public inbox for linux-bcachefs@vger.kernel.org
 help / color / mirror / Atom feed
From: Kent Overstreet <kent.overstreet@linux.dev>
To: linux-bcachefs@vger.kernel.org
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Subject: [PATCH 18/34] bcachefs: Make topology errors autofix
Date: Fri, 29 Nov 2024 15:27:17 -0500	[thread overview]
Message-ID: <20241129202736.2713679-19-kent.overstreet@linux.dev> (raw)
In-Reply-To: <20241129202736.2713679-1-kent.overstreet@linux.dev>

These repair paths are well tested, we can repair them without explicit
user intervention

This also tweaks bch2_topology_error() so that we run topology repair if
we're in recovery, not just fsck.

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
---
 fs/bcachefs/btree_gc.c         |  2 +-
 fs/bcachefs/recovery.c         | 31 +++++++++++++++++++++++++------
 fs/bcachefs/sb-errors_format.h | 12 ++++++------
 3 files changed, 32 insertions(+), 13 deletions(-)

diff --git a/fs/bcachefs/btree_gc.c b/fs/bcachefs/btree_gc.c
index 2e8cfc4d3265..19db4d8aca88 100644
--- a/fs/bcachefs/btree_gc.c
+++ b/fs/bcachefs/btree_gc.c
@@ -348,7 +348,7 @@ static int bch2_btree_repair_topology_recurse(struct btree_trans *trans, struct
 		bch2_bkey_val_to_text(&buf, c, bkey_i_to_s_c(cur_k.k));
 
 		if (mustfix_fsck_err_on(bch2_err_matches(ret, EIO),
-				trans, btree_node_unreadable,
+				trans, btree_node_read_error,
 				"Topology repair: unreadable btree node at\n"
 				"  %s",
 				buf.buf)) {
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index e361057ffad4..64bb330eac86 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -40,19 +40,42 @@ int bch2_btree_lost_data(struct bch_fs *c, enum btree_id btree)
 	int ret = 0;
 
 	mutex_lock(&c->sb_lock);
+	struct bch_sb_field_ext *ext = bch2_sb_field_get(c->disk_sb.sb, ext);
 
 	if (!(c->sb.btrees_lost_data & b)) {
 		struct printbuf buf = PRINTBUF;
 		bch2_btree_id_to_text(&buf, btree);
 		bch_err(c, "flagging btree %s lost data", buf.buf);
 		printbuf_exit(&buf);
-		bch2_sb_field_get(c->disk_sb.sb, ext)->btrees_lost_data |= cpu_to_le64(b);
+		ext->btrees_lost_data |= cpu_to_le64(b);
 	}
 
+	/* Once we have runtime self healing for topology errors we won't need this: */
+	ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_topology) ?: ret;
+
+	/* Btree node accounting will be off: */
+	__set_bit_le64(BCH_FSCK_ERR_accounting_mismatch, ext->errors_silent);
+	ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_allocations) ?: ret;
+
+#ifdef CONFIG_BCACHEFS_DEBUG
+	/*
+	 * These are much more minor, and don't need to be corrected right away,
+	 * but in debug mode we want the next fsck run to be clean:
+	 */
+	ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_lrus) ?: ret;
+	ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_backpointers_to_extents) ?: ret;
+#endif
+
 	switch (btree) {
 	case BTREE_ID_alloc:
-		ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_allocations) ?: ret;
 		ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_alloc_info) ?: ret;
+
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_data_type_wrong, ext->errors_silent);
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_gen_wrong, ext->errors_silent);
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_dirty_sectors_wrong, ext->errors_silent);
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_cached_sectors_wrong, ext->errors_silent);
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_wrong, ext->errors_silent);
+		__set_bit_le64(BCH_FSCK_ERR_alloc_key_stripe_redundancy_wrong, ext->errors_silent);
 		goto out;
 	case BTREE_ID_backpointers:
 		ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_btree_backpointers) ?: ret;
@@ -75,7 +98,6 @@ int bch2_btree_lost_data(struct bch_fs *c, enum btree_id btree)
 		goto out;
 	default:
 		ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_scan_for_btree_nodes) ?: ret;
-		ret = bch2_run_explicit_recovery_pass_persistent_locked(c, BCH_RECOVERY_PASS_check_topology) ?: ret;
 		goto out;
 	}
 out:
@@ -748,9 +770,6 @@ int bch2_fs_recovery(struct bch_fs *c)
 		bch2_write_super(c);
 	mutex_unlock(&c->sb_lock);
 
