summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--fs/bcachefs/checksum.h11
-rw-r--r--fs/bcachefs/data_update.c2
-rw-r--r--fs/bcachefs/ec.c2
-rw-r--r--fs/bcachefs/io_write.c4
-rw-r--r--fs/bcachefs/migrate.c2
-rw-r--r--fs/bcachefs/rebalance.c166
-rw-r--r--fs/bcachefs/rebalance.h10
-rw-r--r--fs/bcachefs/sb-downgrade.c3
-rw-r--r--fs/bcachefs/sb-errors_format.h4
9 files changed, 184 insertions, 20 deletions
diff --git a/fs/bcachefs/checksum.h b/fs/bcachefs/checksum.h
index 10bfadcde80a..362846d5bb87 100644
--- a/fs/bcachefs/checksum.h
+++ b/fs/bcachefs/checksum.h
@@ -143,6 +143,17 @@ static inline enum bch_csum_type bch2_data_checksum_type(struct bch_fs *c,
return bch2_csum_opt_to_type(opts.data_checksum, true);
}
+static inline enum bch_csum_type bch2_data_checksum_type_rb(struct bch_fs *c,
+ struct bch_extent_rebalance opts)
+{
+ if (c->sb.encryption_type)
+ return c->opts.wide_macs
+ ? BCH_CSUM_chacha20_poly1305_128
+ : BCH_CSUM_chacha20_poly1305_80;
+
+ return bch2_csum_opt_to_type(opts.data_checksum, true);
+}
+
static inline enum bch_csum_type bch2_meta_checksum_type(struct bch_fs *c)
{
if (c->sb.encryption_type)
diff --git a/fs/bcachefs/data_update.c b/fs/bcachefs/data_update.c
index 894ba0944fea..dbdf64b40829 100644
--- a/fs/bcachefs/data_update.c
+++ b/fs/bcachefs/data_update.c
@@ -438,7 +438,7 @@ restart_drop_extra_replicas:
bch2_insert_snapshot_whiteouts(trans, m->btree_id,
k.k->p, insert->k.p) ?:
bch2_inum_snapshot_opts_get(trans, k.k->p.inode, k.k->p.snapshot, &opts) ?:
- bch2_bkey_set_needs_rebalance(c, &opts, insert,
+ bch2_bkey_set_needs_rebalance(trans, NULL, &opts, insert,
SET_NEEDS_REBALANCE_foreground,
m->op.opts.change_cookie) ?:
bch2_trans_update(trans, &iter, insert,
diff --git a/fs/bcachefs/ec.c b/fs/bcachefs/ec.c
index 89a95b6c4e51..103719a76c81 100644
--- a/fs/bcachefs/ec.c
+++ b/fs/bcachefs/ec.c
@@ -1134,7 +1134,7 @@ static int ec_stripe_update_extent(struct btree_trans *trans,
ret = bch2_extent_get_io_opts_one(trans, &opts, &iter, bkey_i_to_s_c(n),
SET_NEEDS_REBALANCE_other) ?:
- bch2_bkey_set_needs_rebalance(trans->c, &opts, n,
+ bch2_bkey_set_needs_rebalance(trans, NULL, &opts, n,
SET_NEEDS_REBALANCE_other, 0) ?:
bch2_trans_update(trans, &iter, n, 0);
out:
diff --git a/fs/bcachefs/io_write.c b/fs/bcachefs/io_write.c
index 6a5da02ce266..ccbca802db0b 100644
--- a/fs/bcachefs/io_write.c
+++ b/fs/bcachefs/io_write.c
@@ -365,7 +365,7 @@ int bch2_extent_update(struct btree_trans *trans,
min(k->k.p.offset << 9, new_i_size),
i_sectors_delta, &inode) ?:
(bch2_inode_opts_get_inode(c, &inode, &opts),
- bch2_bkey_set_needs_rebalance(c, &opts, k,
+ bch2_bkey_set_needs_rebalance(trans, NULL, &opts, k,
SET_NEEDS_REBALANCE_foreground,
change_cookie)) ?:
bch2_trans_update(trans, iter, k, 0) ?:
@@ -1271,7 +1271,7 @@ static int bch2_nocow_write_convert_one_unwritten(struct btree_trans *trans,
return bch2_extent_update_i_size_sectors(trans, iter,
min(new->k.p.offset << 9, new_i_size), 0, &inode) ?:
(bch2_inode_opts_get_inode(c, &inode, &opts),
- bch2_bkey_set_needs_rebalance(c, &opts, new,
+ bch2_bkey_set_needs_rebalance(trans, NULL, &opts, new,
SET_NEEDS_REBALANCE_foreground,
op->opts.change_cookie)) ?:
bch2_trans_update(trans, iter, new,
diff --git a/fs/bcachefs/migrate.c b/fs/bcachefs/migrate.c
index 139a6587a64e..9b172af4f8c8 100644
