linux/fs/xfs/xfs_qm.c

1956 lines
46 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
* All Rights Reserved.
*/
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
#include "xfs_bit.h"
#include "xfs_sb.h"
#include "xfs_mount.h"
#include "xfs_inode.h"
#include "xfs_iwalk.h"
#include "xfs_quota.h"
#include "xfs_bmap.h"
#include "xfs_bmap_util.h"
#include "xfs_trans.h"
#include "xfs_trans_space.h"
#include "xfs_qm.h"
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
#include "xfs_trace.h"
#include "xfs_icache.h"
#include "xfs_error.h"
/*
* The global quota manager. There is only one of these for the entire
* system, _not_ one per file system. XQM keeps track of the overall
* quota functionality, including maintaining the freelist and hash
* tables of dquots.
*/
STATIC int xfs_qm_init_quotainos(struct xfs_mount *mp);
STATIC int xfs_qm_init_quotainfo(struct xfs_mount *mp);
STATIC void xfs_qm_destroy_quotainos(struct xfs_quotainfo *qi);
STATIC void xfs_qm_dqfree_one(struct xfs_dquot *dqp);
/*
* We use the batch lookup interface to iterate over the dquots as it
* currently is the only interface into the radix tree code that allows
* fuzzy lookups instead of exact matches. Holding the lock over multiple
* operations is fine as all callers are used either during mount/umount
* or quotaoff.
*/
#define XFS_DQ_LOOKUP_BATCH 32
STATIC int
xfs_qm_dquot_walk(
struct xfs_mount *mp,
int type,
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
int (*execute)(struct xfs_dquot *dqp, void *data),
void *data)
{
struct xfs_quotainfo *qi = mp->m_quotainfo;
struct radix_tree_root *tree = xfs_dquot_tree(qi, type);
uint32_t next_index;
int last_error = 0;
int skipped;
int nr_found;
restart:
skipped = 0;
next_index = 0;
nr_found = 0;
while (1) {
struct xfs_dquot *batch[XFS_DQ_LOOKUP_BATCH];
int error = 0;
int i;
mutex_lock(&qi->qi_tree_lock);
nr_found = radix_tree_gang_lookup(tree, (void **)batch,
next_index, XFS_DQ_LOOKUP_BATCH);
if (!nr_found) {
mutex_unlock(&qi->qi_tree_lock);
break;
}
for (i = 0; i < nr_found; i++) {
struct xfs_dquot *dqp = batch[i];
next_index = be32_to_cpu(dqp->q_core.d_id) + 1;
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
error = execute(batch[i], data);
if (error == -EAGAIN) {
skipped++;
continue;
}
if (error && last_error != -EFSCORRUPTED)
last_error = error;
}
mutex_unlock(&qi->qi_tree_lock);
/* bail out if the filesystem is corrupted. */
if (last_error == -EFSCORRUPTED) {
skipped = 0;
break;
}
/* we're done if id overflows back to zero */
if (!next_index)
break;
}
if (skipped) {
delay(1);
goto restart;
}
return last_error;
}
/*
* Purge a dquot from all tracking data structures and free it.
*/
STATIC int
xfs_qm_dqpurge(
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_dquot *dqp,
void *data)
{
struct xfs_mount *mp = dqp->q_mount;
struct xfs_quotainfo *qi = mp->m_quotainfo;
int error = -EAGAIN;
xfs_dqlock(dqp);
if ((dqp->dq_flags & XFS_DQ_FREEING) || dqp->q_nrefs != 0)
goto out_unlock;
dqp->dq_flags |= XFS_DQ_FREEING;
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
xfs_dqflock(dqp);
/*
* If we are turning this type of quotas off, we don't care
* about the dirty metadata sitting in this dquot. OTOH, if
* we're unmounting, we do care, so we flush it and wait.
*/
if (XFS_DQ_IS_DIRTY(dqp)) {
struct xfs_buf *bp = NULL;
/*
* We don't care about getting disk errors here. We need
* to purge this dquot anyway, so we go ahead regardless.
*/
error = xfs_qm_dqflush(dqp, &bp);
if (!error) {
error = xfs_bwrite(bp);
xfs_buf_relse(bp);
} else if (error == -EAGAIN) {
goto out_unlock;
}
xfs_dqflock(dqp);
}
ASSERT(atomic_read(&dqp->q_pincount) == 0);
ASSERT(XFS_FORCED_SHUTDOWN(mp) ||
!test_bit(XFS_LI_IN_AIL, &dqp->q_logitem.qli_item.li_flags));
xfs_dqfunlock(dqp);
xfs_dqunlock(dqp);
radix_tree_delete(xfs_dquot_tree(qi, dqp->q_core.d_flags),
be32_to_cpu(dqp->q_core.d_id));
qi->qi_dquots--;
/*
* We move dquots to the freelist as soon as their reference count
* hits zero, so it really should be on the freelist here.
*/
ASSERT(!list_empty(&dqp->q_lru));
list_lru_del(&qi->qi_lru, &dqp->q_lru);
XFS_STATS_DEC(mp, xs_qm_dquot_unused);
xfs_qm_dqdestroy(dqp);
xfs: fix infinite loop by detaching the group/project hints from user dquot xfs_quota(8) will hang up if trying to turn group/project quota off before the user quota is off, this could be 100% reproduced by: # mount -ouquota,gquota /dev/sda7 /xfs # mkdir /xfs/test # xfs_quota -xc 'off -g' /xfs <-- hangs up # echo w > /proc/sysrq-trigger # dmesg SysRq : Show Blocked State task PC stack pid father xfs_quota D 0000000000000000 0 27574 2551 0x00000000 [snip] Call Trace: [<ffffffff81aaa21d>] schedule+0xad/0xc0 [<ffffffff81aa327e>] schedule_timeout+0x35e/0x3c0 [<ffffffff8114b506>] ? mark_held_locks+0x176/0x1c0 [<ffffffff810ad6c0>] ? call_timer_fn+0x2c0/0x2c0 [<ffffffffa0c25380>] ? xfs_qm_shrink_count+0x30/0x30 [xfs] [<ffffffff81aa3306>] schedule_timeout_uninterruptible+0x26/0x30 [<ffffffffa0c26155>] xfs_qm_dquot_walk+0x235/0x260 [xfs] [<ffffffffa0c059d8>] ? xfs_perag_get+0x1d8/0x2d0 [xfs] [<ffffffffa0c05805>] ? xfs_perag_get+0x5/0x2d0 [xfs] [<ffffffffa0b7707e>] ? xfs_inode_ag_iterator+0xae/0xf0 [xfs] [<ffffffffa0c22280>] ? xfs_trans_free_dqinfo+0x50/0x50 [xfs] [<ffffffffa0b7709f>] ? xfs_inode_ag_iterator+0xcf/0xf0 [xfs] [<ffffffffa0c261e6>] xfs_qm_dqpurge_all+0x66/0xb0 [xfs] [<ffffffffa0c2497a>] xfs_qm_scall_quotaoff+0x20a/0x5f0 [xfs] [<ffffffffa0c2b8f6>] xfs_fs_set_xstate+0x136/0x180 [xfs] [<ffffffff8136cf7a>] do_quotactl+0x53a/0x6b0 [<ffffffff812fba4b>] ? iput+0x5b/0x90 [<ffffffff8136d257>] SyS_quotactl+0x167/0x1d0 [<ffffffff814cf2ee>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff81abcd19>] system_call_fastpath+0x16/0x1b It's fine if we turn user quota off at first, then turn off other kind of quotas if they are enabled since the group/project dquot refcount is decreased to zero once the user quota if off. Otherwise, those dquots refcount is non-zero due to the user dquot might refer to them as hint(s). Hence, above operation cause an infinite loop at xfs_qm_dquot_walk() while trying to purge dquot cache. This problem has been around since Linux 3.4, it was introduced by: [ b84a3a9675 xfs: remove the per-filesystem list of dquots ] Originally we will release the group dquot pointers because the user dquots maybe carrying around as a hint via xfs_qm_detach_gdquots(). However, with above change, there is no such work to be done before purging group/project dquot cache. In order to solve this problem, this patch introduces a special routine xfs_qm_dqpurge_hints(), and it would release the group/project dquot pointers the user dquots maybe carrying around as a hint, and then it will proceed to purge the user dquot cache if requested. Cc: stable@vger.kernel.org Signed-off-by: Jie Liu <jeff.liu@oracle.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-26 13:38:49 +00:00
return 0;
out_unlock:
xfs_dqunlock(dqp);
return error;
xfs: fix infinite loop by detaching the group/project hints from user dquot xfs_quota(8) will hang up if trying to turn group/project quota off before the user quota is off, this could be 100% reproduced by: # mount -ouquota,gquota /dev/sda7 /xfs # mkdir /xfs/test # xfs_quota -xc 'off -g' /xfs <-- hangs up # echo w > /proc/sysrq-trigger # dmesg SysRq : Show Blocked State task PC stack pid father xfs_quota D 0000000000000000 0 27574 2551 0x00000000 [snip] Call Trace: [<ffffffff81aaa21d>] schedule+0xad/0xc0 [<ffffffff81aa327e>] schedule_timeout+0x35e/0x3c0 [<ffffffff8114b506>] ? mark_held_locks+0x176/0x1c0 [<ffffffff810ad6c0>] ? call_timer_fn+0x2c0/0x2c0 [<ffffffffa0c25380>] ? xfs_qm_shrink_count+0x30/0x30 [xfs] [<ffffffff81aa3306>] schedule_timeout_uninterruptible+0x26/0x30 [<ffffffffa0c26155>] xfs_qm_dquot_walk+0x235/0x260 [xfs] [<ffffffffa0c059d8>] ? xfs_perag_get+0x1d8/0x2d0 [xfs] [<ffffffffa0c05805>] ? xfs_perag_get+0x5/0x2d0 [xfs] [<ffffffffa0b7707e>] ? xfs_inode_ag_iterator+0xae/0xf0 [xfs] [<ffffffffa0c22280>] ? xfs_trans_free_dqinfo+0x50/0x50 [xfs] [<ffffffffa0b7709f>] ? xfs_inode_ag_iterator+0xcf/0xf0 [xfs] [<ffffffffa0c261e6>] xfs_qm_dqpurge_all+0x66/0xb0 [xfs] [<ffffffffa0c2497a>] xfs_qm_scall_quotaoff+0x20a/0x5f0 [xfs] [<ffffffffa0c2b8f6>] xfs_fs_set_xstate+0x136/0x180 [xfs] [<ffffffff8136cf7a>] do_quotactl+0x53a/0x6b0 [<ffffffff812fba4b>] ? iput+0x5b/0x90 [<ffffffff8136d257>] SyS_quotactl+0x167/0x1d0 [<ffffffff814cf2ee>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff81abcd19>] system_call_fastpath+0x16/0x1b It's fine if we turn user quota off at first, then turn off other kind of quotas if they are enabled since the group/project dquot refcount is decreased to zero once the user quota if off. Otherwise, those dquots refcount is non-zero due to the user dquot might refer to them as hint(s). Hence, above operation cause an infinite loop at xfs_qm_dquot_walk() while trying to purge dquot cache. This problem has been around since Linux 3.4, it was introduced by: [ b84a3a9675 xfs: remove the per-filesystem list of dquots ] Originally we will release the group dquot pointers because the user dquots maybe carrying around as a hint via xfs_qm_detach_gdquots(). However, with above change, there is no such work to be done before purging group/project dquot cache. In order to solve this problem, this patch introduces a special routine xfs_qm_dqpurge_hints(), and it would release the group/project dquot pointers the user dquots maybe carrying around as a hint, and then it will proceed to purge the user dquot cache if requested. Cc: stable@vger.kernel.org Signed-off-by: Jie Liu <jeff.liu@oracle.com> Reviewed-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-26 13:38:49 +00:00
}
/*
* Purge the dquot cache.
*/
void
xfs_qm_dqpurge_all(
struct xfs_mount *mp,
uint flags)
{
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
if (flags & XFS_QMOPT_UQUOTA)
xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_dqpurge, NULL);
if (flags & XFS_QMOPT_GQUOTA)
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_dqpurge, NULL);
if (flags & XFS_QMOPT_PQUOTA)
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_dqpurge, NULL);
}
/*
* Just destroy the quotainfo structure.
*/
void
xfs_qm_unmount(
struct xfs_mount *mp)
{
if (mp->m_quotainfo) {
xfs_qm_dqpurge_all(mp, XFS_QMOPT_QUOTALL);
xfs_qm_destroy_quotainfo(mp);
}
}
/*
* Called from the vfsops layer.
*/
void
xfs_qm_unmount_quotas(
xfs_mount_t *mp)
{
/*
* Release the dquots that root inode, et al might be holding,
* before we flush quotas and blow away the quotainfo structure.
*/
ASSERT(mp->m_rootip);
xfs_qm_dqdetach(mp->m_rootip);
if (mp->m_rbmip)
xfs_qm_dqdetach(mp->m_rbmip);
if (mp->m_rsumip)
xfs_qm_dqdetach(mp->m_rsumip);
/*
* Release the quota inodes.
*/
if (mp->m_quotainfo) {
if (mp->m_quotainfo->qi_uquotaip) {
xfs_irele(mp->m_quotainfo->qi_uquotaip);
mp->m_quotainfo->qi_uquotaip = NULL;
}
if (mp->m_quotainfo->qi_gquotaip) {
xfs_irele(mp->m_quotainfo->qi_gquotaip);
mp->m_quotainfo->qi_gquotaip = NULL;
}
if (mp->m_quotainfo->qi_pquotaip) {
xfs_irele(mp->m_quotainfo->qi_pquotaip);
mp->m_quotainfo->qi_pquotaip = NULL;
}
}
}
STATIC int
xfs_qm_dqattach_one(
struct xfs_inode *ip,
xfs_dqid_t id,
uint type,
bool doalloc,
struct xfs_dquot **IO_idqpp)
{
struct xfs_dquot *dqp;
int error;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
error = 0;
/*
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
* See if we already have it in the inode itself. IO_idqpp is &i_udquot
* or &i_gdquot. This made the code look weird, but made the logic a lot
* simpler.
