#include "bkey_buf.h"
#include "btree_update.h"
#include "buckets.h"
+#include "error.h"
#include "extents.h"
#include "inode.h"
-#include "io.h"
+#include "io_misc.h"
+#include "io_write.h"
+#include "rebalance.h"
#include "reflink.h"
#include "subvolume.h"
+#include "super-io.h"
#include <linux/sched/signal.h>
/* reflink pointers */
-int bch2_reflink_p_invalid(const struct bch_fs *c, struct bkey_s_c k,
- int rw, struct printbuf *err)
+int bch2_reflink_p_invalid(struct bch_fs *c, struct bkey_s_c k,
+ enum bkey_invalid_flags flags,
+ struct printbuf *err)
{
struct bkey_s_c_reflink_p p = bkey_s_c_to_reflink_p(k);
- if (bkey_val_bytes(p.k) != sizeof(*p.v)) {
- prt_printf(err, "incorrect value size (%zu != %zu)",
- bkey_val_bytes(p.k), sizeof(*p.v));
- return -EINVAL;
- }
-
if (c->sb.version >= bcachefs_metadata_version_reflink_p_fix &&
le64_to_cpu(p.v->idx) < le32_to_cpu(p.v->front_pad)) {
prt_printf(err, "idx < front_pad (%llu < %u)",
return true;
}
-/* indirect extents */
+static int trans_mark_reflink_p_segment(struct btree_trans *trans,
+ struct bkey_s_c_reflink_p p,
+ u64 *idx, unsigned flags)
+{
+ struct bch_fs *c = trans->c;
+ struct btree_iter iter;
+ struct bkey_i *k;
+ __le64 *refcount;
+ int add = !(flags & BTREE_TRIGGER_OVERWRITE) ? 1 : -1;
+ struct printbuf buf = PRINTBUF;
+ int ret;
+
+ k = bch2_bkey_get_mut_noupdate(trans, &iter,
+ BTREE_ID_reflink, POS(0, *idx),
+ BTREE_ITER_WITH_UPDATES);
+ ret = PTR_ERR_OR_ZERO(k);
+ if (ret)
+ goto err;
+
+ refcount = bkey_refcount(k);
+ if (!refcount) {
+ bch2_bkey_val_to_text(&buf, c, p.s_c);
+ bch2_trans_inconsistent(trans,
+ "nonexistent indirect extent at %llu while marking\n %s",
+ *idx, buf.buf);
+ ret = -EIO;
+ goto err;
+ }
+
+ if (!*refcount && (flags & BTREE_TRIGGER_OVERWRITE)) {
+ bch2_bkey_val_to_text(&buf, c, p.s_c);
+ bch2_trans_inconsistent(trans,
+ "indirect extent refcount underflow at %llu while marking\n %s",
+ *idx, buf.buf);
+ ret = -EIO;
+ goto err;
+ }
-int bch2_reflink_v_invalid(const struct bch_fs *c, struct bkey_s_c k,
- int rw, struct printbuf *err)
+ if (flags & BTREE_TRIGGER_INSERT) {
+ struct bch_reflink_p *v = (struct bch_reflink_p *) p.v;
+ u64 pad;
+
+ pad = max_t(s64, le32_to_cpu(v->front_pad),
+ le64_to_cpu(v->idx) - bkey_start_offset(&k->k));
+ BUG_ON(pad > U32_MAX);
+ v->front_pad = cpu_to_le32(pad);
+
+ pad = max_t(s64, le32_to_cpu(v->back_pad),
+ k->k.p.offset - p.k->size - le64_to_cpu(v->idx));
+ BUG_ON(pad > U32_MAX);
+ v->back_pad = cpu_to_le32(pad);
+ }
+
+ le64_add_cpu(refcount, add);
+
+ bch2_btree_iter_set_pos_to_extent_start(&iter);
+ ret = bch2_trans_update(trans, &iter, k, 0);
+ if (ret)
+ goto err;
+
+ *idx = k->k.p.offset;
+err:
+ bch2_trans_iter_exit(trans, &iter);
+ printbuf_exit(&buf);
+ return ret;
+}
+
+static int __trans_mark_reflink_p(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
+ struct bkey_s_c k, unsigned flags)
{
- struct bkey_s_c_reflink_v r = bkey_s_c_to_reflink_v(k);
