09:09:20 | seletz | hi |
09:27:36 | seletz | erikm: hi |
09:28:04 | erikm | morning |
09:28:44 | seletz | erikm: i got a patch from Urs Kaufmann "KuEdE" for his miniprint board. I'll integrate it, ok? |
09:41:54 | erikm | seletz: sure |
10:00:00 | seletz | ok, done |
10:00:45 | seletz | erikm: been very busy lately? You have'ntg been around a long time ... |
10:02:17 | seletz | ugh, siglunch |
10:12:55 | erikm | seletz: *nod* very busy |
11:26:10 | seletz | hi again |
11:53:06 | Sammy | bye... |
14:12:39 | seletz | prpplague: hi |
14:34:39 | prpplague | seletz: morning |
14:34:39 | prpplague | seletz: how was italy? |
14:52:43 | prpplague | erikm: is the chain booting on blob pretty stable? |
15:01:26 | erikm | prpplague: it should be |
15:01:41 | erikm | prpplague: I use it to chain boot blob on bootldr platforms |
15:18:24 | prpplague | erikm: heres what i'm thinking, using a stable blob as primary bootloader which is modified only to load new blob images in a secondary location |
15:19:10 | prpplague | erikm: this way ppl can update their blob without a worry of loading a bad image and bricking their device |
15:24:31 | seletz | prpplague: hi, Italy was cool. Loads of Pizza, pasta & pesce. Mhhh... |
15:24:59 | seletz | prpplague: and wondeful sights & people. |
15:53:38 | prpplague | seletz: cool |
15:53:48 | prpplague | seletz: try any cool beer? |
16:09:58 | seletz | prpplague: sorry, phone. Got some messy mails .... |
16:10:17 | seletz | prpplague: no, no birra. Vino rosso. |
16:10:31 | seletz | prpplague: (We had a couple of gin tonics, though...) |
16:10:37 | erikm | prpplague: that's already possible. just download a second blob as if it's a kernel, and boot it |
16:12:01 | seletz | erikm: yeah. prpplague just wants to have a fail-save system: He wants to disallow (lock) the first blob (fail-save) and boot into a second one (the one which is updateable). |
16:12:13 | seletz | prpplague: correct me if i'm wrong :) |
16:12:53 | seletz | erikm: that way you dont end up with an system not responding on serial line. |
16:12:54 | prpplague | seletz: that is correct |
16:13:52 | prpplague | i did some testes last week at the local college, i took 3 tuxscreens to the ee department and had them try to get started using them |
16:14:05 | seletz | prpplague: cool |
16:14:08 | prpplague | they bricked all three by the end of the day |
16:14:17 | seletz | prpplague: doh |
16:14:18 | prpplague | of course i had a jtag to recover |
16:14:37 | seletz | prpplague: ok, that was not exactly expected :( |
16:14:44 | prpplague | but when we are talking embedded classes, that might cause a time issue |
16:14:46 | seletz | prpplague: what did they do? |
16:15:46 | seletz | prpplague: (thats why system testing is not for developers but for (l)users) |
16:16:37 | prpplague | seletz: as i understand it, 2 used a blob compiled with gcc no arm-linux-gcc, and the other was unknown |
16:16:50 | seletz | cool |
16:17:32 | seletz | prpplague: ok, but this is solveable with your 2-blob solution IMHO. |
16:17:50 | seletz | prpplague: so, arte they still interested? |
16:17:59 | seletz | s/arte/are/ |
16:18:36 | prpplague | seletz: ya still interested |
16:19:45 | erikm | seletz: already in place |
16:19:53 | erikm | seletz: just lock the first flash image |
16:20:37 | seletz | erikm: yes. And have a second blob image, issue "boot" at the first stage, chain-boot into the second. |
16:21:20 | erikm | seletz: and you can still brick the box, btw. just unlock the flash from the second blob. or do it in linux. or let the kernel crash in interesting ways |
16:21:47 | seletz | erikm: yeah, one does not get 100% safety. |
