From: Filipe Manana <fdmanana@suse.com>
Date: Tue, 20 Jul 2021 16:05:23 +0100
Git-commit: 5a656c3628b241443fd07cda60f3b0587bb8c328
Patch-mainline: v5.15-rc1
References: bsc#1197915
Subject: [PATCH] btrfs: stop doing GFP_KERNEL memory allocations in the ref
verify tool
In commit 351cbf6e4410e7 ("btrfs: use nofs allocations for running delayed
items") we wrapped all btree updates when running delayed items with
memalloc_nofs_save() and memalloc_nofs_restore(), due to a lock inversion
detected by lockdep involving reclaim and the mutex of delayed nodes.
The problem is because the ref verify tool does some memory allocations
with GFP_KERNEL, which can trigger reclaim and reclaim can trigger inode
eviction, which requires locking the mutex of an inode's delayed node.
On the other hand the ref verify tool is called when allocating metadata
extents as part of operations that modify a btree, which is a problem when
running delayed nodes, where we do btree updates while holding the mutex
of a delayed node. This is what caused the lockdep warning.
Instead of wrapping every btree update when running delayed nodes, change
the ref verify tool to never do GFP_KERNEL allocations, because:
1) We get less repeated code, which at the moment does not even have a
comment mentioning why we need to setup the NOFS context, which is a
recommended good practice as mentioned at
Documentation/core-api/gfp_mask-from-fs-io.rst
2) The ref verify tool is something meant only for debugging and not
something that should be enabled on non-debug / non-development
kernels;
3) We may have yet more places outside delayed-inode.c where we have
similar problem: doing btree updates while holding some lock and
then having the GFP_KERNEL memory allocations, from the ref verify
tool, trigger reclaim and trying again to acquire the same lock
through the reclaim path.
Or we could get more such cases in the future, therefore this change
prevents getting into similar cases when using the ref verify tool.
Curiously most of the memory allocations done by the ref verify tool
were already using GFP_NOFS, except a few ones for no apparent reason.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
---
fs/btrfs/delayed-inode.c | 12 ------------
fs/btrfs/ref-verify.c | 10 +++++-----
2 files changed, 5 insertions(+), 17 deletions(-)
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 20f3e748027e..61452f04181a 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -6,7 +6,6 @@
#include <linux/slab.h>
#include <linux/iversion.h>
-#include <linux/sched/mm.h>
#include "misc.h"
#include "delayed-inode.h"
#include "disk-io.h"
@@ -686,7 +685,6 @@ static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans,
const int max_size = BTRFS_LEAF_DATA_SIZE(root->fs_info);
int total_size;
int nitems;
- unsigned int nofs_flag;
char *ins_data = NULL;
struct btrfs_key *ins_keys;
u32 *ins_sizes;
@@ -735,10 +733,8 @@ static int btrfs_insert_delayed_item(struct btrfs_trans_handle *trans,
}
}
- nofs_flag = memalloc_nofs_save();
ret = btrfs_insert_empty_items(trans, root, path, ins_keys, ins_sizes,
nitems);
- memalloc_nofs_restore(nofs_flag);
if (ret)
goto out;
@@ -860,7 +856,6 @@ static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,
struct btrfs_delayed_node *node)
{
struct btrfs_delayed_item *curr, *prev;
- unsigned int nofs_flag;
int ret = 0;
do_again:
@@ -869,9 +864,7 @@ static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,
if (!curr)
goto delete_fail;
- nofs_flag = memalloc_nofs_save();
ret = btrfs_search_slot(trans, root, &curr->key, path, -1, 1);
- memalloc_nofs_restore(nofs_flag);
if (ret < 0)
goto delete_fail;
else if (ret > 0) {
@@ -940,7 +933,6 @@ static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
struct btrfs_key key;
struct btrfs_inode_item *inode_item;
struct extent_buffer *leaf;
- unsigned int nofs_flag;
int mod;
int ret;
@@ -953,9 +945,7 @@ static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
else
mod = 1;
- nofs_flag = memalloc_nofs_save();
ret = btrfs_lookup_inode(trans, root, path, &key, mod);
- memalloc_nofs_restore(nofs_flag);
if (ret > 0)
ret = -ENOENT;
if (ret < 0)
@@ -1012,9 +1002,7 @@ static int __btrfs_update_delayed_inode(struct btrfs_trans_handle *trans,
key.type = BTRFS_INODE_EXTREF_KEY;
key.offset = -1;
- nofs_flag = memalloc_nofs_save();
ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
- memalloc_nofs_restore(nofs_flag);
if (ret < 0)
goto err_out;
ASSERT(ret);
diff --git a/fs/btrfs/ref-verify.c b/fs/btrfs/ref-verify.c
index 8e026de74c44..d2062d5f71dd 100644
--- a/fs/btrfs/ref-verify.c
+++ b/fs/btrfs/ref-verify.c
@@ -264,8 +264,8 @@ static struct block_entry *add_block_entry(struct btrfs_fs_info *fs_info,
struct block_entry *be = NULL, *exist;
struct root_entry *re = NULL;
- re = kzalloc(sizeof(struct root_entry), GFP_KERNEL);
- be = kzalloc(sizeof(struct block_entry), GFP_KERNEL);
+ re = kzalloc(sizeof(struct root_entry), GFP_NOFS);
+ be = kzalloc(sizeof(struct block_entry), GFP_NOFS);
if (!be || !re) {
kfree(re);
kfree(be);
@@ -313,7 +313,7 @@ static int add_tree_block(struct btrfs_fs_info *fs_info, u64 ref_root,
struct root_entry *re;
struct ref_entry *ref = NULL, *exist;
- ref = kmalloc(sizeof(struct ref_entry), GFP_KERNEL);
+ ref = kmalloc(sizeof(struct ref_entry), GFP_NOFS);
if (!ref)
return -ENOMEM;
@@ -358,7 +358,7 @@ static int add_shared_data_ref(struct btrfs_fs_info *fs_info,
struct block_entry *be;
struct ref_entry *ref;
- ref = kzalloc(sizeof(struct ref_entry), GFP_KERNEL);
+ ref = kzalloc(sizeof(struct ref_entry), GFP_NOFS);
if (!ref)
return -ENOMEM;
be = add_block_entry(fs_info, bytenr, num_bytes, 0);
@@ -393,7 +393,7 @@ static int add_extent_data_ref(struct btrfs_fs_info *fs_info,
u64 offset = btrfs_extent_data_ref_offset(leaf, dref);
u32 num_refs = btrfs_extent_data_ref_count(leaf, dref);
- ref = kzalloc(sizeof(struct ref_entry), GFP_KERNEL);
+ ref = kzalloc(sizeof(struct ref_entry), GFP_NOFS);
if (!ref)
return -ENOMEM;
be = add_block_entry(fs_info, bytenr, num_bytes, ref_root);
--
2.26.2