WSL2-Linux-Kernel/fs/xfs/xfs_super.c

2352 строки
59 KiB
C
Исходник Обычный вид История

// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2000-2006 Silicon Graphics, Inc.
* All Rights Reserved.
*/
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-15 02:14:59 +03:00
#include "xfs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_sb.h"
#include "xfs_mount.h"
#include "xfs_inode.h"
#include "xfs_btree.h"
#include "xfs_bmap.h"
#include "xfs_alloc.h"
#include "xfs_fsops.h"
#include "xfs_trans.h"
#include "xfs_buf_item.h"
#include "xfs_log.h"
#include "xfs_log_priv.h"
#include "xfs_dir2.h"
#include "xfs_extfree_item.h"
#include "xfs_mru_cache.h"
#include "xfs_inode_item.h"
#include "xfs_icache.h"
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-15 02:14:59 +03:00
#include "xfs_trace.h"
#include "xfs_icreate_item.h"
#include "xfs_filestream.h"
#include "xfs_quota.h"
#include "xfs_sysfs.h"
#include "xfs_ondisk.h"
#include "xfs_rmap_item.h"
#include "xfs_refcount_item.h"
#include "xfs_bmap_item.h"
#include "xfs_reflink.h"
#include "xfs_pwork.h"
#include "xfs_ag.h"
#include <linux/magic.h>
#include <linux/fs_context.h>
#include <linux/fs_parser.h>
static const struct super_operations xfs_super_operations;
static struct kset *xfs_kset; /* top-level xfs sysfs dir */
#ifdef DEBUG
static struct xfs_kobj xfs_dbg_kobj; /* global debug sysfs attrs */
#endif
#ifdef CONFIG_HOTPLUG_CPU
static LIST_HEAD(xfs_mount_list);
static DEFINE_SPINLOCK(xfs_mount_list_lock);
static inline void xfs_mount_list_add(struct xfs_mount *mp)
{
spin_lock(&xfs_mount_list_lock);
list_add(&mp->m_mount_list, &xfs_mount_list);
spin_unlock(&xfs_mount_list_lock);
}
static inline void xfs_mount_list_del(struct xfs_mount *mp)
{
spin_lock(&xfs_mount_list_lock);
list_del(&mp->m_mount_list);
spin_unlock(&xfs_mount_list_lock);
}
#else /* !CONFIG_HOTPLUG_CPU */
static inline void xfs_mount_list_add(struct xfs_mount *mp) {}
static inline void xfs_mount_list_del(struct xfs_mount *mp) {}
#endif
enum xfs_dax_mode {
XFS_DAX_INODE = 0,
XFS_DAX_ALWAYS = 1,
XFS_DAX_NEVER = 2,
};
static void
xfs_mount_set_dax_mode(
struct xfs_mount *mp,
enum xfs_dax_mode mode)
{
switch (mode) {
case XFS_DAX_INODE:
mp->m_features &= ~(XFS_FEAT_DAX_ALWAYS | XFS_FEAT_DAX_NEVER);
break;
case XFS_DAX_ALWAYS:
mp->m_features |= XFS_FEAT_DAX_ALWAYS;
mp->m_features &= ~XFS_FEAT_DAX_NEVER;
break;
case XFS_DAX_NEVER:
mp->m_features |= XFS_FEAT_DAX_NEVER;
mp->m_features &= ~XFS_FEAT_DAX_ALWAYS;
break;
}
}
static const struct constant_table dax_param_enums[] = {
{"inode", XFS_DAX_INODE },
{"always", XFS_DAX_ALWAYS },
{"never", XFS_DAX_NEVER },
{}
};
/*
* Table driven mount option parser.
*/
enum {
Opt_logbufs, Opt_logbsize, Opt_logdev, Opt_rtdev,
Opt_wsync, Opt_noalign, Opt_swalloc, Opt_sunit, Opt_swidth, Opt_nouuid,
Opt_grpid, Opt_nogrpid, Opt_bsdgroups, Opt_sysvgroups,
Opt_allocsize, Opt_norecovery, Opt_inode64, Opt_inode32, Opt_ikeep,
Opt_noikeep, Opt_largeio, Opt_nolargeio, Opt_attr2, Opt_noattr2,
Opt_filestreams, Opt_quota, Opt_noquota, Opt_usrquota, Opt_grpquota,
Opt_prjquota, Opt_uquota, Opt_gquota, Opt_pquota,
Opt_uqnoenforce, Opt_gqnoenforce, Opt_pqnoenforce, Opt_qnoenforce,
Opt_discard, Opt_nodiscard, Opt_dax, Opt_dax_enum,
};
static const struct fs_parameter_spec xfs_fs_parameters[] = {
fsparam_u32("logbufs", Opt_logbufs),
fsparam_string("logbsize", Opt_logbsize),
fsparam_string("logdev", Opt_logdev),
fsparam_string("rtdev", Opt_rtdev),
fsparam_flag("wsync", Opt_wsync),
fsparam_flag("noalign", Opt_noalign),
fsparam_flag("swalloc", Opt_swalloc),
fsparam_u32("sunit", Opt_sunit),
fsparam_u32("swidth", Opt_swidth),
fsparam_flag("nouuid", Opt_nouuid),
fsparam_flag("grpid", Opt_grpid),
fsparam_flag("nogrpid", Opt_nogrpid),
fsparam_flag("bsdgroups", Opt_bsdgroups),
fsparam_flag("sysvgroups", Opt_sysvgroups),
fsparam_string("allocsize", Opt_allocsize),
fsparam_flag("norecovery", Opt_norecovery),
fsparam_flag("inode64", Opt_inode64),
fsparam_flag("inode32", Opt_inode32),
fsparam_flag("ikeep", Opt_ikeep),
fsparam_flag("noikeep", Opt_noikeep),
fsparam_flag("largeio", Opt_largeio),
fsparam_flag("nolargeio", Opt_nolargeio),
fsparam_flag("attr2", Opt_attr2),
fsparam_flag("noattr2", Opt_noattr2),
fsparam_flag("filestreams", Opt_filestreams),
fsparam_flag("quota", Opt_quota),
fsparam_flag("noquota", Opt_noquota),
fsparam_flag("usrquota", Opt_usrquota),
fsparam_flag("grpquota", Opt_grpquota),
fsparam_flag("prjquota", Opt_prjquota),
fsparam_flag("uquota", Opt_uquota),
fsparam_flag("gquota", Opt_gquota),
fsparam_flag("pquota", Opt_pquota),
fsparam_flag("uqnoenforce", Opt_uqnoenforce),
fsparam_flag("gqnoenforce", Opt_gqnoenforce),
fsparam_flag("pqnoenforce", Opt_pqnoenforce),
fsparam_flag("qnoenforce", Opt_qnoenforce),
fsparam_flag("discard", Opt_discard),
fsparam_flag("nodiscard", Opt_nodiscard),
fsparam_flag("dax", Opt_dax),
fsparam_enum("dax", Opt_dax_enum, dax_param_enums),
{}
};
struct proc_xfs_info {
uint64_t flag;
char *str;
};
static int
xfs_fs_show_options(
struct seq_file *m,
struct dentry *root)
{
static struct proc_xfs_info xfs_info_set[] = {
/* the few simple ones we can get from the mount struct */
{ XFS_FEAT_IKEEP, ",ikeep" },
{ XFS_FEAT_WSYNC, ",wsync" },
{ XFS_FEAT_NOALIGN, ",noalign" },
{ XFS_FEAT_SWALLOC, ",swalloc" },
{ XFS_FEAT_NOUUID, ",nouuid" },
{ XFS_FEAT_NORECOVERY, ",norecovery" },
{ XFS_FEAT_ATTR2, ",attr2" },
{ XFS_FEAT_FILESTREAMS, ",filestreams" },
{ XFS_FEAT_GRPID, ",grpid" },
{ XFS_FEAT_DISCARD, ",discard" },
{ XFS_FEAT_LARGE_IOSIZE, ",largeio" },
{ XFS_FEAT_DAX_ALWAYS, ",dax=always" },
{ XFS_FEAT_DAX_NEVER, ",dax=never" },
{ 0, NULL }
};
struct xfs_mount *mp = XFS_M(root->d_sb);
struct proc_xfs_info *xfs_infop;
for (xfs_infop = xfs_info_set; xfs_infop->flag; xfs_infop++) {
if (mp->m_features & xfs_infop->flag)
seq_puts(m, xfs_infop->str);
}
seq_printf(m, ",inode%d", xfs_has_small_inums(mp) ? 32 : 64);
if (xfs_has_allocsize(mp))
seq_printf(m, ",allocsize=%dk",
(1 << mp->m_allocsize_log) >> 10);
if (mp->m_logbufs > 0)
seq_printf(m, ",logbufs=%d", mp->m_logbufs);
if (mp->m_logbsize > 0)
seq_printf(m, ",logbsize=%dk", mp->m_logbsize >> 10);
if (mp->m_logname)
seq_show_option(m, "logdev", mp->m_logname);
if (mp->m_rtname)
seq_show_option(m, "rtdev", mp->m_rtname);
if (mp->m_dalign > 0)
seq_printf(m, ",sunit=%d",
(int)XFS_FSB_TO_BB(mp, mp->m_dalign));
if (mp->m_swidth > 0)
seq_printf(m, ",swidth=%d",
(int)XFS_FSB_TO_BB(mp, mp->m_swidth));
if (mp->m_qflags & XFS_UQUOTA_ENFD)
seq_puts(m, ",usrquota");
else if (mp->m_qflags & XFS_UQUOTA_ACCT)
seq_puts(m, ",uqnoenforce");
if (mp->m_qflags & XFS_PQUOTA_ENFD)
seq_puts(m, ",prjquota");
else if (mp->m_qflags & XFS_PQUOTA_ACCT)
seq_puts(m, ",pqnoenforce");
if (mp->m_qflags & XFS_GQUOTA_ENFD)
seq_puts(m, ",grpquota");
else if (mp->m_qflags & XFS_GQUOTA_ACCT)
seq_puts(m, ",gqnoenforce");
if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
seq_puts(m, ",noquota");
return 0;
}
/*
* Set parameters for inode allocation heuristics, taking into account
* filesystem size and inode32/inode64 mount options; i.e. specifically
* whether or not XFS_FEAT_SMALL_INUMS is set.
*
* Inode allocation patterns are altered only if inode32 is requested
* (XFS_FEAT_SMALL_INUMS), and the filesystem is sufficiently large.
* If altered, XFS_OPSTATE_INODE32 is set as well.
*
* An agcount independent of that in the mount structure is provided
* because in the growfs case, mp->m_sb.sb_agcount is not yet updated
* to the potentially higher ag count.
*
* Returns the maximum AG index which may contain inodes.
*/
xfs_agnumber_t
xfs_set_inode_alloc(
struct xfs_mount *mp,
xfs_agnumber_t agcount)
{
xfs_agnumber_t index;
xfs_agnumber_t maxagi = 0;
xfs_sb_t *sbp = &mp->m_sb;
xfs_agnumber_t max_metadata;
xfs_agino_t agino;
xfs_ino_t ino;
/*
* Calculate how much should be reserved for inodes to meet
* the max inode percentage. Used only for inode32.
*/
if (M_IGEO(mp)->maxicount) {
uint64_t icount;
icount = sbp->sb_dblocks * sbp->sb_imax_pct;
do_div(icount, 100);
icount += sbp->sb_agblocks - 1;
do_div(icount, sbp->sb_agblocks);
max_metadata = icount;
} else {
max_metadata = agcount;
}
/* Get the last possible inode in the filesystem */
agino = XFS_AGB_TO_AGINO(mp, sbp->sb_agblocks - 1);
ino = XFS_AGINO_TO_INO(mp, agcount - 1, agino);
/*
* If user asked for no more than 32-bit inodes, and the fs is
* sufficiently large, set XFS_OPSTATE_INODE32 if we must alter
* the allocator to accommodate the request.
*/
if (xfs_has_small_inums(mp) && ino > XFS_MAXINUMBER_32)
set_bit(XFS_OPSTATE_INODE32, &mp->m_opstate);
else
clear_bit(XFS_OPSTATE_INODE32, &mp->m_opstate);
for (index = 0; index < agcount; index++) {
struct xfs_perag *pag;
ino = XFS_AGINO_TO_INO(mp, index, agino);
pag = xfs_perag_get(mp, index);
if (xfs_is_inode32(mp)) {
if (ino > XFS_MAXINUMBER_32) {
pag->pagi_inodeok = 0;
pag->pagf_metadata = 0;
} else {
pag->pagi_inodeok = 1;
maxagi++;
if (index < max_metadata)
pag->pagf_metadata = 1;
else
pag->pagf_metadata = 0;
}
} else {
pag->pagi_inodeok = 1;
pag->pagf_metadata = 0;
}
xfs_perag_put(pag);
}
return xfs_is_inode32(mp) ? maxagi : agcount;
}
static bool
xfs_buftarg_is_dax(
struct super_block *sb,
struct xfs_buftarg *bt)
{
return dax_supported(bt->bt_daxdev, bt->bt_bdev, sb->s_blocksize, 0,
bdev_nr_sectors(bt->bt_bdev));
}
STATIC int
xfs_blkdev_get(
xfs_mount_t *mp,
const char *name,
struct block_device **bdevp)
{
int error = 0;
*bdevp = blkdev_get_by_path(name, FMODE_READ|FMODE_WRITE|FMODE_EXCL,
mp);
if (IS_ERR(*bdevp)) {
error = PTR_ERR(*bdevp);
xfs_warn(mp, "Invalid device [%s], error=%d", name, error);
}
return error;
}
STATIC void
xfs_blkdev_put(
struct block_device *bdev)
{
if (bdev)
block: make blkdev_get/put() handle exclusive access Over time, block layer has accumulated a set of APIs dealing with bdev open, close, claim and release. * blkdev_get/put() are the primary open and close functions. * bd_claim/release() deal with exclusive open. * open/close_bdev_exclusive() are combination of open and claim and the other way around, respectively. * bd_link/unlink_disk_holder() to create and remove holder/slave symlinks. * open_by_devnum() wraps bdget() + blkdev_get(). The interface is a bit confusing and the decoupling of open and claim makes it impossible to properly guarantee exclusive access as in-kernel open + claim sequence can disturb the existing exclusive open even before the block layer knows the current open if for another exclusive access. Reorganize the interface such that, * blkdev_get() is extended to include exclusive access management. @holder argument is added and, if is @FMODE_EXCL specified, it will gain exclusive access atomically w.r.t. other exclusive accesses. * blkdev_put() is similarly extended. It now takes @mode argument and if @FMODE_EXCL is set, it releases an exclusive access. Also, when the last exclusive claim is released, the holder/slave symlinks are removed automatically. * bd_claim/release() and close_bdev_exclusive() are no longer necessary and either made static or removed. * bd_link_disk_holder() remains the same but bd_unlink_disk_holder() is no longer necessary and removed. * open_bdev_exclusive() becomes a simple wrapper around lookup_bdev() and blkdev_get(). It also has an unexpected extra bdev_read_only() test which probably should be moved into blkdev_get(). * open_by_devnum() is modified to take @holder argument and pass it to blkdev_get(). Most of bdev open/close operations are unified into blkdev_get/put() and most exclusive accesses are tested atomically at the open time (as it should). This cleans up code and removes some, both valid and invalid, but unnecessary all the same, corner cases. open_bdev_exclusive() and open_by_devnum() can use further cleanup - rename to blkdev_get_by_path() and blkdev_get_by_devt() and drop special features. Well, let's leave them for another day. Most conversions are straight-forward. drbd conversion is a bit more involved as there was some reordering, but the logic should stay the same. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Neil Brown <neilb@suse.de> Acked-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp> Acked-by: Mike Snitzer <snitzer@redhat.com> Acked-by: Philipp Reisner <philipp.reisner@linbit.com> Cc: Peter Osterlund <petero2@telia.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Jan Kara <jack@suse.cz> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <joel.becker@oracle.com> Cc: Alex Elder <aelder@sgi.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: dm-devel@redhat.com Cc: drbd-dev@lists.linbit.com Cc: Leo Chen <leochen@broadcom.com> Cc: Scott Branden <sbranden@broadcom.com> Cc: Chris Mason <chris.mason@oracle.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@linux.vnet.ibm.com> Cc: Joern Engel <joern@logfs.org> Cc: reiserfs-devel@vger.kernel.org Cc: Alexander Viro <viro@zeniv.linux.org.uk>
2010-11-13 13:55:17 +03:00
blkdev_put(bdev, FMODE_READ|FMODE_WRITE|FMODE_EXCL);
}
STATIC void
xfs_close_devices(
struct xfs_mount *mp)
{
struct dax_device *dax_ddev = mp->m_ddev_targp->bt_daxdev;
if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
struct block_device *logdev = mp->m_logdev_targp->bt_bdev;
struct dax_device *dax_logdev = mp->m_logdev_targp->bt_daxdev;
xfs_free_buftarg(mp->m_logdev_targp);
xfs_blkdev_put(logdev);
fs_put_dax(dax_logdev);
}
if (mp->m_rtdev_targp) {
struct block_device *rtdev = mp->m_rtdev_targp->bt_bdev;
struct dax_device *dax_rtdev = mp->m_rtdev_targp->bt_daxdev;
xfs_free_buftarg(mp->m_rtdev_targp);
xfs_blkdev_put(rtdev);
fs_put_dax(dax_rtdev);
}
xfs_free_buftarg(mp->m_ddev_targp);
fs_put_dax(dax_ddev);
}
/*
* The file system configurations are:
* (1) device (partition) with data and internal log
* (2) logical volume with data and log subvolumes.
