linux-stable/fs/gfs2/quota.c

1774 lines
42 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
* Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
*/
/*
* Quota change tags are associated with each transaction that allocates or
* deallocates space. Those changes are accumulated locally to each node (in a
* per-node file) and then are periodically synced to the quota file. This
* avoids the bottleneck of constantly touching the quota file, but introduces
* fuzziness in the current usage value of IDs that are being used on different
* nodes in the cluster simultaneously. So, it is possible for a user on
* multiple nodes to overrun their quota, but that overrun is controlable.
* Since quota tags are part of transactions, there is no need for a quota check
* program to be run on node crashes or anything like that.
*
* There are couple of knobs that let the administrator manage the quota
* fuzziness. "quota_quantum" sets the maximum time a quota change can be
* sitting on one node before being synced to the quota file. (The default is
* 60 seconds.) Another knob, "quota_scale" controls how quickly the frequency
* of quota file syncs increases as the user moves closer to their limit. The
* more frequent the syncs, the more accurate the quota enforcement, but that
* means that there is more contention between the nodes for the quota file.
* The default value is one. This sets the maximum theoretical quota overrun
* (with infinite node with infinite bandwidth) to twice the user's limit. (In
* practice, the maximum overrun you see should be much less.) A "quota_scale"
* number greater than one makes quota syncs more frequent and reduces the
* maximum overrun. Numbers less than one (but greater than zero) make quota
* syncs less frequent.
*
* GFS quotas also use per-ID Lock Value Blocks (LVBs) to cache the contents of
* the quota file, so it is not being constantly read.
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include <linux/spinlock.h>
#include <linux/completion.h>
#include <linux/buffer_head.h>
#include <linux/sort.h>
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
#include <linux/fs.h>
#include <linux/bio.h>
#include <linux/gfs2_ondisk.h>
#include <linux/kthread.h>
#include <linux/freezer.h>
#include <linux/quota.h>
#include <linux/dqblk_xfs.h>
#include <linux/lockref.h>
#include <linux/list_lru.h>
#include <linux/rcupdate.h>
#include <linux/rculist_bl.h>
#include <linux/bit_spinlock.h>
#include <linux/jhash.h>
#include <linux/vmalloc.h>
#include "gfs2.h"
#include "incore.h"
#include "bmap.h"
#include "glock.h"
#include "glops.h"
#include "log.h"
#include "meta_io.h"
#include "quota.h"
#include "rgrp.h"
#include "super.h"
#include "trans.h"
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
#include "inode.h"
#include "util.h"
#define GFS2_QD_HASH_SHIFT 12
#define GFS2_QD_HASH_SIZE BIT(GFS2_QD_HASH_SHIFT)
#define GFS2_QD_HASH_MASK (GFS2_QD_HASH_SIZE - 1)
gfs2: Add quota_change type Function do_qc has two main uses: (1) to re-sync the local quota changes (qd) to the master quotas, and (2) normal quota changes. In the case of normal quota changes, the change can be positive or negative, as the quota usage goes up and down. Before this patch function do_qc was distinguishing one from another by whether the resulting value is or isn't zero: In the case of a re-sync (called do_sync) the quota value is moved from the temporary value to a master value, so the amount is added to one and subtracted from the other. The problem is that since the values can be positive or negative we can occasionally run into situations where we are not doing a re-sync but the quota change just happens to cancel out the previous value. In the case of a re-sync extra references and locks are taken, and so do_qc needs to release them. In the case of a normal quota change, no extra references and locks are taken, so it must not try to release them. The problem is: if the quota change is not a re-sync but the value just happens to cancel out the original quota change, the resulting zero value fools do_qc into thinking this is a re-sync and therefore it must release the extra references. This results in problems, mainly having to do with slot reference numbers going smaller than zero. This patch introduces new constants, QC_SYNC and QC_CHANGE so do_qc can really tell the difference. For QC_SYNC calls it must release the extra references acquired by gfs2_quota_unlock's call to qd_check_sync. For QC_CHANGE calls it does not have extra references to put. Note that this allows quota changes back to a value of zero, and so I removed an assert warning related to that. Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2023-06-28 14:54:41 +00:00
#define QC_CHANGE 0
#define QC_SYNC 1
/* Lock order: qd_lock -> bucket lock -> qd->lockref.lock -> lru lock */
/* -> sd_bitmap_lock */
static DEFINE_SPINLOCK(qd_lock);
struct list_lru gfs2_qd_lru;
static struct hlist_bl_head qd_hash_table[GFS2_QD_HASH_SIZE];
static unsigned int gfs2_qd_hash(const struct gfs2_sbd *sdp,
const struct kqid qid)
{
unsigned int h;
h = jhash(&sdp, sizeof(struct gfs2_sbd *), 0);
h = jhash(&qid, sizeof(struct kqid), h);
return h & GFS2_QD_HASH_MASK;
}
static inline void spin_lock_bucket(unsigned int hash)
{
hlist_bl_lock(&qd_hash_table[hash]);
}
static inline void spin_unlock_bucket(unsigned int hash)
{
hlist_bl_unlock(&qd_hash_table[hash]);
}
static void gfs2_qd_dealloc(struct rcu_head *rcu)
{
struct gfs2_quota_data *qd = container_of(rcu, struct gfs2_quota_data, qd_rcu);
struct gfs2_sbd *sdp = qd->qd_sbd;
kmem_cache_free(gfs2_quotad_cachep, qd);
if (atomic_dec_and_test(&sdp->sd_quota_count))
wake_up(&sdp->sd_kill_wait);
}
static void gfs2_qd_dispose(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
spin_lock(&qd_lock);
list_del(&qd->qd_list);
spin_unlock(&qd_lock);
spin_lock_bucket(qd->qd_hash);
hlist_bl_del_rcu(&qd->qd_hlist);
spin_unlock_bucket(qd->qd_hash);
if (!gfs2_withdrawn(sdp)) {
gfs2_assert_warn(sdp, !qd->qd_change);
gfs2_assert_warn(sdp, !qd->qd_slot_count);
gfs2_assert_warn(sdp, !qd->qd_bh_count);
}
gfs2_glock_put(qd->qd_gl);
call_rcu(&qd->qd_rcu, gfs2_qd_dealloc);
}
static void gfs2_qd_list_dispose(struct list_head *list)
{
struct gfs2_quota_data *qd;
while (!list_empty(list)) {
qd = list_first_entry(list, struct gfs2_quota_data, qd_lru);
list_del(&qd->qd_lru);
gfs2_qd_dispose(qd);
}
}
list_lru: add helpers to isolate items Currently, the isolate callback passed to the list_lru_walk family of functions is supposed to just delete an item from the list upon returning LRU_REMOVED or LRU_REMOVED_RETRY, while nr_items counter is fixed by __list_lru_walk_one after the callback returns. Since the callback is allowed to drop the lock after removing an item (it has to return LRU_REMOVED_RETRY then), the nr_items can be less than the actual number of elements on the list even if we check them under the lock. This makes it difficult to move items from one list_lru_one to another, which is required for per-memcg list_lru reparenting - we can't just splice the lists, we have to move entries one by one. This patch therefore introduces helpers that must be used by callback functions to isolate items instead of raw list_del/list_move. These are list_lru_isolate and list_lru_isolate_move. They not only remove the entry from the list, but also fix the nr_items counter, making sure nr_items always reflects the actual number of elements on the list if checked under the appropriate lock. Signed-off-by: Vladimir Davydov <vdavydov@parallels.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Tejun Heo <tj@kernel.org> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-02-12 22:59:35 +00:00
static enum lru_status gfs2_qd_isolate(struct list_head *item,
struct list_lru_one *lru, spinlock_t *lru_lock, void *arg)
{
struct list_head *dispose = arg;
struct gfs2_quota_data *qd =
list_entry(item, struct gfs2_quota_data, qd_lru);
enum lru_status status;
if (!spin_trylock(&qd->qd_lockref.lock))
return LRU_SKIP;
status = LRU_SKIP;
if (qd->qd_lockref.count == 0) {
lockref_mark_dead(&qd->qd_lockref);
list_lru: add helpers to isolate items Currently, the isolate callback passed to the list_lru_walk family of functions is supposed to just delete an item from the list upon returning LRU_REMOVED or LRU_REMOVED_RETRY, while nr_items counter is fixed by __list_lru_walk_one after the callback returns. Since the callback is allowed to drop the lock after removing an item (it has to return LRU_REMOVED_RETRY then), the nr_items can be less than the actual number of elements on the list even if we check them under the lock. This makes it difficult to move items from one list_lru_one to another, which is required for per-memcg list_lru reparenting - we can't just splice the lists, we have to move entries one by one. This patch therefore introduces helpers that must be used by callback functions to isolate items instead of raw list_del/list_move. These are list_lru_isolate and list_lru_isolate_move. They not only remove the entry from the list, but also fix the nr_items counter, making sure nr_items always reflects the actual number of elements on the list if checked under the appropriate lock. Signed-off-by: Vladimir Davydov <vdavydov@parallels.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Tejun Heo <tj@kernel.org> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Dave Chinner <david@fromorbit.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-02-12 22:59:35 +00:00
