IRC log for #edev on 20070816

00:34.08*** join/#edev g1powermac_PB (n=g1powerm@CPE-76-177-156-202.natcky.res.rr.com)
01:28.01*** join/#edev redblue (i=star@ppp115.108-253-207.mtl.mt.videotron.ca)
02:25.49*** join/#edev linac (n=lin@59.72.77.125.broad.fz.fj.dynamic.163data.com.cn)
03:22.29*** join/#edev redblue (i=star@ppp115.108-253-207.mtl.mt.videotron.ca)
03:47.56*** join/#edev DrZimmerman (n=theo@2001:5c0:8fff:fffe:0:0:0:6af1)
04:13.22*** join/#edev silverpower (n=shrike@atheme/bofh/silverpower)
04:15.42*** join/#edev silverpower_ (n=shrike@adsl-234-207-23.cha.bellsouth.net)
07:25.55*** join/#edev steliosk (n=Stelios@ipa226.211.tellas.gr)
07:39.02*** join/#edev redblue (i=star@ppp240.108-253-207.mtl.mt.videotron.ca)
09:10.25*** join/#edev hw__ (n=hw@p578b3905.dip0.t-ipconnect.de)
11:10.51*** join/#edev g1powermac_PB (n=g1powerm@CPE-76-177-156-202.natcky.res.rr.com)
11:54.13*** join/#edev g1powermac_PB (n=g1powerm@CPE-76-177-156-202.natcky.res.rr.com)
12:10.15*** join/#edev linac (n=lin@125.213.46.243)
12:36.53*** join/#edev prpplague (n=billybob@mail.americanmicrosystems.com)
12:43.20prpplagueg1powermac_PB: ping
12:53.24g1powermac_PBprpplague: pong, but I'm right about to leave
12:53.27g1powermac_PBI'll bbiab
12:57.07*** join/#edev GPSFan (n=kenm@12.10.255.248)
12:57.07*** mode/#edev [+o GPSFan] by ChanServ
13:00.27g1powermacprpplague, back :-)
13:01.18prpplagueg1powermac: the url you listed for a driver project for the onboard wifi is a dead project
13:01.26g1powermacoh
13:01.37g1powermacdo you have a different link?
13:01.38prpplagueg1powermac: the correct location for the source is the git.infradead.org
13:01.42prpplagueg1powermac: libertas
13:01.49g1powermack, thanks
13:02.22g1powermachttp://git.infradead.org/?p=libertas-2.6.git;a=summary ?
13:03.18prpplagueyea
13:03.28g1powermacok, its updated
13:05.24g1powermacprpplague, does anything else need to be changed?
13:05.35prpplaguei think thats the only thing that jumped out
13:05.37prpplagueit looks nice
13:05.43g1powermacthanks :-)
13:08.03prpplaguenp
13:10.20*** join/#edev brad_mssw (n=brad@shop.monetra.com)
13:32.45g1powermacprpplague, hmm, I'm trying to mount one of what appears to be jffs2 images from the firmware of the more expensive sip phone
13:33.25g1powermacprpplague, however, mount is whining about wrong fs type or bad option or bad superblock on /dev/loop0
13:33.42prpplagueg1powermac: depends on what type of image it is
13:33.49g1powermacI'd be kinda surprised if it isn't jffs2, since thats actually in the name of the image
13:34.06g1powermacprpplague, it file extension ends in .raw
13:34.10g1powermacs/it/the
13:34.14prpplagueg1powermac: ahh, you can't directly mount a jffs2 image
13:34.28g1powermacwell, I am doing a loopback mount
13:34.50prpplagueg1powermac: yea, thats what i mean
13:34.58g1powermacoh, so how would I mount it?
13:35.19prpplagueg1powermac: there are some jffs2 utils that you can use to extract a copy of the contents
13:35.29g1powermachmm
13:36.26g1powermacjfsutils ?
13:36.40prpplagueyea
13:37.06prpplagueg1powermac: i generally don't use jffs2 images for that reason
13:39.29g1powermachmm
13:39.39g1powermacwould you happen to know how to do this using jfsutils?
13:41.05prpplagueg1powermac: http://linux-7110.sourceforge.net/howtos/netbook_new/x1125.htm
13:41.17g1powermacthanks
13:41.34Crofton|homeany gumstix gurus handy?
13:42.11prpplagueCrofton|home: hehe, not i
13:42.38Crofton|homejust trying to figure out how to build u-boot for it
13:44.00prpplagueCrofton|home: hehe yea, thats what i dislike about u-boot
13:44.34Crofton|homewell, it would not be an issue if the code had gone upstream
13:47.52prpplagueCrofton|home: yea, that's also a problem i have with gumstix
13:49.34g1powermacCrofton|home, I have a gumstix, but haven't used it
13:50.41Crofton|homeI have seen emails that make me need to care a very small amount about gumstix over the past few days
13:51.03prpplagueCrofton|home: hehe, why aren't you holding out for a hammer board?
13:51.47Crofton|homethis is a case of other people using gumstix and asking me questions
13:52.42Crofton|homepersonally, I do not understand why a SDR person would fool with such a thing :)
13:53.39g1powermacprpplague, woot, awesome, thanks for the info
13:54.51prpplaguegood deal, please document your work
13:54.58g1powermacyup
13:55.02prpplagueg1powermac: start a "how to mount a jffs image"
13:55.02g1powermacI will
13:55.18prpplagueg1powermac: remember a good "scientist" documents his work
13:55.26g1powermachehe, yes
13:55.29prpplagueg1powermac: (if anything to prove you know what you did!)