* (3) logical volume with data, log, and realtime subvolumes.
*
* We only have to handle opening the log and realtime volumes here if
* they are present. The data subvolume has already been opened by
* get_sb_bdev() and is stored in sb->s_bdev.
*/
STATIC int
xfs_open_devices(
struct xfs_mount *mp)
{
struct block_device *ddev = mp->m_super->s_bdev;
struct dax_device *dax_ddev = fs_dax_get_by_bdev(ddev);
struct dax_device *dax_logdev = NULL, *dax_rtdev = NULL;
struct block_device *logdev = NULL, *rtdev = NULL;
int error;
/*
* Open real time and log devices - order is important.
*/
if (mp->m_logname) {
error = xfs_blkdev_get(mp, mp->m_logname, &logdev);
if (error)
goto out;
dax_logdev = fs_dax_get_by_bdev(logdev);
}
if (mp->m_rtname) {
error = xfs_blkdev_get(mp, mp->m_rtname, &rtdev);
if (error)
goto out_close_logdev;
if (rtdev == ddev || rtdev == logdev) {
xfs_warn(mp,
"Cannot mount filesystem with identical rtdev and ddev/logdev.");
error = -EINVAL;
goto out_close_rtdev;
}
dax_rtdev = fs_dax_get_by_bdev(rtdev);
}
/*
* Setup xfs_mount buffer target pointers
*/
error = -ENOMEM;
mp->m_ddev_targp = xfs_alloc_buftarg(mp, ddev, dax_ddev);
if (!mp->m_ddev_targp)
goto out_close_rtdev;
if (rtdev) {
mp->m_rtdev_targp = xfs_alloc_buftarg(mp, rtdev, dax_rtdev);
if (!mp->m_rtdev_targp)
goto out_free_ddev_targ;
}
if (logdev && logdev != ddev) {
mp->m_logdev_targp = xfs_alloc_buftarg(mp, logdev, dax_logdev);
if (!mp->m_logdev_targp)
goto out_free_rtdev_targ;
} else {
mp->m_logdev_targp = mp->m_ddev_targp;
}
return 0;
out_free_rtdev_targ:
if (mp->m_rtdev_targp)
xfs_free_buftarg(mp->m_rtdev_targp);
out_free_ddev_targ:
xfs_free_buftarg(mp->m_ddev_targp);
out_close_rtdev:
xfs_blkdev_put(rtdev);
fs_put_dax(dax_rtdev);
out_close_logdev:
if (logdev && logdev != ddev) {
xfs_blkdev_put(logdev);
fs_put_dax(dax_logdev);
}
out:
fs_put_dax(dax_ddev);
return error;
}
/*
* Setup xfs_mount buffer target pointers based on superblock
*/
STATIC int
xfs_setup_devices(
struct xfs_mount *mp)
{
int error;
error = xfs_setsize_buftarg(mp->m_ddev_targp, mp->m_sb.sb_sectsize);
if (error)
return error;
if (mp->m_logdev_targp && mp->m_logdev_targp != mp->m_ddev_targp) {
unsigned int log_sector_size = BBSIZE;
if (xfs_has_sector(mp))
log_sector_size = mp->m_sb.sb_logsectsize;
error = xfs_setsize_buftarg(mp->m_logdev_targp,
log_sector_size);
if (error)
return error;
}
if (mp->m_rtdev_targp) {
error = xfs_setsize_buftarg(mp->m_rtdev_targp,
mp->m_sb.sb_sectsize);
if (error)
return error;
}
return 0;
}
STATIC int
xfs_init_mount_workqueues(
struct xfs_mount *mp)
{
xfs: replace global xfslogd wq with per-mount wq The xfslogd workqueue is a global, single-job workqueue for buffer ioend processing. This means we allow for a single work item at a time for all possible XFS mounts on a system. fsstress testing in loopback XFS over XFS configurations has reproduced xfslogd deadlocks due to the single threaded nature of the queue and dependencies introduced between the separate XFS instances by online discard (-o discard). Discard over a loopback device converts the discard request to a hole punch (fallocate) on the underlying file. Online discard requests are issued synchronously and from xfslogd context in XFS, hence the xfslogd workqueue is blocked in the upper fs waiting on a hole punch request to be servied in the lower fs. If the lower fs issues I/O that depends on xfslogd to complete, both filesystems end up hung indefinitely. This is reproduced reliabily by generic/013 on XFS->loop->XFS test devices with the '-o discard' mount option. Further, docker implementations appear to use this kind of configuration for container instance filesystems by default (container fs->dm-> loop->base fs) and therefore are subject to this deadlock when running on XFS. Replace the global xfslogd workqueue with a per-mount variant. This guarantees each mount access to a single worker and prevents deadlocks due to inter-fs dependencies introduced by discard. Since the queue is only responsible for buffer iodone processing at this point in time, rename xfslogd to xfs-buf. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-11-28 05:59:58 +03:00
mp->m_buf_workqueue = alloc_workqueue("xfs-buf/%s",
XFS_WQFLAGS(WQ_FREEZABLE | WQ_MEM_RECLAIM),
1, mp->m_super->s_id);
xfs: replace global xfslogd wq with per-mount wq The xfslogd workqueue is a global, single-job workqueue for buffer ioend processing. This means we allow for a single work item at a time for all possible XFS mounts on a system. fsstress testing in loopback XFS over XFS configurations has reproduced xfslogd deadlocks due to the single threaded nature of the queue and dependencies introduced between the separate XFS instances by online discard (-o discard). Discard over a loopback device converts the discard request to a hole punch (fallocate) on the underlying file. Online discard requests are issued synchronously and from xfslogd context in XFS, hence the xfslogd workqueue is blocked in the upper fs waiting on a hole punch request to be servied in the lower fs. If the lower fs issues I/O that depends on xfslogd to complete, both filesystems end up hung indefinitely. This is reproduced reliabily by generic/013 on XFS->loop->XFS test devices with the '-o discard' mount option. Further, docker implementations appear to use this kind of configuration for container instance filesystems by default (container fs->dm-> loop->base fs) and therefore are subject to this deadlock when running on XFS. Replace the global xfslogd workqueue with a per-mount variant. This guarantees each mount access to a single worker and prevents deadlocks due to inter-fs dependencies introduced by discard. Since the queue is only responsible for buffer iodone processing at this point in time, rename xfslogd to xfs-buf. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-11-28 05:59:58 +03:00
if (!mp->m_buf_workqueue)
goto out;
mp->m_unwritten_workqueue = alloc_workqueue("xfs-conv/%s",
XFS_WQFLAGS(WQ_FREEZABLE | WQ_MEM_RECLAIM),
0, mp->m_super->s_id);
if (!mp->m_unwritten_workqueue)
goto out_destroy_buf;
mp->m_reclaim_workqueue = alloc_workqueue("xfs-reclaim/%s",
XFS_WQFLAGS(WQ_FREEZABLE | WQ_MEM_RECLAIM),
0, mp->m_super->s_id);
if (!mp->m_reclaim_workqueue)
goto out_destroy_unwritten;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
mp->m_blockgc_wq = alloc_workqueue("xfs-blockgc/%s",
XFS_WQFLAGS(WQ_UNBOUND | WQ_FREEZABLE | WQ_MEM_RECLAIM),
0, mp->m_super->s_id);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
if (!mp->m_blockgc_wq)
goto out_destroy_reclaim;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
mp->m_inodegc_wq = alloc_workqueue("xfs-inodegc/%s",
XFS_WQFLAGS(WQ_FREEZABLE | WQ_MEM_RECLAIM),
1, mp->m_super->s_id);
if (!mp->m_inodegc_wq)
goto out_destroy_blockgc;
mp->m_sync_workqueue = alloc_workqueue("xfs-sync/%s",
XFS_WQFLAGS(WQ_FREEZABLE), 0, mp->m_super->s_id);
if (!mp->m_sync_workqueue)
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
goto out_destroy_inodegc;
return 0;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
out_destroy_inodegc:
destroy_workqueue(mp->m_inodegc_wq);
out_destroy_blockgc:
destroy_workqueue(mp->m_blockgc_wq);
out_destroy_reclaim:
destroy_workqueue(mp->m_reclaim_workqueue);
out_destroy_unwritten:
destroy_workqueue(mp->m_unwritten_workqueue);
xfs: replace global xfslogd wq with per-mount wq The xfslogd workqueue is a global, single-job workqueue for buffer ioend processing. This means we allow for a single work item at a time for all possible XFS mounts on a system. fsstress testing in loopback XFS over XFS configurations has reproduced xfslogd deadlocks due to the single threaded nature of the queue and dependencies introduced between the separate XFS instances by online discard (-o discard). Discard over a loopback device converts the discard request to a hole punch (fallocate) on the underlying file. Online discard requests are issued synchronously and from xfslogd context in XFS, hence the xfslogd workqueue is blocked in the upper fs waiting on a hole punch request to be servied in the lower fs. If the lower fs issues I/O that depends on xfslogd to complete, both filesystems end up hung indefinitely. This is reproduced reliabily by generic/013 on XFS->loop->XFS test devices with the '-o discard' mount option. Further, docker implementations appear to use this kind of configuration for container instance filesystems by default (container fs->dm-> loop->base fs) and therefore are subject to this deadlock when running on XFS. Replace the global xfslogd workqueue with a per-mount variant. This guarantees each mount access to a single worker and prevents deadlocks due to inter-fs dependencies introduced by discard. Since the queue is only responsible for buffer iodone processing at this point in time, rename xfslogd to xfs-buf. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-11-28 05:59:58 +03:00
out_destroy_buf:
destroy_workqueue(mp->m_buf_workqueue);
out:
return -ENOMEM;
}
STATIC void
xfs_destroy_mount_workqueues(
struct xfs_mount *mp)
{
destroy_workqueue(mp->m_sync_workqueue);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
destroy_workqueue(mp->m_blockgc_wq);
destroy_workqueue(mp->m_inodegc_wq);
destroy_workqueue(mp->m_reclaim_workqueue);
destroy_workqueue(mp->m_unwritten_workqueue);
xfs: replace global xfslogd wq with per-mount wq The xfslogd workqueue is a global, single-job workqueue for buffer ioend processing. This means we allow for a single work item at a time for all possible XFS mounts on a system. fsstress testing in loopback XFS over XFS configurations has reproduced xfslogd deadlocks due to the single threaded nature of the queue and dependencies introduced between the separate XFS instances by online discard (-o discard). Discard over a loopback device converts the discard request to a hole punch (fallocate) on the underlying file. Online discard requests are issued synchronously and from xfslogd context in XFS, hence the xfslogd workqueue is blocked in the upper fs waiting on a hole punch request to be servied in the lower fs. If the lower fs issues I/O that depends on xfslogd to complete, both filesystems end up hung indefinitely. This is reproduced reliabily by generic/013 on XFS->loop->XFS test devices with the '-o discard' mount option. Further, docker implementations appear to use this kind of configuration for container instance filesystems by default (container fs->dm-> loop->base fs) and therefore are subject to this deadlock when running on XFS. Replace the global xfslogd workqueue with a per-mount variant. This guarantees each mount access to a single worker and prevents deadlocks due to inter-fs dependencies introduced by discard. Since the queue is only responsible for buffer iodone processing at this point in time, rename xfslogd to xfs-buf. Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-11-28 05:59:58 +03:00
destroy_workqueue(mp->m_buf_workqueue);
}
static void
xfs_flush_inodes_worker(
struct work_struct *work)
{
struct xfs_mount *mp = container_of(work, struct xfs_mount,
m_flush_inodes_work);
struct super_block *sb = mp->m_super;
if (down_read_trylock(&sb->s_umount)) {
sync_inodes_sb(sb);
up_read(&sb->s_umount);
}
}
xfs: xfs_sync_data is redundant. We don't do any data writeback from XFS any more - the VFS is completely responsible for that, including for freeze. We can replace the remaining caller with a VFS level function that achieves the same thing, but without conflicting with current writeback work. This means we can remove the flush_work and xfs_flush_inodes() - the VFS functionality completely replaces the internal flush queue for doing this writeback work in a separate context to avoid stack overruns. This does have one complication - it cannot be called with page locks held. Hence move the flushing of delalloc space when ENOSPC occurs back up into xfs_file_aio_buffered_write when we don't hold any locks that will stall writeback. Unfortunately, writeback_inodes_sb_if_idle() is not sufficient to trigger delalloc conversion fast enough to prevent spurious ENOSPC whent here are hundreds of writers, thousands of small files and GBs of free RAM. Hence we need to use sync_sb_inodes() to block callers while we wait for writeback like the previous xfs_flush_inodes implementation did. That means we have to hold the s_umount lock here, but because this call can nest inside i_mutex (the parent directory in the create case, held by the VFS), we have to use down_read_trylock() to avoid potential deadlocks. In practice, this trylock will succeed on almost every attempt as unmount/remount type operations are exceedingly rare. Note: we always need to pass a count of zero to generic_file_buffered_write() as the previously written byte count. We only do this by accident before this patch by the virtue of ret always being zero when there are no errors. Make this explicit rather than needing to specifically zero ret in the ENOSPC retry case. Signed-off-by: Dave Chinner <dchinner@redhat.com> Tested-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-10-08 14:56:04 +04:00
/*
* Flush all dirty data to disk. Must not be called while holding an XFS_ILOCK
* or a page lock. We use sync_inodes_sb() here to ensure we block while waiting
* for IO to complete so that we effectively throttle multiple callers to the
* rate at which IO is completing.