list_lru_isolate_move(lru, &qd->qd_lru, dispose);
status = LRU_REMOVED;
}
spin_unlock(&qd->qd_lockref.lock);
return status;
}
static unsigned long gfs2_qd_shrink_scan(struct shrinker *shrink,
struct shrink_control *sc)
{
LIST_HEAD(dispose);
unsigned long freed;
if (!(sc->gfp_mask & __GFP_FS))
return SHRINK_STOP;
list_lru: introduce list_lru_shrink_{count,walk} Kmem accounting of memcg is unusable now, because it lacks slab shrinker support. That means when we hit the limit we will get ENOMEM w/o any chance to recover. What we should do then is to call shrink_slab, which would reclaim old inode/dentry caches from this cgroup. This is what this patch set is intended to do. Basically, it does two things. First, it introduces the notion of per-memcg slab shrinker. A shrinker that wants to reclaim objects per cgroup should mark itself as SHRINKER_MEMCG_AWARE. Then it will be passed the memory cgroup to scan from in shrink_control->memcg. For such shrinkers shrink_slab iterates over the whole cgroup subtree under the target cgroup and calls the shrinker for each kmem-active memory cgroup. Secondly, this patch set makes the list_lru structure per-memcg. It's done transparently to list_lru users - everything they have to do is to tell list_lru_init that they want memcg-aware list_lru. Then the list_lru will automatically distribute objects among per-memcg lists basing on which cgroup the object is accounted to. This way to make FS shrinkers (icache, dcache) memcg-aware we only need to make them use memcg-aware list_lru, and this is what this patch set does. As before, this patch set only enables per-memcg kmem reclaim when the pressure goes from memory.limit, not from memory.kmem.limit. Handling memory.kmem.limit is going to be tricky due to GFP_NOFS allocations, and it is still unclear whether we will have this knob in the unified hierarchy. This patch (of 9): NUMA aware slab shrinkers use the list_lru structure to distribute objects coming from different NUMA nodes to different lists. Whenever such a shrinker needs to count or scan objects from a particular node, it issues commands like this: count = list_lru_count_node(lru, sc->nid); freed = list_lru_walk_node(lru, sc->nid, isolate_func, isolate_arg, &sc->nr_to_scan); where sc is an instance of the shrink_control structure passed to it from vmscan. To simplify this, let's add special list_lru functions to be used by shrinkers, list_lru_shrink_count() and list_lru_shrink_walk(), which consolidate the nid and nr_to_scan arguments in the shrink_control structure. This will also allow us to avoid patching shrinkers that use list_lru when we make shrink_slab() per-memcg - all we will have to do is extend the shrink_control structure to include the target memcg and make list_lru_shrink_{count,walk} handle this appropriately. Signed-off-by: Vladimir Davydov <vdavydov@parallels.com> Suggested-by: Dave Chinner <david@fromorbit.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Greg Thelen <gthelen@google.com> Cc: Glauber Costa <glommer@gmail.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-02-12 22:58:47 +00:00
freed = list_lru_shrink_walk(&gfs2_qd_lru, sc,
gfs2_qd_isolate, &dispose);
gfs2_qd_list_dispose(&dispose);
fs: convert fs shrinkers to new scan/count API Convert the filesystem shrinkers to use the new API, and standardise some of the behaviours of the shrinkers at the same time. For example, nr_to_scan means the number of objects to scan, not the number of objects to free. I refactored the CIFS idmap shrinker a little - it really needs to be broken up into a shrinker per tree and keep an item count with the tree root so that we don't need to walk the tree every time the shrinker needs to count the number of objects in the tree (i.e. all the time under memory pressure). [glommer@openvz.org: fixes for ext4, ubifs, nfs, cifs and glock. Fixes are needed mainly due to new code merged in the tree] [assorted fixes folded in] Signed-off-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Glauber Costa <glommer@openvz.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Acked-by: Jan Kara <jack@suse.cz> Acked-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Cc: Arve Hjønnevåg <arve@android.com> Cc: Carlos Maiolino <cmaiolino@redhat.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Chuck Lever <chuck.lever@oracle.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Rientjes <rientjes@google.com> Cc: Gleb Natapov <gleb@redhat.com> Cc: Greg Thelen <gthelen@google.com> Cc: J. Bruce Fields <bfields@redhat.com> Cc: Jan Kara <jack@suse.cz> Cc: Jerome Glisse <jglisse@redhat.com> Cc: John Stultz <john.stultz@linaro.org> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Kent Overstreet <koverstreet@google.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Thomas Hellstrom <thellstrom@vmware.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2013-08-28 00:18:09 +00:00
return freed;
}
static unsigned long gfs2_qd_shrink_count(struct shrinker *shrink,
struct shrink_control *sc)
fs: convert fs shrinkers to new scan/count API Convert the filesystem shrinkers to use the new API, and standardise some of the behaviours of the shrinkers at the same time. For example, nr_to_scan means the number of objects to scan, not the number of objects to free. I refactored the CIFS idmap shrinker a little - it really needs to be broken up into a shrinker per tree and keep an item count with the tree root so that we don't need to walk the tree every time the shrinker needs to count the number of objects in the tree (i.e. all the time under memory pressure). [glommer@openvz.org: fixes for ext4, ubifs, nfs, cifs and glock. Fixes are needed mainly due to new code merged in the tree] [assorted fixes folded in] Signed-off-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Glauber Costa <glommer@openvz.org> Acked-by: Mel Gorman <mgorman@suse.de> Acked-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Acked-by: Jan Kara <jack@suse.cz> Acked-by: Steven Whitehouse <swhiteho@redhat.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Cc: Arve Hjønnevåg <arve@android.com> Cc: Carlos Maiolino <cmaiolino@redhat.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Chuck Lever <chuck.lever@oracle.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: David Rientjes <rientjes@google.com> Cc: Gleb Natapov <gleb@redhat.com> Cc: Greg Thelen <gthelen@google.com> Cc: J. Bruce Fields <bfields@redhat.com> Cc: Jan Kara <jack@suse.cz> Cc: Jerome Glisse <jglisse@redhat.com> Cc: John Stultz <john.stultz@linaro.org> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Kent Overstreet <koverstreet@google.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Marcelo Tosatti <mtosatti@redhat.com> Cc: Mel Gorman <mgorman@suse.de> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Thomas Hellstrom <thellstrom@vmware.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2013-08-28 00:18:09 +00:00
{
list_lru: introduce list_lru_shrink_{count,walk} Kmem accounting of memcg is unusable now, because it lacks slab shrinker support. That means when we hit the limit we will get ENOMEM w/o any chance to recover. What we should do then is to call shrink_slab, which would reclaim old inode/dentry caches from this cgroup. This is what this patch set is intended to do. Basically, it does two things. First, it introduces the notion of per-memcg slab shrinker. A shrinker that wants to reclaim objects per cgroup should mark itself as SHRINKER_MEMCG_AWARE. Then it will be passed the memory cgroup to scan from in shrink_control->memcg. For such shrinkers shrink_slab iterates over the whole cgroup subtree under the target cgroup and calls the shrinker for each kmem-active memory cgroup. Secondly, this patch set makes the list_lru structure per-memcg. It's done transparently to list_lru users - everything they have to do is to tell list_lru_init that they want memcg-aware list_lru. Then the list_lru will automatically distribute objects among per-memcg lists basing on which cgroup the object is accounted to. This way to make FS shrinkers (icache, dcache) memcg-aware we only need to make them use memcg-aware list_lru, and this is what this patch set does. As before, this patch set only enables per-memcg kmem reclaim when the pressure goes from memory.limit, not from memory.kmem.limit. Handling memory.kmem.limit is going to be tricky due to GFP_NOFS allocations, and it is still unclear whether we will have this knob in the unified hierarchy. This patch (of 9): NUMA aware slab shrinkers use the list_lru structure to distribute objects coming from different NUMA nodes to different lists. Whenever such a shrinker needs to count or scan objects from a particular node, it issues commands like this: count = list_lru_count_node(lru, sc->nid); freed = list_lru_walk_node(lru, sc->nid, isolate_func, isolate_arg, &sc->nr_to_scan); where sc is an instance of the shrink_control structure passed to it from vmscan. To simplify this, let's add special list_lru functions to be used by shrinkers, list_lru_shrink_count() and list_lru_shrink_walk(), which consolidate the nid and nr_to_scan arguments in the shrink_control structure. This will also allow us to avoid patching shrinkers that use list_lru when we make shrink_slab() per-memcg - all we will have to do is extend the shrink_control structure to include the target memcg and make list_lru_shrink_{count,walk} handle this appropriately. Signed-off-by: Vladimir Davydov <vdavydov@parallels.com> Suggested-by: Dave Chinner <david@fromorbit.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Michal Hocko <mhocko@suse.cz> Cc: Greg Thelen <gthelen@google.com> Cc: Glauber Costa <glommer@gmail.com> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: David Rientjes <rientjes@google.com> Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com> Cc: Tejun Heo <tj@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-02-12 22:58:47 +00:00
