@Trev

I just had a look at the 2.4 kernel code in drivers/pci/quirks.c, and the via latency patch is in there, but as far as I can tell, it's only activated on certain northbridges, i.e., not with the ArticaS. The same is true for the 2.6 kernel.

No idea if he's still maintaining the Linux kernel, but rossv used to maintain the A1 Linux kernel. He'll be able to tell you what workarounds were tested.

Hans

_________________
http://hdrlab.org.nz/ - Amiga OS 4 projects, programming articles and more. Home of the RadeonHD driver for Amiga OS 4.x project.
https://keasigmadelta.com/ - More of my work.

@Hans

Just some quick observations/thoughts after slogging through archvies:

* Access to the Articia S Databook was for 1337 hax0rz (under NDA with MAI or Eyetech).
* If your handle isn't Rogue, Rogue thinks you don't know what you're talking about. (Sorry, but dude, this isn't a local C64 BBS where it's all about trumping your buddies.)
* Those with access to information bragged, gloated, and generally made everyone else feel inferior.
* Who owns the old MAI Logic IP? No one? Thought so. Post the databook already.
* AmigaKit bought some or all of what was left of Eyetech, yeah? Maybe Matthew will share what he has. Haven't asked yet.

Anyhow, most of that is in jest. Things sure were messy. Glad I decided to settle down and get married instead.

From an anonymous poster in http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=12656&forum=2#:

Quote:

Well, *roughly* speaking it's about disconnecting [PREQ# and PGNT# on the VT82C686B] from PREQ4(Articia) and PGNT4(Articia) and reroute them to ISAREQ(Articia) and ISAGNT(Articia) instead, which involves scratching off PCB traces and soldering cables on various places on both sides of the motherboard.

On the VIA side, PREQ# and PGNT# go to the north bridge to request and receive PCI bus access, respectively. Without access to the Articia S databook, I don't really know what switching the lines means for the south bridge. I don't really know what I'm talking about, regardless. Maybe the routes were changed because the 686B functions as a PCI-to-ISA bridge?

I also see that by default, the VT82C686B doesn't wait for a PCI grant before allowing IDE DMA. I hope I'm misunderstanding that part.

EDIT:

Oh, I get what you're saying now. They work around the 686B south bridge issue by forcing the north bridge to do master priority rotation after every grant. And it's possible that this can't be changed on the Articia S?

On the 686B, Look at:

IDE
Offset 4A - IDE Interrupt Routing; Bit 7 (in the U-Boot code I have, this bit is cleared--sounds wrong to me, but I don't know)
Offset 5C - DMA Control; Bit 4 (dummy PREQ--necessary? not in U-Boot)

USB (all)
Offset 41 - Miscellaneous Control 2; Bit 4 (U-Boot sets this bit--if PCI grants are an issue, should we be extending grants for USB?)

EDIT2:

I also don't see where 686B interrupts are routed to the Articia PIC by U-Boot. Not necessary? Are all south bridge interrupts shared? That's what it looks like. Anyhow, I need newer U-Boot sources.

Last edited by Trev on 29-Jan-2010 at 07:17 AM.
Last edited by Trev on 29-Jan-2010 at 07:08 AM.

_________________
Sam440ep-flex 733 MHz/1 GB RAM/Radeon 9250/AmigaOS4.1 Update 2
borked A1200/Blizzard1260+SCSI-IV/Z4+MediatorZIV/Deneb/Voodoo3/CatweaselMk3
more borked A1200/MBX1200z/Indivision
A500/clockport/RRNet
A600/A603

One more quick, amusing observation: If you download the latest AmigaOne U-Boot from Hyperion, you'll notice that the only license mentioned is the GPL. With the inclusion of slb_v2 and no other explicit license specified, one might assume that slb_v2 is now GPL. I'm sure that wasn't their intent.

_________________
Sam440ep-flex 733 MHz/1 GB RAM/Radeon 9250/AmigaOS4.1 Update 2
borked A1200/Blizzard1260+SCSI-IV/Z4+MediatorZIV/Deneb/Voodoo3/CatweaselMk3
more borked A1200/MBX1200z/Indivision
A500/clockport/RRNet
A600/A603

@Trev

No luck with AmigaKit re: Articia S documentation. The most likely source is Hyperion, then.

I've also got a friendly request out to Hyperion for U-Boot source. The only email address on the site is webmaster@hyperion-entertainment.biz. If that doesn't work, I'll ask Rogue directly.

While a lot of OS4 development appears to informal, I suspect I'll only be able to get Articia S documentation by signing an NDA with Hyperion. I'm open to that, but this is the Amiga scene, and something tells me I don't have the street cred to be considered. I'll probably have to blow up an Atari web site or something.

EDIT: Replaced a preposition with an article.

Last edited by Trev on 29-Jan-2010 at 09:48 PM.

_________________
Sam440ep-flex 733 MHz/1 GB RAM/Radeon 9250/AmigaOS4.1 Update 2
borked A1200/Blizzard1260+SCSI-IV/Z4+MediatorZIV/Deneb/Voodoo3/CatweaselMk3
more borked A1200/MBX1200z/Indivision
A500/clockport/RRNet
A600/A603