*/
dqp = *IO_idqpp;
if (dqp) {
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dqattach_found(dqp);
return 0;
}
/*
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
* Find the dquot from somewhere. This bumps the reference count of
* dquot and returns it locked. This can return ENOENT if dquot didn't
* exist on disk and we didn't ask it to allocate; ESRCH if quotas got
* turned off suddenly.
*/
error = xfs_qm_dqget_inode(ip, type, doalloc, &dqp);
if (error)
return error;
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dqattach_get(dqp);
/*
* dqget may have dropped and re-acquired the ilock, but it guarantees
* that the dquot returned is the one that should go in the inode.
*/
*IO_idqpp = dqp;
xfs_dqunlock(dqp);
return 0;
}
static bool
xfs_qm_need_dqattach(
struct xfs_inode *ip)
{
struct xfs_mount *mp = ip->i_mount;
if (!XFS_IS_QUOTA_RUNNING(mp))
return false;
if (!XFS_IS_QUOTA_ON(mp))
return false;
if (!XFS_NOT_DQATTACHED(mp, ip))
return false;
if (xfs_is_quota_inode(&mp->m_sb, ip->i_ino))
return false;
return true;
}
/*
* Given a locked inode, attach dquot(s) to it, taking U/G/P-QUOTAON
* into account.
* If @doalloc is true, the dquot(s) will be allocated if needed.
* Inode may get unlocked and relocked in here, and the caller must deal with
* the consequences.
*/
int
xfs_qm_dqattach_locked(
xfs_inode_t *ip,
bool doalloc)
{
xfs_mount_t *mp = ip->i_mount;
int error = 0;
if (!xfs_qm_need_dqattach(ip))
return 0;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
if (XFS_IS_UQUOTA_ON(mp) && !ip->i_udquot) {
error = xfs_qm_dqattach_one(ip, i_uid_read(VFS_I(ip)),
XFS_DQ_USER, doalloc, &ip->i_udquot);
if (error)
goto done;
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
ASSERT(ip->i_udquot);
}
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
if (XFS_IS_GQUOTA_ON(mp) && !ip->i_gdquot) {
error = xfs_qm_dqattach_one(ip, i_gid_read(VFS_I(ip)),
XFS_DQ_GROUP, doalloc, &ip->i_gdquot);
if (error)
goto done;
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
ASSERT(ip->i_gdquot);
}
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
if (XFS_IS_PQUOTA_ON(mp) && !ip->i_pdquot) {
error = xfs_qm_dqattach_one(ip, ip->i_d.di_projid, XFS_DQ_PROJ,
doalloc, &ip->i_pdquot);
if (error)
goto done;
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
ASSERT(ip->i_pdquot);
}
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
done:
/*
xfs: remove dquot hints group and project quota hints are currently stored on the user dquot. If we are attaching quotas to the inode, then the group and project dquots are stored as hints on the user dquot to save having to look them up again later. The thing is, the hints are not used for that inode for the rest of the life of the inode - the dquots are attached directly to the inode itself - so the only time the hints are used is when an inode first has dquots attached. When the hints on the user dquot don't match the dquots being attache dto the inode, they are then removed and replaced with the new hints. If a user is concurrently modifying files in different group and/or project contexts, then this leads to thrashing of the hints attached to user dquot. If user quotas are not enabled, then hints are never even used. So, if the hints are used to avoid the cost of the lookup, is the cost of the lookup significant enough to justify the hint infrstructure? Maybe it was once, when there was a global quota manager shared between all XFS filesystems and was hash table based. However, lookups are now much simpler, requiring only a single lock and radix tree lookup local to the filesystem and no hash or LRU manipulations to be made. Hence the cost of lookup is much lower than when hints were implemented. Turns out that benchmarks show that, too, with thir being no differnce in performance when doing file creation workloads as a single user with user, group and project quotas enabled - the hints do not make the code go any faster. In fact, removing the hints shows a 2-3% reduction in the time it takes to create 50 million inodes.... So, let's just get rid of the hints and the complexity around them. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-05-05 07:30:15 +00:00
* Don't worry about the dquots that we may have attached before any
* error - they'll get detached later if it has not already been done.
*/
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
return error;
}
int
xfs_qm_dqattach(
struct xfs_inode *ip)
{
int error;
if (!xfs_qm_need_dqattach(ip))
return 0;
xfs_ilock(ip, XFS_ILOCK_EXCL);
error = xfs_qm_dqattach_locked(ip, false);
xfs_iunlock(ip, XFS_ILOCK_EXCL);
return error;
}
/*
* Release dquots (and their references) if any.
* The inode should be locked EXCL except when this's called by
* xfs_ireclaim.
*/
void
xfs_qm_dqdetach(
xfs_inode_t *ip)
{
if (!(ip->i_udquot || ip->i_gdquot || ip->i_pdquot))
return;
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dquot_dqdetach(ip);
ASSERT(!xfs_is_quota_inode(&ip->i_mount->m_sb, ip->i_ino));
if (ip->i_udquot) {
xfs_qm_dqrele(ip->i_udquot);
ip->i_udquot = NULL;
}
if (ip->i_gdquot) {
xfs_qm_dqrele(ip->i_gdquot);
ip->i_gdquot = NULL;
}
if (ip->i_pdquot) {
xfs_qm_dqrele(ip->i_pdquot);
ip->i_pdquot = NULL;
}
}
struct xfs_qm_isolate {
struct list_head buffers;
struct list_head dispose;
};
static enum lru_status
xfs_qm_dquot_isolate(
struct list_head *item,
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
struct list_lru_one *lru,
spinlock_t *lru_lock,
void *arg)
__releases(lru_lock) __acquires(lru_lock)
{
struct xfs_dquot *dqp = container_of(item,
struct xfs_dquot, q_lru);
struct xfs_qm_isolate *isol = arg;
if (!xfs_dqlock_nowait(dqp))
goto out_miss_busy;
/*
* This dquot has acquired a reference in the meantime remove it from
* the freelist and try again.
*/
if (dqp->q_nrefs) {
xfs_dqunlock(dqp);
XFS_STATS_INC(dqp->q_mount, xs_qm_dqwants);
trace_xfs_dqreclaim_want(dqp);
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(lru, &dqp->q_lru);
XFS_STATS_DEC(dqp->q_mount, xs_qm_dquot_unused);
return LRU_REMOVED;
}
/*
* If the dquot is dirty, flush it. If it's already being flushed, just
* skip it so there is time for the IO to complete before we try to
* reclaim it again on the next LRU pass.
*/
if (!xfs_dqflock_nowait(dqp)) {
xfs_dqunlock(dqp);
goto out_miss_busy;
}
if (XFS_DQ_IS_DIRTY(dqp)) {
struct xfs_buf *bp = NULL;
int error;
trace_xfs_dqreclaim_dirty(dqp);
/* we have to drop the LRU lock to flush the dquot */
spin_unlock(lru_lock);
error = xfs_qm_dqflush(dqp, &bp);
if (error)
goto out_unlock_dirty;
xfs_buf_delwri_queue(bp, &isol->buffers);
xfs_buf_relse(bp);
goto out_unlock_dirty;
}
xfs_dqfunlock(dqp);
/*
* Prevent lookups now that we are past the point of no return.
*/
dqp->dq_flags |= XFS_DQ_FREEING;
xfs_dqunlock(dqp);
ASSERT(dqp->q_nrefs == 0);
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, &dqp->q_lru, &isol->dispose);
XFS_STATS_DEC(dqp->q_mount, xs_qm_dquot_unused);
trace_xfs_dqreclaim_done(dqp);
XFS_STATS_INC(dqp->q_mount, xs_qm_dqreclaims);
return LRU_REMOVED;
out_miss_busy:
trace_xfs_dqreclaim_busy(dqp);
XFS_STATS_INC(dqp->q_mount, xs_qm_dqreclaim_misses);
return LRU_SKIP;
out_unlock_dirty:
trace_xfs_dqreclaim_busy(dqp);
XFS_STATS_INC(dqp->q_mount, xs_qm_dqreclaim_misses);
xfs_dqunlock(dqp);
spin_lock(lru_lock);
return LRU_RETRY;
}
static unsigned long
xfs_qm_shrink_scan(
struct shrinker *shrink,
struct shrink_control *sc)
{
struct xfs_quotainfo *qi = container_of(shrink,
struct xfs_quotainfo, qi_shrinker);
struct xfs_qm_isolate isol;
unsigned long freed;
int error;
mm, page_alloc: distinguish between being unable to sleep, unwilling to sleep and avoiding waking kswapd __GFP_WAIT has been used to identify atomic context in callers that hold spinlocks or are in interrupts. They are expected to be high priority and have access one of two watermarks lower than "min" which can be referred to as the "atomic reserve". __GFP_HIGH users get access to the first lower watermark and can be called the "high priority reserve". Over time, callers had a requirement to not block when fallback options were available. Some have abused __GFP_WAIT leading to a situation where an optimisitic allocation with a fallback option can access atomic reserves. This patch uses __GFP_ATOMIC to identify callers that are truely atomic, cannot sleep and have no alternative. High priority users continue to use __GFP_HIGH. __GFP_DIRECT_RECLAIM identifies callers that can sleep and are willing to enter direct reclaim. __GFP_KSWAPD_RECLAIM to identify callers that want to wake kswapd for background reclaim. __GFP_WAIT is redefined as a caller that is willing to enter direct reclaim and wake kswapd for background reclaim. This patch then converts a number of sites o __GFP_ATOMIC is used by callers that are high priority and have memory pools for those requests. GFP_ATOMIC uses this flag. o Callers that have a limited mempool to guarantee forward progress clear __GFP_DIRECT_RECLAIM but keep __GFP_KSWAPD_RECLAIM. bio allocations fall into this category where kswapd will still be woken but atomic reserves are not used as there is a one-entry mempool to guarantee progress. o Callers that are checking if they are non-blocking should use the helper gfpflags_allow_blocking() where possible. This is because checking for __GFP_WAIT as was done historically now can trigger false positives. Some exceptions like dm-crypt.c exist where the code intent is clearer if __GFP_DIRECT_RECLAIM is used instead of the helper due to flag manipulations. o Callers that built their own GFP flags instead of starting with GFP_KERNEL and friends now also need to specify __GFP_KSWAPD_RECLAIM. The first key hazard to watch out for is callers that removed __GFP_WAIT and was depending on access to atomic reserves for inconspicuous reasons. In some cases it may be appropriate for them to use __GFP_HIGH. The second key hazard is callers that assembled their own combination of GFP flags instead of starting with something like GFP_KERNEL. They may now wish to specify __GFP_KSWAPD_RECLAIM. It's almost certainly harmless if it's missed in most cases as other activity will wake kswapd. Signed-off-by: Mel Gorman <mgorman@techsingularity.net> Acked-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Michal Hocko <mhocko@suse.com> Acked-by: Johannes Weiner <hannes@cmpxchg.org> Cc: Christoph Lameter <cl@linux.com> Cc: David Rientjes <rientjes@google.com> Cc: Vitaly Wool <vitalywool@gmail.com> Cc: Rik van Riel <riel@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-11-07 00:28:21 +00:00
if ((sc->gfp_mask & (__GFP_FS|__GFP_DIRECT_RECLAIM)) != (__GFP_FS|__GFP_DIRECT_RECLAIM))
return 0;
INIT_LIST_HEAD(&isol.buffers);
INIT_LIST_HEAD(&isol.dispose);
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(&qi->qi_lru, sc,
xfs_qm_dquot_isolate, &isol);
error = xfs_buf_delwri_submit(&isol.buffers);
if (error)
xfs_warn(NULL, "%s: dquot reclaim failed", __func__);
while (!list_empty(&isol.dispose)) {
struct xfs_dquot *dqp;
dqp = list_first_entry(&isol.dispose, struct xfs_dquot, q_lru);
list_del_init(&dqp->q_lru);
xfs_qm_dqfree_one(dqp);
}
return freed;
}
static unsigned long
xfs_qm_shrink_count(
struct shrinker *shrink,
struct shrink_control *sc)
{
struct xfs_quotainfo *qi = container_of(shrink,
struct xfs_quotainfo, qi_shrinker);
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 list_lru_shrink_count(&qi->qi_lru, sc);
}
STATIC void
xfs_qm_set_defquota(
struct xfs_mount *mp,
uint type,
struct xfs_quotainfo *qinf)
{
struct xfs_dquot *dqp;
struct xfs_def_quota *defq;
struct xfs_disk_dquot *ddqp;
int error;
error = xfs_qm_dqget_uncached(mp, 0, type, &dqp);
if (error)
return;
ddqp = &dqp->q_core;
defq = xfs_get_defquota(dqp, qinf);
/*
* Timers and warnings have been already set, let's just set the
* default limits for this quota type
*/
defq->bhardlimit = be64_to_cpu(ddqp->d_blk_hardlimit);
defq->bsoftlimit = be64_to_cpu(ddqp->d_blk_softlimit);
defq->ihardlimit = be64_to_cpu(ddqp->d_ino_hardlimit);
defq->isoftlimit = be64_to_cpu(ddqp->d_ino_softlimit);
defq->rtbhardlimit = be64_to_cpu(ddqp->d_rtb_hardlimit);
defq->rtbsoftlimit = be64_to_cpu(ddqp->d_rtb_softlimit);
xfs_qm_dqdestroy(dqp);
}
/* Initialize quota time limits from the root dquot. */
static void
xfs_qm_init_timelimits(
struct xfs_mount *mp,
struct xfs_quotainfo *qinf)
{
struct xfs_disk_dquot *ddqp;
struct xfs_dquot *dqp;
uint type;
int error;
qinf->qi_btimelimit = XFS_QM_BTIMELIMIT;
qinf->qi_itimelimit = XFS_QM_ITIMELIMIT;
qinf->qi_rtbtimelimit = XFS_QM_RTBTIMELIMIT;
qinf->qi_bwarnlimit = XFS_QM_BWARNLIMIT;
qinf->qi_iwarnlimit = XFS_QM_IWARNLIMIT;
qinf->qi_rtbwarnlimit = XFS_QM_RTBWARNLIMIT;
/*
* We try to get the limits from the superuser's limits fields.