return vfs_pressure_ratio(list_lru_shrink_count(&gfs2_qd_lru, sc));
}
struct shrinker gfs2_qd_shrinker = {
.count_objects = gfs2_qd_shrink_count,
.scan_objects = gfs2_qd_shrink_scan,
.seeks = DEFAULT_SEEKS,
.flags = SHRINKER_NUMA_AWARE,
};
static u64 qd2index(struct gfs2_quota_data *qd)
{
struct kqid qid = qd->qd_id;
return (2 * (u64)from_kqid(&init_user_ns, qid)) +
((qid.type == USRQUOTA) ? 0 : 1);
}
static u64 qd2offset(struct gfs2_quota_data *qd)
{
u64 offset;
offset = qd2index(qd);
offset *= sizeof(struct gfs2_quota);
return offset;
}
static struct gfs2_quota_data *qd_alloc(unsigned hash, struct gfs2_sbd *sdp, struct kqid qid)
{
struct gfs2_quota_data *qd;
int error;
qd = kmem_cache_zalloc(gfs2_quotad_cachep, GFP_NOFS);
if (!qd)
return NULL;
qd->qd_sbd = sdp;
qd->qd_lockref.count = 0;
spin_lock_init(&qd->qd_lockref.lock);
qd->qd_id = qid;
qd->qd_slot = -1;
INIT_LIST_HEAD(&qd->qd_lru);
qd->qd_hash = hash;
error = gfs2_glock_get(sdp, qd2index(qd),
&gfs2_quota_glops, CREATE, &qd->qd_gl);
if (error)
goto fail;
return qd;
fail:
kmem_cache_free(gfs2_quotad_cachep, qd);
return NULL;
}
static struct gfs2_quota_data *gfs2_qd_search_bucket(unsigned int hash,
const struct gfs2_sbd *sdp,
struct kqid qid)
{
struct gfs2_quota_data *qd;
struct hlist_bl_node *h;
hlist_bl_for_each_entry_rcu(qd, h, &qd_hash_table[hash], qd_hlist) {
if (!qid_eq(qd->qd_id, qid))
continue;
if (qd->qd_sbd != sdp)
continue;
if (lockref_get_not_dead(&qd->qd_lockref)) {
list_lru_del(&gfs2_qd_lru, &qd->qd_lru);
return qd;
}
}
return NULL;
}
static int qd_get(struct gfs2_sbd *sdp, struct kqid qid,
struct gfs2_quota_data **qdp)
{
struct gfs2_quota_data *qd, *new_qd;
unsigned int hash = gfs2_qd_hash(sdp, qid);
rcu_read_lock();
*qdp = qd = gfs2_qd_search_bucket(hash, sdp, qid);
rcu_read_unlock();
if (qd)
return 0;
new_qd = qd_alloc(hash, sdp, qid);
if (!new_qd)
return -ENOMEM;
spin_lock(&qd_lock);
spin_lock_bucket(hash);
*qdp = qd = gfs2_qd_search_bucket(hash, sdp, qid);
if (qd == NULL) {
new_qd->qd_lockref.count++;
*qdp = new_qd;
list_add(&new_qd->qd_list, &sdp->sd_quota_list);
hlist_bl_add_head_rcu(&new_qd->qd_hlist, &qd_hash_table[hash]);
atomic_inc(&sdp->sd_quota_count);
}
spin_unlock_bucket(hash);
spin_unlock(&qd_lock);
if (qd) {
gfs2_glock_put(new_qd->qd_gl);
kmem_cache_free(gfs2_quotad_cachep, new_qd);
}
return 0;
}
static void qd_hold(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
gfs2_assert(sdp, !__lockref_is_dead(&qd->qd_lockref));
lockref_get(&qd->qd_lockref);
}
static void qd_put(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp;
if (lockref_put_or_lock(&qd->qd_lockref))
return;
BUG_ON(__lockref_is_dead(&qd->qd_lockref));
sdp = qd->qd_sbd;
if (unlikely(!test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags))) {
lockref_mark_dead(&qd->qd_lockref);
spin_unlock(&qd->qd_lockref.lock);
gfs2_qd_dispose(qd);
return;
}
qd->qd_lockref.count = 0;
list_lru_add(&gfs2_qd_lru, &qd->qd_lru);
spin_unlock(&qd->qd_lockref.lock);
}
static int slot_get(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
unsigned int bit;
int error = 0;
spin_lock(&sdp->sd_bitmap_lock);
if (qd->qd_slot_count != 0)
goto out;
error = -ENOSPC;
bit = find_first_zero_bit(sdp->sd_quota_bitmap, sdp->sd_quota_slots);
if (bit < sdp->sd_quota_slots) {
set_bit(bit, sdp->sd_quota_bitmap);
qd->qd_slot = bit;
error = 0;
out:
qd->qd_slot_count++;
}
spin_unlock(&sdp->sd_bitmap_lock);
return error;
}
static void slot_hold(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
spin_lock(&sdp->sd_bitmap_lock);
gfs2_assert(sdp, qd->qd_slot_count);
qd->qd_slot_count++;
spin_unlock(&sdp->sd_bitmap_lock);
}
static void slot_put(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
spin_lock(&sdp->sd_bitmap_lock);
gfs2_assert(sdp, qd->qd_slot_count);
if (!--qd->qd_slot_count) {
BUG_ON(!test_and_clear_bit(qd->qd_slot, sdp->sd_quota_bitmap));
qd->qd_slot = -1;
}
spin_unlock(&sdp->sd_bitmap_lock);
}
static int bh_get(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
struct inode *inode = sdp->sd_qc_inode;
struct gfs2_inode *ip = GFS2_I(inode);
unsigned int block, offset;
struct buffer_head *bh;
struct iomap iomap = { };
int error;
mutex_lock(&sdp->sd_quota_mutex);
if (qd->qd_bh_count++) {
mutex_unlock(&sdp->sd_quota_mutex);
return 0;
}
block = qd->qd_slot / sdp->sd_qc_per_block;
offset = qd->qd_slot % sdp->sd_qc_per_block;
error = gfs2_iomap_get(inode,
(loff_t)block << inode->i_blkbits,
i_blocksize(inode), &iomap);
if (error)
goto fail;
error = -ENOENT;
if (iomap.type != IOMAP_MAPPED)
goto fail;
error = gfs2_meta_read(ip->i_gl, iomap.addr >> inode->i_blkbits,
DIO_WAIT, 0, &bh);
if (error)
goto fail;
error = -EIO;
if (gfs2_metatype_check(sdp, bh, GFS2_METATYPE_QC))
goto fail_brelse;
qd->qd_bh = bh;
qd->qd_bh_qc = (struct gfs2_quota_change *)
(bh->b_data + sizeof(struct gfs2_meta_header) +
offset * sizeof(struct gfs2_quota_change));
mutex_unlock(&sdp->sd_quota_mutex);
return 0;
fail_brelse:
brelse(bh);
fail:
qd->qd_bh_count--;
mutex_unlock(&sdp->sd_quota_mutex);
return error;
}
static void bh_put(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
mutex_lock(&sdp->sd_quota_mutex);
gfs2_assert(sdp, qd->qd_bh_count);
if (!--qd->qd_bh_count) {
brelse(qd->qd_bh);
qd->qd_bh = NULL;
qd->qd_bh_qc = NULL;
}
mutex_unlock(&sdp->sd_quota_mutex);
}
static int qd_check_sync(struct gfs2_sbd *sdp, struct gfs2_quota_data *qd,
u64 *sync_gen)
{
if (test_bit(QDF_LOCKED, &qd->qd_flags) ||
!test_bit(QDF_CHANGE, &qd->qd_flags) ||
(sync_gen && (qd->qd_sync_gen >= *sync_gen)))
return 0;
if (!lockref_get_not_dead(&qd->qd_lockref))
return 0;
list_move_tail(&qd->qd_list, &sdp->sd_quota_list);
set_bit(QDF_LOCKED, &qd->qd_flags);
qd->qd_change_sync = qd->qd_change;
slot_hold(qd);
return 1;
}
static int qd_bh_get_or_undo(struct gfs2_sbd *sdp, struct gfs2_quota_data *qd)
{
int error;
error = bh_get(qd);
if (!error)
return 0;
clear_bit(QDF_LOCKED, &qd->qd_flags);
slot_put(qd);
qd_put(qd);
return error;
}
static int qd_fish(struct gfs2_sbd *sdp, struct gfs2_quota_data **qdp)
{
struct gfs2_quota_data *qd = NULL, *iter;
int error;
*qdp = NULL;
if (sb_rdonly(sdp->sd_vfs))
return 0;
spin_lock(&qd_lock);
list_for_each_entry(iter, &sdp->sd_quota_list, qd_list) {
if (qd_check_sync(sdp, iter, &sdp->sd_quota_sync_gen)) {
qd = iter;
break;
}
}
spin_unlock(&qd_lock);
if (qd) {
error = qd_bh_get_or_undo(sdp, qd);
if (error)
return error;
*qdp = qd;
}
return 0;
}
static void qdsb_put(struct gfs2_quota_data *qd)
{
bh_put(qd);
slot_put(qd);
qd_put(qd);
}
static void qd_unlock(struct gfs2_quota_data *qd)
{
gfs2_assert_warn(qd->qd_sbd, test_bit(QDF_LOCKED, &qd->qd_flags));
clear_bit(QDF_LOCKED, &qd->qd_flags);
qdsb_put(qd);
}
static int qdsb_get(struct gfs2_sbd *sdp, struct kqid qid,
struct gfs2_quota_data **qdp)
{
int error;
error = qd_get(sdp, qid, qdp);
if (error)
return error;
error = slot_get(*qdp);
if (error)
goto fail;
error = bh_get(*qdp);
if (error)
goto fail_slot;
return 0;
fail_slot:
slot_put(*qdp);
fail:
qd_put(*qdp);
return error;
}
/**
* gfs2_qa_get - make sure we have a quota allocations data structure,
* if necessary
* @ip: the inode for this reservation
*/
int gfs2_qa_get(struct gfs2_inode *ip)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct inode *inode = &ip->i_inode;
if (sdp->sd_args.ar_quota == GFS2_QUOTA_OFF)
return 0;
spin_lock(&inode->i_lock);
if (ip->i_qadata == NULL) {
struct gfs2_qadata *tmp;
spin_unlock(&inode->i_lock);
tmp = kmem_cache_zalloc(gfs2_qadata_cachep, GFP_NOFS);
if (!tmp)
return -ENOMEM;
spin_lock(&inode->i_lock);
if (ip->i_qadata == NULL)
ip->i_qadata = tmp;
else
kmem_cache_free(gfs2_qadata_cachep, tmp);
}
ip->i_qadata->qa_ref++;
spin_unlock(&inode->i_lock);
return 0;
}
void gfs2_qa_put(struct gfs2_inode *ip)
{
struct inode *inode = &ip->i_inode;
spin_lock(&inode->i_lock);
if (ip->i_qadata && --ip->i_qadata->qa_ref == 0) {
kmem_cache_free(gfs2_qadata_cachep, ip->i_qadata);
