_gh;
|
|
|
|
|
2006-02-21 13:51:39 +01:00
|
|
|
mutex_lock(&fp->f_fl_mutex);
|
2015-10-22 19:38:14 +02:00
|
|
|
locks_lock_file_wait(file, fl);
|
2016-06-17 14:31:27 +02:00
|
|
|
if (gfs2_holder_initialized(fl_gh)) {
|
2014-08-20 18:44:45 +02:00
|
|
|
gfs2_glock_dq(fl_gh);
|
2011-03-09 12:14:32 +01:00
|
|
|
gfs2_holder_uninit(fl_gh);
|
|
|
|
}
|
2006-02-21 13:51:39 +01:00
|
|
|
mutex_unlock(&fp->f_fl_mutex);
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* gfs2_flock - acquire/release a flock 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_flock(struct file *file, int cmd, struct file_lock *fl)
|
|
|
|
{
|
|
|
|
if (!(fl->fl_flags & FL_FLOCK))
|
|
|
|
return -ENOLCK;
|
2009-06-01 19:30:03 +02:00
|
|
|
if (fl->fl_type & LOCK_MAND)
|
|
|
|
return -EOPNOTSUPP;
|
2006-01-16 17:50:04 +01:00
|
|
|
|
|
|
|
if (fl->fl_type == F_UNLCK) {
|
|
|
|
do_unflock(file, fl);
|
|
|
|
return 0;
|
2006-10-02 16:28:05 +02:00
|
|
|
} else {
|
2006-01-16 17:50:04 +01:00
|
|
|
return do_flock(file, cmd, fl);
|
2006-10-02 16:28:05 +02:00
|
|
|
}
|
2006-01-16 17:50:04 +01:00
|
|
|
}
|
|
|
|
|
2009-04-07 19:42:17 +02:00
|
|
|
const struct file_operations gfs2_file_fops = {
|
2006-09-04 21:32:10 +02:00
|
|
|
.llseek = gfs2_llseek,
|
2018-06-19 16:08:02 +02:00
|
|
|
.read_iter = gfs2_file_read_iter,
|
2014-04-03 20:11:01 +02:00
|
|
|
.write_iter = gfs2_file_write_iter,
|
2006-09-04 21:32:10 +02:00
|
|
|
.unlocked_ioctl = gfs2_ioctl,
|
|
|
|
.mmap = gfs2_mmap,
|
|
|
|
.open = gfs2_open,
|
2012-04-11 18:56:41 +02:00
|
|
|
.release = gfs2_release,
|
2006-09-04 21:32:10 +02:00
|
|
|
.fsync = gfs2_fsync,
|
|
|
|
.lock = gfs2_lock,
|
|
|
|
.flock = gfs2_flock,
|
2016-09-23 05:35:42 +02:00
|
|
|
.splice_read = generic_file_splice_read,
|
2015-02-19 15:02:16 +01:00
|
|
|
.splice_write = gfs2_file_splice_write,
|
2014-08-27 12:49:41 +02:00
|
|
|
.setlease = simple_nosetlease,
|
2011-01-14 13:07:43 +01:00
|
|
|
.fallocate = gfs2_fallocate,
|
2006-01-16 17:50:04 +01:00
|
|
|
};
|
|
|
|
|
2009-04-07 19:42:17 +02:00
|
|
|
const struct file_operations gfs2_dir_fops = {
|
2016-05-12 23:00:20 +02:00
|
|
|
.iterate_shared = gfs2_readdir,
|
2006-09-04 21:32:10 +02:00
|
|
|
.unlocked_ioctl = gfs2_ioctl,
|
|
|
|
.open = gfs2_open,
|
2012-04-11 18:56:41 +02:00
|
|
|
.release = gfs2_release,
|
2006-09-04 21:32:10 +02:00
|
|
|
.fsync = gfs2_fsync,
|
|
|
|
.lock = gfs2_lock,
|
|
|
|
.flock = gfs2_flock,
|
llseek: automatically add .llseek fop
All file_operations should get a .llseek operation so we can make
nonseekable_open the default for future file operations without a
.llseek pointer.
The three cases that we can automatically detect are no_llseek, seq_lseek
and default_llseek. For cases where we can we can automatically prove that
the file offset is always ignored, we use noop_llseek, which maintains
the current behavior of not returning an error from a seek.