* This is quite hacky, but it is standard quota practice.
*
* Since we may not have done a quotacheck by this point, just read
* the dquot without attaching it to any hashtables or lists.
*
* Timers and warnings are globally set by the first timer found in
* user/group/proj quota types, otherwise a default value is used.
* This should be split into different fields per quota type.
*/
if (XFS_IS_UQUOTA_RUNNING(mp))
type = XFS_DQ_USER;
else if (XFS_IS_GQUOTA_RUNNING(mp))
type = XFS_DQ_GROUP;
else
type = XFS_DQ_PROJ;
error = xfs_qm_dqget_uncached(mp, 0, type, &dqp);
if (error)
return;
ddqp = &dqp->q_core;
/*
* The warnings and timers set the grace period given to
* a user or group before he or she can not perform any
* more writing. If it is zero, a default is used.
*/
if (ddqp->d_btimer)
qinf->qi_btimelimit = be32_to_cpu(ddqp->d_btimer);
if (ddqp->d_itimer)
qinf->qi_itimelimit = be32_to_cpu(ddqp->d_itimer);
if (ddqp->d_rtbtimer)
qinf->qi_rtbtimelimit = be32_to_cpu(ddqp->d_rtbtimer);
if (ddqp->d_bwarns)
qinf->qi_bwarnlimit = be16_to_cpu(ddqp->d_bwarns);
if (ddqp->d_iwarns)
qinf->qi_iwarnlimit = be16_to_cpu(ddqp->d_iwarns);
if (ddqp->d_rtbwarns)
qinf->qi_rtbwarnlimit = be16_to_cpu(ddqp->d_rtbwarns);
xfs_qm_dqdestroy(dqp);
}
/*
* This initializes all the quota information that's kept in the
* mount structure
*/
STATIC int
xfs_qm_init_quotainfo(
struct xfs_mount *mp)
{
struct xfs_quotainfo *qinf;
int error;
ASSERT(XFS_IS_QUOTA_RUNNING(mp));
qinf = mp->m_quotainfo = kmem_zalloc(sizeof(struct xfs_quotainfo), 0);
error = list_lru_init(&qinf->qi_lru);
if (error)
goto out_free_qinf;
list_lru: dynamically adjust node arrays We currently use a compile-time constant to size the node array for the list_lru structure. Due to this, we don't need to allocate any memory at initialization time. But as a consequence, the structures that contain embedded list_lru lists can become way too big (the superblock for instance contains two of them). This patch aims at ameliorating this situation by dynamically allocating the node arrays with the firmware provided nr_node_ids. Signed-off-by: Glauber Costa <glommer@openvz.org> Cc: Dave Chinner <dchinner@redhat.com> Cc: Mel Gorman <mgorman@suse.de> 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 00:18:18 +00:00
/*
* See if quotainodes are setup, and if not, allocate them,
* and change the superblock accordingly.
*/
error = xfs_qm_init_quotainos(mp);
if (error)
goto out_free_lru;
INIT_RADIX_TREE(&qinf->qi_uquota_tree, GFP_NOFS);
INIT_RADIX_TREE(&qinf->qi_gquota_tree, GFP_NOFS);
INIT_RADIX_TREE(&qinf->qi_pquota_tree, GFP_NOFS);
mutex_init(&qinf->qi_tree_lock);
/* mutex used to serialize quotaoffs */
mutex_init(&qinf->qi_quotaofflock);
/* Precalc some constants */
qinf->qi_dqchunklen = XFS_FSB_TO_BB(mp, XFS_DQUOT_CLUSTER_SIZE_FSB);
qinf->qi_dqperchunk = xfs_calc_dquots_per_chunk(qinf->qi_dqchunklen);
mp->m_qflags |= (mp->m_sb.sb_qflags & XFS_ALL_QUOTA_CHKD);
xfs_qm_init_timelimits(mp, qinf);
if (XFS_IS_UQUOTA_RUNNING(mp))
xfs_qm_set_defquota(mp, XFS_DQ_USER, qinf);
if (XFS_IS_GQUOTA_RUNNING(mp))
xfs_qm_set_defquota(mp, XFS_DQ_GROUP, qinf);
if (XFS_IS_PQUOTA_RUNNING(mp))
xfs_qm_set_defquota(mp, XFS_DQ_PROJ, qinf);
qinf->qi_shrinker.count_objects = xfs_qm_shrink_count;
qinf->qi_shrinker.scan_objects = xfs_qm_shrink_scan;
qinf->qi_shrinker.seeks = DEFAULT_SEEKS;
qinf->qi_shrinker.flags = SHRINKER_NUMA_AWARE;
error = register_shrinker(&qinf->qi_shrinker);
if (error)
goto out_free_inos;
return 0;
out_free_inos:
mutex_destroy(&qinf->qi_quotaofflock);
mutex_destroy(&qinf->qi_tree_lock);
xfs_qm_destroy_quotainos(qinf);
out_free_lru:
list_lru_destroy(&qinf->qi_lru);
out_free_qinf:
kmem_free(qinf);
mp->m_quotainfo = NULL;
return error;
}
/*
* Gets called when unmounting a filesystem or when all quotas get
* turned off.
* This purges the quota inodes, destroys locks and frees itself.
*/
void
xfs_qm_destroy_quotainfo(
struct xfs_mount *mp)
{
struct xfs_quotainfo *qi;
qi = mp->m_quotainfo;
ASSERT(qi != NULL);
unregister_shrinker(&qi->qi_shrinker);
list_lru_destroy(&qi->qi_lru);
xfs_qm_destroy_quotainos(qi);
mutex_destroy(&qi->qi_tree_lock);
mutex_destroy(&qi->qi_quotaofflock);
kmem_free(qi);
mp->m_quotainfo = NULL;
}
/*
* Create an inode and return with a reference already taken, but unlocked
* This is how we create quota inodes
*/
STATIC int
xfs_qm_qino_alloc(
xfs_mount_t *mp,
xfs_inode_t **ip,
uint flags)
{
xfs_trans_t *tp;
int error;
xfs: inodes are new until the dentry cache is set up Al Viro noticed a generic set of issues to do with filehandle lookup racing with dentry cache setup. They involve a filehandle lookup occurring while an inode is being created and the filehandle lookup racing with the dentry creation for the real file. This can lead to multiple dentries for the one path being instantiated. There are a host of other issues around this same set of paths. The underlying cause is that file handle lookup only waits on inode cache instantiation rather than full dentry cache instantiation. XFS is mostly immune to the problems discovered due to it's own internal inode cache, but there are a couple of corner cases where races can happen. We currently clear the XFS_INEW flag when the inode is fully set up after insertion into the cache. Newly allocated inodes are inserted locked and so aren't usable until the allocation transaction commits. This, however, occurs before the dentry and security information is fully initialised and hence the inode is unlocked and available for lookups to find too early. To solve the problem, only clear the XFS_INEW flag for newly created inodes once the dentry is fully instantiated. This means lookups will retry until the XFS_INEW flag is removed from the inode and hence avoids the race conditions in questions. THis also means that xfs_create(), xfs_create_tmpfile() and xfs_symlink() need to finish the setup of the inode in their error paths if we had allocated the inode but failed later in the creation process. xfs_symlink(), in particular, needed a lot of help to make it's error handling match that of xfs_create(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-02-23 11:38:08 +00:00
bool need_alloc = true;
*ip = NULL;
/*
* With superblock that doesn't have separate pquotino, we
* share an inode between gquota and pquota. If the on-disk
* superblock has GQUOTA and the filesystem is now mounted
* with PQUOTA, just use sb_gquotino for sb_pquotino and
* vice-versa.
*/
if (!xfs_sb_version_has_pquotino(&mp->m_sb) &&
(flags & (XFS_QMOPT_PQUOTA|XFS_QMOPT_GQUOTA))) {
xfs_ino_t ino = NULLFSINO;
if ((flags & XFS_QMOPT_PQUOTA) &&
(mp->m_sb.sb_gquotino != NULLFSINO)) {
ino = mp->m_sb.sb_gquotino;
if (XFS_IS_CORRUPT(mp,
mp->m_sb.sb_pquotino != NULLFSINO))
return -EFSCORRUPTED;
} else if ((flags & XFS_QMOPT_GQUOTA) &&
(mp->m_sb.sb_pquotino != NULLFSINO)) {
ino = mp->m_sb.sb_pquotino;
if (XFS_IS_CORRUPT(mp,
mp->m_sb.sb_gquotino != NULLFSINO))
return -EFSCORRUPTED;
}
if (ino != NULLFSINO) {
error = xfs_iget(mp, NULL, ino, 0, 0, ip);
if (error)
return error;
mp->m_sb.sb_gquotino = NULLFSINO;
mp->m_sb.sb_pquotino = NULLFSINO;
xfs: inodes are new until the dentry cache is set up Al Viro noticed a generic set of issues to do with filehandle lookup racing with dentry cache setup. They involve a filehandle lookup occurring while an inode is being created and the filehandle lookup racing with the dentry creation for the real file. This can lead to multiple dentries for the one path being instantiated. There are a host of other issues around this same set of paths. The underlying cause is that file handle lookup only waits on inode cache instantiation rather than full dentry cache instantiation. XFS is mostly immune to the problems discovered due to it's own internal inode cache, but there are a couple of corner cases where races can happen. We currently clear the XFS_INEW flag when the inode is fully set up after insertion into the cache. Newly allocated inodes are inserted locked and so aren't usable until the allocation transaction commits. This, however, occurs before the dentry and security information is fully initialised and hence the inode is unlocked and available for lookups to find too early. To solve the problem, only clear the XFS_INEW flag for newly created inodes once the dentry is fully instantiated. This means lookups will retry until the XFS_INEW flag is removed from the inode and hence avoids the race conditions in questions. THis also means that xfs_create(), xfs_create_tmpfile() and xfs_symlink() need to finish the setup of the inode in their error paths if we had allocated the inode but failed later in the creation process. xfs_symlink(), in particular, needed a lot of help to make it's error handling match that of xfs_create(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-02-23 11:38:08 +00:00
need_alloc = false;
}
}
error = xfs_trans_alloc(mp, &M_RES(mp)->tr_create,
need_alloc ? XFS_QM_QINOCREATE_SPACE_RES(mp) : 0,
0, 0, &tp);
if (error)
return error;
xfs: inodes are new until the dentry cache is set up Al Viro noticed a generic set of issues to do with filehandle lookup racing with dentry cache setup. They involve a filehandle lookup occurring while an inode is being created and the filehandle lookup racing with the dentry creation for the real file. This can lead to multiple dentries for the one path being instantiated. There are a host of other issues around this same set of paths. The underlying cause is that file handle lookup only waits on inode cache instantiation rather than full dentry cache instantiation. XFS is mostly immune to the problems discovered due to it's own internal inode cache, but there are a couple of corner cases where races can happen. We currently clear the XFS_INEW flag when the inode is fully set up after insertion into the cache. Newly allocated inodes are inserted locked and so aren't usable until the allocation transaction commits. This, however, occurs before the dentry and security information is fully initialised and hence the inode is unlocked and available for lookups to find too early. To solve the problem, only clear the XFS_INEW flag for newly created inodes once the dentry is fully instantiated. This means lookups will retry until the XFS_INEW flag is removed from the inode and hence avoids the race conditions in questions. THis also means that xfs_create(), xfs_create_tmpfile() and xfs_symlink() need to finish the setup of the inode in their error paths if we had allocated the inode but failed later in the creation process. xfs_symlink(), in particular, needed a lot of help to make it's error handling match that of xfs_create(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-02-23 11:38:08 +00:00
if (need_alloc) {
error = xfs_dir_ialloc(&tp, NULL, S_IFREG, 1, 0, 0, ip);
if (error) {
xfs_trans_cancel(tp);
return error;
}
}
/*
* Make the changes in the superblock, and log those too.
* sbfields arg may contain fields other than *QUOTINO;
* VERSIONNUM for example.
