linux-kernel - Re: [PATCH] hfsplus: don't use BUG_ON() in hfsplus_create_attributes

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

Message-ID: <244c8da9-4c5e-42ed-99c7-ceee3e039a9c@I-love.SAKURA.ne.jp>
Date: Fri, 11 Jul 2025 20:35:41 +0900
From: Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>
To: Viacheslav Dubeyko <Slava.Dubeyko@....com>,
        "frank.li@...o.com" <frank.li@...o.com>,
        "glaubitz@...sik.fu-berlin.de" <glaubitz@...sik.fu-berlin.de>,
        "slava@...eyko.com" <slava@...eyko.com>,
        "brauner@...nel.org" <brauner@...nel.org>,
        "akpm@...ux-foundation.org" <akpm@...ux-foundation.org>
Cc: "linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] hfsplus: don't use BUG_ON() in
 hfsplus_create_attributes_file()

On 2025/07/10 7:03, Tetsuo Handa wrote:
> On 2025/07/10 3:33, Viacheslav Dubeyko wrote:
>> My worry that we could have a race condition here. Let's imagine that two
>> threads are trying to call __hfsplus_setxattr() and both will try to create the
>> Attributes File. Potentially, we could end in situation when inode could have
>> not zero size during hfsplus_create_attributes_file() in one thread because
>> another thread in the middle of Attributes File creation. Could we double check
>> that we don't have the race condition here? Otherwise, we need to make much
>> cleaner fix of this issue.
> 
> I think that there is some sort of race window, for
> https://elixir.bootlin.com/linux/v6.15.5/source/fs/hfsplus/xattr.c#L145
> explains that if more than one thread concurrently reached
> 
> 	if (!HFSPLUS_SB(inode->i_sb)->attr_tree) {
> 		err = hfsplus_create_attributes_file(inode->i_sb);
> 		if (unlikely(err))
> 			goto end_setxattr;
> 	}
> 
> path, all threads except one thread will fail with -EAGAIN.
> 

Do you prefer stricter mount-time validation shown below?
Is vhdr->attr_file.total_blocks == 0 when sbi->attr_tree exists and is empty?

diff --git a/fs/hfsplus/super.c b/fs/hfsplus/super.c
index 948b8aaee33e..f6324a0458f3 100644
--- a/fs/hfsplus/super.c
+++ b/fs/hfsplus/super.c
@@ -482,13 +482,17 @@ static int hfsplus_fill_super(struct super_block *sb, struct fs_context *fc)
 		goto out_close_ext_tree;
 	}
 	atomic_set(&sbi->attr_tree_state, HFSPLUS_EMPTY_ATTR_TREE);
-	if (vhdr->attr_file.total_blocks != 0) {
-		sbi->attr_tree = hfs_btree_open(sb, HFSPLUS_ATTR_CNID);
-		if (!sbi->attr_tree) {
-			pr_err("failed to load attributes file\n");
-			goto out_close_cat_tree;
+	sbi->attr_tree = hfs_btree_open(sb, HFSPLUS_ATTR_CNID);
+	if (sbi->attr_tree) {
+		if (vhdr->attr_file.total_blocks != 0) {
+			atomic_set(&sbi->attr_tree_state, HFSPLUS_VALID_ATTR_TREE);
+		} else {
+			pr_err("found attributes file despite total blocks is 0\n");
+			goto out_close_attr_tree;
 		}
-		atomic_set(&sbi->attr_tree_state, HFSPLUS_VALID_ATTR_TREE);
+	} else if (vhdr->attr_file.total_blocks != 0) {
+		pr_err("failed to load attributes file\n");
+		goto out_close_cat_tree;
 	}
 	sb->s_xattr = hfsplus_xattr_handlers;