--- a/fs/bcachefs/migrate.c
+++ b/fs/bcachefs/migrate.c
@@ -84,7 +84,7 @@ static int bch2_dev_usrdata_drop_key(struct btree_trans *trans,
struct bch_inode_opts opts;
ret = bch2_extent_get_apply_io_opts_one(trans, &opts, iter, k, ctx) ?:
- bch2_bkey_set_needs_rebalance(c, &opts, n, ctx, 0) ?:
+ bch2_bkey_set_needs_rebalance(trans, NULL, &opts, n, ctx, 0) ?:
drop_dev_ptrs(c, bkey_i_to_s(n), dev_idx, flags, err, false);
if (ret)
return ret;
diff --git a/fs/bcachefs/rebalance.c b/fs/bcachefs/rebalance.c
index f2d0040d043c..e850b7d872e2 100644
--- a/fs/bcachefs/rebalance.c
+++ b/fs/bcachefs/rebalance.c
@@ -25,6 +25,8 @@
#include <linux/kthread.h>
#include <linux/sched/cputime.h>
+#define REBALANCE_WORK_SCAN_OFFSET (U64_MAX - 1)
+
/* bch_extent_rebalance: */
static const struct bch_extent_rebalance *bch2_bkey_ptrs_rebalance_opts(struct bkey_ptrs_c ptrs)
@@ -237,6 +239,112 @@ bch2_bkey_needs_rebalance(struct bch_fs *c, struct bkey_s_c k,
return r;
}
+static int check_rebalance_scan_cookie(struct btree_trans *trans, u64 inum, bool *v)
+{
+ if (v && *v)
+ return 1;
+
+ /*
+ * If opts need to be propagated to the extent, a scan cookie should be
+ * present:
+ */
+ CLASS(btree_iter, iter)(trans, BTREE_ID_rebalance_work,
+ SPOS(inum, REBALANCE_WORK_SCAN_OFFSET, U32_MAX),
+ 0);
+ struct bkey_s_c k = bch2_btree_iter_peek_slot(&iter);
+ int ret = bkey_err(k);
+ if (ret)
+ return ret;
+
+ ret = k.k->type == KEY_TYPE_cookie;
+ if (v)
+ *v = ret;
+ return ret;
+}
+
+static int new_needs_rb_allowed(struct btree_trans *trans,
+ struct per_snapshot_io_opts *s,
+ struct bkey_s_c k,
+ enum set_needs_rebalance_ctx ctx,
+ unsigned opt_change_cookie,
+ const struct bch_extent_rebalance *old,
+ const struct bch_extent_rebalance *new,
+ unsigned new_need_rb)
+{
+ struct bch_fs *c = trans->c;
+ /*
+ * New need_rb - pointers that don't match the current io path options -
+ * are only allowed in certain situations:
+ *
+ * Propagating new options: from bch2_set_rebalance_needs_scan
+ *
+ * Foreground writes: background_compression and background_target are
+ * allowed
+ *
+ * Foreground writes: we may have raced with an option change:
+ * opt_change_cookie checks for this
+ *
+ * XXX: foreground writes should still match compression,
+ * foreground_target - figure out how to check for this
+ */
+ if (ctx == SET_NEEDS_REBALANCE_opt_change ||
+ ctx == SET_NEEDS_REBALANCE_opt_change_indirect)
+ return 0;
+
+ if (ctx == SET_NEEDS_REBALANCE_foreground) {
+ new_need_rb &= ~(BIT(BCH_REBALANCE_background_compression)|
+ BIT(BCH_REBALANCE_background_target));
+ if (!new_need_rb)
+ return 0;
+
+ if (opt_change_cookie != atomic_read(&c->opt_change_cookie))
+ return 0;
+ }
+
+ /*
+ * Either the extent data or the extent io options (from
+ * bch_extent_rebalance) should match the io_opts from the
+ * inode/filesystem, unless
+ *
+ * - There's a scan pending to propagate new options
+ * - It's an indirect extent: it may be referenced by inodes
+ * with inconsistent options
+ *
+ * For efficiency (so that we can cache checking for scan
+ * cookies), only check option consistency when we're called
+ * with snapshot_io_opts - don't bother when we're called from
+ * move_data_phys() -> get_io_opts_one()
+ *
+ * Note that we can cache the existence of a cookie, but not the
+ * non-existence, to avoid spurious false positives.
