| @@ -1,6 +1,6 @@ |
| 1 | 1 | // SPDX-License-Identifier: GPL-2.0 |
| 2 | 2 | /* |
| 3 | | − * Copyright (C) 2010 Red Hat, Inc. |
| 3 | + * Copyright (C) 2010, 2023 Red Hat, Inc. |
| 4 | 4 | * All Rights Reserved. |
| 5 | 5 | */ |
| 6 | 6 | #include "xfs.h" |
| @@ -19,21 +19,147 @@ · 12 unmodified lines |
| 19 | 19 | #include "xfs_log.h" |
| 20 | 20 | #include "xfs_ag.h" |
| 21 | 21 | |
| 22 | | −STATIC int |
| 23 | | −xfs_trim_extents( |
| 22 | +/* |
| 23 | + * Notes on an efficient, low latency fstrim algorithm |
| 24 | + * |
| 25 | + * We need to walk the filesystem free space and issue discards on the free |
| 26 | + * space that meet the search criteria (size and location). We cannot issue |
| 27 | + * discards on extents that might be in use, or are so recently in use they are |
| 28 | + * still marked as busy. To serialise against extent state changes whilst we are |
| 29 | + * gathering extents to trim, we must hold the AGF lock to lock out other |
| 30 | + * allocations and extent free operations that might change extent state. |
| 31 | + * |
| 32 | + * However, we cannot just hold the AGF for the entire AG free space walk whilst |
| 33 | + * we issue discards on each free space that is found. Storage devices can have |
| 34 | + * extremely slow discard implementations (e.g. ceph RBD) and so walking a |
| 35 | + * couple of million free extents and issuing synchronous discards on each |
| 36 | + * extent can take a *long* time. Whilst we are doing this walk, nothing else |
| 37 | + * can access the AGF, and we can stall transactions and hence the log whilst |
| 38 | + * modifications wait for the AGF lock to be released. This can lead hung tasks |
| 39 | + * kicking the hung task timer and rebooting the system. This is bad. |
| 40 | + * |
| 41 | + * Hence we need to take a leaf from the bulkstat playbook. It takes the AGI |
| 42 | + * lock, gathers a range of inode cluster buffers that are allocated, drops the |
| 43 | + * AGI lock and then reads all the inode cluster buffers and processes them. It |
| 44 | + * loops doing this, using a cursor to keep track of where it is up to in the AG |
| 45 | + * for each iteration to restart the INOBT lookup from. |
| 46 | + * |
| 47 | + * We can't do this exactly with free space - once we drop the AGF lock, the |
| 48 | + * state of the free extent is out of our control and we cannot run a discard |
| 49 | + * safely on it in this situation. Unless, of course, we've marked the free |
| 50 | + * extent as busy and undergoing a discard operation whilst we held the AGF |
| 51 | + * locked. |
| 52 | + * |
| 53 | + * This is exactly how online discard works - free extents are marked busy when |
| 54 | + * they are freed, and once the extent free has been committed to the journal, |
| 55 | + * the busy extent record is marked as "undergoing discard" and the discard is |
| 56 | + * then issued on the free extent. Once the discard completes, the busy extent |
| 57 | + * record is removed and the extent is able to be allocated again. |
| 58 | + * |
| 59 | + * In the context of fstrim, if we find a free extent we need to discard, we |
| 60 | + * don't have to discard it immediately. All we need to do it record that free |
| 61 | + * extent as being busy and under discard, and all the allocation routines will |
| 62 | + * now avoid trying to allocate it. Hence if we mark the extent as busy under |
| 63 | + * the AGF lock, we can safely discard it without holding the AGF lock because |