ip->i_qadata = NULL;
}
spin_unlock(&inode->i_lock);
}
int gfs2_quota_hold(struct gfs2_inode *ip, kuid_t uid, kgid_t gid)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct gfs2_quota_data **qd;
int error;
if (sdp->sd_args.ar_quota == GFS2_QUOTA_OFF)
return 0;
error = gfs2_qa_get(ip);
if (error)
return error;
qd = ip->i_qadata->qa_qd;
if (gfs2_assert_warn(sdp, !ip->i_qadata->qa_qd_num) ||
gfs2_assert_warn(sdp, !test_bit(GIF_QD_LOCKED, &ip->i_flags))) {
error = -EIO;
gfs2_qa_put(ip);
goto out;
}
error = qdsb_get(sdp, make_kqid_uid(ip->i_inode.i_uid), qd);
if (error)
goto out_unhold;
ip->i_qadata->qa_qd_num++;
qd++;
error = qdsb_get(sdp, make_kqid_gid(ip->i_inode.i_gid), qd);
if (error)
goto out_unhold;
ip->i_qadata->qa_qd_num++;
qd++;
if (!uid_eq(uid, NO_UID_QUOTA_CHANGE) &&
!uid_eq(uid, ip->i_inode.i_uid)) {
error = qdsb_get(sdp, make_kqid_uid(uid), qd);
if (error)
goto out_unhold;
ip->i_qadata->qa_qd_num++;
qd++;
}
if (!gid_eq(gid, NO_GID_QUOTA_CHANGE) &&
!gid_eq(gid, ip->i_inode.i_gid)) {
error = qdsb_get(sdp, make_kqid_gid(gid), qd);
if (error)
goto out_unhold;
ip->i_qadata->qa_qd_num++;
qd++;
}
out_unhold:
if (error)
gfs2_quota_unhold(ip);
out:
return error;
}
void gfs2_quota_unhold(struct gfs2_inode *ip)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
u32 x;
if (ip->i_qadata == NULL)
return;
gfs2_assert_warn(sdp, !test_bit(GIF_QD_LOCKED, &ip->i_flags));
for (x = 0; x < ip->i_qadata->qa_qd_num; x++) {
qdsb_put(ip->i_qadata->qa_qd[x]);
ip->i_qadata->qa_qd[x] = NULL;
}
ip->i_qadata->qa_qd_num = 0;
gfs2_qa_put(ip);
}
static int sort_qd(const void *a, const void *b)
{
const struct gfs2_quota_data *qd_a = *(const struct gfs2_quota_data **)a;
const struct gfs2_quota_data *qd_b = *(const struct gfs2_quota_data **)b;
if (qid_lt(qd_a->qd_id, qd_b->qd_id))
return -1;
if (qid_lt(qd_b->qd_id, qd_a->qd_id))
return 1;
return 0;
}
gfs2: Add quota_change type Function do_qc has two main uses: (1) to re-sync the local quota changes (qd) to the master quotas, and (2) normal quota changes. In the case of normal quota changes, the change can be positive or negative, as the quota usage goes up and down. Before this patch function do_qc was distinguishing one from another by whether the resulting value is or isn't zero: In the case of a re-sync (called do_sync) the quota value is moved from the temporary value to a master value, so the amount is added to one and subtracted from the other. The problem is that since the values can be positive or negative we can occasionally run into situations where we are not doing a re-sync but the quota change just happens to cancel out the previous value. In the case of a re-sync extra references and locks are taken, and so do_qc needs to release them. In the case of a normal quota change, no extra references and locks are taken, so it must not try to release them. The problem is: if the quota change is not a re-sync but the value just happens to cancel out the original quota change, the resulting zero value fools do_qc into thinking this is a re-sync and therefore it must release the extra references. This results in problems, mainly having to do with slot reference numbers going smaller than zero. This patch introduces new constants, QC_SYNC and QC_CHANGE so do_qc can really tell the difference. For QC_SYNC calls it must release the extra references acquired by gfs2_quota_unlock's call to qd_check_sync. For QC_CHANGE calls it does not have extra references to put. Note that this allows quota changes back to a value of zero, and so I removed an assert warning related to that. Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2023-06-28 14:54:41 +00:00
static void do_qc(struct gfs2_quota_data *qd, s64 change, int qc_type)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
struct gfs2_inode *ip = GFS2_I(sdp->sd_qc_inode);
struct gfs2_quota_change *qc = qd->qd_bh_qc;
s64 x;
mutex_lock(&sdp->sd_quota_mutex);
gfs2_trans_add_meta(ip->i_gl, qd->qd_bh);
if (!test_bit(QDF_CHANGE, &qd->qd_flags)) {
qc->qc_change = 0;
qc->qc_flags = 0;
if (qd->qd_id.type == USRQUOTA)
qc->qc_flags = cpu_to_be32(GFS2_QCF_USER);
qc->qc_id = cpu_to_be32(from_kqid(&init_user_ns, qd->qd_id));
}
x = be64_to_cpu(qc->qc_change) + change;
qc->qc_change = cpu_to_be64(x);
spin_lock(&qd_lock);
qd->qd_change = x;
spin_unlock(&qd_lock);
gfs2: Add quota_change type Function do_qc has two main uses: (1) to re-sync the local quota changes (qd) to the master quotas, and (2) normal quota changes. In the case of normal quota changes, the change can be positive or negative, as the quota usage goes up and down. Before this patch function do_qc was distinguishing one from another by whether the resulting value is or isn't zero: In the case of a re-sync (called do_sync) the quota value is moved from the temporary value to a master value, so the amount is added to one and subtracted from the other. The problem is that since the values can be positive or negative we can occasionally run into situations where we are not doing a re-sync but the quota change just happens to cancel out the previous value. In the case of a re-sync extra references and locks are taken, and so do_qc needs to release them. In the case of a normal quota change, no extra references and locks are taken, so it must not try to release them. The problem is: if the quota change is not a re-sync but the value just happens to cancel out the original quota change, the resulting zero value fools do_qc into thinking this is a re-sync and therefore it must release the extra references. This results in problems, mainly having to do with slot reference numbers going smaller than zero. This patch introduces new constants, QC_SYNC and QC_CHANGE so do_qc can really tell the difference. For QC_SYNC calls it must release the extra references acquired by gfs2_quota_unlock's call to qd_check_sync. For QC_CHANGE calls it does not have extra references to put. Note that this allows quota changes back to a value of zero, and so I removed an assert warning related to that. Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2023-06-28 14:54:41 +00:00
if (qc_type == QC_CHANGE) {
if (!test_and_set_bit(QDF_CHANGE, &qd->qd_flags)) {
qd_hold(qd);
slot_hold(qd);
}
} else {
gfs2_assert_warn(sdp, test_bit(QDF_CHANGE, &qd->qd_flags));
clear_bit(QDF_CHANGE, &qd->qd_flags);
qc->qc_flags = 0;
qc->qc_id = 0;
slot_put(qd);
qd_put(qd);
}
if (change < 0) /* Reset quiet flag if we freed some blocks */
clear_bit(QDF_QMSG_QUIET, &qd->qd_flags);
mutex_unlock(&sdp->sd_quota_mutex);
}
static int gfs2_write_buf_to_page(struct gfs2_sbd *sdp, unsigned long index,
unsigned off, void *buf, unsigned bytes)
{
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct inode *inode = &ip->i_inode;
struct address_space *mapping = inode->i_mapping;
struct page *page;
struct buffer_head *bh;
u64 blk;
unsigned bsize = sdp->sd_sb.sb_bsize, bnum = 0, boff = 0;
unsigned to_write = bytes, pg_off = off;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
blk = index << (PAGE_SHIFT - sdp->sd_sb.sb_bsize_shift);
boff = off % bsize;
page = grab_cache_page(mapping, index);
if (!page)
return -ENOMEM;
if (!page_has_buffers(page))
create_empty_buffers(page, bsize, 0);
bh = page_buffers(page);
for(;;) {
/* Find the beginning block within the page */
if (pg_off >= ((bnum * bsize) + bsize)) {
bh = bh->b_this_page;
bnum++;
blk++;
continue;
}
if (!buffer_mapped(bh)) {
gfs2_block_map(inode, blk, bh, 1);
if (!buffer_mapped(bh))
goto unlock_out;
/* If it's a newly allocated disk block, zero it */
if (buffer_new(bh))
zero_user(page, bnum * bsize, bh->b_size);
}
if (PageUptodate(page))
set_buffer_uptodate(bh);
if (bh_read(bh, REQ_META | REQ_PRIO) < 0)
goto unlock_out;
gfs2_trans_add_data(ip->i_gl, bh);
/* If we need to write to the next block as well */
if (to_write > (bsize - boff)) {
pg_off += (bsize - boff);
to_write -= (bsize - boff);
boff = pg_off % bsize;
continue;
}
break;
}
/* Write to the page, now that we have setup the buffer(s) */
memcpy_to_page(page, off, buf, bytes);
flush_dcache_page(page);
unlock_page(page);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(page);
return 0;
unlock_out:
unlock_page(page);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(page);
return -EIO;
}
static int gfs2_write_disk_quota(struct gfs2_sbd *sdp, struct gfs2_quota *qp,
loff_t loc)
{
unsigned long pg_beg;
unsigned pg_off, nbytes, overflow = 0;
int error;
void *ptr;
nbytes = sizeof(struct gfs2_quota);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
pg_beg = loc >> PAGE_SHIFT;
pg_off = offset_in_page(loc);
/* If the quota straddles a page boundary, split the write in two */
if ((pg_off + nbytes) > PAGE_SIZE)
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
overflow = (pg_off + nbytes) - PAGE_SIZE;
ptr = qp;
error = gfs2_write_buf_to_page(sdp, pg_beg, pg_off, ptr,
nbytes - overflow);
/* If there's an overflow, write the remaining bytes to the next page */
if (!error && overflow)
error = gfs2_write_buf_to_page(sdp, pg_beg + 1, 0,
ptr + nbytes - overflow,
overflow);
return error;
}
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
