Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20211011030956.2459172-1-mudongliangabcd@gmail.com>
In-Reply-To: <20211011030956.2459172-1-mudongliangabcd@gmail.com>
From:   Dongliang Mu <mudongliangabcd@gmail.com>
Date:   Thu, 20 Jan 2022 17:28:23 +0800
Message-ID: <CAD-N9QXHwzt9Dui3i6sFF-4d-8Z41LoOJydPykdQQh_jWh+_SA@mail.gmail.com>
Subject: Re: [PATCH] fs: fix GPF in nilfs_mdt_destroy
To:     Alexander Viro <viro@zeniv.linux.org.uk>
Cc:     linux-fsdevel@vger.kernel.org,
        linux-kernel <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Oct 11, 2021 at 11:10 AM Dongliang Mu <mudongliangabcd@gmail.com> wrote:
>
> In alloc_inode, inode_init_always could return -NOMEM if
> security_inode_alloc fails. In its error handling, i_callback and
> nilfs_free_inode will be called. However, because inode->i_private is
> not initialized due to the failure of security_inode_alloc, the function
> nilfs_is_metadata_file_inode can return true and nilfs_mdt_destroy will
> be executed to lead to GPF bug.
>
> Fix this bug by moving the assignment of inode->i_private before
> security_inode_alloc.
>

ping?

> BTW, this bug is triggered by fault injection in the syzkaller.
>
> Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com>
> ---
>  fs/inode.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/inode.c b/fs/inode.c
> index ed0cab8a32db..f6fce84bf550 100644
> --- a/fs/inode.c
> +++ b/fs/inode.c
> @@ -160,6 +160,7 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
>         inode->i_dir_seq = 0;
>         inode->i_rdev = 0;
>         inode->dirtied_when = 0;
> +       inode->i_private = NULL;
>
>  #ifdef CONFIG_CGROUP_WRITEBACK
>         inode->i_wb_frn_winner = 0;
> @@ -194,7 +195,6 @@ int inode_init_always(struct super_block *sb, struct inode *inode)
>         lockdep_set_class_and_name(&mapping->invalidate_lock,
>                                    &sb->s_type->invalidate_lock_key,
>                                    "mapping.invalidate_lock");
> -       inode->i_private = NULL;
>         inode->i_mapping = mapping;
>         INIT_HLIST_HEAD(&inode->i_dentry);      /* buggered by rcu freeing */
>  #ifdef CONFIG_FS_POSIX_ACL
> --
> 2.25.1
>