| 00:00:17 | GoRK | 31337 |
| 00:00:20 | scanline | ha |
| 00:00:21 | SpaceCoaster | Talking I said, the status register, jeez you guys! |
| 00:00:22 | scanline | 5280 |
| 00:00:36 | GoRK | how about 2630 as in IS2630 |
| 00:00:43 | scanline | hey, that's good |
| 00:00:51 | scanline | checks the nmap-services file... |
| 00:01:04 | scanline | yep, probably unused too |
| 00:01:12 | ProtWork | oof. Where's the wiki page to do the Flash WP defeat? |
| 00:01:25 | SpaceCoaster | GoRK: 2630 is good for me too. |
| 00:01:28 | GoRK | ProtWork: FlashUnlock |
| 00:01:36 | ProtWork | GoRK: Thanks. |
| 00:01:50 | ProtWork | Do I need to do this once I have a JTAG? |
| 00:02:09 | ProtWork | (Just wondering if it's O/S control) |
| 00:02:09 | SpaceCoaster | GoRK: do I need to do a block device before an MTD? |
| 00:02:14 | GoRK | i think so yeah |
| 00:02:49 | SpaceCoaster | GoRK: this flash has 264 bit pages, crazy eh? |
| 00:02:59 | GoRK | SpaceCoaster: an mtd driver is easier... all you have to do is provide the read/write/erase routines and everything else falls into place |
| 00:03:31 | GoRK | SpaceCoaster: maybe the extra bits are for checksumming, parity, etc? |
| 00:04:39 | SpaceCoaster | GoRK: maybe |
| 00:05:02 | ProtWork | oof, it would help if there was a legend on where to grab 12V from. :-) |
| 00:05:21 | GoRK | ProtWork: you have to grab it from your own 12v supply |
| 00:05:30 | ProtWork | GoRK: No 12V on board? |
| 00:05:35 | GoRK | ProtWork: nope |
| 00:06:31 | SpaceCoaster | ProtWork: what would the legend say, maybe a map to your nearest Radio Shack? |
| 00:07:15 | other | use the nearest computer |
| 00:07:27 | ProtWork | SpaceCoaster: Unlike the yahoos at Philips, I design in a flash write capability into my products. |
| 00:07:28 | other | the yellow wire on the power cord is generally 12V |
| 00:07:28 | other | :) |
| 00:07:33 | other | [its what i used] |
| 00:08:09 | GoRK | the power supply in the phone has 12V? or do you mean a power cable from your desktop? |
| 00:08:27 | SpaceCoaster | ProtWork: You are right I think they missed a trick or two, but they did quite a bit. |
| 00:08:45 | ProtWork | GoRK: I assumed there was a +12V in the phone, and didn't figure it would require an external 12V. |
| 00:09:05 | SpaceCoaster | woohoo I think I read a block from the serial flash |
| 00:10:36 | other | [other] use the nearest computer |
| 00:10:50 | other | i meant the nearest computer with AT/ATX power supply .. |
| 00:11:11 | ProtWork | Man, this hardware is whack. |
| 00:11:52 | ProtWork | They use a 34 pin connector to go up to the LCD, then a flexiboard in a friction lock connector as well. |
| 00:11:54 | GoRK | well the flash can be written w/o 12v... sboot won't unlock the flash without detecting 12v. if you could recompile sboot to not require 12v it would work w/o it |
| 00:11:59 | ProtWork | shakes his head. |
| 00:12:45 | Alehaz | GoRK: i think it's locked in the chip, not just by sboot |
| 00:13:05 | ProtWork | If it's NAND flash, you'll probably need 12V for it. |
| 00:13:13 | GoRK | hmm |
| 00:13:40 | SpaceCoaster | GoRK: the boot stuff is looking at an io pin waiting for 12V to come in. |
| 00:13:48 | GoRK | well.... all i want to do is build this audio circut |
| 00:13:50 | GoRK | :) |
| 00:14:01 | ProtWork | The 12V is Vpp for the flash most likely. |
| 00:14:10 | ProtWork | GoRK: For the UCB1200? |
| 00:14:13 | GoRK | i dont guess radio shack has lm386 amp ic's do they? |
| 00:14:19 | SpaceCoaster | GoRK: bet you didn't think that you'd get a PCB designed so quick. |
| 00:14:19 | GoRK | ProtWork: yeah |
| 00:14:30 | ProtWork | They'll have TL0xx op-amps iirc. |
| 00:14:48 | GoRK | SpaceCoaster: no way. Reading my post last night makes it sound like i already knew about it!!! |
| 00:14:49 | ProtWork | GoRK: To achieve what, per se? |
| 00:15:28 | SpaceCoaster | GoRK: have you talked to GPSFan about it? |
| 00:15:30 | GoRK | ProtWork: be able to play and record audio from the audio interfaces on the phone in software |
| 00:16:02 | ProtWork | GoRK: Gotcha. |
| 00:16:04 | GoRK | SpaceCoaster: no not yet.. i'd like to see a design modification to that circut to allow for line level audio out though (so people could use it for mp3 playback or whatever |
| 00:16:13 | ProtWork | GoRK: I could fab together a board for that and a blue-wire diagram in about a week. |
| 00:16:18 | GoRK | but of course i dont know how to do it |
| 00:16:24 | GoRK | ProtWork: http://erire.com/eshan/pcb/shanipcb3.pdf |
| 00:16:51 | Alehaz | That reminds me . . |
| 00:17:09 | ProtWork | GoRK: Heh. Check www.digikey.com |
| 00:17:13 | ProtWork | They'll ship you parts. |
| 00:17:21 | GoRK | ProtWork: that is lucent's board for it. If you can figure out a quick addition to give it a line level out jack, that'd be awesome |
| 00:17:41 | ProtWork | GoRK: For audio I/O? |
| 00:17:43 | ProtWork | Hell yes, I can. |
| 00:18:01 | ProtWork | I could also run a 100pc fab set of those boards. |
| 00:18:13 | GoRK | ProtWork: awesome. you are the man!!! |
| 00:18:39 | SpaceCoaster | ProtWork: coolest |
| 00:18:54 | Alehaz | ok, I'm mirroring that site just in case lucent gets annoyed |
| 00:19:14 | scanline | SpaceCoaster: I just dialed a phone number via ethernet |
| 00:19:16 | cory | lucent is too busy firing people |
| 00:19:21 | Alehaz | true |
| 00:19:34 | SpaceCoaster | Alehaz: I did that within 2 minutes :-) |
| 00:19:55 | Alehaz | SpaceCoaster: redundancy never hurts |
| 00:20:01 | GoRK | ProtWork: If I were you i'd make a couple changes... 1) stereo audio out connector (the audio is only mono, but still need to provide a stereo plug)... 2) maybe -- strong maybe -- mic in. 3) F1-8 are designed for an rj45 jack. why they didnt put the holes in the right place for a surface mount jack is beyond me.. all these smart people and no brains :) |
| 00:20:45 | SpaceCoaster | Alehaz: Is that a cruel pun on all those poor people being fired? |
| 00:21:01 | Alehaz | SpaceCoaster: oh gosh, didn't even notice that |
| 00:21:17 | Alehaz | I wonder if that's why this guy finally spilled the beans |
| 00:21:26 | Alehaz | maybe he doesn't work there anymore |
| 00:21:37 | GoRK | oh, and I was also thinking the audio jack (or jacks) could be surface moutn and go out through the left side of the phone (under the handset) to hold the board in place inside the phone... |
| 00:22:06 | jacques | has anyone figured out a way to make a tuxscreen semi-portable? |
| 00:22:11 | GoRK | ProtWork: http://erire.com/eshan/ on the instructions to wire it up |
| 00:22:13 | jacques | I mean running on batteries? |
| 00:22:24 | SpaceCoaster | jacques: RedBaron is the man! |
| 00:22:28 | GoRK | jacques: someone had it running on batteries.. i forgot who |
| 00:22:45 | jacques | SpaceCoaster, GoRK: thanks! that's encouraging! |
| 00:23:33 | ProtWork | GoRK: Oh, I'll only use the lucent design as a reference. I'll re-design it from the ground up. |
| 00:23:38 | Alehaz | some limey . . . |
| 00:23:42 | SpaceCoaster | jacques: if you take the boards out they should fix to the back of the screen, about the right size, then it's just batteries and a case. |
| 00:23:45 | ProtWork | Oh, this is for a phone jack connector. |
| 00:23:47 | ProtWork | That's bogus. |
| 00:23:56 | ProtWork | I was going to tie it right into the UCB1200. |
| 00:23:59 | jacques | SpaceCoaster: ooooh, mmmm. :-D |
| 00:24:03 | GoRK | ProtWork: ? |
| 00:24:13 | ProtWork | That way you won't have to sacrifice the phone jack. |
| 00:24:27 | GoRK | ProtWork: this device bridges the ucb1200 audio to the dsp.. you dont sacrifice the phone jack |
| 00:24:30 | Alehaz | ProtWork: no, the phone jack mod is ethernet |
| 00:24:39 | Alehaz | which is silly |
| 00:24:43 | ProtWork | Okay, so I'm a tad slow. :-) |
| 00:24:51 | SpaceCoaster | jacques: should just add about 1.5 inches as a lump on the back. |
| 00:24:55 | ProtWork | I'm just taking the platform to bits here, so gimme a break. :-) |
| 00:25:29 | jacques | damn, we have 5V PCMCIA mod now??? |
| 00:25:29 | SpaceCoaster | Alehaz: I didn't get the ethernet thing, what did they do? |
| 00:25:41 | jacques | starts to drool. |
| 00:25:45 | Alehaz | SpaceCoaster: hard wired a pcmcia nic to the aux port |
| 00:25:50 | Alehaz | which is silly |
| 00:25:55 | GoRK | ProtWork: just dont stray from their design too much... remember we dont exactly know the internals of the dsp that we plug into |
| 00:26:07 | GoRK | and it wires directly up to the dsp |
| 00:27:01 | GoRK | it shows radio shack sells lm386's.. i may go try to pick one up while i still can.. bbiab |
| 00:27:24 | GoRK | try to build this circut so that we can see how to use it |
| 00:28:03 | ProtWork | GoRK: *nod* But I know the UCB1200 chip. |
| 00:28:22 | ProtWork | is scanning the datasheet right now. |
| 00:29:35 | SpaceCoaster | Alehaz: woz the ethernet just so they could say, "see it has an ethernet socket" |
| 00:29:50 | Alehaz | SpaceCoaster: i imagine |
| 00:29:57 | Alehaz | they hooked it up to an rj11 |
| 00:30:03 | Alehaz | so it needed special cables |
| 00:30:05 | Alehaz | stupdi |
| 00:30:08 | Alehaz | well, silly, at least |
| 00:30:10 | ProtWork | They really could've made it easier to yank the telecom board. |
| 00:30:19 | BZFlag | sitaramgmt 2630/tcp Sitara Management |
| 00:30:21 | SpaceCoaster | indeed stupdi, very stupdi |
| 00:30:35 | scanline | BZFlag: Where's that from? |
