Readme.txt for Periscope/32 for OS/2 2.0, Version 5.40

Tracing of interrupts in V86 mode can result in erratic behavior.
Evidently, OS/2's trap 0dh handler (which gets control on an
interrupt from V86 mode) doesn't work correctly when the single-step
flag is on.  We're pursuing resolution on this problem with IBM. For
now, you can't trace into an int 21h in v86 mode since OS/2 sets it
up as a trap at 0:0. Also, don't trace an interrupt that's not in the
IDT (0-1fh, 50h-57h, and 70h-77h) -- use Periscope's J or TI commands
to step over the interrupt.

The GA and GT commands do not work well with OS/2, because OS/2 tends
to generate spurious interrupt 2's sometimes after the execution of
an interrupt 1. Since these commands use lots of interrupt 1's,
you're likely to get a false stop in Periscope, since it interprets
an int 2 as a 'stop now' interrupt. Even when interrupt 2 is not
hooked by PSR2.SYS, there are still problems, likely related to the
trap 0dh problems mentioned above.

We cannot certify Periscope or Periscope/32 for use in the OS/2 DOS
box, due to significant incompatibilities with a true DOS. The
problems include:

- inconsistent execution -- what works one time will fail in another DOS 
  session
- no NMI or debug register support
- an NMI while in DOS freezes the system
- the QB command may not be usable, since OS/2 doesn't emulate some
  commands (such as LMSW, which is used in the COMPAQ BIOS)
- PS4TEST gets an error on a 486 when it reads CR0

If you're using a non-standard video driver, such as an XGA driver, 
be sure to put the 'device=' statements for Periscope after the video
driver or use the '/S' switch with PSR2.SYS to keep it from using any
video services.