New drivers should normally not use noop_llseek but instead use no_llseek
and call nonseekable_open at open time. Existing drivers can be converted
to do the same when the maintainer knows for certain that no user code
relies on calling seek on the device file.
The generated code is often incorrectly indented and right now contains
comments that clarify for each added line why a specific variant was
chosen. In the version that gets submitted upstream, the comments will
be gone and I will manually fix the indentation, because there does not
seem to be a way to do that using coccinelle.
Some amount of new code is currently sitting in linux-next that should get
the same modifications, which I will do at the end of the merge window.
Many thanks to Julia Lawall for helping me learn to write a semantic
patch that does all this.
===== begin semantic patch =====
// This adds an llseek= method to all file operations,
// as a preparation for making no_llseek the default.
//
// The rules are
// - use no_llseek explicitly if we do nonseekable_open
// - use seq_lseek for sequential files
// - use default_llseek if we know we access f_pos
// - use noop_llseek if we know we don't access f_pos,
// but we still want to allow users to call lseek
//
@ open1 exists @
identifier nested_open;
@@
nested_open(...)
{
<+...
nonseekable_open(...)
...+>
}
@ open exists@
identifier open_f;
identifier i, f;
identifier open1.nested_open;
@@
int open_f(struct inode *i, struct file *f)
{
<+...
(
nonseekable_open(...)
|
nested_open(...)
)
...+>
}
@ read disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ read_no_fpos disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
... when != off
}
@ write @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ write_no_fpos @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
... when != off
}
@ fops0 @
identifier fops;
@@
struct file_operations fops = {
...
};
@ has_llseek depends on fops0 @
identifier fops0.fops;
identifier llseek_f;
@@
struct file_operations fops = {
...
.llseek = llseek_f,
...
};
@ has_read depends on fops0 @
identifier fops0.fops;
identifier read_f;
@@
struct file_operations fops = {
...
.read = read_f,
...
};
@ has_write depends on fops0 @
identifier fops0.fops;
identifier write_f;
@@
struct file_operations fops = {
...
.write = write_f,
...
};
@ has_open depends on fops0 @
identifier fops0.fops;
identifier open_f;
@@
struct file_operations fops = {
...
.open = open_f,
...
};
// use no_llseek if we call nonseekable_open
////////////////////////////////////////////
@ nonseekable1 depends on !has_llseek && has_open @
identifier fops0.fops;
identifier nso ~= "nonseekable_open";
@@
struct file_operations fops = {
... .open = nso, ...
+.llseek = no_llseek, /* nonseekable */
};
@ nonseekable2 depends on !has_llseek @
identifier fops0.fops;
identifier open.open_f;
@@
struct file_operations fops = {
... .open = open_f, ...
+.llseek = no_llseek, /* open uses nonseekable */
};
// use seq_lseek for sequential files
/////////////////////////////////////
@ seq depends on !has_llseek @
identifier fops0.fops;
identifier sr ~= "seq_read";
@@
struct file_operations fops = {
... .read = sr, ...
+.llseek = seq_lseek, /* we have seq_read */
};
// use default_llseek if there is a readdir
///////////////////////////////////////////
@ fops1 depends on !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier readdir_e;
@@
// any other fop is used that changes pos
struct file_operations fops = {
... .readdir = readdir_e, ...
+.llseek = default_llseek, /* readdir is present */
};
// use default_llseek if at least one of read/write touches f_pos
/////////////////////////////////////////////////////////////////
@ fops2 depends on !fops1 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read.read_f;
@@
// read fops use offset
struct file_operations fops = {
... .read = read_f, ...
+.llseek = default_llseek, /* read accesses f_pos */
};
@ fops3 depends on !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write.write_f;
@@
// write fops use offset
struct file_operations fops = {
... .write = write_f, ...
+ .llseek = default_llseek, /* write accesses f_pos */
};
// Use noop_llseek if neither read nor write accesses f_pos
///////////////////////////////////////////////////////////
@ fops4 depends on !fops1 && !fops2 && !fops3 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
identifier write_no_fpos.write_f;
@@
// write fops use offset
struct file_operations fops = {
...
.write = write_f,
.read = read_f,
...
+.llseek = noop_llseek, /* read and write both use no f_pos */
};
@ depends on has_write && !has_read && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write_no_fpos.write_f;
@@
struct file_operations fops = {
... .write = write_f, ...
+.llseek = noop_llseek, /* write uses no f_pos */
};
@ depends on has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
@@
struct file_operations fops = {
... .read = read_f, ...
