Re: [PATCH] libcpp: Optimize AArch64 search_line_fast

On Fri, 14 Aug 2026, Wilco Dijkstra wrote:

> Yes that's why I read from limit to get a full vector and then just shift out any
> chars before 's' - this magic removes 0-15 nibbles from the result mask based
> on the alignment of 's':
> 
> +  mask >>= (uintptr_t)s * 4;

Let me also note that this is undefined behavior due to large shift amount,
would need proper masking of the RHS (otherwise would trigger UBSan).

Alexander