On Sun, 6 Sep 2026 20:40:59 +0000
"Jose Fernandez (Anthropic)" <[email protected]> wrote:

> Will add one in v2, thanks.
> 
> Today arm64 returns 0 only for a mis-aligned patch site. But patch 4 of
> this series replaces that check with two: the two NOPs to patch must be
> there, and the ops pointer must fit in the prefix NOPs (which also
> catches a mis-aligned site). How about this to cover the series?
> 
>       /* The architecture may return 0 to reject a patch site */

That comment is fine. I didn't really like the specifying of mis-alignment
as it did appear to be too specific, and could change.

-- Steve


> 
> Happy to name the mis-aligned case in patch 1 and extend the comment
> in patch 4 if you prefer that.


Reply via email to