+ struct bkey_s_c_reflink_p p = bkey_s_c_to_reflink_p(k);
+ u64 idx, end_idx;
+ int ret = 0;
+
+ idx = le64_to_cpu(p.v->idx) - le32_to_cpu(p.v->front_pad);
+ end_idx = le64_to_cpu(p.v->idx) + p.k->size +
+ le32_to_cpu(p.v->back_pad);
+
+ while (idx < end_idx && !ret)
+ ret = trans_mark_reflink_p_segment(trans, p, &idx, flags);
+ return ret;
+}
+
+int bch2_trans_mark_reflink_p(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
+ struct bkey_s_c old,
+ struct bkey_i *new,
+ unsigned flags)
+{
+ if (flags & BTREE_TRIGGER_INSERT) {
+ struct bch_reflink_p *v = &bkey_i_to_reflink_p(new)->v;
- if (bkey_val_bytes(r.k) < sizeof(*r.v)) {
- prt_printf(err, "incorrect value size (%zu < %zu)",
- bkey_val_bytes(r.k), sizeof(*r.v));
- return -BCH_ERR_invalid_bkey;
+ v->front_pad = v->back_pad = 0;
}
- return bch2_bkey_ptrs_invalid(c, k, rw, err);
+ return trigger_run_overwrite_then_insert(__trans_mark_reflink_p, trans, btree_id, level, old, new, flags);
+}
+
+static s64 __bch2_mark_reflink_p(struct btree_trans *trans,
+ struct bkey_s_c_reflink_p p,
+ u64 start, u64 end,
+ u64 *idx, unsigned flags, size_t r_idx)
+{
+ struct bch_fs *c = trans->c;
+ struct reflink_gc *r;
+ int add = !(flags & BTREE_TRIGGER_OVERWRITE) ? 1 : -1;
+ u64 next_idx = end;
+ s64 ret = 0;
+ struct printbuf buf = PRINTBUF;
+
+ if (r_idx >= c->reflink_gc_nr)
+ goto not_found;
+
+ r = genradix_ptr(&c->reflink_gc_table, r_idx);
+ next_idx = min(next_idx, r->offset - r->size);
+ if (*idx < next_idx)
+ goto not_found;
+
+ BUG_ON((s64) r->refcount + add < 0);
+
+ r->refcount += add;
+ *idx = r->offset;
+ return 0;
+not_found:
+ if (fsck_err(c, reflink_p_to_missing_reflink_v,
+ "pointer to missing indirect extent\n"
+ " %s\n"
+ " missing range %llu-%llu",
+ (bch2_bkey_val_to_text(&buf, c, p.s_c), buf.buf),
+ *idx, next_idx)) {
+ struct bkey_i_error *new;
+
+ new = bch2_trans_kmalloc(trans, sizeof(*new));
+ ret = PTR_ERR_OR_ZERO(new);
+ if (ret)
+ goto err;
+
+ bkey_init(&new->k);
+ new->k.type = KEY_TYPE_error;
+ new->k.p = bkey_start_pos(p.k);
+ new->k.p.offset += *idx - start;
+ bch2_key_resize(&new->k, next_idx - *idx);
+ ret = bch2_btree_insert_trans(trans, BTREE_ID_extents, &new->k_i,
+ BTREE_TRIGGER_NORUN);
+ }
+
+ *idx = next_idx;
+err:
+fsck_err:
+ printbuf_exit(&buf);
+ return ret;
+}
+
+static int __mark_reflink_p(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
+ struct bkey_s_c k, unsigned flags)
+{
+ struct bch_fs *c = trans->c;
+ struct bkey_s_c_reflink_p p = bkey_s_c_to_reflink_p(k);
+ struct reflink_gc *ref;
+ size_t l, r, m;
+ u64 idx = le64_to_cpu(p.v->idx), start = idx;
+ u64 end = le64_to_cpu(p.v->idx) + p.k->size;
+ int ret = 0;
+
+ BUG_ON(!(flags & BTREE_TRIGGER_GC));
+
+ if (c->sb.version_upgrade_complete >= bcachefs_metadata_version_reflink_p_fix) {
+ idx -= le32_to_cpu(p.v->front_pad);
+ end += le32_to_cpu(p.v->back_pad);
+ }
+
+ l = 0;
+ r = c->reflink_gc_nr;
+ while (l < r) {