16:22:32 | seletz | erikm: but i think about bas images or stopped system updates (power-cycled during flashing cramfs). |
16:22:33 | erikm | prpplague: compiling blob with the native gcc compiler is impossible, it will at least barf about the inline assembly |
16:23:22 | erikm | seletz: yes, that's a "don't try this at home, kids" problem. we can't protect against stupid lusers. note that flashing your PC BIOS comes with the same warnings |
16:24:10 | seletz | erikm: Yeah, exactly. I personally just want "technican" safety, not "luser"-safety. |
16:24:11 | erikm | prpplague: oh, and it will also not be able to compile the .S files |
16:24:53 | seletz | prpplague: The technican who powercycles during sysupdate i mean. He should have a secound chance. |
16:25:00 | seletz | s/secound/second/ |
16:25:43 | prpplague | erikm: thats what i thought, but i wasn't on hand when the incident occured, |
16:25:58 | prpplague | erikm: i was only going on what the instructor porvided |
16:26:07 | erikm | seletz: no. we can expect that a technician reads the manual and knows what he is doing |
16:26:42 | erikm | seletz: though I admit that we have to warn in the manual against power cycling during flash |
16:27:04 | erikm | prpplague: I expect that they uploaded blob-start or so instead of blob |
16:27:04 | seletz | erikm: yes. One would expect that. Its not that way in the real world in my exp. |
16:27:40 | erikm | seletz: sure, I know. but a field technician also has the jtag device to put blob back into flash |
16:28:08 | prpplague | erikm: i agree that techs should know better, but the target audience is new embedded techs, so mistakes do happen, i'd be nice to "see if you were working a full prototype your boss would be pissed" |
16:29:32 | seletz | erikm: Thats the question. On some systems JTag pins are hiden and only accessible when opening a case. Case-opening then may require proper sealing when closed again. |
16:29:52 | seletz | erikm: (its that way in our system3 and dafit handheld devices). |
16:30:19 | seletz | erikm: but this is probably "bad design". |
16:32:17 | erikm | seletz: right, so this additional pain for the tecnician to think before he power cycles the device. I consider that a Good Thing [tm] |
16:32:55 | prpplague | i personally feel that failures and bad technique in the class room is acceptable if you have a procedure set in place to point out these problems and learn from them |
16:33:09 | seletz | prpplague: yeah. |
16:33:23 | seletz | prpplague: that is education all about. |
16:33:31 | erikm | seletz: much in the same way you could try to protect a Unix system against stupid mistakes as root, but in real life you should just discover the hard way that you have to think twice before you hit enter |
16:33:56 | seletz | erikm: yeah, i agree fully. You run in open doors ... |
16:34:00 | seletz | erikm: :) |
16:36:13 | seletz | erikm: Its just that i have to think about support costs too. JFlashing is considered as an in-house service. So Devices which need to be JFlashed have to be sent in. All the way from timbuktu. Thats why i want to minimize chances to render devices unusable if possible in an easy way. |
16:36:27 | erikm | looks at his new NTFS recover tool recovering data at > 15MB/s |
16:36:55 | seletz | erikm: that is, its considered in-house for system3 devices. Not for the linux hack kit. |
16:37:00 | prpplague | erikm: nice |
16:37:02 | erikm | seletz: write protect the blocks in which blob lives |
16:37:21 | prpplague | erikm: then you can't upload new blobs |
16:38:03 | erikm | prpplague: of course you can, you only have to unprotect the blob flash blocks |
16:39:11 | erikm | seletz: you could even make a blob .init function that protects the blob flash blocks so that you are sure that the flash is protected each time you boot |