| 64 | + * nothing will attempt to allocate that free space until the discard completes. |
| 65 | + * |
| 66 | + * This also allows us to issue discards asynchronously like we do with online |
| 67 | + * discard, and so for fast devices fstrim will run much faster as we can have |
| 68 | + * multiple discard operations in flight at once, as well as pipeline the free |
| 69 | + * extent search so that it overlaps in flight discard IO. |
| 70 | + */ |
| 71 | + |
| 72 | +struct workqueue_struct *xfs_discard_wq; |
| 73 | + |
| 74 | +static void |
| 75 | +xfs_discard_endio_work( |
| 76 | + struct work_struct *work) |
| 77 | +{ |
| 78 | + struct xfs_busy_extents *extents = |
| 79 | + container_of(work, struct xfs_busy_extents, endio_work); |
| 80 | + |
| 81 | + xfs_extent_busy_clear(extents->mount, &extents->extent_list, false); |
| 82 | + kmem_free(extents->owner); |
| 83 | +} |
| 84 | + |
| 85 | +/* |
| 86 | + * Queue up the actual completion to a thread to avoid IRQ-safe locking for |
| 87 | + * pagb_lock. |
| 88 | + */ |
| 89 | +static void |
| 90 | +xfs_discard_endio( |
| 91 | + struct bio *bio) |
| 92 | +{ |
| 93 | + struct xfs_busy_extents *extents = bio->bi_private; |
| 94 | + |
| 95 | + INIT_WORK(&extents->endio_work, xfs_discard_endio_work); |
| 96 | + queue_work(xfs_discard_wq, &extents->endio_work); |
| 97 | + bio_put(bio); |
| 98 | +} |
| 99 | + |
| 100 | +/* |
| 101 | + * Walk the discard list and issue discards on all the busy extents in the |
| 102 | + * list. We plug and chain the bios so that we only need a single completion |
| 103 | + * call to clear all the busy extents once the discards are complete. |
| 104 | + */ |
| 105 | +int |
| 106 | +xfs_discard_extents( |
| 107 | + struct xfs_mount *mp, |
| 108 | + struct xfs_busy_extents *extents) |
| 109 | +{ |
| 110 | + struct xfs_extent_busy *busyp; |
| 111 | + struct bio *bio = NULL; |
| 112 | + struct blk_plug plug; |
| 113 | + int error = 0; |
| 114 | + |
| 115 | + blk_start_plug(&plug); |
| 116 | + list_for_each_entry(busyp, &extents->extent_list, list) { |
| 117 | + trace_xfs_discard_extent(mp, busyp->agno, busyp->bno, |
| 118 | + busyp->length); |
| 119 | + |
| 120 | + error = __blkdev_issue_discard(mp->m_ddev_targp->bt_bdev, |
| 121 | + XFS_AGB_TO_DADDR(mp, busyp->agno, busyp->bno), |
| 122 | + XFS_FSB_TO_BB(mp, busyp->length), |
| 123 | + GFP_NOFS, &bio); |
| 124 | + if (error && error != -EOPNOTSUPP) { |
| 125 | + xfs_info(mp, |
| 126 | + "discard failed for extent [0x%llx,%u], error %d", |
| 127 | + (unsigned long long)busyp->bno, |
| 128 | + busyp->length, |
| 129 | + error); |
| 130 | + break; |
| 131 | + } |
| 132 | + } |
| 133 | + |
| 134 | + if (bio) { |
| 135 | + bio->bi_private = extents; |
| 136 | + bio->bi_end_io = xfs_discard_endio; |
| 137 | + submit_bio(bio); |
| 138 | + } else { |
| 139 | + xfs_discard_endio_work(&extents->endio_work); |
| 140 | + } |
| 141 | + blk_finish_plug(&plug); |
| 142 | + |
| 143 | + return error; |
| 144 | +} |
| 145 | + |
| 146 | + |
| 147 | +static int |
| 148 | +xfs_trim_gather_extents( |
| 24 | 149 | struct xfs_perag *pag, |
| 25 | 150 | xfs_daddr_t start, |
| 26 | 151 | xfs_daddr_t end, |
| 27 | 152 | xfs_daddr_t minlen, |
| 153 | + struct xfs_alloc_rec_incore *tcur, |
| 154 | + struct xfs_busy_extents *extents, |
| 28 | 155 | uint64_t *blocks_trimmed) |
| 29 | 156 | { |
| 30 | 157 | struct xfs_mount *mp = pag->pag_mount; |
| 31 | | − struct block_device *bdev = mp->m_ddev_targp->bt_bdev; |
| 32 | 158 | struct xfs_btree_cur *cur; |
| 33 | 159 | struct xfs_buf *agbp; |
