I had posted this up a long time ago (like 3 years ago) but never sold it. Time has come to clear it out.

What I have is the MK2A/Riocar 30gb (was a 20GB but upgraded drive and cables), Original tuner (purchased around 2002 via this board), remote control, USB, serial and power cable and sled (cables and power are original units) and original box (*I think*) with the manuals from when I purchased it from another user on the board. Serial is 030102389. Version of firmware is 2.0.1. Pictures available on request.

I'm not interested in parting it out at all. I don't really want to ebay it, and will consider reasonable offers. Any questions or interest, email / PM / post.

Located in Phx, AZ.
Thanks,
Kevin

I have it listed up on Ebay now ... http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=280285080943

I'm not entirely privy to the market value of these units yet (I'm still researching), but I am interested in what you have. If you want to PM me an asking price, I will give it serious consideration.

kastle, I meant to get back to you before. If you're willing to cancel the auction, I'd like to pick this up from you - we can talk about the price (jcm@jonmasters.org) or I can bid on it on ebay I guess...I'd like to pick another one up for development use.

kastle, I'm in Boston, MA btw.

Item sold to Jon(jcm) and I'm betting it's gonna be well loved

Indeed. One empeg clearly wasn't enough for me

My plan for this new one is to dissect mlord's hijack kernel and port the changes to 2.6, and other things.

Please keep us all abreast of your progress!

Heh.. it's messy, organic code, but it works!

Cheers

Hmm, organic I think I will borrow this description for some of my, err, less coldly architected works...

"Organic", in that it just kinda grew on its own, without any particular plan in the beginning.

Cheers

...and in the sense that it's still alive and evolving.

Oh, sure, there' no /reason/ to run a 2.6 kernel, it's just fun. Is there actually a git tree or a quilt tree for those patch files listed on the sf.net project?

So it arrived in the mail today. Interestingly, this one won't upgrade using the windows tools on my KVM image. Probably it's just an artifact of the virtualization and timing, but it means I'll rebuild the disk manually at the weekend. It's running hijack v503 fine though, so I might have a go at building a kernel before then. First thing is to walk by understanding the existing kernel.

The issue might be the serial port. USB-serial timing has always been flakey with these things, and never completely sorted out.

I use a machine with a real ISA serial port when necessary, and the AirCable bluetooth-to-serial dongles for most stuff.

USB-serial *used* to work with download.c on the PC-side on Linux, but it got broken at somepoint, in the 2.6.2x series I think. I can boot a 2.6.16 kernel and do serial download over USB, but not with a modern kernel. Never had time to go hunting there, though.

Cheers

Yeah, but this one won't work with either USB or regular serial (caveat - didn't extensively test USB) for flash upgrades. Some timing issue where the bootloader continues booting anyway.

Not that it matters, I finally ripped apart your "bigdisk" images and looked at partitioning scripts, then duplicated them with the disk mounted via a USB adapter. So "pumping" is actually pretty trivial to do. I'll build a kernel at the w/e once I've got the correct toolchain and so forth.

I can't see any reason why you couldn't increase the root partition over 16MB either. Is anyone actually using that "spare" partition for anything? Your bigdisk image will format it with ext2 but I'm not sure what the purpose is really.

That, is one of the timing issues I mentioned, at least for USB-serial (that is, a USB serial dongle on the PC side, connected to the DB9M on the back of the player).

Note also, that upgrades cannot work in "car mode" (which means they also don't normally work in "docked" mode).

You'll need to have the bare metal player laying on the desktop, with the AC-adapter pluggable directly in the rear of it.

Cheers

There's a discussion of this elsewhere, in the bigdisk builder thread. It can be done, but it then forever requires a Hijack kernel to keep it that way, and one must never again even attempt to apply a factory .upgrade file.

Because the initrd in flash is where the partitioner resides, and ISTR that it attempts to "correct" the partition table on every boot (or is that on every .upgrade?.. don't rememeber).

I've hacked Hijack to intercept and prevent this "correction", but I don't remember whether that change went into the main release, or was just played with out-of-stream.

Cheers

Every upgrade. A corrupted partition table is one of the failure modes that needs to be fixable by applying an upgrade.

Um, I had the impression you abandoned the idea because Hijack isn't running if you're applying a stock upgrade file: the kernel from the upgrade file is. Again, this is because a corrupted kernel in the flash is one of the things that applying an upgrade needs to be able to fix.

Peter

The bigdisk2 .upgrades take care of the partition table themselves, so the initrd actions are counterproductive there. And bigdisk(2) .upgrades have a Hijack kernel already in place.

Cheers