+ m = l + (r - l) / 2;
+
+ ref = genradix_ptr(&c->reflink_gc_table, m);
+ if (ref->offset <= idx)
+ l = m + 1;
+ else
+ r = m;
+ }
+
+ while (idx < end && !ret)
+ ret = __bch2_mark_reflink_p(trans, p, start, end,
+ &idx, flags, l++);
+
+ return ret;
+}
+
+int bch2_mark_reflink_p(struct btree_trans *trans,
+ enum btree_id btree_id, unsigned level,
+ struct bkey_s_c old, struct bkey_s_c new,
+ unsigned flags)
+{
+ return mem_trigger_run_overwrite_then_insert(__mark_reflink_p, trans, btree_id, level, old, new, flags);
+}
+
+/* indirect extents */
+
+int bch2_reflink_v_invalid(struct bch_fs *c, struct bkey_s_c k,
+ enum bkey_invalid_flags flags,
+ struct printbuf *err)
+{
+ return bch2_bkey_ptrs_invalid(c, k, flags, err);
}
void bch2_reflink_v_to_text(struct printbuf *out, struct bch_fs *c,
bch2_bkey_ptrs_to_text(out, c, k);
}
+#if 0
+Currently disabled, needs to be debugged:
+
bool bch2_reflink_v_merge(struct bch_fs *c, struct bkey_s _l, struct bkey_s_c _r)
{
struct bkey_s_reflink_v l = bkey_s_to_reflink_v(_l);
return l.v->refcount == r.v->refcount && bch2_extent_merge(c, _l, _r);
}
+#endif
+
+static inline void check_indirect_extent_deleting(struct bkey_i *new, unsigned *flags)
+{
+ if ((*flags & BTREE_TRIGGER_INSERT) && !*bkey_refcount(new)) {
+ new->k.type = KEY_TYPE_deleted;
+ new->k.size = 0;
+ set_bkey_val_u64s(&new->k, 0);
+ *flags &= ~BTREE_TRIGGER_INSERT;
+ }
+}
int bch2_trans_mark_reflink_v(struct btree_trans *trans,
enum btree_id btree_id, unsigned level,
struct bkey_s_c old, struct bkey_i *new,
unsigned flags)
{
- if (!(flags & BTREE_TRIGGER_OVERWRITE)) {
- struct bkey_i_reflink_v *r = bkey_i_to_reflink_v(new);
-
- if (!r->v.refcount) {
- r->k.type = KEY_TYPE_deleted;
- r->k.size = 0;
- set_bkey_val_u64s(&r->k, 0);
- return 0;
- }
- }
+ check_indirect_extent_deleting(new, &flags);
+
+ if (old.k->type == KEY_TYPE_reflink_v &&
+ new->k.type == KEY_TYPE_reflink_v &&
+ old.k->u64s == new->k.u64s &&
+ !memcmp(bkey_s_c_to_reflink_v(old).v->start,
+ bkey_i_to_reflink_v(new)->v.start,
+ bkey_val_bytes(&new->k) - 8))
+ return 0;
return bch2_trans_mark_extent(trans, btree_id, level, old, new, flags);
}
/* indirect inline data */
-int bch2_indirect_inline_data_invalid(const struct bch_fs *c, struct bkey_s_c k,
- int rw, struct printbuf *err)
+int bch2_indirect_inline_data_invalid(struct bch_fs *c, struct bkey_s_c k,
+ enum bkey_invalid_flags flags,
+ struct printbuf *err)
{
- if (bkey_val_bytes(k.k) < sizeof(struct bch_indirect_inline_data)) {
- prt_printf(err, "incorrect value size (%zu < %zu)",
- bkey_val_bytes(k.k), sizeof(struct bch_indirect_inline_data));
- return -BCH_ERR_invalid_bkey;
- }
-
return 0;
}
void bch2_indirect_inline_data_to_text(struct printbuf *out,
- struct bch_fs *c, struct bkey_s_c k)
+ struct bch_fs *c, struct bkey_s_c k)
{
struct bkey_s_c_indirect_inline_data d = bkey_s_c_to_indirect_inline_data(k);
unsigned datalen = bkey_inline_data_bytes(k.k);
