| 00:09.08 | landley | How would I go about mounting a jffs2 filesystem image via loopback? |
| 00:09.10 | landley | (Can it be done?) |
| 00:09.30 | landley | I have a filesystem image that "file" doesn't identify, but strings is showing me that it does indeed contain busybox. |
| 00:09.46 | landley | Trying to find a more elegant way of mounting it than just chopping out everything before the busybox ELF header... |
| 00:11.38 | ds2 | use the fake block to mtd device converter on top of it |
| 00:14.47 | landley | ds2: how do I do that? |
| 00:15.13 | ds2 | let me see if I can find hat config option |
| 00:15.24 | landley | I do have notes on how to add the mtd utilities to toybox, though. :) |
| 00:15.37 | landley | Pestered Thomas Gleixner about it back around February. |
| 00:17.19 | ds2 | use CONFIG_MTD_BLOCK2MTD |
| 00:17.43 | ds2 | it'll convert the loopX device into a mtdN/mtdblockN pair that will then let you mount w/-t jffs2 |
| 00:22.11 | Redhatter | Silly question... why can't jffs2/yaffs2, etc use a plain block device? I would've thought it'd be a good filesystem to use on SD/CF cards, except that they don't work that way. |
| 00:24.38 | landley | ds2: use CONFIG_MTD_BLOCK2MTD in _what_? |
| 00:24.44 | landley | That's a... kernel configuration option? |
| 00:25.00 | landley | You're saying rebuild the kernel on my laptop, and that will make all loop devices look like mtd devices? |
| 00:25.23 | landley | Or make all loop devices automatically have a corresponding mtd device? |
| 00:28.09 | ds2 | read the docs for it ;) |
| 00:28.34 | ds2 | the summary of it is, if you build it as a module, you can use this to create a mtd device from anyblock device such as a loop device |
| 00:31.55 | *** join/#edev prpplague (n=dave123_@ppp-70-244-175-101.dsl.rcsntx.swbell.net) |
| 00:32.11 | prpplague | landley: how goes the battle? |
| 00:33.00 | prpplague | http://www.linuxdevices.com/news/NS5622936770.html |
| 00:33.03 | prpplague | fun fun |
| 00:33.45 | CosmicPenguin | ahh - yay |
| 00:33.55 | ds2 | prpplague: are you presenting? =) |
| 00:34.38 | landley | Oh yeah, CELF. |
| 00:35.03 | prpplague | ds2 no |
| 00:35.15 | ds2 | attending at least? =) |
| 00:35.15 | prpplague | ds2 never gotten enough nerve up to submit anything |
| 00:35.32 | prpplague | ds2 probably not, AML doesn't pay for conferences |
| 00:35.33 | CosmicPenguin | you should do it |
| 00:35.41 | landley | Hmmm... Penguicon is April 18-20. |
| 00:35.44 | ds2 | wonder how hostile are the audiences at CLEF |
| 00:35.59 | landley | ds2: the one time I presented at CELF, the audience was 50% japanese people. |
| 00:36.04 | prpplague | penguicon is at the same time? |
| 00:36.13 | ds2 | landley: but do they ask questions afterwards? |
| 00:36.21 | landley | By which I mean clustered around in groups speaking Japanese, flown in from japan. |
| 00:36.26 | landley | Very uncomfortable talking in english. |
| 00:36.43 | landley | Each group had a designated speaker, more or less, who communicated with uss cary americans. |
| 00:36.54 | prpplague | hehe |
| 00:37.08 | landley | They'd meet you after your talk and the designated speaker would ask questions. |
| 00:37.15 | prpplague | i noticed a ton of celf docs and posts are all japanesse |
| 00:37.20 | landley | Asking questions during your presentation? Even when you ask for 'em? Not so much. |
| 00:37.29 | prpplague | CosmicPenguin: you gonna submit a paper? |
| 00:37.30 | landley | Well CELF is... sony, is it? |
| 00:37.32 | CosmicPenguin | maybe |
| 00:37.35 | landley | Who does Tim Bird work for? |
| 00:37.39 | prpplague | landley: indeed |
| 00:37.39 | ds2 | oh that isn't bad... just don't want an audience that will start downright challenging you the moment you utter 2 words |
| 00:37.43 | ds2 | (i.e. |
| 00:37.46 | ds2 | good evening) |
| 00:37.56 | landley | Well _I_ don't mind that, but I've hung out on linux-kernel on and off since 1998. |
| 00:37.57 | prpplague | ds2 these are professionals |
| 00:38.11 | prpplague | ds2 its not defcon or such |
| 00:38.11 | *** join/#edev shabble_ (n=shabble@cowu.be) |
| 00:38.17 | ds2 | prpplague: that has happened in talks at ESC :/ |
| 00:38.34 | landley | Usually I learned whatever I'm talking about by breaking it repeatedly and beating it into submission by very small increments. |
| 00:38.42 | prpplague | hehe |
| 00:38.47 | prpplague | landley: indeed as do i |
| 00:38.53 | landley | If they can trip me up, it's just bringing up something I would have run into sooner or later. |
| 00:38.57 | CosmicPenguin | HPSCW at the beginning of april |
| 00:39.08 | prpplague | hpscw? |
| 00:39.09 | landley | I also have no problem saying "wow, no clue on that one. Any ideas?" |
| 00:39.21 | prpplague | landley hehe |
| 00:39.31 | ds2 | would be a useful experience to do at least one talk |
| 00:39.38 | landley | I was planning on driving to Penguicon, which is a day and a half drive from Austin. |
| 00:39.52 | landley | (That way I can visit my sister in minnesota afterwards, plus my wife can come along.) |
| 00:39.57 | prpplague | CosmicPenguin: last conference i went to was OLS back in 2001 |
| 00:40.14 | landley | My first OLS was 2006. |
| 00:40.22 | landley | Only been there twice. Probably not going again this year. |
| 00:40.49 | landley | Yes, Penguicon and CELF are at about the same time, although in our defense we've been approximately that weekend since 2003. |
| 00:40.56 | landley | They scheduled on top of us. :) |
| 00:41.05 | CosmicPenguin | I figure I can beat another year out of OLS - the mini dev conferences at the beginning of the week promise to be useful |
| 00:41.18 | CosmicPenguin | but I am _not_ speaking - I've had that fun |
| 00:41.46 | prpplague | yea i've heard OLS just isn't what it used to be |
| 00:41.57 | prpplague | especially since most of kernel guru's don't go anymore |
| 00:42.07 | landley | I wandered into the CELF bof and tim bird's slides mentioned me by name, and mentioned toybox. |
| 00:42.09 | landley | Boggle. |
| 00:42.15 | prpplague | hehe |
| 00:42.16 | landley | Somebody other than me was paying attention to that? |
| 00:42.34 | CosmicPenguin | Tim probably watches you very carefully |
| 00:42.52 | CosmicPenguin | Can't trust those guys from Sony |
| 00:42.54 | prpplague | landley hehe my boss read the dead on the verizon busybox suit and said "isn't landley the guy we just sent and eval to?" |
| 00:42.54 | landley | It's a hobby project. An outlet for the urges that drove me to work on busybox, now that I can't do busybox anymore thanks to Bruce. |
| 00:43.23 | landley | prpplague: and my questions above about loopback mounting jffs2 have to do with figuring out whether the source High Gain Antennas sent us is compliant or not. :) |
| 00:43.29 | prpplague | landley well you can see a ton of projects are forking to basically two groups |
| 00:43.39 | landley | hmmm? |
| 00:43.54 | prpplague | landley projects that are cutting edge and are very very opensource "do it this way" types |
| 00:43.58 | ds2 | You would think with all that money going to lawyers why not just write a multi call binary w/a BSD license |
| 00:44.09 | ds2 | that has got to be cheaper |
| 00:44.23 | prpplague | landley the other projects tend to be more business oriented, geared towards stability and "what works" |
| 00:44.24 | landley | ds2: BSD projects fork. |
| 00:44.33 | ds2 | might not be a good thing community wise,, it is got to be cheaper then all the lawyers |
| 00:44.34 | prpplague | landley i.e. the difference between OE and Poky |
| 00:44.43 | landley | GPL reduces forking. It's pretty much that simple. |
| 00:44.52 | ds2 | bah |
| 00:44.53 | CosmicPenguin | Poky isn't so much a fork as a customer friendly revision |
| 00:45.10 | ds2 | much rather see money going to a BSD project the lawyers |
| 00:45.15 | landley | prpplague: did I ever point you at the old three waves articles I did back in 2000? |
| 00:45.18 | prpplague | CosmicPenguin i disagree, i think its a true fork in the aspect the whole goal is different |
| 00:45.32 | prpplague | landley sorry no, url? |
| 00:45.32 | ds2 | s/the/then |
| 00:45.37 | landley | ds2: lawyers only come into it when they don't respond when contacted out of court. |
| 00:45.50 | CosmicPenguin | poky is a subset - but its not a fork in the traditional sense because everything in Poky goes into OE |
| 00:45.57 | CosmicPenguin | RP is the defacto master of both |
| 00:45.59 | ds2 | landley: I suspect they get involved at the VZ side as soon as they are contacted |
| 00:46.21 | landley | prpplague: bottom of http://landley.net/writing/ has links. |
| 00:46.23 | CosmicPenguin | but Poky scratches the itch that o-handed needed to deal with their customers who couldn't or wouldn't deal with the everything that OE has |
| 00:46.48 | prpplague | landley great i'll have a read |
| 00:47.32 | landley | ds2: the thing is, "we want to spend money to writing a multicall binary that we won't have to release the source to" is bog-standard proprietary development. |
| 00:47.34 | prpplague | CosmicPenguin well there are aspect of OE that don't make it business friendly |
| 00:47.36 | landley | I'm sure it's been done dozens of times. |
| 00:47.43 | CosmicPenguin | prpplague: sure - its a hackers world |
| 00:47.53 | landley | BSD doesn't enter into it, if you're writing it so you won't have to release the source, it's not open source. |
| 00:47.56 | CosmicPenguin | But that doesn't make poky a fork |
| 00:48.04 | landley | What's poky? |
| 00:48.15 | landley | I've heard of OpenEmbedded but can never remember what it is. |
| 00:48.20 | prpplague | http://www.pokylinux.org iirc |
| 00:48.20 | ds2 | landley: I see it differently... just write it and people can use it if they want |
| 00:48.27 | landley | OpenMoko is a Debian subset for ARM. |
| 00:48.30 | landley | Took me a year to figure that out. |
| 00:48.33 | CosmicPenguin | It serves a similar purpose as buildroot - it twists otherwise difficult code and makes it easier to consume |
| 00:48.37 | landley | ds2: that's BusyBox, yes. |
| 00:48.41 | ds2 | of course, I also expect them to chuck some money at it |
| 00:48.41 | landley | Already exists. |
| 00:48.45 | landley | And toybox. |
| 00:49.01 | landley | If they didn't modify it, all they need to say is "this is version X, unmodified". |
| 00:49.01 | ds2 | landley: busybox is nice but comes with strings attached... it is basically the old BSD vs GPL argument |
| 00:49.11 | landley | As far as I'm concerned, that's compliance. |
| 00:49.12 | landley | The lawyers file lawsuits when they don't respond. |
| 00:49.12 | prpplague | landley a type of build environment |
| 00:49.12 | landley | ds2: yup and GPL won hands down. |
| 00:49.21 | landley | Unless you count that bastion of openness, MacOS X. |
| 00:49.34 | landley | OpenDarwin's died twice now, hasn't it? |
| 00:49.44 | landley | FreeBSD split off NetBSD and OpenBSD and Dragonfly, and they've split off more... |
| 00:49.59 | landley | "Two BSD developers walk into a bar, and immediately fork." |
| 00:50.28 | ds2 | as long as that is followed by an exec... |
| 00:50.32 | landley | prpplague: I could guess that from context, the question is what is it? (An OE fork, sounds like?) |
| 00:50.44 | landley | The real problem with BSD licensed code is your developers can get hired away. |
| 00:50.55 | landley | Apple hired Jordan Hubbard away from FreeBSD, to work on a proprietary fork. |
| 00:51.09 | prpplague | landley yea, poky is basic and OE subset geared toward stability and business interests |
| 00:51.15 | landley | Just like Sun hired Bill Joy away from the original BSD all those years ago, to work on SunOS. |
| 00:51.27 | landley | Any BSD-licensed project that gets any sort of momentum has its developers hired away. |
| 00:51.31 | prpplague | landley its the build environment used for the nokia tablets iirc |
| 00:51.49 | landley | This doesn't happenw ith GPL-licensed code, because they'd have to be put to work on a different codebase if you want to close it up. |
| 00:52.20 | landley | You can throw all the money you want at Alan Cox or Christoph Hellwig, but if you want them workin on Linux it's GPL. |
| 00:52.26 | ds2 | that assumes the GPL code is well intermingled. primarily single owner code has been closed up and relicensed |
| 00:52.40 | landley | Otherwise they're reinventing the wheel rather than being paid to continue doing more or less what they've already been doing. |
| 00:52.52 | landley | ds2: sure, but the old version is still out there. |
| 00:53.02 | landley | Code written almost entirely by one developer, you can hire that developer away. |
| 00:53.07 | landley | Or buy Trolltech. |
| 00:53.12 | prpplague | hehe |
| 00:53.25 | ds2 | TCPMP is one of those with continue promises to release a proper free version |
| 00:53.40 | landley | QT, Mozilla, and Open Office are examples of projects that aren't open source collaborative development. |
| 00:53.47 | landley | The code is licensed open, but the development community isn't. |
| 00:54.01 | landley | So when Netscape released the mozilla source, you had three years of thrashing with negligible development. |
| 00:54.04 | landley | Same with OpenOffice. |
| 00:54.25 | landley | Those sort of things tend to turn into bloated nightmares, because the company goes "hey, free code" and bolts on all submissions. |
| 00:54.44 | ds2 | bloated? heh not like pure open source is any better |
| 00:54.47 | landley | When a project maintainer should be an editor sorting through a slush pile, fighting off sturgeon's law by saying "no" to 90% of it and asking for changes to the rest. |
| 00:55.22 | landley | ds2: the LInux kernel took 17 years to get as big as it is. |
| 00:55.31 | landley | Mozilla got bigger than that in 2 years, and does less. |
| 00:55.37 | ds2 | landley: the kernel is still pretty clean |
| 00:55.54 | ds2 | I am refering to all those projects that want to link in the kitchen sink + more |
| 00:56.03 | landley | If you don't understand the editorial process, you don't understand open source development. |
| 00:56.11 | landley | ds2: emacs is like mozilla and open office. |
| 00:56.20 | landley | You have to sign your code over to the FSF. |
| 00:56.29 | ds2 | X is not the best illustration of this but the current xlogo links font utilities, and rendering crap |
| 00:56.29 | landley | Open source development works like star trek fanzines in the 1970's. |
| 00:56.44 | landley | Ever magazine publisher understands open source. Submissions from the field, most of which are crap. |
| 00:56.52 | landley | But you've got to find the few worth putting in the next issue. |
| 00:57.11 | landley | XFree86 was another cathedral. x.org is fixing it. |
| 00:57.11 | prpplague | landley problem i have is when the process starts to break down, and patches start to be only looked at when the person is "known" to the maintainers |
| 00:57.14 | ds2 | just wish people would stop linking in 23421048023142390849023849023849023849023840923840923804923894 libraries into everything |
| 00:57.23 | landley | Remember that the cathedral in CATB was the FSF, not microsoft. |
| 00:57.37 | landley | Eric wrote the original paper comparing LInux to GNU. :) |
| 00:57.48 | MonMotha | ds2: in many cases, the reason they used a library was to NOT reinvent the wheel |
| 00:57.55 | prpplague | landley similiar to what you have seen with things like tcc |
| 00:58.00 | MonMotha | many commercial apps would do the same, but they can't because of licensing reasons |
| 00:58.15 | ds2 | MonMontha: I much rather have them reinvent the wheel then cause all the problems at run time |
| 00:58.32 | MonMotha | there's a reason Linux has library version management... |
| 00:58.40 | ds2 | a few months developer's work is worth it when it saves all the users time |
| 00:58.52 | davidc__ | ds2: erm... but that means that software would be far more buggy |
| 00:58.55 | MonMotha | not when that developer is unpaid or there are only a few dozen users |
| 00:58.58 | MonMotha | and yeah, then you have to QA your own library |
| 00:59.09 | MonMotha | presumably, the library you are linking to is at least somewhat tested |
| 00:59.12 | davidc__ | ooh, who opped me? Now I can take revenge on prpplague! |
| 00:59.16 | ds2 | I would much rather have that then to fight the dependency problem it creates |
| 00:59.31 | MonMotha | methinks you don't have proper package management |
| 00:59.40 | MonMotha | there's a reason almost all linux distros use packages |
| 00:59.43 | davidc__ | ds2: what dependancy problem? I've never had any dependancy problems when upgrading libs |
| 00:59.48 | ds2 | me thinks it is rediculous to have references to libraries for features I never use |
| 00:59.56 | prpplague | muhahaha |
| 01:00.08 | MonMotha | well, you have the same problem in windows, but every app brings its own libraries with it (either static or dynamic) |
| 01:00.14 | MonMotha | you still have unused code |
| 01:00.17 | *** mode/#edev [+o prpplague] by ChanServ |