*/
spin_lock(&mp->m_sb_lock);
if (flags & XFS_QMOPT_SBVERSION) {
ASSERT(!xfs_sb_version_hasquota(&mp->m_sb));
xfs_sb_version_addquota(&mp->m_sb);
mp->m_sb.sb_uquotino = NULLFSINO;
mp->m_sb.sb_gquotino = NULLFSINO;
mp->m_sb.sb_pquotino = NULLFSINO;
/* qflags will get updated fully _after_ quotacheck */
mp->m_sb.sb_qflags = mp->m_qflags & XFS_ALL_QUOTA_ACCT;
}
if (flags & XFS_QMOPT_UQUOTA)
mp->m_sb.sb_uquotino = (*ip)->i_ino;
else if (flags & XFS_QMOPT_GQUOTA)
mp->m_sb.sb_gquotino = (*ip)->i_ino;
else
mp->m_sb.sb_pquotino = (*ip)->i_ino;
spin_unlock(&mp->m_sb_lock);
xfs_log_sb(tp);
error = xfs_trans_commit(tp);
xfs: inodes are new until the dentry cache is set up Al Viro noticed a generic set of issues to do with filehandle lookup racing with dentry cache setup. They involve a filehandle lookup occurring while an inode is being created and the filehandle lookup racing with the dentry creation for the real file. This can lead to multiple dentries for the one path being instantiated. There are a host of other issues around this same set of paths. The underlying cause is that file handle lookup only waits on inode cache instantiation rather than full dentry cache instantiation. XFS is mostly immune to the problems discovered due to it's own internal inode cache, but there are a couple of corner cases where races can happen. We currently clear the XFS_INEW flag when the inode is fully set up after insertion into the cache. Newly allocated inodes are inserted locked and so aren't usable until the allocation transaction commits. This, however, occurs before the dentry and security information is fully initialised and hence the inode is unlocked and available for lookups to find too early. To solve the problem, only clear the XFS_INEW flag for newly created inodes once the dentry is fully instantiated. This means lookups will retry until the XFS_INEW flag is removed from the inode and hence avoids the race conditions in questions. THis also means that xfs_create(), xfs_create_tmpfile() and xfs_symlink() need to finish the setup of the inode in their error paths if we had allocated the inode but failed later in the creation process. xfs_symlink(), in particular, needed a lot of help to make it's error handling match that of xfs_create(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-02-23 11:38:08 +00:00
if (error) {
ASSERT(XFS_FORCED_SHUTDOWN(mp));
xfs_alert(mp, "%s failed (error %d)!", __func__, error);
}
xfs: inodes are new until the dentry cache is set up Al Viro noticed a generic set of issues to do with filehandle lookup racing with dentry cache setup. They involve a filehandle lookup occurring while an inode is being created and the filehandle lookup racing with the dentry creation for the real file. This can lead to multiple dentries for the one path being instantiated. There are a host of other issues around this same set of paths. The underlying cause is that file handle lookup only waits on inode cache instantiation rather than full dentry cache instantiation. XFS is mostly immune to the problems discovered due to it's own internal inode cache, but there are a couple of corner cases where races can happen. We currently clear the XFS_INEW flag when the inode is fully set up after insertion into the cache. Newly allocated inodes are inserted locked and so aren't usable until the allocation transaction commits. This, however, occurs before the dentry and security information is fully initialised and hence the inode is unlocked and available for lookups to find too early. To solve the problem, only clear the XFS_INEW flag for newly created inodes once the dentry is fully instantiated. This means lookups will retry until the XFS_INEW flag is removed from the inode and hence avoids the race conditions in questions. THis also means that xfs_create(), xfs_create_tmpfile() and xfs_symlink() need to finish the setup of the inode in their error paths if we had allocated the inode but failed later in the creation process. xfs_symlink(), in particular, needed a lot of help to make it's error handling match that of xfs_create(). Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Signed-off-by: Dave Chinner <david@fromorbit.com>
2015-02-23 11:38:08 +00:00
if (need_alloc)
xfs_finish_inode_setup(*ip);
return error;
}
STATIC void
xfs_qm_reset_dqcounts(
xfs_mount_t *mp,
xfs_buf_t *bp,
xfs_dqid_t id,
uint type)
{
struct xfs_dqblk *dqb;
int j;
xfs_failaddr_t fa;
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_reset_dqcounts(bp, _RET_IP_);
/*
* Reset all counters and timers. They'll be
* started afresh by xfs_qm_quotacheck.
*/
#ifdef DEBUG
j = (int)XFS_FSB_TO_B(mp, XFS_DQUOT_CLUSTER_SIZE_FSB) /
sizeof(xfs_dqblk_t);
ASSERT(mp->m_quotainfo->qi_dqperchunk == j);
#endif
dqb = bp->b_addr;
for (j = 0; j < mp->m_quotainfo->qi_dqperchunk; j++) {
struct xfs_disk_dquot *ddq;
ddq = (struct xfs_disk_dquot *)&dqb[j];
/*
* Do a sanity check, and if needed, repair the dqblk. Don't
* output any warnings because it's perfectly possible to
* find uninitialised dquot blks. See comment in
* xfs_dquot_verify.
*/
fa = xfs_dqblk_verify(mp, &dqb[j], id + j, type);
if (fa)
xfs_dqblk_repair(mp, &dqb[j], id + j, type);
/*
* Reset type in case we are reusing group quota file for
* project quotas or vice versa
*/
ddq->d_flags = type;
ddq->d_bcount = 0;
ddq->d_icount = 0;
ddq->d_rtbcount = 0;
/*
* dquot id 0 stores the default grace period and the maximum
* warning limit that were set by the administrator, so we
* should not reset them.
*/
if (ddq->d_id != 0) {
ddq->d_btimer = 0;
ddq->d_itimer = 0;
ddq->d_rtbtimer = 0;
ddq->d_bwarns = 0;
ddq->d_iwarns = 0;
ddq->d_rtbwarns = 0;
}
if (xfs_sb_version_hascrc(&mp->m_sb)) {
xfs_update_cksum((char *)&dqb[j],
sizeof(struct xfs_dqblk),
XFS_DQUOT_CRC_OFF);
}
}
}
STATIC int
xfs_qm_reset_dqcounts_all(
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_mount *mp,
xfs_dqid_t firstid,
xfs_fsblock_t bno,
xfs_filblks_t blkcnt,
uint flags,
struct list_head *buffer_list)
{
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_buf *bp;
int error;
int type;
ASSERT(blkcnt > 0);
type = flags & XFS_QMOPT_UQUOTA ? XFS_DQ_USER :
(flags & XFS_QMOPT_PQUOTA ? XFS_DQ_PROJ : XFS_DQ_GROUP);
error = 0;
/*
* Blkcnt arg can be a very big number, and might even be
* larger than the log itself. So, we have to break it up into
* manageable-sized transactions.
* Note that we don't start a permanent transaction here; we might
* not be able to get a log reservation for the whole thing up front,
* and we don't really care to either, because we just discard
* everything if we were to crash in the middle of this loop.
*/
while (blkcnt--) {
error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, bno),
mp->m_quotainfo->qi_dqchunklen, 0, &bp,
&xfs_dquot_buf_ops);
/*
* CRC and validation errors will return a EFSCORRUPTED here. If
* this occurs, re-read without CRC validation so that we can
* repair the damage via xfs_qm_reset_dqcounts(). This process
* will leave a trace in the log indicating corruption has
* been detected.
*/
if (error == -EFSCORRUPTED) {
error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, bno),
mp->m_quotainfo->qi_dqchunklen, 0, &bp,
NULL);
}
if (error)
break;
xfs: quotacheck leaves dquot buffers without verifiers When running xfs/305, I noticed that quotacheck was flushing dquot buffers that did not have the xfs_dquot_buf_ops verifiers attached: XFS (vdb): _xfs_buf_ioapply: no ops on block 0x1dc8/0x1dc8 ffff880052489000: 44 51 01 04 00 00 65 b8 00 00 00 00 00 00 00 00 DQ....e......... ffff880052489010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ ffff880052489020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ ffff880052489030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ CPU: 1 PID: 2376 Comm: mount Not tainted 3.16.0-rc2-dgc+ #306 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 ffff88006fe38000 ffff88004a0ffae8 ffffffff81cf1cca 0000000000000001 ffff88004a0ffb88 ffffffff814d50ca 000010004a0ffc70 0000000000000000 ffff88006be56dc4 0000000000000021 0000000000001dc8 ffff88007c773d80 Call Trace: [<ffffffff81cf1cca>] dump_stack+0x45/0x56 [<ffffffff814d50ca>] _xfs_buf_ioapply+0x3ca/0x3d0 [<ffffffff810db520>] ? wake_up_state+0x20/0x20 [<ffffffff814d51f5>] ? xfs_bdstrat_cb+0x55/0xb0 [<ffffffff814d513b>] xfs_buf_iorequest+0x6b/0xd0 [<ffffffff814d51f5>] xfs_bdstrat_cb+0x55/0xb0 [<ffffffff814d53ab>] __xfs_buf_delwri_submit+0x15b/0x220 [<ffffffff814d6040>] ? xfs_buf_delwri_submit+0x30/0x90 [<ffffffff814d6040>] xfs_buf_delwri_submit+0x30/0x90 [<ffffffff8150f89d>] xfs_qm_quotacheck+0x17d/0x3c0 [<ffffffff81510591>] xfs_qm_mount_quotas+0x151/0x1e0 [<ffffffff814ed01c>] xfs_mountfs+0x56c/0x7d0 [<ffffffff814f0f12>] xfs_fs_fill_super+0x2c2/0x340 [<ffffffff811c9fe4>] mount_bdev+0x194/0x1d0 [<ffffffff814f0c50>] ? xfs_finish_flags+0x170/0x170 [<ffffffff814ef0f5>] xfs_fs_mount+0x15/0x20 [<ffffffff811ca8c9>] mount_fs+0x39/0x1b0 [<ffffffff811e4d67>] vfs_kern_mount+0x67/0x120 [<ffffffff811e757e>] do_mount+0x23e/0xad0 [<ffffffff8117abde>] ? __get_free_pages+0xe/0x50 [<ffffffff811e71e6>] ? copy_mount_options+0x36/0x150 [<ffffffff811e8103>] SyS_mount+0x83/0xc0 [<ffffffff81cfd40b>] tracesys+0xdd/0xe2 This was caused by dquot buffer readahead not attaching a verifier structure to the buffer when readahead was issued, resulting in the followup read of the buffer finding a valid buffer and so not attaching new verifiers to the buffer as part of the read. Also, when a verifier failure occurs, we then read the buffer without verifiers. Attach the verifiers manually after this read so that if the buffer is then written it will be verified that the corruption has been repaired. Further, when flushing a dquot we don't ask for a verifier when reading in the dquot buffer the dquot belongs to. Most of the time this isn't an issue because the buffer is still cached, but when it is not cached it will result in writing the dquot buffer without having the verfier attached. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-08-04 02:43:26 +00:00
/*
* A corrupt buffer might not have a verifier attached, so
* make sure we have the correct one attached before writeback
* occurs.
*/
bp->b_ops = &xfs_dquot_buf_ops;
xfs_qm_reset_dqcounts(mp, bp, firstid, type);
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
xfs_buf_delwri_queue(bp, buffer_list);
xfs_buf_relse(bp);
/* goto the next block. */
bno++;
firstid += mp->m_quotainfo->qi_dqperchunk;
}
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
return error;
}
/*
* Iterate over all allocated dquot blocks in this quota inode, zeroing all
* counters for every chunk of dquots that we find.
*/
STATIC int
xfs_qm_reset_dqcounts_buf(
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_mount *mp,
struct xfs_inode *qip,
uint flags,
struct list_head *buffer_list)
{
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_bmbt_irec *map;
int i, nmaps; /* number of map entries */
int error; /* return value */
xfs_fileoff_t lblkno;
xfs_filblks_t maxlblkcnt;
xfs_dqid_t firstid;
xfs_fsblock_t rablkno;
xfs_filblks_t rablkcnt;
error = 0;
/*
* This looks racy, but we can't keep an inode lock across a
* trans_reserve. But, this gets called during quotacheck, and that
* happens only at mount time which is single threaded.
*/
if (qip->i_d.di_nblocks == 0)
return 0;
map = kmem_alloc(XFS_DQITER_MAP_SIZE * sizeof(*map), 0);
lblkno = 0;
maxlblkcnt = XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
do {
uint lock_mode;
nmaps = XFS_DQITER_MAP_SIZE;
/*
* We aren't changing the inode itself. Just changing
* some of its data. No new blocks are added here, and
* the inode is never added to the transaction.
*/
lock_mode = xfs_ilock_data_map_shared(qip);
error = xfs_bmapi_read(qip, lblkno, maxlblkcnt - lblkno,
map, &nmaps, 0);
xfs_iunlock(qip, lock_mode);
if (error)
break;
ASSERT(nmaps <= XFS_DQITER_MAP_SIZE);
for (i = 0; i < nmaps; i++) {
ASSERT(map[i].br_startblock != DELAYSTARTBLOCK);
ASSERT(map[i].br_blockcount);
lblkno += map[i].br_blockcount;
if (map[i].br_startblock == HOLESTARTBLOCK)
continue;
firstid = (xfs_dqid_t) map[i].br_startoff *
mp->m_quotainfo->qi_dqperchunk;
/*
* Do a read-ahead on the next extent.