| 00:30:55 | scanline | (and more importantly, is sitaramgmt anything worth avoiding collisions with?) |
| 00:30:56 | Alehaz | etc/services from the look of it |
| 00:31:11 | scanline | yeah, but which /etc/services |
| 00:31:23 | SpaceCoaster | Check first if sitara still exists :-) |
| 00:31:46 | BZFlag | http://www.iana.org/assignments/port-numbers |
| 00:31:58 | scanline | ah |
| 00:32:14 | Alehaz | am i the only person who watched tonight's episode of Enterprise and started chanting "Grup! Grup! Grup! Grup! Grup!" |
| 00:32:19 | Alehaz | ? |
| 00:32:26 | scanline | yeah, it appears that they're still around |
| 00:32:35 | SpaceCoaster | oops, they do VOIP |
| 00:32:39 | scanline | Alehaz: Damn, I missed it! Was it any good? |
| 00:32:43 | Alehaz | haha |
| 00:33:12 | Alehaz | scanline: Yeah, but bore striking resemblance to the episode of oldtrek where they visited the planet where there were no adults |
| 00:33:17 | scanline | So, what was the second choice for port numbers? How about 15263? |
| 00:33:17 | SpaceCoaster | BZFlag: I read the first block from the serial flash! |
| 00:33:35 | scanline | 15263 is kinda like IS2630 without the 0 |
| 00:34:04 | BZFlag | SpaceCoaster: woohoo! |
| 00:34:27 | BZFlag | I'd use 2630. |
| 00:34:56 | SpaceCoaster | BZFlag: it uses the SSP bus for transfers so reasonable throughput, gonna do an MTD driver next! |
| 00:35:03 | scanline | ok. But sitara might sue us!!! ;) |
| 00:35:43 | BZFlag | scanline: ports over 1024 are not "reserved" just "registered" |
| 00:35:50 | BZFlag | so they can not sue us. |
| 00:36:03 | SpaceCoaster | Didn't intel find that you can't copyright/patent/trademark numbers? |
| 00:36:03 | ProtWork | Okay, has anyone else looked at the UCB1200 datasheet? |
| 00:36:04 | ProtWork | around |
| 00:36:05 | ProtWork | <SpaceCoaster> oops, they do VOIP |
| 00:36:11 | ProtWork | Grr. |
| 00:36:12 | Alehaz | OK, got it mirrored |
| 00:36:13 | Alehaz | http://we.areb.org/erire.com/eshan/ |
| 00:36:18 | ProtWork | shoots his right mouse button. |
| 00:36:21 | Alehaz | in case the original should go down |
| 00:36:37 | SpaceCoaster | ProtWork: yes, I speak UCB1200, ish. |
| 00:36:45 | ProtWork | SpaceCoaster: Cool. |
| 00:37:15 | ProtWork | Just going to wire in the audio inputs to the UCB1200 this weekend. :-) |
| 00:37:16 | SpaceCoaster | ProtWork: my wife is fluent tho, she's here if you need her. |
| 00:37:40 | ProtWork | SpaceCoaster: Heheh. Did she work on the Shannon team? |
| 00:38:05 | SpaceCoaster | ProtWork: no but she does now :-) |
| 00:38:09 | ProtWork | *lol* |
| 00:38:47 | BZFlag | Alehaz: I recorded Enterprise, will watch later tonight. |
| 00:39:09 | Alehaz | BZFlag: *nod |
| 00:39:11 | Alehaz | * |
| 00:39:34 | SpaceCoaster | scanline: make it 2630 |
| 00:39:38 | BZFlag | used to work on silverplatter 416/tcp Silverplatter |
| 00:39:49 | scanline | SpaceCoaster: it is |
| 00:40:02 | sammy | Morning all ;-) |
| 00:40:13 | scanline | Hehe. PicoGUI uses 30450, i wonder if anything else does... |
| 00:41:20 | SpaceCoaster | morning sammy, weather good today? |
| 00:42:25 | ProtWork | oof. Putting this back together is going to be twice the chore of dismantling it. |
| 00:42:33 | SpaceCoaster | BZFlag: what do you think 0200080003000dff is on attached to a 256 block of nulls |
| 00:42:53 | SpaceCoaster | says the flash has 264 byte pages |
| 00:44:26 | BZFlag | http://www.sitaranetworks.com/ <- still around, or so it appears. |
| 00:45:58 | ProtWork | Oof. Are the batteries required for anything serious? |
| 00:46:46 | SpaceCoaster | mine are in a flashlight and the TuxScreen seems fine. |
| 00:46:58 | ProtWork | Okay |
| 00:47:01 | ProtWork | grabs the dikes |
| 00:47:30 | SpaceCoaster | ProtWork: ooh ere do they mind? |
| 00:48:40 | BZFlag | ProtWork: the batteries power basic phone (not computer) usage when power is out. |
| 00:49:20 | ProtWork | Hehehe. |
| 00:49:22 | ProtWork | That's funny. |
| 00:49:39 | ProtWork | Doesn't even know how to pull 48V phantom off the line. |
| 00:49:48 | ProtWork | SpaceCoaster: Of course not. They're steel and plastic. |
| 00:50:19 | SpaceCoaster | ProtWork: what are they, I am not from your world. |
| 00:50:36 | ProtWork | SpaceCoaster: Dikes = Diagonal cutters. |
| 00:50:55 | SpaceCoaster | ProtWork: on my world we call them... |
| 00:51:05 | SpaceCoaster | ProtWork: diagonal cutters |
| 00:51:20 | BZFlag | heh, I emailed sitara to ask em for permission. ;-) |
| 00:51:53 | scanline | hahaha |
| 00:54:40 | jacques | has anyone mirrored the http://erire.com/eshan/ site in case it goes away suddenly? |
| 00:55:26 | BZFlag | jacques: good idea. I'll do that on tuxscreen.net ... |
| 00:55:28 | Alehaz | jacques: I just finished a few minutes ago |
| 00:55:32 | Alehaz | http://we.areb.org/erire.com/eshan/ |
| 00:56:01 | jacques | ah, I shold read backlog more carefully :-D |
| 00:56:27 | jacques | any of you guys going to ALS? |
| 00:56:55 | jacques | BZFlag: I bet you could sell a ton of tuxscreens there (are ppl allowed to sell stuff there?) |
| 00:57:45 | BZFlag | jacques: I'd like to, but not going this year. |
| 00:58:12 | BZFlag | any Century folk going that want to show this awesome dev platform? ;-) |
| 01:00:01 | jacques | Century? |
| 01:00:28 | BZFlag | CenturySoftware, home of MicroWindows |
| 01:00:45 | BZFlag | CosmicPenguin and others work there. |
| 01:02:03 | jacques | cool |
| 01:02:10 | BZFlag | http://TuxScreen.net/download/erire.com/ in progress now. |
| 01:04:25 | Alehaz | ah, I get it. the shanip board connects the handset to the ucb . . . or something |
| 01:04:36 | Alehaz | handset ends up connected to the shanip board at least |
| 01:06:51 | Alehaz | er, no |
| 01:06:53 | Alehaz | hmm |
| 01:07:10 | Alehaz | they're using cat5 to connect to the shanip board. guess you've got to use something. |
| 01:08:09 | SpaceCoaster | I think the connector in the middle is bogus, just something that had around. |
| 01:08:42 | Alehaz | yeah |
| 01:09:50 | BZFlag | yeah, the shanip could be wired directly to both boards. that just means soldering if you have to swap anything around. |
| 01:10:41 | SpaceCoaster | Do what the pros do glue a connector to the board with a glue gun! |
| 01:11:13 | Alehaz | connector?! Fah! just solder bare wires to the board and hot glue over 'em :) |
| 01:12:01 | SpaceCoaster | solder, glue, bah, duct tape |
| 01:12:46 | BZFlag | mirror is complete, took some tweaking to not get icons/ and apache/ but I think the rest it there. |
| 01:12:47 | GoRK | back |
| 01:13:13 | GoRK | it's not cat5 it looks like some other multiconductor wire (not twisted) with an rj45 plug on it |
| 01:13:29 | GoRK | anyway i got all the stuff to build it now |
| 01:14:44 | GoRK | rat shack did have lm386's.. i was totally amaszszzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzszzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzssssssssssz |
| 01:15:18 | GoRK | agg sorry some metal piece just got stuck in between the s and z keys |
| 01:16:08 | GoRK | if you are a real pro, you can just coat your whole circut board, components and all with hot glue! |
| 01:16:11 | SpaceCoaster | a tidy workplace for a tidy mind (or is that tiny mind) |
| 01:16:40 | GoRK | my workplace is not tidy. there's too much work that gets done in it |
| 01:17:23 | SpaceCoaster | GoRK: just got my glue gun, haven't worked out all the possibilites, thanks for the advice. |
| 01:18:23 | GoRK | SpaceCoaster: the whole board will bounce if you are careful to make it round enough and use lo-melt |
| 01:23:08 | BZFlag | GoRK: so you are building one? wiki the RS parts list and costs if you could please? |
| 01:24:25 | SpaceCoaster | GoRK: will you need to twiddle a GPIO bit to enable your new toy? |
| 01:24:34 | GoRK | bzflag: the only thing i actually got from rat shack was the lm386 |
| 01:24:53 | GoRK | SpaceCoaster: dunno yet thats why im building one |
| 01:25:17 | GoRK | BZFlag: but i'll put up the part number of it :) |
| 01:37:39 | BZFlag | GoRK: sure. I think we need to start a ShanIP wiki page with additional info as we move along. |
| 01:37:53 | BZFlag | gott head home. cya |
| 02:40:00 | SpaceCoaster | GoRK: I wrote a null MTD driver, you're correct easy peasy |
| 02:45:14 | GoRK | this breadboard is a mess but it looks like it's built |
| 02:46:35 | scanline|PPC | GoRK: I'd still like to add some more features to this thing, but the new tuxphone is usable |
| 02:46:55 | Alehaz | we should figure out what other features we want before we go and print bulk boards |
| 02:51:56 | GoRK | audio out and maybe audio in |
| 02:52:40 | GoRK | is about all i can think of |
| 02:53:12 | Lenolium | *pling* |
| 02:53:32 | Lenolium | *pling* |
| 03:00:55 | GPSFan | GoRK: hi, just got back, & read what's been going on, you have been busy! |
| 03:03:13 | Alehaz | GoRK: shouldn't be tough with those mono jacks w/ built in switches |
| 03:07:15 | GoRK | GPSFan: did you see that url? erire.com/eshan/ ? |
| 03:07:33 | Alehaz | the only problem would be that every time I've bet i can install a jack in a plastic walled device and case mount it, I've had to countersink the hole in order to affix the nut |
| 03:07:51 | Alehaz | the ones the shack sells just aren't long enough |
| 03:09:09 | GPSFan | GoRK: yes, I'm reading all of that now, it looks like a cute little board, lousy integration into the Tux though. I'll have a schematic of the board soon, then we can go about moding it to our own needs. |