| 01:00.21 | MonMotha | there's no good way around that... |
| 01:00.24 | *** mode/#edev [-o prpplague] by davidc__ |
| 01:00.27 | landley | ds2: environmental dependencies are a cost. |
| 01:00.30 | *** mode/#edev [+o prpplague] by davidc__ |
| 01:00.34 | landley | Eric and I wrote about that in "the art of Unix programming". |
| 01:00.38 | ds2 | explain why xlogo pulls in libXrender, libfreetype, libfontconfig, etc? |
| 01:00.41 | davidc__ | prpplague: heh, I'll wait till the moment is more promising ;) |
| 01:01.03 | landley | (My "editing pass" on that book doubled its size. I was almost named a co-author, but the book was Eric's baby, not mine.) |
| 01:01.10 | davidc__ | ds2: well, an easy solution is to use dlopen then |
| 01:01.14 | prpplague | davidc__ hehe |
| 01:01.17 | landley | prpplague: what have I seen with tcc? |
| 01:01.38 | *** mode/#edev [-o davidc__] by davidc__ |
| 01:01.41 | ds2 | davidc__: I can accept dlopen |
| 01:01.51 | landley | I remember back in 2000 seeing an article where the KDE guys were pointing out how insane it was that Gnome needed to link in 8 megabytes of shared libraries in order to beep the speaker. |
| 01:01.53 | landley | Back then, 8 megabytes was a _lot_. :) |
| 01:01.54 | MonMotha | to be fair, that's what a lot of media player apps do is dlopen libraries as needed |
| 01:02.11 | ds2 | to an extent but things like glibc takes that to rediculous levels |
| 01:03.02 | landley | Whatever happened to "do one thing and do it well"? |
| 01:03.23 | MonMotha | landley: people decided to emulate Windows, for some inexplicable reason |
| 01:03.46 | landley | With tcc? That's where 90% of the users of the thing are. Alas. |
| 01:03.50 | landley | Not the ones I'm interested in, though. |
| 01:03.50 | MonMotha | hell, I still run FVWM for my window manager. It is VERY good at one thing: window management |
| 01:03.52 | ds2 | landley: dillo |
| 01:04.15 | ds2 | MonMontha: yes, that's the kind of app we need. not the everything plus kitchen sink garbage |
| 01:04.26 | landley | prpplague: let me know when you read those three waves articles. |
| 01:04.36 | MonMotha | agreed, but haev you seen how many libraries FVWM links? :) |
| 01:04.39 | landley | If you were wondering why there's two types of open source projects, that will probably explain it to you. :) |
| 01:04.42 | ds2 | let me see |
| 01:04.50 | MonMotha | "Error writing hunk 524288 to output file: no error" |
| 01:04.52 | MonMotha | that's really informative... |
| 01:05.02 | ds2 | exactly 5 libraries |
| 01:05.11 | MonMotha | do you have some ancient build? |
| 01:05.19 | MonMotha | or did you build it with 90% of the features turned off/ |
| 01:05.22 | ds2 | probally... at least it is X11R6 ;) |
| 01:05.22 | MonMotha | because mine links like 35 |
| 01:05.40 | ds2 | ack...@#$@#!@%$#!@!@ must be the same thing that infected xlogo |
| 01:05.48 | MonMotha | all but a couple of which I can give a very good reason for |
| 01:05.58 | MonMotha | ds2: part of it is that X is now broken out into smaller libraries |
| 01:06.21 | ds2 | hmm |
| 01:06.53 | MonMotha | there's some GNOME libraries in there that I don't use, but that's just because this build has GNOME support |
| 01:07.02 | MonMotha | as in it can act as a Window Manager for GNOME if you want it to |
| 01:07.08 | landley | There's a reason busybox doesn't link against zlib. It could be smaller by not doing so, and the extra dependency sucks. |
| 01:07.10 | ds2 | YuCK |
| 01:07.15 | MonMotha | and that's only two libraries |
| 01:07.20 | landley | There's a reason busybox doesn't use curses: there's never any reason to use curses. |
| 01:07.33 | landley | Unix still has scar tissue from teletypes in 1975. |
| 01:07.42 | ds2 | curses are a wonderful thing |
| 01:07.45 | landley | These days, ever program outputs to an xterm which handles the output in software. |
| 01:08.12 | landley | So we're emitting data in dozens of different formats, and reading data in dozens of different formats... |
| 01:08.17 | landley | But it's software at both ends doing it. |
| 01:08.32 | landley | The _need_ for dozens of different formats went away around 1983. And yet sheer inertia... |
| 01:08.40 | landley | Just emit ANSI escape sequences and have done with it. |
| 01:08.46 | landley | Which is what busybox does. |
| 01:08.53 | landley | And thus, we don't need curses. |
| 01:09.17 | landley | (Yeah, there's logic to figure out which portions of your screen changed and efficiently emit just the redraws needed.) |
| 01:09.35 | landley | Screw that. Your serial port isn't 300 baud anymore. |
| 01:09.59 | MonMotha | actually, busybox redraws can be quite noticible |
| 01:10.06 | ds2 | *ahem* you don't some wireless links are effectively 300bps :( |
| 01:10.11 | MonMotha | I'll put up with them for the size, but they are very annoying at times |
| 01:10.27 | landley | You're referring to busybox vi? |
| 01:10.30 | davidc__ | landley: it can be 9600baud, at which redraws you can notice |
| 01:10.31 | ds2 | blah |
| 01:10.36 | ds2 | that sentance made no sense |
| 01:10.45 | MonMotha | landley: not just vi |
| 01:10.49 | MonMotha | but yes, vi is the worst |
| 01:11.02 | ds2 | I just drop in nvi ;) |
| 01:11.03 | landley | Know how to do ansi "scroll screen up" and "scroll screen down" commands would take care of most of that. |
| 01:11.35 | landley | Even DOS had those, if I remember correctly... |
| 01:12.20 | MonMotha | I do wish somebody would make a very stripped down curses, though |
| 01:12.38 | landley | Which is how I became so militant about the "just use ANSI already" approach. |
| 01:12.40 | MonMotha | something that would work with the existing library of curses based apps, but only support bare miniumum functionality needed to get them working |
| 01:12.51 | MonMotha | such as only supporting ANSI terminals |
| 01:12.54 | prpplague | landley sorry was doing a few honeydo's |
| 01:13.21 | landley | If I remember (it's been a while), 90% of the curses logic is screen scraping a buffer to determine the minimal updates. |
| 01:13.30 | prpplague | landley as to your question about tcc, i mean the attitude about not taking input from anyone outside the "inner circle" |
| 01:13.34 | Redhatter | What terminals are typically in use these days? vt100, ansi, "linux", xterm...? |
| 01:13.45 | landley | prpplague: I take input from anybody who sends it to me. :) |
| 01:13.58 | Redhatter | You could probably strip out 90% of the unnecessary ones, and just provide support for those in an add-on package. |
| 01:14.02 | landley | tcc had a problem that it wasn't taking input from _anybody_ for a year or so because Fabrice was busy with other things. |
| 01:14.05 | prpplague | landley no i mean the tcc guys |
| 01:14.18 | prpplague | landley i.e. "please do call us we'll call you" attitude |
| 01:14.23 | prpplague | s/do/don't |
| 01:14.30 | landley | prpplague: ? |
| 01:14.35 | landley | I missed this. |
| 01:14.57 | landley | I was following the project circa 2004 and then when I started paying attention again it had more or less died. |
| 01:15.16 | landley | or do you mean the current group now that I've left the old list? |
| 01:15.33 | landley | That was more a problem with the CVS tree being the one true repository. |
| 01:15.49 | landley | It's CVS. |
| 01:16.03 | landley | It's a drug store, not a source control system. |
| 01:16.31 | prpplague | landley i.e. you being asked not to contribute anymore |
| 01:18.06 | landley | prpplague: that was mostly an idiot mouthing off. |
| 01:18.14 | landley | He wasn't even a serious contributor to the project. |
| 01:18.24 | landley | I could easily have ignored him, I was just sick of the stupid CVS tree hanging over my head. |
| 01:18.36 | landley | It's a real pain to see what people have checked into a CVS tree. |
| 01:18.53 | prpplague | landley agreed, but from reading the mailing list i got the feeling they didn't like you mucking around in their "business" |
| 01:18.54 | landley | If the CVS tree started up again, then my tree would get out of sync with it. Guaranteed. |
| 01:19.10 | landley | I'm happy to ignore them. |
| 01:19.10 | prpplague | landley to me, thats the way alot of projects have become, very teritorial |
| 01:19.29 | prpplague | i.e. the gpe vs. handhelds.org battle |
| 01:19.29 | landley | Since I left, they've started up their own mercurial tree, and copied lots of my changes into their CVS tree. |
| 01:19.45 | landley | They've done essentially zero new development. I've done a couple dozen checkins. |
| 01:19.48 | prpplague | landley yea, i;ve been following it |
| 01:20.19 | landley | I'm trying to have a new release by the end of the year, they've talked about having one with no deadline. |
| 01:20.26 | landley | Mine will be 0.9.25, theirs 0.9.24. :) |
| 01:20.33 | prpplague | hehe |
| 01:20.50 | landley | Mine works on an x86-64 host, theirs won't even build on that. |
| 01:21.03 | landley | I need to fix static linking (broken by updates to Ubuntu, not by changes I made). |
| 01:21.14 | prpplague | hopefully arm soon |
| 01:21.20 | landley | libc.a on Ubuntu is full of symbols that live in libgcc_s.so. |
| 01:21.20 | landley | Sigh. |
| 01:21.47 | landley | I need to come up with a libtcc_unbreak_glibc.a. |
| 01:22.09 | landley | Providing stubs for all these stupid symbols that libc.a leaked because it's never been tested with anything but gcc. |
| 01:22.09 | landley | Grrr. |
| 01:22.16 | ds2 | stock glibc cannot produce a static binary :( |
| 01:22.29 | landley | I don't think you could statically link against Ubuntu's libc.a with gcc-2.95 anymore. |
| 01:23.29 | landley | Well, work around it. |
| 01:24.05 | landley | Ideally I want to provide one stub and then _weak link to it for all the leaked symbols. :) |
| 01:24.10 | prpplague | landley i'm not much into userland and compilers, however something that has interested me of late is scriptable language to easy i/o operations |
| 01:24.14 | landley | I can probably do that in libtcc.a. |
| 01:24.15 | prpplague | landley i.e. with robotics |
| 01:24.31 | landley | C makes an excellent scripting language if you want to do low level stuff. :) |
| 01:24.32 | prpplague | landley and in general with devices that might be used in space craft |
| 01:24.47 | landley | Right now -run works fine on x86. |
| 01:24.57 | prpplague | landley really? didn't know that |
| 01:25.03 | ds2 | the complaint stemps from the fact glibc insist on using libnss for everything |
| 01:25.04 | prpplague | learn something new every day |
| 01:25.04 | landley | But not on x86-64, for obvious reasons if you stop and think about it. |
| 01:25.28 | landley | The problem is, if you run the x86 compiler on an x86-64 host, it's a 64-bit binary which generates a buffer full of 32-bit code and then calls it as a function pointer. |
| 01:25.32 | prpplague | landley i have a bizare dream of a generic computing brick |
| 01:25.35 | landley | 32 bit code won't run in a 64 bit program. Segfault. |
| 01:25.50 | landley | Not tcc's fault, it needs a 64-bit backend. |
| 01:25.57 | landley | -run only works for a fully native compileer. |
| 01:26.30 | landley | Right now, tcc runs on an x86-64 host, but hasn't got an x86-64 code generator yet. |
| 01:26.35 | landley | So it still outputs 32-bit binaries. |
| 01:26.39 | landley | Excuse, tinycc does. |
| 01:26.43 | landley | tcc doesn't even compile on x86-64. |
| 01:27.07 | prpplague | hehe |
| 01:31.14 | prpplague | landley you every use LOGO? |
| 01:33.47 | prpplague | CosmicPenguin hey isn't their a logo app on the XO? |
| 01:37.53 | landley | prpplague: long, long ago. |
| 01:44.24 | GPSFan | prpplague: the wifi drivers in the Z2 are authored by marvel and you can find the srting "license=GPL" in both the gspi.ko and gspi8686.ko files. |
| 01:45.24 | GPSFan | prpplague: we'll see what they say about it, they have made the statement that they probably won't be able to release the source for the wifi driver. |
| 01:46.20 | GPSFan | bbl.. |
| 01:54.48 | prpplague | GPSFan OH wow |
| 01:55.15 | prpplague | GPSFan spi based 8686 |
| 01:55.31 | prpplague | GPSFan marvell knows thats a violation of gpl |
| 01:55.50 | prpplague | GPSFan about 90% of the code is GPL, but the section for the spi is not |
| 01:56.11 | prpplague | GPSFan the libertas code in the main tree will work with the 8686 |
| 01:56.25 | prpplague | GPSFan but you'll need to strip the firmware from the binary |
| 01:56.35 | prpplague | GPSFan and you'll need to add the spi interface code |
| 01:56.55 | prpplague | GPSFan currently the 8686 driver only supports sdio |
| 01:57.07 | MonMotha | holy crap, SPI wifi? |
| 01:57.20 | prpplague | MonMotha indeed most sdio chips also support spi |
| 01:57.49 | prpplague | MonMotha problem is it's like pulling teeth to get info about it |
| 01:58.38 | ds2 | wished atheros SD cards were more prevalent and cheaper |
| 01:58.54 | prpplague | ar6k supports spi too |
| 01:59.48 | prpplague | GPSFan can i get a copy of those kernel modules? |
| 02:00.51 | prpplague | GPSFan you might want to get ossman and dan williams involved, because their not suppose to distribute those modules in binary only form |
| 02:00.59 | prpplague | GPSFan and marvell knows that |
| 02:01.42 | flyback- | landley: vm :P |
| 02:03.14 | flyback- | prpplague: anyway to piggback or daisy chain a sd slot |
| 02:03.45 | prpplague | flyback- no the driver assumes you only have on device on the spi bus |
| 02:04.02 | flyback- | I mean't if you modded the driver also |
| 02:04.11 | flyback- | like on my palm so I could have a wireless card and flash |
| 02:04.40 | prpplague | yea you could but your wifi performance is going to suffer |
| 02:05.14 | flyback- | i'll probably just stick with bt then |
| 02:05.26 | flyback- | and get one of those $11.99 routers to use as a ether to bt bridge |
| 02:05.37 | flyback- | mabye even put in a usb hub and have it be a wireless to bt bridge |
| 02:06.01 | MonMotha | drivers that assume only one thing is on the SPI bus need to die |
| 02:06.12 | MonMotha | last design I did had 4 SPI devices and only 2 busses |
| 02:06.20 | prpplague | MonMotha well its not always practical as far as bandwidth goes |
| 02:06.38 | MonMotha | it needs to *work* |
| 02:06.46 | MonMotha | it doesn't need to work well, but it needs to work |
| 02:07.12 | MonMotha | that said, 802.11g is only 54Mbps "wire" speed, so 60MHz SPI should have plenty of room for that and another device |
| 02:07.16 | prpplague | MonMotha indeed, well, the issue also is this code isn't released in code form anyway, its trash code |
| 02:07.20 | flyback- | prpplague: I assumed I would take a nasty hit |
| 02:07.25 | flyback- | but it might be worth it in some cases |
| 02:07.44 | prpplague | MonMotha problem is that most SoC's , i.e. the s3c24xx , the max spi rate is 25MHZ |
| 02:07.45 | flyback- | the other option is |
| 02:07.53 | flyback- | the palm has a serial bus and usb bus on the connector |
| 02:08.01 | flyback- | dunno the speed limits of those |
| 02:08.03 | MonMotha | well, that's their problem :) |
| 02:08.09 | MonMotha | my AT91 can clock the SPI faster |
| 02:08.13 | flyback- | apprentely intel pxa216 usb is OTG |
| 02:08.20 | flyback- | so it's possible my palm could drive a usb wireless device |
| 02:08.22 | MonMotha | 1/2 IO clock, IIRC, which would mean 50MHz |
| 02:08.32 | flyback- | unless palm canucked it to death somehow |
| 02:08.59 | MonMotha | and it does DMA |
| 02:09.13 | flyback- | prpplague: I took apart one of my nicd/nimh smart battery chargers I got for $8 last yr which included a car adaptor and 4 2000MA AA's |
| 02:09.29 | flyback- | it uses a little 4 bit mcu with self programming OTP |
| 02:09.35 | flyback- | i thought that was pretty neat and cute :) |
| 02:09.53 | MonMotha | hum, maybe that's only on the SAM7 (which would be weird) |
| 02:10.02 | flyback- | there might be hope for the industry just yet :) |
| 02:10.07 | prpplague | MonMotha might want to double check |
| 02:10.33 | prpplague | MonMotha the arm module that most manufacturers buy to includin in their core for spi is just 25mhz |
| 02:11.13 | MonMotha | prpplague: actually, max is MCK/1 |
| 02:11.31 | MonMotha | with no indication that it can't hit the full 100MHz |
| 02:11.37 | prpplague | MonMotha interesting |
| 02:11.39 | MonMotha | Atmel makes dataflash chips capable of like 60, so that isn't surprising |
| 02:11.52 | prpplague | indeed they may be using something they had |
| 02:12.05 | prpplague | most companies like samsung just use the arm modules |
| 02:13.10 | MonMotha | "Synchronous Communications at Up to 90Mbits/sec" |
| 02:13.25 | MonMotha | that's quoting the description of the SPI module |
| 02:13.42 | MonMotha | I'm guessing you can safely hit 100 under most conditions |
| 02:13.49 | MonMotha | sounds like it was designed for 100, but didn't make it at worst case |
| 02:13.52 | prpplague | wow thats interesting |
| 02:14.08 | flyback- | did you see the new intel doc thing |
| 02:14.10 | MonMotha | IIRC, the SAM7 series only hits 50 (MCK/1, but MCK is limited to 50 on those) |