+.llseek = noop_llseek, /* read uses no f_pos */
};
@ depends on !has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
@@
struct file_operations fops = {
...
+.llseek = noop_llseek, /* no read or write fn */
};
===== End semantic patch =====
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Julia Lawall <julia@diku.dk>
Cc: Christoph Hellwig <hch@infradead.org>
2010-08-15 18:52:59 +02:00
|
|
|
.llseek = default_llseek,
|
2006-01-16 17:50:04 +01:00
|
|
|
};
|
|
|
|
|
2009-01-12 11:43:39 +01:00
|
|
|
#endif /* CONFIG_GFS2_FS_LOCKING_DLM */
|
|
|
|
|
2009-04-07 19:42:17 +02:00
|
|
|
const struct file_operations gfs2_file_fops_nolock = {
|
2007-11-29 23:56:51 +01:00
|
|
|
.llseek = gfs2_llseek,
|
2018-06-19 16:08:02 +02:00
|
|
|
.read_iter = gfs2_file_read_iter,
|
2014-04-03 20:11:01 +02:00
|
|
|
.write_iter = gfs2_file_write_iter,
|
2007-11-29 23:56:51 +01:00
|
|
|
.unlocked_ioctl = gfs2_ioctl,
|
|
|
|
.mmap = gfs2_mmap,
|
|
|
|
.open = gfs2_open,
|
2012-04-11 18:56:41 +02:00
|
|
|
.release = gfs2_release,
|
2007-11-29 23:56:51 +01:00
|
|
|
.fsync = gfs2_fsync,
|
2016-09-23 05:35:42 +02:00
|
|
|
.splice_read = generic_file_splice_read,
|
2015-02-19 15:02:16 +01:00
|
|
|
.splice_write = gfs2_file_splice_write,
|
2009-01-12 11:43:39 +01:00
|
|
|
.setlease = generic_setlease,
|
2011-01-14 13:07:43 +01:00
|
|
|
.fallocate = gfs2_fallocate,
|
2007-11-29 23:56:51 +01:00
|
|
|
};
|
|
|
|
|
2009-04-07 19:42:17 +02:00
|
|
|
const struct file_operations gfs2_dir_fops_nolock = {
|
2016-05-12 23:00:20 +02:00
|
|
|
.iterate_shared = gfs2_readdir,
|
2007-11-29 23:56:51 +01:00
|
|
|
.unlocked_ioctl = gfs2_ioctl,
|
|
|
|
.open = gfs2_open,
|
2012-04-11 18:56:41 +02:00
|
|
|
.release = gfs2_release,
|
2007-11-29 23:56:51 +01:00
|
|
|
.fsync = gfs2_fsync,
|
llseek: automatically add .llseek fop
All file_operations should get a .llseek operation so we can make
nonseekable_open the default for future file operations without a
.llseek pointer.
The three cases that we can automatically detect are no_llseek, seq_lseek
and default_llseek. For cases where we can we can automatically prove that
the file offset is always ignored, we use noop_llseek, which maintains
the current behavior of not returning an error from a seek.
New drivers should normally not use noop_llseek but instead use no_llseek
and call nonseekable_open at open time. Existing drivers can be converted
to do the same when the maintainer knows for certain that no user code
relies on calling seek on the device file.
The generated code is often incorrectly indented and right now contains
comments that clarify for each added line why a specific variant was
chosen. In the version that gets submitted upstream, the comments will
be gone and I will manually fix the indentation, because there does not
seem to be a way to do that using coccinelle.
Some amount of new code is currently sitting in linux-next that should get
the same modifications, which I will do at the end of the merge window.
Many thanks to Julia Lawall for helping me learn to write a semantic
patch that does all this.
===== begin semantic patch =====
// This adds an llseek= method to all file operations,
// as a preparation for making no_llseek the default.
//
// The rules are
// - use no_llseek explicitly if we do nonseekable_open
// - use seq_lseek for sequential files
// - use default_llseek if we know we access f_pos
// - use noop_llseek if we know we don't access f_pos,
// but we still want to allow users to call lseek
//
@ open1 exists @
identifier nested_open;
@@
nested_open(...)
{
<+...
nonseekable_open(...)
...+>
}
@ open exists@
identifier open_f;
identifier i, f;
identifier open1.nested_open;
@@
int open_f(struct inode *i, struct file *f)
{
<+...
