From: Kent Overstreet <kent.overstreet@linux.dev>
Date: Mon, 10 Jul 2023 21:23:59 +0000 (-0400)
Subject: bcachefs: Don't start copygc until recovery is finished
X-Git-Tag: io_uring-6.7-2023-11-10~119^2~204
X-Git-Url: https://git.kernel.dk/?a=commitdiff_plain;h=ea28c86722954a58e3bab24eec3e7624e8d4956b;p=linux-block.git

bcachefs: Don't start copygc until recovery is finished

With "bcachefs: Snapshot depth, skiplist fields", we now can't run data
move operations until after bch2_check_snapshots() is complete.

Ideally we'd have the copygc (and rebalance) threads wait until
c->curr_recovery_pass has advanced, but the waitlist handling is tricky
- so for now, move starting copygc back to read_write_late().

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

diff --git a/fs/bcachefs/super.c b/fs/bcachefs/super.c
index 6ab98c2299dd..63e9dafa8395 100644
--- a/fs/bcachefs/super.c
+++ b/fs/bcachefs/super.c
@@ -344,6 +344,19 @@ static int bch2_fs_read_write_late(struct bch_fs *c)
 {
 	int ret;
 
+	/*
+	 * Data move operations can't run until after check_snapshots has
+	 * completed, and bch2_snapshot_is_ancestor() is available.
+	 *
+	 * Ideally we'd start copygc/rebalance earlier instead of waiting for
+	 * all of recovery/fsck to complete:
+	 */
+	ret = bch2_copygc_start(c);
+	if (ret) {
+		bch_err(c, "error starting copygc thread");
+		return ret;
+	}
+
 	ret = bch2_rebalance_start(c);
 	if (ret) {
 		bch_err(c, "error starting rebalance thread");
@@ -403,12 +416,6 @@ static int __bch2_fs_read_write(struct bch_fs *c, bool early)
 		return ret;
 	}
 
-	ret = bch2_copygc_start(c);
-	if (ret) {
-		bch_err(c, "error starting copygc thread");
-		return ret;
-	}
-
 	if (!early) {
 		ret = bch2_fs_read_write_late(c);
 		if (ret)