/**
* gfs2_adjust_quota - adjust record of current block usage
* @sdp: The superblock
* @loc: Offset of the entry in the quota file
* @change: The amount of usage change to record
* @qd: The quota data
* @fdq: The updated limits to record
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
*
* This function was mostly borrowed from gfs2_block_truncate_page which was
* in turn mostly borrowed from ext3
*
* Returns: 0 or -ve on error
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
*/
static int gfs2_adjust_quota(struct gfs2_sbd *sdp, loff_t loc,
s64 change, struct gfs2_quota_data *qd,
struct qc_dqblk *fdq)
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
{
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct inode *inode = &ip->i_inode;
struct gfs2_quota q;
int err;
u64 size;
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
if (gfs2_is_stuffed(ip)) {
err = gfs2_unstuff_dinode(ip);
if (err)
return err;
}
memset(&q, 0, sizeof(struct gfs2_quota));
err = gfs2_internal_read(ip, (char *)&q, &loc, sizeof(q));
if (err < 0)
return err;
loc -= sizeof(q); /* gfs2_internal_read would've advanced the loc ptr */
be64_add_cpu(&q.qu_value, change);
if (((s64)be64_to_cpu(q.qu_value)) < 0)
q.qu_value = 0; /* Never go negative on quota usage */
qd->qd_qb.qb_value = q.qu_value;
if (fdq) {
if (fdq->d_fieldmask & QC_SPC_SOFT) {
q.qu_warn = cpu_to_be64(fdq->d_spc_softlimit >> sdp->sd_sb.sb_bsize_shift);
qd->qd_qb.qb_warn = q.qu_warn;
}
if (fdq->d_fieldmask & QC_SPC_HARD) {
q.qu_limit = cpu_to_be64(fdq->d_spc_hardlimit >> sdp->sd_sb.sb_bsize_shift);
qd->qd_qb.qb_limit = q.qu_limit;
}
if (fdq->d_fieldmask & QC_SPACE) {
q.qu_value = cpu_to_be64(fdq->d_space >> sdp->sd_sb.sb_bsize_shift);
qd->qd_qb.qb_value = q.qu_value;
}
}
err = gfs2_write_disk_quota(sdp, &q, loc);
if (!err) {
size = loc + sizeof(struct gfs2_quota);
if (size > inode->i_size)
i_size_write(inode, size);
inode->i_mtime = inode->i_atime = current_time(inode);
mark_inode_dirty(inode);
set_bit(QDF_REFRESH, &qd->qd_flags);
}
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
return err;
}
static int do_sync(unsigned int num_qd, struct gfs2_quota_data **qda)
{
struct gfs2_sbd *sdp = (*qda)->qd_sbd;
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct gfs2_alloc_parms ap = { .aflags = 0, };
unsigned int data_blocks, ind_blocks;
struct gfs2_holder *ghs, i_gh;
unsigned int qx, x;
struct gfs2_quota_data *qd;
unsigned reserved;
loff_t offset;
unsigned int nalloc = 0, blocks;
int error;
gfs2_write_calc_reserv(ip, sizeof(struct gfs2_quota),
&data_blocks, &ind_blocks);
treewide: kmalloc() -> kmalloc_array() The kmalloc() function has a 2-factor argument form, kmalloc_array(). This patch replaces cases of: kmalloc(a * b, gfp) with: kmalloc_array(a * b, gfp) as well as handling cases of: kmalloc(a * b * c, gfp) with: kmalloc(array3_size(a, b, c), gfp) as it's slightly less ugly than: kmalloc_array(array_size(a, b), c, gfp) This does, however, attempt to ignore constant size factors like: kmalloc(4 * 1024, gfp) though any constants defined via macros get caught up in the conversion. Any factors with a sizeof() of "unsigned char", "char", and "u8" were dropped, since they're redundant. The tools/ directory was manually excluded, since it has its own implementation of kmalloc(). The Coccinelle script used for this was: // Fix redundant parens around sizeof(). @@ type TYPE; expression THING, E; @@ ( kmalloc( - (sizeof(TYPE)) * E + sizeof(TYPE) * E , ...) | kmalloc( - (sizeof(THING)) * E + sizeof(THING) * E , ...) ) // Drop single-byte sizes and redundant parens. @@ expression COUNT; typedef u8; typedef __u8; @@ ( kmalloc( - sizeof(u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(__u8) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(unsigned char) * (COUNT) + COUNT , ...) | kmalloc( - sizeof(u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(__u8) * COUNT + COUNT , ...) | kmalloc( - sizeof(char) * COUNT + COUNT , ...) | kmalloc( - sizeof(unsigned char) * COUNT + COUNT , ...) ) // 2-factor product with sizeof(type/expression) and identifier or constant. @@ type TYPE; expression THING; identifier COUNT_ID; constant COUNT_CONST; @@ ( - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_ID) + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_ID + COUNT_ID, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (COUNT_CONST) + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * COUNT_CONST + COUNT_CONST, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_ID) + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_ID + COUNT_ID, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (COUNT_CONST) + COUNT_CONST, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * COUNT_CONST + COUNT_CONST, sizeof(THING) , ...) ) // 2-factor product, only identifiers. @@ identifier SIZE, COUNT; @@ - kmalloc + kmalloc_array ( - SIZE * COUNT + COUNT, SIZE , ...) // 3-factor product with 1 sizeof(type) or sizeof(expression), with // redundant parens removed. @@ expression THING; identifier STRIDE, COUNT; type TYPE; @@ ( kmalloc( - sizeof(TYPE) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(TYPE) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(TYPE)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * (COUNT) * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * (STRIDE) + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) | kmalloc( - sizeof(THING) * COUNT * STRIDE + array3_size(COUNT, STRIDE, sizeof(THING)) , ...) ) // 3-factor product with 2 sizeof(variable), with redundant parens removed. @@ expression THING1, THING2; identifier COUNT; type TYPE1, TYPE2; @@ ( kmalloc( - sizeof(TYPE1) * sizeof(TYPE2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(TYPE2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(THING1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(THING1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * COUNT + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) | kmalloc( - sizeof(TYPE1) * sizeof(THING2) * (COUNT) + array3_size(COUNT, sizeof(TYPE1), sizeof(THING2)) , ...) ) // 3-factor product, only identifiers, with redundant parens removed. @@ identifier STRIDE, SIZE, COUNT; @@ ( kmalloc( - (COUNT) * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * STRIDE * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - (COUNT) * (STRIDE) * (SIZE) + array3_size(COUNT, STRIDE, SIZE) , ...) | kmalloc( - COUNT * STRIDE * SIZE + array3_size(COUNT, STRIDE, SIZE) , ...) ) // Any remaining multi-factor products, first at least 3-factor products, // when they're not all constants... @@ expression E1, E2, E3; constant C1, C2, C3; @@ ( kmalloc(C1 * C2 * C3, ...) | kmalloc( - (E1) * E2 * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * E3 + array3_size(E1, E2, E3) , ...) | kmalloc( - (E1) * (E2) * (E3) + array3_size(E1, E2, E3) , ...) | kmalloc( - E1 * E2 * E3 + array3_size(E1, E2, E3) , ...) ) // And then all remaining 2 factors products when they're not all constants, // keeping sizeof() as the second factor argument. @@ expression THING, E1, E2; type TYPE; constant C1, C2, C3; @@ ( kmalloc(sizeof(THING) * C2, ...) | kmalloc(sizeof(TYPE) * C2, ...) | kmalloc(C1 * C2 * C3, ...) | kmalloc(C1 * C2, ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * (E2) + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(TYPE) * E2 + E2, sizeof(TYPE) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * (E2) + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - sizeof(THING) * E2 + E2, sizeof(THING) , ...) | - kmalloc + kmalloc_array ( - (E1) * E2 + E1, E2 , ...) | - kmalloc + kmalloc_array ( - (E1) * (E2) + E1, E2 , ...) | - kmalloc + kmalloc_array ( - E1 * E2 + E1, E2 , ...) ) Signed-off-by: Kees Cook <keescook@chromium.org>
2018-06-12 20:55:00 +00:00
ghs = kmalloc_array(num_qd, sizeof(struct gfs2_holder), GFP_NOFS);
if (!ghs)
return -ENOMEM;
sort(qda, num_qd, sizeof(struct gfs2_quota_data *), sort_qd, NULL);
inode_lock(&ip->i_inode);
for (qx = 0; qx < num_qd; qx++) {
error = gfs2_glock_nq_init(qda[qx]->qd_gl, LM_ST_EXCLUSIVE,
GL_NOCACHE, &ghs[qx]);
if (error)
goto out_dq;
}
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &i_gh);
if (error)
goto out_dq;
for (x = 0; x < num_qd; x++) {
offset = qd2offset(qda[x]);
if (gfs2_write_alloc_required(ip, offset,
sizeof(struct gfs2_quota)))
nalloc++;
}
/*
* 1 blk for unstuffing inode if stuffed. We add this extra
* block to the reservation unconditionally. If the inode
* doesn't need unstuffing, the block will be released to the
* rgrp since it won't be allocated during the transaction
*/
/* +3 in the end for unstuffing block, inode size update block
* and another block in case quota straddles page boundary and
* two blocks need to be updated instead of 1 */
blocks = num_qd * data_blocks + RES_DINODE + num_qd + 3;
reserved = 1 + (nalloc * (data_blocks + ind_blocks));
ap.target = reserved;
error = gfs2_inplace_reserve(ip, &ap);
if (error)
goto out_alloc;
if (nalloc)
blocks += gfs2_rg_blocks(ip, reserved) + nalloc * ind_blocks + RES_STATFS;
error = gfs2_trans_begin(sdp, blocks, 0);
if (error)