*/
if ((i+1 < nmaps) &&
(map[i+1].br_startblock != HOLESTARTBLOCK)) {
rablkcnt = map[i+1].br_blockcount;
rablkno = map[i+1].br_startblock;
while (rablkcnt--) {
xfs_buf_readahead(mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, rablkno),
mp->m_quotainfo->qi_dqchunklen,
xfs: quotacheck leaves dquot buffers without verifiers When running xfs/305, I noticed that quotacheck was flushing dquot buffers that did not have the xfs_dquot_buf_ops verifiers attached: XFS (vdb): _xfs_buf_ioapply: no ops on block 0x1dc8/0x1dc8 ffff880052489000: 44 51 01 04 00 00 65 b8 00 00 00 00 00 00 00 00 DQ....e......... ffff880052489010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ ffff880052489020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ ffff880052489030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ CPU: 1 PID: 2376 Comm: mount Not tainted 3.16.0-rc2-dgc+ #306 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 ffff88006fe38000 ffff88004a0ffae8 ffffffff81cf1cca 0000000000000001 ffff88004a0ffb88 ffffffff814d50ca 000010004a0ffc70 0000000000000000 ffff88006be56dc4 0000000000000021 0000000000001dc8 ffff88007c773d80 Call Trace: [<ffffffff81cf1cca>] dump_stack+0x45/0x56 [<ffffffff814d50ca>] _xfs_buf_ioapply+0x3ca/0x3d0 [<ffffffff810db520>] ? wake_up_state+0x20/0x20 [<ffffffff814d51f5>] ? xfs_bdstrat_cb+0x55/0xb0 [<ffffffff814d513b>] xfs_buf_iorequest+0x6b/0xd0 [<ffffffff814d51f5>] xfs_bdstrat_cb+0x55/0xb0 [<ffffffff814d53ab>] __xfs_buf_delwri_submit+0x15b/0x220 [<ffffffff814d6040>] ? xfs_buf_delwri_submit+0x30/0x90 [<ffffffff814d6040>] xfs_buf_delwri_submit+0x30/0x90 [<ffffffff8150f89d>] xfs_qm_quotacheck+0x17d/0x3c0 [<ffffffff81510591>] xfs_qm_mount_quotas+0x151/0x1e0 [<ffffffff814ed01c>] xfs_mountfs+0x56c/0x7d0 [<ffffffff814f0f12>] xfs_fs_fill_super+0x2c2/0x340 [<ffffffff811c9fe4>] mount_bdev+0x194/0x1d0 [<ffffffff814f0c50>] ? xfs_finish_flags+0x170/0x170 [<ffffffff814ef0f5>] xfs_fs_mount+0x15/0x20 [<ffffffff811ca8c9>] mount_fs+0x39/0x1b0 [<ffffffff811e4d67>] vfs_kern_mount+0x67/0x120 [<ffffffff811e757e>] do_mount+0x23e/0xad0 [<ffffffff8117abde>] ? __get_free_pages+0xe/0x50 [<ffffffff811e71e6>] ? copy_mount_options+0x36/0x150 [<ffffffff811e8103>] SyS_mount+0x83/0xc0 [<ffffffff81cfd40b>] tracesys+0xdd/0xe2 This was caused by dquot buffer readahead not attaching a verifier structure to the buffer when readahead was issued, resulting in the followup read of the buffer finding a valid buffer and so not attaching new verifiers to the buffer as part of the read. Also, when a verifier failure occurs, we then read the buffer without verifiers. Attach the verifiers manually after this read so that if the buffer is then written it will be verified that the corruption has been repaired. Further, when flushing a dquot we don't ask for a verifier when reading in the dquot buffer the dquot belongs to. Most of the time this isn't an issue because the buffer is still cached, but when it is not cached it will result in writing the dquot buffer without having the verfier attached. Signed-off-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Dave Chinner <david@fromorbit.com>
2014-08-04 02:43:26 +00:00
&xfs_dquot_buf_ops);
rablkno++;
}
}
/*
* Iterate thru all the blks in the extent and
* reset the counters of all the dquots inside them.
*/
error = xfs_qm_reset_dqcounts_all(mp, firstid,
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
map[i].br_startblock,
map[i].br_blockcount,
flags, buffer_list);
if (error)
goto out;
}
} while (nmaps > 0);
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
out:
kmem_free(map);
return error;
}
/*
* Called by dqusage_adjust in doing a quotacheck.
*
* Given the inode, and a dquot id this updates both the incore dqout as well
* as the buffer copy. This is so that once the quotacheck is done, we can
* just log all the buffers, as opposed to logging numerous updates to
* individual dquots.
*/
STATIC int
xfs_qm_quotacheck_dqadjust(
struct xfs_inode *ip,
uint type,
xfs_qcnt_t nblks,
xfs_qcnt_t rtblks)
{
struct xfs_mount *mp = ip->i_mount;
struct xfs_dquot *dqp;
xfs_dqid_t id;
int error;
id = xfs_qm_id_for_quotatype(ip, type);
error = xfs_qm_dqget(mp, id, type, true, &dqp);
if (error) {
/*
* Shouldn't be able to turn off quotas here.
*/
ASSERT(error != -ESRCH);
ASSERT(error != -ENOENT);
return error;
}
xfs: event tracing support Convert the old xfs tracing support that could only be used with the out of tree kdb and xfsidbg patches to use the generic event tracer. To use it make sure CONFIG_EVENT_TRACING is enabled and then enable all xfs trace channels by: echo 1 > /sys/kernel/debug/tracing/events/xfs/enable or alternatively enable single events by just doing the same in one event subdirectory, e.g. echo 1 > /sys/kernel/debug/tracing/events/xfs/xfs_ihold/enable or set more complex filters, etc. In Documentation/trace/events.txt all this is desctribed in more detail. To reads the events do a cat /sys/kernel/debug/tracing/trace Compared to the last posting this patch converts the tracing mostly to the one tracepoint per callsite model that other users of the new tracing facility also employ. This allows a very fine-grained control of the tracing, a cleaner output of the traces and also enables the perf tool to use each tracepoint as a virtual performance counter, allowing us to e.g. count how often certain workloads git various spots in XFS. Take a look at http://lwn.net/Articles/346470/ for some examples. Also the btree tracing isn't included at all yet, as it will require additional core tracing features not in mainline yet, I plan to deliver it later. And the really nice thing about this patch is that it actually removes many lines of code while adding this nice functionality: fs/xfs/Makefile | 8 fs/xfs/linux-2.6/xfs_acl.c | 1 fs/xfs/linux-2.6/xfs_aops.c | 52 - fs/xfs/linux-2.6/xfs_aops.h | 2 fs/xfs/linux-2.6/xfs_buf.c | 117 +-- fs/xfs/linux-2.6/xfs_buf.h | 33 fs/xfs/linux-2.6/xfs_fs_subr.c | 3 fs/xfs/linux-2.6/xfs_ioctl.c | 1 fs/xfs/linux-2.6/xfs_ioctl32.c | 1 fs/xfs/linux-2.6/xfs_iops.c | 1 fs/xfs/linux-2.6/xfs_linux.h | 1 fs/xfs/linux-2.6/xfs_lrw.c | 87 -- fs/xfs/linux-2.6/xfs_lrw.h | 45 - fs/xfs/linux-2.6/xfs_super.c | 104 --- fs/xfs/linux-2.6/xfs_super.h | 7 fs/xfs/linux-2.6/xfs_sync.c | 1 fs/xfs/linux-2.6/xfs_trace.c | 75 ++ fs/xfs/linux-2.6/xfs_trace.h | 1369 +++++++++++++++++++++++++++++++++++++++++ fs/xfs/linux-2.6/xfs_vnode.h | 4 fs/xfs/quota/xfs_dquot.c | 110 --- fs/xfs/quota/xfs_dquot.h | 21 fs/xfs/quota/xfs_qm.c | 40 - fs/xfs/quota/xfs_qm_syscalls.c | 4 fs/xfs/support/ktrace.c | 323 --------- fs/xfs/support/ktrace.h | 85 -- fs/xfs/xfs.h | 16 fs/xfs/xfs_ag.h | 14 fs/xfs/xfs_alloc.c | 230 +----- fs/xfs/xfs_alloc.h | 27 fs/xfs/xfs_alloc_btree.c | 1 fs/xfs/xfs_attr.c | 107 --- fs/xfs/xfs_attr.h | 10 fs/xfs/xfs_attr_leaf.c | 14 fs/xfs/xfs_attr_sf.h | 40 - fs/xfs/xfs_bmap.c | 507 +++------------ fs/xfs/xfs_bmap.h | 49 - fs/xfs/xfs_bmap_btree.c | 6 fs/xfs/xfs_btree.c | 5 fs/xfs/xfs_btree_trace.h | 17 fs/xfs/xfs_buf_item.c | 87 -- fs/xfs/xfs_buf_item.h | 20 fs/xfs/xfs_da_btree.c | 3 fs/xfs/xfs_da_btree.h | 7 fs/xfs/xfs_dfrag.c | 2 fs/xfs/xfs_dir2.c | 8 fs/xfs/xfs_dir2_block.c | 20 fs/xfs/xfs_dir2_leaf.c | 21 fs/xfs/xfs_dir2_node.c | 27 fs/xfs/xfs_dir2_sf.c | 26 fs/xfs/xfs_dir2_trace.c | 216 ------ fs/xfs/xfs_dir2_trace.h | 72 -- fs/xfs/xfs_filestream.c | 8 fs/xfs/xfs_fsops.c | 2 fs/xfs/xfs_iget.c | 111 --- fs/xfs/xfs_inode.c | 67 -- fs/xfs/xfs_inode.h | 76 -- fs/xfs/xfs_inode_item.c | 5 fs/xfs/xfs_iomap.c | 85 -- fs/xfs/xfs_iomap.h | 8 fs/xfs/xfs_log.c | 181 +---- fs/xfs/xfs_log_priv.h | 20 fs/xfs/xfs_log_recover.c | 1 fs/xfs/xfs_mount.c | 2 fs/xfs/xfs_quota.h | 8 fs/xfs/xfs_rename.c | 1 fs/xfs/xfs_rtalloc.c | 1 fs/xfs/xfs_rw.c | 3 fs/xfs/xfs_trans.h | 47 + fs/xfs/xfs_trans_buf.c | 62 - fs/xfs/xfs_vnodeops.c | 8 70 files changed, 2151 insertions(+), 2592 deletions(-) Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Alex Elder <aelder@sgi.com>
2009-12-14 23:14:59 +00:00
trace_xfs_dqadjust(dqp);
/*
* Adjust the inode count and the block count to reflect this inode's
* resource usage.
*/
be64_add_cpu(&dqp->q_core.d_icount, 1);
dqp->q_res_icount++;
if (nblks) {
be64_add_cpu(&dqp->q_core.d_bcount, nblks);
dqp->q_res_bcount += nblks;
}
if (rtblks) {
be64_add_cpu(&dqp->q_core.d_rtbcount, rtblks);
dqp->q_res_rtbcount += rtblks;
}
/*
* Set default limits, adjust timers (since we changed usages)
*
* There are no timers for the default values set in the root dquot.
*/
if (dqp->q_core.d_id) {
xfs_qm_adjust_dqlimits(mp, dqp);
xfs_qm_adjust_dqtimers(mp, &dqp->q_core);
}
dqp->dq_flags |= XFS_DQ_DIRTY;
xfs_qm_dqput(dqp);
return 0;
}
/*
* callback routine supplied to bulkstat(). Given an inumber, find its
* dquots and update them to account for resources taken by that inode.
*/
/* ARGSUSED */
STATIC int
xfs_qm_dqusage_adjust(
struct xfs_mount *mp,
struct xfs_trans *tp,
xfs_ino_t ino,
void *data)
{
struct xfs_inode *ip;
xfs_qcnt_t nblks;
xfs_filblks_t rtblks = 0; /* total rt blks */
int error;
ASSERT(XFS_IS_QUOTA_RUNNING(mp));
/*
* rootino must have its resources accounted for, not so with the quota
* inodes.
*/
if (xfs_is_quota_inode(&mp->m_sb, ino))
return 0;
/*
* We don't _need_ to take the ilock EXCL here because quotacheck runs
* at mount time and therefore nobody will be racing chown/chproj.
*/
error = xfs_iget(mp, tp, ino, XFS_IGET_DONTCACHE, 0, &ip);
if (error == -EINVAL || error == -ENOENT)
return 0;
if (error)
return error;
ASSERT(ip->i_delayed_blks == 0);
if (XFS_IS_REALTIME_INODE(ip)) {
struct xfs_ifork *ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
if (!(ifp->if_flags & XFS_IFEXTENTS)) {
error = xfs_iread_extents(tp, ip, XFS_DATA_FORK);
if (error)
goto error0;
}
xfs_bmap_count_leaves(ifp, &rtblks);
}
nblks = (xfs_qcnt_t)ip->i_d.di_nblocks - rtblks;
/*
* Add the (disk blocks and inode) resources occupied by this
* inode to its dquots. We do this adjustment in the incore dquot,
* and also copy the changes to its buffer.
* We don't care about putting these changes in a transaction
* envelope because if we crash in the middle of a 'quotacheck'
* we have to start from the beginning anyway.
* Once we're done, we'll log all the dquot bufs.
*
* The *QUOTA_ON checks below may look pretty racy, but quotachecks
* and quotaoffs don't race. (Quotachecks happen at mount time only).