struct bkey_s_c old, struct bkey_i *new,
unsigned flags)
{
- if (!(flags & BTREE_TRIGGER_OVERWRITE)) {
- struct bkey_i_indirect_inline_data *r =
- bkey_i_to_indirect_inline_data(new);
-
- if (!r->v.refcount) {
- r->k.type = KEY_TYPE_deleted;
- r->k.size = 0;
- set_bkey_val_u64s(&r->k, 0);
- }
- }
+ check_indirect_extent_deleting(new, &flags);
return 0;
}
if (orig->k.type == KEY_TYPE_inline_data)
bch2_check_set_feature(c, BCH_FEATURE_reflink_inline_data);
- for_each_btree_key_norestart(trans, reflink_iter, BTREE_ID_reflink,
- POS(0, c->reflink_hint),
- BTREE_ITER_INTENT|BTREE_ITER_SLOTS, k, ret) {
- if (reflink_iter.pos.inode) {
- bch2_btree_iter_set_pos(&reflink_iter, POS_MIN);
- continue;
- }
-
- if (bkey_deleted(k.k) && orig->k.size <= k.k->size)
- break;
- }
-
+ bch2_trans_iter_init(trans, &reflink_iter, BTREE_ID_reflink, POS_MAX,
+ BTREE_ITER_INTENT);
+ k = bch2_btree_iter_peek_prev(&reflink_iter);
+ ret = bkey_err(k);
if (ret)
goto err;
- /* rewind iter to start of hole, if necessary: */
- bch2_btree_iter_set_pos_to_extent_start(&reflink_iter);
-
r_v = bch2_trans_kmalloc(trans, sizeof(__le64) + bkey_bytes(&orig->k));
ret = PTR_ERR_OR_ZERO(r_v);
if (ret)
orig->k.type = KEY_TYPE_reflink_p;
r_p = bkey_i_to_reflink_p(orig);
set_bkey_val_bytes(&r_p->k, sizeof(r_p->v));
+
+ /* FORTIFY_SOURCE is broken here, and doesn't provide unsafe_memset() */
+#if !defined(__NO_FORTIFY) && defined(__OPTIMIZE__) && defined(CONFIG_FORTIFY_SOURCE)
+ __underlying_memset(&r_p->v, 0, sizeof(r_p->v));
+#else
memset(&r_p->v, 0, sizeof(r_p->v));
+#endif
r_p->v.idx = cpu_to_le64(bkey_start_offset(&r_v->k));
ret = bch2_trans_update(trans, extent_iter, &r_p->k_i,
BTREE_UPDATE_INTERNAL_SNAPSHOT_NODE);
err:
- c->reflink_hint = reflink_iter.pos.offset;
bch2_trans_iter_exit(trans, &reflink_iter);
return ret;
u64 remap_sectors,
u64 new_i_size, s64 *i_sectors_delta)
{
- struct btree_trans trans;
+ struct btree_trans *trans;
struct btree_iter dst_iter, src_iter;
struct bkey_s_c src_k;
struct bkey_buf new_dst, new_src;
struct bpos dst_start = POS(dst_inum.inum, dst_offset);
struct bpos src_start = POS(src_inum.inum, src_offset);
struct bpos dst_end = dst_start, src_end = src_start;
+ struct bch_io_opts opts;
struct bpos src_want;
- u64 dst_done;
+ u64 dst_done = 0;
u32 dst_snapshot, src_snapshot;
int ret = 0, ret2 = 0;
- if (!percpu_ref_tryget_live(&c->writes))
- return -EROFS;
+ if (!bch2_write_ref_tryget(c, BCH_WRITE_REF_reflink))
+ return -BCH_ERR_erofs_no_writes;
bch2_check_set_feature(c, BCH_FEATURE_reflink);
bch2_bkey_buf_init(&new_dst);
bch2_bkey_buf_init(&new_src);
- bch2_trans_init(&trans, c, BTREE_ITER_MAX, 4096);
+ trans = bch2_trans_get(c);
+
+ ret = bch2_inum_opts_get(trans, src_inum, &opts);
+ if (ret)
+ goto err;
- bch2_trans_iter_init(&trans, &src_iter, BTREE_ID_extents, src_start,
+ bch2_trans_iter_init(trans, &src_iter, BTREE_ID_extents, src_start,
BTREE_ITER_INTENT);
