Re: [PATCH] crypto : ecc - Fix carry overflow in vli multiplication

[Lukas Wunner <lukas@wunner.de>]

On Fri, May 08, 2026 at 02:48:44PM +0300, Anastasia Tishchenko wrote:
> The carry flag calculation fails when r01.m_high is saturated
> (0xFFFFFFFFFFFFFFFF) and addition of lower bits overflows.
> 
> The condition (r01.m_high < product.m_high) doesn't handle the case
> where r01.m_high == product.m_high and an additional carry exists
> from lower-bit overflow.
> 
> Add proper handling for this boundary by accounting for the carry
> from the lower addition.
[...]
> +++ b/crypto/ecc.c
> @@ -427,7 +427,10 @@ static void vli_mult(u64 *result, const u64 *left, const u64 *right,
>  			product = mul_64_64(left[i], right[k - i]);
>  
>  			r01 = add_128_128(r01, product);
> -			r2 += (r01.m_high < product.m_high);
> +			if (r01.m_high != product.m_high)
> +				r2 += (r01.m_high < product.m_high);
> +			else
> +				r2 += (r01.m_low < product.m_low);
>  		}
>  
>  		result[k] = r01.m_low;

ICYMI, sashiko's AI-generated review alleges that the if-else condition
may cause a timing side channel vis-à-vis binary arithmetic:

https://sashiko.dev/#/patchset/20260508114844.29694-1-sv3iry%40gmail.com

You may want to address this if/when respinning your patch.  If you do,
a code comment is probably merited to explain this subtlety.

Thanks,

Lukas