| 02:14.22 | MonMotha | still, very fast SPI |
| 02:14.31 | MonMotha | makes routing the SPI nets a challenge |
| 02:14.36 | prpplague | indeed |
| 02:14.39 | MonMotha | you practically need controlled impedance and termination |
| 02:14.45 | flyback- | hey prpplague how well does your arm board handle soft dsp |
| 02:14.50 | flyback- | this might be a fun project for one |
| 02:14.56 | flyback- | needs a soundcard or at least a adc though |
| 02:15.00 | prpplague | no clue, not something i've ever done |
| 02:15.04 | flyback- | lemme find it |
| 02:15.09 | flyback- | I am going to do this on my pc asap |
| 02:15.20 | prpplague | flyback- you can use a hiband filter with the pwm to generate audio |
| 02:15.30 | flyback- | no audio in |
| 02:15.32 | flyback- | not out |
| 02:15.35 | prpplague | flyback- the quality is just around 22khz sample rate |
| 02:15.38 | flyback- | http://en.wikipedia.org/wiki/Digital_Radio_Mondiale |
| 02:15.38 | prpplague | flyback- oh |
| 02:16.13 | prpplague | flyback- the hammer could do a DRM , but it would have to be tweaked |
| 02:16.48 | MonMotha | prpplague: all that said, I'm not sure I can recommend the 91SAM9263 for a design, at least not until they get the errata worked out |
| 02:16.53 | MonMotha | there's some REALLY nasty errata |
| 02:17.12 | MonMotha | in particular, booting the damn thing is not trivial.. |
| 02:17.18 | flyback- | prpplague: I did a spin test in the parking lot of one of my windmills I built |
| 02:17.33 | MonMotha | it can only boot from NOR Flash (or mask rom, or whatever) or Dataflash |
| 02:17.46 | flyback- | it lit up 3 white leds in a moderate wind and that's without rectifiers or caps which will 10x the performance |
| 02:17.51 | flyback- | not bad for about $5 in junk :) |
| 02:17.52 | MonMotha | don't be fooled by the datasheet claiming SD card boot and NAND boot: it doesn't work (per the errata) |
| 02:17.57 | ds2 | Mmmmmm dataflash boot |
| 02:18.12 | prpplague | MonMotha interesting |
| 02:18.13 | ds2 | they do make dataflash in a SD formfactor ;) |
| 02:18.23 | flyback- | prpplague: my idea being small windmills for remote areas or places that don't have reliable power |
| 02:18.31 | flyback- | to be able to charge up some batteries |
| 02:18.57 | flyback- | then a low power mcu or devel board like the hammer can power up, get as much work done as it can on a charge and go back to sleep |
| 02:19.05 | prpplague | flyback- might check out the circuit cellar archives, there is a great article on a small board to help manage solar and wind power voltage |
| 02:19.06 | flyback- | also it's just bragging rights :P |
| 02:19.17 | MonMotha | ds2: yes, and you can use that |
| 02:19.26 | MonMotha | the physical nature of the chip is also TERRIBLE |
| 02:19.32 | MonMotha | 0.8mm BGA with no center cutout, 324 pads |
| 02:19.35 | flyback- | I got a canuckload of caps from motherboards so I can get a really nice buffer between the gen and the batteries |
| 02:19.38 | ds2 | flyback-: come back when prpplague's pals release the version with power management ;) |
| 02:19.44 | flyback- | so that I catch every little twitch of the gen |
| 02:19.53 | MonMotha | giant 18x18 BGA, and the pinout sucks |
| 02:19.55 | flyback- | ds2 well my gen is still in raw form anyways :) |
| 02:19.57 | flyback- | but thx |
| 02:20.18 | ds2 | I been thinking of a potatoe powered Linux system |
| 02:20.32 | ds2 | but finding a chip that is happy running at 5-10mA and have a MMU + decent memory is hard |
| 02:20.44 | flyback- | used some steel spine with aluminum blades from a fan and a ibm proprinter sanko denki step syn stepper |
| 02:20.59 | flyback- | it gets about 15 vac/30vac per one of 2 coils in a moderate wind |
| 02:21.20 | flyback- | hmm that reminds me I got batteries tonight lemme fire up my cam and upload some pics |
| 02:22.28 | landley | HA! |
| 02:22.41 | landley | An hour and a half of the _weirdest_ debugging, and what do I find? |
| 02:22.51 | landley | "killall minicom.sh" kills the wrapper script, but not the toybox instance. |
| 02:22.59 | flyback- | hoffa's body in c? |
| 02:23.06 | landley | So even though I got my shell prompt back, old instances of netcat were still listening to the console. |
| 02:23.32 | landley | So I was getting debugging printfs that weren't in the current version (very confusing), and hangs as the key I'd just typed was read back by one of the other instances... |
| 02:23.49 | landley | Yet another fun way to break things... |
| 02:24.14 | flyback- | heh |
| 02:25.30 | prpplague | landley hehe |
| 02:26.07 | flyback- | oh prpplague |
| 02:26.10 | flyback- | you will like this |
| 02:26.16 | flyback- | some company released a mini-itx board |
| 02:26.26 | flyback- | what is insteresting about it and applies to you guys |
| 02:26.31 | flyback- | they ran it underwater np |
| 02:26.40 | flyback- | some kinda protective coating tech |
| 02:26.53 | ds2 | inside an otter box? |
| 02:27.00 | flyback- | eh? |
| 02:27.02 | flyback- | otter? |
| 02:30.21 | flyback- | prpplague: how are sales btw? |
| 02:31.55 | ds2 | a la USBstix? |
| 02:31.56 | landley | If I could take the sucker with me, I'd have a much easier time playing with it and developing for it... |
| 02:33.03 | prpplague | landley easy enough |
| 02:33.14 | prpplague | landley takes about two minutes to wire up |
| 02:33.23 | landley | Not for me. :) |
| 02:33.41 | prpplague | landley remind me tomorrow, i'll get some proto's made |
| 02:33.47 | landley | ok. Thanks. |
| 02:33.58 | prpplague | landley anything else you want on the board? |
| 02:34.03 | prpplague | landley nows the time to ask |
| 02:34.09 | landley | hmmm? |
| 02:34.15 | landley | Ah. |
| 02:34.27 | ds2 | why about a little mint? ;) |
| 02:34.41 | landley | When it plugs in, could it present a USB and a JTAG interface, the way the flyswatter board does when it's plugged in? |
| 02:34.44 | ds2 | say, one that makes gold dimes ;) |
| 02:34.55 | landley | What I'm looking for is a portable development thingy. |
| 02:35.20 | landley | A USB plug on the other end would be marvelous too, but might be pushing it. :) |
| 02:35.30 | landley | (That way I could plug an ethernet thing into it the way I'm doing now.) |
| 02:35.36 | landley | Just wire it back around to the end of my laptop... |
| 02:35.44 | prpplague | landley well, i can add a usb host on one side, and a usb slave on the other |
| 02:35.46 | landley | With a 1 foot cat5 cable. |
| 02:35.59 | landley | Cool. |
| 02:36.10 | landley | Basically what I want to do from a use case perspective: |
| 02:36.12 | prpplague | landley when you plug in the usb slave the device powers up and presents itself as a serial port |
| 02:36.19 | landley | I want to take this sucker with me, wherever I take my laptop. |
| 02:36.25 | landley | Throw it in my backpack along with the laptop. |
| 02:36.35 | prpplague | completely doable |
| 02:36.37 | landley | I want to plug it in and be able to re-flash it if I've built a new image. |
| 02:36.41 | flyback- | mints even |
| 02:36.59 | landley | I don't want to have to carry around a separate power cord, both from a fiddly bits perspective and because my laptop is often running off battery. |
| 02:37.13 | prpplague | landley you can reflash anytime the kernel and flash with the exception of the bootloader |
| 02:37.19 | landley | I want to be able to get a console from the device, through the USB port, or I can't talk to it. |
| 02:37.24 | landley | prpplague: probably good enough. |
| 02:37.30 | landley | ALthough brickability is still a concern. |
| 02:37.36 | prpplague | indeed |
| 02:37.51 | prpplague | i'll run a couple ideas around tomorrow |
| 02:37.52 | landley | And then I need the USB plug on the other end so I can put network onto it. |
| 02:37.56 | flyback- | landley: you are welcome to drill #electronics |
| 02:37.58 | flyback- | for any ideas |
| 02:37.59 | landley | Because a device without ethernet is much less interesting... |
| 02:38.30 | landley | I suppose if it simply _had_ internet (10baseT even) on the other end, that would be plenty. |
| 02:38.42 | landley | Heck, I could see people making little portable linksys-like devices for windows with that. |
| 02:38.46 | landley | "network condom". |
| 02:38.54 | prpplague | hehe |
| 02:38.57 | prpplague | one sec phone |
| 02:38.57 | flyback- | prpplague: |
| 02:39.12 | flyback- | what about one of those $10 routers |
| 02:39.20 | prpplague | ~lart flyback- |
| 02:39.21 | purl | eats flyback-'s liver with some fava beans and a nice chianti |
| 02:39.29 | flyback- | no hear me out |
| 02:39.40 | flyback- | to feed the hammer a serial link and give it ether |
| 02:39.51 | flyback- | if he doesn't want to carry it around |
| 02:40.05 | landley | I want to carry the hammer board around. |
| 02:40.08 | flyback- | guess youc ould do that with a pc too |
| 02:40.09 | flyback- | nm |
| 02:40.10 | prpplague | landley any specific style of case you want? |
| 02:40.39 | landley | prpplague: something vaguely enclosed that takes a certain amount of effort to break? |
| 02:41.03 | landley | By which I mean throwing it in a side pocket of my backpack wouldn't be deeply inadvisable. |
| 02:41.42 | landley | I dunno what the packaging options are. USB key? The "blob with cord sticking out of it" type? |
| 02:41.49 | flyback- | do think about esd if you do that |
| 02:41.52 | flyback- | even in a case |
| 02:41.59 | flyback- | unless the case is esd proof |
| 02:42.22 | landley | As long as the exposed metal is just USB ports, it should be ok. |
| 02:42.29 | prpplague | landley well i'll probably stick it in a generic case for starters , but if its something people are interested in i'll have our IE do a solid works design |
| 02:42.51 | landley | Cool. |
| 02:42.57 | landley | I'm looking at the actual hammer board. |
| 02:43.01 | flyback- | prpplague: quick question |
| 02:43.07 | prpplague | landley 10baseT enough? |
| 02:43.14 | flyback- | did I tell you about the fabric softner research I did ? |
| 02:43.22 | prpplague | flyback- uh no |
| 02:43.28 | landley | If that was enclosed in plastic with a USB plug on one end, my main concern would be it not resting on the table properly and breaking if it was leaned on. |
| 02:43.38 | flyback- | if you take fabric software dillute with water |
| 02:43.46 | flyback- | and apply to carpets, plastic containers, etc |
| 02:43.48 | landley | 10baseT is probably plenty. |
| 02:43.53 | flyback- | it forms a static disappative coating |
| 02:43.58 | landley | That's 1.1 megabytes/second. |
| 02:44.17 | flyback- | I did that on some seats at work cause I was getting nasty bolts and was scared for customer gear |
| 02:44.44 | prpplague | landley main problem is going to be power, i'll have to look, some laptops limit their usb ports to 100mA |
| 02:45.13 | landley | USB spec says 2 amps, but I doubt you can expect that from anything battery powered. :) |
| 02:45.14 | prpplague | landley with the hammer plus a 10baseT you are going to be looking at about 250mA |
| 02:45.31 | landley | 100baseT takes more power than 10baseT? |
| 02:45.32 | MonMotha | prpplague: USB spec allows that when not on battery? |
| 02:45.47 | prpplague | MonMotha non battery is 500mA per port |
| 02:45.50 | MonMotha | (and no other devices connected) |
| 02:45.51 | MonMotha | ah, ok |
| 02:45.54 | prpplague | battery is 100mA |
| 02:46.06 | MonMotha | that makes more sense |
| 02:46.30 | prpplague | thats a powered hub |
| 02:46.37 | MonMotha | 500mA max per port |
| 02:46.39 | prpplague | usually 4 ports * 500mA |
| 02:46.41 | landley | What does it take to get the generic I/O stuff to fake 10baseT in software? |
| 02:46.46 | MonMotha | and that's only after you ask for permission |
| 02:46.48 | flyback- | no |
| 02:46.52 | flyback- | it's 100ma |
| 02:46.56 | landley | I vaguely remember something doing that... |
| 02:46.57 | prpplague | landley pretty difficuly |
| 02:46.59 | flyback- | then the device is supposted to ask for the 500ma max |
| 02:47.01 | MonMotha | I would not advise bitbanging ethernet... |
| 02:47.06 | prpplague | landley but the enc28j60 is cheap |
| 02:47.13 | flyback- | but a lot of ignorant cheap chinese usb toys just rape the port anyways |
| 02:47.33 | prpplague | http://www.elinux.org/Enc28j60 |
| 02:47.49 | landley | Query: how much of that power is the ethernet and how much the hammer? |
| 02:47.50 | MonMotha | prpplague: and a breeze to use! :) |
| 02:48.15 | flyback- | prpplague: |
| 02:48.17 | flyback- | beat me too it :) |
| 02:48.25 | prpplague | landley hammer via usb is going to be about 125 and the enc28j60 about 125 average |
| 02:48.34 | prpplague | landley should be less than 500mA |
| 02:48.44 | prpplague | landley i guess you could use two usb ports |
| 02:48.52 | flyback- | prpplague: |
| 02:49.05 | flyback- | man that's like 3 hrs on a high end AA nimh 3a |
| 02:49.19 | flyback- | well not including the loss due to voltage boost |
| 02:49.28 | prpplague | no boost needed |
| 02:49.28 | flyback- | so you would need 3 or so |
| 02:49.36 | landley | prpplague: I suppose you could draw 100 amps with the arm clocked down a bit, and then ask for more. |
| 02:49.37 | flyback- | it will run on 1.2 ? |
| 02:49.46 | landley | If you get it, clock up the board and power up the ethernet... |
| 02:50.12 | prpplague | landley i'll do the test in the morning at the office, just be a few minutes to wire up and test |
| 02:50.25 | landley | I suppose many laptops _will_ give more than 100 when running on battery power, because some devices just won't work otherwise. |
| 02:50.36 | landley | Ok. I look forward to hearing about the results. |
| 02:51.10 | prpplague | landley if you can, try this, plug in the most power hungry usb device you have to you laptop and the cat the /proc/bus/usb/devices file |
| 02:51.11 | MonMotha | most devices which are at all aware of current limits on their USB ports will turn you off if you try to draw appreciably moer |
| 02:51.19 | landley | I suppose the jtag pins would still be accessible if you opened the case. |
| 02:51.20 | prpplague | should list the max power for all the devices |
| 02:51.39 | prpplague | landley indeed, i think i can work out something |
| 02:51.50 | flyback- | wow prpplague |
| 02:51.57 | flyback- | that ether chip even has wake on lan |
| 02:51.58 | flyback- | nice |
| 02:52.00 | prpplague | landley i have an idea cranking |
| 02:52.16 | flyback- | landley: so you can have the chip go to sleep when no data going in/out |
| 02:53.26 | landley | Possibly add a small capacitor to make the laptop's USB cry less. |
| 02:54.08 | prpplague | landley i can do some tweaking of the power requirements |
| 02:54.26 | MonMotha | wait, doesn't the hammer use a linear regulator? |
| 02:54.27 | landley | How do I see power consumption figures in the usb/devices file? |
| 02:54.37 | prpplague | MonMotha yea its a standard ldo to 3.3v |
| 02:54.43 | MonMotha | hum, only 3.3... |
| 02:54.44 | prpplague | MonMotha which gobbles power |
| 02:54.49 | landley | MonMotha: I was thinking if the ethernet chip has wake on lan and powers down between packets. |
| 02:54.52 | MonMotha | hum... |
| 02:54.57 | MonMotha | you may be able to get a switcher for that |
| 02:55.02 | prpplague | one sec kids duty |
| 02:55.14 | MonMotha | would improve efficiency, but only 1.7V isn't much for a buck |
| 02:57.00 | MonMotha | if you could get a switching solution to work, 80-95% eff. isn't uncommon (caveat here being the low Vin), much better than 66% for your LDO |
| 02:57.11 | MonMotha | would ease the current requirements and might let you sneak in under 100mA |
| 02:57.39 | landley | Last time I did threading was under Java, and OS/2 before that. |
| 02:58.11 | landley | But bunzip is easy to break into parallel chunks (each 900k block is totally independent, and the run length encoding pass that finds blocks isn't very cpu intensive). |
| 02:58.30 | landley | So I could double the speed of bunzip on my core 2 duo by having both processors involved in decompression. |
| 02:58.39 | landley | But this involves threading... |
| 02:58.52 | MonMotha | bzip2 is block based, so that shouldn't be too hard |
| 02:59.24 | MonMotha | ah |
| 02:59.34 | landley | Mine's under 7k, and 10% faster than Julian Seward's. |
| 02:59.38 | landley | Plus, you can _read_ mine. :) |
| 02:59.50 | landley | Had to learn what a burrows-wheeler transform was, but I was bored. |
| 02:59.56 | prpplague | landley the proc entry should have a mxpwr entry |
| 02:59.57 | landley | It was either that or do grad school homework... |
| 03:00.21 | landley | mxps, perhaps? |
| 03:00.22 | prpplague | landley MxPwr= |
| 03:00.40 | prpplague | yea it might be listed that way depending on the kernel |
| 03:01.02 | landley | Ah, found it. |
| 03:01.17 | landley | Ok, flyswatter is 200mA. |
| 03:01.33 | landley | My usb/serial is 100mA. |
| 03:02.00 | landley | My USB/ethernet is 156mA. |
| 03:02.05 | landley | That's a remarkably specific figure... |