goto out_ipres;
for (x = 0; x < num_qd; x++) {
qd = qda[x];
offset = qd2offset(qd);
error = gfs2_adjust_quota(sdp, offset, qd->qd_change_sync, qd,
NULL);
[GFS2] Make journaled data files identical to normal files on disk This is a very large patch, with a few still to be resolved issues so you might want to check out the previous head of the tree since this is known to be unstable. Fixes for the various bugs will be forthcoming shortly. This patch removes the special data format which has been used up till now for journaled data files. Directories still retain the old format so that they will remain on disk compatible with earlier releases. As a result you can now do the following with journaled data files: 1) mmap them 2) export them over NFS 3) convert to/from normal files whenever you want to (the zero length restriction is gone) In addition the level at which GFS' locking is done has changed for all files (since they all now use the page cache) such that the locking is done at the page cache level rather than the level of the fs operations. This should mean that things like loopback mounts and other things which touch the page cache directly should now work. Current known issues: 1. There is a lock mode inversion problem related to the resource group hold function which needs to be resolved. 2. Any significant amount of I/O causes an oops with an offset of hex 320 (NULL pointer dereference) which appears to be related to a journaled data buffer appearing on a list where it shouldn't be. 3. Direct I/O writes are disabled for the time being (will reappear later) 4. There is probably a deadlock between the page lock and GFS' locks under certain combinations of mmap and fs operation I/O. 5. Issue relating to ref counting on internally used inodes causes a hang on umount (discovered before this patch, and not fixed by it) 6. One part of the directory metadata is different from GFS1 and will need to be resolved before next release. Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
2006-02-08 11:50:51 +00:00
if (error)
goto out_end_trans;
gfs2: Add quota_change type Function do_qc has two main uses: (1) to re-sync the local quota changes (qd) to the master quotas, and (2) normal quota changes. In the case of normal quota changes, the change can be positive or negative, as the quota usage goes up and down. Before this patch function do_qc was distinguishing one from another by whether the resulting value is or isn't zero: In the case of a re-sync (called do_sync) the quota value is moved from the temporary value to a master value, so the amount is added to one and subtracted from the other. The problem is that since the values can be positive or negative we can occasionally run into situations where we are not doing a re-sync but the quota change just happens to cancel out the previous value. In the case of a re-sync extra references and locks are taken, and so do_qc needs to release them. In the case of a normal quota change, no extra references and locks are taken, so it must not try to release them. The problem is: if the quota change is not a re-sync but the value just happens to cancel out the original quota change, the resulting zero value fools do_qc into thinking this is a re-sync and therefore it must release the extra references. This results in problems, mainly having to do with slot reference numbers going smaller than zero. This patch introduces new constants, QC_SYNC and QC_CHANGE so do_qc can really tell the difference. For QC_SYNC calls it must release the extra references acquired by gfs2_quota_unlock's call to qd_check_sync. For QC_CHANGE calls it does not have extra references to put. Note that this allows quota changes back to a value of zero, and so I removed an assert warning related to that. Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2023-06-28 14:54:41 +00:00
do_qc(qd, -qd->qd_change_sync, QC_SYNC);
set_bit(QDF_REFRESH, &qd->qd_flags);
}
error = 0;
out_end_trans:
gfs2_trans_end(sdp);
out_ipres:
gfs2_inplace_release(ip);
out_alloc:
gfs2_glock_dq_uninit(&i_gh);
out_dq:
while (qx--)
gfs2_glock_dq_uninit(&ghs[qx]);
inode_unlock(&ip->i_inode);
kfree(ghs);
gfs2_log_flush(ip->i_gl->gl_name.ln_sbd, ip->i_gl,
GFS2_LOG_HEAD_FLUSH_NORMAL | GFS2_LFC_DO_SYNC);
if (!error) {
for (x = 0; x < num_qd; x++)
qda[x]->qd_sync_gen = sdp->sd_quota_sync_gen;
}
return error;
}
static int update_qd(struct gfs2_sbd *sdp, struct gfs2_quota_data *qd)
{
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct gfs2_quota q;
struct gfs2_quota_lvb *qlvb;
loff_t pos;
int error;
memset(&q, 0, sizeof(struct gfs2_quota));
pos = qd2offset(qd);
error = gfs2_internal_read(ip, (char *)&q, &pos, sizeof(q));
if (error < 0)
return error;
qlvb = (struct gfs2_quota_lvb *)qd->qd_gl->gl_lksb.sb_lvbptr;
qlvb->qb_magic = cpu_to_be32(GFS2_MAGIC);
qlvb->__pad = 0;
qlvb->qb_limit = q.qu_limit;
qlvb->qb_warn = q.qu_warn;
qlvb->qb_value = q.qu_value;
qd->qd_qb = *qlvb;
return 0;
}
static int do_glock(struct gfs2_quota_data *qd, int force_refresh,
struct gfs2_holder *q_gh)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct gfs2_holder i_gh;
int error;
gfs2_assert_warn(sdp, sdp == qd->qd_gl->gl_name.ln_sbd);
restart:
error = gfs2_glock_nq_init(qd->qd_gl, LM_ST_SHARED, 0, q_gh);
if (error)
return error;
if (test_and_clear_bit(QDF_REFRESH, &qd->qd_flags))
force_refresh = FORCE;
qd->qd_qb = *(struct gfs2_quota_lvb *)qd->qd_gl->gl_lksb.sb_lvbptr;
if (force_refresh || qd->qd_qb.qb_magic != cpu_to_be32(GFS2_MAGIC)) {
gfs2_glock_dq_uninit(q_gh);
error = gfs2_glock_nq_init(qd->qd_gl, LM_ST_EXCLUSIVE,
GL_NOCACHE, q_gh);
if (error)
return error;
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, 0, &i_gh);
if (error)
goto fail;
error = update_qd(sdp, qd);
if (error)
goto fail_gunlock;
gfs2_glock_dq_uninit(&i_gh);
gfs2_glock_dq_uninit(q_gh);
force_refresh = 0;
goto restart;
}
return 0;
fail_gunlock:
gfs2_glock_dq_uninit(&i_gh);
fail:
gfs2_glock_dq_uninit(q_gh);
return error;
}
int gfs2_quota_lock(struct gfs2_inode *ip, kuid_t uid, kgid_t gid)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct gfs2_quota_data *qd;
u32 x;
int error;
if (sdp->sd_args.ar_quota != GFS2_QUOTA_ON &&
sdp->sd_args.ar_quota != GFS2_QUOTA_QUIET)
return 0;
error = gfs2_quota_hold(ip, uid, gid);
if (error)
return error;
sort(ip->i_qadata->qa_qd, ip->i_qadata->qa_qd_num,
sizeof(struct gfs2_quota_data *), sort_qd, NULL);
for (x = 0; x < ip->i_qadata->qa_qd_num; x++) {
qd = ip->i_qadata->qa_qd[x];
error = do_glock(qd, NO_FORCE, &ip->i_qadata->qa_qd_ghs[x]);
if (error)
break;
}
if (!error)
set_bit(GIF_QD_LOCKED, &ip->i_flags);
else {
while (x--)
gfs2_glock_dq_uninit(&ip->i_qadata->qa_qd_ghs[x]);
gfs2_quota_unhold(ip);
}
return error;
}
static bool need_sync(struct gfs2_quota_data *qd)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
struct gfs2_tune *gt = &sdp->sd_tune;
s64 value;
unsigned int num, den;
if (!qd->qd_qb.qb_limit)
return false;
spin_lock(&qd_lock);
value = qd->qd_change;
spin_unlock(&qd_lock);
spin_lock(&gt->gt_spin);
num = gt->gt_quota_scale_num;
den = gt->gt_quota_scale_den;
spin_unlock(&gt->gt_spin);
if (value <= 0)
return false;
else if ((s64)be64_to_cpu(qd->qd_qb.qb_value) >=
(s64)be64_to_cpu(qd->qd_qb.qb_limit))
return false;
else {
value *= gfs2_jindex_size(sdp) * num;
value = div_s64(value, den);
value += (s64)be64_to_cpu(qd->qd_qb.qb_value);
if (value < (s64)be64_to_cpu(qd->qd_qb.qb_limit))
return false;
}
return true;
}
void gfs2_quota_unlock(struct gfs2_inode *ip)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct gfs2_quota_data *qda[2 * GFS2_MAXQUOTAS];
unsigned int count = 0;
u32 x;
int found;
if (!test_and_clear_bit(GIF_QD_LOCKED, &ip->i_flags))
return;
for (x = 0; x < ip->i_qadata->qa_qd_num; x++) {
struct gfs2_quota_data *qd;
bool sync;
qd = ip->i_qadata->qa_qd[x];
sync = need_sync(qd);
gfs2_glock_dq_uninit(&ip->i_qadata->qa_qd_ghs[x]);
if (!sync)
continue;
spin_lock(&qd_lock);
found = qd_check_sync(sdp, qd, NULL);
spin_unlock(&qd_lock);
if (!found)
continue;
if (!qd_bh_get_or_undo(sdp, qd))
qda[count++] = qd;
}
if (count) {
do_sync(count, qda);
for (x = 0; x < count; x++)
qd_unlock(qda[x]);
}
gfs2_quota_unhold(ip);
}
#define MAX_LINE 256
static int print_message(struct gfs2_quota_data *qd, char *type)
{
struct gfs2_sbd *sdp = qd->qd_sbd;
if (sdp->sd_args.ar_quota != GFS2_QUOTA_QUIET)
fs_info(sdp, "quota %s for %s %u\n",
type,
(qd->qd_id.type == USRQUOTA) ? "user" : "group",
from_kqid(&init_user_ns, qd->qd_id));
return 0;
}
/**
* gfs2_quota_check - check if allocating new blocks will exceed quota
* @ip: The inode for which this check is being performed
* @uid: The uid to check against
* @gid: The gid to check against
* @ap: The allocation parameters. ap->target contains the requested
* blocks. ap->min_target, if set, contains the minimum blks
* requested.
*
* Returns: 0 on success.
* min_req = ap->min_target ? ap->min_target : ap->target;
* quota must allow at least min_req blks for success and
* ap->allowed is set to the number of blocks allowed
*
* -EDQUOT otherwise, quota violation. ap->allowed is set to number
* of blocks available.