13:57.02g1powermacnow the one thing I'm a bit confused about with the firmware images is actually there's multiple images
13:57.14GPSFang1powermac: here is another link for you to look at re. mounting jffs2 http://www.linux-mtd.infradead.org/faq/jffs2.html#L_mtdblock
13:57.32g1powermacgot rootfs, homefs, u-boot, and some encrypted ones
13:58.16g1powermacGPSFan, ahh, even nicer
13:58.30GPSFanand: http://www.handhelds.org/hypermail/familiar/62/6232.html
14:08.44g1powermacthink this might deserve a simple bash script. . .
14:17.24*** join/#edev ico2 (n=ico2@modem-2160.chimpanzee.dialup.pol.co.uk)
14:24.25*** join/#edev CosmicPenguin (n=nobody@crank.laptop.org)
14:35.26prpplagueCosmicPenguin: greetings earthling
14:35.31prpplagueCosmicPenguin: hows life in CO?
14:36.30CosmicPenguinsameol
14:36.37CosmicPenguinnothing ever changesa round here
14:36.57prpplaguehehe
14:36.59chouimatmorning
14:37.07prpplagueCosmicPenguin: i'd like to go do a beer tour sometime
14:37.16CosmicPenguinany time
14:37.33prpplaguechouimat: greetings my canadian friend
14:37.34CosmicPenguinOdells, New Belgium, Left Hand, Flying Dog, Avery, Boulder
14:37.42CosmicPenguinOskar Blues
14:37.42chouimatprpplague: hi
14:37.45prpplagueplenty to choose from
14:38.30CosmicPenguinSome good local brew joints too
14:40.11prpplaguethere are only about 20 microbrews in the entire state of texas
14:40.23wmatprpplague: that sucks
14:40.33brad_msswnot much here if FL either
14:40.37brad_mssws/if/in/
14:41.37prpplagueCosmicPenguin: basically the entire island of barbados will be shutdown tomorrow in preparation for the rain from dean
14:41.50CosmicPenguinYeah - I saw they were under a ts warning
14:42.58prpplagueCosmicPenguin: hehe, rhinna bought her grandmother a new house about a block from donna's mothers house
14:43.14prpplagueCosmicPenguin: hehe she was there helping her move
14:43.20prpplagueCosmicPenguin: rhinna that is
14:43.39prpplaguerhianna, i always leave the "a" out
14:49.25prpplagueas well as push button
14:53.49chouimat2 huge hdd to go ... :(
14:55.19prpplaguehmm
14:55.33prpplaguei can't seem to find where to declare the led gpio in the leds driver
15:02.54g1powermacneed to make it a bit more robust, but it does work
15:05.06g1powermaca bit interesting note for it
15:05.21g1powermacyou have to give the kernel a bit of time to load the module first before doing the dd command
15:06.41g1powermacheh
15:06.59agaffneyheh
15:07.08chouimatCosmicPenguin: have fun :)
15:07.11agaffneyI've done many unspeakable things with tar in the past
15:07.45chouimatagaffney: tar and the neighbor's dog is a great combination :)
15:07.48agaffneytar -cp . | ssh somehost 'tar -C /somedir -xp' is one of them :P
15:08.00agaffneygreat when you don't have rsync on both ends
15:08.43chouimatagaffney: I have seem a file where the guy use shell scripts and dd to implement tar
15:08.48CosmicPenguinMy eviil is mainly centereda round --transform
15:11.47agaffney--transform?
15:12.13agaffneymy gnu tar doesn't have that option
15:12.43chouimatagaffney: it's not evil enough then
15:12.45CosmicPenguinwhat version?
15:15.00CosmicPenguinwow - that works neat
15:16.08*** join/#edev likewise (n=chatzill@82-217-243-87.cable.quicknet.nl)
15:17.42agaffneyhmm, I've got tar 1.18 and the online manual I found for 1.18 mentions transform
15:17.56agaffneyhowever, I bet my tar man page comes from the man-pages package
15:18.00agaffneywhich probably isn't updated
15:18.11likewiseWhich channel is for C/POSIX/Linux developer questions? I have a mmap() question...
15:18.17agaffneyah, yep
15:18.21agaffney'tar --help' shows it
15:18.28CosmicPenguinagaffney: ubuntu/debian?
15:18.32g1powermacprpplague, http://www.elinux.org/Mount_jffs2
15:18.32agaffneygentoo
15:18.36CosmicPenguinahhh
15:18.53agaffney<PROTECTED>
15:18.58agaffneythat's kinda nifty
15:18.58g1powermacvery simple bash script since I don't do much bash scripting :-)
15:19.05CosmicPenguinthe bottom of my tar man page digresses into the evilness of the GNU people and their documentation license
15:19.14prpplagueg1powermac: lovely
15:19.30CosmicPenguinagaffney: also, I've never used --strip-components either
15:19.33CosmicPenguinthat seems useful
15:19.43agaffneyCosmicPenguin: hah
15:19.59agaffneyyeah, I just saw --strip-components when I did 'tar --help | grep -C2 transform'
15:20.09agaffneythat's handy, too
15:20.49chouimatBUGS
15:20.49chouimat<PROTECTED>
15:20.49chouimat<PROTECTED>
15:20.49chouimat<PROTECTED>
15:20.49chouimat<PROTECTED>
15:20.52chouimatoops
15:21.07agaffneyheh
15:22.22chouimatagaffney: it's  a great bug btw :)
15:22.29agaffneyindeed
15:22.43agaffneyI despise `info`
15:22.47agaffneyman ftw
15:23.35chouimatagaffney: info is only useable in konqueror ...