| 03:02.33 | prpplague | landley the device has to enumerate the max power config |
| 03:02.33 | landley | Another point is I really doubt we need the full 1500 feet of cat 5 the spec allows. :) |
| 03:02.56 | prpplague | landley most companies max it out to round numbers, i.e. 50,75,100,150,200 |
| 03:03.05 | prpplague | landley others seem to be pretty exact |
| 03:03.21 | prpplague | landley your laptop ok with the 200mA for the flyswatter? |
| 03:03.32 | landley | Seems to be, but I'm runnin gfrom wall current at the moment. |
| 03:03.41 | landley | (THis is the non-portable flyswatter, so I'm at home on the kitchen table.) |
| 03:03.50 | prpplague | landley hehe |
| 03:04.40 | prpplague | i suspect that it will be ok portable, but i'll see what i can do to cut down the power |
| 03:05.09 | landley | Woot. |
| 03:05.23 | landley | I can do docs for "So you've never done embedded development before? Here's how you start!" |
| 03:05.31 | landley | I'm looking for hobbyists who want to get _into_ this kind of thing. |
| 03:05.42 | landley | Including those like me who really aren't at home with a soldering iron. |
| 03:06.16 | MonMotha | prpplague: have you guys looked into a switching solution for the power on that thing? |
| 03:07.49 | prpplague | hmm this is a little too big - http://www.pactecenclosures.com/Plastic-Enclosures/PP.html |
| 03:07.58 | landley | An ARM device you can plug into your USB port and run Linux on is perfect for this. |
| 03:08.34 | landley | Um... I guess? |
| 03:08.36 | prpplague | landley hehe, seriously if you want to do it, we can go thru lulu |
| 03:08.40 | landley | Can't judge scale from the picture. |
| 03:08.53 | prpplague | landley sorry i always have calipers to measure, hehe |
| 03:09.32 | prpplague | landley http://www.lulu.com/ print on demand services |
| 03:09.39 | MonMotha | prpplague: oh, do you know where I might find an off-the-shelf "usb keyfob" enclosure? |
| 03:10.03 | prpplague | MonMotha pactec makes them as well as okwusa |
| 03:10.08 | prpplague | MonMotha one sec |
| 03:10.41 | landley | lulu I know. |
| 03:10.51 | landley | I'm pondering inviting Robert Young to be a GoH at Penguicon one of these years. |
| 03:11.32 | landley | The old hawking USB adapter I have has ~6 inches of USB cable, and then a largeish wart thing. |
| 03:11.35 | landley | It's something like 5 years old. |
| 03:12.12 | landley | That case would be about big enough to hold the hammer, and is very unlikely to break due to the laptop being too tall or too flat, but I dunno if anybody makes that sort of thing anymore. |
| 03:12.15 | landley | Or how much it would cost. |
| 03:12.18 | prpplague | http://www.okwusa.com/products/okw/smart.htm |
| 03:12.20 | landley | (This sucker was expensive when I got it.) |
| 03:12.47 | MonMotha | prpplague: any of those have cutouts for the USB 'A' connectors |
| 03:12.48 | MonMotha | ? |
| 03:12.48 | landley | Cool. |
| 03:12.51 | joe_bleau | How about http://www.newageenclosures.com/default.aspx ? |
| 03:13.24 | prpplague | MonMotha not directly, easy enough to get a punch made |
| 03:13.26 | landley | The bottom-right one looks interesting. |
| 03:13.36 | MonMotha | ah, like that |
| 03:14.04 | prpplague | landley we can get a cheap tool made for about 4k and that'll make the parts around $1-2 each |
| 03:14.14 | prpplague | landley which url? |
| 03:14.44 | prpplague | landley ahh |
| 03:14.50 | prpplague | landley yea i see now |
| 03:15.03 | prpplague | joe_bleau those guys want to do the engineering too |
| 03:15.18 | joe_bleau | Min order, $50. |
| 03:15.26 | landley | http://www.newageenclosures.com/ProductInfo.aspx?productid=I-401108 might fit |
| 03:15.35 | prpplague | interesting |
| 03:16.09 | prpplague | problem is that i'll need to get an additional pcb in there |
| 03:16.38 | landley | An additional PCB the size of the hammer itself? |
| 03:16.47 | prpplague | landley yea |
| 03:17.02 | prpplague | landley with additional connections and electronics |
| 03:17.11 | landley | Because if they can piggy-back or go face to face on the pinouts, then you have a rectangle. |
| 03:17.15 | prpplague | landley basically two hammer boards pigback each other |
| 03:17.30 | landley | Ok. |
| 03:17.45 | landley | How much would 2-3 inches of USB cable add to the cost? |
| 03:18.03 | landley | You can get away with a much bigger enclosure if it's on a tether instead of its weight hanging from the laptop. |
| 03:18.23 | landley | (Speaking as an end-user. :) |
| 03:18.38 | landley | Wouldn't fit i pants pocket, would still fit in jacket pocket. |
| 03:19.33 | prpplague | landley probably easiest to get a short usb host to slave cable, i.e. 6 inches, and on the hammer have a usb mini connector |
| 03:19.42 | prpplague | landley i.e. whats on a camera |
| 03:19.50 | prpplague | landley or do you think thats too small? |
| 03:20.42 | prpplague | http://www.sparkfun.com/commerce/product_info.php?products_id=8401 |
| 03:20.46 | landley | 6 inches is plenty. |
| 03:21.02 | landley | It's not a question of needing length, it's a question of being able to fit it into a wide range of laptops. |
| 03:21.24 | MonMotha | joe_bleau: you know any distributors for them that don't have a minimum order? |
| 03:21.27 | landley | The direct-connect dongles have a goldilocks problem trying to rest on the table. |
| 03:22.33 | landley | You could even have the camera-type USB connector be embedded in epoxy in the "lump" part of it. |
| 03:22.40 | landley | It doesn't have to come off. |
| 03:22.52 | landley | The cable is just to ease connecting a larger lump up to your laptop. |
| 03:23.01 | landley | Without being quite so awkward. |
| 03:23.02 | prpplague | indeed |
| 03:23.19 | prpplague | i can have a rough prototype done tomorrow |
| 03:23.20 | landley | Your link says 100mA, though. |
| 03:23.29 | landley | Back to "hmmm, power limits". |
| 03:23.30 | prpplague | which link? |
| 03:23.35 | landley | the sparkfun one? |
| 03:23.38 | landley | above. |
| 03:23.45 | prpplague | oh, let me look missed that |
| 03:23.55 | landley | Also good for an un-regulated, non-limited 5V supply up to 100mA. |
| 03:24.16 | landley | Maybe it can be used above that and just needs... cooling or something. |
| 03:24.51 | prpplague | landley thats just something they suggested, the connector is good up to 1A |
| 03:24.56 | landley | Oh good. |
| 03:24.59 | landley | Way over. :) |
| 03:25.13 | joe_bleau | MonMotha: nope, but I'm sure they'll sample a few... |
| 03:25.53 | landley | prpplague: how much extra work would it be to also have a jtag in the thing like flyswatter does? |
| 03:26.00 | landley | Because I brick. |
| 03:26.23 | landley | Today's "why do I have 15 netcat instances hanging off the same serial port" thing is _normal_. |
| 03:26.29 | cajaka | I have been reading back on what you guys are talking about was wondering what all would be involved. I am glad to see it disscused. |
| 03:26.33 | landley | I get into situations nobody's ever _seen_ before, as a matter of course. :) |
| 03:27.22 | prpplague | landley not bad at all |
| 03:27.44 | landley | Earlier somebody mentioned mounting jffs2 via loopback? |
| 03:27.45 | prpplague | landley just reverse the jtag pins to bottom side |
| 03:27.58 | prpplague | landley that way they plug into the bottom board |
| 03:28.28 | landley | prpplague: what I meant was, when I plug in a flyswatter to my laptop's USB, it shows me two serial ports, one of which is a jtag. |
| 03:28.44 | prpplague | landley yea, sorry i'm thinking physical at the momment |
| 03:28.48 | prpplague | landley no problem |
| 03:28.49 | landley | Would would it take to get the USB dongle version of hammer to (at least as an option) have a jtag built in, ala flyswatter? |
| 03:28.56 | landley | I think virtual. :) |
| 03:28.57 | prpplague | landley the ft2232 chipset can go on the bottom board |
| 03:29.05 | landley | Cool. |
| 03:29.12 | prpplague | landley indeed thats what i'm thinking |
| 03:29.29 | landley | If the sucker is unbrickable, and the total re-flash procedure is well documented, software people can relax a bit. |
| 03:29.40 | landley | "So you reduced your hammer to a puddle of molten slag? Here's how to recover from that." |
| 03:29.49 | landley | Step by step... |
| 03:30.13 | landley | Heck, "run this script". |
| 03:30.31 | landley | hammertime.sh kernel.img rootfs.ext2 |
| 03:31.14 | landley | Actually "hammertime.sh kernel.img rootfs/" |
| 03:31.21 | prpplague | landley yea, what i'd do is do a small c app that will reflash the bootloader |
| 03:31.25 | landley | As a directory. The script should image it. |
| 03:31.57 | landley | The ability to rebuild the bootloader from source is nice, but it's also less interesting. |
| 03:32.03 | prpplague | landley do kernel and rootfs? |
| 03:32.04 | landley | It's a transient piece of software, doesn't stay running. |
| 03:32.18 | landley | Provided by user, although the stock ones would come on the CD. |
| 03:32.43 | landley | For example, right now I plan to build my own kernel and root filesystem for the board. |
| 03:32.58 | landley | But re-flashing the board is currently enough of a production it's still on my todo list. |
| 03:32.59 | prpplague | landley: generally speaking, you can reload the kernel and rootfs via the bootloader, it has it's own "shell like" interface |
| 03:33.15 | landley | And doing just a new kernel, while reflashing with existing bootloader and root filesystem is non-obvious how to do. |
| 03:33.28 | landley | prpplague: with a half hour of study, sure. |
| 03:33.47 | landley | "Run this script, which you can read if you want to know how it works because it's a well-commented shell script" would be easier. :) |
| 03:34.20 | landley | Knowing that I can do something and being able to sit down at a prompt and do so are two different things. |
| 03:34.42 | landley | "Great, I have a bootloader prompt. I need to type 5 longish command lines, three of which are in this file over here and the other two I have to come up with via long division". |
| 03:35.13 | landley | And of course if I get it wrong, which I will the first few times, I'm in for a fabulous night of debugging... |
| 03:35.26 | prpplague | landley indeed i'm still thinking of things on the dev side, ok, let me get the hardware working tomorrow, and then i can see where want to go with |
| 03:35.39 | landley | I should let you think about hardware. |
| 03:35.44 | prpplague | hehe |
| 03:35.46 | landley | I can do the software stuff myself, it's just time consuming. |
| 03:35.59 | landley | (The research part being the most time consuming.) |
| 03:36.08 | prpplague | landley naw, i usually work on the end that our customers (AML) are total idiots |
| 03:36.31 | landley | on the assumption, you mean? |
| 03:36.37 | landley | I dont' assume they're idiots, I assume they're ignorant. |
| 03:36.42 | landley | Don't confuse ignorance with stupidity. |
| 03:36.48 | landley | There are some darn clever ignorant people. |
| 03:36.49 | prpplague | landley hehe indeed, my mistake, hehe |
| 03:36.54 | landley | They tend to let out the magic smoke. |
| 03:37.01 | prpplague | landley the M5900 has two kernels and two rootfs's |
| 03:37.02 | landley | And then get it to do _tricks_. |
| 03:37.08 | CP|home | And there are many more people who are both ignorant, and happy to remain so |
| 03:37.19 | prpplague | landley the bootloader, backup kernel, and backup initrd, never get reflashed |
| 03:37.22 | landley | People will pay for the luxury of ignorance. |
| 03:37.29 | prpplague | landley basically boots up into a reflash mode |
| 03:37.35 | landley | I need to know 8 gazillion things today. How to change the oil in my car is not one of them. |
| 03:37.40 | prpplague | hehe |
| 03:37.41 | landley | I hand people money to do that for me. |
| 03:37.45 | prpplague | indeed |
| 03:37.59 | landley | What's an M5900? |
| 03:38.18 | landley | Googled for it. |
| 03:38.20 | landley | Bar code reader. |
| 03:38.22 | landley | Ok. |
| 03:38.26 | landley | That's nice for a production system. |
| 03:38.30 | landley | Not so nice for a development system. |
| 03:38.38 | landley | Because I _will_ screw up that backup kernel. |
| 03:38.42 | landley | I don't know how, I don't know when... |
| 03:38.50 | landley | But unless you've burned it into ROM... |
| 03:39.09 | landley | I have faith in my ability to break things. |
| 03:39.17 | landley | I broke "echo". |
| 03:39.35 | landley | I've currently managed to break "patch" on my laptop. No idea how... |
| 03:39.38 | landley | Wasn't root at the time. |
| 03:39.51 | landley | But it's insisting a patch that isn't reversed is reversed. |
| 03:39.55 | prpplague | landley hehe, the backup kernel is write protected |
| 03:40.00 | landley | Decided to implement patch in toybox rathe rthan fiddle with it. |
| 03:40.21 | landley | prpplague: is it theoretically possible to screw that up in software once the linux kernel is running? |
| 03:40.37 | prpplague | landley indeed |
| 03:40.38 | prpplague | http://www.elinux.org/M5900 |
| 03:40.59 | landley | There there's a reasonable chance that me, wandering around on the system as root, will break it. |
| 03:41.07 | prpplague | hehe |
| 03:41.12 | landley | I copied a floppy image to /dev/hda once. By accident. |
| 03:41.35 | landley | And then had to copy all my files off before I could reboot and reinstall. |
| 03:42.02 | prpplague | hehe, no listing in the rma log of anyone yet blowing up the flash |
| 03:42.04 | prpplague | hehe |
| 03:42.12 | prpplague | doesn't mean it wont happen |
| 03:42.19 | landley | prpplague: do people log into this device as root and use it for development? |
| 03:42.25 | jbevren | its kinda hard unless you write to the mtdblock devices |
| 03:42.40 | prpplague | landley neither, hehe, just use the apps |
| 03:42.45 | landley | Exactly. |
| 03:42.52 | prpplague | landley but they do update the firmware regularly |
| 03:43.00 | jbevren | I just need to find the spare time to build a working kernel and redo my flash layout |
| 03:43.05 | jbevren | (I dont like the one it came with lol) |
| 03:43.08 | landley | I'm talking about a development system hobbyists can use to learn embedded development and more experienced people can prototype hammer board stuff with. |
| 03:43.25 | prpplague | landley so basically we need something about 1/2 way between using command lines to reflash and using something totally automated |
| 03:43.26 | jbevren | lad: cool |
| 03:43.36 | jbevren | prpplague: nod |
| 03:43.44 | prpplague | landley yep, i'm withya |
| 03:43.45 | jbevren | something less automated than the jtag script? |
| 03:44.03 | landley | So you have people who don't know what they're doing and others who aren't paying attention (or are simply inexperienced with this particular system) performing major surgery in an exploratory fashion. |
| 03:44.17 | landley | prpplague: yup. |
| 03:44.31 | landley | The _laptop_ is going to stay up. |
| 03:44.38 | landley | If you fry the laptop, that's not hammer's problem. |
| 03:44.45 | landley | Presumably there are other laptops, or you already know how to reinstall that. |
| 03:44.53 | joe_bleau | You seem to be awfully concerned about brickage. |
| 03:45.02 | joe_bleau | Yet, if you have JTAG, there's really no risk. |
| 03:45.09 | landley | It's lobotomizing the hammer board that needs to be recoverable from a "don't fear the hardware" perspective. |
| 03:45.16 | prpplague | hehe, joe_bleau thats always a concern because most people don't understand JTAG |
| 03:45.22 | landley | joe_bleau: I still don't. |
| 03:45.34 | landley | I have a jtag here. Still havent' used it. |
| 03:45.38 | jbevren | I barely do, but you dont need to know much to unbrick the hammer |
| 03:45.49 | landley | And the jtag just gets the bootloader back on, and then I'm supposed to use the bootloader to reflash the kernel and filesystem. |
| 03:45.57 | joe_bleau | Bold print? Don't worry, if you blow away the entire flash, you can get it back? |
| 03:46.10 | prpplague | hehe |
| 03:46.15 | jbevren | joe_bleau: * physical damage excluded |
| 03:46.17 | jbevren | lol |
| 03:46.22 | landley | joe_bleau: More like "when you accidentally blow away the entire flash, here's how to fix it." |
| 03:46.35 | joe_bleau | Contrast with something like the Walnut board, where you could screw yourself if you didn't have JTAG. |
| 03:46.42 | prpplague | landley actually the script does the bootloader,kernel, initrd, and formats the jffs area |
| 03:46.51 | jbevren | joe_bleau: nod, one of my walnuts is currently in that state |
| 03:46.57 | landley | prpplague: which script is this? |
| 03:47.06 | prpplague | hammer.ocd |
| 03:47.24 | jbevren | landley: hammer.ocd in binaries/ |
| 03:47.26 | *** join/#edev dijenerate (n=dijenera@72.51.93.247) |
| 03:47.38 | prpplague | jbevren the cd layout has changed a little |
| 03:47.42 | jbevren | ah |
| 03:47.46 | jbevren | he has a release version? |
| 03:48.01 | joe_bleau | jbevren: maybe you should download the latest... |
| 03:48.03 | prpplague | yea, the early release, we are making alot of changes now |