-	if (c->opts.fsck && IS_ENABLED(CONFIG_BCACHEFS_DEBUG))
-		c->opts.recovery_passes |= BIT_ULL(BCH_RECOVERY_PASS_check_topology);
-
 	if (c->opts.fsck)
 		set_bit(BCH_FS_fsck_running, &c->flags);
 	if (c->sb.clean)
diff --git a/fs/bcachefs/sb-errors_format.h b/fs/bcachefs/sb-errors_format.h
index 89d9dc2c859b..917ef6aa4a23 100644
--- a/fs/bcachefs/sb-errors_format.h
+++ b/fs/bcachefs/sb-errors_format.h
@@ -72,12 +72,12 @@ enum bch_fsck_flags {
 	x(btree_root_read_error,				 59,	FSCK_AUTOFIX)	\
 	x(btree_root_bad_min_key,				 60,	0)		\
 	x(btree_root_bad_max_key,				 61,	0)		\
-	x(btree_node_read_error,				 62,	0)		\
-	x(btree_node_topology_bad_min_key,			 63,	0)		\
-	x(btree_node_topology_bad_max_key,			 64,	0)		\
-	x(btree_node_topology_overwritten_by_prev_node,		 65,	0)		\
-	x(btree_node_topology_overwritten_by_next_node,		 66,	0)		\
-	x(btree_node_topology_interior_node_empty,		 67,	0)		\
+	x(btree_node_read_error,				 62,	FSCK_AUTOFIX)	\
+	x(btree_node_topology_bad_min_key,			 63,	FSCK_AUTOFIX)	\
+	x(btree_node_topology_bad_max_key,			 64,	FSCK_AUTOFIX)	\
+	x(btree_node_topology_overwritten_by_prev_node,		 65,	FSCK_AUTOFIX)	\
+	x(btree_node_topology_overwritten_by_next_node,		 66,	FSCK_AUTOFIX)	\
+	x(btree_node_topology_interior_node_empty,		 67,	FSCK_AUTOFIX)	\
 	x(fs_usage_hidden_wrong,				 68,	FSCK_AUTOFIX)	\
 	x(fs_usage_btree_wrong,					 69,	FSCK_AUTOFIX)	\
 	x(fs_usage_data_wrong,					 70,	FSCK_AUTOFIX)	\
-- 
2.45.2


