I would like to know if parts list are available for these boards and where I get a parts list.

stvngrm@netscape.net wrote on 2024-08-31, 18:08:

I would like to know if parts list are available for these boards and where I get a parts list.

KiCAD can export BOM in a way that'll at least describe what parts you need to look for.

You should check out the v0.2 version of the board as that's the known working one. You need to populate all components except R6, and an extra resistor is needed under the ISA slot.

The main issue would be the F85226 chips. You may have to look for online sellers if the usual part shops don't have any in stock... And of course, be careful of potentially defective parts.

Back then I got a few F85226FG and F85226AF from different sellers, and it appeared the former (FG) were defective that it would suddenly stop operating. It took me quite a while to find out the issue was the chip itself. The AF chips, on the other hand, were okay.

EDIT: Sorry, should have said about the extra resistor under the ISA slot sooner... see the picture below (which could be found several pages back as well).

LSS10999 wrote on 2024-09-01, 03:44:

You should check out the v0.2 version of the board as that's the known working one, and you need to populate all components except R6.

0.2 is a known NOT working version of the board, you have to add an extra resistor :

rasteri wrote on 2023-10-23, 20:54:

OK, installing an 8.2K resistor between MASTER# and 5V makes it boot perfectly every time again.

I imagine 0.3 is fine but LSS10999 had some trouble with it, and I wouldn't recommend building any version unless you're willing to do a lot of debugging.

I will release a final tested version but I'm finalizing another project just now

Super cool rasteri!
Like GrafWasili, it sure is intriguing to know exactly whether DMA can be coaxed from a 100+ series PCH and whether some vestigial hardware DMA functionality remains in the PCH in spite of documentation clearly stating to the contrary. Handing over TPM and (more relevant) flash to eSPI sure doesn't provide much hope. Close to the edge though, 11th-gen intel seems to be the barrier for CSM support for DOS booting. Pretty incredible to be here talking 6th-gen ISA DMA!
Meanwhile, I'm actually wondering about the other direction, how early can the F85226 be used - i have a GA-G31/E8600 here i might experiment with, pulling LPC sigs from the SuperIO and tracking down LDRQ1#.
I continue to be incredibly time poor, but I'll report back if i do get any traction.

Kenpachi wrote on 2024-09-07, 13:38:

Close to the edge though, 11th-gen intel seems to be the barrier for CSM support for DOS booting. Pretty incredible to be here talking 6th-gen ISA DMA!

I keep name-dropping TKChia's mUEFircate project, in the hope that other knowledgeable software hackers start contributing to that project. The concept is quite promising: a piece of software that can boot from pure UEFI (even without a CSM), sets up the resources that a legacy BIOS would, and then boots older pre-UEFI operating systems such as DOS.

I know, eventually Intel will drop native 16-bit and 32-bit compatibility from its ISA completely, but until they do, let's see how much further we can take this legacy compatibility. Let's see if the dISAppointment can be made to work on 11th gen and beyond!

Kenpachi wrote on 2024-09-07, 13:38:

Like GrafWasili, it sure is intriguing to know exactly whether DMA can be coaxed from a 100+ series PCH and whether some vestigial hardware DMA functionality remains in the PCH in spite of documentation clearly stating to the contrary.

I would love to find a way to do 8237-style DMA on later chipsets but it ain't gonna happen through LPC, the LDRQ line is physically gone from the chip.

I don't know if the DMA controllers exist inside the chipset still. In many ways it would be better for us if they didn't, since we could simulate them externally, perhaps via PCIe.

But it's a complex prospect, requiring pretty intimate knowledge of PCIe and ISA DMA, and would need to be implemented on an FPGA.

disappointment is a really simple project by comparison, just a readymade IC and an ISA slot.

There are STILL some motherboards with current chipsets including ECP compatible parallel ports so yes.... the 8237 lives on somewhere. I suspect it will still hang on as long as platforms still support booting DOS via CSM.

digger wrote on 2024-09-07, 15:46:

I keep name-dropping TKChia's mUEFircate project, in the hope that other knowledgeable software hackers start contributing to that project. The concept is quite promising: a piece of software that can boot from pure UEFI (even without a CSM), sets up the resources that a legacy BIOS would, and then boots older pre-UEFI operating systems such as DOS.

Yes! mUEFIrcate, i always have that project in mind, though i don't often recall the name. I think it'll become a household name around here once post-11th-gen PCs become more commonplace.

rasteri wrote on 2024-09-07, 16:08:

I would love to find a way to do 8237-style DMA on later chipsets but it ain't gonna happen through LPC, the LDRQ line is physic […]
Show full quote

I would love to find a way to do 8237-style DMA on later chipsets but it ain't gonna happen through LPC, the LDRQ line is physically gone from the chip.