| 03:48.05 | jbevren | btw did the hardware change at all? |
| 03:48.17 | prpplague | jbevren only the carrier board |
| 03:48.20 | jbevren | cool :) |
| 03:48.28 | landley | Ah yes, the script that assumed that ./apex.bin, ./zImage, and rootfs.arm.ext2.gz were available in the current directory. |
| 03:48.35 | prpplague | jbevren the 5v regulator went away and we use a 5v wallwart |
| 03:48.36 | jbevren | landley: right |
| 03:48.51 | jbevren | prpplague: nod, I have an old dlink regulatd 5v unit Ive been pondering using |
| 03:49.15 | jbevren | I think I have a fair grasp on the 1541's memory decode logic though |
| 03:49.17 | jbevren | oops |
| 03:49.33 | prpplague | landley i've changed that so that when you start openocd it takes the three files as args and then generates the script |
| 03:49.54 | landley | Is the new ISO available anywhere? |
| 03:49.58 | landley | Or its contents? |
| 03:50.10 | jbevren | prpplague: btw, nice design on the flyswatter.. I used it last week for jtag (for the first time) |
| 03:50.11 | landley | (Mercurial archive would be nice, so I could see what's changed. :) |
| 03:50.16 | prpplague | landley not as of yet, my target is wed. afternoon |
| 03:50.17 | landley | But that's probably wishful thinking... |
| 03:50.20 | joe_bleau | sorta OT, but can you run multiple jtags dongles on the same PC, at the same time? |
| 03:50.20 | landley | Cool. |
| 03:50.32 | landley | joe_bleau: it shows up as a serial port. |
| 03:50.47 | landley | At least the flyswatter does. |
| 03:50.56 | jbevren | landley: More specifically, its a special serial port with GPIOs that you can flutter :) |
| 03:50.57 | prpplague | joe_bleau indeed we gang program our devices |
| 03:51.05 | prpplague | joe_bleau we have 8 flyswatters |
| 03:51.22 | jbevren | prpplague: ..hooked to ZIF sockets? (drool) |
| 03:51.23 | jbevren | hehe |
| 03:51.29 | landley | How you'd got about telling mww and such _which_ serial port to use is something I dunno. |
| 03:51.37 | prpplague | jbevren ziff and sodimm sockets |
| 03:51.38 | joe_bleau | I'm thinking more in terms of debugging both ends of a comm protocol |
| 03:51.49 | landley | Ok, hang on. |
| 03:51.50 | landley | What do I run? |
| 03:52.07 | landley | I'm sitting at my laptop, I have the flyswatter and hammer plugged together and the flyswatter plugged into my laptop. |
| 03:52.17 | landley | What do I _do_? |
| 03:52.24 | landley | hammer.ocd is a script I feed to the bootloader, right? |
| 03:52.33 | landley | So for it to get the files I need to set up a tftp server...? |
| 03:52.38 | prpplague | right, it requires you use openocd |
| 03:52.58 | landley | openocd is this big executable in this directory/ |
| 03:52.59 | prpplague | landley one sec, let me get you the howto |
| 03:53.18 | landley | Which said "error while loading shared libraries: libftd2xx.so.0 blah blah blah" when I ran it? |
| 03:53.37 | landley | jtag_how.txt? Read it... |
| 03:54.04 | prpplague | landley didn't help? |
| 03:54.41 | landley | Ok, mount usbfs (as root), extract two tarballs, copy two .h files from one tarball to another... |
| 03:54.51 | landley | Install a random shared library into my laptop's /usr/lib directory by hand... |
| 03:54.58 | landley | Create symlinks... |
| 03:55.08 | landley | recompile openocd... |
| 03:55.39 | landley | Now copy that openocd executable into the directory with hammer.cfg, hammer.ocd, and apex.bin. |
| 03:55.53 | landley | (There already is one in there, but it didn't work. Possibly it just needed the shared library from step 6.) |
| 03:55.53 | prpplague | yep |
| 03:56.17 | landley | Then run openocd -f hammer.cfg as root. |
| 03:56.24 | prpplague | yep |
| 03:56.34 | landley | "14) This will load apex into ram & run it. then you can flash the original apex.bin back." |
| 03:56.59 | landley | You now have the ability to do this. How you invoke it is left as an exercise for the reader. |
| 03:57.07 | prpplague | oops |
| 03:57.11 | landley | Possibly there's a menu... |
| 03:57.12 | prpplague | that aint right, hehe |
| 03:57.18 | joe_bleau | Wow, snarky. |
| 03:57.30 | joe_bleau | Oh, you were joking, right? |
| 03:58.22 | landley | Nowhere in here mentions the filenames "apex.bin", "zImage", or "rootfs.arm.ext2.gz" which you have to read hammer.ocd to see, and which have to come from _somewhere_... |
| 03:58.23 | prpplague | <PROTECTED> |
| 03:58.32 | landley | Load them fromwhere? |
| 03:58.45 | landley | If I wanted to supply a different kernel but leave the bootloader and initrd alone for the moment, what do I do? |
| 03:59.02 | prpplague | step 12 |
| 03:59.28 | landley | Ah, copy them into the current directory? |
| 03:59.40 | landley | Using the names "zImage" and "rootfs.arm.ext2.gz" which this document doesn't mention? |
| 03:59.56 | landley | (Also, "apex.bin is the bootloader" would be nice to know.) |
| 04:00.23 | landley | Everything up through step 12 should be an "install" script of some kind. |
| 04:00.29 | prpplague | landley indeed, that document is incorrect, as the original was intended to only flash the apex |
| 04:00.55 | landley | I don't suppose there's a way to statically link libftd2xx into openocd? |
| 04:01.06 | prpplague | landley no, the lib is closed source |
| 04:01.12 | landley | Oh JOY. |
| 04:01.18 | landley | Bliss, rapture. |
| 04:01.29 | landley | Ahem. |
| 04:01.33 | prpplague | hehe, there is an open source version, but it doesn't have the performance |
| 04:01.48 | joe_bleau | Could be worse. JTAG info for PPC isn't open at all. |
| 04:01.53 | landley | How much performance difference are we talking about? |
| 04:02.01 | landley | 3 seconds vs 3 minutes? |
| 04:02.35 | landley | How about this: statically link against the open source version but try to dlopen() the closed source version, and use it instead if you can. |
| 04:02.51 | landley | Have a note "if this is too slow you can grab the closed source library BLAH and put it in your library path". |
| 04:02.57 | prpplague | basicically a 66% difference |
| 04:03.13 | landley | How long does reflashing the whole thing (entire memory, worst case scenario) take? |
| 04:03.14 | prpplague | on a 1 minute flash, you are looking at 3 minutes with the open source lib |
| 04:03.54 | landley | Ship a statically linked openocd, and have an openocd_fast that requires the library to be in your libpath. |
| 04:04.02 | prpplague | landley depending on size you could be up to around 4 minutes with libftdxxx, and around 12 with the open source lib |
| 04:04.15 | landley | prpplague: I can get a beverage. |
| 04:04.24 | landley | I had a 1541 disk drive! |
| 04:04.41 | landley | Why the speed difference? Better batching? |
| 04:04.51 | landley | 's probably it. |
| 04:04.57 | prpplague | landley hehe, indeed, however i guess thinking production since we need to program them as fast as possible |
| 04:05.05 | jbevren | landley: We're adding a PIO IDE interface to it |
| 04:05.19 | jbevren | so I've worked out IO decode logic heh |
| 04:05.27 | prpplague | landley not totally sure, drath over in #openjtag could elaborate |
| 04:05.38 | prpplague | landley drath is the primary author of openocd |
| 04:06.16 | landley | prpplague: for end users, a statically linked openocd is cool even if it is 1/3 the speed. |
| 04:06.24 | *** join/#edev agaffney (n=agaffney@gentoo/developer/pdpc.active.agaffney) |
| 04:06.39 | flyback- | neat |
| 04:06.49 | landley | Instructions on how to put together a faster one is nice, but "fire and forget" processes that you can do something else and come back can be slow without really bothering anybody too much. |
| 04:06.49 | flyback- | I seen ide tricks before |
| 04:07.06 | landley | jbevren: what's involved in doing this? |
| 04:07.16 | jbevren | landley: the proof of concept is 3 ICs |
| 04:07.20 | jbevren | discrete logics |
| 04:07.28 | landley | So it needs extra hardware, then? |
| 04:07.36 | jbevren | taht only decodes 8 bits of ide data, so you only get to use half a sector |
| 04:07.39 | landley | Not just clever PIO programming and an extra plug? |
| 04:07.42 | jbevren | with 3 more ICs you get all 16 |
| 04:07.59 | landley | Still, cool. |
| 04:08.00 | prpplague | landley lots of hardware geeks here |
| 04:08.00 | flyback- | heh I seen those also |
| 04:08.04 | flyback- | the 8 bit ide hack |
| 04:08.07 | jbevren | flyback-: nod |
| 04:08.08 | landley | prpplague: I'd noticed this thing. |
| 04:08.10 | flyback- | jbevren: you do know |
| 04:08.13 | flyback- | if you use a cf to ide |
| 04:08.19 | flyback- | cf are required to support 8 bit mode :) |
| 04:08.31 | landley | I am a software geek. I need hardware geeks, but it's really hard to restore broken hardware from a backup... |
| 04:08.42 | prpplague | landley hehe, indeed |
| 04:08.43 | jbevren | landley: define 'broken' :) |
| 04:08.46 | flyback- | landley: anytime you need pointers just holler :) |
| 04:08.48 | jbevren | cause imho bricked isnt broken :) |
| 04:09.02 | landley | I've let the magic smoke out of ethernet hubs? |
| 04:09.11 | flyback- | hubs? |
| 04:09.14 | flyback- | no loss :) |
| 04:09.15 | prpplague | landley well, you give us wise lists and we'll create hardware |
| 04:09.15 | landley | (Strangely, some of them still work afterwards!) |
| 04:09.15 | jbevren | landley: at any rate, we're now going to modify the hardcoded disk geometry in the DOS rom and replace the sector read/write routines with ide routines |
| 04:09.17 | jbevren | lol |
| 04:09.22 | jbevren | easy 16meg hack |
| 04:09.29 | jbevren | (CBMFS is limited to 16megs) |
| 04:09.38 | flyback- | jbevren: did you see what I said about cf to ide |
| 04:09.40 | jbevren | landley: easy. |
| 04:09.44 | prpplague | landley hehe |
| 04:09.58 | jbevren | landley: hub: hollaring across the street at your neighbor (all neigbors hear it) |
| 04:10.03 | landley | My aforementioned 1541 C64 disk drive, I forgot to use an insulator one time I was realigning it. |
| 04:10.03 | flyback- | jbe if you only have 8 bit decoding and use a cf module you get the whole module |
| 04:10.05 | jbevren | landley: Switch: Calling your neighbor on the phone |
| 04:10.09 | jbevren | <PROTECTED> |
| 04:10.18 | landley | Turned the circuit board 45 degrees and put it down on the metal case. |
| 04:10.26 | flyback- | wow |
| 04:10.34 | landley | Ran the calibration program, fiddling with the screwdriver, and wondered why it was making so much noise... |
| 04:10.38 | prpplague | or the difference between a mass market mailing, and a single mail going to a single address |
| 04:10.41 | landley | Turned around, this trail o smoke coming up from one of the components. |
| 04:10.44 | jbevren | prpplague: true |
| 04:10.48 | prpplague | landley ha |
| 04:10.52 | landley | Turned it off FAST |
| 04:10.53 | jbevren | flyback-: Ive done taht before too |
| 04:10.56 | landley | Moved the board back in position. |
| 04:11.01 | landley | Strangely, the drive still worked afterwards. |
| 04:11.04 | jbevren | heh |
| 04:11.14 | jbevren | if you dont hit the +12v line, the 1541 is amazingly forgiving |
| 04:11.16 | landley | Except that if I turned it on with a disk in the drive, it would wipe whatever sector was under the head while powering up. |
| 04:11.19 | jbevren | at any rate, it comes with 2k ram |
| 04:11.26 | jbevren | with four ICs I can make that 42k |
| 04:11.26 | landley | Learned to take the disk out of the drive before turning it off... |
| 04:11.28 | jbevren | lol |
| 04:11.40 | jbevren | and I'll probably add a serial chip for tty debugging |
| 04:11.48 | jbevren | the DOS runs two threads simultaneously |
| 04:11.48 | prpplague | landley ok so, here is my list of stuff, small case with usb slave on one side(connector or short cable), 10baseT rj-45 on the other side |
| 04:11.51 | flyback- | prpplague: can I call landley a stupid canuck just this once? please? |
| 04:11.53 | landley | jbevren: I know the difference between hub and switch, I just don't _care_. :) |
| 04:11.57 | jbevren | hehe |
| 04:12.08 | flyback- | j/k |
| 04:12.09 | prpplague | flyback- only warning |
| 04:12.17 | landley | Also, you ccna degrade a switch into hub mode by repeatedly lying about your mac address until its ARP cache fills up and it starts to punt. |
| 04:12.26 | prpplague | flyback- you know we find that word offensive here |
| 04:12.26 | landley | This is good for packet sniffing. |
| 04:12.40 | flyback- | landley: putting a hub in series is easier :P |
| 04:12.48 | landley | flyback: they have texas canucks? |
| 04:12.54 | landley | I thought it was a type of waterfowl. |
| 04:12.57 | flyback- | some newer switches don't let you muck the mac buffers anymore |
| 04:12.57 | landley | Or perhaps a hockey team. |
| 04:13.00 | prpplague | hehe |
| 04:13.36 | prpplague | landley what about the gpio lines? you want some leds or something on the outside of the case? |
| 04:13.57 | landley | prpplague: the USB connector will instantiate a serial console and a serial jtag thingy for the laptop it's plugged into? |
| 04:14.04 | landley | LEDs are nice. |
| 04:14.07 | landley | Blinky lights are always fun. |
| 04:14.11 | prpplague | landley indeed |
| 04:14.20 | landley | "My first embedded program: make it blink its lights!" |
| 04:14.43 | landley | I remember dealing with my headless box, and finding a program that would beep the speaker. |
| 04:14.51 | prpplague | hehe |
| 04:14.53 | landley | So I had some OUTPUT if it died before the web server came up. |
| 04:15.03 | landley | I taught that little program to play a half-dozen different tunes. |
| 04:15.18 | flyback- | WTF |
| 04:15.29 | flyback- | PRPP the microchip ether chip comes in dip? |
| 04:15.34 | prpplague | yea |
| 04:15.39 | flyback- | wow :) |
| 04:16.26 | landley | How much is the little sound generating diode thingy, anyway? |
| 04:16.53 | landley | It's the hardware equivalent of the "blink" tag, and including it would probably be tacky, but I'm curious... |
| 04:16.55 | prpplague | landley i'm sorry, i didn't understand your question? |
| 04:17.05 | flyback- | landly make a resistor dac |
| 04:17.09 | flyback- | or use a usb sound chip |
| 04:17.21 | landley | You know how a modern "pc speaker" in desktop systems is a little noisy two-wire component that stands up from the board? |
| 04:17.21 | flyback- | prpplague: I think he means piezo generation |
| 04:17.25 | prpplague | landley the buzzer on the carrier board is easy |
| 04:17.29 | flyback- | yep I was right |
| 04:17.34 | landley | Ooh, does it already have one? |
| 04:17.35 | prpplague | landley couple cents |
| 04:17.39 | prpplague | landley yea |
| 04:17.44 | landley | Oh, and a button. |
| 04:17.47 | landley | It should probably have a button. |
| 04:17.52 | prpplague | yea |
| 04:18.06 | landley | I dunno what the button should DO, but it should have one on general principles. |
| 04:18.12 | landley | If it has blinky lights, it should have a button. :) |
| 04:18.35 | landley | (I mean if you want to reset the sucker, unplug it and plug it back in.) |
| 04:18.39 | landley | USB == power, piece of cake. |
| 04:18.41 | prpplague | landley indeed |
| 04:18.54 | landley | The button would be for software, like the lights. |
| 04:19.16 | landley | Heck, if you wanted to go the full LULU route, we could do a book about "learning embedded development". |
| 04:19.26 | joe_bleau | Jeez, next he'll want a knob, too. ;-) |
| 04:19.39 | prpplague | landley indeed |
| 04:19.50 | landley | Knobs I can take or leave. |
| 04:19.59 | prpplague | hehe |
| 04:20.20 | landley | The little black reset button on the current carrier board is what I'm thinking of. Probably a 50 cent part... |
| 04:20.31 | joe_bleau | They exemplify a third class of I/O: analog input. |
| 04:20.38 | prpplague | landley ha, try $0.003 |
| 04:21.18 | landley | Hmmm... |
| 04:21.30 | landley | It should probably have one LED that says "I am plugged in and receiving power". |
| 04:21.52 | landley | The rest software controlled, but one the kernel should switche on to say "hey, I'm up and have a CONSOLE". |
| 04:21.58 | landley | Or the init scripts should. |
| 04:22.43 | landley | Basically, light 1 says hardware is working okay-ish. |
| 04:23.07 | landley | Light 2 says "if you can't get a console on the serial port right now, the problem is at your end. Linux is up and running." |
| 04:23.17 | jbevren | landley: like the linsksys router LEDs? |
| 04:23.20 | landley | Other lights? Up to the user. |
| 04:23.22 | jbevren | blinking=loading kernel |
| 04:23.25 | landley | A bit, yes. |
| 04:23.28 | landley | Sure, why not. :) |
| 04:23.29 | jbevren | solid=kernel going |
| 04:23.31 | jbevren | etc |
| 04:23.33 | landley | The bootloader can blink LEDs? |
| 04:23.34 | joe_bleau | bicolor? |
| 04:23.34 | jbevren | I'm not sure thats their behavior |
| 04:23.36 | cajaka | 2 lights or a R/G? |
| 04:23.43 | landley | Up to you. |
| 04:24.05 | jbevren | my HP 725/50 has a hearbeat LED |
| 04:24.06 | jbevren | hehe |
| 04:24.10 | landley | I dunno what's involved in dimming LEDs, from an expense standpoint. |