+ */
+ int ret = check_rebalance_scan_cookie(trans, 0, s ? &s->fs_scan_cookie : NULL) ?:
+ check_rebalance_scan_cookie(trans, k.k->p.inode, s ? &s->inum_scan_cookie : NULL);
+ if (ret < 0)
+ return ret;
+ if (ret)
+ return 0;
+
+ CLASS(printbuf, buf)();
+
+ prt_printf(&buf, "extent with incorrect/missing rebalance opts:\n");
+ bch2_bkey_val_to_text(&buf, c, k);
+
+ const struct bch_extent_rebalance _old = {};
+ if (!old)
+ old = &_old;
+
+#define x(_name) \
+ if (new_need_rb & BIT(BCH_REBALANCE_##_name)) \
+ prt_printf(&buf, "\n" #_name " %u != %u", old->_name, new->_name);
+ BCH_REBALANCE_OPTS()
+#undef x
+
+ fsck_err(trans, extent_io_opts_not_set, "%s", buf.buf);
+fsck_err:
+ return ret;
+}
+
static inline bool bkey_should_have_rb_opts(struct bkey_s_c k,
struct bch_extent_rebalance new)
{
@@ -248,7 +356,8 @@ static inline bool bkey_should_have_rb_opts(struct bkey_s_c k,
return new.need_rb;
}
-int bch2_bkey_set_needs_rebalance(struct bch_fs *c,
+int bch2_bkey_set_needs_rebalance(struct btree_trans *trans,
+ struct per_snapshot_io_opts *snapshot_io_opts,
struct bch_inode_opts *opts,
struct bkey_i *_k,
enum set_needs_rebalance_ctx ctx,
@@ -257,6 +366,7 @@ int bch2_bkey_set_needs_rebalance(struct bch_fs *c,
if (!bkey_extent_is_direct_data(&_k->k))
return 0;
+ struct bch_fs *c = trans->c;
struct bkey_s k = bkey_i_to_s(_k);
struct bch_extent_rebalance *old =
(struct bch_extent_rebalance *) bch2_bkey_rebalance_opts(k.s_c);
@@ -273,6 +383,16 @@ int bch2_bkey_set_needs_rebalance(struct bch_fs *c,
(should_have_rb ? !memcmp(old, &new, sizeof(new)) : !old))
return 0;
+ unsigned new_need_rb = new.need_rb & ~(old ? old->need_rb : 0);
+
+ if (unlikely(new_need_rb)) {
+ int ret = new_needs_rb_allowed(trans, snapshot_io_opts,
+ k.s_c, ctx, opt_change_cookie,
+ old, &new, new_need_rb);
+ if (ret)
+ return ret;
+ }
+
if (should_have_rb) {
if (!old) {
old = bkey_val_end(k);
@@ -287,6 +407,7 @@ int bch2_bkey_set_needs_rebalance(struct bch_fs *c,
}
static int bch2_get_update_rebalance_opts(struct btree_trans *trans,
+ struct per_snapshot_io_opts *snapshot_io_opts,
struct bch_inode_opts *io_opts,
struct btree_iter *iter,
struct bkey_s_c k,
@@ -324,7 +445,7 @@ static int bch2_get_update_rebalance_opts(struct btree_trans *trans,
/* On successfull transaction commit, @k was invalidated: */
- return bch2_bkey_set_needs_rebalance(c, io_opts, n, ctx, 0) ?:
+ return bch2_bkey_set_needs_rebalance(trans, snapshot_io_opts, io_opts, n, ctx, 0) ?:
bch2_trans_update(trans, iter, n, BTREE_UPDATE_internal_snapshot_node) ?:
bch2_trans_commit(trans, NULL, NULL, 0) ?:
bch_err_throw(c, transaction_restart_commit);
@@ -367,7 +488,8 @@ static struct bch_inode_opts *bch2_extent_get_io_opts(struct btree_trans *trans,
darray_push(&io_opts->d, e);
}));
- io_opts->cur_inum = extent_pos.inode;