| 03:10:07 | GoRK | i dont know how you integrate it any better though... |
| 03:14:18 | GPSFan | GoRK: has anyone tried the PCMCIA mod yet? it looks like they just switch 5V or 3.3V to the power pins, and provide a level shifter for the POR. Still doesn't have hot plug capability. |
| 03:14:56 | GPSFan | GoRK: maybe I'm too critical, I'd be happy for any 5v mod that works at the moment. |
| 03:16:16 | Alehaz | GPSFan: should still be good enough for most folks |
| 03:16:33 | Alehaz | I have no motivation to do a 5v hack, having good stock of 3.3v cards |
| 03:17:05 | GPSFan | Alehaz: what sort of cards and are you selling them? |
| 03:18:15 | scanline|PPC | GoRK: I think I'm done with my first cut at client-server tuxphone. Should I email it to you, or the list? |
| 03:19:47 | GoRK | email it to me and ill put it in cvs |
| 03:20:04 | scanline|PPC | cool |
| 03:20:26 | scanline|PPC | It's fun dialing the speakerphone via 802.11b :) |
| 03:20:56 | Alehaz | GPSFan: sorry for the confusion, just personal stock |
| 03:21:05 | Alehaz | I've got at least as many as i have phones |
| 03:22:55 | GPSFan | Alehaz: ah got it, still, what kind of card(s) ar they? |
| 03:23:40 | Alehaz | GPSFan: one socket lp-e, one te100pc16r |
| 03:23:51 | Alehaz | had a cio10t, sold it to duck |
| 03:25:21 | GPSFan | Alehaz: thanks... |
| 03:25:54 | Alehaz | np |
| 03:30:50 | scanline | GoRK: tarball should be in your inbox |
| 03:30:54 | BZFlag | GPSFan: what are the addresses you want again? |
| 03:31:29 | BZFlag | 0xa0000000-20 or so? |
| 03:32:14 | BZFlag | GPSFan: I'm making a wiki page. |
| 03:36:54 | scanline | GoRK: I just tested the ring, it works. We don't have caller ID here though |
| 03:38:53 | GPSFan | BZFlag: |
| 03:39:22 | GPSFan | 0xA0000000,04,08,0c,10,14 |
| 03:39:38 | GPSFan | BZFlag: thanks |
| 03:40:34 | BZFlag | http://TuxScreen.net/wiki/view/Main/InfernoSODIMM |
| 03:42:35 | GPSFan | BZFlag: did you do the same with the 32M? |
| 03:45:22 | BZFlag | just did. |
| 03:46:11 | BZFlag | that's what you're after? I have a 64M too, but it does not even boot in the phone under either Inferno or Linux. |
| 03:46:38 | BZFlag | What was the deal with a12 ? that should get added to the wiki page. |
| 03:46:43 | GoRK | ok this time my breadboard will work :) heh |
| 03:48:15 | GPSFan | BZFlag: what you got is great, it tells me at once that Inferno can sense the presance of the 32M card and knows what to do. Once we decode those values we should be able to set blob up to run with 40M. |
| 03:49:35 | GPSFan | BZFlag: the deal with A12 is that it isn't used on the 32M Crucial card. SpaceCoaster read me off the part number from his card and I was able to find it on the Micron web site, and it doesn't use A12. |
| 03:49:57 | BZFlag | GPSFan: Inferno shows the memory config on the bottom of the boot screen. It shows 40M with the 32M sodimm installed. |
| 03:50:13 | BZFlag | what about the 64M ? |
| 03:50:30 | Lenolium | *plong* |
| 03:50:38 | BZFlag | The 64M board does not seem to have a Micron part number. There are 8 4LC8M8C2-5 (64MB 3.3V DRAM, 8Mx8 EDO 4K REF) chips on it. Crucial lists the |
| 03:50:38 | BZFlag | SODIMM as 16Meg by 32. |
| 03:50:54 | BZFlag | from http://TuxScreen.net/wiki/view/Main/SODIMM |
| 03:52:33 | GPSFan | BZFlag: 40 is better than 16, I'v got to spend some time looking at those #s you just put up and see if how they correspond with the blob #s. then we can move on to try to figure out why the 64M won't work. |
| 03:53:36 | BZFlag | pin 46 is A12? the 64M does label that. |
| 03:55:45 | GPSFan | BZFlag: yes and it is connected to nCS1 on the SA1100. I don't know what that is for. |
| 03:55:47 | BZFlag | yep, the working 32M has 46 listed as not connected, but the 64M has it as A12. |
| 03:59:38 | GoRK | ok im off to solder. this circut seems ok on the breadboard |
| 04:00:11 | BZFlag | GPSFan: any chance of getting a12 to pin 46? ;-) |
| 04:00:40 | BZFlag | 8 + 64 = 74M would be even nicer. ;-) |
| 04:00:53 | BZFlag | er 72, math would help too. |
| 04:04:44 | GPSFan | BZFlag: there may be a cut & jumper that we can do, but I still have to review the SA1100 specs and the refresh requirements of the 8M onboard and the 64M SODIMM to see if they will operate like the 8M onboard and the 32M SODIMM. It may be that the 8M onboard may not work, but we can get the 64M working by itself. Don't have to unsolder the 8M just disable it in blob. |
| 04:07:54 | BZFlag | yes, but if the a12 is required by the sodimm, we won't even get there without a board mod. |
| 04:09:05 | BZFlag | disabling the base ram on a board will be a bit tricky. blob is not setup to handle that now. and the kernel wants mmu off at boot. |
| 04:09:15 | GPSFan | BZFlag: Phillips tied pin 46 to nCS1 for some reason, maybe they had a software paging memory driver (ugg) but I can't see how it would work in a dynamic ram system. |
| 04:09:36 | BZFlag | so kernel mods would be required too such that the kernel would be in other that the default boot location. |
| 04:10:16 | BZFlag | that mean fwict that a different kernel would be required for a phone with the 64M installed if we can not leave the onboard enabled. |
| 04:11:21 | BZFlag | GPSFan: the 64M sodimms were not available in early '98 when the design was made. perhaps they thought they could hack up board that worked with the smaller pins by paging. |
| 04:11:27 | GPSFan | if we were able to get only the 64M SODIMM working by itself, there might be kernel mods, depending on how and when the MMU remaps the non-contiguous memory into a flat model. |
| 04:12:14 | BZFlag | yes, the initial code in the kernel (decompressor, etc) would need to be changed. perhaps it could be written to be relocatable, but it is not now. |
| 04:13:54 | GPSFan | when blob sizes memory, it makes a table of where real and aliased pages are, normally there holes after the 8M pieces, the MMU remaps these. if the physical memory didn't start till the 2nd page, maybe the MMU could handle that, maybe not. |
| 04:14:27 | GPSFan | since that all happens before the kernel loads. |
| 04:20:20 | GPSFan | BZFlag: just pup a schematic of the Lucent board up http://tuxscreen.net/download/schematics/shanip.pdf |
| 04:20:50 | GPSFan | s/pup/put up |
| 04:21:43 | BZFlag | GPSFan: nice. you really should decide on a license. ;-) |
| 04:22:38 | BZFlag | GPSFan: the mapping takes place first, but the kernel get's passed that info and does the MMU setup later. |
| 04:22:56 | BZFlag | Problem is the kernel will be running in the page that we want to disable. |
| 04:26:38 | GPSFan | You are right, but the guy is in Germany, I'v just started using the program, and haven't gotten around to sending him a check. I didn't know when the MMU was set up, if it is in the kernel, then the kernel must run where it wants, so the 8M needs to work. |
| 04:31:49 | as2 | Nite all. |
| 04:56:03 | BZFlag | is a member of the blob project. |
| 04:56:18 | BZFlag | Russ is a member too. |
| 04:59:49 | GoRK | man those connections are damned impossible to see in those photos |
| 04:59:55 | GoRK | brb |
| 05:00:56 | GPSFan | BZFlag: I've been looking at the memory config reg's and I think we can reset the values in memsetup.S and have the Tux work with the 40M. |
| 05:05:54 | BZFlag | sweet. we need to do some config detection first. have you seen the disassembled inferno boot code? |
| 05:06:13 | BZFlag | http://TuxScreen.net/download/inferno/inferno.lst |
| 05:07:48 | GPSFan | not yet, I'll look.. |
| 05:07:53 | BZFlag | It have register settings for 4 different configurations. I presume that the extra one is 8+16 which I don't have. |
| 05:07:59 | BZFlag | it also has detection code. |
| 05:10:51 | GPSFan | cool!, I wiwh I had looked at that earlier, oh well 20-20 hindsight... yes we can hack out part of that code to do detection, but if we want to try Linux in 40M JTAG a new blob with the new memsetup.S |
| 05:12:39 | GPSFan | decoding the bits, it looks like the Inferno config was set up for a bit fster access, to both DRAM & flash. |
| 05:12:43 | BZFlag | right. you have jtag, right? |
| 05:12:52 | GPSFan | s/fster/faster |
| 05:13:15 | GoRK | ok i think i have all the connections in place for the shanip mod |
| 05:13:28 | GPSFan | I don't nor do I have Linux on my tux... |
| 05:13:34 | BZFlag | GPSFan: yes, Inferno has a kernel that will run at 206mhz too, and linux crashes with that at present. |
| 05:13:53 | BZFlag | I see no reason why 221mhz should not work, the chip runs so cool. |
| 05:13:54 | GPSFan | probably because of the memory timings. |
| 05:14:16 | BZFlag | we should get the kernel cpu scaling working and then we can modify the clock on the fly. |
| 05:14:43 | BZFlag | yes, I think the memory timings need work under linux. |
| 05:15:26 | BZFlag | GPSFan: ah, no jtag makes blob development tough. |
| 05:16:04 | BZFlag | I wonder if olivier will release the modified inferno image too. |
| 05:16:17 | BZFlag | have you installed the new images? that can be done without jtag. |
| 05:16:19 | GPSFan | that would be too cool! I just put a modified memsetup.S in /downloads ...newmemsetup.S |
| 05:16:47 | BZFlag | GPSFan: you added detection? or just changed the registers? |
| 05:17:15 | GPSFan | should compile with BIGSHANNON, no just changed the regs. I'm not a programmer. |