| 04:24.21 | jbevren | as long as the kernel's going it blips twice and watis a sec |
| 04:24.33 | jbevren | landley: PWM output. $0. :) |
| 04:24.36 | jbevren | hehe |
| 04:24.46 | landley | jbevren: that should be up to the software. |
| 04:24.49 | prpplague | yea we can do several things |
| 04:24.50 | jbevren | nod |
| 04:24.52 | landley | Some people find blinking LEDs distracting. |
| 04:25.01 | jbevren | landley: not a 1hz led |
| 04:25.04 | jbevren | (generally) |
| 04:25.09 | landley | If you put together dimmable RBG leds, you can get any color out of it. |
| 04:25.20 | landley | This isn't _useful_, but it's _cool_. |
| 04:25.36 | prpplague | landley try out the buzzer |
| 04:26.02 | landley | And it's another fun diagnostic tool. "Right, the LED is puce. That means my program got the wrong checksum value in part 3." |
| 04:26.20 | landley | prpplague: how? |
| 04:26.25 | landley | Normal linux pcspkr? |
| 04:26.39 | prpplague | landley naw, its under the sys under input |
| 04:26.51 | landley | Is this documented somewhere? |
| 04:26.53 | prpplague | landley there is a freq entry and a duration entry |
| 04:27.08 | joe_bleau | man input? |
| 04:27.16 | prpplague | landley nope, not as of yet, not enough people asked for it yet |
| 04:27.45 | landley | What's jp1? |
| 04:28.02 | landley | What's the button that _isn't_ reset? |
| 04:28.12 | jbevren | thats a user input button |
| 04:28.14 | landley | And there's more jumpers over near the serial port... |
| 04:28.20 | jbevren | press it heh |
| 04:28.30 | jbevren | the console (on mine anyway) lets me know I pressed it |
| 04:29.29 | landley | Nope, didn't say anything about me having pressed it. |
| 04:29.55 | prpplague | landley the default just does a keyboard event |
| 04:30.09 | prpplague | jbevren yours had an example script installed |
| 04:30.14 | landley | looking at /sys/class/input/input0... |
| 04:30.40 | prpplague | landley one of the inputs is the button, the other is the speaker |
| 04:30.53 | landley | The speaker is an input? |
| 04:30.59 | prpplague | landley if you look under the input options you'll see freq and duration |
| 04:31.06 | prpplague | landley yea go figure |
| 04:31.10 | joe_bleau | weird. |
| 04:31.13 | landley | Ah, /dev/input |
| 04:31.19 | landley | mice? |
| 04:31.20 | prpplague | landley i asked on the mailing list like 20 times why? |
| 04:31.51 | prpplague | landley: no one ever would answer why that driver is in the input class |
| 04:31.54 | landley | Ooh, if I cat /dev/input/event0 I get binary gunk when I press the button! |
| 04:31.57 | landley | I wonder what it means? |
| 04:32.12 | landley | Because Greg Kroah-Hartman. |
| 04:32.20 | landley | Yes, that is the end of the sentence. |
| 04:32.25 | prpplague | landley hehe |
| 04:32.35 | prpplague | there is a script file in /sbin |
| 04:32.39 | prpplague | pb0.sh |
| 04:32.45 | prpplague | have a look at that |
| 04:33.04 | prpplague | thats coded into the button driver |
| 04:33.13 | jbevren | heh |
| 04:33.16 | prpplague | it gets executed when your press the button |
| 04:33.43 | landley | Ooh, there's a little blinky LED next to it. |
| 04:33.49 | landley | Hadn't noticed, my finger was in the way. |
| 04:33.53 | prpplague | hehe |
| 04:34.04 | landley | The driver is doing an exec usermodehelper? |
| 04:34.05 | landley | Ew. |
| 04:34.10 | prpplague | landley indeed ew |
| 04:34.13 | landley | Run a daemon. |
| 04:34.27 | CP|home | what do you mean indeed ew? |
| 04:34.31 | CP|home | you argued violently for that |
| 04:34.47 | landley | I'm all for the existence of exec usermodehelper. |
| 04:34.54 | landley | I just think this is a bad use for it. |
| 04:35.07 | prpplague | CP|home i mean that its not clean, but its basically the best way to do what i need |
| 04:35.27 | prpplague | just a simple way of kicking off userland app |
| 04:35.29 | jbevren | landley: I think its an awesome proof of concept |
| 04:35.35 | jbevren | tis dirty yeah |
| 04:35.38 | jbevren | but it works ;) |
| 04:36.02 | prpplague | hehe |
| 04:37.45 | jbevren | careful, CP.. he'd do it |
| 04:37.46 | jbevren | :) |
| 04:37.50 | prpplague | hehe |
| 04:37.56 | jbevren | Hm |
| 04:38.09 | landley | Who needs anything fancy? |
| 04:38.12 | landley | Shell script is plenty. |
| 04:38.16 | jbevren | so I add 40k sram, another IO chip, and a UART to my commodore drive. next thing I guess is wiring the hammer to it lol |
| 04:38.27 | jbevren | prpplague: go ahead |
| 04:38.28 | jbevren | cause I did |
| 04:38.32 | jbevren | :D |
| 04:38.39 | landley | If I knew what the blob from /dev/input/event0 was, I could have a script read it right now and do the echo. |
| 04:38.40 | prpplague | jbevren: where? |
| 04:38.43 | jbevren | elinux.org/wiki |
| 04:38.53 | prpplague | jbevren url? |
| 04:38.57 | landley | Except that if my script toggles the light, the driver calling pb0.sh will _also_ toggle the light... |
| 04:39.06 | prpplague | landley there are several small event sample code you can look at |
| 04:39.39 | landley | Ah, mv pb0.sh pb0.bak... |
| 04:39.46 | jbevren | http://elinux.org/Hammer_SID_Interface |
| 04:40.34 | prpplague | jbevren hmm, not exactly what i was looking for, but i suspose it will satisfy the boss |
| 04:41.01 | prpplague | landley if you look at the script it has a check to see if there are custom scripts in the jffs partition |
| 04:41.17 | prpplague | landley you can just put scripts in there that do nothing |
| 04:42.17 | jbevren | prpplague: I'll update with much more detail as I proceed |
| 04:43.05 | landley | Commodore's toast, where do you get a sid chip? |
| 04:43.16 | jbevren | landley: Ive rescued them from dead c64's |
| 04:43.25 | landley | Commodore went out of business when QVC (the home shopping network) started up. |
| 04:43.29 | landley | That was the next tenant in that building. |
| 04:43.40 | jbevren | Commodore went out of business when it got sold to an Indian busines stycoon |
| 04:43.50 | landley | Not quite... |
| 04:44.10 | landley | Read http://www.variantpress.com/books/on-the-edge |
| 04:44.18 | landley | Marvelous book, I'm only halfway through it myself. |
| 04:44.27 | landley | Did I mention that one of my hobbies is computer historian? |
| 04:44.34 | landley | http://landley.net/history/mirror |
| 04:44.49 | jbevren | landley: I watched the DVD produced by ex-commodore employees |
| 04:44.53 | jbevren | I know the truth :) |
| 04:44.55 | landley | And in the case of commodore, http://landley.net/history/scans/ |
| 04:45.00 | landley | I have that DVD somewhere. |
| 04:45.09 | landley | Packed, I think. I still have to watch the atari one first. |
| 04:45.59 | prpplague | hehe |
| 04:46.11 | jbevren | but my understanding at any rate is taht it got sold and dismantled |
| 04:46.13 | jbevren | which is a shame |
| 04:46.19 | jbevren | commodore was at one point big enough to buy MS out |
| 04:46.31 | jbevren | they had their on chipfab even |
| 04:46.44 | prpplague | landley ok, well i've got my work cut out for me tomorrow, you be online tomorrow evening? |
| 04:48.32 | landley | I should be. |
| 04:48.50 | landley | jbevren: you're kind of misunderstanding things. |
| 04:48.54 | landley | Read the book. |
| 04:49.13 | landley | I've talked to ex-commodore employees (used to live near there). Most of the old hands were already gone by the time it shut down. |
| 04:50.26 | landley | At one point, MITS was in a position to buy MS out. They were a tiny company until about 1980. Yes, commodore bought an early chip fab (MOStek), but they were never really going toe to toe with Intel. |
| 04:50.35 | landley | The more amusing part was when they were in a position to buy APPLE out. |
| 04:51.16 | prpplague | landley ok, i'll give you a shout tomorrow, think of any more questions, let me know |
| 04:51.28 | prpplague | later folks |
| 04:52.07 | wmat | mark: do we actually say anywhere on the website that we support Ulink2? |
| 04:52.24 | wmat | ugh, wrong window |
| 04:55.16 | jbevren | time to hit the sack for me too |
| 05:20.08 | *** join/#edev T0mW (n=Tom@24.238.86.158.res-cmts.sth.ptd.net) |
| 05:20.33 | *** part/#edev T0mW (n=Tom@24.238.86.158.res-cmts.sth.ptd.net) |
| 05:20.33 | landley | These old computer magazines always have weird interesting stuff in them. |
| 05:20.59 | landley | For example, here'a mathematical explanation of why the Amiga lasted around 10 years as a viable platform: |
| 05:21.02 | landley | http://landley.net/history/scans/icpug-1986-jul/page290-291.jpg |
| 05:21.27 | landley | Excuse me, around 15 years. |
| 05:21.39 | landley | Wasn't really seriously displaced until almost 2000. |
| 05:36.16 | Redhatter | That'd explain that bloody annoying whine that some monitors kick up after a while operating. |
| 05:36.43 | Redhatter | Particularly television sets, but also some low-resolution VGA monitors ... seem to kick up a bit of noise. |
| 05:39.25 | jbevren | heh flyback whine |
| 05:39.35 | jbevren | re-glue the flyback or tighten the bolts on it |
| 05:39.37 | jbevren | ;) |
| 05:43.53 | ds2 | chuck the CRT |
| 05:44.19 | Redhatter | I like to leave the HV parts of CRTs alone ... I don't fancy becoming a human capacitor. |
| 05:45.11 | ds2 | what's a few coulombs between friends ;) |
| 05:45.28 | jbevren | lol |
| 05:45.30 | Redhatter | Mmmm, but it's those killer volts that I worry about. |
| 05:45.36 | jbevren | we're not nearly dielectric enough tob e a cap |
| 05:46.34 | jbevren | ok bedtime for real |
| 05:46.38 | Redhatter | They're not beyond leaping out to attack. :-) |
| 05:48.56 | MonMotha | if your VGA or higher monitor is audibly whining, something is wrong with it |
| 05:49.18 | MonMotha | if you have a TV or "standard res" (480i/CGA) monitor that's whining, well, then that's normal and you have halfway decent hearing |
| 05:50.46 | Redhatter | Yeah, it's rare that I hear it with a VGA monitor, and it's almost always, when working in either standard VGA, or some older monitors when running SVGA (1024×768 @ 43Hz interlaced was one that comes to mind) |
| 05:51.41 | Redhatter | It's mainly TVs that irritate me, but I can convince my brain to filter it out if there's enough other sounds to tune into. |
| 05:51.59 | MonMotha | that's not SVGA... |
| 05:52.28 | MonMotha | that's actually CGA timings, just higher res (and much lower vertical refresh) |
| 05:52.34 | MonMotha | ~16kHz |
| 05:52.37 | MonMotha | which is why you could hear it |
| 05:52.58 | MonMotha | what happens is that the flyback transformer vibrates at the frequency of the horizontal sync |
| 05:53.12 | Redhatter | Makes sense. |
| 05:53.33 | Redhatter | I knew it was to do with the HV components of the CRT monitor, which is why I'm not keen to go touch them. |
| 05:55.16 | MonMotha | depending on tube size, it's generally in the neighborhood of 20-30kV |
| 05:55.21 | MonMotha | not exactly something you want to play with |
| 05:55.32 | MonMotha | really small tubes can be lower |
| 05:55.40 | MonMotha | usually still a dozen kV or so |
| 05:55.49 | MonMotha | really tiny tubes (like 1" or smaller) can be a few hundred volts |
| 06:53.50 | *** join/#edev redblue (i=star@ppp091.108-253-207.mtl.mt.videotron.ca) |
| 07:06.46 | *** join/#edev chouimat|sick (n=dieu@r2351064.cidc.net) |
| 07:08.46 | *** join/#edev DrZimmerman (n=theo@2001:5c0:8fff:fffe:0:0:0:6af1) |
| 09:27.50 | *** join/#edev hw_ (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 09:35.09 | *** join/#edev lyakh (n=lyakh@p57BD2521.dip0.t-ipconnect.de) |
| 09:37.44 | *** join/#edev bbradley (n=bbradley@5ac8083e.bb.sky.com) |
| 10:18.04 | *** join/#edev linac (n=lin@222.76.90.80) |
| 10:40.14 | *** join/#edev redblue (i=star@ppp091.108-253-207.mtl.mt.videotron.ca) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev dijenerate (n=dijenera@72.51.93.247) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev dfarnsworth (n=dfarnswo@65.200.49.156) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev marty_p (i=1000@cp1292005-a.tilbu1.nb.home.nl) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev SpeedEvil (n=mauve@tor/regular/SpeedEvil) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev dbtid (i=qjgns0eb@cpe-71-72-252-171.columbus.res.rr.com) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev cbrake_away (n=cbrake@oh-69-34-21-229.sta.embarqhsd.net) [NETSPLIT VICTIM] |
| 10:40.14 | *** join/#edev adyer (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) [NETSPLIT VICTIM] |
| 11:40.54 | *** join/#edev sjdsphreaker (n=sjdsphre@72.51.93.247) |
| 11:42.18 | sjdsphreaker | ooopp!!! |
| 11:42.33 | sjdsphreaker | wrong channel, server... day |
| 11:42.34 | sjdsphreaker | lol |
| 12:46.05 | *** join/#edev steliosk (n=Stelios@ipa226.211.tellas.gr) |
| 13:53.14 | *** join/#edev sjhill (n=sjhill@real.realitydiluted.com) |
| 13:57.50 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 14:01.33 | prpplague | ho ho hum |
| 14:04.29 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 14:04.29 | *** mode/#edev [+o GPSFan] by ChanServ |
| 14:40.21 | cajaka | <PROTECTED> |
| 14:41.10 | cajaka | Aparently I'm not awake. |
| 14:56.26 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 14:58.03 | prpplague | cajaka: greetings |
| 15:13.00 | *** join/#edev chouimat|work (n=mchouina@64.26.181.250) |
| 15:13.47 | chouimat|work | MORNING |
| 15:17.03 | prpplague | GPSFan: ping |
| 15:21.54 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 15:31.28 | GPSFan | prpplague: g'day.. |
| 15:34.09 | prpplague | GPSFan: is aeronix including source in their package or an offer for source? |
| 15:41.31 | GPSFan | prpplague: in the included doc's ... neither. the unit does display a screen that talks about GPL software being included. I can't look right now to confirm the wording. THey have a website, linux.zipitwireless.com that is supposed to distribute all the GPL sources etc, but it hasn't gone live yet. Aeronix is no longer involved except as a contract design house. zipitwireless is now a seperate company, and filed their fcc registration as such. |
| 15:41.59 | GPSFan | bbl... |
| 15:47.29 | wmat | prpplague: looks like I'll be ordering a Hammer devkit :) |
| 15:47.55 | prpplague | wmat: cool, i spoke to boss |
| 15:48.07 | prpplague | wmat: if you add support, we'll add a link for you guys |
| 15:50.08 | wmat | prpplague: cool |
| 15:50.32 | wmat | prpplague: now tell him to send me one for free too ;) |
| 15:51.23 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 15:58.05 | *** join/#edev ico2 (n=ico2@modem-826.cheetah.dialup.pol.co.uk) |
| 16:06.09 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 16:06.09 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 16:06.46 | GPSFan | prpplague: on the device, there is a copyrights screen which says (among other things) that they make the sources available to the public at www.zipitwireless.com/linux.html. There is nothing for the Z2 at that address (the new site is supposed to provide the Z2 specific stuff), just the old "in the spirit of" but not really GPL compliant Z1 stuff. |
| 16:07.15 | prpplague | GPSFan: yea i noticed |
| 16:08.00 | GPSFan | btw I have asked the question about the GPL strings in the wireless drivers. Stay tuned... |
| 16:08.09 | prpplague | GPSFan: ok dokie |
| 16:16.50 | prpplague | GPSFan: http://www.elinux.org/Image:Nail_board.jpg |
| 16:16.55 | prpplague | oops wrong url |
| 16:17.02 | prpplague | GPSFan: http://www.elinux.org/Nail_Board |
| 16:18.46 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 16:18.59 | GPSFan | prpplague: what's the usb thingy? a built-in flyswatter? |
| 16:19.10 | prpplague | GPSFan: yea |
| 16:19.34 | GPSFan | sort a development system in a cigar-tube. |
| 16:19.45 | prpplague | hehe indeed |
| 16:19.57 | prpplague | GPSFan: on the back side of the board have some led's and some push buttons |
| 16:20.26 | GPSFan | put an ultrasonic transducet on one end, add a few buttons an you have a "sonic screwdriver" |
| 16:23.29 | prpplague | hehe |
| 16:23.44 | prpplague | i think the doctor who folks might have a TM problem with that |
| 16:24.06 | GPSFan | prpplague: on the nail, I'd swap out the ehternet interface for a usb host connector. then you could plug in a usb-ehternet, or usb wireless, or usb-bluetooth, or usb-audio..... |
| 16:24.33 | GPSFan | thinkgeeks has a sonic screwdriver for sale. |
| 16:24.56 | prpplague | GPSFan: yea thats kinda what i was thinking too, but in our discussions last night, it seemed more people were interested in the 10baseT |
| 16:25.00 | prpplague | maybe i was wrong |
| 16:26.02 | SpeedEvil | I'd wonder about making the hammer with turned pin IC sockets |
| 16:26.17 | SpeedEvil | so you can plug modules into the top |
| 16:26.24 | GPSFan | I watched them but didn't participate. was busy doing other stuff. but I'm sure there would be interest in either option, but IMHO the usb host connector version would be more flexible. I haven't researched the power consumption issues though. |
| 16:26.56 | cbrake | prpplague: back to developing with some wifi modules |