I don't know if the DMA controllers exist inside the chipset still. In many ways it would be better for us if they didn't, since we could simulate them externally, perhaps via PCIe.

But it's a complex prospect, requiring pretty intimate knowledge of PCIe and ISA DMA, and would need to be implemented on an FPGA.

disappointment is a really simple project by comparison, just a readymade IC and an ISA slot.

dISAppointment is awesome and i am keen to try it out! Sapphire goodness for those (me) who can't fork out for a Ruby!

I agree, i think solution for the 100 series and beyond is going to be some kind of convoluted bit-banging of reserved registers.

Edit - or as you stated, an entirely different bridging method...

Personally, my dream system is now finally covered, which is c2d with PCIe+ISA while retaining FDD & PS2 in win98. Super cool!

Is there any progress with mUEFIrcate? I found it years ago but we didnt make it work on 11th gen HP zbook laptop due to missing legacy vga bios... how about new amd zen5 boards, do they still have csm? Anybody tried run dos/w9x there?

RayeR wrote on 2024-09-10, 13:08:

Is there any progress with mUEFIrcate? I found it years ago but we didnt make it work on 11th gen HP zbook laptop due to missing legacy vga bios... how about new amd zen5 boards, do they still have csm? Anybody tried run dos/w9x there?

For systems that lack a legacy VGA BIOS, a solution might be to use the generic SeaVGABIOS. It should be compatible with GPUs that still have register-level VGA compatibility. For instance for Coreboot, SeaVGABIOS exists as a "fallback" option when the native VGA BIOS of a GPU isn't extracted from the target system before Coreboot is built for it and flashed on it. Granted, it might only offer "generic" (legacy) VGA compatibility, which would e no higher resolutions and color depths. But for most DOS games, that would be sufficient.

(UPDATE: Apparently, SeaVGABIOS also supports some VBE modes, but I'm not entirely sure how that works. It seems to defer to whatever framebuffer device Coreboot makes available, if I understand correctly how that works. Maybe SeaVGABIOS could at least in theory also support VBE modes by deferring to the UEFI framebuffer, but I'm not sure if that framebuffer is still accessible once mUEFIrcate is booted from UEFI and has switched to legacy VGA text mode through SeaVGABIOS.)

I think the biggest problem with the project at the moment is the lack of developers. TK Chia has been working on it alone until now, and apparently doesn't always have time to work on it.

A healthy community of skilled contributors to this project would do it good.

I believe it could really be taken to the next level with some reasonable group effort! More publicity around the project would help.

Eep, well it seems i lost the DDR2 for my c2d E8600 board, and i couldn't seem to find any cheaper than about AU$60. So i kinda ended up with a "machinist" x99 xeon + 16GB DDR4 purchase instead, and surprisingly, change.
That escalated quickly.
Be interesting to see what that's like when it arrives.

I'm still quite curious as to the AM3+ option, did that eventuate? I definitely read, and enjoyed, the entire thread, but i think i had my c2d in mind the entire time, with perhaps an eye on the x99 info.
It did occur that it would be pretty incredible to have a pcie4.0 and ISA, after noting pcie3.0 with the x99.

I've got an F85226AF en route(two actually), though finding time to order some boards/ assemble a dISAppointment isn't likely to appear soon.

It's been quite some time since I fired up Altium (i haven't check out KiCad), but adding another slot or two is also playing on my mind, against a backdrop of other crazy ideas. I was still hoping for an FDC, and on an x99 won't be easy.

But in any case, before i get ahead of myself, there'll be at least one LDRQ1# to find!

Kenpachi wrote on 2024-09-15, 11:36:

Eep, well it seems i lost the DDR2 for my c2d E8600 board, and i couldn't seem to find any cheaper than about AU$60. So i kinda ended up with a "machinist" x99 xeon + 16GB DDR4 purchase instead, and surprisingly, change.
That escalated quickly.
Be interesting to see what that's like when it arrives.

I'm afraid those machinist boards don't route either LDRQ pin anywhere - Re: dISAppointment - LPC to ISA adapter - ISA on modern motherboards

Unless you fancy some extremely acrobatic board rework you might be out of luck 🙁

I'm still quite curious as to the AM3+ option, did that eventuate? I definitely read, and enjoyed, the entire thread, but i think i had my c2d in mind the entire time, with perhaps an eye on the x99 info.
It did occur that it would be pretty incredible to have a pcie4.0 and ISA, after noting pcie3.0 with the x99.