+ io_opts->cur_inum = extent_pos.inode;
+ io_opts->inum_scan_cookie = false;
}
ret = ret ?: trans_was_restarted(trans, restart_count);
@@ -395,7 +517,8 @@ struct bch_inode_opts *bch2_extent_get_apply_io_opts(struct btree_trans *trans,
if (IS_ERR(opts) || btree_iter_path(trans, extent_iter)->level)
return opts;
- int ret = bch2_get_update_rebalance_opts(trans, opts, extent_iter, extent_k, ctx);
+ int ret = bch2_get_update_rebalance_opts(trans, snapshot_io_opts, opts,
+ extent_iter, extent_k, ctx);
return ret ? ERR_PTR(ret) : opts;
}
@@ -439,11 +562,9 @@ int bch2_extent_get_apply_io_opts_one(struct btree_trans *trans,
if (ret || btree_iter_path(trans, extent_iter)->level)
return ret;
- return bch2_get_update_rebalance_opts(trans, io_opts, extent_iter, extent_k, ctx);
+ return bch2_get_update_rebalance_opts(trans, NULL, io_opts, extent_iter, extent_k, ctx);
}
-#define REBALANCE_WORK_SCAN_OFFSET (U64_MAX - 1)
-
static const char * const bch2_rebalance_state_strs[] = {
#define x(t) #t,
BCH_REBALANCE_STATES()
@@ -691,6 +812,7 @@ out:
static int do_rebalance_scan_indirect(struct btree_trans *trans,
struct bkey_s_c_reflink_p p,
+ struct per_snapshot_io_opts *snapshot_io_opts,
struct bch_inode_opts *opts)
{
u64 idx = REFLINK_P_IDX(p.v) - le32_to_cpu(p.v->front_pad);
@@ -703,7 +825,7 @@ static int do_rebalance_scan_indirect(struct btree_trans *trans,
BTREE_ITER_not_extents, k, ({
if (bpos_ge(bkey_start_pos(k.k), POS(0, end)))
break;
- bch2_get_update_rebalance_opts(trans, opts, &iter, k,
+ bch2_get_update_rebalance_opts(trans, snapshot_io_opts, opts, &iter, k,
SET_NEEDS_REBALANCE_opt_change_indirect);
}));
if (ret)
@@ -751,7 +873,8 @@ static int do_rebalance_scan(struct moving_context *ctxt,
(inum &&
k.k->type == KEY_TYPE_reflink_p &&
REFLINK_P_MAY_UPDATE_OPTIONS(bkey_s_c_to_reflink_p(k).v)
- ? do_rebalance_scan_indirect(trans, bkey_s_c_to_reflink_p(k), opts)
+ ? do_rebalance_scan_indirect(trans, bkey_s_c_to_reflink_p(k),
+ snapshot_io_opts, opts)
: 0);
}));
if (ret)
@@ -1050,6 +1173,7 @@ int bch2_fs_rebalance_init(struct bch_fs *c)
static int check_rebalance_work_one(struct btree_trans *trans,
struct btree_iter *extent_iter,
struct btree_iter *rebalance_iter,
+ struct per_snapshot_io_opts *snapshot_io_opts,
struct bkey_buf *last_flushed)
{
struct bch_fs *c = trans->c;
@@ -1119,6 +1243,21 @@ static int check_rebalance_work_one(struct btree_trans *trans,
return ret;
}
+ struct bch_inode_opts *opts = bch2_extent_get_apply_io_opts(trans,
+ snapshot_io_opts, extent_iter->pos, extent_iter, extent_k,
+ SET_NEEDS_REBALANCE_other);
+ ret = PTR_ERR_OR_ZERO(opts);
+ if (ret == -BCH_ERR_transaction_restart_commit) {
+ /*
+ * If get_apply_io_opts() did work, just advance and check the
+ * next key; it may have updated the rebalance_work btree so
+ * we'd need a write buffer flush to check what it just did.
