Re: [PATCH v2] audit: fix potential integer overflow in audit_log_n_hex()

Paul Moore <[email protected]> Tue, 30 Jun 2026 15:46:50 -0400
Newsgroups org.kernel.vger.audit,org.kernel.vger.linux-kernel
Message-ID <[email protected]>
On Jun  1, 2026 Ricardo Robaina <[email protected]> wrote:
> 
> The function calculates new_len as len << 1 for hex encoding. This
> has two overflow risks: the shift itself can overflow when len is
> large, and the result can be truncated when assigned to new_len
> (declared as int) from the size_t calculation.
> 
> Fix by using check_shl_overflow() to catch shift overflow and
> changing new_len and loop counter i to size_t to prevent truncation.
> 
> Fixes: 168b7173959f ("AUDIT: Clean up logging of untrusted strings")
> Signed-off-by: Ricardo Robaina <[email protected]>
> Reviewed-by: Richard Guy Briggs <[email protected]>
> ---
> Changes in v2:
> - Use check_shl_overflow() instead of manual overflow check.
> 
>  kernel/audit.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/audit.c b/kernel/audit.c
> index e1d489bc2dff..8ca268610641 100644
> --- a/kernel/audit.c
> +++ b/kernel/audit.c
> @@ -62,6 +62,7 @@
>  #include <net/ip.h>
>  #include <net/ipv6.h>
>  #include <linux/sctp.h>
> +#include <linux/overflow.h>
>  
>  #include "audit.h"
>  
> @@ -2076,7 +2077,8 @@ void audit_log_format(struct audit_buffer *ab, const char *fmt, ...)
>  void audit_log_n_hex(struct audit_buffer *ab, const unsigned char *buf,
>  		size_t len)
>  {
> -	int i, avail, new_len;
> +	int avail;
> +	size_t i, new_len;
>  	unsigned char *ptr;
>  	struct sk_buff *skb;
>  
> @@ -2084,9 +2086,13 @@ void audit_log_n_hex(struct audit_buffer *ab, const unsigned char *buf,
>  		return;
>  
>  	BUG_ON(!ab->skb);
> +
>  	skb = ab->skb;
>  	avail = skb_tailroom(skb);
> -	new_len = len<<1;
> +
> +	if (check_shl_overflow(len, 1, &new_len))
> +		return;

By returning without logging a value I worry we could end up with an
oddly formatted audit record, e.g. "... A=foo B= C=bar ...".  Instead of
simply returning, should we log a '?' for the value and then return?

>  	if (new_len >= avail) {
>  		/* Round the buffer request up to the next multiple */
>  		new_len = AUDIT_BUFSIZ*(((new_len-avail)/AUDIT_BUFSIZ) + 1);
> -- 
> 2.53.0

--
paul-moore.com