16:40:03 | seletz | erikm: yeah. i'll probably lock the whole system. I defined CF as the only RW media. |
16:41:57 | prpplague | erikm: ya but that introduces the problem of destroying blob again |
16:42:33 | prpplague | erikm: i mean i see your point, i'm on your side with the basic idea, but....... |
16:43:27 | prpplague | erikm: i have to think about the possiblity of accidents |
16:43:41 | erikm | prpplague: yes, but you have to issue a specific command to unprotect the blob flash blocks. the probability that joe random luser a) unprotects the correct blocks and b) writes a flash image is very low. |
16:44:53 | prpplague | erikm: but see thats just it, i my target audience is the education enviroment where newbie will be experimenting on things just like flashing a bootloader |
16:45:57 | prpplague | of course, this might be a built in source of income....., flash reload server $25.00 |
16:46:03 | prpplague | s/server/service |
16:46:29 | erikm | prpplague: in an education environment you can expect an instructor with the flash hardware |
16:49:06 | prpplague | erikm: ya i considered an instructor kit with flash hardware |
16:49:39 | prpplague | erikm: anyway, i dodn't mean to pull you away from work |
16:49:48 | prpplague | s/dodn't/didn't |
16:52:22 | prpplague | erikm: maybe we can discuss it more at OLS |
16:52:28 | erikm | sure |
16:52:37 | prpplague | seletz: still think you might have a prototype by OLS |
16:52:49 | seletz | prpplague: ... |
16:52:55 | prpplague | erikm: don't forget, i'm buying a round of brew for you guys |
16:53:08 | erikm | hehe, no I don't :) |
16:53:58 | prpplague | seletz: well, i've killed a PO for some trizeps untill october |
16:54:20 | seletz | prpplague: sorry? PO? |
16:54:27 | prpplague | seletz: Purchase Order |
16:54:42 | seletz | prpplague: ok, october is reasonable. |
16:55:02 | seletz | prpplague: how many # did you order ? :) |
16:55:22 | prpplague | seletz: the original order was for 10 |
16:55:37 | prpplague | seletz: but by october we will probably need 25 |
16:56:00 | seletz | prpplague: prototypes? |
16:56:15 | seletz | prpplague: or for edu market already? |
16:56:24 | prpplague | seletz: ya for prototype development |
16:56:32 | seletz | prpplague: ok |
16:56:53 | seletz | prpplague: for edu market the numbers are quite low, actually. |
16:57:26 | prpplague | seletz: we are really starting to move towards arm as the core of choice |
16:58:06 | prpplague | seletz: well, once we get the initial board ready, i expect along with our own needs, to see about 100-200 core units per month |
17:00:10 | prpplague | seletz: quick question, the uarts from the core board are going 3.3v levels right? |
17:00:47 | seletz | prpplague: i've included a MAX chip for the first uart un the core board, so we'll have TTL |
17:00:59 | seletz | prpplague: for the first serial, that is. |
17:01:28 | seletz | prpplague: so one can have a core board and connect power, serial and JTAG and off you go. |
17:01:53 | prpplague | seletz: cool |
17:02:08 | prpplague | seletz: which max chip are you using? max3232 ? |
17:02:21 | seletz | prpplague: one moment ... |
17:02:27 | seletz | digs out specs |
17:03:15 | seletz | prpplague: basically the same as on assabet |
17:04:30 | seletz | prpplague: its a MAX3244 |
17:05:21 | prpplague | seletz: ya thats the one with low battery function IIRC |
17:07:53 | seletz | btw, anyone knows good (i.e. big name, well-known) companies which use ARM Linux and StrongARM cores? |
17:07:53 | prpplague | seletz: ahh no thats the one with full hardware handshake |
17:08:35 | prpplague | seletz: no, but there is a pretty full list at www.arm.com |
