On Dec 10, 2025, at 9:14 PM, Theo de Raadt <[email protected]> wrote: > >> I got curious and looked at what it would take to support both the old >> and new structure sizes, and the diff below appears to work fine. The >> issue is more of a problem on slower architectures where kernel and >> packages can stay unsynchronized for a longer time. Thoughts? > > I really don't believe in that 'support model'. The people using those > machines build their own packages (or they don't, but that's not our > problem).
If I remove the backward compat in the devel/gdb patches (the #ifdef PT_PTS_NAMELEN conditionals) then ports gdb won’t build unless base has the PT_PTS_NAMELEN define. This alone should ensure that ports gdb is built with a base that has the new define. The conditionals I put in the ports/gdb patches are only needed if we upstream the changes so they can be left out in the port. In fact, I think either way they should be removed from the port patches to ensure the p1 package is built with an updated base system. -Kurt
