From: Kent Overstreet <kent.overstreet@linux.dev>
Date: Wed, 4 Dec 2024 23:16:25 +0000 (-0500)
Subject: bcachefs: Fix failure to allocate journal write on discard retry
X-Git-Tag: io_uring-6.14-20240131~27^2~101
X-Git-Url: https://git.kernel.dk/?a=commitdiff_plain;h=49833ce27ed2eed91915a4c25690d82aae5b6a0b;p=linux-block.git

bcachefs: Fix failure to allocate journal write on discard retry

When allocating a journal write fails, then retries after doing
discards, we were failing to count already allocated replicas.

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

diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index e5fce5e497f2..d7dfea5f0181 100644
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -1498,6 +1498,15 @@ static int journal_write_alloc(struct journal *j, struct journal_buf *w)
 				       READ_ONCE(c->opts.metadata_replicas_required));
 
 	rcu_read_lock();
+
+	/* We might run more than once if we have to stop and do discards: */
+	struct bkey_ptrs_c ptrs = bch2_bkey_ptrs_c(bkey_i_to_s_c(&w->key));
+	bkey_for_each_ptr(ptrs, p) {
+		struct bch_dev *ca = bch2_dev_rcu_noerror(c, p->dev);
+		if (ca)
+			replicas += ca->mi.durability;
+	}
+
 retry:
 	devs = target_rw_devs(c, BCH_DATA_journal, target);