  parent reply	other threads:[~2024-11-29 20:28 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-11-29 20:26 [PATCH 00/34] a whole raft of bugfixes Kent Overstreet
2024-11-29 20:27 ` [PATCH 01/34] bcachefs: BCH_ERR_btree_node_read_error_cached Kent Overstreet
2024-11-29 20:27 ` [PATCH 02/34] bcachefs: Use separate rhltable for bch2_inode_or_descendents_is_open() Kent Overstreet
2024-11-29 20:27 ` [PATCH 03/34] bcachefs: errcode cleanup: journal errors Kent Overstreet
2024-11-29 20:27 ` [PATCH 04/34] bcachefs: disk_accounting: bch2_dev_rcu -> bch2_dev_rcu_noerror Kent Overstreet
2024-11-29 20:27 ` [PATCH 05/34] bcachefs: Fix accounting_read when we rewind Kent Overstreet
2024-11-29 20:27 ` [PATCH 06/34] bcachefs: backpointer_to_missing_ptr is now autofix Kent Overstreet
2024-11-29 20:27 ` [PATCH 07/34] bcachefs: Fix btree node scan when unknown btree IDs are present Kent Overstreet
2024-11-29 20:27 ` [PATCH 08/34] bcachefs: Kill bch2_bucket_alloc_new_fs() Kent Overstreet
2024-11-29 20:27 ` [PATCH 09/34] bcachefs: Bad btree roots are now autofix Kent Overstreet
2024-11-29 20:27 ` [PATCH 10/34] bcachefs: Fix dup/misordered check in btree node read Kent Overstreet
2024-11-29 20:27 ` [PATCH 11/34] bcachefs: Don't try to en/decrypt when encryption not available Kent Overstreet
2024-11-29 20:27 ` [PATCH 12/34] bcachefs: Change "disk accounting version 0" check to commit only Kent Overstreet
2024-11-29 20:27 ` [PATCH 13/34] bcachefs: Fix bch2_btree_node_update_key_early() Kent Overstreet
2024-11-29 20:27 ` [PATCH 14/34] bcachefs: Go RW earlier, for normal rw mount Kent Overstreet
2024-11-29 20:27 ` [PATCH 15/34] bcachefs: Fix null ptr deref in btree_path_lock_root() Kent Overstreet
2024-11-29 20:27 ` [PATCH 16/34] bcachefs: Ignore empty btree root journal entries Kent Overstreet
2024-11-29 20:27 ` [PATCH 17/34] bcachefs: struct bkey_validate_context Kent Overstreet
2024-11-29 20:27 ` Kent Overstreet [this message]
2024-11-29 20:27 ` [PATCH 19/34] bcachefs: BCH_FS_recovery_running Kent Overstreet
2024-11-29 20:27 ` [PATCH 20/34] bcachefs: dio write: Take ref on mm_struct when using asynchronously Kent Overstreet
2024-11-29 22:38   ` Jens Axboe
2024-12-05  1:55   ` Kent Overstreet
2024-11-29 20:27 ` [PATCH 21/34] bcachefs: Guard against journal seq overflow Kent Overstreet
2024-11-29 20:27 ` [PATCH 22/34] bcachefs: Issue a transaction restart after commit in repair Kent Overstreet
2024-11-29 20:27 ` [PATCH 23/34] bcachefs: Guard against backpointers to unknown btrees Kent Overstreet
2024-11-29 20:27 ` [PATCH 24/34] bcachefs: Fix journal_iter list corruption Kent Overstreet
2024-11-29 20:27 ` [PATCH 25/34] bcachefs: add missing printbuf_reset() Kent Overstreet
2024-11-29 20:27 ` [PATCH 26/34] bcachefs: mark more errors AUTOFIX Kent Overstreet
2024-11-29 20:27 ` [PATCH 27/34] bcachefs: Don't error out when logging fsck error Kent Overstreet
2024-11-29 20:27 ` [PATCH 28/34] bcachefs: do_fsck_ask_yn() Kent Overstreet
2024-11-29 20:27 ` [PATCH 29/34] bcachefs: Check for bucket journal seq in the future Kent Overstreet
2024-11-29 20:27 ` [PATCH 30/34] bcachefs: Check for inode " Kent Overstreet
2024-11-29 20:27 ` [PATCH 31/34] bcachefs: cryptographic MACs on superblock are not (yet?) supported Kent Overstreet
2024-11-29 20:27 ` [PATCH 32/34] bcachefs: bch2_trans_relock() is trylock for lockdep Kent Overstreet
2024-11-29 20:27 ` [PATCH 33/34] bcachefs: Check for extent crc uncompressed/compressed size mismatch Kent Overstreet
2024-11-29 20:27 ` [PATCH 34/34] bcachefs: Don't recurse in check_discard_freespace_key Kent Overstreet

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20241129202736.2713679-19-kent.overstreet@linux.dev \
    --to=kent.overstreet@linux.dev \
    --cc=linux-bcachefs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox