Re: [PATCH] e1000: fix OOB in e1000_tbi_should_accept()

[Tony Nguyen <anthony.l.nguyen@intel.com>]

On 11/23/2025 7:45 PM, Guangshuo Li wrote:
> Hi Tony, all,
> 
> Thanks for the review. As suggested by Tony, I’ll keep the declarations 
> at the top and place the bounds checks before assigning last_byte. I’ll 
> send a v2 with the following change:
> 
> static bool e1000_tbi_should_accept(struct e1000_adapter *adapter,
>                                      u8 status, u8 errors,
>                                      u32 length, const u8 *data)
> {
>      struct e1000_hw *hw = &adapter->hw;
>      u8 last_byte;
> 
>      /* Guard against OOB on data[length - 1] */
>      if (unlikely(!length))
>          return false;
> 
>      /* Upper bound: length must not exceed rx_buffer_len */
>      if (unlikely(length > adapter->rx_buffer_len))
>          return false;
> 
>      last_byte = data[length - 1];
> 
>      /* existing logic follows ... */
> }
> Please let me know if further adjustments are preferred.

This looks along the lines of what I was expecting.

Thanks,
Tony

> Best regards,
> Guangshuo Li