[BUG] NSSI hangs AMI WinBIOS 486 machine!

[ppgrainbow]

Is anyone trying to get Navratil System Software Information working on the AMI WinBIOS 486 compatible machine?

When NSSI tries to determine the co-processor type, the emulated machine locks up! The emulated machine runs on a Intel 80486DX2 running at 66 MHz w/16 MB RAM and a Trident TGUI9440 graphics card with 1 MB display RAM.

Here's a screenshot for proof:

NSSI lockup on AMI WinBIOS 486 machine

20170323_212505.png (11.39 KiB) Viewed 19474 times

I suspect that even if the CONFIG.SYS and AUTOEXEC.BAT files don't contain anything, the issue still persists and I have Dynarec disabled.

This issue shouldn't even occur on a real 486 or early Pentium-based PC. I suspect that the attempted co-processor detection might be causing an illegal instruction exception.

Is there a way to fix this?

[ppgrainbow]

UPDATE: If I select a 486SX or 486SX2 for the WinBIOS 486 and if I run NSSI, then it will run as usual by skipping co-processor detection:

NSSI_486SX2.png (13.31 KiB) Viewed 19439 times

There needs to be a way to make co-processor detection work properly with having Navratil Software System Information lock the emulated machine up!

[szadycbr]

Thanks for the post , but if you feel like , then you must do more tests, cos it seem that it is not an ami WINbios board fault , but the CPU itself, i tryed dx4 100, pentium overdrive and they works fine, i did not check other cpus but the problem is the same using Award SIS motherboard/bios so it looks as if the fault is in cpu, not the bios version , and the very dx2 cpu only, so far. please do more tests if you want this to be fixed and check every other cpu, that will take like 30-60 min. but you started the topic. some opcode error is displayed at the bottom maybe it will help.

[omarsis81]

As a matter of fact you began a similar thread some months ago
viewtopic.php?f=2&t=483&p=3278

[szadycbr]

Good catch, Omarsis81. anyway the issue seems to exist only on dx2 cpus. ill do some more tests, will see.

[SarahWalker]

Rev 665 should fix this.

[ppgrainbow]

Thank you so much for the heads up and fixing the problem! It turned out that thex87 instructions are NOT supposed to cause an illegal exception when trying to detect the co-processor!

I did post a similar thread last year, but it was never brought to our attention. Once again, thank you so much for fix this bug!

In any event I get updated, I will need to run more tests to make sure that the illegal instruction set bug does not persist.

[A. Naim]

One to three days between most bug reports and the fixes...Rather amazing.

[omarsis81]

and one week to emulate a hard disk controller! That's even more amazing!
I've read about Bill Gates, John Carmack, Michael Abrash as being good programmers, but arguably Sarah Walker is the best programmer I've seen! Hats off

[SarahWalker]

Err, it wasn't exactly a difficult bug...

[omarsis81]

Not this one in particular, but I'm talking in general, that's why I said about the HD controllers.
Anyway, I admire you. Keep up the outstanding work

PS: Oh, and to make this post a little more on-topic and talking about the HD controllers, do you think we can see selectable IDE controllers for the 286 and 386 clones? There no rush of course, and current "Internal IDE" work perfect, but as you know, internal IDE controllers began with the 486. Just to make thing a little more historical.

[EluanCM]

I have a 286 with internal IDE (SCAT chipset). But since the chipset is from 1989, it's on par with the 486 timeframe.

That makes you half-right

[omarsis81]

I have a 286 with internal IDE (SCAT chipset). But since the chipset is from 1989, it's on par with the 486 timeframe.

That makes you half-right

Can you post a pic of the motherboard? I never seen a 286 with an onboard IDE controller

[JohnElliott]

286 and 386 portables like the Compaq SLT would have had onboard IDE.

And Amstrad's late XT clones had XTA on the motherboard, so I expect the 286 and 386 versions would have had ATA on the motherboard. There's certainly a 40-pin header on the Amstrad 5286 motherboard.

[Battler]

My first PC was an Amstrad MegaPC (PC70386) and it definitely used IDE. And PCem also emulates the AMI and MR 386 clones that also have IDE.

[omarsis81]

You people didn't understand my point, or I expressed myself in the wrong way.
Of course all current PCem support IDE! What I meant was clone 286 and 386 (by clone I mean those motherboards you purchase alone so you can assemble a PC yourself) don't have an integrated/onboard IDE interface, you'll have to plug in an ISA card to have IDE, and PCem lists those boards as I they'd come with an internal IDE.

So, I'm not talking about Compaq portables, nor Amstrad Computers

Eg. here's a guy reviewing a 386 board and as you can see there is no IDE, he'd need an ISA controller
https://www.youtube.com/watch?v=-sMMIxYPGp8

What I suggested in a distant future is to have those generic ISA IDE controllers for the 286/386 clones and "Integrated IDE" for the 486 onward

[EluanCM]

I have a 286 with internal IDE (SCAT chipset). But since the chipset is from 1989, it's on par with the 486 timeframe.

That makes you half-right

Can you post a pic of the motherboard? I never seen a 286 with an onboard IDE controller

It's on storage, unfortunately (it was my first PC and nowadays I don't even risk turning it on before doing a full restoration (think capacitors, etc))

But I've found this not so clear picture in a russian forum:

At the center of the picture is the floppy and IDE connector, very hard to see.
The chipset is Chips & Technologies F82C235 / CHIPS F82C710. Everything is integrated, including IDE.