*/
if (XFS_IS_UQUOTA_ON(mp)) {
error = xfs_qm_quotacheck_dqadjust(ip, XFS_DQ_USER, nblks,
rtblks);
if (error)
goto error0;
}
if (XFS_IS_GQUOTA_ON(mp)) {
error = xfs_qm_quotacheck_dqadjust(ip, XFS_DQ_GROUP, nblks,
rtblks);
if (error)
goto error0;
}
if (XFS_IS_PQUOTA_ON(mp)) {
error = xfs_qm_quotacheck_dqadjust(ip, XFS_DQ_PROJ, nblks,
rtblks);
if (error)
goto error0;
}
error0:
xfs_irele(ip);
return error;
}
STATIC int
xfs_qm_flush_one(
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct xfs_dquot *dqp,
void *data)
{
xfs: push buffer of flush locked dquot to avoid quotacheck deadlock Reclaim during quotacheck can lead to deadlocks on the dquot flush lock: - Quotacheck populates a local delwri queue with the physical dquot buffers. - Quotacheck performs the xfs_qm_dqusage_adjust() bulkstat and dirties all of the dquots. - Reclaim kicks in and attempts to flush a dquot whose buffer is already queud on the quotacheck queue. The flush succeeds but queueing to the reclaim delwri queue fails as the backing buffer is already queued. The flush unlock is now deferred to I/O completion of the buffer from the quotacheck queue. - The dqadjust bulkstat continues and dirties the recently flushed dquot once again. - Quotacheck proceeds to the xfs_qm_flush_one() walk which requires the flush lock to update the backing buffers with the in-core recalculated values. It deadlocks on the redirtied dquot as the flush lock was already acquired by reclaim, but the buffer resides on the local delwri queue which isn't submitted until the end of quotacheck. This is reproduced by running quotacheck on a filesystem with a couple million inodes in low memory (512MB-1GB) situations. This is a regression as of commit 43ff2122e6 ("xfs: on-stack delayed write buffer lists"), which removed a trylock and buffer I/O submission from the quotacheck dquot flush sequence. Quotacheck first resets and collects the physical dquot buffers in a delwri queue. Then, it traverses the filesystem inodes via bulkstat, updates the in-core dquots, flushes the corrected dquots to the backing buffers and finally submits the delwri queue for I/O. Since the backing buffers are queued across the entire quotacheck operation, dquot reclaim cannot possibly complete a dquot flush before quotacheck completes. Therefore, quotacheck must submit the buffer for I/O in order to cycle the flush lock and flush the dirty in-core dquot to the buffer. Add a delwri queue buffer push mechanism to submit an individual buffer for I/O without losing the delwri queue status and use it from quotacheck to avoid the deadlock. This restores quotacheck behavior to as before the regression was introduced. Reported-by: Martin Svec <martin.svec@zoner.cz> Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2017-06-15 04:21:45 +00:00
struct xfs_mount *mp = dqp->q_mount;
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
struct list_head *buffer_list = data;
struct xfs_buf *bp = NULL;
int error = 0;
xfs_dqlock(dqp);
if (dqp->dq_flags & XFS_DQ_FREEING)
goto out_unlock;
if (!XFS_DQ_IS_DIRTY(dqp))
goto out_unlock;
xfs: push buffer of flush locked dquot to avoid quotacheck deadlock Reclaim during quotacheck can lead to deadlocks on the dquot flush lock: - Quotacheck populates a local delwri queue with the physical dquot buffers. - Quotacheck performs the xfs_qm_dqusage_adjust() bulkstat and dirties all of the dquots. - Reclaim kicks in and attempts to flush a dquot whose buffer is already queud on the quotacheck queue. The flush succeeds but queueing to the reclaim delwri queue fails as the backing buffer is already queued. The flush unlock is now deferred to I/O completion of the buffer from the quotacheck queue. - The dqadjust bulkstat continues and dirties the recently flushed dquot once again. - Quotacheck proceeds to the xfs_qm_flush_one() walk which requires the flush lock to update the backing buffers with the in-core recalculated values. It deadlocks on the redirtied dquot as the flush lock was already acquired by reclaim, but the buffer resides on the local delwri queue which isn't submitted until the end of quotacheck. This is reproduced by running quotacheck on a filesystem with a couple million inodes in low memory (512MB-1GB) situations. This is a regression as of commit 43ff2122e6 ("xfs: on-stack delayed write buffer lists"), which removed a trylock and buffer I/O submission from the quotacheck dquot flush sequence. Quotacheck first resets and collects the physical dquot buffers in a delwri queue. Then, it traverses the filesystem inodes via bulkstat, updates the in-core dquots, flushes the corrected dquots to the backing buffers and finally submits the delwri queue for I/O. Since the backing buffers are queued across the entire quotacheck operation, dquot reclaim cannot possibly complete a dquot flush before quotacheck completes. Therefore, quotacheck must submit the buffer for I/O in order to cycle the flush lock and flush the dirty in-core dquot to the buffer. Add a delwri queue buffer push mechanism to submit an individual buffer for I/O without losing the delwri queue status and use it from quotacheck to avoid the deadlock. This restores quotacheck behavior to as before the regression was introduced. Reported-by: Martin Svec <martin.svec@zoner.cz> Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2017-06-15 04:21:45 +00:00
/*
* The only way the dquot is already flush locked by the time quotacheck
* gets here is if reclaim flushed it before the dqadjust walk dirtied
* it for the final time. Quotacheck collects all dquot bufs in the
* local delwri queue before dquots are dirtied, so reclaim can't have
* possibly queued it for I/O. The only way out is to push the buffer to
* cycle the flush lock.
*/
if (!xfs_dqflock_nowait(dqp)) {
/* buf is pinned in-core by delwri list */
bp = xfs_buf_incore(mp->m_ddev_targp, dqp->q_blkno,
mp->m_quotainfo->qi_dqchunklen, 0);
xfs: push buffer of flush locked dquot to avoid quotacheck deadlock Reclaim during quotacheck can lead to deadlocks on the dquot flush lock: - Quotacheck populates a local delwri queue with the physical dquot buffers. - Quotacheck performs the xfs_qm_dqusage_adjust() bulkstat and dirties all of the dquots. - Reclaim kicks in and attempts to flush a dquot whose buffer is already queud on the quotacheck queue. The flush succeeds but queueing to the reclaim delwri queue fails as the backing buffer is already queued. The flush unlock is now deferred to I/O completion of the buffer from the quotacheck queue. - The dqadjust bulkstat continues and dirties the recently flushed dquot once again. - Quotacheck proceeds to the xfs_qm_flush_one() walk which requires the flush lock to update the backing buffers with the in-core recalculated values. It deadlocks on the redirtied dquot as the flush lock was already acquired by reclaim, but the buffer resides on the local delwri queue which isn't submitted until the end of quotacheck. This is reproduced by running quotacheck on a filesystem with a couple million inodes in low memory (512MB-1GB) situations. This is a regression as of commit 43ff2122e6 ("xfs: on-stack delayed write buffer lists"), which removed a trylock and buffer I/O submission from the quotacheck dquot flush sequence. Quotacheck first resets and collects the physical dquot buffers in a delwri queue. Then, it traverses the filesystem inodes via bulkstat, updates the in-core dquots, flushes the corrected dquots to the backing buffers and finally submits the delwri queue for I/O. Since the backing buffers are queued across the entire quotacheck operation, dquot reclaim cannot possibly complete a dquot flush before quotacheck completes. Therefore, quotacheck must submit the buffer for I/O in order to cycle the flush lock and flush the dirty in-core dquot to the buffer. Add a delwri queue buffer push mechanism to submit an individual buffer for I/O without losing the delwri queue status and use it from quotacheck to avoid the deadlock. This restores quotacheck behavior to as before the regression was introduced. Reported-by: Martin Svec <martin.svec@zoner.cz> Signed-off-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2017-06-15 04:21:45 +00:00
if (!bp) {
error = -EINVAL;
goto out_unlock;
}
xfs_buf_unlock(bp);
xfs_buf_delwri_pushbuf(bp, buffer_list);
xfs_buf_rele(bp);
error = -EAGAIN;
goto out_unlock;
}
error = xfs_qm_dqflush(dqp, &bp);
if (error)
goto out_unlock;
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
xfs_buf_delwri_queue(bp, buffer_list);
xfs_buf_relse(bp);
out_unlock:
xfs_dqunlock(dqp);
return error;
}
/*
* Walk thru all the filesystem inodes and construct a consistent view
* of the disk quota world. If the quotacheck fails, disable quotas.
*/
STATIC int
xfs_qm_quotacheck(
xfs_mount_t *mp)
{
int error, error2;
uint flags;
LIST_HEAD (buffer_list);
struct xfs_inode *uip = mp->m_quotainfo->qi_uquotaip;
struct xfs_inode *gip = mp->m_quotainfo->qi_gquotaip;
struct xfs_inode *pip = mp->m_quotainfo->qi_pquotaip;
flags = 0;
ASSERT(uip || gip || pip);
ASSERT(XFS_IS_QUOTA_RUNNING(mp));
xfs_notice(mp, "Quotacheck needed: Please wait.");
/*
* First we go thru all the dquots on disk, USR and GRP/PRJ, and reset
* their counters to zero. We need a clean slate.
* We don't log our changes till later.
*/
if (uip) {
error = xfs_qm_reset_dqcounts_buf(mp, uip, XFS_QMOPT_UQUOTA,
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
&buffer_list);
if (error)
goto error_return;
flags |= XFS_UQUOTA_CHKD;
}
if (gip) {
error = xfs_qm_reset_dqcounts_buf(mp, gip, XFS_QMOPT_GQUOTA,
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
&buffer_list);
if (error)
goto error_return;
flags |= XFS_GQUOTA_CHKD;
}
if (pip) {
error = xfs_qm_reset_dqcounts_buf(mp, pip, XFS_QMOPT_PQUOTA,
&buffer_list);
if (error)
goto error_return;
flags |= XFS_PQUOTA_CHKD;
}
error = xfs_iwalk_threaded(mp, 0, 0, xfs_qm_dqusage_adjust, 0, true,
NULL);
if (error)
goto error_return;
/*
* We've made all the changes that we need to make incore. Flush them
* down to disk buffers if everything was updated successfully.
*/
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
if (XFS_IS_UQUOTA_ON(mp)) {
error = xfs_qm_dquot_walk(mp, XFS_DQ_USER, xfs_qm_flush_one,
&buffer_list);
}
if (XFS_IS_GQUOTA_ON(mp)) {
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
error2 = xfs_qm_dquot_walk(mp, XFS_DQ_GROUP, xfs_qm_flush_one,
&buffer_list);
if (!error)
error = error2;
}
if (XFS_IS_PQUOTA_ON(mp)) {
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
error2 = xfs_qm_dquot_walk(mp, XFS_DQ_PROJ, xfs_qm_flush_one,
&buffer_list);
if (!error)
error = error2;
}
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
error2 = xfs_buf_delwri_submit(&buffer_list);
if (!error)
error = error2;
/*
* We can get this error if we couldn't do a dquot allocation inside
* xfs_qm_dqusage_adjust (via bulkstat). We don't care about the
* dirty dquots that might be cached, we just want to get rid of them
* and turn quotaoff. The dquots won't be attached to any of the inodes
* at this point (because we intentionally didn't in dqget_noattach).
*/
if (error) {
xfs_qm_dqpurge_all(mp, XFS_QMOPT_QUOTALL);
goto error_return;
}
/*
* If one type of quotas is off, then it will lose its
* quotachecked status, since we won't be doing accounting for
* that type anymore.
*/
mp->m_qflags &= ~XFS_ALL_QUOTA_CHKD;
mp->m_qflags |= flags;
error_return:
xfs_buf_delwri_cancel(&buffer_list);
xfs: on-stack delayed write buffer lists Queue delwri buffers on a local on-stack list instead of a per-buftarg one, and write back the buffers per-process instead of by waking up xfsbufd. This is now easily doable given that we have very few places left that write delwri buffers: - log recovery: Only done at mount time, and already forcing out the buffers synchronously using xfs_flush_buftarg - quotacheck: Same story. - dquot reclaim: Writes out dirty dquots on the LRU under memory pressure. We might want to look into doing more of this via xfsaild, but it's already more optimal than the synchronous inode reclaim that writes each buffer synchronously. - xfsaild: This is the main beneficiary of the change. By keeping a local list of buffers to write we reduce latency of writing out buffers, and more importably we can remove all the delwri list promotions which were hitting the buffer cache hard under sustained metadata loads. The implementation is very straight forward - xfs_buf_delwri_queue now gets a new list_head pointer that it adds the delwri buffers to, and all callers need to eventually submit the list using xfs_buf_delwi_submit or xfs_buf_delwi_submit_nowait. Buffers that already are on a delwri list are skipped in xfs_buf_delwri_queue, assuming they already are on another delwri list. The biggest change to pass down the buffer list was done to the AIL pushing. Now that we operate on buffers the trylock, push and pushbuf log item methods are merged into a single push routine, which tries to lock the item, and if possible add the buffer that needs writeback to the buffer list. This leads to much simpler code than the previous split but requires the individual IOP_PUSH instances to unlock and reacquire the AIL around calls to blocking routines. Given that xfsailds now also handle writing out buffers, the conditions for log forcing and the sleep times needed some small changes. The most important one is that we consider an AIL busy as long we still have buffers to push, and the other one is that we do increment the pushed LSN for buffers that are under flushing at this moment, but still count them towards the stuck items for restart purposes. Without this we could hammer on stuck items without ever forcing the log and not make progress under heavy random delete workloads on fast flash storage devices. [ Dave Chinner: - rebase on previous patches. - improved comments for XBF_DELWRI_Q handling - fix XBF_ASYNC handling in queue submission (test 106 failure) - rename delwri submit function buffer list parameters for clarity - xfs_efd_item_push() should return XFS_ITEM_PINNED ] Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Dave Chinner <dchinner@redhat.com> Reviewed-by: Mark Tinguely <tinguely@sgi.com> Signed-off-by: Ben Myers <bpm@sgi.com>
2012-04-23 05:58:39 +00:00
if (error) {
xfs_warn(mp,
"Quotacheck: Unsuccessful (Error %d): Disabling quotas.",
error);
/*
* We must turn off quotas.
*/
ASSERT(mp->m_quotainfo != NULL);
xfs_qm_destroy_quotainfo(mp);
if (xfs_mount_reset_sbqflags(mp)) {
xfs_warn(mp,
"Quotacheck: Failed to reset quota flags.");
}
} else
xfs_notice(mp, "Quotacheck: Done.");
return error;
}
/*
* This is called from xfs_mountfs to start quotas and initialize all
* necessary data structures like quotainfo. This is also responsible for
* running a quotacheck as necessary. We are guaranteed that the superblock
* is consistently read in at this point.
*
* If we fail here, the mount will continue with quota turned off. We don't
* need to inidicate success or failure at all.
*/
void
xfs_qm_mount_quotas(
struct xfs_mount *mp)
{
int error = 0;
uint sbf;
/*
* If quotas on realtime volumes is not supported, we disable
* quotas immediately.
*/
if (mp->m_sb.sb_rextents) {
xfs_notice(mp, "Cannot turn on quotas for realtime filesystem");
mp->m_qflags = 0;
goto write_changes;
}
ASSERT(XFS_IS_QUOTA_RUNNING(mp));
/*
* Allocate the quotainfo structure inside the mount struct, and
* create quotainode(s), and change/rev superblock if necessary.
*/
error = xfs_qm_init_quotainfo(mp);
if (error) {
/*
* We must turn off quotas.
*/
ASSERT(mp->m_quotainfo == NULL);
mp->m_qflags = 0;
goto write_changes;
}
/*
* If any of the quotas are not consistent, do a quotacheck.
*/
if (XFS_QM_NEED_QUOTACHECK(mp)) {
error = xfs_qm_quotacheck(mp);
if (error) {
/* Quotacheck failed and disabled quotas. */
return;
}
}
/*
* If one type of quotas is off, then it will lose its
* quotachecked status, since we won't be doing accounting for
* that type anymore.
