linux-kernel - [PATCH] btrfs: push memalloc_nofs_save/restore() out of alloc

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20251002115427.98773-2-rtapadia730@gmail.com>
Date: Thu,  2 Oct 2025 17:24:28 +0530
From: rtapadia730@...il.com
To: dsterba@...e.com,
	clm@...com,
	fdmanana@...e.com
Cc: linux-btrfs@...r.kernel.org,
	linux-kernel@...r.kernel.org,
	skhan@...uxfoundation.org,
	khalid@...nel.org,
	david.hunter.linux@...il.com,
	Rajeev Tapadia <rtapadia730@...il.com>
Subject: [PATCH] btrfs: push memalloc_nofs_save/restore() out of alloc_bitmap()

From: Rajeev Tapadia <rtapadia730@...il.com>

alloc_bitmap() currently wraps its allocation in memalloc_nofs_save
/restore(), but this hides allocation context from callers. GFP_NOFS is
required to avoid recursion into the filesystem during transaction commits,
but the correct place to enforce that is at the call sites where we know
recursion is unsafe.

So now alloc_bitmap() just allocates a bitmap. Also completing the TODO
comment.

Signed-off-by: Rajeev Tapadia <rtapadia730@...il.com>
---

The patch was tested by enabling CONFIG_BTRFS_FS_RUN_SANITY_TESTS
All tests passed while booting the kernel in qemu.

 fs/btrfs/free-space-tree.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/fs/btrfs/free-space-tree.c b/fs/btrfs/free-space-tree.c
index dad0b492a663..abdbdc74edf8 100644
--- a/fs/btrfs/free-space-tree.c
+++ b/fs/btrfs/free-space-tree.c
@@ -159,8 +159,6 @@ static inline u32 free_space_bitmap_size(const struct btrfs_fs_info *fs_info,
 
 static unsigned long *alloc_bitmap(u32 bitmap_size)
 {
-	unsigned long *ret;
-	unsigned int nofs_flag;
 	u32 bitmap_rounded_size = round_up(bitmap_size, sizeof(unsigned long));
 
 	/*
@@ -168,13 +166,11 @@ static unsigned long *alloc_bitmap(u32 bitmap_size)
 	 * into the filesystem as the free space bitmap can be modified in the
 	 * critical section of a transaction commit.
 	 *
-	 * TODO: push the memalloc_nofs_{save,restore}() to the caller where we
-	 * know that recursion is unsafe.
+	 * This function's caller is responsible for setting the appropriate
+	 * allocation context (e.g., using memalloc_nofs_save/restore())
+	 * to prevent recursion.
 	 */
-	nofs_flag = memalloc_nofs_save();
-	ret = kvzalloc(bitmap_rounded_size, GFP_KERNEL);
-	memalloc_nofs_restore(nofs_flag);
-	return ret;
+	return kvzalloc(bitmap_rounded_size, GFP_KERNEL);
 }
 
 static void le_bitmap_set(unsigned long *map, unsigned int start, int len)
@@ -217,7 +213,9 @@ int btrfs_convert_free_space_to_bitmaps(struct btrfs_trans_handle *trans,
 	int ret;
 
 	bitmap_size = free_space_bitmap_size(fs_info, block_group->length);
+	unsigned int nofs_flag = memalloc_nofs_save();
 	bitmap = alloc_bitmap(bitmap_size);
+	memalloc_nofs_restore(nofs_flag);
 	if (unlikely(!bitmap)) {
 		ret = -ENOMEM;
 		btrfs_abort_transaction(trans, ret);
@@ -360,7 +358,9 @@ int btrfs_convert_free_space_to_extents(struct btrfs_trans_handle *trans,
 	int ret;
 
 	bitmap_size = free_space_bitmap_size(fs_info, block_group->length);
+	unsigned int nofs_flag = memalloc_nofs_save();
 	bitmap = alloc_bitmap(bitmap_size);
+	memalloc_nofs_restore(nofs_flag);
 	if (unlikely(!bitmap)) {
 		ret = -ENOMEM;
 		btrfs_abort_transaction(trans, ret);
-- 
2.51.0