| 05:17:52 | BZFlag | ahh. I won't get to it tonight. |
| 05:18:59 | GoRK | which modified inferno? the voip one? |
| 05:19:24 | BZFlag | I hope they will release the Inferno with ethernet drivers. |
| 05:19:34 | GPSFan | I'm off to Zzzz land lots to do tomorrow, maybe someone else can give it a shot. It should work with 8 x 8 or 8 x 32. |
| 05:19:36 | BZFlag | though VoIP would be nice too. ;-) |
| 05:21:52 | GoRK | ugh these stupid wires keep breaking off -- the ones the idiots who soldered up the gangly comm board put on |
| 05:22:16 | GoRK | the green ones are the weakest! |
| 05:28:11 | BZFlag | "What is it about the green ones?" |
| 05:29:08 | BZFlag | andersee: http://erire.com/eshan/ |
| 05:33:16 | andersee | looking |
| 05:44:08 | andersee | Holy cow! |
| 05:44:29 | BZFlag | nice huh? |
| 05:44:42 | BZFlag | remembers that he needs to mail the list... |
| 05:45:54 | andersee | Where did this come from? Do we need to mirror it before it disappears? |
| 05:47:52 | BZFlag | http://TuxScreen.net/download/erire.com/eshan/ |
| 05:47:54 | scanline | GoRK: I just committed a working version of pgtuxphone into CVS. No new features, but it now uses the tuxphone daemon. |
| 05:49:06 | BZFlag | olivier sent it to me. He did not explain whether Lucent actually released the info or not. Perhaps he assumes they no longer care. |
| 05:50:30 | andersee | A reasonable assumption I would guess. |
| 05:50:36 | andersee | Sweet stuff though |
| 05:51:04 | andersee | The voip hack seems pretty easy to do |
| 05:51:20 | andersee | I'd love to make these into intercoms throughout the house |
| 05:59:01 | GoRK | argh i think i have a shanip board done |
| 05:59:16 | GoRK | pretty easy... hardly :) blood, sweat and tears |
| 05:59:16 | scanline | argh? |
| 05:59:43 | scanline | so you burned yourself with the soldering iron, eh? |
| 06:00:19 | GoRK | again :( |
| 06:00:39 | GoRK | heh no i didnt burn myself but it was not easy |
| 06:00:52 | GoRK | it'd probably be a lot easier with a prefab board |
| 06:01:22 | scanline | yeah |
| 06:01:48 | scanline | do you know if it works yet? |
| 06:02:52 | GoRK | nope about to boot |
| 06:04:10 | GoRK | booting |
| 06:05:15 | GoRK | ok it booted up.. gonna test the dsp.. |
| 06:06:04 | GoRK | that worked |
| 06:08:42 | andersee | what worked? |
| 06:09:28 | GoRK | the dsp continues to function |
| 06:09:36 | GoRK | audio out doesnt seem to be doing anything |
| 06:18:29 | GoRK | audio in might be though.. hang on |
| 06:21:19 | GoRK | heh yeah i can record from the telephone device |
| 06:21:26 | GoRK | i hear myself |
| 06:23:49 | andersee | Cool! |
| 06:24:03 | andersee | So then voip is almost ready to go... :) |
| 06:24:30 | GoRK | im not so sure i wired it up to the right pins |
| 06:24:34 | GoRK | i cant tell what is what |
| 06:24:59 | GoRK | my cicrcut is garbage that's for sure |
| 06:25:06 | GoRK | anyway i have to go.. talk to you all later |
| 06:25:16 | scanline | bye |
| 06:37:45 | BZFlag | http://www.toshiba.com/taecdpd/products/features/MK5002mpl-Over.shtml <- HD for your tux? |
| 06:39:02 | BZFlag | 5G pcmcia drive. Works at 3.3v too. ;-) |
| 06:40:03 | scanline | $400, not as much as I might expect |
| 06:40:17 | scanline | how much are the microdrives usually? |
| 06:40:54 | BZFlag | they have dropped in price, checking... |
| 06:40:54 | collord | 1 GB microdrives are about $350 aren't they? |
| 06:41:39 | scanline | hopefully some competition between IBM and toshiba could lower the price of microdrives and this thing |
| 06:41:43 | BZFlag | found 340M for $150 |
| 06:42:03 | BZFlag | 1G for $300 |
| 06:42:18 | collord | kingston has a pcmcia drive too, sounds the same as tosh |
| 06:43:40 | BZFlag | found the toshiba for $350 on pricewatch. |
| 06:44:11 | collord | what's the speed differences between microdrive and flash? |
| 06:45:53 | BZFlag | collord: as they are different technology, it's tough to say. initial read/writes on HD will be slow as spinup may have to happen. |
| 06:46:20 | BZFlag | flash writes are fast if an erase does not have to happen and slow if it does. |
| 06:46:42 | BZFlag | I think flash will probably always win on reads though. |
| 06:48:29 | collord | i assume flash has an order of magnitude longer lifespan? |
| 06:52:59 | BZFlag | I doubt it. flash is limited to 100k writes. longer read lifespan probably. |
| 06:53:24 | BZFlag | So I guess it depends. |
| 06:53:46 | BZFlag | under light usage, flash will last longer, under heavy usage hd might last longer. |
| 06:56:47 | collord | huh. thanks and g'night |
| 07:04:27 | jacques | the toshibas are a lot more power-hungry than the microdrives |
| 07:05:00 | jacques | supposedly the 4-5GB microdrive will be out RSN |
| 07:34:21 | scanline | is away: zzzzzzz |
| 10:45:10 | SpaceCoaster | mallum: hi |
| 10:45:38 | mallum | SpaceCoaster: he spacecoaster :) |
| 10:45:47 | mallum | SpaceCoaster: that lucent mod page is pretty wierd |
| 10:45:57 | SpaceCoaster | hows that? |
| 10:46:41 | mallum | SpaceCoaster: wierd is a good way - just odd that people were hacking it to 5v and stuff a while back |
| 10:48:44 | SpaceCoaster | design flaws, glad that they tried to correct them, probably stuff like that caused the project to be canned, we should be grateful! |
| 10:49:28 | SpaceCoaster | mallum: I have read the extra 512k of flash memory! |
| 10:49:54 | mallum | SpaceCoaster: cool ! |
| 10:50:14 | mallum | SpaceCoaster: yes indeed, its a good piece of info. |
| 10:50:32 | SpaceCoaster | seems to read 512k in about 6 seconds, not bad for 1 bit at a time! |
| 10:50:47 | mallum | heh |
| 10:51:06 | mallum | thats fast enough to play mp3's :) |
| 10:51:12 | SpaceCoaster | gonna double the clock speed, see what we get, stay online |
| 10:52:31 | SpaceCoaster | don't get much of an mp3 in 512k, doh |
| 10:53:26 | mallum | just got your mail to the list |
| 10:53:38 | SpaceCoaster | mee too |
| 10:56:34 | SpaceCoaster | 5 seconds that time, lets double it again :-) |
| 11:00:43 | SpaceCoaster | mallum: 3 seconds now, that's the end of the line, maxed out the clock |
| 11:01:19 | mallum | :) |
| 11:01:55 | SpaceCoaster | i'd be happier if the data was the same :-( |
| 12:03:58 | hypaStyle | does anyone know how the daemon is progressing ? |
| 12:52:06 | GiuseppeOttaviano | Why are we stuck at kernel version 2.4.7? |
| 12:53:00 | SpaceCoaster | Because you haven't ported 2.4.12-ac93 to the TuxScreen yet :-) |
| 12:54:01 | GiuseppeOttaviano | Can I try? :-)) |
| 12:54:45 | GiuseppeOttaviano | Why aren't the shannon-specific patches sent to the arm linux patch system? We could use the straight rmk patch |
| 12:55:45 | SpaceCoaster | GiuseppeOttaviano: sorry couldn't resist |
| 12:55:57 | GiuseppeOttaviano | ??? |
| 12:56:57 | SpaceCoaster | GiuseppeOttaviano: I saw Russ with a 2.4.10 patch on his web site http://russ.dhs.org/files/tuxpatches/ |
| 12:57:03 | GPSFan | SpaceCoaster: 'mornin |
| 12:57:26 | GiuseppeOttaviano | I can't see it |
| 12:57:59 | SpaceCoaster | GPSFan: howdy, enquiring minds need to know "what are the critters"? |
| 12:58:12 | SpaceCoaster | GiuseppeOttaviano: http://russ.dhs.org/files/tuxpatches/patch-2.4.10-ac11-rmk1-np1-tux1.gz |
| 12:59:21 | GPSFan | Ah, My wife & I have a small farm in Colorado, we raise sheep, horses and some cows. ~120 sheep, 8 horses 4 cows 3 dogs 31/2 cats. |
| 13:00:27 | GiuseppeOttaviano | strange, it isn't listed in http://russ.dhs.org/files/tuxpatches/ |
| 13:01:10 | SpaceCoaster | GPSFan: wow, we have some chipmunks, max, trevor and adrian so I know how much work it can be :-) |
| 13:01:32 | SpaceCoaster | GiuseppeOttaviano: ssshh, don't tell anyone |
| 13:01:47 | GiuseppeOttaviano | ?? |
| 13:02:28 | GPSFan | SpaceCoaster: lst night BZ got the mem controller numbers, and the Inferno ones are quite different from the blob ones. they look reasonable though, and I put up a mod'ed memsetup.S. Is your Tux alive? |
| 13:02:39 | SpaceCoaster | GiuseppeOttaviano: the file is there, honest, but I think it's a work in progress, maybe |
| 13:03:04 | SpaceCoaster | GPSFan: indeedy, and transferring data from the serial flash at 100kbytes/sec! |
| 13:03:44 | GPSFan | SpaceCoaster: re: farm, yes it is a lot of work, but the NorthFork Valley is nearly self sufficient wrt locally produced food. |
| 13:04:09 | GPSFan | SpaceCoaster: I saw your posts re: the serial flash. Good work! |
| 13:05:13 | SpaceCoaster | thanks, I writing an MTD driver, so we should be able to put JFFS2 on it, writing will be a bit slow 10kbytes/sec. |
| 13:05:27 | SpaceCoaster | GPSFan: where is the file? |
| 13:05:47 | GPSFan | in /downloads it is called newmemsetup.S |
| 13:06:02 | GPSFan | I added a define for BIGSHANNON |
| 13:06:59 | GPSFan | SpaceCoaster: after looking at the bits, for the memcontroller, I think they should work with 8, 8 + 8 or 8 + 32.. |
| 13:07:05 | GiuseppeOttaviano | could someone post me the last wheaties.c file? Thanks |
| 13:08:49 | SpaceCoaster | GiuseppeOttaviano: sure is you email on the wiki |
| 13:08:58 | GPSFan | SpaceCoaster: as I told BZ last night I'm not a programmer, I have recompiled kernels though when I was working on a wireless Linux router. but that was just cookbook stuff. I don't normally have the tools nor the inclination to do extensive programming. |
| 13:09:42 | SpaceCoaster | GPSFan: now you tell me I just flashed my bootloader! |
| 13:10:11 | SpaceCoaster | GPSFan: well it boots the 8mb, time for the real test! |
