Is it MSCDEX, maybe? MS-DOS 6.22 ships with a more recent MSCDEX.
MS-DOS 5 doesn't, at all. Maybe that's why the setup program has copied over its own version?
It's just an idea. Also, here's a CD-ROM driver that has an DOS 5/6 style installer: Re: Get CD-ROM running under MS-DOS

I think its an CHIPSET or CABLE or JUMPERS issue (MASTER/SLAVE/CABLE SELECT)
Not all old systems compatible with HDD and CD on one cable (usual for 286/386 systems but who knows).
Give us more details of connections and jumpers settings on CD and HDD

Babasha wrote on 2024-12-23, 20:59:

I think its an CHIPSET or CABLE or JUMPERS issue (MASTER/SLAVE/CABLE SELECT)
Not all old systems compatible with HDD and CD on one cable (usual for 286/386 systems but who knows).
Give us more details of connections and jumpers settings on CD and HDD

It cannot be… I can make it work flawlessly on DOS 5.0 (and Windows 3.1) with Atapicd.sys. So it’s nothing hardware related…

Jo22 wrote on 2024-12-23, 20:45:

Is it MSCDEX, maybe? MS-DOS 6.22 ships with a more recent MSCDEX.
MS-DOS 5 doesn't, at all. Maybe that's why the setup program has copied over its own version?
It's just an idea. Also, here's a CD-ROM driver that has an DOS 5/6 style installer: Re: Get CD-ROM running under MS-DOS

I forgot to mention that under dos 5.0 , I use the mscdex from Windows 3.1.

But for the Windows 95 setup, I use the one provided in the ‘command’ folder of Windows. Is there a more recent version? Maybe I can try the one from Windows 3.1…

I should have mentioned it’s really the Windows 95 installation I want fix… For my DOS installation, I prefer using msdos 5.0 over msdos 6.22 for this machine.

OK! If u so sure - no questions at all.
Or try connect it thru external secondary IDE card or soundcard with additional IDE channel.

My experience with 486/early Pentijm motherboards - 50% of them comes with some IDE or ATAPI "miracles" 😉

Making the drive a master on its own controller may be worth a shot.

Anonymous Coward wrote on 2024-12-23, 23:46:

Making the drive a master on its own controller may be worth a shot.

That's the thing... I only have one channel on this motherboard.

I do have a SB16 that has IDE... But setting up the all the required drivers in DOS eats up so much conventional RAM that it was absurd (35-40k) ! On top of that, it was weirdly fragile and the drive would drop off after 15 minutes of operation...

What did you name the cdrom ? Do not use /D:MSCD000, try like /D:MSCD001 (have seen "some" Win9x that do not work with it all zeros, not sure why).
Also some HD's are poor at mastering certain cdroms as slave, has to do with their specific built in drive controllers.....

Linoleum wrote on 2024-12-24, 01:45:

Anonymous Coward wrote on 2024-12-23, 23:46:

Making the drive a master on its own controller may be worth a shot.

That's the thing... I only have one channel on this motherboard.

I do have a SB16 that has IDE... But setting up the all the required drivers in DOS eats up so much conventional RAM that it was absurd (35-40k) ! On top of that, it was weirdly fragile and the drive would drop off after 15 minutes of operation...

I don't know if I've ever used a Sound Blaster card with an IDE connector, but I wasn't under the impression that it required special drivers to activate it. Don't you basically just have to plug in the drive and use the oakcdrom driver that comes with Windows 9x? I don't really understand why it would matter if the IDE connector is on the motherboard, soundcard or extra controller. The procedure should be the same. The only issue I could see is if perhaps the soundcard is configuring the port as tertiary or quaternary, but the driver is only able to see drives on primary or secondary controllers.

I did some more tests and it's even weirder than I thought... I've used DOS 6.22 to do my tests as a proxy for Windows 95 (they both seem to behave similarly, but DOS is easier and quicker).

I took everything that could come into play from my working DOS 5.0 setup (CEMM.exe, HIMEM.exe, MSCDEX) to gradually replace those memory and cdrom managers from DOS 6.22 to see which one would prevent me from using another CD-ROM driver than ATAPICD.sys. I ended up running everything I had from my DOS 5 setup into DOS 6.22 and I was still unable to use VIDE-CDD.sys or MTMCDAI.sys. So it's COMMAND.com or something like that?!?

So I went back to my MS-DOS 5.0 and see if my drivers were disclosing some kind of address that I could "force" on my MS-DOS 6.22 setup... And it does! Setup on slave of Primary Channel is indeed port 1F0 and IRQ 14! (see attached pic)

So I tried forcing these parameters to VIDE-CDD and MTMCDAI with /P:1F0,14 and none of them worked on MS-DOS 6.22!! Urgh!! I am going mad here!

Anonymous Coward wrote on 2024-12-24, 02:33:

I don't know if I've ever used a Sound Blaster card with an IDE connector, but I wasn't under the impression that it required special drivers to activate it. Don't you basically just have to plug in the drive and use the oakcdrom driver that comes with Windows 9x? I don't really understand why it would matter if the IDE connector is on the motherboard, soundcard or extra controller. The procedure should be the same. The only issue I could see is if perhaps the soundcard is configuring the port as tertiary or quaternary, but the driver is only able to see drives on primary or secondary controllers.