[vbdasc]

You people didn't understand my point, or I expressed myself in the wrong way.
Of course all current PCem support IDE! What I meant was clone 286 and 386 (by clone I mean those motherboards you purchase alone so you can assemble a PC yourself) don't have an integrated/onboard IDE interface, you'll have to plug in an ISA card to have IDE, and PCem lists those boards as I they'd come with an internal IDE.

So, I'm not talking about Compaq portables, nor Amstrad Computers

Eg. here's a guy reviewing a 386 board and as you can see there is no IDE, he'd need an ISA controller
https://www.youtube.com/watch?v=-sMMIxYPGp8

What I suggested in a distant future is to have those generic ISA IDE controllers for the 286/386 clones and "Integrated IDE" for the 486 onward

AFAIK, these ISA SuperIO and IDE "controllers" are nothing more than dumb bridges between the ISA bus and the ATA bus of the disk. So, emulating them, as far as disks are concerned, makes no sense, for there is nothing to emulate. AFAIK, the "integrated IDE" of some 286/386/486 ISA/VLB systems is in fact the same "dumb ISA/ATA bridge" too, and it can't be distinguished from an ISA IDE "controller" from the software POV, and emulating it makes no sense either.

P.S. By the way, I once had a 486 and VLB-based computer that didn't have onboard IDE.

[SarahWalker]

Indeed. You can pretend that 'Internal IDE controller' says 'Generic IDE controller' if it makes you feel any better.

[EluanCM]

After all, that's what integrated drive electronics is supposed to mean, with later controllers being yet another layer of abstraction (afaik)

[ecksemmess]

Indeed. You can pretend that 'Internal IDE controller' says 'Generic IDE controller' if it makes you feel any better.

Perhaps it actually should be changed to "Generic IDE controller" to avoid any further confusion along these lines?

[ppgrainbow]

I'm wondering if anyone is running into a issue when NSSI misdetects the 50MHz 486 as a 486DX2 due to a CPUID issue:

486-50.png (31.63 KiB) Viewed 19126 times

80486 processors running from 20 MHz to 50 MHz operate at par clock rate of the external bus clock. 486DX2 processors run at twice the clock rate of the external bus clock. I'm wondering if it's possible to look in this bug.

Did we forgot about adding the 386(DX) 12MHz, 486(DX) 20 MHz and 486SX2 66MHz machine configuration...?

[omarsis81]

I get PCem crashing since last build each time I change a 486 processor
Reports an error on msvcrt.dll
Does the same happen to you?

[szadycbr]

OMARSIS81, something really happend with last build, mine crashes every second time, when i do change CPU, no matter which.

[SarahWalker]

Did we forgot about adding the 386(DX) 12MHz, 486(DX) 20 MHz and 486SX2 66MHz machine configuration...?

I'm unaware of the 386/12 ever being used in a PC. I couldn't find any evidence of the 486DX/20 existing, and the 486SX2/66 is implemented, but only for AMD as they seem to be the only company that made one.

[ppgrainbow]

Did we forgot about adding the 386(DX) 12MHz, 486(DX) 20 MHz and 486SX2 66MHz machine configuration...?

I'm unaware of the 386/12 ever being used in a PC. I couldn't find any evidence of the 486DX/20 existing, and the 486SX2/66 is implemented, but only for AMD as they seem to be the only company that made one.

Intel did make the 66MHz i486SX2 (i486SX2-66) processor, but it was hardly used. When the Intel 80486 was introduced in April 1989, it had clock speeds of 20 MHz and 25 MHz. The slower variant the i486(DX)-20 was almost nowhere being used.

Intel was originally intended to introduce the 80386 processor at 16 MHz, but due to poor yields, the 80386 was introduced at 12 MHz. Here's what a early 80386 processor at 12 MHz looks like: https://en.wikipedia.org/wiki/File:Intel_A80386-12.jpg

[omarsis81]

Did we forgot about adding the 386(DX) 12MHz, 486(DX) 20 MHz and 486SX2 66MHz machine configuration...?

I'm unaware of the 386/12 ever being used in a PC. I couldn't find any evidence of the 486DX/20 existing, and the 486SX2/66 is implemented, but only for AMD as they seem to be the only company that made one.

Intel did make the 66MHz i486SX2 (i486SX2-66) processor, but it was hardly used. When the Intel 80486 was introduced in April 1989, it had clock speeds of 20 MHz and 25 MHz. The slower variant the i486(DX)-20 was almost nowhere being used.

Intel was originally intended to introduce the 80386 processor at 16 MHz, but due to poor yields, the 80386 was introduced at 12 MHz. Here's what a early 80386 processor at 12 MHz looks like: https://en.wikipedia.org/wiki/File:Intel_A80386-12.jpg

She didn't say it doesn't exit (386@12) but that it wasn't used at all), even the first 386 computer ever, the Compaq Deskpro 386 from 1987 was launched at 16 mhz

[SarahWalker]

Rev 666 should fix the crash in the config window.

[SarahWalker]

If you want me to add emulation of the 486DX/20 and i486SX2/66, then show me some actual evidence that they exist - a photo would do...

[szadycbr]

Hell'ish Revision 666!!!