bcachefs: Add a feature bit for indicating fs has no alloc info WIP

If a filesystem is going to only be used read-only, and will be a
deployable image, we can strip out alloc info for a substantial
reduction in metadata size - around half, due to backpointers.

Alloc info will be regenerated on first read-write mount.

Remounting RW is disallowed for now, since we don't yet have
check_allocations running in RW mode.

XXX make scrub work (no repair)
XXX look at mount wall clock time/profile

XXX we also really don't want to be pinning btree roots in memory
anymore, since a lot of them aren't used

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

6 files changed, 47 insertions, 15 deletions

diff --git a/fs/bcachefs/bcachefs_format.h b/fs/bcachefs/bcachefs_format.h
index 8a1366aabebd..792dc8c3d54e 100644
--- a/fs/bcachefs/bcachefs_format.h
+++ b/fs/bcachefs/bcachefs_format.h
@@ -917,7 +917,8 @@ static inline void SET_BCH_SB_BACKGROUND_COMPRESSION_TYPE(struct bch_sb *sb, __u
 	x(alloc_v2,			17)	\
 	x(extents_across_btree_nodes,	18)	\
 	x(incompat_version_field,	19)	\
-	x(casefolding,			20)
+	x(casefolding,			20)	\
+	x(no_alloc_info,		21)
 
 #define BCH_SB_FEATURES_ALWAYS				\
 	(BIT_ULL(BCH_FEATURE_new_extent_overwrite)|	\
diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
index c4eb0ed9838d..197d49a478b9 100644
--- a/fs/bcachefs/errcode.h
+++ b/fs/bcachefs/errcode.h
@@ -221,6 +221,7 @@
 	x(EROFS,			erofs_unfixed_errors)			\
 	x(EROFS,			erofs_norecovery)			\
 	x(EROFS,			erofs_nochanges)			\
+	x(EROFS,			erofs_no_alloc_info)			\
 	x(EROFS,			insufficient_devices)			\
 	x(0,				operation_blocked)			\
 	x(BCH_ERR_operation_blocked,	btree_cache_cannibalize_lock_blocked)	\
diff --git a/fs/bcachefs/recovery.c b/fs/bcachefs/recovery.c
index d2b07f602da9..abcc888630c5 100644
--- a/fs/bcachefs/recovery.c
+++ b/fs/bcachefs/recovery.c
@@ -32,7 +32,6 @@
 #include <linux/sort.h>
 #include <linux/stat.h>
 
-
 int bch2_btree_lost_data(struct bch_fs *c, enum btree_id btree)
 {
 	u64 b = BIT_ULL(btree);
@@ -115,9 +114,6 @@ static void kill_btree(struct bch_fs *c, enum btree_id btree)
 /* for -o reconstruct_alloc: */
 static void bch2_reconstruct_alloc(struct bch_fs *c)
 {
-	bch2_journal_log_msg(c, "dropping alloc info");
-	bch_info(c, "dropping and reconstructing all alloc info");
-
 	mutex_lock(&c->sb_lock);
 	struct bch_sb_field_ext *ext = bch2_sb_field_get(c->disk_sb.sb, ext);
 
@@ -159,6 +155,8 @@ static void bch2_reconstruct_alloc(struct bch_fs *c)
 
 	c->opts.recovery_passes |= bch2_recovery_passes_from_stable(le64_to_cpu(ext->recovery_passes_required[0]));
 
+	c->disk_sb.sb->features[0] &= ~cpu_to_le64(BIT_ULL(BCH_FEATURE_no_alloc_info));
+
 	bch2_write_super(c);
 	mutex_unlock(&c->sb_lock);
 
@@ -888,8 +886,26 @@ use_clean:
 	if (ret)
 		goto err;
 
-	if (c->opts.reconstruct_alloc)
+	if (!c->opts.read_only &&
+	    (c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info))) {
+		bch_info(c, "mounting a filesystem with no alloc info read-write; will recreate");
+
 		bch2_reconstruct_alloc(c);
+	} else if (c->opts.reconstruct_alloc) {
+		bch2_journal_log_msg(c, "dropping alloc info");
+		bch_info(c, "dropping and reconstructing all alloc info");
+
+		bch2_reconstruct_alloc(c);
+	}
+
+	if (c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info)) {
+		/* We can't go RW to fix errors without alloc info */
+		if (c->opts.fix_errors == FSCK_FIX_yes ||
+		    c->opts.fix_errors == FSCK_FIX_ask)
+			c->opts.fix_errors = FSCK_FIX_no;
+		if (c->opts.errors == BCH_ON_ERROR_fix_safe)
+			c->opts.errors = BCH_ON_ERROR_continue;
+	}
 
 	/*
 	 * After an unclean shutdown, skip then next few journal sequence
diff --git a/fs/bcachefs/recovery_passes.c b/fs/bcachefs/recovery_passes.c
index 593ff142530d..e16d1ea2c2c5 100644
--- a/fs/bcachefs/recovery_passes.c
+++ b/fs/bcachefs/recovery_passes.c
@@ -46,7 +46,11 @@ static int bch2_set_may_go_rw(struct bch_fs *c)
 
 	set_bit(BCH_FS_may_go_rw, &c->flags);
 