| 16:27.17 | prpplague | cbrake: fun fun |
| 16:27.24 | cbrake | prpplague: the EWWM3236AB-KIT we received from Embedded Works seems to be completely dead |
| 16:27.29 | cbrake | prpplague: does not respond to CMD5 |
| 16:27.51 | cbrake | prpplague: zcomax modules seems to be responding fine in the same system |
| 16:28.01 | prpplague | cbrake: fun fun |
| 16:28.05 | cbrake | prpplague: :-) |
| 16:28.13 | prpplague | cbrake: the wi2wi stuff seems to be pretty darn solid |
| 16:28.40 | prpplague | cbrake: just incase you mised it, the new zipit v2 has a 8686 wifi chip running via spi |
| 16:29.20 | prpplague | wmat: let me know if you have any problems ordering |
| 16:29.30 | prpplague | wmat: you should be able to choose USPS for deliver |
| 16:29.30 | wmat | prpplague: will do |
| 16:29.32 | prpplague | delivery |
| 16:30.07 | prpplague | wmat: you look at the nail board? |
| 16:31.05 | cbrake | prpplague: I'll have to check wi2wi out again |
| 16:33.57 | cbrake | prpplague: does the liberatas in mainstream build and run for you? |
| 16:45.05 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 16:50.28 | *** join/#edev prpplague^2 (n=dave@mail.americanmicrosystems.com) |
| 16:54.05 | cbrake | prpplague^2: this is what I get when I try libertas in 2.6.24rc5: http://pastebin.ca/822250 |
| 16:54.06 | *** mode/#edev [+o prpplague] by ChanServ |
| 16:54.27 | cbrake | prpplague: trying git HEAD next ... |
| 16:54.44 | prpplague | cbrake: thats odd |
| 16:54.57 | prpplague | cbrake: i know dwmw2 has pushed a ton of patches in the last few days |
| 16:54.58 | cbrake | prpplague: so it works for you? |
| 16:55.13 | prpplague | cbrake: i'm using an older version |
| 16:55.20 | chouimat|work | hi prpplague |
| 16:55.22 | prpplague | cbrake: i've not upgraded yet |
| 16:55.27 | prpplague | chouimat|work: greetings |
| 16:56.02 | cbrake | prpplague: are you running mainline, or did you get SDIO/libertas patches from another tree? |
| 16:56.24 | prpplague | cbrake: i'm using ossmans tree from way back |
| 16:56.36 | cbrake | prpplague: ok, thanks |
| 16:58.21 | prpplague | wmat: you look at the nail board? |
| 16:58.46 | CosmicPenguin | dwmw2 has been banging the crap of of libertas for like two weeks now |
| 16:59.00 | CosmicPenguin | s/of/out/ |
| 16:59.12 | CosmicPenguin | and cursing a blue streak to go along with it |
| 16:59.26 | prpplague | bluestreak? |
| 17:00.01 | prpplague | CosmicPenguin: the sharp/nxp arm cores? |
| 17:01.14 | *** join/#edev FireEgl (i=FireEgl@FireEgl.CJB.Net) |
| 17:01.45 | kay2 | hey |
| 17:01.50 | kay2 | when booting my kernel I get that: |
| 17:01.51 | kay2 | Synthesizing the initial hotplug events...net eth1: device_rename: sysfs_create_ |
| 17:01.51 | kay2 | symlink failed (-17) |
| 17:02.03 | kay2 | any idea ? |
| 17:05.02 | cbrake | prpplague: ahh, I did not have CONFIG_WIRELESS_EXT enabled |
| 17:05.14 | prpplague | cbrake: fun fun |
| 17:05.20 | prpplague | bbiab |
| 17:10.19 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 17:27.55 | *** join/#edev ico2_ (n=ico2@modem-2197.cougar.dialup.pol.co.uk) |
| 17:33.50 | hapchap_ | prpplague: default baud rate on the bluetooth module? |
| 17:36.25 | prpplague | hapchap_: 9600,n,8,1 |
| 17:56.49 | *** join/#edev landley (n=landley@adsl-76-197-33-46.dsl.austtx.sbcglobal.net) |
| 17:57.30 | landley | So how do I loopback mount a jffs2 partition again? |
| 18:02.44 | prpplague | landley: jffs2 mounting isn't as easy as something like an iso |
| 18:02.59 | prpplague | landley: in fact the procedure is a real PITA |
| 18:03.08 | prpplague | landley: greetings btw |
| 18:03.16 | prpplague | landley: http://www.elinux.org/Nail_Board |
| 18:04.38 | prpplague | landley: http://osdir.com/ml/linux.debian.devel.embedded/2005-10/msg00037.html |
| 18:05.53 | landley | prpplague: I noticed the PITA-ness. :) |
| 18:06.06 | prpplague | landley: http://www.yargybot.com/Familiar/BackupHowto.html |
| 18:06.16 | prpplague | landley: doing some testing for busybox compliancy? |
| 18:06.18 | landley | This is your prototype? |
| 18:06.24 | landley | prpplague: trying to. |
| 18:06.28 | *** join/#edev T0mW (n=Tom@24.238.86.158.res-cmts.sth.ptd.net) |
| 18:06.40 | prpplague | landley: yea, got it working just a few minutes ago |
| 18:06.47 | landley | Although the high gain suit is now "settled", their license to use busybox remains suspended until I sign off that the source they gave us is complete and corresponding. |
| 18:07.01 | landley | I can confirm it builds busybox, but unless I can grab their binary I dunno if it's corresponding. |
| 18:07.04 | landley | Corresponding to what? |
| 18:07.08 | landley | I dunno what they distributed... |
| 18:07.32 | landley | prpplague: I take it the slightly-less-prototype version will have the hammer and other board mounted vertically instead of horizontally? :) |
| 18:08.07 | landley | It does look cool... |
| 18:08.12 | prpplague | landley: i'm looking at both options |
| 18:08.33 | prpplague | landley: with the horizontal mounting the back side is left open for things like leds and buttons |
| 18:08.51 | prpplague | landley: i.e. the back side of the board is blank right now |
| 18:08.55 | landley | (Reading the above howto...) This looks like a job for User Mode Linux! |
| 18:09.21 | landley | http://www.thefump.com/fump.php?id=86 |
| 18:10.24 | prpplague | hehe |
| 18:10.25 | landley | prpplague: While that's cool, you have a long thin narrow board that has more leverage to break itself accidentally, and is less likely to fit in a jacket pocket. |
| 18:10.45 | prpplague | hmm, good points, |
| 18:10.51 | landley | You shouldn't have a heat dissipation problem, so packing the components more tightly in two layers has less downside than normal. |
| 18:10.59 | prpplague | landley: you think being as long as a standard pen is too long? |
| 18:11.19 | landley | It's thicker than a pen. |
| 18:11.34 | prpplague | landley: indeed |
| 18:11.42 | landley | It's also longer than the pen in the picture. |
| 18:12.03 | landley | Plus you're going to have a cable dangling off the end, which won't bend at a 90 degree angle immediately off the end. |
| 18:12.10 | landley | Plus case. |
| 18:12.11 | prpplague | landley: yea thats because of it being a prototype, after doing a custom board, it would be around 1 to 1.25 inches shorter |
| 18:12.27 | prpplague | landley: indeed |
| 18:12.39 | landley | You could probably get away with it, but shorter (at the expense of thicker) would be preferable. |
| 18:12.50 | prpplague | landley: the other question was on the 10baseT, several folks thought it should be a usb host connector |
| 18:13.06 | landley | I see you fit the speaker in there. :) |
| 18:13.14 | landley | A USB host connector was my initial suggestion. |
| 18:13.15 | prpplague | hehe indeed |
| 18:13.31 | landley | I'm torn on that one. |
| 18:13.37 | landley | ON the one hand, it's more flexible. |
| 18:13.45 | prpplague | landley: yea, i remember, |
| 18:14.02 | landley | On the other hand, 90% of the use is a network connection, and that means you need an extra piece of hardware. |
| 18:14.16 | landley | Plus you're powering the board _and_ who knows what other device off the same USB port. |
| 18:14.25 | prpplague | exactly |
| 18:14.27 | landley | How does the board pass the power request upstream to the laptop? |
| 18:14.52 | SpeedEvil | the USB host on the laptop talks to the USB peripheral |
| 18:14.58 | SpeedEvil | it reads the config descriptors |
| 18:15.04 | landley | If it's easily modular to plug one or the other, you could always do both. |
| 18:15.07 | SpeedEvil | in these is a 'I need xxx ma' |
| 18:15.17 | prpplague | landley: when the usb device first enumerates it only draws about 10mA, it then says what the max power is going to be, the host side then says if thats ok or not |
| 18:15.19 | landley | SpeedEvil: so the USB device can change its max power request retroactively? |
| 18:15.20 | SpeedEvil | it then configures the USB device, and it can read it |
| 18:15.27 | SpeedEvil | landley: I don't think so |
| 18:15.29 | prpplague | landley: if its ok the device powers the rest of the device up |
| 18:15.38 | landley | The hammer board won't know what other USB devices you could plug into its downstream connection. |
| 18:15.43 | SpeedEvil | landley: it's typically at plug-in-time. It could fake an unplug |
| 18:16.00 | landley | SpeedEvil: if the hammer fakes an unplug, does it stop getting power for its arm processor? |
| 18:16.31 | landley | "Dear phone company, please disconnect my phone. Now, I want you to... hello?" |
| 18:16.36 | T0mW | help |
| 18:16.46 | landley | T0mW: hmmm? |
| 18:16.59 | prpplague | landley: basically we will set the main usb interface up to the max request of 500mA, then we will tell the hammer not to allow anything to enumberate with a power request greater than 300mA |
| 18:17.01 | T0mW | Why, at start_kernel(), is console_drivers == NULL |
| 18:17.17 | landley | T0mW: I'm guessing a config issue? |
| 18:17.21 | landley | Do you have early_printk enabled? |
| 18:17.36 | prpplague | landley: its doable |
| 18:17.48 | landley | prpplague: I'll happily leave that up to you. |
| 18:17.49 | T0mW | well, I have no console output after "bootin" |
| 18:18.10 | prpplague | landley: based on a sampling of bluetooth, thumbdrives, wifi and ethernet dongles, most are below the 300mA mark |
| 18:18.15 | T0mW | it hits start_kernel, but without console drivers. |
| 18:18.19 | landley | The upside of a downstream USB connection is extra flexibility. The downside is you need more hardware to use it for the common case of ethernet, and have power management issues. |
| 18:18.35 | prpplague | landley: indeed |
| 18:18.41 | landley | USB does let you mount a thumb drive, which is highly cool. |
| 18:18.59 | prpplague | landley: yep |
| 18:19.19 | landley | Heck, you could plug in an externally powered USB hub and have mouse, keyboard, video, thumb drive, and ethernet. |
| 18:19.40 | landley | If you plugged the device into said USB hub for power... Wow, that might confuse it. :) |
| 18:19.43 | prpplague | landley: well the video would be a little difficult |
| 18:19.50 | prpplague | hehe |
| 18:19.56 | landley | prpplague: says who? CompUSA's going out of business sale has a USB video card. |
| 18:20.02 | landley | "Add a second monitor to any machine." |
| 18:20.15 | landley | Boggles the mind. |
| 18:20.17 | prpplague | landley: i don't believe those are supported under linux current |
| 18:20.20 | prpplague | currently |
| 18:20.25 | landley | Give it three months. |
| 18:20.28 | prpplague | landley: hehe |
| 18:20.41 | landley | Or I could just buy one and mail it to Greg KH's "we write drivers" project. :) |
| 18:20.50 | cbrake | prpplague: heh libertas: can't load helper firmware |
| 18:20.57 | landley | I already need to send them one of the sprint USB cellular dongles. |
| 18:21.05 | landley | "No unsupported devices"? Ha! |
| 18:21.12 | *** join/#edev sunrunner20 (n=jamesdoe@pool-71-170-156-64.dllstx.fios.verizon.net) |
| 18:21.19 | prpplague | landley: plus iirc the usb video card requires usb2 high speed connections |
| 18:21.26 | landley | T0mW: what console devices do you have configured in? |
| 18:21.41 | T0mW | ttySAC0, essentially |
| 18:21.43 | landley | prpplague: requires, or is just dog slow without it? |
| 18:22.04 | landley | T0mW: dunno that one. |
| 18:22.07 | landley | What board is this? |
| 18:22.11 | prpplague | landley: thats just it, i i'm not sure |
| 18:22.36 | sunrunner20 | hey |
| 18:22.39 | landley | prpplague: so you're calling it the nail board? |
| 18:22.48 | sunrunner20 | GPIO is digital not analog right? |
| 18:22.54 | T0mW | landley: mine |
| 18:23.18 | landley | T0mW: the bootloader can talk to this serial port? |
| 18:23.22 | cbrake | prpplague: where does one get 8385 firmware? |
| 18:23.37 | sunrunner20 | lol |
| 18:23.44 | prpplague | cbrake: extract it from winblows drivers |
| 18:23.46 | landley | After 25 years, _still_... |
| 18:24.16 | T0mW | landley: s3c2410, and yes, I get messages all the way up to, and including, "Uncompressing Linux...done. bootin." |
| 18:24.18 | prpplague | landley: yea, Hammer and Nail |
| 18:24.21 | prpplague | landley: haha |
| 18:24.41 | prpplague | T0mW: are you sure its completely booting? |
| 18:24.59 | prpplague | T0mW: my suggestion is for you to enable printascii and shove one in the printk.c file |
| 18:25.05 | T0mW | if I break on start_kernel via JTAG, I can manually send chars out the uart via the JTAG debgger (fill TXDATA). |
| 18:25.18 | T0mW | prpplague: oh, it is up. |
| 18:25.18 | landley | prpplague: works for me. :) |
| 18:25.31 | prpplague | T0mW: how do you know? |
| 18:26.29 | prpplague | T0mW: i'd be willing to bet $5 that you have the incorrect cpu id between the atag and your kernel |
| 18:26.55 | T0mW | I put a breakpoint, via the BDI 2000 not gdb, on 0xc000885c, which is start_kernel, it hits that breakpoint, I can then connect gdb and single-step. |
| 18:27.28 | T0mW | prpplague: that is something for me to look at, thanks, it is a roll-your-own kernel. |
| 18:27.34 | T0mW | bbl, looking |
| 18:28.33 | T0mW | prpplague: that is what I was thinking, there is a point in the arch/arm/kernel source that setsup an initial uart, dunno, been a while. |
| 18:28.38 | landley | Darn it. |
| 18:28.40 | hapchap_ | December 18th and it is 63 degrees outside |
| 18:28.46 | landley | User Mode Linux is being clever. :( |
| 18:28.51 | T0mW | hapchap_: where are you!? |
| 18:28.52 | prpplague | T0mW: easiest is to just add a printascii |
| 18:29.00 | T0mW | k |
| 18:29.00 | hapchap_ | T0mW: north texas |
| 18:29.08 | hapchap_ | T0mW: DFW area |
| 18:29.21 | prpplague | T0mW: hapchap_ is a aml flunky like me |
| 18:29.24 | prpplague | T0mW: http://www.abcsinc.com/small-linux/printascii.txt |
| 18:29.26 | T0mW | hapchap_: wow, I have ice out there on top of three inches of snow. |
| 18:29.43 | prpplague | T0mW: all you have to do is the bottom portion |
| 18:29.47 | hapchap_ | T0mW: kind of hard to get into the christmas spirit when its this hot |
| 18:29.57 | T0mW | prpplague: didn't they put printascii into the kernel yet? |
| 18:30.12 | prpplague | T0mW: its in the kernel, but not added to the printk.c file |
| 18:30.21 | prpplague | T0mW: which kernel are you using? |
| 18:30.55 | T0mW | prpplague: 2.6.20.4, with Systemmap saying: c00227a4 T printascii |
| 18:31.08 | prpplague | T0mW: that just means you have printascii enabled |
| 18:31.10 | T0mW | prpplague: I'll check it out |
| 18:31.16 | T0mW | thanks guys, bbl. |
| 18:31.16 | prpplague | T0mW: doesn't mean that its in the printk.c |
| 18:31.19 | prpplague | T0mW: later |
| 18:31.33 | T0mW | I just ran out of ideas last night |
| 18:32.27 | T0mW | prpplague: you know that 2.6.20.4 is the last release before they did bonehead surgery on the DMA, right? |
| 18:32.31 | T0mW | :) |
| 18:33.00 | sunrunner20 | bonehead surgyer? |
| 18:33.05 | sunrunner20 | A*surgery |
| 18:35.20 | prpplague | T0mW: yea |
| 18:35.38 | landley | Everybody's always talking about "the good old linux kernels of old", but nobody can ever agree on which versions those were... |
| 18:35.40 | prpplague | T0mW: well all you need to do is just add the printascii call in printk |
| 18:35.52 | landley | prpplague: what is printascii? |
| 18:35.54 | landley | what does it do? |
| 18:36.13 | prpplague | landley: extremelly lowlevel access to uart |
| 18:36.28 | prpplague | landley: basically a printk function but super super lowlevel |
| 18:36.51 | prpplague | landley: mainly used for debugging kernel booting before uart drivers are registered |
| 18:37.00 | landley | I've heard of that. I thought early_printk used it... |
| 18:37.38 | prpplague | landley: iirc it does, but early_printk doesn't always get problems such as the CPU id issue |
| 18:37.55 | prpplague | landley: or some of the early arm macros for machine identification |
| 18:38.07 | prpplague | landley: its a delicate dance between the bootloader and the kernel |
| 18:38.50 | landley | One of these days, I want to build just enough of the kernel infrastructure to print "hello world" and then HLT. |
| 18:38.55 | adyer | speaking of the dance between kernel and loader, anybody know about the ppc flat device tree? |
| 18:39.04 | landley | I want to see what the truly minimal set of files needed to do this _is_... |
| 18:39.16 | prpplague | hehe |
| 18:39.18 | landley | I know a tiny bit about it... |
| 18:39.26 | landley | They hang out on #mklinux if I recall... |
| 18:39.46 | landley | That's all the Linux ppc guys. |
| 18:39.52 | adyer | i see alot of work in u-boot, but i mostly do mips and arm, so I'm wondering what it's all about |
| 18:40.05 | landley | adyer: wander over to #mklinux and ask. |
| 18:40.10 | landley | Actually, there's a Documentation file on it now... |
| 18:40.47 | adyer | is it pretty much atags on steroids? |
| 18:40.56 | landley | Look at Documentation/powerpc/booting-without-of.txt |
| 18:41.03 | landley | That's for booting without Open Firmware. |