15:24.36agaffneykonqueror does info? neat
15:24.52chouimatagaffney: just type info:/
15:25.15agaffneywell, I don't use kde
15:25.22agaffneybut I've played with konqueror before
15:25.26agaffneyjust as a file manager
15:26.15chouimatagaffney: if you have the right ioslave you can do anything ... I build one that enable me to browse a CANOpen network using konqueror
15:46.35*** join/#edev ico2_ (n=ico2@modem-2147.chameleon.dialup.pol.co.uk)
15:46.50g1powermacheh
15:47.20g1powermacecho "haha">/tmp/donot_reboot;
15:47.30agaffneyo_O
15:47.47g1powermacwhich was written by a guy named roy on 20060915
15:47.50agaffneyheh
15:48.08prpplaguelovely, leds working
15:48.30wmatprpplague: your boards get mailed out yet?
15:48.50prpplaguewmat: no not yet, i'm a week behind on code since i had a dental problem last week
15:48.57prpplaguewmat: boards are ready, code isn't
15:49.08prpplaguewmat: doing the final touches on 2.6.22 now
15:49.24wmatprpplague: ah, no worries, i was just wondering if it should have arrived here yet
15:50.23g1powermachmm
15:50.33g1powermacthey definitely kept the modules for the kernel down
15:50.47g1powermaconly 7 modules in /lib/modules
15:50.54prpplagueCosmicPenguin: still debating what to do about the push button
15:51.13CosmicPenguinwhats the debate?
15:53.05prpplagueCosmicPenguin: basically if i should keep that driver or look to intergrate it more into the input subsystem
15:53.25CosmicPenguinyou're going to need a driver for it regardless
15:53.35CosmicPenguinthe input subsystem doesn't grok how to actually talk to the hardware
15:54.15prpplagueCosmicPenguin: correct, but there are some examples of using buttons under the input/misc
15:54.19CosmicPenguinright
15:54.26CosmicPenguinbut you can do that anywhere
15:54.39prpplagueCosmicPenguin: the thing is i want it easy for people to have the buttons do stuff from userland
15:54.45CosmicPenguinlike for example?
15:55.44CosmicPenguinokay - well, we'll get to that
15:55.58CosmicPenguinfirst, the input subysstem integration
15:56.09CosmicPenguinthis is a slightly different example, because its a switch, but the concept is similar
15:56.14prpplagueCosmicPenguin: right, i can intergrate it into that easy
15:56.29prpplagueCosmicPenguin: the thing is, that the hammer board really doesn't have a console perse
15:56.47prpplagueCosmicPenguin: so having it work like a "keyboard" really doesn't make sense
15:56.51CosmicPenguinhttp://dev.laptop.org/git?p=olpc-2.6;a=blob;f=arch/i386/kernel/olpc-pm.c;h=67c78775ca9eb45aa1c57b95f414e2d8192bdf08;hb=HEAD#l538
15:56.58CosmicPenguinSee line 558
15:57.24prpplagueCosmicPenguin: yea, i can do a keyboard style interface np
15:57.34CosmicPenguinand see where we set it (or not) in the interrupt function on line 196
15:57.41CosmicPenguinbut that file can live anywher
15:57.42CosmicPenguine
15:58.23CosmicPenguinApparently, I'm not groking your problem
16:00.02prpplagueCosmicPenguin: hehe, np, i'll try again
16:00.50prpplagueCosmicPenguin: think of it more like a switch connected to a door (i.e. an alarm system)
16:01.19prpplagueCosmicPenguin: when something occurs with the switch(button), the userland wants an application to kick off
16:01.52prpplagueCosmicPenguin: i.e. similiar to what you'd expect for a multimedia keyboard when you press the "music" button to open up a music player
16:02.13prpplagueCosmicPenguin: but since you don't have a console perse you can really work like a keyboard
16:02.39CosmicPenguinokay
16:03.43CosmicPenguinYou should definately do that through the input subsystem
16:06.04prpplaguethats what i thought
16:06.22prpplaguebut not sure how to implement something similiar to what i already have inside the input system
16:06.47*** join/#edev Crofton (n=balister@hc65216f2.dhcp.vt.edu)
16:07.11CosmicPenguinwhat do you mean?
16:07.35CosmicPenguinyou make a little daemon that hangs out, listens on your input device, and does something when the bit toggles
16:07.57prpplagueCosmicPenguin: with the original driver, what i'm doing is when the isr kicks off, it schedules a task to do a helper app
16:08.13CosmicPenguinright
16:08.33CosmicPenguinbut my way it all happens in user space, so you can change things without changing the kernel
16:09.00prpplagueCosmicPenguin: right, what i'm saying is that i don't know how that occurs in userspace
16:09.11prpplagueCosmicPenguin: i'm not very familiar with the sysfs events
16:09.15CosmicPenguinIts not a sysfs event
16:09.20CosmicPenguinits an input device
16:09.43prpplagueCosmicPenguin: ok, maybe i'm confused then
16:09.56prpplagueCosmicPenguin: how do you know an event has occured from userspace then?
16:10.03CosmicPenguinselect()
16:10.58prpplagueCosmicPenguin: you mean from a c app?
16:11.32CosmicPenguinsure
16:11.34CosmicPenguinor whatever
16:11.47prpplaguehmm, that means something has to be running
16:11.47CosmicPenguinyou basically watch a device file like you do anything else
16:11.51CosmicPenguinyes
16:11.54CosmicPenguinyou need a daemon running
16:12.11prpplaguei was thinking more like what you'd expect to happen with the hotplug interface
16:12.25CosmicPenguinYeah, but that behavior has been deprecated
16:12.25prpplaguewhere when the event occurs it goes out and executes something
16:12.38prpplagueCosmicPenguin: in favor of what?