- bch2_trans_iter_init(&trans, &dst_iter, BTREE_ID_extents, dst_start,
+ bch2_trans_iter_init(trans, &dst_iter, BTREE_ID_extents, dst_start,
BTREE_ITER_INTENT);
while ((ret == 0 ||
bkey_lt(dst_iter.pos, dst_end)) {
struct disk_reservation disk_res = { 0 };
- bch2_trans_begin(&trans);
+ bch2_trans_begin(trans);
if (fatal_signal_pending(current)) {
ret = -EINTR;
break;
}
- ret = bch2_subvolume_get_snapshot(&trans, src_inum.subvol,
+ ret = bch2_subvolume_get_snapshot(trans, src_inum.subvol,
&src_snapshot);
if (ret)
continue;
bch2_btree_iter_set_snapshot(&src_iter, src_snapshot);
- ret = bch2_subvolume_get_snapshot(&trans, dst_inum.subvol,
+ ret = bch2_subvolume_get_snapshot(trans, dst_inum.subvol,
&dst_snapshot);
if (ret)
continue;
continue;
if (bkey_lt(src_want, src_iter.pos)) {
- ret = bch2_fpunch_at(&trans, &dst_iter, dst_inum,
+ ret = bch2_fpunch_at(trans, &dst_iter, dst_inum,
min(dst_end.offset,
dst_iter.pos.offset +
src_iter.pos.offset - src_want.offset),
bch2_bkey_buf_reassemble(&new_src, c, src_k);
src_k = bkey_i_to_s_c(new_src.k);
- ret = bch2_make_extent_indirect(&trans, &src_iter,
+ ret = bch2_make_extent_indirect(trans, &src_iter,
new_src.k);
if (ret)
continue;
min(src_k.k->p.offset - src_want.offset,
dst_end.offset - dst_iter.pos.offset));
- ret = bch2_extent_update(&trans, dst_inum, &dst_iter,
- new_dst.k, &disk_res,
- new_i_size, i_sectors_delta,
- true);
+ ret = bch2_bkey_set_needs_rebalance(c, new_dst.k,
+ opts.background_target,
+ opts.background_compression) ?:
+ bch2_extent_update(trans, dst_inum, &dst_iter,
+ new_dst.k, &disk_res,
+ new_i_size, i_sectors_delta,
+ true);
bch2_disk_reservation_put(c, &disk_res);
}
- bch2_trans_iter_exit(&trans, &dst_iter);
- bch2_trans_iter_exit(&trans, &src_iter);
+ bch2_trans_iter_exit(trans, &dst_iter);
+ bch2_trans_iter_exit(trans, &src_iter);
BUG_ON(!ret && !bkey_eq(dst_iter.pos, dst_end));
BUG_ON(bkey_gt(dst_iter.pos, dst_end));
struct bch_inode_unpacked inode_u;
struct btree_iter inode_iter = { NULL };
- bch2_trans_begin(&trans);
+ bch2_trans_begin(trans);
- ret2 = bch2_inode_peek(&trans, &inode_iter, &inode_u,
+ ret2 = bch2_inode_peek(trans, &inode_iter, &inode_u,
dst_inum, BTREE_ITER_INTENT);
if (!ret2 &&
inode_u.bi_size < new_i_size) {
inode_u.bi_size = new_i_size;
- ret2 = bch2_inode_write(&trans, &inode_iter, &inode_u) ?:
- bch2_trans_commit(&trans, NULL, NULL,
- BTREE_INSERT_NOFAIL);
+ ret2 = bch2_inode_write(trans, &inode_iter, &inode_u) ?:
+ bch2_trans_commit(trans, NULL, NULL,
+ BCH_TRANS_COMMIT_no_enospc);
}
- bch2_trans_iter_exit(&trans, &inode_iter);
+ bch2_trans_iter_exit(trans, &inode_iter);
} while (bch2_err_matches(ret2, BCH_ERR_transaction_restart));
-
- bch2_trans_exit(&trans);
+err:
+ bch2_trans_put(trans);
bch2_bkey_buf_exit(&new_src, c);
bch2_bkey_buf_exit(&new_dst, c);
- percpu_ref_put(&c->writes);
+ bch2_write_ref_put(c, BCH_WRITE_REF_reflink);
return dst_done ?: ret ?: ret2;
}
projects / bcachefs-tools-debian / blobdiff