-	if (keys->nr || !c->opts.read_only || c->opts.fsck || !c->sb.clean || c->opts.recovery_passes)
+	if (keys->nr ||
+	    !c->opts.read_only ||
+	    !c->sb.clean ||
+	    c->opts.recovery_passes ||
+	    (c->opts.fsck && !(c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info))))
 		return bch2_fs_read_write_early(c);
 	return 0;
 }
@@ -202,6 +206,8 @@ static bool should_run_recovery_pass(struct bch_fs *c, enum bch_recovery_pass pa
 {
 	struct recovery_pass_fn *p = recovery_pass_fns + pass;
 
+	if ((p->when & PASS_ALLOC) && (c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info)))
+		return false;
 	if (c->opts.recovery_passes_exclude & BIT_ULL(pass))
 		return false;
 	if (c->opts.recovery_passes & BIT_ULL(pass))
diff --git a/fs/bcachefs/recovery_passes_types.h b/fs/bcachefs/recovery_passes_types.h
index e89b9c783285..d8e166623189 100644
--- a/fs/bcachefs/recovery_passes_types.h
+++ b/fs/bcachefs/recovery_passes_types.h
@@ -7,6 +7,8 @@
 #define PASS_UNCLEAN		BIT(2)
 #define PASS_ALWAYS		BIT(3)
 #define PASS_ONLINE		BIT(4)
+#define PASS_ALLOC		BIT(5)
+#define PASS_FSCK_ALLOC		(PASS_FSCK|PASS_ALLOC)
 
 #ifdef CONFIG_BCACHEFS_DEBUG
 #define PASS_FSCK_DEBUG		BIT(1)
@@ -27,17 +29,17 @@
 	x(stripes_read,				 1, 0)					\
 	x(initialize_subvolumes,		 2, 0)					\
 	x(snapshots_read,			 3, PASS_ALWAYS)			\
-	x(check_allocations,			 5, PASS_FSCK)				\
-	x(trans_mark_dev_sbs,			 6, PASS_ALWAYS|PASS_SILENT)		\
-	x(fs_journal_alloc,			 7, PASS_ALWAYS|PASS_SILENT)		\
+	x(check_allocations,			 5, PASS_FSCK_ALLOC)			\
+	x(trans_mark_dev_sbs,			 6, PASS_ALWAYS|PASS_SILENT|PASS_ALLOC)	\
+	x(fs_journal_alloc,			 7, PASS_ALWAYS|PASS_SILENT|PASS_ALLOC)	\
 	x(set_may_go_rw,			 8, PASS_ALWAYS|PASS_SILENT)		\
 	x(journal_replay,			 9, PASS_ALWAYS)			\
-	x(check_alloc_info,			10, PASS_ONLINE|PASS_FSCK)		\
-	x(check_lrus,				11, PASS_ONLINE|PASS_FSCK)		\
-	x(check_btree_backpointers,		12, PASS_ONLINE|PASS_FSCK)		\
+	x(check_alloc_info,			10, PASS_ONLINE|PASS_FSCK_ALLOC)	\
+	x(check_lrus,				11, PASS_ONLINE|PASS_FSCK_ALLOC)	\
+	x(check_btree_backpointers,		12, PASS_ONLINE|PASS_FSCK_ALLOC)	\
 	x(check_backpointers_to_extents,	13, PASS_ONLINE|PASS_FSCK_DEBUG)	\
-	x(check_extents_to_backpointers,	14, PASS_ONLINE|PASS_FSCK)		\
-	x(check_alloc_to_lru_refs,		15, PASS_ONLINE|PASS_FSCK)		\
+	x(check_extents_to_backpointers,	14, PASS_ONLINE|PASS_FSCK_ALLOC)	\
+	x(check_alloc_to_lru_refs,		15, PASS_ONLINE|PASS_FSCK_ALLOC)	\
 	x(fs_freespace_init,			16, PASS_ALWAYS|PASS_SILENT)		\
 	x(bucket_gens_init,			17, 0)					\
 	x(reconstruct_snapshots,		38, 0)					\
diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
index f0c6fe0247a2..80cadecd37e9 100644
--- a/fs/bcachefs/super.c
+++ b/fs/bcachefs/super.c
@@ -543,11 +543,17 @@ int bch2_fs_read_write(struct bch_fs *c)
 	if (c->opts.nochanges)
 		return -BCH_ERR_erofs_nochanges;
 
+	if (c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info))
+		return -BCH_ERR_erofs_no_alloc_info;
+
 	return __bch2_fs_read_write(c, false);
 }
 
 int bch2_fs_read_write_early(struct bch_fs *c)
 {
+	if (WARN_ON(c->sb.features & BIT_ULL(BCH_FEATURE_no_alloc_info)))
+		return -BCH_ERR_erofs_no_alloc_info;
+
 	down_write(&c->state_lock);
 	int ret = __bch2_fs_read_write(c, true);
 	up_write(&c->state_lock);