*/
void
xfs_flush_inodes(
struct xfs_mount *mp)
{
/*
* If flush_work() returns true then that means we waited for a flush
* which was already in progress. Don't bother running another scan.
*/
if (flush_work(&mp->m_flush_inodes_work))
return;
queue_work(mp->m_sync_workqueue, &mp->m_flush_inodes_work);
flush_work(&mp->m_flush_inodes_work);
xfs: xfs_sync_data is redundant. We don't do any data writeback from XFS any more - the VFS is completely responsible for that, including for freeze. We can replace the remaining caller with a VFS level function that achieves the same thing, but without conflicting with current writeback work. This means we can remove the flush_work and xfs_flush_inodes() - the VFS functionality completely replaces the internal flush queue for doing this writeback work in a separate context to avoid stack overruns. This does have one complication - it cannot be called with page locks held. Hence move the flushing of delalloc space when ENOSPC occurs back up into xfs_file_aio_buffered_write when we don't hold any locks that will stall writeback. Unfortunately, writeback_inodes_sb_if_idle() is not sufficient to trigger delalloc conversion fast enough to prevent spurious ENOSPC whent here are hundreds of writers, thousands of small files and GBs of free RAM. Hence we need to use sync_sb_inodes() to block callers while we wait for writeback like the previous xfs_flush_inodes implementation did. That means we have to hold the s_umount lock here, but because this call can nest inside i_mutex (the parent directory in the create case, held by the VFS), we have to use down_read_trylock() to avoid potential deadlocks. In practice, this trylock will succeed on almost every attempt as unmount/remount type operations are exceedingly rare. Note: we always need to pass a count of zero to generic_file_buffered_write() as the previously written byte count. We only do this by accident before this patch by the virtue of ret always being zero when there are no errors. Make this explicit rather than needing to specifically zero ret in the ENOSPC retry case. Signed-off-by: Dave Chinner <dchinner@redhat.com> Tested-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-10-08 14:56:04 +04:00
}
/* Catch misguided souls that try to use this interface on XFS */
STATIC struct inode *
xfs_fs_alloc_inode(
struct super_block *sb)
{
BUG();
return NULL;
}
/*
* Now that the generic code is guaranteed not to be accessing
xfs: remove xfs_fs_evict_inode() Joe Lawrence reported a list_add corruption with 4.6-rc1 when testing some custom md administration code that made it's own block device nodes for the md array. The simple test loop of: for i in {0..100}; do mknod --mode=0600 $tmp/tmp_node b $MAJOR $MINOR mdadm --detail --export $tmp/tmp_node > /dev/null rm -f $tmp/tmp_node done Would produce this warning in bd_acquire() when mdadm opened the device node: list_add double add: new=ffff88043831c7b8, prev=ffff8804380287d8, next=ffff88043831c7b8. And then produce this from bd_forget from kdevtmpfs evicting a block dev inode: list_del corruption. prev->next should be ffff8800bb83eb10, but was ffff88043831c7b8 This is a regression caused by commit c19b3b05 ("xfs: mode di_mode to vfs inode"). The issue is that xfs_inactive() frees the unlinked inode, and the above commit meant that this freeing zeroed the mode in the struct inode. The problem is that after evict() has called ->evict_inode, it expects the i_mode to be intact so that it can call bd_forget() or cd_forget() to drop the reference to the block device inode attached to the XFS inode. In reality, the only thing we do in xfs_fs_evict_inode() that is not generic is call xfs_inactive(). We can move the xfs_inactive() call to xfs_fs_destroy_inode() without any problems at all, and this will leave the VFS inode intact until it is completely done with it. So, remove xfs_fs_evict_inode(), and do the work it used to do in ->destroy_inode instead. cc: <stable@vger.kernel.org> # 4.6 Reported-by: Joe Lawrence <joe.lawrence@stratus.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2016-05-18 06:52:42 +03:00
* the linux inode, we can inactivate and reclaim the inode.
*/
STATIC void
xfs_fs_destroy_inode(
struct inode *inode)
{
struct xfs_inode *ip = XFS_I(inode);
trace_xfs_destroy_inode(ip);
ASSERT(!rwsem_is_locked(&inode->i_rwsem));
xfs: remove xfs_fs_evict_inode() Joe Lawrence reported a list_add corruption with 4.6-rc1 when testing some custom md administration code that made it's own block device nodes for the md array. The simple test loop of: for i in {0..100}; do mknod --mode=0600 $tmp/tmp_node b $MAJOR $MINOR mdadm --detail --export $tmp/tmp_node > /dev/null rm -f $tmp/tmp_node done Would produce this warning in bd_acquire() when mdadm opened the device node: list_add double add: new=ffff88043831c7b8, prev=ffff8804380287d8, next=ffff88043831c7b8. And then produce this from bd_forget from kdevtmpfs evicting a block dev inode: list_del corruption. prev->next should be ffff8800bb83eb10, but was ffff88043831c7b8 This is a regression caused by commit c19b3b05 ("xfs: mode di_mode to vfs inode"). The issue is that xfs_inactive() frees the unlinked inode, and the above commit meant that this freeing zeroed the mode in the struct inode. The problem is that after evict() has called ->evict_inode, it expects the i_mode to be intact so that it can call bd_forget() or cd_forget() to drop the reference to the block device inode attached to the XFS inode. In reality, the only thing we do in xfs_fs_evict_inode() that is not generic is call xfs_inactive(). We can move the xfs_inactive() call to xfs_fs_destroy_inode() without any problems at all, and this will leave the VFS inode intact until it is completely done with it. So, remove xfs_fs_evict_inode(), and do the work it used to do in ->destroy_inode instead. cc: <stable@vger.kernel.org> # 4.6 Reported-by: Joe Lawrence <joe.lawrence@stratus.com> Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2016-05-18 06:52:42 +03:00
XFS_STATS_INC(ip->i_mount, vn_rele);
XFS_STATS_INC(ip->i_mount, vn_remove);
xfs_inode_mark_reclaimable(ip);
}
static void
xfs_fs_dirty_inode(
struct inode *inode,
int flag)
{
struct xfs_inode *ip = XFS_I(inode);
struct xfs_mount *mp = ip->i_mount;
struct xfs_trans *tp;
if (!(inode->i_sb->s_flags & SB_LAZYTIME))
return;
if (flag != I_DIRTY_SYNC || !(inode->i_state & I_DIRTY_TIME))
return;
if (xfs_trans_alloc(mp, &M_RES(mp)->tr_fsyncts, 0, 0, 0, &tp))
return;
xfs_ilock(ip, XFS_ILOCK_EXCL);
xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
xfs_trans_log_inode(tp, ip, XFS_ILOG_TIMESTAMP);
xfs_trans_commit(tp);
}
/*
* Slab object creation initialisation for the XFS inode.
* This covers only the idempotent fields in the XFS inode;
* all other fields need to be initialised on allocation
* from the slab. This avoids the need to repeatedly initialise
* fields in the xfs inode that left in the initialise state
* when freeing the inode.
*/
STATIC void
xfs_fs_inode_init_once(
void *inode)
{
struct xfs_inode *ip = inode;
memset(ip, 0, sizeof(struct xfs_inode));
/* vfs inode */
inode_init_once(VFS_I(ip));
/* xfs inode */
atomic_set(&ip->i_pincount, 0);
spin_lock_init(&ip->i_flags_lock);
mrlock_init(&ip->i_lock, MRLOCK_ALLOW_EQUAL_PRI|MRLOCK_BARRIER,
"xfsino", ip->i_ino);
}
/*
* We do an unlocked check for XFS_IDONTCACHE here because we are already
* serialised against cache hits here via the inode->i_lock and igrab() in
* xfs_iget_cache_hit(). Hence a lookup that might clear this flag will not be
* racing with us, and it avoids needing to grab a spinlock here for every inode
* we drop the final reference on.
*/
STATIC int
xfs_fs_drop_inode(
struct inode *inode)
{
struct xfs_inode *ip = XFS_I(inode);
/*
* If this unlinked inode is in the middle of recovery, don't
* drop the inode just yet; log recovery will take care of
* that. See the comment for this inode flag.
*/
if (ip->i_flags & XFS_IRECOVERY) {
ASSERT(xlog_recovery_needed(ip->i_mount->m_log));
return 0;
}
return generic_drop_inode(inode);
}
static void
xfs_mount_free(
struct xfs_mount *mp)
{
kfree(mp->m_rtname);
kfree(mp->m_logname);
kmem_free(mp);
}
STATIC int
xfs_fs_sync_fs(
struct super_block *sb,
int wait)
{
struct xfs_mount *mp = XFS_M(sb);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
trace_xfs_fs_sync_fs(mp, __return_address);
/*
* Doing anything during the async pass would be counterproductive.
*/
if (!wait)
return 0;
xfs_log_force(mp, XFS_LOG_SYNC);
if (laptop_mode) {
/*
* The disk must be active because we're syncing.
* We schedule log work now (now that the disk is
* active) instead of later (when it might not be).
*/
flush_delayed_work(&mp->m_log->l_work);
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
/*
* If we are called with page faults frozen out, it means we are about
* to freeze the transaction subsystem. Take the opportunity to shut
* down inodegc because once SB_FREEZE_FS is set it's too late to
* prevent inactivation races with freeze. The fs doesn't get called
* again by the freezing process until after SB_FREEZE_FS has been set,
* so it's now or never. Same logic applies to speculative allocation
* garbage collection.
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
*
* We don't care if this is a normal syncfs call that does this or
* freeze that does this - we can run this multiple times without issue
* and we won't race with a restart because a restart can only occur
* when the state is either SB_FREEZE_FS or SB_FREEZE_COMPLETE.
*/
if (sb->s_writers.frozen == SB_FREEZE_PAGEFAULT) {
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_stop(mp);
xfs_blockgc_stop(mp);
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
return 0;
}
STATIC int
xfs_fs_statfs(
struct dentry *dentry,
struct kstatfs *statp)
{
struct xfs_mount *mp = XFS_M(dentry->d_sb);
xfs_sb_t *sbp = &mp->m_sb;
struct xfs_inode *ip = XFS_I(d_inode(dentry));
uint64_t fakeinos, id;
uint64_t icount;
uint64_t ifree;
uint64_t fdblocks;
xfs_extlen_t lsize;
int64_t ffree;
/* Wait for whatever inactivations are in progress. */
xfs_inodegc_flush(mp);
statp->f_type = XFS_SUPER_MAGIC;
statp->f_namelen = MAXNAMELEN - 1;
id = huge_encode_dev(mp->m_ddev_targp->bt_dev);
statp->f_fsid = u64_to_fsid(id);
icount = percpu_counter_sum(&mp->m_icount);
ifree = percpu_counter_sum(&mp->m_ifree);
fdblocks = percpu_counter_sum(&mp->m_fdblocks);
spin_lock(&mp->m_sb_lock);
statp->f_bsize = sbp->sb_blocksize;
lsize = sbp->sb_logstart ? sbp->sb_logblocks : 0;
statp->f_blocks = sbp->sb_dblocks - lsize;
spin_unlock(&mp->m_sb_lock);
/* make sure statp->f_bfree does not underflow */
statp->f_bfree = max_t(int64_t, fdblocks - mp->m_alloc_set_aside, 0);
statp->f_bavail = statp->f_bfree;
fakeinos = XFS_FSB_TO_INO(mp, statp->f_bfree);
statp->f_files = min(icount + fakeinos, (uint64_t)XFS_MAXINUMBER);
if (M_IGEO(mp)->maxicount)
statp->f_files = min_t(typeof(statp->f_files),
statp->f_files,
M_IGEO(mp)->maxicount);
/* If sb_icount overshot maxicount, report actual allocation */
statp->f_files = max_t(typeof(statp->f_files),
statp->f_files,
sbp->sb_icount);
/* make sure statp->f_ffree does not underflow */
ffree = statp->f_files - (icount - ifree);
statp->f_ffree = max_t(int64_t, ffree, 0);
if ((ip->i_diflags & XFS_DIFLAG_PROJINHERIT) &&
((mp->m_qflags & (XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))) ==
(XFS_PQUOTA_ACCT|XFS_PQUOTA_ENFD))
xfs_qm_statvfs(ip, statp);
if (XFS_IS_REALTIME_MOUNT(mp) &&
(ip->i_diflags & (XFS_DIFLAG_RTINHERIT | XFS_DIFLAG_REALTIME))) {
statp->f_blocks = sbp->sb_rblocks;
statp->f_bavail = statp->f_bfree =
sbp->sb_frextents * sbp->sb_rextsize;
}
return 0;
}
STATIC void
xfs_save_resvblks(struct xfs_mount *mp)
{
uint64_t resblks = 0;
mp->m_resblks_save = mp->m_resblks;
xfs_reserve_blocks(mp, &resblks, NULL);
}
STATIC void
xfs_restore_resvblks(struct xfs_mount *mp)
{
uint64_t resblks;
if (mp->m_resblks_save) {
resblks = mp->m_resblks_save;
mp->m_resblks_save = 0;
} else
resblks = xfs_default_resblks(mp);
xfs_reserve_blocks(mp, &resblks, NULL);
}
/*
* Second stage of a freeze. The data is already frozen so we only
* need to take care of the metadata. Once that's done sync the superblock
* to the log to dirty it in case of a crash while frozen. This ensures that we
* will recover the unlinked inode lists on the next mount.
*/
STATIC int
xfs_fs_freeze(
struct super_block *sb)
{
struct xfs_mount *mp = XFS_M(sb);
xfs: Fix false positive lockdep warning with sb_internal & fs_reclaim Depending on the workloads, the following circular locking dependency warning between sb_internal (a percpu rwsem) and fs_reclaim (a pseudo lock) may show up: ====================================================== WARNING: possible circular locking dependency detected 5.0.0-rc1+ #60 Tainted: G W ------------------------------------------------------ fsfreeze/4346 is trying to acquire lock: 0000000026f1d784 (fs_reclaim){+.+.}, at: fs_reclaim_acquire.part.19+0x5/0x30 but task is already holding lock: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 which lock already depends on the new lock. : Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal); lock(fs_reclaim); lock(sb_internal); lock(fs_reclaim); *** DEADLOCK *** 4 locks held by fsfreeze/4346: #0: 00000000b478ef56 (sb_writers#8){++++}, at: percpu_down_write+0xb4/0x650 #1: 000000001ec487a9 (&type->s_umount_key#28){++++}, at: freeze_super+0xda/0x290 #2: 000000003edbd5a0 (sb_pagefaults){++++}, at: percpu_down_write+0xb4/0x650 #3: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 stack backtrace: Call Trace: dump_stack+0xe0/0x19a print_circular_bug.isra.10.cold.34+0x2f4/0x435 check_prev_add.constprop.19+0xca1/0x15f0 validate_chain.isra.14+0x11af/0x3b50 __lock_acquire+0x728/0x1200 lock_acquire+0x269/0x5a0 fs_reclaim_acquire.part.19+0x29/0x30 fs_reclaim_acquire+0x19/0x20 kmem_cache_alloc+0x3e/0x3f0 kmem_zone_alloc+0x79/0x150 xfs_trans_alloc+0xfa/0x9d0 xfs_sync_sb+0x86/0x170 xfs_log_sbcount+0x10f/0x140 xfs_quiesce_attr+0x134/0x270 xfs_fs_freeze+0x4a/0x70 freeze_super+0x1af/0x290 do_vfs_ioctl+0xedc/0x16c0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x73/0xa9 do_syscall_64+0x18f/0xd23 entry_SYSCALL_64_after_hwframe+0x49/0xbe This is a false positive as all the dirty pages are flushed out before the filesystem can be frozen. One way to avoid this splat is to add GFP_NOFS to the affected allocation calls by using the memalloc_nofs_save()/memalloc_nofs_restore() pair. This shouldn't matter unless the system is really running out of memory. In that particular case, the filesystem freeze operation may fail while it was succeeding previously. Without this patch, the command sequence below will show that the lock dependency chain sb_internal -> fs_reclaim exists. # fsfreeze -f /home # fsfreeze --unfreeze /home # grep -i fs_reclaim -C 3 /proc/lockdep_chains | grep -C 5 sb_internal After applying the patch, such sb_internal -> fs_reclaim lock dependency chain can no longer be found. Because of that, the locking dependency warning will not be shown. Suggested-by: Dave Chinner <david@fromorbit.com> Signed-off-by: Waiman Long <longman@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-07-08 20:21:44 +03:00
unsigned int flags;
int ret;
xfs: Fix false positive lockdep warning with sb_internal & fs_reclaim Depending on the workloads, the following circular locking dependency warning between sb_internal (a percpu rwsem) and fs_reclaim (a pseudo lock) may show up: ====================================================== WARNING: possible circular locking dependency detected 5.0.0-rc1+ #60 Tainted: G W ------------------------------------------------------ fsfreeze/4346 is trying to acquire lock: 0000000026f1d784 (fs_reclaim){+.+.}, at: fs_reclaim_acquire.part.19+0x5/0x30 but task is already holding lock: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 which lock already depends on the new lock. : Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal); lock(fs_reclaim); lock(sb_internal); lock(fs_reclaim); *** DEADLOCK *** 4 locks held by fsfreeze/4346: #0: 00000000b478ef56 (sb_writers#8){++++}, at: percpu_down_write+0xb4/0x650 #1: 000000001ec487a9 (&type->s_umount_key#28){++++}, at: freeze_super+0xda/0x290 #2: 000000003edbd5a0 (sb_pagefaults){++++}, at: percpu_down_write+0xb4/0x650 #3: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 stack backtrace: Call Trace: dump_stack+0xe0/0x19a print_circular_bug.isra.10.cold.34+0x2f4/0x435 check_prev_add.constprop.19+0xca1/0x15f0 validate_chain.isra.14+0x11af/0x3b50 __lock_acquire+0x728/0x1200 lock_acquire+0x269/0x5a0 fs_reclaim_acquire.part.19+0x29/0x30 fs_reclaim_acquire+0x19/0x20 kmem_cache_alloc+0x3e/0x3f0 kmem_zone_alloc+0x79/0x150 xfs_trans_alloc+0xfa/0x9d0 xfs_sync_sb+0x86/0x170 xfs_log_sbcount+0x10f/0x140 xfs_quiesce_attr+0x134/0x270 xfs_fs_freeze+0x4a/0x70 freeze_super+0x1af/0x290 do_vfs_ioctl+0xedc/0x16c0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x73/0xa9 do_syscall_64+0x18f/0xd23 entry_SYSCALL_64_after_hwframe+0x49/0xbe This is a false positive as all the dirty pages are flushed out before the filesystem can be frozen. One way to avoid this splat is to add GFP_NOFS to the affected allocation calls by using the memalloc_nofs_save()/memalloc_nofs_restore() pair. This shouldn't matter unless the system is really running out of memory. In that particular case, the filesystem freeze operation may fail while it was succeeding previously. Without this patch, the command sequence below will show that the lock dependency chain sb_internal -> fs_reclaim exists. # fsfreeze -f /home # fsfreeze --unfreeze /home # grep -i fs_reclaim -C 3 /proc/lockdep_chains | grep -C 5 sb_internal After applying the patch, such sb_internal -> fs_reclaim lock dependency chain can no longer be found. Because of that, the locking dependency warning will not be shown. Suggested-by: Dave Chinner <david@fromorbit.com> Signed-off-by: Waiman Long <longman@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-07-08 20:21:44 +03:00
/*
* The filesystem is now frozen far enough that memory reclaim
* cannot safely operate on the filesystem. Hence we need to
* set a GFP_NOFS context here to avoid recursion deadlocks.
*/
flags = memalloc_nofs_save();
xfs_save_resvblks(mp);
ret = xfs_log_quiesce(mp);
xfs: Fix false positive lockdep warning with sb_internal & fs_reclaim Depending on the workloads, the following circular locking dependency warning between sb_internal (a percpu rwsem) and fs_reclaim (a pseudo lock) may show up: ====================================================== WARNING: possible circular locking dependency detected 5.0.0-rc1+ #60 Tainted: G W ------------------------------------------------------ fsfreeze/4346 is trying to acquire lock: 0000000026f1d784 (fs_reclaim){+.+.}, at: fs_reclaim_acquire.part.19+0x5/0x30 but task is already holding lock: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 which lock already depends on the new lock. : Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal); lock(fs_reclaim); lock(sb_internal); lock(fs_reclaim); *** DEADLOCK *** 4 locks held by fsfreeze/4346: #0: 00000000b478ef56 (sb_writers#8){++++}, at: percpu_down_write+0xb4/0x650 #1: 000000001ec487a9 (&type->s_umount_key#28){++++}, at: freeze_super+0xda/0x290 #2: 000000003edbd5a0 (sb_pagefaults){++++}, at: percpu_down_write+0xb4/0x650 #3: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 stack backtrace: Call Trace: dump_stack+0xe0/0x19a print_circular_bug.isra.10.cold.34+0x2f4/0x435 check_prev_add.constprop.19+0xca1/0x15f0 validate_chain.isra.14+0x11af/0x3b50 __lock_acquire+0x728/0x1200 lock_acquire+0x269/0x5a0 fs_reclaim_acquire.part.19+0x29/0x30 fs_reclaim_acquire+0x19/0x20 kmem_cache_alloc+0x3e/0x3f0 kmem_zone_alloc+0x79/0x150 xfs_trans_alloc+0xfa/0x9d0 xfs_sync_sb+0x86/0x170 xfs_log_sbcount+0x10f/0x140 xfs_quiesce_attr+0x134/0x270 xfs_fs_freeze+0x4a/0x70 freeze_super+0x1af/0x290 do_vfs_ioctl+0xedc/0x16c0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x73/0xa9 do_syscall_64+0x18f/0xd23 entry_SYSCALL_64_after_hwframe+0x49/0xbe This is a false positive as all the dirty pages are flushed out before the filesystem can be frozen. One way to avoid this splat is to add GFP_NOFS to the affected allocation calls by using the memalloc_nofs_save()/memalloc_nofs_restore() pair. This shouldn't matter unless the system is really running out of memory. In that particular case, the filesystem freeze operation may fail while it was succeeding previously. Without this patch, the command sequence below will show that the lock dependency chain sb_internal -> fs_reclaim exists. # fsfreeze -f /home # fsfreeze --unfreeze /home # grep -i fs_reclaim -C 3 /proc/lockdep_chains | grep -C 5 sb_internal After applying the patch, such sb_internal -> fs_reclaim lock dependency chain can no longer be found. Because of that, the locking dependency warning will not be shown. Suggested-by: Dave Chinner <david@fromorbit.com> Signed-off-by: Waiman Long <longman@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-07-08 20:21:44 +03:00
memalloc_nofs_restore(flags);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
/*
* For read-write filesystems, we need to restart the inodegc on error
* because we stopped it at SB_FREEZE_PAGEFAULT level and a thaw is not
* going to be run to restart it now. We are at SB_FREEZE_FS level
* here, so we can restart safely without racing with a stop in
* xfs_fs_sync_fs().
*/
if (ret && !xfs_is_readonly(mp)) {
xfs_blockgc_start(mp);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_start(mp);
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs: Fix false positive lockdep warning with sb_internal & fs_reclaim Depending on the workloads, the following circular locking dependency warning between sb_internal (a percpu rwsem) and fs_reclaim (a pseudo lock) may show up: ====================================================== WARNING: possible circular locking dependency detected 5.0.0-rc1+ #60 Tainted: G W ------------------------------------------------------ fsfreeze/4346 is trying to acquire lock: 0000000026f1d784 (fs_reclaim){+.+.}, at: fs_reclaim_acquire.part.19+0x5/0x30 but task is already holding lock: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 which lock already depends on the new lock. : Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(sb_internal); lock(fs_reclaim); lock(sb_internal); lock(fs_reclaim); *** DEADLOCK *** 4 locks held by fsfreeze/4346: #0: 00000000b478ef56 (sb_writers#8){++++}, at: percpu_down_write+0xb4/0x650 #1: 000000001ec487a9 (&type->s_umount_key#28){++++}, at: freeze_super+0xda/0x290 #2: 000000003edbd5a0 (sb_pagefaults){++++}, at: percpu_down_write+0xb4/0x650 #3: 0000000072bfc54b (sb_internal){++++}, at: percpu_down_write+0xb4/0x650 stack backtrace: Call Trace: dump_stack+0xe0/0x19a print_circular_bug.isra.10.cold.34+0x2f4/0x435 check_prev_add.constprop.19+0xca1/0x15f0 validate_chain.isra.14+0x11af/0x3b50 __lock_acquire+0x728/0x1200 lock_acquire+0x269/0x5a0 fs_reclaim_acquire.part.19+0x29/0x30 fs_reclaim_acquire+0x19/0x20 kmem_cache_alloc+0x3e/0x3f0 kmem_zone_alloc+0x79/0x150 xfs_trans_alloc+0xfa/0x9d0 xfs_sync_sb+0x86/0x170 xfs_log_sbcount+0x10f/0x140 xfs_quiesce_attr+0x134/0x270 xfs_fs_freeze+0x4a/0x70 freeze_super+0x1af/0x290 do_vfs_ioctl+0xedc/0x16c0 ksys_ioctl+0x41/0x80 __x64_sys_ioctl+0x73/0xa9 do_syscall_64+0x18f/0xd23 entry_SYSCALL_64_after_hwframe+0x49/0xbe This is a false positive as all the dirty pages are flushed out before the filesystem can be frozen. One way to avoid this splat is to add GFP_NOFS to the affected allocation calls by using the memalloc_nofs_save()/memalloc_nofs_restore() pair. This shouldn't matter unless the system is really running out of memory. In that particular case, the filesystem freeze operation may fail while it was succeeding previously. Without this patch, the command sequence below will show that the lock dependency chain sb_internal -> fs_reclaim exists. # fsfreeze -f /home # fsfreeze --unfreeze /home # grep -i fs_reclaim -C 3 /proc/lockdep_chains | grep -C 5 sb_internal After applying the patch, such sb_internal -> fs_reclaim lock dependency chain can no longer be found. Because of that, the locking dependency warning will not be shown. Suggested-by: Dave Chinner <david@fromorbit.com> Signed-off-by: Waiman Long <longman@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de>
2020-07-08 20:21:44 +03:00
return ret;
}
STATIC int
xfs_fs_unfreeze(
struct super_block *sb)
{
struct xfs_mount *mp = XFS_M(sb);
xfs_restore_resvblks(mp);
xfs_log_work_queue(mp);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
/*
* Don't reactivate the inodegc worker on a readonly filesystem because
* inodes are sent directly to reclaim. Don't reactivate the blockgc
* worker because there are no speculative preallocations on a readonly
* filesystem.
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
*/
if (!xfs_is_readonly(mp)) {
xfs_blockgc_start(mp);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_start(mp);
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
return 0;
}
/*
* This function fills in xfs_mount_t fields based on mount args.
* Note: the superblock _has_ now been read in.
*/
STATIC int
xfs_finish_flags(
struct xfs_mount *mp)
{
/* Fail a mount where the logbuf is smaller than the log stripe */
if (xfs_has_logv2(mp)) {
if (mp->m_logbsize <= 0 &&
mp->m_sb.sb_logsunit > XLOG_BIG_RECORD_BSIZE) {
mp->m_logbsize = mp->m_sb.sb_logsunit;
} else if (mp->m_logbsize > 0 &&
mp->m_logbsize < mp->m_sb.sb_logsunit) {
xfs_warn(mp,
"logbuf size must be greater than or equal to log stripe size");
return -EINVAL;
}
} else {
/* Fail a mount if the logbuf is larger than 32K */
if (mp->m_logbsize > XLOG_BIG_RECORD_BSIZE) {
xfs_warn(mp,
"logbuf size for version 1 logs must be 16K or 32K");
return -EINVAL;
}
}
/*
* V5 filesystems always use attr2 format for attributes.
