On newer OpenBSD releases there is a bug in /usr/include/ctype.h,
it #defines EOF ignoring POSIX rules. As the schily make environmemnt
relies on EOF as an indicator to check whether stdio.h has been included,
there may be problems to compile things. If you find compile problems,
you may try to include 

#undef EOF 

directly after #include <ctype.h>

Cdrecord has been tested and known to work on OpenBSD, however,
the current OpenBSD port will not be able to scan the SCSI bus
for a CD-Recorder as it does on other operating systems.

In other words, 'cdrecord -scanbus' most likely will not work.

Both regular SCSI and ATAPI devices are supported. For ATAPI support
you need OpenBSD-2.6 or newer.

Libscg sends Generic scsi commands that are only supported for the
"whole disk" partition.  For example, 'cdrecord dev=/dev/rcd0c -checkdrive'.
See the cd(4), atapiscsi(4), and cdrecord(1) manpages for more details.

There are at least two ways to figure out what SCSI CD devices are
on the machine:

  From /var/run/dmesg.boot:
  $ egrep "^cd.+ at scsibus" /var/run/dmesg.boot
  cd0 at scsibus0 targ 0 lun 0: <SONY, CD-RW CRX175E2, S002> SCSI0 5/cdrom removable
  cd1 at scsibus1 targ 0 lun 0: <ATAPI, CD-ROM MARY500, 1.21> SCSI0 5/cdrom removable

  From sysctl:
  $ sysctl hw.disknames
  hw.disknames = wd0,cd0,cd1,fd0

In these examples, 'cd0' and 'cd1' are the SCSI CD devices on the
system, and they should be accessed by '/dev/rcd0c' and '/dev/rcd1c',
respectively.

It seems that the atapiscsi(4) driver has some bugs that may cause
cdrecord to hang for a while.  While this is annoying, it is harmless.
The command will eventually timeout (the default timeout is 40 seconds,
but can be changed with timeout=%d).  Trying to kill the "hung" process
may leave the device unusable until the system is rebooted (wich would
be another bug of the atapiscsi(4) driver as killing a process should
do all necessary housekeeping for the dead process), so just be patient.