The issue would be if the sound card IDE port isn't configured statically (with jumpers or EEPROM) but sits in inhibited at power on until woken up, whether that's through ISA Plug and Play, or a proprietary initialization sequence through something stuck in config.sys or autoexec.bat

If you have any ISA slots to spare, you can just get some run of the mill ISA multi I/O card and use that. Most of the ones I've seen allow you to set them as secondary.

jakethompson1 wrote on 2024-12-24, 02:51:

Anonymous Coward wrote on 2024-12-24, 02:33:

I don't know if I've ever used a Sound Blaster card with an IDE connector, but I wasn't under the impression that it required special drivers to activate it. Don't you basically just have to plug in the drive and use the oakcdrom driver that comes with Windows 9x? I don't really understand why it would matter if the IDE connector is on the motherboard, soundcard or extra controller. The procedure should be the same. The only issue I could see is if perhaps the soundcard is configuring the port as tertiary or quaternary, but the driver is only able to see drives on primary or secondary controllers.

The issue would be if the sound card IDE port isn't configured statically (with jumpers or EEPROM) but sits in inhibited at power on until woken up, whether that's through ISA Plug and Play, or a proprietary initialization sequence through something stuck in config.sys or autoexec.bat

Now that explanation makes sense... I have a CT2940 PNP SB16. The best way I was I able to make it work was to first load CTCM.exe as a device in config.sys, then load SBIDE.sys; both that an insane amount of RAM. I was hoping to avoid that...

Why are MSDOS 6 and Win95 killing my primary slave port!?!

Linoleum wrote on 2024-12-24, 03:04:

jakethompson1 wrote on 2024-12-24, 02:51:

Anonymous Coward wrote on 2024-12-24, 02:33:

I don't know if I've ever used a Sound Blaster card with an IDE connector, but I wasn't under the impression that it required special drivers to activate it. Don't you basically just have to plug in the drive and use the oakcdrom driver that comes with Windows 9x? I don't really understand why it would matter if the IDE connector is on the motherboard, soundcard or extra controller. The procedure should be the same. The only issue I could see is if perhaps the soundcard is configuring the port as tertiary or quaternary, but the driver is only able to see drives on primary or secondary controllers.

The issue would be if the sound card IDE port isn't configured statically (with jumpers or EEPROM) but sits in inhibited at power on until woken up, whether that's through ISA Plug and Play, or a proprietary initialization sequence through something stuck in config.sys or autoexec.bat

Now that explanation makes sense... I have a CT2940 PNP SB16. The best way I was I able to make it work was to first load CTCM.exe as a device in config.sys, then load SBIDE.sys; both that an insane amount of RAM. I was hoping to avoid that...

Why are MSDOS 6 and Win95 killing my primary slave port!?!

There is a port of Linux pnptools to DOS. It is not a TSR so it does not consume any memory once it runs and configures your devices. Since it doesn't run in config.sys, you would then have to use a devload-like tool to load your ATAPI CD driver. Beware it uses a text config file and is not easy to use. And you would never want to use it in conjunction with Windows 95.

It does not make sense that MS-DOS 5 vs. 6 makes a difference. They don't mess with hardware at this level, deferring that to the BIOS.

jakethompson1 wrote on 2024-12-24, 03:28:

It does not make sense that MS-DOS 5 vs. 6 makes a difference. They don't mess with hardware at this level, deferring that to the BIOS.

I know... That's what's killing me! Although, I am using COMPAQ's DOS 5.0... Have they tweaked something in their version compare to the plain old MS-DOS 5.0 that makes my CDROM magically work on Primary Slave? I might give real MS-DOS 5.0 for my curiosity...

jakethompson1 wrote on 2024-12-24, 03:28:

It does not make sense that MS-DOS 5 vs. 6 makes a difference. They don't mess with hardware at this level, deferring that to the BIOS.

Agree ! If it were me I would make a simple DOS 6 boot floppy (just himem.sys and cdrom driver in config, just mscdex in auto..bat). Remove HD and set CDROM as master and boot from floppy.
If that does not work there must be something odd about this Compaq's bios as the 6x cd is well within it's time frame.....

I wonder if a disk cache difference, or if none is loaded, something about the disk buffering config in the two versions, could exercise some hardware fault when disk activity is switching between the C: drive and the cd-rom in rapid-fire fashion. The IRQ 14s have to be properly hooked by both the bios int13h handler and the cd-rom driver. I assume the cd-rom driver takes over that vector, and keeps track of whether a request is active and passes it back to the HDD hander if not.

Horun wrote on 2024-12-24, 03:54:

jakethompson1 wrote on 2024-12-24, 03:28:

It does not make sense that MS-DOS 5 vs. 6 makes a difference. They don't mess with hardware at this level, deferring that to the BIOS.

Agree ! If it were me I would make a simple DOS 6 boot floppy (just himem.sys and cdrom driver in config, just mscdex in auto..bat). Remove HD and set CDROM as master and boot from floppy.
If that does not work there must be something odd about this Compaq's bios as the 6x cd is well within it's time frame.....

Yup I will! I’ll do this test on top of the real msdos 5.0 just in case…

Keep ideas coming and I’ll report back in few days (after Xmas).