List of Outstanding Projects

Hardware/Driver List

Listed here are projects that deal with writing drivers for a certain bit of hardware that I for some reason have. No code is probably actually written, but most likely I still have the hardware, in case I ever feel like starting. Basically, I'm going through my Big Drawer of Hardware (anyone who's been in here knows what I'm talking about) and picking out the stuff I don't have drivers for (yet). Mostly listed by manufacturing year.

• Early Novell EISA ethernet adapter (1989) This is a rather large board, and a very complex one. However, the chips on it are quite documented: i82355 EISA BMIC, i82586 (shared-memory ethernet controller, used on the Intel EtherExpress 16 ISA cards, in this board, probably used only by the code thats in the ROMs), i80186 (10mhz!), lots of RAM, several PALs, several connectors (one labelled "Monitor" -- may be an rs232 console to the 186!), and 2x2764 ROMS (in DIPs that I can read with my burner). Would be interested hack indeed. I wish I had a faster EISA machine to do devevlopement on. No code available.
• IBM Expanded Memory Adapter for MCA (1991?) I have several of these. Early PS/2's (like the 8580) take only planar RAM, which is difficult to find these days. These boards were created so that you could put more RAM in a 32bit MCA slot. However, they need drivers to run. I'd like to get these working in linux, if not as real memory (unlikely) then as a ramdisk that I can make a swapfile on. If nothing else, its still another 16mb or so thats faster than a hard disk, however slightly slower than real planar RAM. I have yet to find the dos drivers. There was rumor of an alpha driver for linux floating around but I have yet to find it. Probably the only way to figure them out would be to disassemble the dos drivers.
• First-generation Intel TokenLink ISA token ring card (1992) The simplest TMS380 board there is. Its attached to ISA via simple logic only, and uses the Second-Generation TMS380 chipset (the two-chip version: TMS380C16 and the TMS38053). I'm not even sure if this card is functional, as I don't have any drivers for it even for DOS. Shouldn't be too difficult to get it working with the tms380tr driver library once its I/O model is figured out (assuming its functional at all).
• Creative Labs VideoBlaster (CT6000) (1992?) First of the PC video digitizers, way overpriced, and slightly overdesigned. It was as good as to be expected for the time. Uses a Chips&Tech overlay chip along with the standard Philips video interfacing. There's documentation available for everything, as well as userspace utilities for linux floating around somewhere. These boards make nice webcam capture boards. Writing a kernel v4l driver for them would just be an ego booster (which is why its on my list).
• Creative Labs VideoSpigot (CT6030) (1993) This is a very simple video digitization board (on ISA, so its usefulness is limited). Its based on the standard Philips chipset: SAA7191, TDA8708, and TDA7197. Has Svideo and composite inputs, no sound. Also has a small amount of frame buffer on it. I actually have some old windows source code for this. I just haven't gotten around to writing the driver yet. It shouldn't be too complex. The i2c bus is connected directly to ISA, and everything except the buffers is controlled through that. No code available (I have some somewhere, I don't remember where it went).
• Second-Generation(?) Intel ISA token ring adapter (1993) Based on the same TMS380 chipset as the first-generation Intel board, however the ISA interfacing is cleaned up a lot. The pseudo-PnP glus chip (MB620228) looks suspiciously like it came from Madge. I've made basic attempts at this, and could probably get it working in at least pDMA mode if I could figure out how to enable interrupts on the glue chip.
• IBM RS/6000 Model 7011-250 (1993?) A MicroChannel based RS/6000. I'd like to get Linux/PPC running on this. It's got a 66mhz PowerPC 601. More info here. I'd also like to work on the pre-PPC POWER/POWER2 based RS/6ks if I can get the hardware and CPU docs.
• SMC EISA token ring adapter (1994) Looks to be a very well designed card. I have no documentation for it, however, Jay has gotten documentation out of SMC in the past (for the smctr driver for ISA boards), so it may not be out of reach.