*/
if (!XFS_IS_UQUOTA_ON(mp))
mp->m_qflags &= ~XFS_UQUOTA_CHKD;
if (!XFS_IS_GQUOTA_ON(mp))
mp->m_qflags &= ~XFS_GQUOTA_CHKD;
if (!XFS_IS_PQUOTA_ON(mp))
mp->m_qflags &= ~XFS_PQUOTA_CHKD;
write_changes:
/*
* We actually don't have to acquire the m_sb_lock at all.
* This can only be called from mount, and that's single threaded. XXX
*/
spin_lock(&mp->m_sb_lock);
sbf = mp->m_sb.sb_qflags;
mp->m_sb.sb_qflags = mp->m_qflags & XFS_MOUNT_QUOTA_ALL;
spin_unlock(&mp->m_sb_lock);
if (sbf != (mp->m_qflags & XFS_MOUNT_QUOTA_ALL)) {
if (xfs_sync_sb(mp, false)) {
/*
* We could only have been turning quotas off.
* We aren't in very good shape actually because
* the incore structures are convinced that quotas are
* off, but the on disk superblock doesn't know that !
*/
ASSERT(!(XFS_IS_QUOTA_RUNNING(mp)));
xfs_alert(mp, "%s: Superblock update failed!",
__func__);
}
}
if (error) {
xfs_warn(mp, "Failed to initialize disk quotas.");
return;
}
}
/*
* This is called after the superblock has been read in and we're ready to
* iget the quota inodes.
*/
STATIC int
xfs_qm_init_quotainos(
xfs_mount_t *mp)
{
struct xfs_inode *uip = NULL;
struct xfs_inode *gip = NULL;
struct xfs_inode *pip = NULL;
int error;
uint flags = 0;
ASSERT(mp->m_quotainfo);
/*
* Get the uquota and gquota inodes
*/
if (xfs_sb_version_hasquota(&mp->m_sb)) {
if (XFS_IS_UQUOTA_ON(mp) &&
mp->m_sb.sb_uquotino != NULLFSINO) {
ASSERT(mp->m_sb.sb_uquotino > 0);
error = xfs_iget(mp, NULL, mp->m_sb.sb_uquotino,
0, 0, &uip);
if (error)
return error;
}
if (XFS_IS_GQUOTA_ON(mp) &&
mp->m_sb.sb_gquotino != NULLFSINO) {
ASSERT(mp->m_sb.sb_gquotino > 0);
error = xfs_iget(mp, NULL, mp->m_sb.sb_gquotino,
0, 0, &gip);
if (error)
goto error_rele;
}
if (XFS_IS_PQUOTA_ON(mp) &&
mp->m_sb.sb_pquotino != NULLFSINO) {
ASSERT(mp->m_sb.sb_pquotino > 0);
error = xfs_iget(mp, NULL, mp->m_sb.sb_pquotino,
0, 0, &pip);
if (error)
goto error_rele;
}
} else {
flags |= XFS_QMOPT_SBVERSION;
}
/*
* Create the three inodes, if they don't exist already. The changes
* made above will get added to a transaction and logged in one of
* the qino_alloc calls below. If the device is readonly,
* temporarily switch to read-write to do this.
*/
if (XFS_IS_UQUOTA_ON(mp) && uip == NULL) {
error = xfs_qm_qino_alloc(mp, &uip,
flags | XFS_QMOPT_UQUOTA);
if (error)
goto error_rele;
flags &= ~XFS_QMOPT_SBVERSION;
}
if (XFS_IS_GQUOTA_ON(mp) && gip == NULL) {
error = xfs_qm_qino_alloc(mp, &gip,
flags | XFS_QMOPT_GQUOTA);
if (error)
goto error_rele;
flags &= ~XFS_QMOPT_SBVERSION;
}
if (XFS_IS_PQUOTA_ON(mp) && pip == NULL) {
error = xfs_qm_qino_alloc(mp, &pip,
flags | XFS_QMOPT_PQUOTA);
if (error)
goto error_rele;
}
mp->m_quotainfo->qi_uquotaip = uip;
mp->m_quotainfo->qi_gquotaip = gip;
mp->m_quotainfo->qi_pquotaip = pip;
return 0;
error_rele:
if (uip)
xfs_irele(uip);
if (gip)
xfs_irele(gip);
if (pip)
xfs_irele(pip);
return error;
}
STATIC void
xfs_qm_destroy_quotainos(
struct xfs_quotainfo *qi)
{
if (qi->qi_uquotaip) {
xfs_irele(qi->qi_uquotaip);
qi->qi_uquotaip = NULL; /* paranoia */
}
if (qi->qi_gquotaip) {
xfs_irele(qi->qi_gquotaip);
qi->qi_gquotaip = NULL;
}
if (qi->qi_pquotaip) {
xfs_irele(qi->qi_pquotaip);
qi->qi_pquotaip = NULL;
}
}
STATIC void
xfs_qm_dqfree_one(
struct xfs_dquot *dqp)
{
struct xfs_mount *mp = dqp->q_mount;
struct xfs_quotainfo *qi = mp->m_quotainfo;
mutex_lock(&qi->qi_tree_lock);
radix_tree_delete(xfs_dquot_tree(qi, dqp->q_core.d_flags),
be32_to_cpu(dqp->q_core.d_id));
qi->qi_dquots--;
mutex_unlock(&qi->qi_tree_lock);
xfs_qm_dqdestroy(dqp);
}
/* --------------- utility functions for vnodeops ---------------- */
/*
* Given an inode, a uid, gid and prid make sure that we have
* allocated relevant dquot(s) on disk, and that we won't exceed inode
* quotas by creating this file.
* This also attaches dquot(s) to the given inode after locking it,
* and returns the dquots corresponding to the uid and/or gid.
*
* in : inode (unlocked)
* out : udquot, gdquot with references taken and unlocked
*/
int
xfs_qm_vop_dqalloc(
struct xfs_inode *ip,
kuid_t uid,
kgid_t gid,
prid_t prid,
uint flags,
struct xfs_dquot **O_udqpp,
struct xfs_dquot **O_gdqpp,
struct xfs_dquot **O_pdqpp)
{
struct xfs_mount *mp = ip->i_mount;
struct inode *inode = VFS_I(ip);
struct user_namespace *user_ns = inode->i_sb->s_user_ns;
struct xfs_dquot *uq = NULL;
struct xfs_dquot *gq = NULL;
struct xfs_dquot *pq = NULL;
int error;
uint lockflags;
if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
return 0;
lockflags = XFS_ILOCK_EXCL;
xfs_ilock(ip, lockflags);
if ((flags & XFS_QMOPT_INHERIT) && XFS_INHERIT_GID(ip))
gid = inode->i_gid;
/*
* Attach the dquot(s) to this inode, doing a dquot allocation
* if necessary. The dquot(s) will not be locked.
*/
if (XFS_NOT_DQATTACHED(mp, ip)) {
error = xfs_qm_dqattach_locked(ip, true);
if (error) {
xfs_iunlock(ip, lockflags);
return error;
}
}
if ((flags & XFS_QMOPT_UQUOTA) && XFS_IS_UQUOTA_ON(mp)) {
if (!uid_eq(inode->i_uid, uid)) {
/*
* What we need is the dquot that has this uid, and
* if we send the inode to dqget, the uid of the inode
* takes priority over what's sent in the uid argument.
* We must unlock inode here before calling dqget if
* we're not sending the inode, because otherwise
* we'll deadlock by doing trans_reserve while
* holding ilock.
*/
xfs_iunlock(ip, lockflags);
error = xfs_qm_dqget(mp, from_kuid(user_ns, uid),
XFS_DQ_USER, true, &uq);
if (error) {
ASSERT(error != -ENOENT);
return error;
}
/*
* Get the ilock in the right order.
*/
xfs_dqunlock(uq);
lockflags = XFS_ILOCK_SHARED;
xfs_ilock(ip, lockflags);
} else {
/*
* Take an extra reference, because we'll return
* this to caller
*/
ASSERT(ip->i_udquot);
uq = xfs_qm_dqhold(ip->i_udquot);
}
}
if ((flags & XFS_QMOPT_GQUOTA) && XFS_IS_GQUOTA_ON(mp)) {
if (!gid_eq(inode->i_gid, gid)) {
xfs_iunlock(ip, lockflags);
error = xfs_qm_dqget(mp, from_kgid(user_ns, gid),
XFS_DQ_GROUP, true, &gq);
if (error) {
ASSERT(error != -ENOENT);
goto error_rele;
}
xfs_dqunlock(gq);
lockflags = XFS_ILOCK_SHARED;
xfs_ilock(ip, lockflags);
} else {
ASSERT(ip->i_gdquot);
gq = xfs_qm_dqhold(ip->i_gdquot);
}
}
if ((flags & XFS_QMOPT_PQUOTA) && XFS_IS_PQUOTA_ON(mp)) {
if (ip->i_d.di_projid != prid) {
xfs_iunlock(ip, lockflags);
error = xfs_qm_dqget(mp, (xfs_dqid_t)prid, XFS_DQ_PROJ,
true, &pq);
if (error) {
ASSERT(error != -ENOENT);
goto error_rele;
}
xfs_dqunlock(pq);
lockflags = XFS_ILOCK_SHARED;
xfs_ilock(ip, lockflags);
} else {
ASSERT(ip->i_pdquot);
pq = xfs_qm_dqhold(ip->i_pdquot);
}
}
trace_xfs_dquot_dqalloc(ip);
xfs_iunlock(ip, lockflags);
if (O_udqpp)
*O_udqpp = uq;
else
xfs_qm_dqrele(uq);
if (O_gdqpp)
*O_gdqpp = gq;
else
xfs_qm_dqrele(gq);
if (O_pdqpp)
*O_pdqpp = pq;
else
xfs_qm_dqrele(pq);
return 0;
error_rele:
xfs_qm_dqrele(gq);
xfs_qm_dqrele(uq);
return error;
}
/*
* Actually transfer ownership, and do dquot modifications.
* These were already reserved.
*/
struct xfs_dquot *
xfs_qm_vop_chown(
struct xfs_trans *tp,
struct xfs_inode *ip,
struct xfs_dquot **IO_olddq,
struct xfs_dquot *newdq)
{
struct xfs_dquot *prevdq;
uint bfield = XFS_IS_REALTIME_INODE(ip) ?
XFS_TRANS_DQ_RTBCOUNT : XFS_TRANS_DQ_BCOUNT;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
ASSERT(XFS_IS_QUOTA_RUNNING(ip->i_mount));
/* old dquot */
prevdq = *IO_olddq;
ASSERT(prevdq);
ASSERT(prevdq != newdq);
xfs_trans_mod_dquot(tp, prevdq, bfield, -(ip->i_d.di_nblocks));
xfs_trans_mod_dquot(tp, prevdq, XFS_TRANS_DQ_ICOUNT, -1);
/* the sparkling new dquot */
xfs_trans_mod_dquot(tp, newdq, bfield, ip->i_d.di_nblocks);
xfs_trans_mod_dquot(tp, newdq, XFS_TRANS_DQ_ICOUNT, 1);
/*
* Take an extra reference, because the inode is going to keep
* this dquot pointer even after the trans_commit.
*/
*IO_olddq = xfs_qm_dqhold(newdq);
return prevdq;
}
/*
* Quota reservations for setattr(AT_UID|AT_GID|AT_PROJID).
*/
int
xfs_qm_vop_chown_reserve(
struct xfs_trans *tp,
struct xfs_inode *ip,
struct xfs_dquot *udqp,
struct xfs_dquot *gdqp,
struct xfs_dquot *pdqp,
uint flags)
{
struct xfs_mount *mp = ip->i_mount;
uint64_t delblks;
xfs: always return -ENOSPC on project quota reservation failure XFS project quota treats project hierarchies as "mini filesysems" and so rather than -EDQUOT, the intent is to return -ENOSPC when a quota reservation fails, but this behavior is not consistent. The only place we make a decision between -EDQUOT and -ENOSPC returns based on quota type is in xfs_trans_dqresv(). This behavior is currently controlled by whether or not the XFS_QMOPT_ENOSPC flag gets passed into the quota reservation. However, its use is not consistent; paths such as xfs_create() and xfs_symlink() don't set the flag, so a reservation failure will return -EDQUOT for project quota reservation failures rather than -ENOSPC for these sorts of operations, even for project quota: # mkdir mnt/project # xfs_quota -x -c "project -s -p mnt/project 42" mnt # xfs_quota -x -c 'limit -p isoft=2 ihard=3 42' mnt # touch mnt/project/file{1,2,3} touch: cannot touch ‘mnt/project/file3’: Disk quota exceeded We can make this consistent by not requiring the flag to be set at the top of the callchain; instead we can simply test whether we are reserving a project quota with XFS_QM_ISPDQ in xfs_trans_dqresv and if so, return -ENOSPC for that failure. This removes the need for the XFS_QMOPT_ENOSPC altogether and simplifies the code a fair bit. Signed-off-by: Eric Sandeen <sandeen@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 20:06:59 +00:00
unsigned int blkflags;
struct xfs_dquot *udq_unres = NULL;
struct xfs_dquot *gdq_unres = NULL;
struct xfs_dquot *pdq_unres = NULL;
struct xfs_dquot *udq_delblks = NULL;
struct xfs_dquot *gdq_delblks = NULL;
struct xfs_dquot *pdq_delblks = NULL;
int error;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED));
ASSERT(XFS_IS_QUOTA_RUNNING(mp));
delblks = ip->i_delayed_blks;
blkflags = XFS_IS_REALTIME_INODE(ip) ?
XFS_QMOPT_RES_RTBLKS : XFS_QMOPT_RES_REGBLKS;
if (XFS_IS_UQUOTA_ON(mp) && udqp &&
i_uid_read(VFS_I(ip)) != be32_to_cpu(udqp->q_core.d_id)) {
udq_delblks = udqp;
/*
* If there are delayed allocation blocks, then we have to
* unreserve those from the old dquot, and add them to the
* new dquot.
