Mark Kettenis <mark.kette...@xs4all.nl> wrote: > > Date: Sat, 21 May 2022 10:31:37 -0400 > > From: Johan Huldtgren <johan+openbsd-b...@huldtgren.com> > > > > hello, > > > > Details below, but commenting out 'ttyflags -a' from /etc/rc lets > > this host boot. I wrote much of this e-mail while going through it, > > so while we know now what the issue is I'm leaving my responses in > > case it sheds light on anything. > > So it seems your machine incorrectly advertises a serial port that > doesn't actually exist: > > > com1 at acpi0 UAR1 addr 0x2f8/0x8 irq 3: ti16750, 64 byte fifo > > com1: probed fifo depth: 0 bytes > > This may be a bug in our APCI code. Can you send the contents of > /var/db/acpi on your machine?
Once that is resolved, we need to decide how com.c behaves for either open or ioctl, in such a circumstance. Maybe the "probe" function can be more cynical, preventing an attach. Anyways, this locking-up behaviour today is inconvient.