16:14.19CosmicPenguinuevents
16:16.14prpplaguehmm
16:20.41CosmicPenguinno
16:20.41CosmicPenguinits more correct when you really think of it
16:23.51prpplagueCosmicPenguin: hehe, just trying to understand how what i have translates to uevents
16:24.07Exileshmmm
16:24.25CosmicPenguinI would just use the daemon
16:24.27CosmicPenguinit has many advantages
16:24.54CosmicPenguinit is super simple to implement
16:25.09ExilesI've mainly seen a small daemon in embeded systems
16:25.16CosmicPenguinand it gives the user full control
16:26.16CosmicPenguinthe daemon only runs when you want it to - it can be written in any language you want, it can do anything you want
16:26.21CosmicPenguinyou want to fire another app - done
16:26.29CosmicPenguinyou want to handle it within the daemon?  done
16:26.46CosmicPenguinyou want to send it to somebody else with dbus? done
16:27.41prpplagueCosmicPenguin: hmm, just feels overly complicated for what i need
16:28.22CosmicPenguinhow is that more complicated then your plan?
16:29.31prpplagueCosmicPenguin: much more so, now i not only have to have a driver but i also have to have a daemon
16:29.43CosmicPenguinyou had to have a trigger application anyway
16:30.08CosmicPenguinBut think of it this way - what is your goal?
16:30.13prpplagueCosmicPenguin: but the trigger application is nothing more than a shell script that the user can configure
16:30.15CosmicPenguinWhat is best for your customers?
16:30.41prpplagueCosmicPenguin: the goal is to make it easy for low end developers to kick of a specific application when the button/switch is triggered
16:31.19CosmicPenguinand your only intent is to ever fire a single application - and to always have it fire?
16:31.58prpplagueCosmicPenguin: either a single app with an arg of either "up" or "down" or two apps, one for up and one for down
16:32.22CosmicPenguinahh - you're already changing the goal... :)
16:32.42CosmicPenguinAll I'm saying is - if it was me, I would go the input subsystem route
16:32.49CosmicPenguinand move the complexity to the userspace, where it belongs
16:32.54prpplagueCosmicPenguin: how am i changing the goal?
16:32.54CosmicPenguinymmv
16:33.31prpplagueCosmicPenguin: "move the complexity to the userspace" thats what i don't want
16:33.43Exilesokay, go the way of having a daemon do it, in the source file include extra lines commented out, on hoto add another button, or howto multiplex a few switches
16:33.43prpplagueCosmicPenguin: i want the complexity in the kernel
16:33.50Exilesproblem solved
16:34.14CosmicPenguinThats where we differ then - I'm a strong advocate of keeping policy in userspace
16:34.26CosmicPenguinanybody can hack on userspace - few can hack on the kernel
16:34.34CosmicPenguinBut again, thats me
16:34.39CosmicPenguinLinus and I are tight like that
16:36.51prpplagueCosmicPenguin: yea
16:39.04prpplagueCosmicPenguin: thats the thing, for us, making it easier in userspace adds sale-ability
16:40.15CosmicPenguindoes it though?
16:40.41CosmicPenguinit encourages design wins
16:40.53CosmicPenguinbut it makes it harder for the customer to go to production
16:42.04CosmicPenguinIn my experience, customers are more confortable changing userspace then kernel space
16:42.11prpplagueCosmicPenguin: for production, the customer isn't interested in anything other than the amount of time it takes to get their app running on the device
16:42.24prpplagueCosmicPenguin: correct, thats the whole reason for this
16:42.59CosmicPenguinbut by pushing policy into the kernel, you are forcing the customer into whatever you decided today
16:43.07CosmicPenguinoriginally you said you wanted one app
16:43.11CosmicPenguinthen you said, well, maybe two apps
16:43.15CosmicPenguinwhat if the customer wants three apps?
16:43.19CosmicPenguinwhat if they want a timer?
16:43.32prpplagueCosmicPenguin: no, what i meant was, either do it with one app and two args, or two apps with no args
16:43.37prpplagueCosmicPenguin: doesn't matter which
16:43.55prpplagueCosmicPenguin: the contraints are pretty limited
16:44.10CosmicPenguinBut thats a decision that you made
16:44.13prpplagueCosmicPenguin: just need to do x when button is down and do y when button is up
16:44.22CosmicPenguinis it?
16:44.38prpplagueCosmicPenguin: correct, based on the needs of the device
16:44.45CosmicPenguinWell - you've obviously made your decision
16:45.00prpplagueCosmicPenguin: no not yet, but i do like discussing it :)
16:45.07prpplagueCosmicPenguin: i like to get feedback on stuff
16:45.08Croftonthe needs of the many outweigh the needs of the few
16:45.32prpplagueCosmicPenguin: i'm thinking more in the lines of how stuff like the Stamp from parallax is supported
16:45.44prpplagueCosmicPenguin: which is what alot of hardware devrs are familiar with
16:49.43CosmicPenguinBut hardware developers aren't driving the bus to production
16:52.03prpplagueCosmicPenguin: i think the problem is we are looking at it from two different types of "production" evironments
16:52.29CosmicPenguinpossibly
16:52.31prpplagueCosmicPenguin: most of our customers will be handed a completed product, all they are concerned about is getting their 3rd party app running in the least amount of time
16:53.15prpplagueCosmicPenguin: situation 1) customer has to write a c app to do ioctls as well as read or write to a device
16:53.35prpplagueCosmicPenguin: situation 2) customer "cat"'s or "echo"'s to sysfs or procfs
16:53.45prpplagueCosmicPenguin: they are going to choose 2 everytime
16:54.05CosmicPenguinWe're not discussing configuration here
16:54.12CosmicPenguinwe're discussing getting the event
16:54.20CosmicPenguinthose are two very, very different concepts
16:55.23CosmicPenguinBut, I would submit that the compexity for an application to output something to sysfs or procfs is just as much as writing to a device file
16:56.10prpplagueCosmicPenguin: http://www.pastebin.ca/659855 less than 100 lines and pretty straight forward
16:56.34CosmicPenguinstraightforward for who?