| 34 | | − struct xfs_agf *agf; |
| 35 | 160 | int error; |
| 36 | 161 | int i; |
| 162 | + int batch = 100; |
| 37 | 163 | |
| 38 | 164 | /* |
| 39 | 165 | * Force out the log. This means any transactions that might have freed |
| @@ -45,20 +171,28 @@ xfs_trim_extents( · 5 unmodified lines |
| 45 | 171 | error = xfs_alloc_read_agf(pag, NULL, 0, &agbp); |
| 46 | 172 | if (error) |
| 47 | 173 | return error; |
| 48 | | − agf = agbp->b_addr; |
| 49 | 174 | |
| 50 | 175 | cur = xfs_allocbt_init_cursor(mp, NULL, agbp, pag, XFS_BTNUM_CNT); |
| 51 | 176 | |
| 52 | 177 | /* |
| 53 | | − * Look up the longest btree in the AGF and start with it. |
| 178 | + * Look up the extent length requested in the AGF and start with it. |
| 54 | 179 | */ |
| 55 | | − error = xfs_alloc_lookup_ge(cur, 0, be32_to_cpu(agf->agf_longest), &i); |
| 180 | + if (tcur->ar_startblock == NULLAGBLOCK) |
| 181 | + error = xfs_alloc_lookup_ge(cur, 0, tcur->ar_blockcount, &i); |
| 182 | + else |
| 183 | + error = xfs_alloc_lookup_le(cur, tcur->ar_startblock, |
| 184 | + tcur->ar_blockcount, &i); |
| 56 | 185 | if (error) |
| 57 | 186 | goto out_del_cursor; |
| 187 | + if (i == 0) { |
| 188 | + /* nothing of that length left in the AG, we are done */ |
| 189 | + tcur->ar_blockcount = 0; |
| 190 | + goto out_del_cursor; |
| 191 | + } |
| 58 | 192 | |
| 59 | 193 | /* |
| 60 | 194 | * Loop until we are done with all extents that are large |
| 61 | | − * enough to be worth discarding. |
| 195 | + * enough to be worth discarding or we hit batch limits. |
| 62 | 196 | */ |
| 63 | 197 | while (i) { |
| 64 | 198 | xfs_agblock_t fbno; |
| @@ -73,7 +207,16 @@ xfs_trim_extents( · 8 unmodified lines |
| 73 | 207 | error = -EFSCORRUPTED; |
| 74 | 208 | break; |
| 75 | 209 | } |
| 76 | | − ASSERT(flen <= be32_to_cpu(agf->agf_longest)); |
| 210 | + |
| 211 | + if (--batch <= 0) { |
| 212 | + /* |
| 213 | + * Update the cursor to point at this extent so we |
| 214 | + * restart the next batch from this extent. |
| 215 | + */ |
| 216 | + tcur->ar_startblock = fbno; |
| 217 | + tcur->ar_blockcount = flen; |
| 218 | + break; |
| 219 | + } |
| 77 | 220 | |
| 78 | 221 | /* |
| 79 | 222 | * use daddr format for all range/len calculations as that is |
| @@ -88,6 +231,7 @@ xfs_trim_extents( · 8 unmodified lines |
| 88 | 231 | */ |
| 89 | 232 | if (dlen < minlen) { |
| 90 | 233 | trace_xfs_discard_toosmall(mp, pag->pag_agno, fbno, flen); |
| 234 | + tcur->ar_blockcount = 0; |
| 91 | 235 | break; |
| 92 | 236 | } |
| 93 | 237 | |
| @@ -110,29 +254,103 @@ xfs_trim_extents( · 16 unmodified lines |
| 110 | 254 | goto next_extent; |
| 111 | 255 | } |
| 112 | 256 | |
| 113 | | − trace_xfs_discard_extent(mp, pag->pag_agno, fbno, flen); |
| 114 | | − error = blkdev_issue_discard(bdev, dbno, dlen, GFP_NOFS); |
| 115 | | − if (error) |
| 116 | | − break; |
| 257 | + xfs_extent_busy_insert_discard(pag, fbno, flen, |
| 258 | + &extents->extent_list); |
| 117 | 259 | *blocks_trimmed += flen; |
| 118 | | − |
| 119 | 260 | next_extent: |
| 120 | 261 | error = xfs_btree_decrement(cur, 0, &i); |
| 121 | 262 | if (error) |
| 122 | 263 | break; |
| 123 | 264 | |
| 124 | | − if (fatal_signal_pending(current)) { |
| 125 | | − error = -ERESTARTSYS; |
| 126 | | − break; |
| 127 | | − } |
| 265 | + /* |
| 266 | + * If there's no more records in the tree, we are done. Set the |
| 267 | + * cursor block count to 0 to indicate to the caller that there |
| 268 | + * is no more extents to search. |
| 269 | + */ |
| 270 | + if (i == 0) |
