gentoo-overlay/sys-kernel/hardened-kernel/files/linux-6.12/0119-bcachefs-Fix-accountin...

48 lines
1.6 KiB
Diff

From ba91f39cd42bcfac86b9a7165d56b6964c314c81 Mon Sep 17 00:00:00 2001
From: Kent Overstreet <kent.overstreet@linux.dev>
Date: Sun, 24 Nov 2024 22:28:41 -0500
Subject: [PATCH 119/233] bcachefs: Fix accounting_read when we rewind
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit
If we rewind recovery to run topology repair, that causes
accounting_read to run twice.
This fixes accounting being double counted.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Alexander Miroshnichenko <alex@millerson.name>
---
fs/bcachefs/disk_accounting.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/fs/bcachefs/disk_accounting.c b/fs/bcachefs/disk_accounting.c
index fa821d278c45..bb5dbbf71d04 100644
--- a/fs/bcachefs/disk_accounting.c
+++ b/fs/bcachefs/disk_accounting.c
@@ -700,6 +700,21 @@ int bch2_accounting_read(struct bch_fs *c)
struct btree_trans *trans = bch2_trans_get(c);
struct printbuf buf = PRINTBUF;
+ /*
+ * We might run more than once if we rewind to start topology repair or
+ * btree node scan - and those might cause us to get different results,
+ * so we can't just skip if we've already run.
+ *
+ * Instead, zero out any accounting we have:
+ */
+ percpu_down_write(&c->mark_lock);
+ darray_for_each(acc->k, e)
+ percpu_memset(e->v[0], 0, sizeof(u64) * e->nr_counters);
+ for_each_member_device(c, ca)
+ percpu_memset(ca->usage, 0, sizeof(*ca->usage));
+ percpu_memset(c->usage, 0, sizeof(*c->usage));
+ percpu_up_write(&c->mark_lock);
+
int ret = for_each_btree_key(trans, iter,
BTREE_ID_accounting, POS_MIN,
BTREE_ITER_prefetch|BTREE_ITER_all_snapshots, k, ({
--
2.45.2