16:56.36CosmicPenguinthe application writer?
16:56.45CosmicPenguinwho has to figure out how to make the script interact with his already running application?
16:57.00CosmicPenguinThe kernel writer?
16:57.41CosmicPenguinA input subystem driver would completely eliminate all of button_event_task_handle
16:57.42CosmicPenguinr
16:58.04CosmicPenguinand is even more straightforward because you don't need to grok how the usermode helper stuff works
16:58.29prpplagueCosmicPenguin: easier for the application writer
16:58.36CosmicPenguinwhy?
16:58.53CosmicPenguineasier then IPC?
16:59.01prpplagueCosmicPenguin: no in depth knowledge is need to kick of an application
16:59.04prpplagueCosmicPenguin: much
16:59.28CosmicPenguinIPC is easier then a simple select() loop?
16:59.44prpplagueCosmicPenguin: but requires more coding
16:59.46CosmicPenguinwhich you would probably end up implementing anyway
17:00.19CosmicPenguinbecause the smart way to communicate these events to a running application would be through an UNIX socket, and so you're right back to square one
17:00.41prpplagueCosmicPenguin: right, but their isn't a running application
17:00.46CosmicPenguinHow do you konw?
17:01.13CosmicPenguinYou see my point?  You're basing your decisions on a set of unknown parameters
17:01.17CosmicPenguinThats policy
17:01.29prpplagueCosmicPenguin: what parameters are unknown?
17:01.51CosmicPenguinYou said that the customer would want to get their third party app running as fast as possible
17:01.57CosmicPenguinwhat you don't know is how that third party application works
17:02.32prpplagueCosmicPenguin: that is correct to some extent
17:02.38CosmicPenguinIf its an already running daemon, for example
17:03.02CosmicPenguinthen the developers would need to work up some method of communicating the event to the daemon through dbus, or unix sockets or whatever
17:03.27prpplagueCosmicPenguin: but again, what i'm comparing to is the usage of small pic/avr devices that are not structured in that way
17:03.46prpplagueCosmicPenguin: i see you point though
17:03.49CosmicPenguinbut you're defining that based on your own experience
17:04.05CosmicPenguinIt has always been my policy that I don't understand my customers as well as I think I do
17:04.21CosmicPenguinSo I try to push these decisions to the customer so they can make the right choise
17:04.30CosmicPenguinyes - that kills some customers who don't want to make the decision
17:04.59prpplagueCosmicPenguin: yea, the problem we have with that is that they customers have too many choices and are unable to get an item working, they basically just give up
17:05.07CosmicPenguinBut my personal opnion is - those customers usually don't make it to production anyway
17:05.31prpplagueCosmicPenguin: hehe, thats about 90% of our customers, only about 10% of them are high end users
17:05.40CosmicPenguinEvery single one of our tier 1 customers for Geode end up rolling their own kernel and rootfs
17:05.57CosmicPenguinso they demand the ability to make a choice
17:06.17prpplagueCosmicPenguin: yea, i can see that
17:06.43prpplagueCosmicPenguin: 90% of our customers when asked about a "kernel" would refer to KFC
17:07.04CosmicPenguinWell, then you need to adjust your expectations accordingly
17:07.54CosmicPenguinAll of my actions are predicated on enabling as many of my customers as possible
17:08.06CosmicPenguinI assume yours are as well
17:08.28prpplagueCosmicPenguin: yea
17:09.02CosmicPenguinIn my case, that means generic drivers that are in the kernel mainline and stay on the edge of development
17:09.03prpplagueCosmicPenguin: most are familiar with making changes to shell scripts, most are familiar with compiling "their" applications for the handheld
17:09.19prpplagueCosmicPenguin: but most are totally unfamiliar with the workings of linux
17:09.28CosmicPenguinThats the market you are in
17:09.34prpplagueyea
17:10.30prpplagueCosmicPenguin: which is also why i often have conflicts with things like OE folks, because the market is so much different that what many people deal with
17:12.23prpplagueCosmicPenguin: thanks for the conversation it though, i like to discuss the concepts
17:13.31CosmicPenguinNo problem
17:21.13prpplagueCosmicPenguin: can you "echo" to a sysfs uevent entry?
17:21.43CosmicPenguinI think its only outgoing
17:22.03*** join/#edev T0mW (n=Tom@24.238.86.158.res-cmts.sth.ptd.net)
17:23.20prpplagueCosmicPenguin: hmm the entry shows as writable
17:26.18prpplagueCosmicPenguin: looks like you can, just don't what the syntax is
17:28.10*** join/#edev lyakh (n=lyakh@dynamic-unidsl-85-197-26-24.westend.de)
17:34.30prpplagueCosmicPenguin: probably need to just add an attr
17:40.55prpplagueT0mW: you use sysfs?