| 13:10:33 | GPSFan | SpaceCoaster: !!! try again with the 32M SODIMM |
| 13:10:34 | GiuseppeOttaviano | SpaceCoaster: yes, my email is on wiki. |
| 13:10:41 | GiuseppeOttaviano | giuott@tin.it |
| 13:13:01 | SpaceCoaster | GPSFan: memory in ... |
| 13:13:50 | SpaceCoaster | GPSFan: it boots what should I look for. |
| 13:15:34 | GPSFan | Wow, BZ posted something a while ago about the memory map, I don't know how he got it. it is on the SODIMM wiki page. |
| 13:18:22 | SpaceCoaster | GiuseppeOttaviano: on its way |
| 13:19:38 | GiuseppeOttaviano | thanks! I didn't know it was in the cvs |
| 13:20:35 | SpaceCoaster | GiuseppeOttaviano: no problem, it isn't part of buildroot because we don't intend to make things backwards compatible, possibly |
| 13:21:04 | SpaceCoaster | GiuseppeOttaviano: I mean it doesn't build when you run buildroot |
| 13:21:12 | CosmicPenguin | morning |
| 13:21:29 | mallum | Hi CosmicPenguin |
| 13:21:45 | GPSFan | CosmicPenguin: 'mornin |
| 13:22:10 | CosmicPenguin | Any news? |
| 13:22:34 | GiuseppeOttaviano | I have some troubles with the file you sent, it's illegible |
| 13:22:48 | GPSFan | SpaceCoaster: is there a memory map command in either blob or Linux? |
| 13:23:05 | CosmicPenguin | SpaceCoaster: Just saw your post to the list - Is it time for a serial flash spec? |
| 13:23:57 | SpaceCoaster | CosmicPenguin: might be too late :-) |
| 13:24:53 | CosmicPenguin | SpaceCoaster: Did you already scratch one up? |
| 13:24:59 | GiuseppeOttaviano | ok, I downloaded it with cvs |
| 13:25:11 | SpaceCoaster | GPSFan: blob displays that memory map when it boots, your memsetup.S isn't any different. |
| 13:25:56 | SpaceCoaster | GiuseppeOttaviano: Cool, I sent you gzipped/html of the driver by accident, sorry |
| 13:26:08 | GiuseppeOttaviano | :) |
| 13:26:16 | GPSFan | SpaceCoaster: read out A0000000,4,8,c,10 and psot them. |
| 13:26:27 | GPSFan | s/psot/post |
| 13:26:47 | SpaceCoaster | CosmicPenguin: yes, I have a MTD driver, read only at the moment, 100kbytes/sec though! |
| 13:27:10 | CosmicPenguin | Nice..... |
| 13:27:28 | SpaceCoaster | GPSFan: I read them out, did you hear, what do you mean? |
| 13:27:29 | GiuseppeOttaviano | Rotfl, gzipped html |
| 13:27:41 | CosmicPenguin | So that would be a raw dump of the DSP code, eh? Any chance we can disassemble that? |
| 13:27:48 | SpaceCoaster | GiuseppeOttaviano: yeah |
| 13:27:58 | GiuseppeOttaviano | an the header is: |
| 13:28:00 | GiuseppeOttaviano | --------------020408090602060309010002 |
| 13:28:00 | GiuseppeOttaviano | Content-Type: application/msword; |
| 13:28:01 | GiuseppeOttaviano | name="wheaties.c" |
| 13:28:06 | GiuseppeOttaviano | msword :))) |
| 13:28:25 | GPSFan | SpaceCoaster: those are the memory controll registers I want to see if they got set to what I thought they should. |
| 13:28:42 | CosmicPenguin | GiuseppeOttaviano: You should see how many different things get encoded as msword on my kMail.... :) |
| 13:28:51 | SpaceCoaster | GiuseppeOttaviano: woa, not here, blame mozilla and webcvs for that. |
| 13:28:53 | GiuseppeOttaviano | :) |
| 13:28:56 | GiuseppeOttaviano | Ok I go |
| 13:28:57 | GiuseppeOttaviano | bye |
| 13:29:41 | SpaceCoaster | GPSFan: sorry, I'' concentrate, I can get at all the SA1100 registers from Linux, which registers do you want? |
| 13:30:38 | SpaceCoaster | are these the MDCAS?? registers |
| 13:30:39 | GPSFan | Memory locations 0xA0000000, etc are the memory control registers, just dump the long words with debug (?) |
| 13:30:47 | GPSFan | yes. |
| 13:31:19 | SpaceCoaster | GPSFan: MDCAS00 = 0xC71C703F |
| 13:31:39 | SpaceCoaster | MDCAS01 = 0xFFC71C71 |
| 13:31:51 | SpaceCoaster | MDCAS02 = 0xFFFFFFFF |
| 13:32:04 | SpaceCoaster | MDCAS20 = 0x00000000 |
| 13:32:18 | SpaceCoaster | MDCAS21 = 0x00000000 |
| 13:32:30 | SpaceCoaster | MDCAS22 = 0x00000000 |
| 13:32:36 | GPSFan | SpaceCoaster: that's enough, it looks like my mod's didn't get applied, look at newmemsetup.S in the BIGSHANNON define and see how badly I screwed up. |
| 13:33:38 | GPSFan | SpaceCoaster: maybe I should have just substituted values in the SHANNON define. |
| 13:37:14 | SpaceCoaster | I did it again with SHANNON removed and BIGSHANNON #defined, no difference |
| 13:38:19 | GPSFan | The MDCAS00 should read 0xcccccccf |
| 13:39:02 | SpaceCoaster | it still reads 0xC71C703F |
| 13:39:23 | SpaceCoaster | does the kernel still need changes? |
| 13:40:03 | GPSFan | Idon't think so, right now we are dealing with blob initializing and sizing the DRAM |
| 13:41:01 | SpaceCoaster | hmmm, correction, when I downloaded/flashed blob, I just did a boot, this doesn't reload blob, I should have done a reboot |
| 13:41:14 | GPSFan | Ah.. |
| 13:41:18 | SpaceCoaster | When I reboot now it just hangs in blob, |
| 13:41:54 | GPSFan | does anything come out the serial port? I mean to say is debug enabled in blob? |
| 13:53:20 | SpaceCoaster | GPSFan: back, any other tests/changes you'd like to try? |
| 13:54:09 | GPSFan | SpaceCoaster: I blew it, the mdcncfg: should be 0x00c8993f... |
| 13:55:04 | GPSFan | I was looking at the wrong string when I typed it in, that will really screw up the mem controller. SORRY, you will have to JTAG blob back in. |
| 13:55:51 | SpaceCoaster | just did, making the change... |
| 13:56:21 | GPSFan | It was after midnight when we did this... ;>P |
| 13:57:20 | GPSFan | that is mdcnfg: .long 0x00c8993f |
| 14:14:24 | SpaceCoaster | GPSFan: back again! |
| 14:14:30 | GPSFan | SpaceCoaster: any luck? |
| 14:15:02 | SpaceCoaster | GPSFan: fraid not. |
| 14:15:51 | SpaceCoaster | GPSFan: this jtag stuff is a blessing tho' :-) |
| 14:15:53 | GPSFan | :>( where are we now, did anything come out the serial port? |
| 14:16:23 | GPSFan | Yes JTAG is, blob debugging would be nearly impossible without it. |
| 14:16:28 | SpaceCoaster | GPSFan: nothing at all |
| 14:17:05 | GPSFan | I guess I might ask if anything is supposed to at that point in the boot. |
| 14:18:03 | SpaceCoaster | well it doesn't boot at all, reset and nothing happens, no output |
| 14:18:05 | GPSFan | in memsetup.S someone talks ablut blob writing to the serial port after setting up the DRAM but befor accessing it. |
| 14:18:53 | GPSFan | but then there is a comment from Erik that they don't do that anymore. |
| 14:19:50 | GPSFan | SpaceCoaster: I was hoping that just substuting the memory controller init would be enough. but there must be more somewhere. |
| 14:21:48 | SpaceCoaster | Do you want to talk through the reasoning for the values you chose? |
| 14:22:36 | GPSFan | SpaceCoaster: sure. BZ read out the reg's with 8, 8 + 8, and 8 + 32 Megs. he posted them on the wiki. |
| 14:23:16 | SpaceCoaster | found them |
| 14:24:45 | GPSFan | I looked at those values, and what blob set things up to and the 8 + 8 and 8 + 32 only differed in the mdcnfg part. |
| 14:25:31 | GPSFan | ie. the mdcas0,1,2 were the same in Inferno in all config's. |
| 14:25:47 | GPSFan | so the differences were in mdcnfg. |
| 14:26:04 | SpaceCoaster | The first 4 bytes are the MDCNFG, correct/ |
| 14:27:12 | GPSFan | yes but the order is reversed from what is in memsetup.S |
| 14:27:20 | SpaceCoaster | and these are little endian |
| 14:27:30 | GPSFan | ie first byte is 00 next 01... |
| 14:27:41 | SpaceCoaster | ok |
| 14:27:59 | GPSFan | I took the Inferno settings and decoded the bits as per the SA1100 docs. |
| 14:28:16 | SpaceCoaster | for mdcnfg? |
| 14:28:41 | GPSFan | 1f for both Inferno and blob 8M 3f for Inferno 32M |
| 14:29:04 | GPSFan | that bit sets the row/column organization. |
| 14:29:34 | GPSFan | section 10.2.1 of the SA1100 spec |
| 14:29:49 | GPSFan | bit 5 |
| 14:29:53 | SpaceCoaster | drac1? |
| 14:30:21 | GPSFan | DRAC(1:0) 10---10 row 11--12 row. |
| 14:30:58 | SpaceCoaster | ok so we are using 12 row now |
| 14:31:15 | GPSFan | for the 8 x 32 setup in inferno. |
| 14:31:24 | SpaceCoaster | yes |
| 14:31:29 | GPSFan | and that's what the 32M datasheet says. |
| 14:31:35 | SpaceCoaster | good! |
| 14:31:47 | GPSFan | But, it aslo seems to work for the lower 8M |
| 14:32:07 | GPSFan | since in inferno the lower 8 works with both setups. |
| 14:32:46 | SpaceCoaster | so why did all the other bits change? |
| 14:33:21 | GPSFan | the MDCAS0,1,2 set the timing waveform for refresh, and I think that is how Inferno can run at 206M whereas Linux cannot. the timings are faster. |
| 14:33:54 | GPSFan | same with the remaining bits in mdcnfg |
| 14:34:13 | SpaceCoaster | ahh, so I try it with just the 1 bit changed? |
| 14:35:00 | GPSFan | that is certainly a starting point. thanks to JTAG we can always recover. |
| 14:35:31 | SpaceCoaster | registers are |
| 14:35:35 | SpaceCoaster | mdcas0:.long0xc71c703f |
| 14:35:39 | SpaceCoaster | mdcas1:.long0xffc71c71 |
| 14:35:42 | SpaceCoaster | mdcas2:.long0xffffffff |
| 14:35:45 | SpaceCoaster | mdcnfg:.long0x0334b23f |
| 14:35:49 | SpaceCoaster | mcs0:.long0xfff84458 |
| 14:36:04 | SpaceCoaster | just 1f --> 3f in mdcnfg, ok? |
| 14:36:12 | GPSFan | that's for a working 8 + 8 blob |
| 14:36:34 | GPSFan | try it, you can't do it on the fly |
| 14:37:17 | SpaceCoaster | jtagging... |
| 14:37:54 | SpaceCoaster | ooh it booted! |
| 14:38:06 | GPSFan | what does the mem map say? |
| 14:39:26 | SpaceCoaster | MDCNFG = 0x0334B21F |
| 14:39:26 | SpaceCoaster | :-( |