*/
if (xfs_has_crc(mp) && xfs_has_noattr2(mp)) {
xfs_warn(mp, "Cannot mount a V5 filesystem as noattr2. "
"attr2 is always enabled for V5 filesystems.");
return -EINVAL;
}
/*
* prohibit r/w mounts of read-only filesystems
*/
if ((mp->m_sb.sb_flags & XFS_SBF_READONLY) && !xfs_is_readonly(mp)) {
xfs_warn(mp,
"cannot mount a read-only filesystem as read-write");
return -EROFS;
}
if ((mp->m_qflags & XFS_GQUOTA_ACCT) &&
(mp->m_qflags & XFS_PQUOTA_ACCT) &&
!xfs_has_pquotino(mp)) {
xfs_warn(mp,
"Super block does not support project and group quota together");
return -EINVAL;
}
return 0;
}
static int
xfs_init_percpu_counters(
struct xfs_mount *mp)
{
int error;
error = percpu_counter_init(&mp->m_icount, 0, GFP_KERNEL);
if (error)
return -ENOMEM;
error = percpu_counter_init(&mp->m_ifree, 0, GFP_KERNEL);
if (error)
goto free_icount;
error = percpu_counter_init(&mp->m_fdblocks, 0, GFP_KERNEL);
if (error)
goto free_ifree;
error = percpu_counter_init(&mp->m_delalloc_blks, 0, GFP_KERNEL);
if (error)
goto free_fdblocks;
return 0;
free_fdblocks:
percpu_counter_destroy(&mp->m_fdblocks);
free_ifree:
percpu_counter_destroy(&mp->m_ifree);
free_icount:
percpu_counter_destroy(&mp->m_icount);
return -ENOMEM;
}
void
xfs_reinit_percpu_counters(
struct xfs_mount *mp)
{
percpu_counter_set(&mp->m_icount, mp->m_sb.sb_icount);
percpu_counter_set(&mp->m_ifree, mp->m_sb.sb_ifree);
percpu_counter_set(&mp->m_fdblocks, mp->m_sb.sb_fdblocks);
}
static void
xfs_destroy_percpu_counters(
struct xfs_mount *mp)
{
percpu_counter_destroy(&mp->m_icount);
percpu_counter_destroy(&mp->m_ifree);
percpu_counter_destroy(&mp->m_fdblocks);
ASSERT(xfs_is_shutdown(mp) ||
percpu_counter_sum(&mp->m_delalloc_blks) == 0);
percpu_counter_destroy(&mp->m_delalloc_blks);
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
static int
xfs_inodegc_init_percpu(
struct xfs_mount *mp)
{
struct xfs_inodegc *gc;
int cpu;
mp->m_inodegc = alloc_percpu(struct xfs_inodegc);
if (!mp->m_inodegc)
return -ENOMEM;
for_each_possible_cpu(cpu) {
gc = per_cpu_ptr(mp->m_inodegc, cpu);
init_llist_head(&gc->list);
gc->items = 0;
INIT_WORK(&gc->work, xfs_inodegc_worker);
}
return 0;
}
static void
xfs_inodegc_free_percpu(
struct xfs_mount *mp)
{
if (!mp->m_inodegc)
return;
free_percpu(mp->m_inodegc);
}
static void
xfs_fs_put_super(
struct super_block *sb)
{
struct xfs_mount *mp = XFS_M(sb);
/* if ->fill_super failed, we have no mount to tear down */
if (!sb->s_fs_info)
return;
xfs_notice(mp, "Unmounting Filesystem");
xfs_filestream_unmount(mp);
xfs_unmountfs(mp);
xfs_freesb(mp);
free_percpu(mp->m_stats.xs_stats);
xfs_mount_list_del(mp);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_free_percpu(mp);
xfs_destroy_percpu_counters(mp);
xfs_destroy_mount_workqueues(mp);
xfs_close_devices(mp);
sb->s_fs_info = NULL;
xfs_mount_free(mp);
}
static long
xfs_fs_nr_cached_objects(
struct super_block *sb,
struct shrink_control *sc)
{
/* Paranoia: catch incorrect calls during mount setup or teardown */
if (WARN_ON_ONCE(!sb->s_fs_info))
return 0;
return xfs_reclaim_inodes_count(XFS_M(sb));
}
static long
xfs_fs_free_cached_objects(
struct super_block *sb,
struct shrink_control *sc)
{
return xfs_reclaim_inodes_nr(XFS_M(sb), sc->nr_to_scan);
}
static const struct super_operations xfs_super_operations = {
.alloc_inode = xfs_fs_alloc_inode,
.destroy_inode = xfs_fs_destroy_inode,
.dirty_inode = xfs_fs_dirty_inode,
.drop_inode = xfs_fs_drop_inode,
.put_super = xfs_fs_put_super,
.sync_fs = xfs_fs_sync_fs,
.freeze_fs = xfs_fs_freeze,
.unfreeze_fs = xfs_fs_unfreeze,
.statfs = xfs_fs_statfs,
.show_options = xfs_fs_show_options,
.nr_cached_objects = xfs_fs_nr_cached_objects,
.free_cached_objects = xfs_fs_free_cached_objects,
};
static int
suffix_kstrtoint(
const char *s,
unsigned int base,
int *res)
{
int last, shift_left_factor = 0, _res;
char *value;
int ret = 0;
value = kstrdup(s, GFP_KERNEL);
if (!value)
return -ENOMEM;
last = strlen(value) - 1;
if (value[last] == 'K' || value[last] == 'k') {
shift_left_factor = 10;
value[last] = '\0';
}
if (value[last] == 'M' || value[last] == 'm') {
shift_left_factor = 20;
value[last] = '\0';
}
if (value[last] == 'G' || value[last] == 'g') {
shift_left_factor = 30;
value[last] = '\0';
}
if (kstrtoint(value, base, &_res))
ret = -EINVAL;
kfree(value);
*res = _res << shift_left_factor;
return ret;
}
static inline void
xfs_fs_warn_deprecated(
struct fs_context *fc,
struct fs_parameter *param,
uint64_t flag,
bool value)
{
/* Don't print the warning if reconfiguring and current mount point
* already had the flag set
*/
if ((fc->purpose & FS_CONTEXT_FOR_RECONFIGURE) &&
!!(XFS_M(fc->root->d_sb)->m_features & flag) == value)
return;
xfs_warn(fc->s_fs_info, "%s mount option is deprecated.", param->key);
}
/*
* Set mount state from a mount option.
*
* NOTE: mp->m_super is NULL here!
*/
static int
xfs_fs_parse_param(
struct fs_context *fc,
struct fs_parameter *param)
{
struct xfs_mount *parsing_mp = fc->s_fs_info;
struct fs_parse_result result;
int size = 0;
int opt;
opt = fs_parse(fc, xfs_fs_parameters, param, &result);
if (opt < 0)
return opt;
switch (opt) {
case Opt_logbufs:
parsing_mp->m_logbufs = result.uint_32;
return 0;
case Opt_logbsize:
if (suffix_kstrtoint(param->string, 10, &parsing_mp->m_logbsize))
return -EINVAL;
return 0;
case Opt_logdev:
kfree(parsing_mp->m_logname);
parsing_mp->m_logname = kstrdup(param->string, GFP_KERNEL);
if (!parsing_mp->m_logname)
return -ENOMEM;
return 0;
case Opt_rtdev:
kfree(parsing_mp->m_rtname);
parsing_mp->m_rtname = kstrdup(param->string, GFP_KERNEL);
if (!parsing_mp->m_rtname)
return -ENOMEM;
return 0;
case Opt_allocsize:
if (suffix_kstrtoint(param->string, 10, &size))
return -EINVAL;
parsing_mp->m_allocsize_log = ffs(size) - 1;
parsing_mp->m_features |= XFS_FEAT_ALLOCSIZE;
return 0;
case Opt_grpid:
case Opt_bsdgroups:
parsing_mp->m_features |= XFS_FEAT_GRPID;
return 0;
case Opt_nogrpid:
case Opt_sysvgroups:
parsing_mp->m_features &= ~XFS_FEAT_GRPID;
return 0;
case Opt_wsync:
parsing_mp->m_features |= XFS_FEAT_WSYNC;
return 0;
case Opt_norecovery:
parsing_mp->m_features |= XFS_FEAT_NORECOVERY;
return 0;
case Opt_noalign:
parsing_mp->m_features |= XFS_FEAT_NOALIGN;
return 0;
case Opt_swalloc:
parsing_mp->m_features |= XFS_FEAT_SWALLOC;
return 0;
case Opt_sunit:
parsing_mp->m_dalign = result.uint_32;
return 0;
case Opt_swidth:
parsing_mp->m_swidth = result.uint_32;
return 0;
case Opt_inode32:
parsing_mp->m_features |= XFS_FEAT_SMALL_INUMS;
return 0;
case Opt_inode64:
parsing_mp->m_features &= ~XFS_FEAT_SMALL_INUMS;
return 0;
case Opt_nouuid:
parsing_mp->m_features |= XFS_FEAT_NOUUID;
return 0;
case Opt_largeio:
parsing_mp->m_features |= XFS_FEAT_LARGE_IOSIZE;
return 0;
case Opt_nolargeio:
parsing_mp->m_features &= ~XFS_FEAT_LARGE_IOSIZE;
return 0;
case Opt_filestreams:
parsing_mp->m_features |= XFS_FEAT_FILESTREAMS;
return 0;
case Opt_noquota:
parsing_mp->m_qflags &= ~XFS_ALL_QUOTA_ACCT;
parsing_mp->m_qflags &= ~XFS_ALL_QUOTA_ENFD;
return 0;
case Opt_quota:
case Opt_uquota:
case Opt_usrquota:
parsing_mp->m_qflags |= (XFS_UQUOTA_ACCT | XFS_UQUOTA_ENFD);
return 0;
case Opt_qnoenforce:
case Opt_uqnoenforce:
parsing_mp->m_qflags |= XFS_UQUOTA_ACCT;
parsing_mp->m_qflags &= ~XFS_UQUOTA_ENFD;
return 0;
case Opt_pquota:
case Opt_prjquota:
parsing_mp->m_qflags |= (XFS_PQUOTA_ACCT | XFS_PQUOTA_ENFD);
return 0;
case Opt_pqnoenforce:
parsing_mp->m_qflags |= XFS_PQUOTA_ACCT;
parsing_mp->m_qflags &= ~XFS_PQUOTA_ENFD;
return 0;
case Opt_gquota:
case Opt_grpquota:
parsing_mp->m_qflags |= (XFS_GQUOTA_ACCT | XFS_GQUOTA_ENFD);
return 0;
case Opt_gqnoenforce:
parsing_mp->m_qflags |= XFS_GQUOTA_ACCT;
parsing_mp->m_qflags &= ~XFS_GQUOTA_ENFD;
return 0;
case Opt_discard:
parsing_mp->m_features |= XFS_FEAT_DISCARD;
return 0;
case Opt_nodiscard:
parsing_mp->m_features &= ~XFS_FEAT_DISCARD;
return 0;
#ifdef CONFIG_FS_DAX
case Opt_dax:
xfs_mount_set_dax_mode(parsing_mp, XFS_DAX_ALWAYS);
return 0;
case Opt_dax_enum:
xfs_mount_set_dax_mode(parsing_mp, result.uint_32);
return 0;
#endif
/* Following mount options will be removed in September 2025 */
case Opt_ikeep:
xfs_fs_warn_deprecated(fc, param, XFS_FEAT_IKEEP, true);
parsing_mp->m_features |= XFS_FEAT_IKEEP;
return 0;
case Opt_noikeep:
xfs_fs_warn_deprecated(fc, param, XFS_FEAT_IKEEP, false);
parsing_mp->m_features &= ~XFS_FEAT_IKEEP;
return 0;
case Opt_attr2:
xfs_fs_warn_deprecated(fc, param, XFS_FEAT_ATTR2, true);
parsing_mp->m_features |= XFS_FEAT_ATTR2;
return 0;
case Opt_noattr2:
xfs_fs_warn_deprecated(fc, param, XFS_FEAT_NOATTR2, true);
parsing_mp->m_features |= XFS_FEAT_NOATTR2;
return 0;
default:
xfs_warn(parsing_mp, "unknown mount option [%s].", param->key);
return -EINVAL;
}
return 0;
}
static int
xfs_fs_validate_params(
struct xfs_mount *mp)
{
/* No recovery flag requires a read-only mount */
if (xfs_has_norecovery(mp) && !xfs_is_readonly(mp)) {
xfs_warn(mp, "no-recovery mounts must be read-only.");
return -EINVAL;
}
/*
* We have not read the superblock at this point, so only the attr2
* mount option can set the attr2 feature by this stage.