• Madge Smart 16/4 MC Ringnode (1994) Second-generation TMS380-based token ring adapter for the MicroChannel bus. Uses two Madge glue chips: MB606207 and MB621206. I don't have any documentation on the glue chips at all. However, from playing with the config options and the register space, it looks like Madge might have actually tried to take full(ish) advantage of the features of MCA. They talk about 40Mb/sec Host-to-Card DMA, which would surpass the bandwidth capable of even their PCI Mk2 boards (though not their Presto Knossos-based boards). It would be really nifty to get this working, however the fastest MCA machine I have is a model 80 (386-20). Anyone have an old MCA RS/6000 I can borrow?
• Diamond Multimeda DTV1100 (1995) Addon board for the Stealth series of video cards, however since it uses the standard LPB bus, it should be ussable with any S3 video card. It takes up an ISA slot. Contains a standard Temic tuner and a Micronas VPX3220 video decoder (I have datasheets on it). Everything happens across LPB. The decoder is controlled using the i2c bus thats part of the LPB spec.
• IBM Wireless LAN Entry PCMCIA cards (1995) No real documentation on these. They're quite slow, but I still use them (granted, only in Win95). They work well for what I wanted, and I got a good deal on them. They're based on an unnamed Intel ethernet chipset, and use a proprietary undocumented IBM ASIC for radio and interfacing. There's other people working on these, but very little code has come out of those attempts. I doubt mine will be much different.
• Madge Smart 16/4 PCMCIA Ringnode Mk2 (1996) [Madge supplied] TMS380-based token ring controller. Have only Madge's limited source code to go off of.
• Olicom OC-3137 PCI token ring adapter (1997) [Madge supplied] Fourth-generation TMS380-based board, with Olicom glue (P/N 119000026). No documentation.
• Olicom OC-3118/512k ISA token ring adapter (1997) [Madge supplied] Fourth-generation TMS380-based board, with Olicom glu (P/N 119000023). No documentation.
• Madge Smart 16/4 AT Plus Ringnode (1998) [Madge supplied] Fourth-generation TMS380-baesd, with Madge glue (MB606259 -- this chip has both the Madge logo and the F logo that is on the Intel chip above, further proof that the Intel chip came from Madge). This board uses four DIP switches and an Atmel AT29C512 EEPROM. Probably the simplest Madge board there is for ISA, but still complicated enough to make reverse engineering difficult. No documentation.
• Madge Smart 16/4 ISA Client PnP Ringnode (1998) [Madge supplied] Fourth-generation TMS380-based, with Madge glue (1616F1237). No documentation.
• HP PCMCIA SCSI controller that comes with the CD-Writer M820e (1999) This uses an unknown Panasonic chipset. None of Panasonic's engineers seem to speak enough english to understand what I mean by "documentation".
• Madge Smart 16/4 PCI Ringnode Mk2 (1999) [Madge supplied] The "Abyss" board. Uses the fourth-generation TMS380 chipset, plus a Madge glue chip (140-019-02 [new numbering scheme]). No documentation except the Madge public source. This is actually already supported by the in-beta tms380tr driver.
• Olicom RapidFire 3540 (OC-3540) 100/16/4 PCI token ring adapter (1999) [Madge supplied] Olicom custom ASIC, with TMS38054 and Broadcomm BC5201 chips for the final wire output. No documentation.

Undone/Unstarted Userland Projects

These are projects that I either thought of a really really good idea for and/or actually started coding, but never finished because of various difficulties (probably time or hardware constraints).

• netbridge: Userspace generalized network bridger. Should be able to translate across topologies. Started sometime in late Decemeber 99, as the software component in David/Brocks invisible machine. I decided to go for userspace after discovering that I didn't really want to hack a translational bridge into the kernel bridging code, which at the time was in strange disrepair. I have all the required I/O routines does (using libnet and libpcap), and the basic translation code done. However, it still needs the logic that bridges have (learning, loop discovery, etc). Source available.