| 14:40:00 | SpaceCoaster | changed the wrong file, only got 4 hours sleep! |
| 14:40:13 | GPSFan | I don't see anywhere in blob that changes that other than in memsetup.S |
| 14:41:01 | GPSFan | oh.. ;>P |
| 14:41:08 | CosmicPenguin | SpaceCoaster: Go to bed, man! |
| 14:41:14 | SpaceCoaster | no I had oldmemsetup.S and newmemsetup.S |
| 14:41:35 | SpaceCoaster | CosmicPenguin: I did, just not for long enough, |
| 14:41:49 | GPSFan | ther may be something in DRI(14:0) DRAM refresh control that needs to change as well |
| 14:41:50 | SpaceCoaster | woohoo, woohoo who needs sleep |
| 14:42:07 | SpaceCoaster | The memory map is different |
| 14:42:16 | SpaceCoaster | 0x00100000 @ 0xC0000000 (1 MB) |
| 14:42:19 | SpaceCoaster | 0x00100000 @ 0xC0200000 (1 MB) |
| 14:42:23 | SpaceCoaster | 0x00100000 @ 0xC0400000 (1 MB) |
| 14:42:26 | SpaceCoaster | 0x00100000 @ 0xC0600000 (1 MB) |
| 14:42:29 | SpaceCoaster | 0x00100000 @ 0xC8000000 (1 MB) |
| 14:42:33 | SpaceCoaster | 0x00100000 @ 0xC8200000 (1 MB) |
| 14:42:37 | SpaceCoaster | 0x00100000 @ 0xC8400000 (1 MB) |
| 14:42:41 | SpaceCoaster | 0x00100000 @ 0xC8600000 (1 MB) |
| 14:42:44 | SpaceCoaster | 0x00800000 @ 0xD0000000 (8 MB) |
| 14:42:49 | SpaceCoaster | 0x00800000 @ 0xD8000000 (8 MB) |
| 14:43:12 | SpaceCoaster | linux doesn'l load, stops when Uncompressing Linux.... |
| 14:43:13 | GPSFan | is that with a 32M |
| 14:43:19 | SpaceCoaster | yes |
| 14:44:04 | GPSFan | and mdcnfg ... is 3f? |
| 14:44:49 | SpaceCoaster | can't confirm it from linux but i believe so |
| 14:45:20 | GPSFan | ok, let's move on th the refresh control part of mdcnfg. |
| 14:45:56 | SpaceCoaster | ok |
| 14:47:13 | GPSFan | or just for grins just change all of mdcnfg to 0x00c8993f |
| 14:47:33 | GPSFan | leaving the other reg's as in orig. blob. |
| 14:47:43 | SpaceCoaster | ok, but it goes against the 1 bit at a time philosophy |
| 14:47:52 | GPSFan | yep, |
| 14:48:49 | SpaceCoaster | but at my error rate that could take forever |
| 14:49:55 | SpaceCoaster | hmmm, different, some 1mb mappings then.. |
| 14:50:01 | SpaceCoaster | 0x01000000 @ 0xD0000000 (16 MB) |
| 14:50:06 | SpaceCoaster | 0x01000000 @ 0xD8000000 (16 MB) |
| 14:50:52 | SpaceCoaster | still not booting linux, hang when uncompressing |
| 14:51:12 | GPSFan | but the 32 seems to be regognized. |
| 14:51:16 | GPSFan | as 32. |
| 14:51:33 | SpaceCoaster | GPSFan: 16+16=32, cool |
| 14:52:15 | SpaceCoaster | and we have 8 x 1mb mappings |
| 14:52:50 | GPSFan | that may be how blob interprets it. so far so good. |
| 14:53:07 | SpaceCoaster | what next maestro |
| 14:53:31 | GPSFan | I would like it better if we had the 2 4M mappings. |
| 14:54:26 | SpaceCoaster | I guess that the kernel is being decompressed into C0000000 |
| 14:55:16 | GPSFan | the c0000000 should be contiguous 4M it looks like it is fragmented 1m spaces.] |
| 14:55:36 | SpaceCoaster | GPSFan: the mappings are contiguous... |
| 14:55:41 | SpaceCoaster | 0x00100000 @ 0xC0000000 (1 MB) |
| 14:55:45 | SpaceCoaster | 0x00100000 @ 0xC0400000 (1 MB) |
| 14:55:49 | SpaceCoaster | 0x00100000 @ 0xC0800000 (1 MB) |
| 14:55:54 | SpaceCoaster | 0x00100000 @ 0xC0C00000 (1 MB) |
| 14:55:58 | SpaceCoaster | 0x00100000 @ 0xC8000000 (1 MB) |
| 14:56:02 | SpaceCoaster | 0x00100000 @ 0xC8400000 (1 MB) |
| 14:56:05 | SpaceCoaster | 0x00100000 @ 0xC8800000 (1 MB) |
| 14:56:08 | SpaceCoaster | 0x00100000 @ 0xC8C00000 (1 MB) |
| 14:56:10 | GPSFan | yeah, your right, but somehow Linux doesn't like it. |
| 14:56:47 | SpaceCoaster | No I'm not right, 1mb == 0x100000, we have gaps |
| 14:57:02 | GPSFan | BZ said that the kernel is hard coded for where it is to run. |
| 14:57:32 | SpaceCoaster | Sounds better than trying to relocate it as we load it |
| 14:57:58 | CosmicPenguin | SpaceCoaster: For somebody who had 4 hours of sleep, you're doing an awful lot of typing |
| 14:58:18 | GPSFan | if it won't fit into 1m then crash, with a 4m block at c0000000 it is ok. |
| 14:58:53 | SpaceCoaster | GPSFan: do you agree that the memory is NOT contiguous |
| 14:59:46 | GPSFan | that's what the map says 100000 # 400000 |
| 15:00:49 | SpaceCoaster | The kernel has a . = 0xC0008000; |
| 15:00:50 | SpaceCoaster | as part of it's load map |
| 15:01:22 | SpaceCoaster | CosmicPenguin: I wan't my memory to work! |
| 15:01:34 | SpaceCoaster | s/wan't/want/ |
| 15:01:53 | CosmicPenguin | SpaceCoaster: I know, I know.... just joshing with you.... :) |
| 15:02:41 | GPSFan | SpaceCoaster: I've got to run, the wife is in need of help outside. I'll be back later... take a look at the SA1100 docs and try changing the CAS waveforms. |
| 15:02:55 | SpaceCoaster | ooo, ok |
| 15:05:15 | pattieja | is away: I'm busy |
| 15:05:35 | pattieja | is back (gone 00:00:02) |
| 15:27:21 | Lenolium | *pling* |
| 15:27:37 | Lenolium | *pling* |
| 15:28:07 | Lenolium | *pling* |
| 15:28:29 | Lenolium | *pling* |
| 15:28:51 | Lenolium | *pling* |
| 15:29:37 | Lenolium | *pling* |
| 15:30:15 | Lenolium | *plong* |
| 16:07:05 | CosmicPenguin | ipaq mailing list is talking us up today.... :) |
| 16:08:25 | AlHazWork | heh |
| 16:43:10 | AlHazWork | there's a bunch of 340 meg microdrives w/ pcmcia adapters on ebay for $120 |
| 17:05:44 | CosmicPenguin | Check this out: http://freshmeat.net/projects/voiceapp/ |
| 17:06:03 | CosmicPenguin | That would be some cool eye candy to have on the tuxscreen while you are on the phone... you could watch your voice go up and down as you talked |
| 17:20:50 | SpaceCoaster | GPSFan_: ring, ring, ring, ring |
| 17:21:12 | GPSFan_ | Hello! just got back from eating |
| 17:22:03 | SpaceCoaster | no real progress but some more understanding and more coffee |
| 17:22:21 | GPSFan | Ah coffee! that should help! |
| 17:22:30 | GPSFan | what have you figured out? |
| 17:22:48 | SpaceCoaster | GPSFan: SilverBullet is on the case or is that CAS too. |
| 17:23:18 | SpaceCoaster | well we've figured out that our sodimms have 2 chips per side! |
| 17:23:20 | GPSFan | ;>) good to have more folks |
| 17:23:50 | GPSFan | yes, the 8M and the 32M each have 2 banks on board, |
| 17:24:07 | GPSFan | the 8 has 2 banks of 4 the 32 2 banks of 16. |
| 17:24:17 | SpaceCoaster | and that makes them 1Mx16 and 4Mx16 with 2 banks each |
| 17:24:41 | SpaceCoaster | so the allowed configurations are 12x8 and 12x10 |
| 17:24:58 | GPSFan | the 8m SODIMMs chips are 1M x 16 just like on the Tux CPU board. |
| 17:25:38 | SpaceCoaster | yeah, so we have 12 rows and either 8 or 10 columns |
| 17:26:00 | GPSFan | from table 53 in the SA1100 spec. |
| 17:26:34 | SpaceCoaster | we have table 10-5 in the developers manual |
| 17:27:07 | GPSFan | oops I'm looking at the older DEC manual. |
| 17:27:46 | SilverBullet | GPSFan: Could you look at the following http://www.linuxhq.com/kernel/v2.4/patch/patch-2.4.9/linux/arch/arm/mach-sa1100/cpu-sa1100.c.html |
| 17:27:57 | GPSFan | section 10.3.1 DRAM Overview. |
| 17:28:05 | SpaceCoaster | yes |
| 17:29:21 | SilverBullet | GPSFan: there is an arra of sa1100_dram_settings, which I thought you might be able to clarify. |
| 17:29:33 | GPSFan | SilverBullet: ooo! the timings are related to clock speed. |
| 17:29:49 | SilverBullet | GPSFan: yes |
| 17:30:39 | SpaceCoaster | Do you know our clock speed? |
| 17:31:49 | GPSFan | Two Inferno kernels allowed 133 (?) & 206MHz a real early version allowed 103 |
| 17:32:06 | GPSFan | I don't know what Linux was set to. |
| 17:32:20 | GPSFan | But it is probably early in blob |
| 17:33:01 | SpaceCoaster | we are 191.7 mhz |
| 17:34:24 | SpaceCoaster | shall I plug in those values chaps? |
| 17:34:32 | GPSFan | I've got to run off again... be back later |
| 18:10:00 | pattieja | is away: rebooting |
| 18:49:10 | SpaceCoaster | BZFlag: do we have the sa1100-techref.pdf on the wiki, if not it can be found at http://www.mit.edu/afs/sipb/contrib/doc/specs/ic/cpu/arm/sa1100-techref.pdf |
| 18:54:34 | SpaceCoaster | BZFlag: The directory looks good for all things ARM |
| 18:55:24 | BZFlag | SpaceCoaster: add it to the Links wiki page. |
| 18:55:51 | SpaceCoaster | ok |
| 19:23:33 | BZFlag | to work |
| 19:29:30 | mallum | Hmmm, Im getting a wierd error whilst trying to build uCLibc on ARM |
| 19:30:04 | mallum | lots of these -> malloc.o(.text+0x30): undefined reference to `getpagesize' |
| 19:30:13 | mallum | any idea what that could be ? |
| 20:35:26 | collord | prpplague: any luck? |
| 20:38:11 | AlHazWork | So, anyway, Duck tells me he showed the 5v pcmcia page to his friend Curt, who's an EE, and has some tuxscreens |
| 20:38:41 | AlHazWork | and he says that if you follow the instructions & diagrams, you'll smoke your phone. disgrams are inconsistent with themselves, don't match the pictures, etc |
| 20:38:53 | AlHazWork | so he's working on coming up with something similar that might actually work |
| 20:45:03 | BZFlag | AlHazWork: please have him mail any questions to olivier. He will probably be glad to update the web page if there are inconsistancies. |
| 21:09:06 | Lenolium | *pling* |
| 21:09:31 | Lenolium | So, has anyone tried the mods yet? |
| 21:27:37 | Lenolium | SpaceCoaster: Anything new and interesting happening with the DSP driver? |
| 21:32:33 | Lenolium | The pictures on the site don't do all that much good. :) I hope GPSFan can draw us up some better looking schematics of what needs to go on here. |
| 22:02:29 | BZFlag | Lenolium: GPSFan did make up schematics for the shanip board. have you seen that? |