17:08:47 | seletz | looking |
17:09:01 | seletz | has to answer silly emails :( |
17:09:28 | erikm | prpplague: you mean the 3223 |
17:10:38 | erikm | .eu people, sign this: http://www.stop1984.com/index2.php?lang=en&text=letter.txt |
17:10:59 | prpplague | erikm: ya thats the one i was thinking of |
17:12:07 | prpplague | erikm: you sure know you IC's :) |
17:12:43 | erikm | prpplague: heh, this is the one we used on the LART |
17:19:51 | prpplague | ya, i'm looking forward to meeting jdb as well as the rest of you guys at OLS |
17:20:26 | prpplague | sounds like LART was a great learning experience for you guys |
17:20:53 | seletz | prpplague: i hope the hack kit core will be for us ... |
17:21:07 | prpplague | seletz: lol, me too |
17:22:18 | erikm | prpplague: AFAIK jdb won't come to OLS |
17:27:48 | seletz | prpplague: have any specs/proposals/ideas 'bout the hack kit i could include on my web page? |
17:58:00 | prpplague | seletz: ya, i could get some stuff together |
17:58:15 | prpplague | erikm: ahh, thats right you told me that earlier |
17:58:21 | prpplague | is slow today |
17:59:29 | prpplague | http://www.ratebeer.com/ViewUser.asp?UserID=1210 |
17:59:43 | prpplague | i hope to break 200 at OLS |
17:59:44 | prpplague | lol |
18:00:15 | seletz | prpplague: man, what the hell are you drinking there?? |
18:10:05 | seletz | argl |
18:10:10 | seletz | tired |
18:10:17 | seletz | going home |
18:11:42 | prpplague | seletz: a yard of bass ale |
18:16:25 | seletz | prpplague: ugh? |
18:16:32 | seletz | prpplague: whats that? |
18:17:12 | prpplague | seletz: a yard? english style glass size |
18:17:14 | seletz | prpplague: looks like you drinking out of a lenghty cow horn |
18:17:58 | seletz | anyway, i'm off |
18:18:01 | seletz | see ya |
19:05:48 | Russ | afternoon erikm |
19:08:24 | erikm | hey Russ |
19:20:34 | prpplague | Russ: howdy |
19:30:52 | prpplague | Russ: you get rid of /proc/sys/gf ? |
19:56:50 | Russ | yup |
19:57:20 | Russ | /proc/loadavg has gone down considerably |
20:12:33 | prpplague | wahoo! |
20:12:46 | prpplague | has been victorious over my pcmcia issue! |
20:33:39 | prpplague | mmatten|home: ! |
20:38:02 | Russ | which blob image do I load if I want to chain-boot? blob-rest? |
20:39:53 | mmatten|home | prpplague: ! |
20:41:15 | mmatten|home | just checkin my mail :) |
20:47:10 | mmatten|home | KuEdE: btw i've played around with your sourcecode from the miniprint, added a couple of bitmasks to make lcd.h a bit more intuitive when it comes to configuring those LCCR Registers... if seletz_zz agrees with what i did we'll probably add that to the cvs as well (i'm fiddling with the lcd stuff for our board as well, so i just included your stuff in the bitmask work) |
20:47:10 | Russ--- | er, duh, blob-chain |
20:47:39 | mmatten|home | i hope he's not idling or i wrote all for naught :) |
21:02:01 | Russ | Consider yourself BLOBed! |
21:05:04 | mmatten|home | ugh i'm doomed :) |
21:05:11 | prpplague | Russ: m$bootloader - Consider yourself BLOAted |
21:05:29 | mmatten|home | prpplague: ugh we're all gunna die :> |
21:06:33 | prpplague | mmatten|home: why? |
21:06:56 | mmatten|home | i got m$ here :P |
21:07:26 | mmatten|home | (preparing to deflect thrown stones) |
21:08:40 | prpplague | mmatten|home: hmm, friends don't let friends use winblows |
21:08:57 | prpplague | mmatten|home: can i recommend a good counsler? |
21:19:34 | Russ | I musta missed something |
21:25:40 | mmatten|home | prpplague: *unfortunately i need windoze :) |
21:26:17 | Russ | no one "needs" windows |
21:26:38 | mmatten|home | 1. there are no linux ports for my audio software |
21:26:47 | mmatten|home | 2. i'm a student and need to do stupid tasks :) |
