Appeal for Linux Help from Pedal-Powered Internet?
lma asks: "As has been reported previously on Slashdot, Lee Felsenstein is involved in a project to bring wireless internet connectivity (including VOIP) to
isolated villages in Laos. Lee is in Laos this week, trying to install and configure the initial
systems. He's having a problem getting Linux installed and booting off a M-Systems DiskOnChip(DOC) as described here. Perhaps Slashdot readers could help with this problem, and make themselves available for future technical support issues as well."
used the February 7 2003 cvs snapshot to upgrade the 2.4.21-pre2 mtd drivers. Ran the patchin.sh, which worked nicely
Tried this "latest source" amalgamation with both the doc2000 and doc2001 driver, same results:
doc2000 driver detects no flash chips doc2001 driver detects 3 flash chips.
Not to be the Grinch, and I applaud the intentions and effort of the people working on this, but -- isn't the real answer that installing some crazy-ass combination of patched Linux kernels and "Imagine a Beowulf cluster of these!" hardware should be left to bored CS majors, instead of being left in the hands of users with no relevant technological knowledge whatsoever?
OK, let's say that given enough Ask Slashdots from Vientiane you can get this thing to work. Then what? And how will this help anyone else in the world?
For that matter, how are detecting no chips and detecting 3 chips the same result?
What I'm listening to now on Pandora...
OK, not to be a bastard, but isn't this the kind of shit you learn how to do *before* you go "on-site," as it were, to install it? I have set up for m-systems disk-on-chip stuff and it's not that easy. I can see how he might be having problems with it, as it's an unnecessarily complicated interface. He should be contacting M-Systems and Tri-M to get this resolved, though, and not slashdot.
#1) Your 20/20 hindsight is not useful. Live in the now.
/. picked up on his cry for help and presented it to the community, because wider exposure may lead to resolution - or to put it another way, "Given enough eyeballs, all bugs are shallow".
#2) He didn't contact slashdot, as far as I can see; rather, somebody at
Incidentally, I too have failed to get Linux running on a disk-on-chip. I was trying it in my basement on salvaged equipment, and eventually I shelved the project after a few weeks of frustration.
I suspect the ltsp guys might be the most accessible experts on this. I'm going to forward the link to them...
Please see http://leaf.sourceforge.net/devel/jnilo/budiskonch ip.html
It's the "Installing and booting Bering from a M-Systems DiskOnChip" chapter (written by Brad Fritz) of the LEAF Bering User's Guide.
Bering is one of the branches (the currently most active one) of the LEAF project that is building on the strong LEAF heritage and adding some advanced stuff to get: kernel 2.4.19 (and 2.4.20), PPP[OE], firewalling (using Shorewall), bridging, wireless utils, linux-wlan, Host AP, DHCP (client and server), DJB's dnscache, pcmcia, Freee Swan, .... Bering main developers are Jacques Nilo and Eric Wolzak
LEAF is the Linux Embedded Appliance Firewall (http://leaf.sourceforge.net/).
...not sure why "this solution" is being used, my guess is because of the nature of the power supply, and the costs involved, the sheer low quantity of the electric that's able to be produced, and the sophistication of the end users. Set up hard, running it after set up easy? And maybe travel back and forth to remote regions is harder than bopping on the tube or jumping in the belchfire for a spin to the store. That could be why he's doing it onsite now. Heck, I've lived places in the US it was quite the adventure to get to any "store" at all, let alone any tech store or place to get help on a tech question. This ws back pre internet days. I guess being even more remote than that causes some occassional problems, but I will agree the boxes should have been developed off site and QAed better. I think it's called making a mistake, bet everyone reading this has done that. The glass houses-stones postulate appplies as well as murphy's law..
Really just guessing on all this though.
I think a better deal would have been to get low powered, best battery life functional laptops donated from someplace and use them, perhaps modded for pure DC use, eliminate all that inverting, converting and perverting jazz that goes on to get juice to where it needs to be. Using the laptops built in battery as a cushion for the intermittent pedal electrical power seems the best idea. That and see if any of the big alternate energy suppliers would donate some stuff for more juice on site. I know from just one of my PV panels I can get enough juice during the day time to run one of my older laptops easily. Umm, I do it all the time. I also know in monsoon season in the tropics this isn't practical, so that's my guess on the pedal powered deal.
I had trouble getting a DoC recognised on my webplayer too, with the same kind of symptoms.
/* Virgin-Linux patch for WebPlayer DoC timing, multiply cycles by 4 */
Eventually I found that adding a delay in the DoC code fixed it - see patch.
--- linux/drivers/mtd/doc2000.c Tue May 1 08:19:14 2001
+++ linux/drivers/mtd/doc2000.c Tue May 1 08:21:25 2001
@@ -71,7 +71,8 @@
volatile char dummy;
int i;
- for (i = 0; i < cycles; i++) {
+
+ for (i = 0; i < cycles*4; i++) {
if (DoC_is_Millennium(doc))
dummy = ReadDOC(doc->virtadr, NOP);
else
You are in a twisty maze of processor lines, all alike.
There is a lot of hype here.
I guess it's just typical of this mostly American and war-obsessed /. crowd that an altruistic project such as this could only muster a full NINE comments, not including this one. Crawl back into your insular paranoia laden trenches, Yanks. No wonder the rest of the world hates you!