| 22:04:34 | BZFlag | http://TuxScreen.net/download/schematics/shanip.pdf |
| 22:05:49 | Lenolium | Has anyone talked about getting someone to build a run of 100-200 of those boards? |
| 22:06:16 | ProtWork | What, the audio I/O boards? |
| 22:06:37 | Lenolium | I'm looking at my board, and most of those wires are going onto places that have nice spots for soldering. |
| 22:06:40 | BZFlag | Lenolium: not yet. I still have 800 phones, I'd be interested in doing a run. I'd like to do more testing first though and see if it does everything we need. |
| 22:07:14 | BZFlag | I'd like to determine what's involved in turning the shannon into a h323 gateway. |
| 22:07:21 | Lenolium | ProtWork: No, these shanip boards let the DSP in/out of the UCB1200 talk to the line interface of the phone. So speakerphone and handset would work properly. |
| 22:07:29 | BZFlag | then wire aux to a set of relays as well. |
| 22:07:39 | ProtWork | BZFlag: I've got the facilities and resources to do a run. |
| 22:08:09 | BZFlag | so that aux can get routed to the dsp and then use the h323 out the ethernet. this much should be easy. |
| 22:08:09 | Lenolium | ProtWork: Take a look at the schematic that he pasted for me, what do you think a run of 200 would cost? |
| 22:08:35 | ProtWork | Lenolium: Give me a moment and I'll tell you. What kind of headers are there? |
| 22:08:42 | BZFlag | the next thing would be to have an incoming h323 call be able to ring the other phones connected to the aux jack. |
| 22:09:15 | BZFlag | that requires that something be able to generate ring voltage. |
| 22:09:18 | Lenolium | ProtWork: Two eight soldering position connectors. :) |
| 22:09:36 | Lenolium | BZFlag: Ain't ring voltage around 48V? |
| 22:10:53 | Lenolium | ProtWork: We need to run eight wires from various locations on the bottom of the board, through this little board, and onto a pretty close section of wire on the top of the board. |
| 22:11:09 | Lenolium | can almost tell where to put the wires onto the top of the board, except for the grey one. |
| 22:13:01 | ProtWork | Lenolium: So you'd blue-wire the board to the Shannon PCB? |
| 22:13:04 | Lenolium | and where does that brown wire from the PCMCIA slot go? |
| 22:13:20 | Lenolium | ProtWork: Yup. |
| 22:13:24 | ProtWork | God, whoever drew this schematic schlocked it together. |
| 22:13:50 | Lenolium | http://www.tuxscreen.net/download/erire.com/eshan/emain66b.jpg <-- There is the idea |
| 22:14:18 | Lenolium | except, that rj-45 port in the middle of it all seems a bit too odd for me. :) |
| 22:14:42 | ProtWork | *nod* That was probably b/c it's what they had lying around. |
| 22:15:00 | ProtWork | Where'd this schematic come from? |
| 22:18:10 | BZFlag | Lenolium: iirc yes, like 48v is required for US ring. |
| 22:18:53 | Lenolium | ProtWork: Feel lucky you didn't get this one: http://www.tuxscreen.net/download/erire.com/eshan/fivevolt.jpg |
| 22:19:35 | GPSFan | BZFlag: did you follow SpaceCoaster's & my exploits with the 32M SODIMM this morning? |
| 22:20:15 | BZFlag | GPSFan: no, what was the outcome? |
| 22:21:06 | ProtWork | Lenolium: I can understand that one. |
| 22:21:27 | ProtWork | Lenolium: This OrCAD Schematic is nonsense, though. |
| 22:21:48 | ProtWork | Unless the LM386 is in a *really* funky configuration. |
| 22:22:42 | GPSFan | BZFlag: well we got blob to recognize the 32M SODIMM, but in doing so the onboard 8M appears as 8 1M chunks as opposed to 2 4M chunks. Linux has problems loading into c0000000 in only 1M of memory. Inferno looks like it sets up the MMU in sboot, so the OS has a nice contiguous space when it starts. |
| 22:23:47 | Lenolium | ProtWork: Then try this one: http://www.tuxscreen.net/download/erire.com/eshan/pcb/shanipcb3.pdf |
| 22:23:53 | bhima | hrm. so my tuxscreen seems confused. it worked at first, now it keeps demanding that it dials up for an update, and after three or four reboots, it flashes a red error screen. |
| 22:24:09 | bhima | (still running linux; just got it today) |
| 22:24:12 | Lenolium | bhima: You unlocked the flash, right? |
| 22:24:14 | GPSFan | BTW I drew the Shannonip scjhematic from the board jpg and it was quick, just to get something that people could look at and try to understand what was going on. |
| 22:24:40 | GPSFan | it is also rev .5 which is far from a finished product. |
| 22:24:52 | bhima | lenolium: naw. only just started using it and playing with it. it was working happily at first. |
| 22:25:05 | Lenolium | bhima: Did you pay to have it flash unlocked? |
| 22:25:10 | bhima | lenol: no. |
| 22:25:17 | Lenolium | Hrm... |
| 22:25:25 | ProtWork | Lenolium: I can understand this one too - don't know what values they have, but between this board layout and the foil that's on this, I can wire one up easily. |
| 22:25:27 | Lenolium | I've only noticed that it would do that when the flash was unlocked. |
| 22:25:47 | Lenolium | ProtWork: That was the original. |
| 22:25:48 | bhima | there were two things. a power failure, and I did open it up, examine the boards, and close it up. I was grounded, no static, was careful, etc. |
| 22:25:58 | ProtWork | Lenolium: I could throw this together in a bit. |
| 22:26:08 | bhima | it did manage to dial their 800 number and do PPP connection stuff and so on. |
| 22:26:09 | Lenolium | ProtWork: Estimated cost for a run of 200? |
| 22:26:20 | ProtWork | Lenolium: Give me a minute to work up a BOM, and amoritize NRE. |
| 22:26:21 | bhima | so I'm pretty sure it is ok in terms of hardware. |
| 22:26:39 | bhima | I mean, displaying LCD graphics, working on the touch screen, keyboard, and modem dialling should be a good self test :) |
| 22:27:07 | Lenolium | bhima: Hrm... that's weird, I'm running one phone here in normal mode, and it works just fine, my other phone did that a lot before I got a null modem cable and linux installed. |
| 22:27:11 | bhima | unrelated question: is there any estimate for how long there will still be tuxscreens for sale? |
| 22:27:12 | ProtWork | Lenolium: I'd also re-do the product into a surface mount package so I could use cheaper parts. |
| 22:27:31 | bhima | lenol: it did what mine's doing? |
| 22:28:11 | Lenolium | ProtWork: Uhm... okay. |
| 22:28:19 | Lenolium | bhima: Yup. |
| 22:28:39 | bhima | leno: ahh, ok. yeah, I'll just get linux on it, all will be happy, I hope :) |
| 22:28:46 | ProtWork | Lenolium: It's easier to lay out, and makes the board MUCH cheaper. Would there be any incentive to have mini-jacks for the Audio I/O instead of thse point to point soldering connections? |
| 22:28:47 | bhima | is installing it in his car. |
| 22:29:30 | bhima | gonna build one in to a DIN-sized outlet, and link via ethernet to a computer in the trunk. |
| 22:29:56 | Lenolium | ProtWork: Hrm... no, we need to do the wiring, the whole point is to get it to do VoIP and link into the existing handset/speakerphone. |
| 22:30:11 | ProtWork | Lenolium: Okay. |
| 22:31:13 | Lenolium | ProtWork: Like in this photo http://www.tuxscreen.net/download/erire.com/eshan/emain66f.jpg |
| 22:31:46 | BZFlag | bhima: on boot, do ESC and then 2 |
| 22:32:20 | bhima | bzf: ok. will try that when I get home. |
| 22:33:00 | bhima | realizes who bzflag is and waves. thanks for the really cool toy! :) |
| 22:33:11 | ProtWork | Lenolium: kk. I got it. I'd like to have discrete audio I/O's on this thing, so I might add it on for the other audio port. |
| 22:33:45 | bhima | I have at least threee friends who are gonna buy these. very nifty gadgets. |
| 22:34:32 | BZFlag | bhima: watch the web site for availability. I've selling about 40 a week now, so 800 should go in 20 weeks or 4 months at this rate. |
| 22:35:33 | bhima | bzf: ok. thanks. I have a few project ideas that could use more than one. some ideas even more reasonable that the refrigerator browser :) |
| 22:37:23 | BZFlag | ProtWork: I expect there will be more things we'd like to add. If you are planning on shipping populated boards, it might make sense to have connectors on them and include cables with the connector on one end and the user would solder the other end to the boards. |
| 22:38:28 | BZFlag | you saw my comments about wanting dial tone, ring voltage generation, and relays for the aux port? |
| 22:39:07 | ProtWork | BZFlag: I figured so. I won't produce these boards until we have a consensus on what we'd want to do. |
| 22:39:15 | BZFlag | agreed. |
| 22:39:19 | ProtWork | BZFlag: Yes, but that might not work. You're talking the aux port on the UCB1200. |
| 22:39:21 | ProtWork | right? |
| 22:39:40 | BZFlag | ProtWork: I was referring to the aux rj11 on the back. |
| 22:39:53 | ProtWork | Ah, I see. I'll have to dive in and check the schematics on this. |
| 22:40:16 | BZFlag | as shipped the line and aux rj11 are just wired together. |
| 22:40:48 | BZFlag | but it would be nice to be able to use the phone as a h323 gateway for other phones in the house. |
| 22:41:48 | ProtWork | Very true - but you'd have to do A/D and D/A conversion on the aux jack - and there's only one phone input on the UCB1200. THe other ports on it don't have SB suppression and/or enough resolution for modem communications. |
| 22:42:24 | ProtWork | Plus, you'd need an external DAA. |
| 22:42:26 | BZFlag | hence, the desire for relays to switch over to the aux port. ;-) |
| 22:42:41 | ProtWork | Okay - you're not talking about simultaneous communications on both lines, right? |