*/
int gfs2_quota_check(struct gfs2_inode *ip, kuid_t uid, kgid_t gid,
struct gfs2_alloc_parms *ap)
{
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
struct gfs2_quota_data *qd;
s64 value, warn, limit;
u32 x;
int error = 0;
ap->allowed = UINT_MAX; /* Assume we are permitted a whole lot */
if (!test_bit(GIF_QD_LOCKED, &ip->i_flags))
return 0;
for (x = 0; x < ip->i_qadata->qa_qd_num; x++) {
qd = ip->i_qadata->qa_qd[x];
if (!(qid_eq(qd->qd_id, make_kqid_uid(uid)) ||
qid_eq(qd->qd_id, make_kqid_gid(gid))))
continue;
warn = (s64)be64_to_cpu(qd->qd_qb.qb_warn);
limit = (s64)be64_to_cpu(qd->qd_qb.qb_limit);
value = (s64)be64_to_cpu(qd->qd_qb.qb_value);
spin_lock(&qd_lock);
value += qd->qd_change;
spin_unlock(&qd_lock);
if (limit > 0 && (limit - value) < ap->allowed)
ap->allowed = limit - value;
/* If we can't meet the target */
if (limit && limit < (value + (s64)ap->target)) {
/* If no min_target specified or we don't meet
* min_target, return -EDQUOT */
if (!ap->min_target || ap->min_target > ap->allowed) {
if (!test_and_set_bit(QDF_QMSG_QUIET,
&qd->qd_flags)) {
print_message(qd, "exceeded");
quota_send_warning(qd->qd_id,
sdp->sd_vfs->s_dev,
QUOTA_NL_BHARDWARN);
}
error = -EDQUOT;
break;
}
} else if (warn && warn < value &&
time_after_eq(jiffies, qd->qd_last_warn +
gfs2_tune_get(sdp, gt_quota_warn_period)
* HZ)) {
quota_send_warning(qd->qd_id,
sdp->sd_vfs->s_dev, QUOTA_NL_BSOFTWARN);
error = print_message(qd, "warning");
qd->qd_last_warn = jiffies;
}
}
return error;
}
void gfs2_quota_change(struct gfs2_inode *ip, s64 change,
kuid_t uid, kgid_t gid)
{
struct gfs2_quota_data *qd;
u32 x;
struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
if ((sdp->sd_args.ar_quota != GFS2_QUOTA_ON &&
sdp->sd_args.ar_quota != GFS2_QUOTA_QUIET) ||
gfs2_assert_warn(sdp, change))
return;
if (ip->i_diskflags & GFS2_DIF_SYSTEM)
return;
if (gfs2_assert_withdraw(sdp, ip->i_qadata &&
ip->i_qadata->qa_ref > 0))
return;
for (x = 0; x < ip->i_qadata->qa_qd_num; x++) {
qd = ip->i_qadata->qa_qd[x];
if (qid_eq(qd->qd_id, make_kqid_uid(uid)) ||
qid_eq(qd->qd_id, make_kqid_gid(gid))) {
gfs2: Add quota_change type Function do_qc has two main uses: (1) to re-sync the local quota changes (qd) to the master quotas, and (2) normal quota changes. In the case of normal quota changes, the change can be positive or negative, as the quota usage goes up and down. Before this patch function do_qc was distinguishing one from another by whether the resulting value is or isn't zero: In the case of a re-sync (called do_sync) the quota value is moved from the temporary value to a master value, so the amount is added to one and subtracted from the other. The problem is that since the values can be positive or negative we can occasionally run into situations where we are not doing a re-sync but the quota change just happens to cancel out the previous value. In the case of a re-sync extra references and locks are taken, and so do_qc needs to release them. In the case of a normal quota change, no extra references and locks are taken, so it must not try to release them. The problem is: if the quota change is not a re-sync but the value just happens to cancel out the original quota change, the resulting zero value fools do_qc into thinking this is a re-sync and therefore it must release the extra references. This results in problems, mainly having to do with slot reference numbers going smaller than zero. This patch introduces new constants, QC_SYNC and QC_CHANGE so do_qc can really tell the difference. For QC_SYNC calls it must release the extra references acquired by gfs2_quota_unlock's call to qd_check_sync. For QC_CHANGE calls it does not have extra references to put. Note that this allows quota changes back to a value of zero, and so I removed an assert warning related to that. Signed-off-by: Bob Peterson <rpeterso@redhat.com> Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
2023-06-28 14:54:41 +00:00
do_qc(qd, change, QC_CHANGE);
}
}
}
int gfs2_quota_sync(struct super_block *sb, int type)
{
struct gfs2_sbd *sdp = sb->s_fs_info;
struct gfs2_quota_data **qda;
unsigned int max_qd = PAGE_SIZE / sizeof(struct gfs2_holder);
unsigned int num_qd;
unsigned int x;
int error = 0;
qda = kcalloc(max_qd, sizeof(struct gfs2_quota_data *), GFP_KERNEL);
if (!qda)
return -ENOMEM;
mutex_lock(&sdp->sd_quota_sync_mutex);
sdp->sd_quota_sync_gen++;
do {
num_qd = 0;
for (;;) {
error = qd_fish(sdp, qda + num_qd);
if (error || !qda[num_qd])
break;
if (++num_qd == max_qd)
break;
}
if (num_qd) {
if (!error)
error = do_sync(num_qd, qda);
for (x = 0; x < num_qd; x++)
qd_unlock(qda[x]);
}
} while (!error && num_qd == max_qd);
mutex_unlock(&sdp->sd_quota_sync_mutex);
kfree(qda);
return error;
}
int gfs2_quota_refresh(struct gfs2_sbd *sdp, struct kqid qid)
{
struct gfs2_quota_data *qd;
struct gfs2_holder q_gh;
int error;
error = qd_get(sdp, qid, &qd);
if (error)
return error;
error = do_glock(qd, FORCE, &q_gh);
if (!error)
gfs2_glock_dq_uninit(&q_gh);
qd_put(qd);
return error;
}
int gfs2_quota_init(struct gfs2_sbd *sdp)
{
struct gfs2_inode *ip = GFS2_I(sdp->sd_qc_inode);
u64 size = i_size_read(sdp->sd_qc_inode);
unsigned int blocks = size >> sdp->sd_sb.sb_bsize_shift;
unsigned int x, slot = 0;
unsigned int found = 0;
unsigned int hash;
unsigned int bm_size;
u64 dblock;
u32 extlen = 0;
int error;
if (gfs2_check_internal_file_size(sdp->sd_qc_inode, 1, 64 << 20))
return -EIO;
sdp->sd_quota_slots = blocks * sdp->sd_qc_per_block;
bm_size = DIV_ROUND_UP(sdp->sd_quota_slots, 8 * sizeof(unsigned long));
bm_size *= sizeof(unsigned long);
error = -ENOMEM;
sdp->sd_quota_bitmap = kzalloc(bm_size, GFP_NOFS | __GFP_NOWARN);
if (sdp->sd_quota_bitmap == NULL)
sdp->sd_quota_bitmap = __vmalloc(bm_size, GFP_NOFS |
2020-06-02 04:51:40 +00:00
__GFP_ZERO);
if (!sdp->sd_quota_bitmap)
return error;
for (x = 0; x < blocks; x++) {
struct buffer_head *bh;
const struct gfs2_quota_change *qc;
unsigned int y;
if (!extlen) {
extlen = 32;
error = gfs2_get_extent(&ip->i_inode, x, &dblock, &extlen);
if (error)
goto fail;
}
error = -EIO;
bh = gfs2_meta_ra(ip->i_gl, dblock, extlen);
if (!bh)
goto fail;
if (gfs2_metatype_check(sdp, bh, GFS2_METATYPE_QC)) {
brelse(bh);
goto fail;
}
qc = (const struct gfs2_quota_change *)(bh->b_data + sizeof(struct gfs2_meta_header));
for (y = 0; y < sdp->sd_qc_per_block && slot < sdp->sd_quota_slots;
y++, slot++) {
struct gfs2_quota_data *qd;
s64 qc_change = be64_to_cpu(qc->qc_change);
u32 qc_flags = be32_to_cpu(qc->qc_flags);
enum quota_type qtype = (qc_flags & GFS2_QCF_USER) ?
USRQUOTA : GRPQUOTA;
struct kqid qc_id = make_kqid(&init_user_ns, qtype,
be32_to_cpu(qc->qc_id));
qc++;
if (!qc_change)
continue;
hash = gfs2_qd_hash(sdp, qc_id);
qd = qd_alloc(hash, sdp, qc_id);
if (qd == NULL) {
brelse(bh);
goto fail;
}
set_bit(QDF_CHANGE, &qd->qd_flags);
qd->qd_change = qc_change;
qd->qd_slot = slot;
qd->qd_slot_count = 1;
spin_lock(&qd_lock);
BUG_ON(test_and_set_bit(slot, sdp->sd_quota_bitmap));
list_add(&qd->qd_list, &sdp->sd_quota_list);
atomic_inc(&sdp->sd_quota_count);
spin_unlock(&qd_lock);
spin_lock_bucket(hash);
hlist_bl_add_head_rcu(&qd->qd_hlist, &qd_hash_table[hash]);
spin_unlock_bucket(hash);
found++;
}
brelse(bh);
dblock++;
extlen--;
}
if (found)
fs_info(sdp, "found %u quota changes\n", found);
return 0;
fail:
gfs2_quota_cleanup(sdp);
return error;
}
void gfs2_quota_cleanup(struct gfs2_sbd *sdp)
{
struct gfs2_quota_data *qd;
LIST_HEAD(dispose);
int count;
BUG_ON(test_bit(SDF_JOURNAL_LIVE, &sdp->sd_flags));
spin_lock(&qd_lock);
list_for_each_entry(qd, &sdp->sd_quota_list, qd_list) {
spin_lock(&qd->qd_lockref.lock);
if (qd->qd_lockref.count != 0) {
spin_unlock(&qd->qd_lockref.lock);
continue;
}
lockref_mark_dead(&qd->qd_lockref);
spin_unlock(&qd->qd_lockref.lock);
list_lru_del(&gfs2_qd_lru, &qd->qd_lru);
list_add(&qd->qd_lru, &dispose);
}
spin_unlock(&qd_lock);
gfs2_qd_list_dispose(&dispose);
wait_event_timeout(sdp->sd_kill_wait,
(count = atomic_read(&sdp->sd_quota_count)) == 0,
HZ * 60);
if (count != 0)
fs_err(sdp, "%d left-over quota data objects\n", count);
kvfree(sdp->sd_quota_bitmap);
sdp->sd_quota_bitmap = NULL;
}
static void quotad_error(struct gfs2_sbd *sdp, const char *msg, int error)
{
if (error == 0 || error == -EROFS)
return;
if (!gfs2_withdrawn(sdp)) {
if (!cmpxchg(&sdp->sd_log_error, 0, error))
fs_err(sdp, "gfs2_quotad: %s error %d\n", msg, error);
wake_up(&sdp->sd_logd_waitq);
}
}
static void quotad_check_timeo(struct gfs2_sbd *sdp, const char *msg,
int (*fxn)(struct super_block *sb, int type),
unsigned long t, unsigned long *timeo,
unsigned int *new_timeo)
{
if (t >= *timeo) {
int error = fxn(sdp->sd_vfs, 0);
quotad_error(sdp, msg, error);
*timeo = gfs2_tune_get_i(&sdp->sd_tune, new_timeo) * HZ;
} else {
*timeo -= t;
}
}
void gfs2_wake_up_statfs(struct gfs2_sbd *sdp) {
if (!sdp->sd_statfs_force_sync) {
sdp->sd_statfs_force_sync = 1;
wake_up(&sdp->sd_quota_wait);
}
}
/**
* gfs2_quotad - Write cached quota changes into the quota file
* @data: Pointer to GFS2 superblock
*
*/
int gfs2_quotad(void *data)
{
struct gfs2_sbd *sdp = data;
struct gfs2_tune *tune = &sdp->sd_tune;