17:41.35T0mWprpplague: I don't remember
17:41.41T0mWheh
17:41.51T0mWmy head is into pascal today
17:43.34T0mWprpplague: yeah, I use sysfs on the zipit
17:43.58T0mWprpplague: devfs is pita
17:47.11*** join/#edev leoncamel (n=leoncame@p1239-ipbfp1105fukuokachu.fukuoka.ocn.ne.jp)
17:47.29prpplagueT0mW: you ever try to manually inject uevents into a sysfs device?
17:47.47T0mWprpplague: like the uinput?
17:48.54prpplagueT0mW: yea
17:49.49T0mWprpplague: http://www.openhardware.net/Misc_Stuff/IR_Keyboard/
17:50.05T0mWprpplague: that inserts key events into uinput
17:50.45T0mWprpplague: more precisely: http://www.openhardware.net/Misc_Stuff/IR_Keyboard/LircTechnicalInfo/
17:51.16prpplagueT0mW: yea i was thinking more like "echo 123 > uevent"
17:51.33T0mWprpplague: no, you need to give it "keystroke" codes
17:52.02T0mWprpplague: IIRC, the keys are 16bits
17:53.23T0mWprpplague: what you could do, is take my code and use that to translate your ascii string into appropriate codes, e.g.: echo "123" | toms_stuff
17:54.42T0mWprpplague: although it is written as a daemon, you could just make lirc_keybd.c an app
17:55.31prpplagueT0mW: yea
17:56.49T0mWprpplague: I forget how it works really, been a while since I did that.
17:57.02chouimatyay thunder!!
17:57.09T0mWyipe!
17:57.30T0mWchouimat: lightning too?
17:57.31prpplagueT0mW: yea i'm familiar with the daemon portion, was jsut curious about the echo part
17:57.36chouimatT0mW: yes
18:01.20T0mWprpplague: /dev/uinput is a file stream, you write short ints to it representing the keycodes
18:01.56*** join/#edev chouimat (n=dieu@kde/developer/chouinard)
18:02.17T0mWprpplague: for example, to do a CAPITAL-A, you would give it: SHIFTKEY_DOWN_CODE, KEY_A_CODE and finally, SHIFTKEY_UP_CODE
18:03.39T0mWprpplague: errr, SHIFTKEY_DOWN_CODE, KEY_A_CODE_DOWN, KEY_A_CODE_UP, and finally, SHIFTKEY_UP_CODE
18:03.44prpplagueT0mW: yea, i have all that working, this was off on another tangent
18:04.14T0mWprpplague: the advantage to /dev/uinput is the key is entered into the kernel level
18:04.58T0mWprpplague: then ALT-SYSREQ-K and the like would also work
18:16.19prpplaguelets see if that did the trick
18:16.37prpplagueCosmicPenguin: i think its just a learning curve getting used to using sysfs instead of procfs
18:16.47prpplagueCosmicPenguin: i still like how easy procfs is though
18:21.12*** join/#edev chouimat|busy (n=dieu@kde/developer/chouinard)
18:21.30prpplaguehmm i see my new entry but nothing seems to be working :(
18:27.03*** join/#edev ico2 (n=ico2@modem-3513.cougar.dialup.pol.co.uk)
18:32.18prpplaguehmm this just isn't working
18:34.26g1powermacprpplague, I also put up a simple umount script for jffs2
18:35.03g1powermacit basically umounts the image and gets rid of the extra modules loaded so you could mount other jffs2 images
18:35.36g1powermaconly issue is that you can mount only one image at a time
18:35.55g1powermacthere's prolly a way to do multiple ones, but really don't need it
18:40.41prpplagueCosmicPenguin: all the buzzer/beeper stuff is in the input section, but the input section doesn't have a class_device entry for you to be able to initiate a beep
18:45.16*** join/#edev Crofton|work (n=balister@ossie.wireless.vt.edu)
18:45.17*** join/#edev Crofton|work_ (n=balister@ossie.wireless.vt.edu)
18:47.05*** join/#edev g1powermac_DP (n=g1powerm@adsl-210-184-158.sdf.bellsouth.net)
18:52.19prpplagueCosmicPenguin: any reason i shouldnt be able to add a sysfs entry under the driver section?
19:00.12g1powermac_DPI actually found a ssid and password for an AP in the firmware for SMCs SIP phone
19:00.35g1powermac_DPI wonder if its what they were using for testing, or just a random thing they put in there
19:01.08prpplagueg1powermac_DP: probably, that happens often
19:01.25g1powermac_DPyea, its just funny to see :-)
19:01.38g1powermac_DPI still like what I found before
19:01.51g1powermac_DPecho of "haha" to some file
19:03.27g1powermac_DPwell, I just found they're using Qtopia for the GUI
19:04.15g1powermac_DPlook at all those lovely apps they made for the taking. . . . ;-)
19:04.35CosmicPenguinprpplague: no reason at all
19:07.20g1powermac_DPinteresting stuff
19:10.05*** join/#edev Crofton|work (n=balister@ossie.wireless.vt.edu)
19:10.38prpplagueCosmicPenguin: hmm, i keep getting an incompatible pointer error :(
19:12.05CosmicPenguinpastebinit
19:13.53prpplagueCosmicPenguin: nm, got it , i needed to do a class_device addition, not a driver
19:14.28prpplagueCosmicPenguin: i think i got it now, sysfs is sooo messy
19:24.04g1powermachehe, they left some of the generic default Qtopia images in the firmware
19:26.22g1powermachah, and a big tux logo :-)
19:32.59*** join/#edev LyosNorezel2 (n=lyosnore@unaffiliated/lyosnorezel)
19:39.11MothaWorkprpplague: you're sure you had Advanced Circuits doing a <5mil annular ring on a via?
