Thanks for testing this Ken. I wasn't even aware of the "mystery opcodes" prior to your email. I'll add that snipet of code to the next release.
Does TS-DOS even use them? I'm not sure I've seen them when trapping for requests I haven't accounted for. I'll have to take another look at my logs. Kurt On Thu, Apr 2, 2020, at 3:19 PM, Ken Pettit wrote: > Hey Guys (and especially Kurt), > > I tested NEWDOS using VirtualT by connected to mComm on Ubuntu via two > pseudo terminals created with 'socat'. I had to make a couple of minor > changes to mComm to get it to run with /dev/pts terminals, but no big deal. > > I ran into the same issue with NEWDOS running against mComm where it > would lock up and not return. I discovered that mComm is not responding > to the TS-DOS "mystery opcodes". It is unclear why vanilla TS-DOS has > no issue with this ... perhaps I accidentally removed a timeout > somewhere waiting for response to these opcodes. > > When I modified mComm to respond to the "mystery opcodes", NEWDOS sprung > to life and displayed the file list of the DISK. > > The "mystery opcodes" that TS-DOS / NEWDOS sends are: > > "ZZ1" 04h 01h 00h 83h FFh 46h > "ZZ1" 04h 01h 00h 96h 0Fh 24h > "ZZ1" 04h 01h 00h 94h 0Fh 26h > > I supported this by adding this code to tpdd.py: > > elif TPDD_Command[2] == 0x31: # Mystery opcode (Send a > valid response) > #print("Mystery opcode") > self.LogMessage("Mystery Opcode", True) > TPDD_Response = bytearray([0x12, 0x01, 0x00, 0x00]) # 0x00 > Operation complete > TPDD_Response[3] = self.CalcChecksum(TPDD_Response) > > > I will try to figure out why NEWDOS doesn't timeout if the mystery > opcodes have no response. It is unclear what a true TPDD does with > these opcodes, or if they can simply be removed from NEWDOS with no > side-effects. > > Ken > > On 4/1/20 1:15 PM, Joshua O'Keefe wrote: > >> On Apr 1, 2020, at 10:13 AM, Ken Pettit <petti...@gmail.com> wrote: > >> Hi Joshua, > >> > >> Okay, this is good information. Then I need to figure out what is > >> different between NEWDOS autobaud and regular TS-DOS autobaud. > > To add some further twists to this: if I have DOS-ON and issue a LFILES > > command, I get a hang, with no visible I/O. If I open and close COM:98n1d > > and then issue LFILES, I get some visible RX/TX...and then a hang. > > > > All of the above done with delay off. > > > > Also, just loading a page of DISK (F4) seems a little slower than vanilla, > > as though that outer loop you mentioned is still adding some overhead even > > in non delay mode. It's subtle but noticeable. > > > > And just by way of comparison, yes, vanilla TSDOS (and SARDOS, which I like > > because it's the other rom besides NEWDOS with the shift-arrow fix) > > exhibits none of this. > > > > Let me know if there's more I can report to help! > >