*/
if (xfs_has_attr2(mp) && xfs_has_noattr2(mp)) {
xfs: rework attr2 feature and mount options The attr2 feature is somewhat unique in that it has both a superblock feature bit to enable it and mount options to enable and disable it. Back when it was first introduced in 2005, attr2 was disabled unless either the attr2 superblock feature bit was set, or the attr2 mount option was set. If the superblock feature bit was not set but the mount option was set, then when the first attr2 format inode fork was created, it would set the superblock feature bit. This is as it should be - the superblock feature bit indicated the presence of the attr2 on disk format. The noattr2 mount option, however, did not affect the superblock feature bit. If noattr2 was specified, the on-disk superblock feature bit was ignored and the code always just created attr1 format inode forks. If neither of the attr2 or noattr2 mounts option were specified, then the behaviour was determined by the superblock feature bit. This was all pretty sane. Fast foward 3 years, and we are dealing with fallout from the botched sb_features2 addition and having to deal with feature mismatches between the sb_features2 and sb_bad_features2 fields. The attr2 feature bit was one of these flags. The reconciliation was done well after mount option parsing and, unfortunately, the feature reconciliation had a bug where it ignored the noattr2 mount option. For reasons lost to the mists of time, it was decided that resolving this issue in commit 7c12f296500e ("[XFS] Fix up noattr2 so that it will properly update the versionnum and features2 fields.") required noattr2 to clear the superblock attr2 feature bit. This greatly complicated the attr2 behaviour and broke rules about feature bits needing to be set when those specific features are present in the filesystem. By complicated, I mean that it introduced problems due to feature bit interactions with log recovery. All of the superblock feature bit checks are done prior to log recovery, but if we crash after removing a feature bit, then on the next mount we see the feature bit in the unrecovered superblock, only to have it go away after the log has been replayed. This means our mount time feature processing could be all wrong. Hence you can mount with noattr2, crash shortly afterwards, and mount again without attr2 or noattr2 and still have attr2 enabled because the second mount sees attr2 still enabled in the superblock before recovery runs and removes the feature bit. It's just a mess. Further, this is all legacy code as the v5 format requires attr2 to be enabled at all times and it cannot be disabled. i.e. the noattr2 mount option returns an error when used on v5 format filesystems. To straighten this all out, this patch reverts the attr2/noattr2 mount option behaviour back to the original behaviour. There is no reason for disabling attr2 these days, so we will only do this when the noattr2 mount option is set. This will not remove the superblock feature bit. The superblock bit will provide the default behaviour and only track whether attr2 is present on disk or not. The attr2 mount option will enable the creation of attr2 format inode forks, and if the superblock feature bit is not set it will be added when the first attr2 inode fork is created. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-19 04:46:25 +03:00
xfs_warn(mp, "attr2 and noattr2 cannot both be specified.");
return -EINVAL;
}
xfs: rework attr2 feature and mount options The attr2 feature is somewhat unique in that it has both a superblock feature bit to enable it and mount options to enable and disable it. Back when it was first introduced in 2005, attr2 was disabled unless either the attr2 superblock feature bit was set, or the attr2 mount option was set. If the superblock feature bit was not set but the mount option was set, then when the first attr2 format inode fork was created, it would set the superblock feature bit. This is as it should be - the superblock feature bit indicated the presence of the attr2 on disk format. The noattr2 mount option, however, did not affect the superblock feature bit. If noattr2 was specified, the on-disk superblock feature bit was ignored and the code always just created attr1 format inode forks. If neither of the attr2 or noattr2 mounts option were specified, then the behaviour was determined by the superblock feature bit. This was all pretty sane. Fast foward 3 years, and we are dealing with fallout from the botched sb_features2 addition and having to deal with feature mismatches between the sb_features2 and sb_bad_features2 fields. The attr2 feature bit was one of these flags. The reconciliation was done well after mount option parsing and, unfortunately, the feature reconciliation had a bug where it ignored the noattr2 mount option. For reasons lost to the mists of time, it was decided that resolving this issue in commit 7c12f296500e ("[XFS] Fix up noattr2 so that it will properly update the versionnum and features2 fields.") required noattr2 to clear the superblock attr2 feature bit. This greatly complicated the attr2 behaviour and broke rules about feature bits needing to be set when those specific features are present in the filesystem. By complicated, I mean that it introduced problems due to feature bit interactions with log recovery. All of the superblock feature bit checks are done prior to log recovery, but if we crash after removing a feature bit, then on the next mount we see the feature bit in the unrecovered superblock, only to have it go away after the log has been replayed. This means our mount time feature processing could be all wrong. Hence you can mount with noattr2, crash shortly afterwards, and mount again without attr2 or noattr2 and still have attr2 enabled because the second mount sees attr2 still enabled in the superblock before recovery runs and removes the feature bit. It's just a mess. Further, this is all legacy code as the v5 format requires attr2 to be enabled at all times and it cannot be disabled. i.e. the noattr2 mount option returns an error when used on v5 format filesystems. To straighten this all out, this patch reverts the attr2/noattr2 mount option behaviour back to the original behaviour. There is no reason for disabling attr2 these days, so we will only do this when the noattr2 mount option is set. This will not remove the superblock feature bit. The superblock bit will provide the default behaviour and only track whether attr2 is present on disk or not. The attr2 mount option will enable the creation of attr2 format inode forks, and if the superblock feature bit is not set it will be added when the first attr2 inode fork is created. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-19 04:46:25 +03:00
if (xfs_has_noalign(mp) && (mp->m_dalign || mp->m_swidth)) {
xfs_warn(mp,
"sunit and swidth options incompatible with the noalign option");
return -EINVAL;
}
if (!IS_ENABLED(CONFIG_XFS_QUOTA) && mp->m_qflags != 0) {
xfs_warn(mp, "quota support not available in this kernel.");
return -EINVAL;
}
if ((mp->m_dalign && !mp->m_swidth) ||
(!mp->m_dalign && mp->m_swidth)) {
xfs_warn(mp, "sunit and swidth must be specified together");
return -EINVAL;
}
if (mp->m_dalign && (mp->m_swidth % mp->m_dalign != 0)) {
xfs_warn(mp,
"stripe width (%d) must be a multiple of the stripe unit (%d)",
mp->m_swidth, mp->m_dalign);
return -EINVAL;
}
if (mp->m_logbufs != -1 &&
mp->m_logbufs != 0 &&
(mp->m_logbufs < XLOG_MIN_ICLOGS ||
mp->m_logbufs > XLOG_MAX_ICLOGS)) {
xfs_warn(mp, "invalid logbufs value: %d [not %d-%d]",
mp->m_logbufs, XLOG_MIN_ICLOGS, XLOG_MAX_ICLOGS);
return -EINVAL;
}
if (mp->m_logbsize != -1 &&
mp->m_logbsize != 0 &&
(mp->m_logbsize < XLOG_MIN_RECORD_BSIZE ||
mp->m_logbsize > XLOG_MAX_RECORD_BSIZE ||
!is_power_of_2(mp->m_logbsize))) {
xfs_warn(mp,
"invalid logbufsize: %d [not 16k,32k,64k,128k or 256k]",
mp->m_logbsize);
return -EINVAL;
}
if (xfs_has_allocsize(mp) &&
(mp->m_allocsize_log > XFS_MAX_IO_LOG ||
mp->m_allocsize_log < XFS_MIN_IO_LOG)) {
xfs_warn(mp, "invalid log iosize: %d [not %d-%d]",
mp->m_allocsize_log, XFS_MIN_IO_LOG, XFS_MAX_IO_LOG);
return -EINVAL;
}
return 0;
}
static int
xfs_fs_fill_super(
struct super_block *sb,
struct fs_context *fc)
{
struct xfs_mount *mp = sb->s_fs_info;
struct inode *root;
int flags = 0, error;
mp->m_super = sb;
error = xfs_fs_validate_params(mp);
if (error)
goto out_free_names;
sb_min_blocksize(sb, BBSIZE);
sb->s_xattr = xfs_xattr_handlers;
sb->s_export_op = &xfs_export_operations;
#ifdef CONFIG_XFS_QUOTA
sb->s_qcop = &xfs_quotactl_operations;
sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
#endif
sb->s_op = &xfs_super_operations;
xfs: add mount delay debug option Similar to log_recovery_delay, this delay occurs between the VFS superblock being initialised and the xfs_mount being fully initialised. It also poisons the per-ag radix tree node so that it can be used for triggering shrinker races during mount such as the following: <run memory pressure workload in background> $ cat dirty-mount.sh #! /bin/bash umount -f /dev/pmem0 mkfs.xfs -f /dev/pmem0 mount /dev/pmem0 /mnt/test rm -f /mnt/test/foo xfs_io -fxc "pwrite 0 4k" -c fsync -c "shutdown" /mnt/test/foo umount /dev/pmem0 # let's crash it now! echo 30 > /sys/fs/xfs/debug/mount_delay mount /dev/pmem0 /mnt/test echo 0 > /sys/fs/xfs/debug/mount_delay umount /dev/pmem0 $ sudo ./dirty-mount.sh ..... [ 60.378118] CPU: 3 PID: 3577 Comm: fs_mark Tainted: G D W 4.16.0-rc5-dgc #440 [ 60.378120] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014 [ 60.378124] RIP: 0010:radix_tree_next_chunk+0x76/0x320 [ 60.378127] RSP: 0018:ffffc9000276f4f8 EFLAGS: 00010282 [ 60.383670] RAX: a5a5a5a5a5a5a5a4 RBX: 0000000000000010 RCX: 000000000000001a [ 60.385277] RDX: 0000000000000000 RSI: ffffc9000276f540 RDI: 0000000000000000 [ 60.386554] RBP: 0000000000000000 R08: 0000000000000000 R09: a5a5a5a5a5a5a5a5 [ 60.388194] R10: 0000000000000006 R11: 0000000000000001 R12: ffffc9000276f598 [ 60.389288] R13: 0000000000000040 R14: 0000000000000228 R15: ffff880816cd6458 [ 60.390827] FS: 00007f5c124b9740(0000) GS:ffff88083fc00000(0000) knlGS:0000000000000000 [ 60.392253] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 60.393423] CR2: 00007f5c11bba0b8 CR3: 000000035580e001 CR4: 00000000000606e0 [ 60.394519] Call Trace: [ 60.395252] radix_tree_gang_lookup_tag+0xc4/0x130 [ 60.395948] xfs_perag_get_tag+0x37/0xf0 [ 60.396522] xfs_reclaim_inodes_count+0x32/0x40 [ 60.397178] xfs_fs_nr_cached_objects+0x11/0x20 [ 60.397837] super_cache_count+0x35/0xc0 [ 60.399159] shrink_slab.part.66+0xb1/0x370 [ 60.400194] shrink_node+0x7e/0x1a0 [ 60.401058] try_to_free_pages+0x199/0x470 [ 60.402081] __alloc_pages_slowpath+0x3a1/0xd20 [ 60.403729] __alloc_pages_nodemask+0x1c3/0x200 [ 60.404941] cache_grow_begin+0x20b/0x2e0 [ 60.406164] fallback_alloc+0x160/0x200 [ 60.407088] kmem_cache_alloc+0x111/0x4e0 [ 60.408038] ? xfs_buf_rele+0x61/0x430 [ 60.408925] kmem_zone_alloc+0x61/0xe0 [ 60.409965] xfs_inode_alloc+0x24/0x1d0 ..... Signed-Off-By: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2018-05-11 07:50:23 +03:00
/*
* Delay mount work if the debug hook is set. This is debug
* instrumention to coordinate simulation of xfs mount failures with
* VFS superblock operations
*/
if (xfs_globals.mount_delay) {
xfs_notice(mp, "Delaying mount for %d seconds.",
xfs_globals.mount_delay);
msleep(xfs_globals.mount_delay * 1000);
}
if (fc->sb_flags & SB_SILENT)
flags |= XFS_MFSI_QUIET;
error = xfs_open_devices(mp);
if (error)
goto out_free_names;
error = xfs_init_mount_workqueues(mp);
if (error)
goto out_close_devices;
error = xfs_init_percpu_counters(mp);
if (error)
goto out_destroy_workqueues;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
error = xfs_inodegc_init_percpu(mp);
if (error)
goto out_destroy_counters;
/*
* All percpu data structures requiring cleanup when a cpu goes offline
* must be allocated before adding this @mp to the cpu-dead handler's
* mount list.
*/
xfs_mount_list_add(mp);
/* Allocate stats memory before we do operations that might use it */
mp->m_stats.xs_stats = alloc_percpu(struct xfsstats);
if (!mp->m_stats.xs_stats) {
error = -ENOMEM;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
goto out_destroy_inodegc;
}
error = xfs_readsb(mp, flags);
if (error)
goto out_free_stats;
error = xfs_finish_flags(mp);
if (error)
goto out_free_sb;
error = xfs_setup_devices(mp);
if (error)
goto out_free_sb;
/* V4 support is undergoing deprecation. */
if (!xfs_has_crc(mp)) {
#ifdef CONFIG_XFS_SUPPORT_V4
xfs_warn_once(mp,
"Deprecated V4 format (crc=0) will not be supported after September 2030.");
#else
xfs_warn(mp,
"Deprecated V4 format (crc=0) not supported by kernel.");
error = -EINVAL;
goto out_free_sb;
#endif
}
/* Filesystem claims it needs repair, so refuse the mount. */
if (xfs_has_needsrepair(mp)) {
xfs_warn(mp, "Filesystem needs repair. Please run xfs_repair.");
error = -EFSCORRUPTED;
goto out_free_sb;
}
/*
* Don't touch the filesystem if a user tool thinks it owns the primary
* superblock. mkfs doesn't clear the flag from secondary supers, so
* we don't check them at all.
*/
if (mp->m_sb.sb_inprogress) {
xfs_warn(mp, "Offline file system operation in progress!");
error = -EFSCORRUPTED;
goto out_free_sb;
}
/*
* Until this is fixed only page-sized or smaller data blocks work.
*/
if (mp->m_sb.sb_blocksize > PAGE_SIZE) {
xfs_warn(mp,
"File system with blocksize %d bytes. "
"Only pagesize (%ld) or less will currently work.",
mp->m_sb.sb_blocksize, PAGE_SIZE);
error = -ENOSYS;
goto out_free_sb;
}
/* Ensure this filesystem fits in the page cache limits */
if (xfs_sb_validate_fsb_count(&mp->m_sb, mp->m_sb.sb_dblocks) ||
xfs_sb_validate_fsb_count(&mp->m_sb, mp->m_sb.sb_rblocks)) {
xfs_warn(mp,
"file system too large to be mounted on this system.");
error = -EFBIG;
goto out_free_sb;
}
/*
* XFS block mappings use 54 bits to store the logical block offset.
* This should suffice to handle the maximum file size that the VFS
* supports (currently 2^63 bytes on 64-bit and ULONG_MAX << PAGE_SHIFT
* bytes on 32-bit), but as XFS and VFS have gotten the s_maxbytes
* calculation wrong on 32-bit kernels in the past, we'll add a WARN_ON
* to check this assertion.
*
* Avoid integer overflow by comparing the maximum bmbt offset to the
* maximum pagecache offset in units of fs blocks.
*/
if (!xfs_verify_fileoff(mp, XFS_B_TO_FSBT(mp, MAX_LFS_FILESIZE))) {
xfs_warn(mp,
"MAX_LFS_FILESIZE block offset (%llu) exceeds extent map maximum (%llu)!",
XFS_B_TO_FSBT(mp, MAX_LFS_FILESIZE),
XFS_MAX_FILEOFF);
error = -EINVAL;
goto out_free_sb;
}
error = xfs_filestream_mount(mp);
if (error)
goto out_free_sb;
/*
* we must configure the block size in the superblock before we run the
* full mount process as the mount process can lookup and cache inodes.