| 18:41.08 | adyer | k, thnaks |
| 18:41.17 | landley | By compiling your own device tree and statically linking it into the kernel image. |
| 18:41.18 | adyer | s/na/an/ |
| 18:50.19 | *** join/#edev MothaWork (n=monmotha@149.164.193.61) |
| 19:00.30 | T0mW | hmm, my machine type (reg R1 from apex) is not listed in arch/arm/tools/mach-type |
| 19:00.47 | T0mW | prpplague: hey, where can I get a new machine type ID? |
| 19:01.05 | T0mW | prpplague: is there someone who is giving these out? |
| 19:04.47 | prpplague | T0mW: hehe |
| 19:04.51 | prpplague | T0mW: so i was right? |
| 19:05.03 | prpplague | T0mW: what number is apex spitting out? |
| 19:05.10 | T0mW | 725 |
| 19:05.32 | T0mW | that is not in arch/arm/tools/mach-type of 2.6.20.4 |
| 19:05.44 | prpplague | littlechips? interesting |
| 19:05.51 | T0mW | in fact, that file looks suspiciously empty |
| 19:05.56 | prpplague | i didn't know i had added littlechips to apex |
| 19:06.21 | T0mW | there is only 366 entries in that file |
| 19:06.26 | prpplague | T0mW: http://www.arm.linux.org.uk/developer/machines/ |
| 19:06.29 | prpplague | T0mW: thats odd |
| 19:06.33 | T0mW | I know |
| 19:06.59 | T0mW | it is from ftp.kernel.org though |
| 19:07.03 | prpplague | T0mW: anyway, be 100% certain you want to register a board |
| 19:07.27 | T0mW | prpplague: well, we will need a type number as "something" will go into production soon. |
| 19:07.33 | T0mW | won't we? |
| 19:07.37 | prpplague | T0mW: this is a custom s3c2410 board you've done? |
| 19:08.00 | T0mW | prpplague: embest did the board, they never got a type number, as yet. |
| 19:08.13 | prpplague | T0mW: you sure of that? |
| 19:08.26 | T0mW | prpplague: it is not in russell's list either |
| 19:08.42 | T0mW | http://www.arm.linux.org.uk/developer/machines/ |
| 19:09.09 | prpplague | T0mW: okie dokie |
| 19:09.30 | prpplague | T0mW: well, just read up on the procedure, rmk gets pissy when you don't RTFM |
| 19:09.45 | T0mW | LOL, pissy is a mild word |
| 19:10.29 | prpplague | T0mW: btw, where is my $5 ??? |
| 19:10.37 | T0mW | I think rmk told Dale Carnagy what to do with himself. |
| 19:10.46 | T0mW | prpplague: profit from hammer sale. |
| 19:10.50 | prpplague | hehe |
| 19:11.57 | *** join/#edev ico2_ (n=ico2@modem-2197.cougar.dialup.pol.co.uk) |
| 19:13.27 | T0mW | prpplague: ok, why is the mach-type number important? I know that I can just "steal" a number that works, but why should I even bother with getting a new number for a new board? |
| 19:14.21 | prpplague | T0mW: along with just a number you get a dedicated macro that identifies your machine, so that you can do board specific code |
| 19:15.35 | prpplague | T0mW: i.e. if you have a special power management feature you can do things like machine_is_amlm5900() |
| 19:15.49 | T0mW | prpplague: yeah, I know all that. |
| 19:16.43 | prpplague | T0mW: the other aspect is that if you push you base configuration into the tree, it gets tested and updated to some extent for each kernel release |
| 19:16.56 | adyer | i think the idea is to be able to make one kernel that will do the "right thing" on multiple boards as determined by the id |
| 19:17.09 | T0mW | so, CONFIG_MACH_TYPE == xxx has to be unique so distro-ed code doesn't collide. Where you have the same type number, you could end up activating code segments that you didn't want, correct? |
| 19:17.31 | prpplague | T0mW: so to answer your question, you don't have to register a number, but its more likely you may run into problems where it things its a littlechip board and needs to do x , when its a embest board and needs to do y |
| 19:17.41 | T0mW | k |
| 19:17.47 | prpplague | T0mW: exactly |
| 19:19.49 | prpplague | jeeze - http://www.foxnews.com/story/0,2933,317202,00.html |
| 19:20.04 | prpplague | the US school system is totally broken |
| 19:20.05 | *** join/#edev Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) |
| 19:21.11 | hapchap_ | prpplague: felony weapons charges?!?!?! |
| 19:21.35 | hapchap_ | prpplague: zero tolerance = no common sense? |
| 19:21.36 | landley | prpplague: that made Fark last week. |
| 19:21.53 | landley | The US school system has been totally broken since the funding from "sputnik" tailed off. |
| 19:22.00 | prpplague | landley: indeed |
| 19:22.12 | landley | Keep in mind it's primarily a government funded day care system. |
| 19:22.13 | chouimat|work | prpplague: WOW |
| 19:22.21 | prpplague | i'm wondering what has happened to the kid that got in trouble for using firefox |
| 19:22.29 | landley | Once upon a time, child labor was the norm. That's why the summer vacation: so kids could help with the crops. |
| 19:23.30 | landley | The current education system is tailored to mass production, invented by good old Eli Whitney in the late 1800's (who called it the "american system"). |
| 19:23.42 | adyer | prpplague the firefox thing was a hoax, a photoshopped letter |
| 19:24.05 | prpplague | adyer: ahh i was reading now |
| 19:24.12 | chouimat|work | landley: yup and in Canada it will make more sense now to close the school in winter (will save a lot of money because they won't need to heat the building that much) |
| 19:24.13 | landley | All the desks facing front, read your book, shut up and listen to the one teacher who has chalk and a blackboard (the height of 1800's technology)... |
| 19:24.36 | landley | The problem is, mass production stopped being interesting in about the 1970's. |
| 19:24.43 | prpplague | landley: indeed, and even more so in the english style schools |
| 19:24.52 | landley | These days the jobs that pay need creativity. We automated the rest. |
| 19:25.01 | prpplague | landley: exactly |
| 19:25.30 | landley | El shrubbo's "no child left behind" act was probably the nail in the coffin. (The entire class will move at the pace of the slowest student! Brilliant!) |
| 19:25.50 | landley | I plan to home school. |
| 19:26.05 | prpplague | landley: yep, my kids classes are at a crawl due the illegals |
| 19:26.15 | landley | I skipped my senior year of high school to go to a community college. |
| 19:26.20 | prpplague | landley: i do alot of educational stuff with our kids |
| 19:26.30 | landley | My main regret is I didn't skip straight from about fifth grade into the community college. |
| 19:26.37 | prpplague | hehe |
| 19:26.47 | landley | At BCC I tutored for extra money. |
| 19:26.47 | prpplague | landley: indeed my highschool was a total waste |
| 19:26.57 | landley | I tutored "math 001". Addition, subtraction, multiplication, and division. |
| 19:27.16 | landley | There are students in a community college who don't know that stuff. So why the heck does middle school even bother? |
| 19:27.23 | prpplague | landley: uh at the cc ? |
| 19:27.28 | landley | Yup. |
| 19:27.35 | prpplague | omg |
| 19:27.39 | landley | Yup. |
| 19:27.45 | adyer | my old high school is on a watch list because the 5% of the school that is latino is underperforming - rest of the school is well above average |
| 19:27.47 | landley | They had multiple sections of that course... |
| 19:28.05 | landley | You can teach division to 3 year olds. |
| 19:28.26 | landley | Grab four 3 year olds and give them 12 M&M's. Tell them they can only eat them if they each get the same amount. |
| 19:28.30 | landley | They'll work it out pretty quickly. |
| 19:28.35 | adyer | lol |
| 19:28.39 | landley | No, it's true. |
| 19:28.45 | prpplague | landley: when we were in barbados, the school there was soooooo uninterested in creativity |
| 19:28.46 | landley | I used to work for a company making educational software. |
| 19:28.53 | landley | One of the co-founders was an education researcher for UT. |
| 19:28.54 | chouimat|work | landley: with how many bruses? (???) |
| 19:28.55 | adyer | i know it is, i have 7 & 5 at home |
| 19:29.08 | landley | chouimat|work: I didn't say unsupervised. :) |
| 19:29.22 | chouimat|work | landley: then it less fun for the kids ;) |
| 19:29.38 | landley | I repeat, public school is federally funded day care, with only a vestigial educational purpose. |
| 19:30.06 | prpplague | landley: my son used to get marks on his conduct card for not doing math homework as he was told, yet he always got the answer correct |
| 19:31.10 | landley | Paul graham wrote a marvelous essay about this: http://www.paulgraham.com/nerds.html |
| 19:31.30 | chouimat|work | landley: I agree what the grade mean anyway? that you understand the "stuff" or that you're able to remember what the moron explained? |
| 19:31.33 | landley | First time I've heard a detailed comparison between public school and the 17th century french court. :) |
| 19:41.43 | chouimat|work | adyer: me too but I don't want to pay to replace the thing I will break :) |
| 19:41.45 | adyer | client can't figure out how to install git |
| 19:42.05 | adyer | client can't figure out how to use patches |
| 19:42.20 | adyer | client says "oh, just send me that one file" |
| 19:42.35 | chouimat|work | adyer: tell the client to repack his computer and return it :) |
| 19:43.43 | adyer | I spent a fair amount of time setting up a git repository, just so we could all be running off the same source :-( gggggrrrrrrr |
| 19:44.18 | adyer | I want to send a Dalek to his office |
| 19:45.13 | ds2 | but they were all destoryed at the end of the last time war... |
| 19:45.43 | adyer | "exterrrrrrrminate!" |
| 20:23.19 | landley | ds2: says who? |
| 20:23.25 | landley | They've popped up in at least 3 episodes since then. |
| 20:23.58 | landley | http://www.youtube.com/watch?v=XEtMNrK9nHQ |
| 20:26.59 | landley | adyer: http://kernel.org/doc/local/git-quick.html |
| 20:27.05 | landley | I wrote a git quickstart. :) |
| 20:27.23 | landley | Back when I was doing documentation for the linux foundation... |
| 20:27.47 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 20:27.50 | prpplague | landley: hehe, my son loves the daleks |
| 20:27.58 | prpplague | landley: he knows their entire history, hehe |
| 20:28.09 | prpplague | he's been pestering me to build one using the hammer |
| 20:28.47 | adyer | landley, his problem is installing the package on the host os. (he mumbled something about ubuntu) |
| 20:29.22 | landley | adyer: that says how to build it from source. :) |
| 20:29.26 | landley | Step by step. |
| 20:29.31 | landley | Installing the ubuntu package sucked. |
| 20:30.06 | adyer | if he can't use synaptic or the ubuntu forums to get it done, I have little hope for that... |
| 20:31.31 | adyer | at least he's apologetic about his failings :-| |
| 20:31.40 | landley | adyer: point him at my quickstart. |
| 20:31.52 | landley | I'm always interested in feedback from the truly ignorant, when it comes to documentation. :) |
| 20:33.15 | prpplague | landley: as am i |
| 20:33.26 | prpplague | hehe |
| 20:36.33 | landley | I have the hard-fought ability to emulate ignorance. |
| 20:36.36 | landley | Ok, some of it's natural. |
| 20:36.40 | landley | "What was that bit again?" |
| 20:36.41 | adyer | landley btw if you're looking for gpl violations with busybox, I'm pretty sure these guys http://www.tothegreen.com/ are in violation :-) |
| 20:36.58 | landley | adyer: gpl@busybox.org is the report email address. |
| 20:37.04 | landley | That goes directly to the SFLC lawyers. |
| 20:37.18 | landley | I don't even try to keep track anymore, there were too many (and some went to Erik, some to me, some to Denis...) |
| 20:37.22 | landley | Please tell them. :) |
| 20:37.31 | landley | er, busybox.net I think. |
| 20:37.32 | landley | :P |
| 20:37.35 | landley | gpl@busybox.net |
| 20:38.28 | landley | Yeah, busybox.org is some guy's personal website with photoes of a canoe. |
| 21:14.39 | hapchap_ | i dont think its a canoe (looks to be powered) |
| 21:18.46 | *** join/#edev roxfan (n=dunno@180.217-200-80.adsl-dyn.isp.belgacom.be) |
| 21:21.45 | T0mW | prpplague: mach-type 1583 |
| 21:35.02 | *** join/#edev ico2_ (n=ico2@modem-3994.cougar.dialup.pol.co.uk) |
| 21:42.30 | prpplague | hmm, by being x86 compatible by default imply the processor has a FPU? |
| 21:43.13 | adyer | 486SX, 386EX, 286 all didn't |
| 21:43.25 | prpplague | ahh |
| 21:43.37 | prpplague | i forgot about those |
| 21:43.44 | prpplague | been a long time since i did x86 |
| 21:43.54 | adyer | but for practical purposes these days I would say yes |
| 21:53.02 | adyer | on linux shouldn't it trap back to a kernel handler? |
| 21:55.19 | prpplague | CosmicPenguin: ping |
| 21:56.18 | CosmicPenguin | pong |
| 21:59.38 | adyer | ahh, the great s/w excuse, "I'm compiling" |
| 22:01.30 | T0mW | hmm, printascii doesn't work |
| 22:02.03 | prpplague | CosmicPenguin: you ever use the logo on the XO? |
| 22:02.17 | prpplague | T0mW: uh pastebin.ca me a copy of your printk.c |
| 22:02.22 | CosmicPenguin | prpplague: yeah |
| 22:02.46 | prpplague | CosmicPenguin: just looking at it now, what kind of size is that? |
| 22:03.09 | prpplague | CosmicPenguin: some of the basics of logo be able to put on a small arm like the hammer? |
| 22:03.41 | prpplague | T0mW: still there? |
| 22:03.51 | T0mW | prpplague: yeah |
| 22:04.11 | prpplague | T0mW: paste me your printk.c file |
| 22:05.07 | CosmicPenguin | no - its not really hammer friendly |
| 22:05.14 | T0mW | prpplague: nm |
| 22:05.15 | prpplague | CosmicPenguin: big? |
| 22:05.25 | prpplague | hmm |
| 22:05.33 | CosmicPenguin | no - its more graphically oriented then you need |
| 22:05.34 | prpplague | i noticed there are some python extensions |
| 22:06.04 | prpplague | CosmicPenguin: well what i was wanting is robotics based, which the graphics movements would correspond to |
| 22:06.10 | T0mW | prpplague: http://rafb.net/p/ijRix237.html |
| 22:06.11 | CosmicPenguin | yeah - no |
| 22:06.14 | CosmicPenguin | its not even close to that |
| 22:06.20 | prpplague | CosmicPenguin: oh ok |
| 22:06.53 | prpplague | T0mW: yea, you don't have the your cpu init data |
| 22:07.03 | prpplague | T0mW: paste me your mach-xxxx.c file |
| 22:08.06 | T0mW | prpplague: http://rafb.net/p/jZi4y918.html |
| 22:08.35 | T0mW | doh |
| 22:08.39 | T0mW | I see it |
| 22:09.08 | T0mW | static void __init digi2410_init(void) |
| 22:09.09 | T0mW | { |
| 22:09.09 | T0mW | digi2410_map_io(); |
| 22:09.09 | T0mW | } |
| 22:09.14 | CosmicPenguin | prpplague: the xo logo is really graphically oriented |
| 22:09.19 | T0mW | I need to call the init stuff |
| 22:09.35 | CosmicPenguin | you put together blocks of code and the turtle does it - all mouse driven, point and click visual stuff |
| 22:09.50 | CosmicPenguin | what you want is more of just the basic interpeter running on text files |
| 22:10.00 | CosmicPenguin | and hooks coming out for the software |
| 22:11.10 | prpplague | T0mW: yea i don't see where your devices are added |
| 22:11.17 | prpplague | T0mW: have a look at the tct hammer init |
| 22:11.20 | T0mW | barebones |
| 22:11.33 | T0mW | just want it to panic at no init found |
| 22:12.53 | prpplague | T0mW: ?? |
| 22:14.52 | prpplague | T0mW: if you are using the __init portion and the function is empty, i.e. it does nothing, it gets thrown away |
| 22:21.08 | T0mW | prpplague: I'm now calling digi2410_map_io from digi2410_init. digi2410_map_io() has the call to s3c24xx_init_clocks(0), that is right, right? |
| 22:24.29 | prpplague | T0mW: are you planning to do something special with the irqs? |
| 22:27.02 | T0mW | doh |
| 22:30.11 | prpplague | T0mW: find something? |
| 22:31.27 | T0mW | yeah, I needed to create an empty iodesc struct, then call s3c24xx_init_io passing the empty struct from within init. Now I have printing from console |
| 22:32.09 | T0mW | It now inits the cpu var and s3c24xx_init_clocks() is happy. I get all the way down to: |
| 22:32.09 | prpplague | T0mW: hehe |
| 22:32.27 | T0mW | RAMDISK: Compressed image found at block 0 |
| 22:32.27 | T0mW | VFS: Mounted root (ext2 filesystem) readonly. |
| 22:32.27 | T0mW | Freeing init memory: 84K |
| 22:32.27 | T0mW | BUG: soft lockup detected on CPU#0! |
| 22:32.34 | T0mW | and then it crash-dumps |
| 22:33.06 | prpplague | T0mW: well thats interesting |
| 22:33.14 | T0mW | prpplague: NO IT IS NOT! |
| 22:33.19 | T0mW | heheh |
| 22:33.28 | T0mW | getting closer though |
| 22:34.43 | T0mW | prpplague: all I'm adding is a barren mach-digi2410.c into arch/arc/mach-s3c2410. Nothing is turned on: spi, net, i2c, etc. only the: console on s3c2410 uart |
| 22:35.01 | prpplague | T0mW: yea i noticed |
| 22:35.33 | T0mW | prpplague: I figured that I would get it to a login (/sbin/init) prompt, then add in the hardware I need. |
| 22:36.03 | T0mW | the s3c2410 serial ports are also enabled. |
| 22:36.10 | prpplague | T0mW: yea |
| 22:43.54 | *** join/#edev dijenerate (n=dijenera@72.51.93.247) |
| 23:01.53 | *** join/#edev davidc__ (n=davidc__@S01060002b360aacd.vc.shawcable.net) |
| 23:04.32 | *** join/#edev sunrunner20|prom (n=jamesdoe@pool-71-123-227-162.dllstx.fios.verizon.net) |
| 23:05.16 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 23:05.16 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
