2 * Copyright (C) 2010 Red Hat, Inc.
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation.
9 * This program is distributed in the hope that it would be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write the Free Software Foundation,
16 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
23 #include "xfs_mount.h"
24 #include "xfs_quota.h"
25 #include "xfs_trans.h"
26 #include "xfs_alloc_btree.h"
27 #include "xfs_bmap_btree.h"
28 #include "xfs_ialloc_btree.h"
29 #include "xfs_btree.h"
30 #include "xfs_inode.h"
31 #include "xfs_alloc.h"
32 #include "xfs_error.h"
33 #include "xfs_discard.h"
34 #include "xfs_trace.h"
43 __uint64_t
*blocks_trimmed
)
45 struct block_device
*bdev
= mp
->m_ddev_targp
->bt_bdev
;
46 struct xfs_btree_cur
*cur
;
48 struct xfs_perag
*pag
;
52 pag
= xfs_perag_get(mp
, agno
);
54 error
= xfs_alloc_read_agf(mp
, NULL
, agno
, 0, &agbp
);
58 cur
= xfs_allocbt_init_cursor(mp
, NULL
, agbp
, agno
, XFS_BTNUM_CNT
);
61 * Force out the log. This means any transactions that might have freed
62 * space before we took the AGF buffer lock are now on disk, and the
63 * volatile disk cache is flushed.
65 xfs_log_force(mp
, XFS_LOG_SYNC
);
68 * Look up the longest btree in the AGF and start with it.
70 error
= xfs_alloc_lookup_le(cur
, 0,
71 XFS_BUF_TO_AGF(agbp
)->agf_longest
, &i
);
76 * Loop until we are done with all extents that are large
77 * enough to be worth discarding.
83 error
= xfs_alloc_get_rec(cur
, &fbno
, &flen
, &i
);
86 XFS_WANT_CORRUPTED_GOTO(i
== 1, out_del_cursor
);
87 ASSERT(flen
<= XFS_BUF_TO_AGF(agbp
)->agf_longest
);
93 trace_xfs_discard_toosmall(mp
, agno
, fbno
, flen
);
98 * If the extent is entirely outside of the range we are
99 * supposed to discard skip it. Do not bother to trim
100 * down partially overlapping ranges for now.
102 if (XFS_AGB_TO_FSB(mp
, agno
, fbno
) + flen
< start
||
103 XFS_AGB_TO_FSB(mp
, agno
, fbno
) > end
) {
104 trace_xfs_discard_exclude(mp
, agno
, fbno
, flen
);
109 * If any blocks in the range are still busy, skip the
110 * discard and try again the next time.
112 if (xfs_alloc_busy_search(mp
, agno
, fbno
, flen
)) {
113 trace_xfs_discard_busy(mp
, agno
, fbno
, flen
);
117 trace_xfs_discard_extent(mp
, agno
, fbno
, flen
);
118 error
= -blkdev_issue_discard(bdev
,
119 XFS_AGB_TO_DADDR(mp
, agno
, fbno
),
120 XFS_FSB_TO_BB(mp
, flen
),
124 *blocks_trimmed
+= flen
;
127 error
= xfs_btree_decrement(cur
, 0, &i
);
133 xfs_btree_del_cursor(cur
, error
? XFS_BTREE_ERROR
: XFS_BTREE_NOERROR
);
142 struct xfs_mount
*mp
,
143 struct fstrim_range __user
*urange
)
145 struct request_queue
*q
= mp
->m_ddev_targp
->bt_bdev
->bd_disk
->queue
;
146 unsigned int granularity
= q
->limits
.discard_granularity
;
147 struct fstrim_range range
;
148 xfs_fsblock_t start
, end
, minlen
;
149 xfs_agnumber_t start_agno
, end_agno
, agno
;
150 __uint64_t blocks_trimmed
= 0;
151 int error
, last_error
= 0;
153 if (!capable(CAP_SYS_ADMIN
))
154 return -XFS_ERROR(EPERM
);
155 if (!blk_queue_discard(q
))
156 return -XFS_ERROR(EOPNOTSUPP
);
157 if (copy_from_user(&range
, urange
, sizeof(range
)))
158 return -XFS_ERROR(EFAULT
);
161 * Truncating down the len isn't actually quite correct, but using
162 * XFS_B_TO_FSB would mean we trivially get overflows for values
163 * of ULLONG_MAX or slightly lower. And ULLONG_MAX is the default
164 * used by the fstrim application. In the end it really doesn't
165 * matter as trimming blocks is an advisory interface.
167 start
= XFS_B_TO_FSBT(mp
, range
.start
);
168 end
= start
+ XFS_B_TO_FSBT(mp
, range
.len
) - 1;
169 minlen
= XFS_B_TO_FSB(mp
, max_t(u64
, granularity
, range
.minlen
));
171 if (start
>= mp
->m_sb
.sb_dblocks
)
172 return -XFS_ERROR(EINVAL
);
173 if (end
> mp
->m_sb
.sb_dblocks
- 1)
174 end
= mp
->m_sb
.sb_dblocks
- 1;
176 start_agno
= XFS_FSB_TO_AGNO(mp
, start
);
177 end_agno
= XFS_FSB_TO_AGNO(mp
, end
);
179 for (agno
= start_agno
; agno
<= end_agno
; agno
++) {
180 error
= -xfs_trim_extents(mp
, agno
, start
, end
, minlen
,
189 range
.len
= XFS_FSB_TO_B(mp
, blocks_trimmed
);
190 if (copy_to_user(urange
, &range
, sizeof(range
)))
191 return -XFS_ERROR(EFAULT
);
197 struct xfs_mount
*mp
,
198 struct list_head
*list
)
200 struct xfs_busy_extent
*busyp
;
203 list_for_each_entry(busyp
, list
, list
) {
204 trace_xfs_discard_extent(mp
, busyp
->agno
, busyp
->bno
,
207 error
= -blkdev_issue_discard(mp
->m_ddev_targp
->bt_bdev
,
208 XFS_AGB_TO_DADDR(mp
, busyp
->agno
, busyp
->bno
),
209 XFS_FSB_TO_BB(mp
, busyp
->length
),
211 if (error
&& error
!= EOPNOTSUPP
) {
213 "discard failed for extent [0x%llu,%u], error %d",
214 (unsigned long long)busyp
->bno
,