| 22:42:47 | BZFlag | no. |
| 22:43:05 | BZFlag | just on ethernet as well as ( aux or line ) |
| 22:43:23 | ProtWork | Okay - I'm fuzzy on what h.323 is. :-) If I have a spec, I can design it. I'm an audio guy, not a telecom guy. :-) |
| 22:44:18 | BZFlag | ok, phone is connected from "line" to wall, to internet via pcmcia ethernet, and to the rest of the phones in the house via the "aux" rj11 |
| 22:44:46 | BZFlag | incoming pots call rings all phones, ie: line and aux are conencted. |
| 22:44:51 | ProtWork | Okay. So the phone isn't connected to the DSP - just the aux output, right? |
| 22:44:53 | ProtWork | Simpe. |
| 22:44:55 | ProtWork | er, simple. |
| 22:45:06 | ProtWork | They don't have that already? I thought I saw a relay on the DAA. |
| 22:45:09 | BZFlag | incoming internet call disconnects phone from "line" and rings phones off "aux" |
| 22:45:20 | ProtWork | Ah, that's where the gateway comes in. |
| 22:45:21 | ProtWork | Gotcha. |
| 22:45:32 | bhima | thinks about the number of phones in his house, and the number of computers, and imagines the opposite. lots of computers via ethernet to the single tuxscreen on the phone. |
| 22:46:15 | BZFlag | yes, reverse should work, h323 call from ethernet should be able to pick up "line" and dial outside number. ;-) |
| 22:47:22 | BZFlag | so the relay needs to allow at least two states, connected to "line" and "aux", or just connected to "aux" |
| 22:47:50 | BZFlag | we phone power off state would be aux and line are connected. |
| 22:47:58 | BZFlag | s/we/where/ |
| 22:48:19 | bhima | cool. I have ethernet in more rooms than POTS, so, yeah, that would be easier. |
| 22:48:33 | BZFlag | it's not a problem to leave "aux" connected when phone uses "line" to dial out. |
| 22:48:56 | bhima | hmm. I should go. I'm gonna install linux soon. bzflag: thanks again for filling your house with insane numbers of cool toys for us :) |
| 22:48:58 | bhima | & |
| 22:49:12 | BZFlag | bhima: no prob! happy hacking |
| 22:49:29 | ProtWork | Lenolium: About $30 for the board. |
| 22:50:01 | BZFlag | ProtWork: populated? including the relay enhancements or just as the online version is now? |
| 22:50:24 | ProtWork | BZFlag: Just the way it is now. If I do the relay enhancements, I'd probably re-design the DAA for the UCB1200 and then quote you. |
| 22:50:42 | ProtWork | This is board only, populated. No cables, no extras. If it was as-is with cables, I'd say about $35-$40. |
| 22:51:04 | ProtWork | Re-designed, I'd add about 20% of the cost to that. |
| 22:51:28 | ProtWork | But, I can't give you a hard quote until I have a full schematic and a BOM. |
| 22:51:44 | BZFlag | agreed. |
| 22:52:20 | ProtWork | Would you want 2-line support as well? And also, do you want to use the audio I/O on the UCB1200 for stuff like MP3 playback, etc...? |
| 22:52:31 | ProtWork | is still familiarizing himself with Shannon, so please be patient. |
| 22:52:56 | ProtWork | Oh, this also assumes I build and sell 200 boards. |
| 22:53:07 | ProtWork | If I only do 100 boards, the cost nearly doubles. |
| 22:53:11 | BZFlag | ProtWork: hmm... one line should suffice IMHO |
| 22:53:20 | ProtWork | (the PCB is costs more than the components put together) |
| 22:53:42 | BZFlag | the designed implementation should allow mp3 playback out speaker port no? |
| 22:54:08 | ProtWork | BZFlag: Dunno, need to check the Shannon telecom schematics. I've got the UCB1200 datasheet - do they have the speaker wired on the audio out of the UCB1200? |
| 22:54:56 | BZFlag | not now. cables to the shanip is will be the only thing wired to the ucb1200 audio in and out. |
| 22:55:28 | BZFlag | they might be wired to the phone in and out on the ucb, in which case the audio in/out might be extra wiring. |
| 22:56:02 | BZFlag | in other words the ucb audio is not wired to anything now. afaik. |
| 23:04:45 | ProtWork | Okay. |
| 23:07:02 | BZFlag | ProtWork: thanx for the efforts! |
| 23:07:11 | BZFlag | going home. |
| 23:08:01 | Lenolium | ProtWork: $30 per board? Wow. So someone would have to lay down 6,000 to build these things? |
| 23:08:31 | ProtWork | Lenolium: Yep. |
| 23:08:37 | Lenolium | Eep. |
| 23:09:11 | SpaceCoaster | GPSFan: seen the SA-1100 technical ref it has more memory details http://www.mit.edu/afs/sipb/contrib/doc/specs/ic/cpu/arm/sa1100-techref.pdf |
| 23:34:10 | SpaceCoaster | hi Alehaz |
| 23:34:54 | GPSFan | SpaceCoaster: hi, just saw your post , was off @ dinner, section 10 is just about the same detail as what I have. did you see my eariler post to BZ? |
| 23:37:18 | SpaceCoaster | GPSFan: yes I did |
| 23:37:49 | GPSFan | SpaceCoaster: I think what we saw with the 1M chunkswas consistant with looking at the 8M memories using the 32M configuration. Linux won't boot because the kernel won't fit into 1M and Inferno works because sboot sets the MMU before starting the OS. |
| 23:38:59 | SpaceCoaster | GPSFan: yes I agree, do you think some of the row/column waveforms could be changed to make the 8m memories behave like 4+4 as before? |
| 23:40:34 | SpaceCoaster | I think the clue to it all is in 10.2.2 but it's inpenetrable |
| 23:40:56 | GPSFan | SpaceCoaster: I dont think the timing registers would have that effect. the * + * used 10 bit addressing, the 8 + 32 uses 12 bit, those 2 bits shifted from the CAS to the RAS address, that works for the 32M, but results in the 8M becomming fragmented. As we saw. |
| 23:41:30 | SpaceCoaster | so is the solution to alter the way that they are mapped in blob? |
| 23:41:33 | Alehaz | hi |
| 23:41:42 | SpaceCoaster | Alehaz: yo |
| 23:42:27 | GPSFan | SpaceCoaster: if the MMU were setup in blob, or there was a preboot loader that would work in <1m and setup the MMU, then I think it would work. |
| 23:43:29 | GPSFan | SpaceCoaster: I am a kernel & Linux newbie, I don't know enough of the internal structure to rationally comment on what is possible. |
| 23:43:30 | bhima | hrm. esc-2 doesn't seem to get it out of the intense desire to connect to a remote server or reboot when I cancel. |
| 23:43:59 | Lenolium | bhima: Did you let it attempt to connect to the server? |
| 23:44:13 | bhima | lenolium: I did before. I'll try it again after esc-2. |
| 23:46:49 | bhima | access unauthorized by the server. "Status: Try to connect to the server again." |
| 23:47:17 | SpaceCoaster | GPSFan: I don't know when the MMU is configured, but the memory is not contigous wit the 8+8, its 4x4 chunks on 8m boundaries, so something handles it. |
| 23:47:47 | SpaceCoaster | oops make that 8m << 4 boundaries |
| 23:49:12 | SpaceCoaster | I tried to make the kernel execute at D0000000 where we had 16M of contig memory it loaded but trapped out, I think the memory map is pretty fixed. |
| 23:49:12 | GPSFan | SpaceCoaster: I read through the partial dis-assembly of sboot, and saw several memory configurations, for 9, 10, 11, & 12 bit modes. Also saw where it tested memory & the mmu setup. Yes, the 8 + 8 is non contiguous (4M chunks) but the kernel will fit and run in 4m where it apparently won't in 1M. |
| 23:50:15 | SpaceCoaster | If we could map the 16+16 at the bottom, maybe the 1m chunks could be used still |
| 23:50:15 | GPSFan | SpaceCoaster: bz told me last night that the MMU is setup in the kernel, and that blob passes it a memory map showing where the chunks are. |
| 23:50:54 | SpaceCoaster | Hmm I saw the memory map, maybe if I hardwire it, it's in C code, much easier to handle, then we ca reverse engineer from there |
| 23:50:57 | GPSFan | SpaceCoaster: unfortunately the banks are fixed by what CAS & RAS lines are wired to the socket. |
| 23:52:08 | GPSFan | SpaceCoaster: the SODIMM banks are 2 & 3 starting at D0000000. |
| 23:52:53 | bhima | wow. ok. I suck. I needed to wait longer to press 2. |
| 23:54:28 | SpaceCoaster | I need to investigate blob a little more, see what it is doing to the MMU and if the kernel is being loaded into virtual memory or physical. They may be using a 1:1 mapping from physical to virtual which we may be able to kludge. |
| 23:54:52 | GPSFan | SpaceCoaster: if we wanted to hack the hardware, we could disable the 8M onboard, and rewire the socket to start at C0000000, then have a 64M only. |
| 23:55:34 | GPSFan | It would be nice if a real programmer (like you) look at blob and figure out what you just said. |
| 23:56:17 | SpaceCoaster | Hmm, thanks for the compliment, I'll give it a go. |
| 23:57:03 | SpaceCoaster | GPSFan: ok a simple hardware question the ATMEL flash is write protected if the WP pin is tied to something. Is it or is it not? |
| 23:57:25 | GPSFan | SpaceCoaster: you and GoRk ,& CP have gotten more working in tha last week than everyone else put together. |
| 23:57:41 | SpaceCoaster | GPSFan: you have it on the schematics but I don't know which way is up (1 or 0). |
| 23:58:27 | SpaceCoaster | Yeah but I need the memory to work, my java app won't fit in 16M :-( |
| 23:58:34 | GPSFan | SpaceCoaster: depends on whether you like positive of negative logic, you can have your choice of up or down, Think of it as being weightless. |
| 23:59:16 | SpaceCoaster | the inferno stuff is full of if (0 != (a == b)), arrgh a hardware engineer |
| 23:59:17 | GPSFan | SpaceCoaster: 16M is pretty lame considering the SA1100 will support 4 banks of 128M each. |
| 23:59:57 | Russ | SpaceCoaster: I hate that stuff |
