Re: [Intel-wired-lan] [PATCH] e100: prevent shift out-of-bounds in e100_eeprom_load

On 08/07 20:26, [email protected] wrote:
> From: Yalagada Pavan Kumar <[email protected]>
> 
> When reading the EEPROM address length, e100_eeprom_read() can return
> an invalid length (0 or >= 16). Passing an invalid addr_len to bit-shift
> operations causes a shift out-of-bounds, triggering a kernel panic or
> UBSAN warning.
> 
> Validate addr_len after reading it from EEPROM and return -EINVAL if
> the value is out of bounds. Additionally, use 1U to prevent
> signed integer overflow when calculating eeprom_wc.
> 
> Reported-by: [email protected]
> Tested-by: [email protected]
> Closes: https://syzkaller.appspot.com/bug?extid=e0abb1d45ac291ebebeb
> Signed-off-by: Yalagada Pavan Kumar <[email protected]>
> ---
> Tested using syzbot c reproducer.
> ---
>  drivers/net/ethernet/intel/e100.c | 17 +++++++++++++++--
>  1 file changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/e100.c b/drivers/net/ethernet/intel/e100.c
> index 29960762e64a..1de5cd41ea0c 100644
> --- a/drivers/net/ethernet/intel/e100.c
> +++ b/drivers/net/ethernet/intel/e100.c
> @@ -744,7 +744,12 @@ static __le16 e100_eeprom_read(struct nic *nic, u16 *addr_len, u16 addr)
>  		 * complete address.  Use this to adjust addr_len. */
>  		ctrl = ioread8(&nic->csr->eeprom_ctrl_lo);
>  		if (!(ctrl & eedo) && i > 16) {
> -			*addr_len -= (i - 16);
> +			u16 len = i - 16;
> +
> +			if (len > *addr_len)
> +				*addr_len = 0;
> +			else
> +				*addr_len -= len;
>  			i = 17;
>  		}
>  
> @@ -765,7 +770,15 @@ static int e100_eeprom_load(struct nic *nic)
>  
>  	/* Try reading with an 8-bit addr len to discover actual addr len */
>  	e100_eeprom_read(nic, &addr_len, 0);
> -	nic->eeprom_wc = 1 << addr_len;
> +
> +	if (!addr_len || addr_len >= 16) {
> +		netif_err(nic, probe, nic->netdev,
> +			"Invalid EEPROM address length %u\n",
> +			addr_len);
> +		return -EINVAL;

An invalid address length here appears to indicate an unexpected
response from the EEPROM, rather than an invalid argument. Therefore,
-EINVAL seems somewhat misleading. Would -EIO be more appropriate here?

Also, e100_eeprom_save() seems to have the same pattern. Shouldn't we
fix that as well?

> +	}
> +
> +	nic->eeprom_wc = 1U << addr_len;

The commit message says that the U suffix prevents signed integer
overflow. However, after the validation addr_len is in [1, 15], so 1 <<
addr_len cannot overflow an int. I think what prevents the out-of-bounds
shift is the validation, not the U.

>  
>  	for (addr = 0; addr < nic->eeprom_wc; addr++) {
>  		nic->eeprom[addr] = e100_eeprom_read(nic, &addr_len, addr);
> -- 
> 2.43.0
>