21:27:07 | mmatten|home | so there ain't much choice... |
21:27:09 | Russ | I made it though all four years not needing doze |
21:27:33 | mmatten|home | but, to make it up i got a 2nd (linux) box in the other room :) |
21:27:52 | mmatten|home | hmm well i guess u didn't have my windows-fanatic professors :P |
21:28:34 | Russ | we had to write code to run on sco for most of my classes (or m6800 for some) |
21:29:05 | KuEdE | mmatten: im back... cool some guys awake |
21:29:10 | Russ | one of my classes, I used gtk instead of msvc++ and demoed with a laptop |
21:30:00 | mmatten|home | russ: they made me use borland c++ builder, which to be honest, would be reason enough for me to quit a company |
21:30:13 | Russ | another one, I was in the team, and I did the backend, and everyone else did the frontend in msvc++ |
21:30:23 | mmatten|home | i didn't have a choice <sniff> |
21:30:25 | Russ | I rarely had to write anything in c++ |
21:30:39 | mmatten|home | well if it's c++ class.... |
21:30:49 | Russ | about half of my coding hours were spent with assembly |
21:31:16 | prpplague | college days were spent with fortran77 |
21:31:20 | Russ | course, now after I left, its all java |
21:31:33 | mmatten|home | that we didn't do to much, unfortunately... i rarely learned any assembler which is a pity considering the low level linux stuff |
21:31:37 | Russ | even for computer systems engineers, those damn cs freaks are taking over stuff |
21:31:41 | mmatten|home | assembler i mean |
21:32:31 | prpplague | Russ: lol |
21:32:39 | mmatten|home | well i guess thats the fate of the late-born, struggle with the overhead instead to learn the basics |
21:32:44 | prpplague | is not java fan |
21:32:52 | mmatten|home | neither |
21:33:07 | KuEdE | ups...had a crash |
21:33:44 | prpplague | KuEdE: was the driver ok? i hate to see what happened to that big brown truck,lol |
21:34:00 | KuEdE | mmatten: im working too on lcd.c ... |
21:34:10 | mmatten|home | i know i've seen it |
21:34:14 | KuEdE | so ive got to get newest CVS |
21:34:27 | KuEdE | i added 12bit rgb support |
21:34:37 | KuEdE | for my nice boot logo. or so |
21:34:56 | KuEdE | but its still messy |
21:35:43 | KuEdE | Q: is in diag data-cache enabled ? no ..ithink |
21:36:05 | Russ | no, because it would require setting up the MMU |
21:36:19 | KuEdE | ahh.. i c |
21:36:23 | Russ | the instruction cache generally does the job though |
21:37:06 | KuEdE | played around with some high mem bandwith graphic effects |
21:37:41 | KuEdE | so it seemed not so fast... trying to do som asm for core grx routines |
21:38:31 | mmatten|home | like what ? |
21:39:11 | Russ | dude, why are you doing that in blob? |
21:39:40 | KuEdE | cos i want to have some simple boot gfx... so its not only "dark" for some seconds |
21:39:53 | mmatten|home | guess for the same reason as usual: customer requirement |
21:39:57 | mmatten|home | ehm |
21:40:00 | mmatten|home | a sec |
21:40:09 | Russ | KuEdE: I have a patch for that |
21:40:11 | KuEdE | nope... building my own mp3 player ;-) for my car.. |
21:40:15 | mmatten|home | u do the image for the boot in blob ? |
21:40:35 | Russ | I just stored a logo in the paramater block, in the form that the fb memory needs it (palette and all) |
21:40:45 | Russ | then just enabled the the fb code |
21:40:47 | KuEdE | yes.....after reset |
21:40:57 | Russ | its very simple |
21:41:01 | KuEdE | ahh |
21:41:13 | Russ | also used mini_inflate so I could store the image deflated |
21:41:26 | mmatten|home | russ: that ets pretty mem intensive with color display no ? |
21:41:31 | mmatten|home | ets=gets |
21:41:47 | KuEdE | good idea...but i want some cool stoff... |