unsigned long statfs_timeo = 0;
unsigned long quotad_timeo = 0;
unsigned long t = 0;
while (!kthread_should_stop()) {
gfs2: Force withdraw to replay journals and wait for it to finish When a node withdraws from a file system, it often leaves its journal in an incomplete state. This is especially true when the withdraw is caused by io errors writing to the journal. Before this patch, a withdraw would try to write a "shutdown" record to the journal, tell dlm it's done with the file system, and none of the other nodes know about the problem. Later, when the problem is fixed and the withdrawn node is rebooted, it would then discover that its own journal was incomplete, and replay it. However, replaying it at this point is almost guaranteed to introduce corruption because the other nodes are likely to have used affected resource groups that appeared in the journal since the time of the withdraw. Replaying the journal later will overwrite any changes made, and not through any fault of dlm, which was instructed during the withdraw to release those resources. This patch makes file system withdraws seen by the entire cluster. Withdrawing nodes dequeue their journal glock to allow recovery. The remaining nodes check all the journals to see if they are clean or in need of replay. They try to replay dirty journals, but only the journals of withdrawn nodes will be "not busy" and therefore available for replay. Until the journal replay is complete, no i/o related glocks may be given out, to ensure that the replay does not cause the aforementioned corruption: We cannot allow any journal replay to overwrite blocks associated with a glock once it is held. The "live" glock which is now used to signal when a withdraw occurs. When a withdraw occurs, the node signals its withdraw by dequeueing the "live" glock and trying to enqueue it in EX mode, thus forcing the other nodes to all see a demote request, by way of a "1CB" (one callback) try lock. The "live" glock is not granted in EX; the callback is only just used to indicate a withdraw has occurred. Note that all nodes in the cluster must wait for the recovering node to finish replaying the withdrawing node's journal before continuing. To this end, it checks that the journals are clean multiple times in a retry loop. Also note that the withdraw function may be called from a wide variety of situations, and therefore, we need to take extra precautions to make sure pointers are valid before using them in many circumstances. We also need to take care when glocks decide to withdraw, since the withdraw code now uses glocks. Also, before this patch, if a process encountered an error and decided to withdraw, if another process was already withdrawing, the second withdraw would be silently ignored, which set it free to unlock its glocks. That's correct behavior if the original withdrawer encounters further errors down the road. But if secondary waiters don't wait for the journal replay, unlocking glocks will allow other nodes to use them, despite the fact that the journal containing those blocks is being replayed. The replay needs to finish before our glocks are released to other nodes. IOW, secondary withdraws need to wait for the first withdraw to finish. For example, if an rgrp glock is unlocked by a process that didn't wait for the first withdraw, a journal replay could introduce file system corruption by replaying a rgrp block that has already been granted to a different cluster node. Signed-off-by: Bob Peterson <rpeterso@redhat.com>
2020-01-28 19:23:45 +00:00
if (gfs2_withdrawn(sdp))
break;
/* Update the master statfs file */
if (sdp->sd_statfs_force_sync) {
int error = gfs2_statfs_sync(sdp->sd_vfs, 0);
quotad_error(sdp, "statfs", error);
statfs_timeo = gfs2_tune_get(sdp, gt_statfs_quantum) * HZ;
}
else
quotad_check_timeo(sdp, "statfs", gfs2_statfs_sync, t,
&statfs_timeo,
&tune->gt_statfs_quantum);
/* Update quota file */
quotad_check_timeo(sdp, "sync", gfs2_quota_sync, t,
&quotad_timeo, &tune->gt_quota_quantum);
try_to_freeze();
t = min(quotad_timeo, statfs_timeo);
t = wait_event_interruptible_timeout(sdp->sd_quota_wait,
sdp->sd_statfs_force_sync ||
gfs2_withdrawn(sdp) ||
kthread_should_stop(),
t);
if (sdp->sd_statfs_force_sync)
t = 0;
}
return 0;
}
static int gfs2_quota_get_state(struct super_block *sb, struct qc_state *state)
{
struct gfs2_sbd *sdp = sb->s_fs_info;
memset(state, 0, sizeof(*state));
switch (sdp->sd_args.ar_quota) {
case GFS2_QUOTA_QUIET:
fallthrough;
case GFS2_QUOTA_ON:
state->s_state[USRQUOTA].flags |= QCI_LIMITS_ENFORCED;
state->s_state[GRPQUOTA].flags |= QCI_LIMITS_ENFORCED;
fallthrough;
case GFS2_QUOTA_ACCOUNT:
state->s_state[USRQUOTA].flags |= QCI_ACCT_ENABLED |
QCI_SYSFILE;
state->s_state[GRPQUOTA].flags |= QCI_ACCT_ENABLED |
QCI_SYSFILE;
break;
case GFS2_QUOTA_OFF:
break;
}
if (sdp->sd_quota_inode) {
state->s_state[USRQUOTA].ino =
GFS2_I(sdp->sd_quota_inode)->i_no_addr;
state->s_state[USRQUOTA].blocks = sdp->sd_quota_inode->i_blocks;
}
state->s_state[USRQUOTA].nextents = 1; /* unsupported */
state->s_state[GRPQUOTA] = state->s_state[USRQUOTA];
state->s_incoredqs = list_lru_count(&gfs2_qd_lru);
return 0;
}
static int gfs2_get_dqblk(struct super_block *sb, struct kqid qid,
struct qc_dqblk *fdq)
{
struct gfs2_sbd *sdp = sb->s_fs_info;
struct gfs2_quota_lvb *qlvb;
struct gfs2_quota_data *qd;
struct gfs2_holder q_gh;
int error;
memset(fdq, 0, sizeof(*fdq));
if (sdp->sd_args.ar_quota == GFS2_QUOTA_OFF)
return -ESRCH; /* Crazy XFS error code */
if ((qid.type != USRQUOTA) &&
(qid.type != GRPQUOTA))
return -EINVAL;
error = qd_get(sdp, qid, &qd);
if (error)
return error;
error = do_glock(qd, FORCE, &q_gh);
if (error)
goto out;
qlvb = (struct gfs2_quota_lvb *)qd->qd_gl->gl_lksb.sb_lvbptr;
fdq->d_spc_hardlimit = be64_to_cpu(qlvb->qb_limit) << sdp->sd_sb.sb_bsize_shift;
fdq->d_spc_softlimit = be64_to_cpu(qlvb->qb_warn) << sdp->sd_sb.sb_bsize_shift;
fdq->d_space = be64_to_cpu(qlvb->qb_value) << sdp->sd_sb.sb_bsize_shift;
gfs2_glock_dq_uninit(&q_gh);
out:
qd_put(qd);
return error;
}
/* GFS2 only supports a subset of the XFS fields */
#define GFS2_FIELDMASK (QC_SPC_SOFT|QC_SPC_HARD|QC_SPACE)
static int gfs2_set_dqblk(struct super_block *sb, struct kqid qid,
struct qc_dqblk *fdq)
{
struct gfs2_sbd *sdp = sb->s_fs_info;
struct gfs2_inode *ip = GFS2_I(sdp->sd_quota_inode);
struct gfs2_quota_data *qd;
struct gfs2_holder q_gh, i_gh;
unsigned int data_blocks, ind_blocks;
unsigned int blocks = 0;
int alloc_required;
loff_t offset;
int error;
if (sdp->sd_args.ar_quota == GFS2_QUOTA_OFF)
return -ESRCH; /* Crazy XFS error code */
if ((qid.type != USRQUOTA) &&
(qid.type != GRPQUOTA))
return -EINVAL;
if (fdq->d_fieldmask & ~GFS2_FIELDMASK)
return -EINVAL;
error = qd_get(sdp, qid, &qd);
if (error)
return error;
error = gfs2_qa_get(ip);
if (error)
goto out_put;
inode_lock(&ip->i_inode);
error = gfs2_glock_nq_init(qd->qd_gl, LM_ST_EXCLUSIVE, 0, &q_gh);
if (error)
goto out_unlockput;
error = gfs2_glock_nq_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, &i_gh);
if (error)
goto out_q;
/* Check for existing entry, if none then alloc new blocks */
error = update_qd(sdp, qd);
if (error)
goto out_i;
/* If nothing has changed, this is a no-op */
if ((fdq->d_fieldmask & QC_SPC_SOFT) &&
((fdq->d_spc_softlimit >> sdp->sd_sb.sb_bsize_shift) == be64_to_cpu(qd->qd_qb.qb_warn)))
fdq->d_fieldmask ^= QC_SPC_SOFT;
if ((fdq->d_fieldmask & QC_SPC_HARD) &&
((fdq->d_spc_hardlimit >> sdp->sd_sb.sb_bsize_shift) == be64_to_cpu(qd->qd_qb.qb_limit)))
fdq->d_fieldmask ^= QC_SPC_HARD;
if ((fdq->d_fieldmask & QC_SPACE) &&
((fdq->d_space >> sdp->sd_sb.sb_bsize_shift) == be64_to_cpu(qd->qd_qb.qb_value)))
fdq->d_fieldmask ^= QC_SPACE;
if (fdq->d_fieldmask == 0)
goto out_i;
offset = qd2offset(qd);
alloc_required = gfs2_write_alloc_required(ip, offset, sizeof(struct gfs2_quota));
if (gfs2_is_stuffed(ip))
alloc_required = 1;
if (alloc_required) {
struct gfs2_alloc_parms ap = { .aflags = 0, };
gfs2_write_calc_reserv(ip, sizeof(struct gfs2_quota),
&data_blocks, &ind_blocks);
blocks = 1 + data_blocks + ind_blocks;
ap.target = blocks;
error = gfs2_inplace_reserve(ip, &ap);
if (error)
goto out_i;
blocks += gfs2_rg_blocks(ip, blocks);
}
/* Some quotas span block boundaries and can update two blocks,
adding an extra block to the transaction to handle such quotas */
error = gfs2_trans_begin(sdp, blocks + RES_DINODE + 2, 0);
if (error)
goto out_release;
/* Apply changes */
error = gfs2_adjust_quota(sdp, offset, 0, qd, fdq);
if (!error)
clear_bit(QDF_QMSG_QUIET, &qd->qd_flags);
gfs2_trans_end(sdp);
out_release:
if (alloc_required)
gfs2_inplace_release(ip);
out_i:
gfs2_glock_dq_uninit(&i_gh);
out_q:
gfs2_glock_dq_uninit(&q_gh);
out_unlockput:
gfs2_qa_put(ip);
inode_unlock(&ip->i_inode);
out_put:
qd_put(qd);
return error;
}
const struct quotactl_ops gfs2_quotactl_ops = {
.quota_sync = gfs2_quota_sync,
.get_state = gfs2_quota_get_state,
.get_dqblk = gfs2_get_dqblk,
.set_dqblk = gfs2_set_dqblk,
};
void __init gfs2_quota_hash_init(void)
{
unsigned i;
for(i = 0; i < GFS2_QD_HASH_SIZE; i++)
INIT_HLIST_BL_HEAD(&qd_hash_table[i]);
}