*/
if (delblks) {
ASSERT(ip->i_udquot);
udq_unres = ip->i_udquot;
}
}
if (XFS_IS_GQUOTA_ON(ip->i_mount) && gdqp &&
i_gid_read(VFS_I(ip)) != be32_to_cpu(gdqp->q_core.d_id)) {
gdq_delblks = gdqp;
if (delblks) {
ASSERT(ip->i_gdquot);
gdq_unres = ip->i_gdquot;
}
}
if (XFS_IS_PQUOTA_ON(ip->i_mount) && pdqp &&
ip->i_d.di_projid != be32_to_cpu(pdqp->q_core.d_id)) {
pdq_delblks = pdqp;
if (delblks) {
ASSERT(ip->i_pdquot);
pdq_unres = ip->i_pdquot;
}
}
error = xfs_trans_reserve_quota_bydquots(tp, ip->i_mount,
udq_delblks, gdq_delblks, pdq_delblks,
xfs: always return -ENOSPC on project quota reservation failure XFS project quota treats project hierarchies as "mini filesysems" and so rather than -EDQUOT, the intent is to return -ENOSPC when a quota reservation fails, but this behavior is not consistent. The only place we make a decision between -EDQUOT and -ENOSPC returns based on quota type is in xfs_trans_dqresv(). This behavior is currently controlled by whether or not the XFS_QMOPT_ENOSPC flag gets passed into the quota reservation. However, its use is not consistent; paths such as xfs_create() and xfs_symlink() don't set the flag, so a reservation failure will return -EDQUOT for project quota reservation failures rather than -ENOSPC for these sorts of operations, even for project quota: # mkdir mnt/project # xfs_quota -x -c "project -s -p mnt/project 42" mnt # xfs_quota -x -c 'limit -p isoft=2 ihard=3 42' mnt # touch mnt/project/file{1,2,3} touch: cannot touch ‘mnt/project/file3’: Disk quota exceeded We can make this consistent by not requiring the flag to be set at the top of the callchain; instead we can simply test whether we are reserving a project quota with XFS_QM_ISPDQ in xfs_trans_dqresv and if so, return -ENOSPC for that failure. This removes the need for the XFS_QMOPT_ENOSPC altogether and simplifies the code a fair bit. Signed-off-by: Eric Sandeen <sandeen@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 20:06:59 +00:00
ip->i_d.di_nblocks, 1, flags | blkflags);
if (error)
return error;
/*
* Do the delayed blks reservations/unreservations now. Since, these
* are done without the help of a transaction, if a reservation fails
* its previous reservations won't be automatically undone by trans
* code. So, we have to do it manually here.
*/
if (delblks) {
/*
* Do the reservations first. Unreservation can't fail.
*/
ASSERT(udq_delblks || gdq_delblks || pdq_delblks);
ASSERT(udq_unres || gdq_unres || pdq_unres);
error = xfs_trans_reserve_quota_bydquots(NULL, ip->i_mount,
udq_delblks, gdq_delblks, pdq_delblks,
xfs: always return -ENOSPC on project quota reservation failure XFS project quota treats project hierarchies as "mini filesysems" and so rather than -EDQUOT, the intent is to return -ENOSPC when a quota reservation fails, but this behavior is not consistent. The only place we make a decision between -EDQUOT and -ENOSPC returns based on quota type is in xfs_trans_dqresv(). This behavior is currently controlled by whether or not the XFS_QMOPT_ENOSPC flag gets passed into the quota reservation. However, its use is not consistent; paths such as xfs_create() and xfs_symlink() don't set the flag, so a reservation failure will return -EDQUOT for project quota reservation failures rather than -ENOSPC for these sorts of operations, even for project quota: # mkdir mnt/project # xfs_quota -x -c "project -s -p mnt/project 42" mnt # xfs_quota -x -c 'limit -p isoft=2 ihard=3 42' mnt # touch mnt/project/file{1,2,3} touch: cannot touch ‘mnt/project/file3’: Disk quota exceeded We can make this consistent by not requiring the flag to be set at the top of the callchain; instead we can simply test whether we are reserving a project quota with XFS_QM_ISPDQ in xfs_trans_dqresv and if so, return -ENOSPC for that failure. This removes the need for the XFS_QMOPT_ENOSPC altogether and simplifies the code a fair bit. Signed-off-by: Eric Sandeen <sandeen@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
2020-05-21 20:06:59 +00:00
(xfs_qcnt_t)delblks, 0, flags | blkflags);
if (error)
return error;
xfs_trans_reserve_quota_bydquots(NULL, ip->i_mount,
udq_unres, gdq_unres, pdq_unres,
-((xfs_qcnt_t)delblks), 0, blkflags);
}
return 0;
}
int
xfs_qm_vop_rename_dqattach(
struct xfs_inode **i_tab)
{
struct xfs_mount *mp = i_tab[0]->i_mount;
int i;
if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
return 0;
for (i = 0; (i < 4 && i_tab[i]); i++) {
struct xfs_inode *ip = i_tab[i];
int error;
/*
* Watch out for duplicate entries in the table.
*/
if (i == 0 || ip != i_tab[i-1]) {
if (XFS_NOT_DQATTACHED(mp, ip)) {
error = xfs_qm_dqattach(ip);
if (error)
return error;
}
}
}
return 0;
}
void
xfs_qm_vop_create_dqattach(
struct xfs_trans *tp,
struct xfs_inode *ip,
struct xfs_dquot *udqp,
struct xfs_dquot *gdqp,
struct xfs_dquot *pdqp)
{
struct xfs_mount *mp = tp->t_mountp;
if (!XFS_IS_QUOTA_RUNNING(mp) || !XFS_IS_QUOTA_ON(mp))
return;
ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
xfs: fix false assertion at xfs_qm_vop_create_dqattach After the previous fix, there still has another ASSERT failure if turning off any type of quota while fsstress is running at the same time. Backtrace in this case: [ 50.867897] XFS: Assertion failed: XFS_IS_GQUOTA_ON(mp), file: fs/xfs/xfs_qm.c, line: 2118 [ 50.867924] ------------[ cut here ]------------ ... <snip> [ 50.867957] Kernel BUG at ffffffffa0b55a32 [verbose debug info unavailable] [ 50.867999] invalid opcode: 0000 [#1] SMP [ 50.869407] Call Trace: [ 50.869446] [<ffffffffa0bc408a>] xfs_qm_vop_create_dqattach+0x19a/0x2d0 [xfs] [ 50.869512] [<ffffffffa0b9cc45>] xfs_create+0x5c5/0x6a0 [xfs] [ 50.869564] [<ffffffffa0b5307c>] xfs_vn_mknod+0xac/0x1d0 [xfs] [ 50.869615] [<ffffffffa0b531d6>] xfs_vn_mkdir+0x16/0x20 [xfs] [ 50.869655] [<ffffffff811becd5>] vfs_mkdir+0x95/0x130 [ 50.869689] [<ffffffff811bf63a>] SyS_mkdirat+0xaa/0xe0 [ 50.869723] [<ffffffff811bf689>] SyS_mkdir+0x19/0x20 [ 50.869757] [<ffffffff8170f7dd>] system_call_fastpath+0x1a/0x1f [ 50.869793] Code: 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 <snip> [ 50.870003] RIP [<ffffffffa0b55a32>] assfail+0x22/0x30 [xfs] [ 50.870050] RSP <ffff88002941fd60> [ 50.879251] ---[ end trace c93a2b342341c65b ]--- We're hitting the ASSERT(XFS_IS_*QUOTA_ON(mp)) in xfs_qm_vop_create_dqattach(), however the assertion itself is not right IMHO. While performing quota off, we firstly clear the XFS_*QUOTA_ACTIVE bit(s) from struct xfs_mount without taking any special locks, see xfs_qm_scall_quotaoff(). Hence there is no guarantee that the desired quota is still active. Signed-off-by: Jie Liu <jeff.liu@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-26 13:38:54 +00:00
if (udqp && XFS_IS_UQUOTA_ON(mp)) {
ASSERT(ip->i_udquot == NULL);
ASSERT(i_uid_read(VFS_I(ip)) == be32_to_cpu(udqp->q_core.d_id));
ip->i_udquot = xfs_qm_dqhold(udqp);
xfs_trans_mod_dquot(tp, udqp, XFS_TRANS_DQ_ICOUNT, 1);
}
xfs: fix false assertion at xfs_qm_vop_create_dqattach After the previous fix, there still has another ASSERT failure if turning off any type of quota while fsstress is running at the same time. Backtrace in this case: [ 50.867897] XFS: Assertion failed: XFS_IS_GQUOTA_ON(mp), file: fs/xfs/xfs_qm.c, line: 2118 [ 50.867924] ------------[ cut here ]------------ ... <snip> [ 50.867957] Kernel BUG at ffffffffa0b55a32 [verbose debug info unavailable] [ 50.867999] invalid opcode: 0000 [#1] SMP [ 50.869407] Call Trace: [ 50.869446] [<ffffffffa0bc408a>] xfs_qm_vop_create_dqattach+0x19a/0x2d0 [xfs] [ 50.869512] [<ffffffffa0b9cc45>] xfs_create+0x5c5/0x6a0 [xfs] [ 50.869564] [<ffffffffa0b5307c>] xfs_vn_mknod+0xac/0x1d0 [xfs] [ 50.869615] [<ffffffffa0b531d6>] xfs_vn_mkdir+0x16/0x20 [xfs] [ 50.869655] [<ffffffff811becd5>] vfs_mkdir+0x95/0x130 [ 50.869689] [<ffffffff811bf63a>] SyS_mkdirat+0xaa/0xe0 [ 50.869723] [<ffffffff811bf689>] SyS_mkdir+0x19/0x20 [ 50.869757] [<ffffffff8170f7dd>] system_call_fastpath+0x1a/0x1f [ 50.869793] Code: 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 <snip> [ 50.870003] RIP [<ffffffffa0b55a32>] assfail+0x22/0x30 [xfs] [ 50.870050] RSP <ffff88002941fd60> [ 50.879251] ---[ end trace c93a2b342341c65b ]--- We're hitting the ASSERT(XFS_IS_*QUOTA_ON(mp)) in xfs_qm_vop_create_dqattach(), however the assertion itself is not right IMHO. While performing quota off, we firstly clear the XFS_*QUOTA_ACTIVE bit(s) from struct xfs_mount without taking any special locks, see xfs_qm_scall_quotaoff(). Hence there is no guarantee that the desired quota is still active. Signed-off-by: Jie Liu <jeff.liu@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-26 13:38:54 +00:00
if (gdqp && XFS_IS_GQUOTA_ON(mp)) {
ASSERT(ip->i_gdquot == NULL);
ASSERT(i_gid_read(VFS_I(ip)) == be32_to_cpu(gdqp->q_core.d_id));
ip->i_gdquot = xfs_qm_dqhold(gdqp);
xfs_trans_mod_dquot(tp, gdqp, XFS_TRANS_DQ_ICOUNT, 1);
}
xfs: fix false assertion at xfs_qm_vop_create_dqattach After the previous fix, there still has another ASSERT failure if turning off any type of quota while fsstress is running at the same time. Backtrace in this case: [ 50.867897] XFS: Assertion failed: XFS_IS_GQUOTA_ON(mp), file: fs/xfs/xfs_qm.c, line: 2118 [ 50.867924] ------------[ cut here ]------------ ... <snip> [ 50.867957] Kernel BUG at ffffffffa0b55a32 [verbose debug info unavailable] [ 50.867999] invalid opcode: 0000 [#1] SMP [ 50.869407] Call Trace: [ 50.869446] [<ffffffffa0bc408a>] xfs_qm_vop_create_dqattach+0x19a/0x2d0 [xfs] [ 50.869512] [<ffffffffa0b9cc45>] xfs_create+0x5c5/0x6a0 [xfs] [ 50.869564] [<ffffffffa0b5307c>] xfs_vn_mknod+0xac/0x1d0 [xfs] [ 50.869615] [<ffffffffa0b531d6>] xfs_vn_mkdir+0x16/0x20 [xfs] [ 50.869655] [<ffffffff811becd5>] vfs_mkdir+0x95/0x130 [ 50.869689] [<ffffffff811bf63a>] SyS_mkdirat+0xaa/0xe0 [ 50.869723] [<ffffffff811bf689>] SyS_mkdir+0x19/0x20 [ 50.869757] [<ffffffff8170f7dd>] system_call_fastpath+0x1a/0x1f [ 50.869793] Code: 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 55 48 89 <snip> [ 50.870003] RIP [<ffffffffa0b55a32>] assfail+0x22/0x30 [xfs] [ 50.870050] RSP <ffff88002941fd60> [ 50.879251] ---[ end trace c93a2b342341c65b ]--- We're hitting the ASSERT(XFS_IS_*QUOTA_ON(mp)) in xfs_qm_vop_create_dqattach(), however the assertion itself is not right IMHO. While performing quota off, we firstly clear the XFS_*QUOTA_ACTIVE bit(s) from struct xfs_mount without taking any special locks, see xfs_qm_scall_quotaoff(). Hence there is no guarantee that the desired quota is still active. Signed-off-by: Jie Liu <jeff.liu@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ben Myers <bpm@sgi.com>
2013-11-26 13:38:54 +00:00
if (pdqp && XFS_IS_PQUOTA_ON(mp)) {
ASSERT(ip->i_pdquot == NULL);
ASSERT(ip->i_d.di_projid == be32_to_cpu(pdqp->q_core.d_id));
ip->i_pdquot = xfs_qm_dqhold(pdqp);
xfs_trans_mod_dquot(tp, pdqp, XFS_TRANS_DQ_ICOUNT, 1);
}
}