On Sat, Oct 10, 2020 at 01:33:42PM +0000, nia wrote: > syzkaller found a bug in tty_43.c which i noticed could lock up my > system after i wrote a reproducer for it. > > i was curious why 43bsd compat is enabled by default on amd64 then > noticed it gets loaded by tty if it detects an unrecognized ioctl > > https://github.com/NetBSD/src/blob/trunk/sys/kern/tty.c#L1420 > > this seems to be a strange choice, especially on archs that never ran > 43bsd binaries. were applications relying on these ioctls more recently? > > i'm not familiar with the history of this code, but maybe someone here is
it looks like maxv disabled compat43 in amd64 GENERIC in 2017 https://freshbsd.org/commit/netbsd/src/460aa6eb5b4d8484a9a626a429cbbc6be167d8dca7be2361e93ce118eb4d86ab is its autoloading by tty an oversight?
