Breno Leitao wrote: > Hello Willem, > > On Fri, Jun 12, 2026 at 10:58:19AM -0400, Willem de Bruijn wrote: > > Breno Leitao wrote: > > > In preparation for converting the proto-layer getsockopt callbacks to the > > > sockopt_t interface, switch udp_lib_getsockopt() to take a sockopt_t. > > > > What is the benefit of this conversion? > > This enables kernel callers (io_uring, bpf, etc.) to invoke getsockopt > directly. The setsockopt() conversion previously used sockptr, but > Linus objected to that approach and suggested iov_iter instead. > > For full context, see: > https://lore.kernel.org/all/[email protected]/ > > > It does add a lot more complexity and makes the code less obvious. > > I agree this adds complexity. Unfortunately, I don't see a simpler way > to enable getsockopt to work with non-__user addresses.
Oh right, this is already applied for other protocols such as l2tp_ppp rds and smc.
