Re: [PATCH net v3] e100: prevent shift-out-of-bounds in EEPROM access

On 8/16/2026 4:37 AM, [email protected] wrote:
> From: Yalagada Pavan Kumar <[email protected]>
> 
> When reading the EEPROM address length, e100_eeprom_read() can return
> an invalid length. This value is then used as a shift count when
> calculating eeprom_wc, resulting in a shift-out-of-bounds UBSAN warning.
> 
> Stop EEPROM address probing if the detected address length exceeds
> the initial address length to prevent addr_len from underflowing.
> 
> Validate addr_len after reading it from the EEPROM in both
> e100_eeprom_load() and e100_eeprom_save(), and return -EIO
> if the value is zero or greater than 8.
> 
> Reported-by: [email protected]
> Closes: https://syzkaller.appspot.com/bug?extid=e0abb1d45ac291ebebeb
> Signed-off-by: Yalagada Pavan Kumar <[email protected]>
> ---
> v3:
>   - Prevent addr_len underflow during EEPROM probing.
>   - Use BIT() and explicitly cast the result to u16 when assigning eeprom_wc.
>   - Update the commit message and subject to reflect the scope of the fix.
>   - verify the UBSAN failure with syzkaller reproducer before the fix
>     and its absence after the fix.
> 
> v2: https://lore.kernel.org/all/[email protected]/
>   - Return -EIO instead of -EINVAL for an invalid EEPROM address length.
>   - Validate addr_len in e100_eeprom_save() as well.
>   - Drop the unnecessary 1U change in the shift.
>   - Update the commit message.
> 
> v1: https://lore.kernel.org/all/[email protected]/
> ---
>   drivers/net/ethernet/intel/e100.c | 16 +++++++++-------
>   1 file changed, 9 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/e100.c b/drivers/net/ethernet/intel/e100.c
> index 464dc2d6cdb5..5165f1d3c98f 100644
> --- a/drivers/net/ethernet/intel/e100.c
> +++ b/drivers/net/ethernet/intel/e100.c
> @@ -746,10 +746,12 @@ static __le16 e100_eeprom_read(struct nic *nic, u16 *addr_len, u16 addr)
>   		if (!(ctrl & eedo) && i > 16) {
>   			u16 len = i - 16;
>   
> -			if (len > *addr_len)
> +			if (len > *addr_len) {
>   				*addr_len = 0;
> -			else
> -				*addr_len -= len;
> +				break;
> +			}
> +
> +			*addr_len -= len;
>   			i = 17;

It seems each version you are sending is change on top of the previous 
version. Please send all the changes in a single patch so that it can be 
reviewed, and applied, standalone.

Also, Andrew had requested on the other patch that this go via *-next 
since the issue is a theoretical one. Can you target this to iwl-next?

Thanks,
Tony

>   		}
>