19:39.14MothaWorkthey insist they can't do it
19:39.36MothaWorkthey'll run it as is if you like, but if it fails e test, they wont' make them again
19:40.40prpplagueMothaWork: thats what the layout guy told me
19:40.45prpplagueMothaWork: i'll ask again
19:42.17prpplagueMothaWork: 10 mil drill, 16.5 pad, 5 trace , 5space
19:42.33prpplagueMothaWork: said he's been getting them from 4pcb.com for quick turns
19:43.29MothaWorkyeah, that's advanced circuits
19:43.51prpplagueMothaWork: that should be a 6.5 annular ring
19:43.57MothaWorkno, that's 3.25
19:44.07MothaWorkthe pad diameter is 6.5 over, but the annular ring is radius
19:45.13prpplagueMothaWork: ahh your right
19:47.18prpplagueMothaWork: checked again, definetly was 4pcb.com, it was their standard service, not the prototype service
19:47.36MothaWorkthat doesn't change the minimum annular ring, though you can then go down to 8 mil holes
19:47.56prpplagueMonMotha: interesting
19:48.33prpplagueMothaWork: i can only pass along what the layout guy is telling me, there is another local company we use as well, called W-Tech
19:56.12MothaWorkmy math says that ~15.5mil pad is the minimum to do this with 5/5 on a 0.8mm via
19:56.17MothaWorker 0.8mm bga
20:06.30prpplaguewmat: you still good for working on the Hammer?
20:08.16wmatprpplague: I'll certainly poke at it a bit
20:16.19prpplagueT0mW: ping
20:20.07jbevrenlol!
20:20.15jbevrenFoster is playing RIAA tactics with the RIAA
20:20.20jbevrenthey refused to pay judgement and to reply
20:20.31jbevrenso now foster is demanding 210k instead of 68k
20:24.38prpplaguejbevren: you still in for the hammer board?
20:24.49prpplaguejbevren: just double checking everyone before shipping
20:26.14jbevrenI am :)
20:26.18jbevrenhow is it being shipped?
20:26.50jbevrenI thought of another application for the board heh
20:26.57prpplaguejbevren: probably UPS
20:27.00jbevrenkk
20:27.11jbevrenups will leave it on my porch if no delivery verification is required
20:27.19jbevrenI'm not normally there during the week
20:27.24jbevrensince I work so far from home atm
20:27.39prpplaguejbevren: hmm
20:27.45jbevrenI just wanted to make sure it doesnt get sent back :)
20:29.29jbevrenI'd be disappointed to find out I missed it hehe
20:29.43jbevrenany idea when they head out?
20:30.02prpplaguejbevren: probably tuesday
20:30.06jbevrenahh cool
20:30.30prpplaguelater folks
20:35.16*** join/#edev Exiled (n=Exiled@68-113-40-5.dhcp.klmt.or.charter.com)
22:16.34*** join/#edev pierrelux (n=pierre-l@144-125.sh.cgocable.ca)
22:16.41*** join/#edev flyback (n=n@c-24-3-31-106.hsd1.pa.comcast.net)
22:16.43*** join/#edev g1powermac_ (n=g1powerm@adsl-80-80-250.sdf.bellsouth.net)
22:16.44flybackwtf
22:16.50flybackbanned then unbanned
22:16.54flybackI wish someone would make up their mind
22:18.17*** join/#edev sjdsphreaker (n=sjdsphre@72.22.131.185)
22:21.37agaffneyprobably only unbanned because chanserv is a bit forgetful
22:21.50*** join/#edev g1powermac__ (n=g1powerm@adsl-210-143-149.sdf.bellsouth.net)
22:22.27flybackwell I am not complaining :P
22:23.04*** join/#edev Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net)
22:23.15*** join/#edev g1powermac__ (n=g1powerm@adsl-210-143-149.sdf.bellsouth.net)
22:41.02*** join/#edev joe_bleau (n=joe_blea@adsl-68-89-94-3.dsl.fyvlar.swbell.net)
22:46.51*** join/#edev g1powermac_ (n=g1powerm@adsl-18-80-74.sdf.bellsouth.net)
22:56.39GPSFang1powermac: where did you get the code for your phone?
22:57.53g1powermacGPSFan, no source, I was just going through the binary firmware for the SMC sip phone (almost identical to mine, except it costs twice the amount and has unencrypted firmware images)
22:58.20GPSFang1powermac: ah, that's what I meant, the firmware image.
22:59.21g1powermacoh, hmm, lemme get a link
22:59.49GPSFanyou probably put it on the wiki, but I'm feeling lazy atm ;>)
22:59.54g1powermachttp://www.smc.com/index.cfm?event=downloads.doSearchCriteria&localeCode=EN_USA&productCategory=14&modelNumber=1564&partNumber=3564&downloadType=8&knowsPartNumber=false
23:00.06g1powermacactually, I didn't put a link to the firmware of the SIP phone
23:01.58g1powermacGPSFan, its not so simple to extract the actual jffs2 images however
23:02.03GPSFanthere is no firmware at that url, lots of other stuff, manuals, user guides, no firmware that I can tell
23:02.14g1powermacGPSFan, you will need windows to run the firmware flasher
23:02.27g1powermacdang it, wrong link
23:02.43GPSFan;>)
23:04.20g1powermachttp://www.smc.com/index.cfm?event=downloads.doSearchCriteria&localeCode=EN_USA&knowsPartNumber=false&productCategory=14&modelNumber=1620&partNumber=4050&downloadType=0&os=0
23:04.23g1powermacbetter link :-)
23:07.57GPSFanindeed...