*/
sb->s_magic = XFS_SUPER_MAGIC;
sb->s_blocksize = mp->m_sb.sb_blocksize;
sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1;
sb->s_maxbytes = MAX_LFS_FILESIZE;
sb->s_max_links = XFS_MAXLINK;
sb->s_time_gran = 1;
if (xfs_has_bigtime(mp)) {
sb->s_time_min = xfs_bigtime_to_unix(XFS_BIGTIME_TIME_MIN);
sb->s_time_max = xfs_bigtime_to_unix(XFS_BIGTIME_TIME_MAX);
} else {
sb->s_time_min = XFS_LEGACY_TIME_MIN;
sb->s_time_max = XFS_LEGACY_TIME_MAX;
}
trace_xfs_inode_timestamp_range(mp, sb->s_time_min, sb->s_time_max);
sb->s_iflags |= SB_I_CGROUPWB;
set_posix_acl_flag(sb);
/* version 5 superblocks support inode version counters. */
if (xfs_has_crc(mp))
sb->s_flags |= SB_I_VERSION;
if (xfs_has_dax_always(mp)) {
bool rtdev_is_dax = false, datadev_is_dax;
xfs_warn(mp,
"DAX enabled. Warning: EXPERIMENTAL, use at your own risk");
datadev_is_dax = xfs_buftarg_is_dax(sb, mp->m_ddev_targp);
if (mp->m_rtdev_targp)
rtdev_is_dax = xfs_buftarg_is_dax(sb,
mp->m_rtdev_targp);
if (!rtdev_is_dax && !datadev_is_dax) {
xfs_alert(mp,
"DAX unsupported by block device. Turning off DAX.");
xfs_mount_set_dax_mode(mp, XFS_DAX_NEVER);
}
if (xfs_has_reflink(mp)) {
xfs_alert(mp,
"DAX and reflink cannot be used together!");
error = -EINVAL;
goto out_filestream_unmount;
}
}
if (xfs_has_discard(mp)) {
struct request_queue *q = bdev_get_queue(sb->s_bdev);
if (!blk_queue_discard(q)) {
xfs_warn(mp, "mounting with \"discard\" option, but "
"the device does not support discard");
mp->m_features &= ~XFS_FEAT_DISCARD;
}
}
if (xfs_has_reflink(mp)) {
xfs: introduce an always_cow mode Add a mode where XFS never overwrites existing blocks in place. This is to aid debugging our COW code, and also put infatructure in place for things like possible future support for zoned block devices, which can't support overwrites. This mode is enabled globally by doing a: echo 1 > /sys/fs/xfs/debug/always_cow Note that the parameter is global to allow running all tests in xfstests easily in this mode, which would not easily be possible with a per-fs sysfs file. In always_cow mode persistent preallocations are disabled, and fallocate will fail when called with a 0 mode (with our without FALLOC_FL_KEEP_SIZE), and not create unwritten extent for zeroed space when called with FALLOC_FL_ZERO_RANGE or FALLOC_FL_UNSHARE_RANGE. There are a few interesting xfstests failures when run in always_cow mode: - generic/392 fails because the bytes used in the file used to test hole punch recovery are less after the log replay. This is because the blocks written and then punched out are only freed with a delay due to the logging mechanism. - xfs/170 will fail as the already fragile file streams mechanism doesn't seem to interact well with the COW allocator - xfs/180 xfs/182 xfs/192 xfs/198 xfs/204 and xfs/208 will claim the file system is badly fragmented, but there is not much we can do to avoid that when always writing out of place - xfs/205 fails because overwriting a file in always_cow mode will require new space allocation and the assumption in the test thus don't work anymore. - xfs/326 fails to modify the file at all in always_cow mode after injecting the refcount error, leading to an unexpected md5sum after the remount, but that again is expected Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-02-18 20:38:49 +03:00
if (mp->m_sb.sb_rblocks) {
xfs_alert(mp,
"reflink not compatible with realtime device!");
xfs: introduce an always_cow mode Add a mode where XFS never overwrites existing blocks in place. This is to aid debugging our COW code, and also put infatructure in place for things like possible future support for zoned block devices, which can't support overwrites. This mode is enabled globally by doing a: echo 1 > /sys/fs/xfs/debug/always_cow Note that the parameter is global to allow running all tests in xfstests easily in this mode, which would not easily be possible with a per-fs sysfs file. In always_cow mode persistent preallocations are disabled, and fallocate will fail when called with a 0 mode (with our without FALLOC_FL_KEEP_SIZE), and not create unwritten extent for zeroed space when called with FALLOC_FL_ZERO_RANGE or FALLOC_FL_UNSHARE_RANGE. There are a few interesting xfstests failures when run in always_cow mode: - generic/392 fails because the bytes used in the file used to test hole punch recovery are less after the log replay. This is because the blocks written and then punched out are only freed with a delay due to the logging mechanism. - xfs/170 will fail as the already fragile file streams mechanism doesn't seem to interact well with the COW allocator - xfs/180 xfs/182 xfs/192 xfs/198 xfs/204 and xfs/208 will claim the file system is badly fragmented, but there is not much we can do to avoid that when always writing out of place - xfs/205 fails because overwriting a file in always_cow mode will require new space allocation and the assumption in the test thus don't work anymore. - xfs/326 fails to modify the file at all in always_cow mode after injecting the refcount error, leading to an unexpected md5sum after the remount, but that again is expected Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2019-02-18 20:38:49 +03:00
error = -EINVAL;
goto out_filestream_unmount;
}
if (xfs_globals.always_cow) {
xfs_info(mp, "using DEBUG-only always_cow mode.");
mp->m_always_cow = true;
}
}
if (xfs_has_rmapbt(mp) && mp->m_sb.sb_rblocks) {
xfs_alert(mp,
"reverse mapping btree not compatible with realtime device!");
error = -EINVAL;
goto out_filestream_unmount;
}
error = xfs_mountfs(mp);
if (error)
goto out_filestream_unmount;
root = igrab(VFS_I(mp->m_rootip));
if (!root) {
error = -ENOENT;
goto out_unmount;
}
sb->s_root = d_make_root(root);
if (!sb->s_root) {
error = -ENOMEM;
goto out_unmount;
}
return 0;
out_filestream_unmount:
xfs_filestream_unmount(mp);
out_free_sb:
xfs_freesb(mp);
out_free_stats:
free_percpu(mp->m_stats.xs_stats);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
out_destroy_inodegc:
xfs_mount_list_del(mp);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_free_percpu(mp);
out_destroy_counters:
xfs_destroy_percpu_counters(mp);
out_destroy_workqueues:
xfs_destroy_mount_workqueues(mp);
out_close_devices:
xfs_close_devices(mp);
out_free_names:
sb->s_fs_info = NULL;
xfs_mount_free(mp);
return error;
out_unmount:
xfs_filestream_unmount(mp);
xfs_unmountfs(mp);
goto out_free_sb;
}
static int
xfs_fs_get_tree(
struct fs_context *fc)
{
return get_tree_bdev(fc, xfs_fs_fill_super);
}
static int
xfs_remount_rw(
struct xfs_mount *mp)
{
struct xfs_sb *sbp = &mp->m_sb;
int error;
if (xfs_has_norecovery(mp)) {
xfs_warn(mp,
"ro->rw transition prohibited on norecovery mount");
return -EINVAL;
}
if (xfs_sb_is_v5(sbp) &&
xfs_sb_has_ro_compat_feature(sbp, XFS_SB_FEAT_RO_COMPAT_UNKNOWN)) {
xfs_warn(mp,
"ro->rw transition prohibited on unknown (0x%x) ro-compat filesystem",
(sbp->sb_features_ro_compat &
XFS_SB_FEAT_RO_COMPAT_UNKNOWN));
return -EINVAL;
}
clear_bit(XFS_OPSTATE_READONLY, &mp->m_opstate);
/*
* If this is the first remount to writeable state we might have some
* superblock changes to update.
*/
if (mp->m_update_sb) {
error = xfs_sync_sb(mp, false);
if (error) {
xfs_warn(mp, "failed to write sb changes");
return error;
}
mp->m_update_sb = false;
}
/*
* Fill out the reserve pool if it is empty. Use the stashed value if
* it is non-zero, otherwise go with the default.
*/
xfs_restore_resvblks(mp);
xfs_log_work_queue(mp);
/* Recover any CoW blocks that never got remapped. */
error = xfs_reflink_recover_cow(mp);
if (error) {
xfs_err(mp,
"Error %d recovering leftover CoW allocations.", error);
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
return error;
}
xfs_blockgc_start(mp);
/* Create the per-AG metadata reservation pool .*/
error = xfs_fs_reserve_ag_blocks(mp);
if (error && error != -ENOSPC)
return error;
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
/* Re-enable the background inode inactivation worker. */
xfs_inodegc_start(mp);
return 0;
}
static int
xfs_remount_ro(
struct xfs_mount *mp)
{
int error;
/*
* Cancel background eofb scanning so it cannot race with the final
* log force+buftarg wait and deadlock the remount.
*/
xfs_blockgc_stop(mp);
/* Get rid of any leftover CoW reservations... */
error = xfs_blockgc_free_space(mp, NULL);
if (error) {
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
return error;
}
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
/*
* Stop the inodegc background worker. xfs_fs_reconfigure already
* flushed all pending inodegc work when it sync'd the filesystem.
* The VFS holds s_umount, so we know that inodes cannot enter
* xfs_fs_destroy_inode during a remount operation. In readonly mode
* we send inodes straight to reclaim, so no inodes will be queued.
*/
xfs_inodegc_stop(mp);
/* Free the per-AG metadata reservation pool. */
error = xfs_fs_unreserve_ag_blocks(mp);
if (error) {
xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
return error;
}
/*
* Before we sync the metadata, we need to free up the reserve block
* pool so that the used block count in the superblock on disk is
* correct at the end of the remount. Stash the current* reserve pool
* size so that if we get remounted rw, we can return it to the same
* size.
*/
xfs_save_resvblks(mp);
xfs_log_clean(mp);
set_bit(XFS_OPSTATE_READONLY, &mp->m_opstate);
return 0;
}
/*
* Logically we would return an error here to prevent users from believing
* they might have changed mount options using remount which can't be changed.
*
* But unfortunately mount(8) adds all options from mtab and fstab to the mount
* arguments in some cases so we can't blindly reject options, but have to
* check for each specified option if it actually differs from the currently
* set option and only reject it if that's the case.
*
* Until that is implemented we return success for every remount request, and
* silently ignore all options that we can't actually change.
*/
static int
xfs_fs_reconfigure(
struct fs_context *fc)
{
struct xfs_mount *mp = XFS_M(fc->root->d_sb);
struct xfs_mount *new_mp = fc->s_fs_info;
int flags = fc->sb_flags;
int error;
/* version 5 superblocks always support version counters. */
if (xfs_has_crc(mp))
fc->sb_flags |= SB_I_VERSION;
error = xfs_fs_validate_params(new_mp);
if (error)
return error;
sync_filesystem(mp->m_super);
/* inode32 -> inode64 */
if (xfs_has_small_inums(mp) && !xfs_has_small_inums(new_mp)) {
mp->m_features &= ~XFS_FEAT_SMALL_INUMS;
mp->m_maxagi = xfs_set_inode_alloc(mp, mp->m_sb.sb_agcount);
}
/* inode64 -> inode32 */
if (!xfs_has_small_inums(mp) && xfs_has_small_inums(new_mp)) {
mp->m_features |= XFS_FEAT_SMALL_INUMS;
mp->m_maxagi = xfs_set_inode_alloc(mp, mp->m_sb.sb_agcount);
}
/* ro -> rw */
if (xfs_is_readonly(mp) && !(flags & SB_RDONLY)) {
error = xfs_remount_rw(mp);
if (error)
return error;
}
/* rw -> ro */
if (!xfs_is_readonly(mp) && (flags & SB_RDONLY)) {
error = xfs_remount_ro(mp);
if (error)
return error;
}
return 0;
}
static void xfs_fs_free(
struct fs_context *fc)
{
struct xfs_mount *mp = fc->s_fs_info;
/*
* mp is stored in the fs_context when it is initialized.
* mp is transferred to the superblock on a successful mount,
* but if an error occurs before the transfer we have to free
* it here.
*/
if (mp)
xfs_mount_free(mp);
}
static const struct fs_context_operations xfs_context_ops = {
.parse_param = xfs_fs_parse_param,
.get_tree = xfs_fs_get_tree,
.reconfigure = xfs_fs_reconfigure,
.free = xfs_fs_free,
};
static int xfs_init_fs_context(
struct fs_context *fc)
{
struct xfs_mount *mp;
mp = kmem_alloc(sizeof(struct xfs_mount), KM_ZERO);
if (!mp)
return -ENOMEM;
spin_lock_init(&mp->m_sb_lock);
spin_lock_init(&mp->m_agirotor_lock);
INIT_RADIX_TREE(&mp->m_perag_tree, GFP_ATOMIC);
spin_lock_init(&mp->m_perag_lock);
mutex_init(&mp->m_growlock);
INIT_WORK(&mp->m_flush_inodes_work, xfs_flush_inodes_worker);
INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
mp->m_kobj.kobject.kset = xfs_kset;
/*
* We don't create the finobt per-ag space reservation until after log
* recovery, so we must set this to true so that an ifree transaction
* started during log recovery will not depend on space reservations
* for finobt expansion.
*/
mp->m_finobt_nores = true;
/*
* These can be overridden by the mount option parsing.
*/
mp->m_logbufs = -1;
mp->m_logbsize = -1;
mp->m_allocsize_log = 16; /* 64k */
/*
* Copy binary VFS mount flags we are interested in.
*/
if (fc->sb_flags & SB_RDONLY)
set_bit(XFS_OPSTATE_READONLY, &mp->m_opstate);
if (fc->sb_flags & SB_DIRSYNC)
mp->m_features |= XFS_FEAT_DIRSYNC;
if (fc->sb_flags & SB_SYNCHRONOUS)
mp->m_features |= XFS_FEAT_WSYNC;
fc->s_fs_info = mp;
fc->ops = &xfs_context_ops;
return 0;
}
static struct file_system_type xfs_fs_type = {
.owner = THIS_MODULE,
.name = "xfs",
.init_fs_context = xfs_init_fs_context,
.parameters = xfs_fs_parameters,
.kill_sb = kill_block_super,
.fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
};
fs: Limit sys_mount to only request filesystem modules. Modify the request_module to prefix the file system type with "fs-" and add aliases to all of the filesystems that can be built as modules to match. A common practice is to build all of the kernel code and leave code that is not commonly needed as modules, with the result that many users are exposed to any bug anywhere in the kernel. Looking for filesystems with a fs- prefix limits the pool of possible modules that can be loaded by mount to just filesystems trivially making things safer with no real cost. Using aliases means user space can control the policy of which filesystem modules are auto-loaded by editing /etc/modprobe.d/*.conf with blacklist and alias directives. Allowing simple, safe, well understood work-arounds to known problematic software. This also addresses a rare but unfortunate problem where the filesystem name is not the same as it's module name and module auto-loading would not work. While writing this patch I saw a handful of such cases. The most significant being autofs that lives in the module autofs4. This is relevant to user namespaces because we can reach the request module in get_fs_type() without having any special permissions, and people get uncomfortable when a user specified string (in this case the filesystem type) goes all of the way to request_module. After having looked at this issue I don't think there is any particular reason to perform any filtering or permission checks beyond making it clear in the module request that we want a filesystem module. The common pattern in the kernel is to call request_module() without regards to the users permissions. In general all a filesystem module does once loaded is call register_filesystem() and go to sleep. Which means there is not much attack surface exposed by loading a filesytem module unless the filesystem is mounted. In a user namespace filesystems are not mounted unless .fs_flags = FS_USERNS_MOUNT, which most filesystems do not set today. Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Acked-by: Kees Cook <keescook@chromium.org> Reported-by: Kees Cook <keescook@google.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
2013-03-03 07:39:14 +04:00
MODULE_ALIAS_FS("xfs");
STATIC int __init
xfs_init_zones(void)
{
xfs_log_ticket_zone = kmem_cache_create("xfs_log_ticket",
sizeof(struct xlog_ticket),
0, 0, NULL);
if (!xfs_log_ticket_zone)
goto out;
xfs_bmap_free_item_zone = kmem_cache_create("xfs_bmap_free_item",
sizeof(struct xfs_extent_free_item),
0, 0, NULL);
if (!xfs_bmap_free_item_zone)
goto out_destroy_log_ticket_zone;
xfs_btree_cur_zone = kmem_cache_create("xfs_btree_cur",
sizeof(struct xfs_btree_cur),
0, 0, NULL);
if (!xfs_btree_cur_zone)
goto out_destroy_bmap_free_item_zone;
xfs_da_state_zone = kmem_cache_create("xfs_da_state",
sizeof(struct xfs_da_state),
0, 0, NULL);
if (!xfs_da_state_zone)
goto out_destroy_btree_cur_zone;
xfs_ifork_zone = kmem_cache_create("xfs_ifork",
sizeof(struct xfs_ifork),
0, 0, NULL);
if (!xfs_ifork_zone)
goto out_destroy_da_state_zone;
xfs_trans_zone = kmem_cache_create("xfs_trans",
sizeof(struct xfs_trans),
0, 0, NULL);
if (!xfs_trans_zone)
goto out_destroy_ifork_zone;
/*
* The size of the zone allocated buf log item is the maximum
* size possible under XFS. This wastes a little bit of memory,
* but it is much faster.
