linux-kernel - Re: [PATCH] bcachefs: return the error code instead of the return value of IS_ERR_OR

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] [day] [month] [year] [list]

Message-ID: <ywccsi7uqqnsfha4yvg6lycsbafeqwylwabwvkjprjrv66q4sb@kxuz3ygc5qp7>
Date: Wed, 25 Sep 2024 11:20:09 -0400
From: Kent Overstreet <kent.overstreet@...ux.dev>
To: Edward Adam Davis <eadavis@...com>
Cc: syzbot+c0360e8367d6d8d04a66@...kaller.appspotmail.com, 
	brauner@...nel.org, jack@...e.cz, jfs-discussion@...ts.sourceforge.net, 
	linux-bcachefs@...r.kernel.org, linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org, 
	shaggy@...nel.org, syzkaller-bugs@...glegroups.com, viro@...iv.linux.org.uk
Subject: Re: [PATCH] bcachefs: return the error code instead of the return
 value of IS_ERR_OR_NULL

On Wed, Sep 25, 2024 at 09:53:00PM GMT, Edward Adam Davis wrote:
> Syzbot report a kernel BUG in vfs_get_tree.
> The root cause is that read_btree_nodes() returned 1 and returned -EINTR
> due to kthread_run() execution failure.
> 
> The -EINTR needs to be returnned to bch2_fs_recovery(), not return to
> "ret = IS_ERR_OR_NULL(t)".
> 
> Reported-and-tested-by: syzbot+c0360e8367d6d8d04a66@...kaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=c0360e8367d6d8d04a66
> Signed-off-by: Edward Adam Davis <eadavis@...com>
> ---
>  fs/bcachefs/btree_node_scan.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c
> index b28c649c6838..df7090ca1e81 100644
> --- a/fs/bcachefs/btree_node_scan.c
> +++ b/fs/bcachefs/btree_node_scan.c
> @@ -281,6 +281,10 @@ static int read_btree_nodes(struct find_btree_nodes *f)
>  			closure_put(&cl);
>  			f->ret = ret;
>  			bch_err(c, "error starting kthread: %i", ret);
> +			if (IS_ERR(t)) {
> +				closure_sync(&cl);
> +				return PTR_ERR(t);
> +			}
>  			break;
>  		}
>  	}
> -- 
> 2.43.0
> 

I fixed this last night with the patch below...

commit c1a6f5ca052b7f8609917d13cd11fc60c94396aa
Author: Kent Overstreet <kent.overstreet@...ux.dev>
Date:   Tue Sep 24 19:31:22 2024 -0400

    bcachefs: Fix incorrect IS_ERR_OR_NULL usage
    
    Returning a positive integer instead of an error code causes error paths
    to become very confused.
    
    Closes: syzbot+c0360e8367d6d8d04a66@...kaller.appspotmail.com
    Signed-off-by: Kent Overstreet <kent.overstreet@...ux.dev>

diff --git a/fs/bcachefs/btree_node_scan.c b/fs/bcachefs/btree_node_scan.c
index b28c649c6838..1e694fedc5da 100644
--- a/fs/bcachefs/btree_node_scan.c
+++ b/fs/bcachefs/btree_node_scan.c
@@ -275,7 +275,7 @@ static int read_btree_nodes(struct find_btree_nodes *f)
 		w->ca		= ca;
 
 		t = kthread_run(read_btree_nodes_worker, w, "read_btree_nodes/%s", ca->name);
-		ret = IS_ERR_OR_NULL(t);
+		ret = PTR_ERR_OR_ZERO(t);
 		if (ret) {
 			percpu_ref_put(&ca->io_ref);
 			closure_put(&cl);