23:10.51g1powermachmm, has anyone here ever tried to compile qtopia?
23:10.57g1powermacits giving me grief
23:11.39GPSFang1powermac: how'd you extract it from the .exe file? does running it under windoze creat an intermediate .jffs2 file?
23:12.06g1powermaccheck the wiki page
23:12.19g1powermacit describes the process, but yea, you have to run it under windows
23:12.57g1powermaconce it runs to the point right before it flashes the phone, it spits out the jffs2 files to a tmp dir
23:13.01GPSFanok thanks, just thought I'd browse through it to see if there is anything of interest.
23:13.20GPSFantry it under wine?
23:13.39g1powermacdidn't try that
23:14.40GPSFandoesn
23:14.59GPSFandoesn't work, missing a dll setupapi or some such
23:15.00g1powermacthere might be ways to directly extract files from the .exe
23:15.18g1powermacI was trying that, but after a few tries, I went with just running it under windows
23:16.33*** join/#edev g1powermac_ (n=g1powerm@adsl-78-220-113.sdf.bellsouth.net)
23:17.08GPSFanI'll try that.
23:23.49GPSFang1powermac: there is homefs and rootfs, uboot and uImage. I'd bet the kernel is uImage, uboot is of course the bootloader rootfs & homefs are /  and /home.
23:24.06g1powermacyea
23:24.13g1powermacthough there is a linuxrc in rootfs
23:24.35g1powermachomefs is a bit weird, there isn't much in it for the 10mb it consumes
23:24.48agaffney~quote scox
23:24.49g1powermacI haven't played with uImage or uboot yet
23:25.25GPSFanhmm, are these jffs2.raw  files just .jffs2?
23:25.33g1powermacyup
23:25.40g1powermacnot sure why there's a .raw at the end
23:25.43GPSFangotcha.
23:25.53g1powermacmaybe its actual images of the partitions
23:26.01g1powermacso it includes freespace
23:26.11g1powermaccould explain why homefs is so large
23:26.27g1powermacbut they mount like normal jffs2 images
23:34.33g1powermacargh, still can't get qtopia to compile
23:37.43*** join/#edev redblue (i=star@ppp091.110-253-207.mtl.mt.videotron.ca)
23:38.06GPSFaninteresting, very very little in /home one onders why. btw you need to be root to modprobe the mtd drivers.
23:38.21GPSFans/onders/wonders
23:38.28g1powermacyea
23:39.00g1powermactechnically you need to be root to mount too
23:39.14g1powermacat least on my machine
23:39.51GPSFanand who is alex kao ?
23:40.11g1powermacwho?
23:40.19g1powermacwhere did you find that name?
23:40.57GPSFanin the libs dir, there are several symlinks that point to /home/alexkao/sip/......
23:41.10g1powermacnow that is interesting
23:41.32g1powermacguess he was one of the devs
23:43.08g1powermacits funny to see leftovers like that :-)
23:43.51GPSFanyikes, 20M in /opt
23:44.05g1powermacyea, thats qtopia, iirc
23:45.11g1powermachah, yea, look at those symlinks, thats funny
23:45.50GPSFanI'm always curious to see what folks have left in their dists. that they don't really mean to.
23:46.01g1powermachmm, think I might have found the cause to the qtopia compile problem, kinda helps to have the xorg-dev packages installed :-)
23:46.16g1powermacnow its actually taking a lot longer to compile
23:47.25GPSFanthat's a good sign.
23:49.40g1powermacI wish I could get the jffs2 images for the actual phone I'm going to hack
23:50.47g1powermacbut hopefully I should be able to browse around the filesystem of the phone once I hook a serial cable to it
23:50.52GPSFanI thought there was a serial hack that dropped you to a root shell
23:51.01g1powermacyup, there is :-0
23:51.28g1powermacpersonally, I really want to see if JTAG is taken out to the COMM headers
23:51.44g1powermacthen I can grab the whole flash image
23:51.46GPSFanwell, as root you should be able to ftp. or ssh or something to a server and just dump everything.
23:52.10g1powermacpossibly, depends on whats actually on the phone
23:52.54GPSFanthis image has ftpput
23:53.10g1powermacthats a good sign at least
23:53.19g1powermacthough I'm not sure what they kept on the other image
23:53.31g1powermacsince they have half the flash to work with for my phone
23:53.33GPSFanneed to see what the skype version has
23:53.48g1powermacthe skype phone only has 16mb of flash
23:53.58g1powermacwhile the one you're looking at has 32mb
23:54.14GPSFanyeah, but they probably left busybox as it is.
23:54.21g1powermacprobably
23:54.31GPSFanwhen does you phone arive?
23:54.36g1powermacnot sure
23:55.35g1powermacbah, got about 10 days
23:55.42GPSFan;>(
23:55.54g1powermacstupid amazon and their ultra slow shipping for their free shipping option
23:56.00g1powermacbut hey, its free :-0
23:56.12GPSFanI guess they are shipping it from the Amazon...
23:56.23g1powermacheh
23:56.34g1powermacactually, it hasn't even shipped yet
23:56.49g1powermacI guess cause the shipping is free they don't rush things
23:57.34g1powermacheh, yup, something is definitely better now, its been compiling now for about 20 mins
23:58.19g1powermaceven on this intel C2D iMac
23:58.58g1powermacwould compile for days on end
23:59.10ds2heh... X11 used to compile for days
23:59.52g1powermacds2, I have machines where the 2.4 linux kernel would take 4 hours to compile, and I stopped using them not so long ago :-)

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.