On Fri, Nov 13, 2026 at 11:58:31AM +0800, Longjun Luo wrote:
I understand the concern about readability. Since English is not my
native language, I drafted the content myself and used AI to help with
the English writing before reviewing and sending it out. I'll keep the
information density in mind going forward.
Just to clarify the background: this design comes from my practical work
on userspace livepatch, and v1/v2 were meant to present the two
layers of that design as a whole. While dealing with LTO in practice, I
ran into many recurring correlation issues, so I tried to abstract and
model the problem systematically rather than hunting for corner cases,
and I brought that modeling process into this review.
As noted in the RFC, this patchset is intended to validate the
feasibility of the approach. I wanted feedback on the overall design
before deciding whether to take it further.
Thanks for taking the time to review and share your thoughts.
Thanks, that makes sense, but for upstreaming I think we will need to
take a more iterative approach, even if it ends up causing some
intermediate patches to be obsoleted by later ones.
--
Josh