In article <ah70jy88njct6...@internode.on.net>, Brett Lymn <bl...@internode.on.net> wrote: >On Tue, Jul 22, 2025 at 04:12:13AM +0300, Valery Ushakov wrote: >> On Mon, Jul 21, 2025 at 23:13:59 +0200, Thomas Klausner wrote: >> >> > Does gdb work on this for anyone else? >> > Or can someone please fix gdb? :-) >> >> gdb has been in a sad state for years, especially for threaded >> programs (amd64 is in a better shape than most). it's not rocket >> science, but it's just a lot of slog and nobody had time or energy to >> step up. >> > >Also, the gdb developers have been very sloppy with curses calls and >happily pass NULL pointers into libcurses. They get away with this when >linking against ncurses because that has checks for NULL. I went >through and added checks for NULL into our curses in -current too which >I don't believe that we should need to do but there is no way to avoid it >because our curses will be blamed (and has been) for what is actually >improper use of the interface - the checks do add overhead, not really >noticable but, still, it wastes cycles because of sloppy coding on the >caller's side. > >Anyway, is the crash of gdb in a curses call? If so, please let me know >and I will fix that.
I've fixed the gdb tui code to not abuse curses. Can we have the NULL checking code in curses conditionally compiled it? christos
