Re: [PATCH] mm: fix dereferencing possible ERR_PTR

[Ammar Faizi <ammarfaizi2@xxxxxxxxxxx>]

On 9/4/22 2:46 PM, Binyi Han wrote:
> Smatch checker complains that 'secretmem_mnt' dereferencing possible
> ERR_PTR().
> Let the function return if 'secretmem_mnt' is ERR_PTR, to avoid
> deferencing it.
>
> Signed-off-by: Binyi Han <dantengknight@xxxxxxxxx>
> ---

Fixes: 1507f51255c9ff07d75909a84e7c0d7f3c4b2f49 ("mm: introduce memfd_secret system call to create "secret" memory areas")

>   mm/secretmem.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/secretmem.c b/mm/secretmem.c
> index e3e9590c6fb3..3f7154099795 100644
> --- a/mm/secretmem.c
> +++ b/mm/secretmem.c
> @@ -285,7 +285,7 @@ static int secretmem_init(void)
>   
>   	secretmem_mnt = kern_mount(&secretmem_fs);
>   	if (IS_ERR(secretmem_mnt))
> -		ret = PTR_ERR(secretmem_mnt);
> +		return PTR_ERR(secretmem_mnt);
>   
>   	/* prevent secretmem mappings from ever getting PROT_EXEC */
>   	secretmem_mnt->mnt_flags |= MNT_NOEXEC;

I agree that doing:

   secretmem_mnt->mnt_flags |= MNT_NOEXEC;

when IS_ERR(secretmem_mnt) evaluates to true is wrong. But I have
a question: what happen if you invoke memfd_secret() syscall when
@secretmem_mnt is an ERR_PTR?

Shouldn't we also guard the memfd_secret() path?

diff --git a/mm/secretmem.c b/mm/secretmem.c
index e3e9590c6fb3..2d52508d47a9 100644
--- a/mm/secretmem.c
+++ b/mm/secretmem.c
@@ -230,18 +230,21 @@ static struct file *secretmem_file_create(unsigned long flags)
 
 SYSCALL_DEFINE1(memfd_secret, unsigned int, flags)
 {
 	struct file *file;
 	int fd, err;
 
 	/* make sure local flags do not confict with global fcntl.h */
 	BUILD_BUG_ON(SECRETMEM_FLAGS_MASK & O_CLOEXEC);
 
+	if (IS_ERR(secretmem_mnt))
+		return PTR_ERR(secretmem_mnt);
+
 	if (!secretmem_enable)
 		return -ENOSYS;
 
 	if (flags & ~(SECRETMEM_FLAGS_MASK | O_CLOEXEC))
 		return -EINVAL;
 	if (atomic_read(&secretmem_users) < 0)
 		return -ENFILE;
 
 	fd = get_unused_fd_flags(flags & O_CLOEXEC);


--
Ammar Faizi