(
nonseekable_open(...)
|
nested_open(...)
)
...+>
}
@ read disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ read_no_fpos disable optional_qualifier exists @
identifier read_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t read_f(struct file *f, char *p, size_t s, loff_t *off)
{
... when != off
}
@ write @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
expression E;
identifier func;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
<+...
(
*off = E
|
*off += E
|
func(..., off, ...)
|
E = *off
)
...+>
}
@ write_no_fpos @
identifier write_f;
identifier f, p, s, off;
type ssize_t, size_t, loff_t;
@@
ssize_t write_f(struct file *f, const char *p, size_t s, loff_t *off)
{
... when != off
}
@ fops0 @
identifier fops;
@@
struct file_operations fops = {
...
};
@ has_llseek depends on fops0 @
identifier fops0.fops;
identifier llseek_f;
@@
struct file_operations fops = {
...
.llseek = llseek_f,
...
};
@ has_read depends on fops0 @
identifier fops0.fops;
identifier read_f;
@@
struct file_operations fops = {
...
.read = read_f,
...
};
@ has_write depends on fops0 @
identifier fops0.fops;
identifier write_f;
@@
struct file_operations fops = {
...
.write = write_f,
...
};
@ has_open depends on fops0 @
identifier fops0.fops;
identifier open_f;
@@
struct file_operations fops = {
...
.open = open_f,
...
};
// use no_llseek if we call nonseekable_open
////////////////////////////////////////////
@ nonseekable1 depends on !has_llseek && has_open @
identifier fops0.fops;
identifier nso ~= "nonseekable_open";
@@
struct file_operations fops = {
... .open = nso, ...
+.llseek = no_llseek, /* nonseekable */
};
@ nonseekable2 depends on !has_llseek @
identifier fops0.fops;
identifier open.open_f;
@@
struct file_operations fops = {
... .open = open_f, ...
+.llseek = no_llseek, /* open uses nonseekable */
};
// use seq_lseek for sequential files
/////////////////////////////////////
@ seq depends on !has_llseek @
identifier fops0.fops;
identifier sr ~= "seq_read";
@@
struct file_operations fops = {
... .read = sr, ...
+.llseek = seq_lseek, /* we have seq_read */
};
// use default_llseek if there is a readdir
///////////////////////////////////////////
@ fops1 depends on !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier readdir_e;
@@
// any other fop is used that changes pos
struct file_operations fops = {
... .readdir = readdir_e, ...
+.llseek = default_llseek, /* readdir is present */
};
// use default_llseek if at least one of read/write touches f_pos
/////////////////////////////////////////////////////////////////
@ fops2 depends on !fops1 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read.read_f;
@@
// read fops use offset
struct file_operations fops = {
... .read = read_f, ...
+.llseek = default_llseek, /* read accesses f_pos */
};
@ fops3 depends on !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write.write_f;
@@
// write fops use offset
struct file_operations fops = {
... .write = write_f, ...
+ .llseek = default_llseek, /* write accesses f_pos */
};
// Use noop_llseek if neither read nor write accesses f_pos
///////////////////////////////////////////////////////////
@ fops4 depends on !fops1 && !fops2 && !fops3 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
identifier write_no_fpos.write_f;
@@
// write fops use offset
struct file_operations fops = {
...
.write = write_f,
.read = read_f,
...
+.llseek = noop_llseek, /* read and write both use no f_pos */
};
@ depends on has_write && !has_read && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier write_no_fpos.write_f;
@@
struct file_operations fops = {
... .write = write_f, ...
+.llseek = noop_llseek, /* write uses no f_pos */
};
@ depends on has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
identifier read_no_fpos.read_f;
@@
struct file_operations fops = {
... .read = read_f, ...
+.llseek = noop_llseek, /* read uses no f_pos */
};
@ depends on !has_read && !has_write && !fops1 && !fops2 && !has_llseek && !nonseekable1 && !nonseekable2 && !seq @
identifier fops0.fops;
@@
struct file_operations fops = {
...
+.llseek = noop_llseek, /* no read or write fn */
};
===== End semantic patch =====
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Julia Lawall <julia@diku.dk>
Cc: Christoph Hellwig <hch@infradead.org>
2010-08-15 18:52:59 +02:00
|
|
|
.llseek = default_llseek,
|
2007-11-29 23:56:51 +01:00
|
|
|
};
|
|
|
|
|