| 271 | + tcur->ar_blockcount = 0; |
| 128 | 272 | } |
| 129 | 273 | |
| 274 | + /* |
| 275 | + * If there was an error, release all the gathered busy extents because |
| 276 | + * we aren't going to issue a discard on them any more. |
| 277 | + */ |
| 278 | + if (error) |
| 279 | + xfs_extent_busy_clear(mp, &extents->extent_list, false); |
| 130 | 280 | out_del_cursor: |
| 131 | 281 | xfs_btree_del_cursor(cur, error); |
| 132 | 282 | xfs_buf_relse(agbp); |
| 133 | 283 | return error; |
| 134 | 284 | } |
| 135 | 285 | |
| 286 | +static bool |
| 287 | +xfs_trim_should_stop(void) |
| 288 | +{ |
| 289 | + return fatal_signal_pending(current) || freezing(current); |
| 290 | +} |
| 291 | + |
| 292 | +/* |
| 293 | + * Iterate the free list gathering extents and discarding them. We need a cursor |
| 294 | + * for the repeated iteration of gather/discard loop, so use the longest extent |
| 295 | + * we found in the last batch as the key to start the next. |
| 296 | + */ |
| 297 | +static int |
| 298 | +xfs_trim_extents( |
| 299 | + struct xfs_perag *pag, |
| 300 | + xfs_daddr_t start, |
| 301 | + xfs_daddr_t end, |
| 302 | + xfs_daddr_t minlen, |
| 303 | + uint64_t *blocks_trimmed) |
| 304 | +{ |
| 305 | + struct xfs_alloc_rec_incore tcur = { |
| 306 | + .ar_blockcount = pag->pagf_longest, |
| 307 | + .ar_startblock = NULLAGBLOCK, |
| 308 | + }; |
| 309 | + int error = 0; |
| 310 | + |
| 311 | + do { |
| 312 | + struct xfs_busy_extents *extents; |
| 313 | + |
| 314 | + extents = kzalloc(sizeof(*extents), GFP_KERNEL); |
| 315 | + if (!extents) { |
| 316 | + error = -ENOMEM; |
| 317 | + break; |
| 318 | + } |
| 319 | + |
| 320 | + extents->mount = pag->pag_mount; |
| 321 | + extents->owner = extents; |
| 322 | + INIT_LIST_HEAD(&extents->extent_list); |
| 323 | + |
| 324 | + error = xfs_trim_gather_extents(pag, start, end, minlen, |
| 325 | + &tcur, extents, blocks_trimmed); |
| 326 | + if (error) { |
| 327 | + kfree(extents); |
| 328 | + break; |
| 329 | + } |
| 330 | + |
| 331 | + /* |
| 332 | + * We hand the extent list to the discard function here so the |
| 333 | + * discarded extents can be removed from the busy extent list. |
| 334 | + * This allows the discards to run asynchronously with gathering |
| 335 | + * the next round of extents to discard. |
| 336 | + * |
| 337 | + * However, we must ensure that we do not reference the extent |
| 338 | + * list after this function call, as it may have been freed by |
| 339 | + * the time control returns to us. |
| 340 | + */ |
| 341 | + error = xfs_discard_extents(pag->pag_mount, extents); |
| 342 | + if (error) |
| 343 | + break; |
| 344 | + |
| 345 | + if (xfs_trim_should_stop()) |
| 346 | + break; |
| 347 | + |
| 348 | + } while (tcur.ar_blockcount != 0); |
| 349 | + |
| 350 | + return error; |
| 351 | + |
| 352 | +} |
| 353 | + |
| 136 | 354 | /* |
| 137 | 355 | * trim a range of the filesystem. |
| 138 | 356 | * |
| @@ -195,12 +413,12 @@ xfs_ioc_trim( · 56 unmodified lines |
| 195 | 413 | for_each_perag_range(mp, agno, xfs_daddr_to_agno(mp, end), pag) { |
| 196 | 414 | error = xfs_trim_extents(pag, start, end, minlen, |
| 197 | 415 | &blocks_trimmed); |
| 198 | | − if (error) { |
| 416 | + if (error) |
| 199 | 417 | last_error = error; |
| 200 | | − if (error == -ERESTARTSYS) { |
| 201 | | − xfs_perag_rele(pag); |
| 202 | | − break; |
| 203 | | − } |
| 418 | + |
| 419 | + if (xfs_trim_should_stop()) { |
| 420 | + xfs_perag_rele(pag); |
| 421 | + break; |
| 204 | 422 | } |
| 205 | 423 | } |
| 206 | 424 | |