21:41:49 | mmatten|home | i mean if he did it from blob |
21:42:13 | mmatten|home | oh sry parameter block |
21:42:25 | mmatten|home | i take everyting back :) |
21:42:44 | Russ | I just did BLOB! in big white letters |
21:43:06 | Russ | compressed, that takes up very little space |
21:43:20 | mmatten|home | hmm well we need to do something similar to Kuede, screen for sysupdate for example |
21:43:36 | KuEdE | i want to do some animated (but low mem) "linux" than small in the corner a "powered by blob".... thats what i think of |
21:43:45 | mmatten|home | that will have to reside in blob directly |
21:43:54 | Russ | KuEdE: how long to you plan on being in blob? |
21:44:32 | KuEdE | Russ: 1/2 year i think.. because i do my "Diplomarbeit" on this stuff |
21:44:42 | KuEdE | ..later.. |
21:44:46 | mmatten|home | kuede: hmm russ is right, i mean, there are already a couple of solutions to display logo during bootup (tux-logo i.e) |
21:45:33 | Russ | I mean, when most boards boot, blob is running for about 2 seconds |
21:45:47 | Russ | so to say you want not just a logo, but an animated logo... |
21:46:01 | KuEdE | you dont have to include it in the CVS ;-)) |
21:46:29 | mmatten|home | i think erikm would send paid assains if u did :) |
21:46:46 | mmatten|home | (at least if u put it in the standard build) :) |
21:46:55 | KuEdE | hmm.. ure right .. ill think about it |
21:47:09 | KuEdE | :-)) |
21:47:37 | mmatten|home | KuEdE: you're still student as well ? |
21:47:52 | KuEdE | yes |
21:47:56 | KuEdE | in .ch |
21:47:57 | prpplague | Russ: so when blob turns over the boot to the kernel does the "BLOB" image clear? |
21:48:14 | Russ | if you make it |
21:48:30 | Russ | otherwise, your image data will probaly get overwritten by random data |
21:48:43 | Russ | so its a good idea to disable the LCD before booting the kernel |
21:49:16 | KuEdE | i think too... when mmu is enable things will corrupt i think |
21:49:21 | mmatten|home | hmm doesn't the display/fb gets reinitialised on kernelboot? (newbie question i know) |
21:49:41 | Russ | not immediately |
21:50:23 | mmatten|home | well before that it can go corrupt and afterwards blank... |
21:50:26 | prpplague | Russ: btw, done any more with the Camion? |
21:50:36 | Russ | working on the kernel right now |
21:50:43 | prpplague | Russ: cool |
21:50:55 | prpplague | Russ: its a shame we can't some more of those |
21:51:12 | mmatten|home | ack - almost midnight |
21:51:16 | prpplague | mizi, wants to seem me some exilen104 units |
21:51:36 | prpplague | s/seem/send |
21:52:07 | mmatten|home | i better get some sleep -.- |
21:52:17 | prpplague | mmatten|home: night |
21:52:26 | mmatten|home | night all :) |
21:54:02 | KuEdE | night... |
21:57:05 | prpplague | Russ: so what do you think of the Camion? |
21:57:16 | Russ | its a nice unit |
21:58:03 | prpplague | Russ: ya thats what i thought, we would have bought 50 if we could get them |
21:58:26 | Russ | I think we'll be able to extend battery life in suspend quite a bit too |
22:00:25 | prpplague | Russ: find some bugs? |
22:00:31 | prpplague | opps, brb |
22:09:42 | Russ | prpplague: there was a recend discrussion on arm-kernel about pm |
22:09:51 | prpplague | Russ: ahh |
22:10:13 | prpplague | thats the 4th time the latest mozilla has kernel paniced my system |
22:10:29 | Russ | umm...thats a bad sign for X |
22:11:22 | prpplague | Russ: so whats the story on pm? |
22:11:43 | Russ | just a lot of improvements |
22:13:11 | prpplague | Russ: cool |
22:14:25 | prpplague | Russ: i need to get with those triad guys for a price quote |
22:43:42 | prpplague | Russ: http://www.toaei-usa.com/page683639.htm |
22:52:37 | Russ | nice, too bad it doesn't have more ram |