*/
xfs_buf_item_zone = kmem_cache_create("xfs_buf_item",
sizeof(struct xfs_buf_log_item),
0, 0, NULL);
if (!xfs_buf_item_zone)
goto out_destroy_trans_zone;
xfs_efd_zone = kmem_cache_create("xfs_efd_item",
(sizeof(struct xfs_efd_log_item) +
(XFS_EFD_MAX_FAST_EXTENTS - 1) *
sizeof(struct xfs_extent)),
0, 0, NULL);
if (!xfs_efd_zone)
goto out_destroy_buf_item_zone;
xfs_efi_zone = kmem_cache_create("xfs_efi_item",
(sizeof(struct xfs_efi_log_item) +
(XFS_EFI_MAX_FAST_EXTENTS - 1) *
sizeof(struct xfs_extent)),
0, 0, NULL);
if (!xfs_efi_zone)
goto out_destroy_efd_zone;
xfs_inode_zone = kmem_cache_create("xfs_inode",
sizeof(struct xfs_inode), 0,
(SLAB_HWCACHE_ALIGN |
SLAB_RECLAIM_ACCOUNT |
SLAB_MEM_SPREAD | SLAB_ACCOUNT),
xfs_fs_inode_init_once);
if (!xfs_inode_zone)
goto out_destroy_efi_zone;
xfs_ili_zone = kmem_cache_create("xfs_ili",
sizeof(struct xfs_inode_log_item), 0,
SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
NULL);
if (!xfs_ili_zone)
goto out_destroy_inode_zone;
xfs_icreate_zone = kmem_cache_create("xfs_icr",
sizeof(struct xfs_icreate_item),
0, 0, NULL);
if (!xfs_icreate_zone)
goto out_destroy_ili_zone;
xfs_rud_zone = kmem_cache_create("xfs_rud_item",
sizeof(struct xfs_rud_log_item),
0, 0, NULL);
if (!xfs_rud_zone)
goto out_destroy_icreate_zone;
xfs_rui_zone = kmem_cache_create("xfs_rui_item",
xfs_rui_log_item_sizeof(XFS_RUI_MAX_FAST_EXTENTS),
0, 0, NULL);
if (!xfs_rui_zone)
goto out_destroy_rud_zone;
xfs_cud_zone = kmem_cache_create("xfs_cud_item",
sizeof(struct xfs_cud_log_item),
0, 0, NULL);
if (!xfs_cud_zone)
goto out_destroy_rui_zone;
xfs_cui_zone = kmem_cache_create("xfs_cui_item",
xfs_cui_log_item_sizeof(XFS_CUI_MAX_FAST_EXTENTS),
0, 0, NULL);
if (!xfs_cui_zone)
goto out_destroy_cud_zone;
xfs_bud_zone = kmem_cache_create("xfs_bud_item",
sizeof(struct xfs_bud_log_item),
0, 0, NULL);
if (!xfs_bud_zone)
goto out_destroy_cui_zone;
xfs_bui_zone = kmem_cache_create("xfs_bui_item",
xfs_bui_log_item_sizeof(XFS_BUI_MAX_FAST_EXTENTS),
0, 0, NULL);
if (!xfs_bui_zone)
goto out_destroy_bud_zone;
return 0;
out_destroy_bud_zone:
kmem_cache_destroy(xfs_bud_zone);
out_destroy_cui_zone:
kmem_cache_destroy(xfs_cui_zone);
out_destroy_cud_zone:
kmem_cache_destroy(xfs_cud_zone);
out_destroy_rui_zone:
kmem_cache_destroy(xfs_rui_zone);
out_destroy_rud_zone:
kmem_cache_destroy(xfs_rud_zone);
out_destroy_icreate_zone:
kmem_cache_destroy(xfs_icreate_zone);
out_destroy_ili_zone:
kmem_cache_destroy(xfs_ili_zone);
out_destroy_inode_zone:
kmem_cache_destroy(xfs_inode_zone);
out_destroy_efi_zone:
kmem_cache_destroy(xfs_efi_zone);
out_destroy_efd_zone:
kmem_cache_destroy(xfs_efd_zone);
out_destroy_buf_item_zone:
kmem_cache_destroy(xfs_buf_item_zone);
out_destroy_trans_zone:
kmem_cache_destroy(xfs_trans_zone);
out_destroy_ifork_zone:
kmem_cache_destroy(xfs_ifork_zone);
out_destroy_da_state_zone:
kmem_cache_destroy(xfs_da_state_zone);
out_destroy_btree_cur_zone:
kmem_cache_destroy(xfs_btree_cur_zone);
out_destroy_bmap_free_item_zone:
kmem_cache_destroy(xfs_bmap_free_item_zone);
out_destroy_log_ticket_zone:
kmem_cache_destroy(xfs_log_ticket_zone);
out:
return -ENOMEM;
}
STATIC void
xfs_destroy_zones(void)
{
/*
* Make sure all delayed rcu free are flushed before we
* destroy caches.
*/
rcu_barrier();
kmem_cache_destroy(xfs_bui_zone);
kmem_cache_destroy(xfs_bud_zone);
kmem_cache_destroy(xfs_cui_zone);
kmem_cache_destroy(xfs_cud_zone);
kmem_cache_destroy(xfs_rui_zone);
kmem_cache_destroy(xfs_rud_zone);
kmem_cache_destroy(xfs_icreate_zone);
kmem_cache_destroy(xfs_ili_zone);
kmem_cache_destroy(xfs_inode_zone);
kmem_cache_destroy(xfs_efi_zone);
kmem_cache_destroy(xfs_efd_zone);
kmem_cache_destroy(xfs_buf_item_zone);
kmem_cache_destroy(xfs_trans_zone);
kmem_cache_destroy(xfs_ifork_zone);
kmem_cache_destroy(xfs_da_state_zone);
kmem_cache_destroy(xfs_btree_cur_zone);
kmem_cache_destroy(xfs_bmap_free_item_zone);
kmem_cache_destroy(xfs_log_ticket_zone);
}
STATIC int __init
xfs_init_workqueues(void)
{
/*
* The allocation workqueue can be used in memory reclaim situations
* (writepage path), and parallelism is only limited by the number of
* AGs in all the filesystems mounted. Hence use the default large
* max_active value for this workqueue.
*/
xfs_alloc_wq = alloc_workqueue("xfsalloc",
XFS_WQFLAGS(WQ_MEM_RECLAIM | WQ_FREEZABLE), 0);
if (!xfs_alloc_wq)
return -ENOMEM;
xfs_discard_wq = alloc_workqueue("xfsdiscard", XFS_WQFLAGS(WQ_UNBOUND),
0);
if (!xfs_discard_wq)
goto out_free_alloc_wq;
return 0;
out_free_alloc_wq:
destroy_workqueue(xfs_alloc_wq);
return -ENOMEM;
}
STATIC void
xfs_destroy_workqueues(void)
{
destroy_workqueue(xfs_discard_wq);
destroy_workqueue(xfs_alloc_wq);
}
#ifdef CONFIG_HOTPLUG_CPU
static int
xfs_cpu_dead(
unsigned int cpu)
{
struct xfs_mount *mp, *n;
spin_lock(&xfs_mount_list_lock);
list_for_each_entry_safe(mp, n, &xfs_mount_list, m_mount_list) {
spin_unlock(&xfs_mount_list_lock);
xfs: per-cpu deferred inode inactivation queues Move inode inactivation to background work contexts so that it no longer runs in the context that releases the final reference to an inode. This will allow process work that ends up blocking on inactivation to continue doing work while the filesytem processes the inactivation in the background. A typical demonstration of this is unlinking an inode with lots of extents. The extents are removed during inactivation, so this blocks the process that unlinked the inode from the directory structure. By moving the inactivation to the background process, the userspace applicaiton can keep working (e.g. unlinking the next inode in the directory) while the inactivation work on the previous inode is done by a different CPU. The implementation of the queue is relatively simple. We use a per-cpu lockless linked list (llist) to queue inodes for inactivation without requiring serialisation mechanisms, and a work item to allow the queue to be processed by a CPU bound worker thread. We also keep a count of the queue depth so that we can trigger work after a number of deferred inactivations have been queued. The use of a bound workqueue with a single work depth allows the workqueue to run one work item per CPU. We queue the work item on the CPU we are currently running on, and so this essentially gives us affine per-cpu worker threads for the per-cpu queues. THis maintains the effective CPU affinity that occurs within XFS at the AG level due to all objects in a directory being local to an AG. Hence inactivation work tends to run on the same CPU that last accessed all the objects that inactivation accesses and this maintains hot CPU caches for unlink workloads. A depth of 32 inodes was chosen to match the number of inodes in an inode cluster buffer. This hopefully allows sequential allocation/unlink behaviours to defering inactivation of all the inodes in a single cluster buffer at a time, further helping maintain hot CPU and buffer cache accesses while running inactivations. A hard per-cpu queue throttle of 256 inode has been set to avoid runaway queuing when inodes that take a long to time inactivate are being processed. For example, when unlinking inodes with large numbers of extents that can take a lot of processing to free. Signed-off-by: Dave Chinner <dchinner@redhat.com> [djwong: tweak comments and tracepoints, convert opflags to state bits] Reviewed-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Darrick J. Wong <djwong@kernel.org>
2021-08-06 21:05:39 +03:00
xfs_inodegc_cpu_dead(mp, cpu);
spin_lock(&xfs_mount_list_lock);
}
spin_unlock(&xfs_mount_list_lock);
return 0;
}
static int __init
xfs_cpu_hotplug_init(void)
{
int error;
error = cpuhp_setup_state_nocalls(CPUHP_XFS_DEAD, "xfs:dead", NULL,
xfs_cpu_dead);
if (error < 0)
xfs_alert(NULL,
"Failed to initialise CPU hotplug, error %d. XFS is non-functional.",
error);
return error;
}
static void
xfs_cpu_hotplug_destroy(void)
{
cpuhp_remove_state_nocalls(CPUHP_XFS_DEAD);
}
#else /* !CONFIG_HOTPLUG_CPU */
static inline int xfs_cpu_hotplug_init(void) { return 0; }
static inline void xfs_cpu_hotplug_destroy(void) {}
#endif
STATIC int __init
init_xfs_fs(void)
{
int error;
xfs_check_ondisk_structs();
printk(KERN_INFO XFS_VERSION_STRING " with "
XFS_BUILD_OPTIONS " enabled\n");
xfs_dir_startup();
error = xfs_cpu_hotplug_init();
if (error)
goto out;
error = xfs_init_zones();
if (error)
goto out_destroy_hp;
error = xfs_init_workqueues();
if (error)
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-15 02:14:59 +03:00
goto out_destroy_zones;
error = xfs_mru_cache_init();
if (error)
goto out_destroy_wq;
error = xfs_buf_init();
if (error)
goto out_mru_cache_uninit;
error = xfs_init_procfs();
if (error)
goto out_buf_terminate;
error = xfs_sysctl_register();
if (error)
goto out_cleanup_procfs;
xfs_kset = kset_create_and_add("xfs", NULL, fs_kobj);
if (!xfs_kset) {
error = -ENOMEM;
goto out_sysctl_unregister;
}
xfsstats.xs_kobj.kobject.kset = xfs_kset;
xfsstats.xs_stats = alloc_percpu(struct xfsstats);
if (!xfsstats.xs_stats) {
error = -ENOMEM;
goto out_kset_unregister;
}
error = xfs_sysfs_init(&xfsstats.xs_kobj, &xfs_stats_ktype, NULL,
"stats");
if (error)
goto out_free_stats;
#ifdef DEBUG
xfs_dbg_kobj.kobject.kset = xfs_kset;
error = xfs_sysfs_init(&xfs_dbg_kobj, &xfs_dbg_ktype, NULL, "debug");
if (error)
goto out_remove_stats_kobj;
#endif
error = xfs_qm_init();
if (error)
goto out_remove_dbg_kobj;
error = register_filesystem(&xfs_fs_type);
if (error)
goto out_qm_exit;
return 0;
out_qm_exit:
xfs_qm_exit();
out_remove_dbg_kobj:
#ifdef DEBUG
xfs_sysfs_del(&xfs_dbg_kobj);
out_remove_stats_kobj:
#endif
xfs_sysfs_del(&xfsstats.xs_kobj);
out_free_stats:
free_percpu(xfsstats.xs_stats);
out_kset_unregister:
kset_unregister(xfs_kset);
out_sysctl_unregister:
xfs_sysctl_unregister();
out_cleanup_procfs:
xfs_cleanup_procfs();
out_buf_terminate:
xfs_buf_terminate();
out_mru_cache_uninit:
xfs_mru_cache_uninit();
out_destroy_wq:
xfs_destroy_workqueues();
out_destroy_zones:
xfs_destroy_zones();
out_destroy_hp:
xfs_cpu_hotplug_destroy();
out:
return error;
}
STATIC void __exit
exit_xfs_fs(void)
{
xfs_qm_exit();
unregister_filesystem(&xfs_fs_type);
#ifdef DEBUG
xfs_sysfs_del(&xfs_dbg_kobj);
#endif
xfs_sysfs_del(&xfsstats.xs_kobj);
free_percpu(xfsstats.xs_stats);
kset_unregister(xfs_kset);
xfs_sysctl_unregister();
xfs_cleanup_procfs();
xfs_buf_terminate();
xfs_mru_cache_uninit();
xfs_destroy_workqueues();
xfs_destroy_zones();
xfs_uuid_table_free();
xfs_cpu_hotplug_destroy();
}
module_init(init_xfs_fs);
module_exit(exit_xfs_fs);
MODULE_AUTHOR("Silicon Graphics, Inc.");
MODULE_DESCRIPTION(XFS_VERSION_STRING " with " XFS_BUILD_OPTIONS " enabled");
MODULE_LICENSE("GPL");