+ */
+ ret = 0;
+ }
+ if (ret)
+ return ret;
+
if (cmp <= 0)
bch2_btree_iter_advance(extent_iter);
if (cmp >= 0)
@@ -1131,10 +1270,14 @@ int bch2_check_rebalance_work(struct bch_fs *c)
{
CLASS(btree_trans, trans)(c);
CLASS(btree_iter, extent_iter)(trans, BTREE_ID_reflink, POS_MIN,
+ BTREE_ITER_not_extents|
BTREE_ITER_prefetch);
CLASS(btree_iter, rebalance_iter)(trans, BTREE_ID_rebalance_work, POS_MIN,
BTREE_ITER_prefetch);
+ struct per_snapshot_io_opts snapshot_io_opts;
+ per_snapshot_io_opts_init(&snapshot_io_opts, c);
+
struct bkey_buf last_flushed;
bch2_bkey_buf_init(&last_flushed);
bkey_init(&last_flushed.k->k);
@@ -1148,12 +1291,15 @@ int bch2_check_rebalance_work(struct bch_fs *c)
bch2_trans_begin(trans);
- ret = check_rebalance_work_one(trans, &extent_iter, &rebalance_iter, &last_flushed);
+ ret = check_rebalance_work_one(trans, &extent_iter, &rebalance_iter,
+ &snapshot_io_opts, &last_flushed) ?:
+ bch2_trans_commit(trans, NULL, NULL, 0);
if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
ret = 0;
}
+ per_snapshot_io_opts_exit(&snapshot_io_opts);
bch2_bkey_buf_exit(&last_flushed, c);
return ret < 0 ? ret : 0;
}
diff --git a/fs/bcachefs/rebalance.h b/fs/bcachefs/rebalance.h
index ae576d8af609..767de863fb53 100644
--- a/fs/bcachefs/rebalance.h
+++ b/fs/bcachefs/rebalance.h
@@ -56,9 +56,6 @@ enum set_needs_rebalance_ctx {
SET_NEEDS_REBALANCE_other,
};
-int bch2_bkey_set_needs_rebalance(struct bch_fs *, struct bch_inode_opts *,
- struct bkey_i *, enum set_needs_rebalance_ctx, u32);
-
/* Inodes in different snapshots may have different IO options: */
struct snapshot_io_opts_entry {
u32 snapshot;
@@ -67,6 +64,9 @@ struct snapshot_io_opts_entry {
struct per_snapshot_io_opts {
u64 cur_inum;
+ bool fs_scan_cookie;
+ bool inum_scan_cookie;
+
struct bch_inode_opts fs_io_opts;
DARRAY(struct snapshot_io_opts_entry) d;
};
@@ -82,6 +82,10 @@ static inline void per_snapshot_io_opts_exit(struct per_snapshot_io_opts *io_opt
darray_exit(&io_opts->d);
}
+int bch2_bkey_set_needs_rebalance(struct btree_trans *,
+ struct per_snapshot_io_opts *, struct bch_inode_opts *,
+ struct bkey_i *, enum set_needs_rebalance_ctx, u32);
+
struct bch_inode_opts *bch2_extent_get_apply_io_opts(struct btree_trans *,
struct per_snapshot_io_opts *, struct bpos,
struct btree_iter *, struct bkey_s_c,
diff --git a/fs/bcachefs/sb-downgrade.c b/fs/bcachefs/sb-downgrade.c
index 0d47e83c28a6..66b7f19f0437 100644
--- a/fs/bcachefs/sb-downgrade.c
+++ b/fs/bcachefs/sb-downgrade.c
@@ -109,7 +109,8 @@
BIT_ULL(BCH_RECOVERY_PASS_check_allocations), \
BCH_FSCK_ERR_accounting_mismatch) \
x(rebalance_v2, \
- BIT_ULL(BCH_RECOVERY_PASS_check_rebalance_work))
+ BIT_ULL(BCH_RECOVERY_PASS_check_rebalance_work), \
+ BCH_FSCK_ERR_extent_io_opts_not_set)
#define DOWNGRADE_TABLE() \
x(bucket_stripe_sectors, \
diff --git a/fs/bcachefs/sb-errors_format.h b/fs/bcachefs/sb-errors_format.h
index 77e3fc92e39b..04c1da1e664a 100644
--- a/fs/bcachefs/sb-errors_format.h
+++ b/fs/bcachefs/sb-errors_format.h
@@ -339,7 +339,9 @@ enum bch_fsck_flags {
x(dirent_stray_data_after_cf_name, 305, 0) \
x(rebalance_work_incorrectly_set, 309, FSCK_AUTOFIX) \
x(rebalance_work_incorrectly_unset, 310, FSCK_AUTOFIX) \
- x(MAX, 328, 0)
+ x(extent_io_opts_not_set, 328, FSCK_AUTOFIX) \
+ x(extent_io_opts_unneeded, 329, FSCK_AUTOFIX) \
+ x(MAX, 330, 0)
enum bch_sb_error_id {
#define x(t, n, ...) BCH_FSCK_ERR_##t = n,
generated by cgit v1.2.3 (git 2.51.0) at 2025-09-20 03:55:48 +0000