No I'm afraid nobody has figured out AMD chipsets yet. I'm sure it must be possible but it will likely require some deep PCI register hacking

Hi rasteri,

Thanks for your reply!
Oh dear yes, i recall you looking under the PCH but my memory omitted that crucial detail. Hard to stay focused with an apparent x99 bargain while looking for DDR2. It's a machinist K9 which is a little different, though I don't see why there's any particular hope vs the PR9.
Oh well, better order that DDR2 for the c2d.
When time permits I'll have a dig into the K9 and report back when i do manage to get around to it.

Right, ok. Seems the best path really is just finding an Intel, sub-100 series board with a LDRQ-routed TPM header... the easy path to victory!

Kenpachi wrote on 2024-09-20, 13:39:

It's a machinist K9 which is a little different, though I don't see why there's any

Oh maybe - if you check which superIO it has then if that SuperIO uses LDRQ then there may be hope...

Kenpachi wrote on 2024-09-20, 13:39:

Oh well, better order that DDR2 for the c2d.
When time permits I'll have a dig into the K9 and report back when i do manage to get around to it.
Right, ok. Seems the best path really is just finding an Intel, sub-100 series board with a LDRQ-routed TPM header... the easy path to victory!

I think you could go easy for some Gigabyte MB with 5x - 9x chipsets, there are pretty good chances you can reach LDRQ# at least on unpopulated resistor to tie a wire there. At least Gigabyte has leaked a lot of schematics that will help you to find it. And some GB MBs has even precious legacy ports (COM, LPT).
My GB P67 is quite perfect, only lack floppy controller but I added it via secondary SuperIO chip at LPC...
So why to bother with some noname chinese crap...

Hi rasteri and RayeR,

I did find a boardview for the G31M-ES2L, yes superIO has LDRQ0, and LDRQ1 is NC, which is not ideal, but not a total loss.
Of interest though, LDRQ1 on the ICH7 is right next to INTRUDER#, and the G31M-ES2L has a chassis intruder header, apparently, and those ICH pads would be easy to bridge: that would trip XOR Chain 3 output RI# (Ring Indicator) with every LDRQ1- i guess you could bridge pads and mask to reroute it.
As long as RI_EN in GPE0_EN isn't set, pad masking shouldn't be needed. Theoretically.
Of course, that might incorrectly indicate I'm up for an ICH desolder... and I don't even have a heatgun at the moment!

RayeR,
So cool you added the FDD to the P67, i read that on your site. And PS2!
I guess if i chase LDRQ1, it'll be a compromise with both options, as the G31M-ESL2 has FDD...
I suppose i have a few options, at least - pull LDRQ0 from superIO and test for problems sharing the request line; toggle LDRQ1 and hope for a pad that Gigabyte magically routed to LDRQ1 (still waiting for DDR2); win the long-shot lottery with the Machinist K9, or go and source a more suitable board...

Until the mail arrives...

Re: vga bios

There are various versions of the intel iGPU vbios in the linux-firmware package that may be of interest.

Most of these newer machines boot in intel video mode, OR, only have the iGPU to begin with.

So, stop me if I’m wrong, but could this method be used to adapt the LPC on a vortex86MX to attach one of the Crystal chips used in the WeeCee?

DJNW wrote on 2024-10-14, 12:01:

So, stop me if I’m wrong, but could this method be used to adapt the LPC on a vortex86MX to attach one of the Crystal chips used in the WeeCee?

According to MX datasheet, all necessary LPC and PCI signal pins are present.
So, in theory, it is possible (pitfalls included, of course).
Another question why would anyone bother doing it, though.

Paul_V wrote on 2024-10-14, 13:12:

According to MX datasheet, all necessary LPC and PCI signal pins are present. So, in theory, it is possible (pitfalls included, […]
Show full quote

DJNW wrote on 2024-10-14, 12:01:

So, stop me if I’m wrong, but could this method be used to adapt the LPC on a vortex86MX to attach one of the Crystal chips used in the WeeCee?

According to MX datasheet, all necessary LPC and PCI signal pins are present.
So, in theory, it is possible (pitfalls included, of course).
Another question why would anyone bother doing it, though.

Okay, so hear me out - the onboard audio of the MX is HD Audio, so it can only be activated via SBEMU, which can cause issues with certain dos extenders (eg TD2192), right?
Someone could get hold of one of the various V86MX based Little Guys out there (eg ebox3350MX) and (assuming the right pins are accessible on the board) make up and attach a little LPC-attached sound card. If it's doable, it would be both cheaper and faster than having a new SOM shipped from Taiwan, easier to assemble and encourages reuse of old parts.