irclog2html for blob on 2001.10.12

00:26:43sammymorning everyone
01:30:40Russmorning
04:54:24sammyhello Russ , amything new ?
07:30:06RussSammy: not really anything new
08:09:09erikmmorning, *
08:09:22Sammyevering erikm
08:09:42WeirdArmsls
08:13:46Russtotal 0
08:14:54erikmRuss: could have been "lectori salutem" as well :)
08:20:15SammyI am start porting blob on sa-1110 develop board ...
08:21:22Sammymy friend borraw that board for me
08:22:27erikmSammy: cool
08:22:36erikmSammy: should be quite easy to port nowadays
08:23:26SammyI try
08:25:10Sammythis board running Angelboot ...and download with ethernet ...
08:25:30Sammyerikm : can blob download use ethernet ?
08:27:07Sammywhen ?
08:33:42erikmSammy: not yet
08:34:05erikmSammy: if BZFlag gets a contract to port it, it can be done quite fast
08:34:23erikmSammy: (two months or so)
08:34:43erikmSammy: otherwise it'll take at least half a year, I expect
08:37:56Sammyall I can do is watting , maybe someday I can help a little bit ..
08:39:57Sammyerikm : can you add run image from ram into blob , is that necessary ?
08:41:19SammyI think this maybe not let blob too big , right ?
08:41:53erikmSammy: we already do that in the CVS version
08:42:18erikmSammy: blob got split up in blob the bootloader and diag the diagnostics tool
08:42:19Sammyreally ?
08:42:36erikmSammy: diag is loaded with "load kernel" and started with "boot"
08:45:30Sammyconnect to the CVS ...
10:59:11Sammyerikm: is I want to download blob from CVS , in this time what command should I add after /cvsroot ?
11:15:46erikmSammy: see the blob project pages on sourceforge
13:19:40prpplaguemorning all
13:20:31prpplagueWeirdArms: things settle down on the linux-arm-kernel list?
13:20:46WeirdArms?
13:21:03WeirdArmsI think Nico see's my point a bit more clearly
13:21:18WeirdArmsalthough I was a bit snappy
13:22:55prpplagueWeirdArms: nothing like (virtual) room full of opionated programmers....
13:23:37WeirdArmsIt wasn't like that
13:24:01WeirdArmsbasically I haven't explained the work clearly to everyone and I'm asking for a bit of faith that it will work
13:24:17WeirdArmsthats a bit much to ask people who don't really know how good you are
13:28:29prpplagueWeirdArms: i understand that all to well
13:29:31WeirdArmsI've just found it frustraiting that I never get an answer to my questions because not many people know the specifics of the ARM memory implementation
13:30:00WeirdArmsSo it's taking me a long time to implement
13:33:18WeirdArmsBut it's progressing
13:37:01prpplagueWeirdArms: well i'm just a newbie in the ARM world, i'm experimenting on getting linux running on the ELF board from InHand, it amazing what you can learn when don't have any instructions
13:42:59WeirdArmsWell I'm pretty new to Linux internals
14:21:51sammyhi :-)
14:26:27WeirdArmsHi
14:37:52prpplaguehowdy sammy
14:38:10prpplaguesammy: have you been LARTed today?
14:38:46prpplaguei need bumber sticker that says that...or maybe "Got Blob?"
14:39:05prpplagueor maybe "have blob, will boot"
15:13:46WeirdArmswb erikm
15:13:49WeirdArmshow was your meeting?
15:14:05erikmunproductive, as usual with a meeting > four people
15:14:14WeirdArmshehehe
15:57:54seletzhi folks!
16:01:40prpplaguehowdy
16:02:22WeirdArmswell time for me to get some sleep
16:02:44prpplaguelater
16:02:45seletzweirdarms: goodnight!
16:09:36erikmthinks WeirdArms does weird things in his WeirdSleep
16:09:44prpplaguelol
16:10:02prpplaguethis dlink 802.11b access point has 16m flash
16:11:12prpplaguethe specs say its sa-1100 core but i don't see one
16:16:38prpplaguehmm, its got solder points labeled jtag
16:16:50erikmw00t!
16:17:19erikmget the device ID and see if it's an sa1100 :)
16:17:51prpplagueerikm: haven't done anything with jtag yet...., i'll have to some reading
16:18:58erikmprpplague: short jtag introduction: it's designed by a comittee, therefore it: a) overly complex and b) broken by design
16:19:22prpplaguehmm, there doesn't appear to be an ic even close to the size of sa-1100
16:19:28prpplagueerikm: ya thats what i hear
16:19:46erikmprpplague: it won't take long to discover that the JTAG state machine actually has twice the amount of states it really needs
16:20:44erikmprpplague: an sa1110 is quite small
16:22:08prpplagueas small as a flash ic ?
16:22:53erikmyes, it's BGA
16:23:45prpplaguehmm, guess i'll try to hack this thing, http://www.dlink.com/products/wireless/dwl1000ap/
16:29:32prpplaguehttp://www.abcsinc.com/small-linux/dsc00043.jpg
16:29:38prpplaguetheres a pics of the board
16:30:01prpplaguethe two dark chips i've not been able to identify
16:30:50erikmwgets
16:31:28erikmthe dark one at the bottom is not a SA1100
16:31:37seletzeating kebap
16:33:06erikmthe one in the middle could be an sa1110, but the package is a bit unusual
16:35:35prpplagueit has www.nwn.com on it
16:36:29prpplaguelooks like they've sold out to intersil
16:37:13prpplagueARM940 core controller
16:38:40sammyhoo ,everyone ^_^
16:47:31prpplagueerikm: found a pdf of the cpu - http://www.abcsinc.com/small-linux/FN4932.pdf
16:48:45erikmwgets and looks
16:50:50erikmprpplague: btw, the black chip at the bottom could be an UCB 1200
16:51:18prpplagueerikm: i've got the specs on it too, let me check
16:52:36erikmthe data sheets talks about an SCP port, and the only sensible device I would know to connect to that is the philips ucb 1[123]000
16:52:54prpplagueerikm: pdf says "direct sequence spread spectrum baseband processor"
16:53:46prpplagueprism 2.4 chipset
16:54:54prpplaguehows the support for arm9 under linux?
16:57:17erikmquite good, afaik
16:57:24erikmbut this particular chip won't work
16:57:32erikmit doesn't have a real MMU
16:58:01prpplagueya i was wondering about
16:58:02prpplaguethat
16:58:24prpplaguewhat about uclinux?
16:58:44prpplagueisn't that geared to mmu-less cpus?
16:58:55erikmyes, it's for mmu-less stuff.
16:59:05erikmI don't like mmu-less CPUs
16:59:36prpplaguehmm, guess i'll pass on this one, maybe better luck next time
17:01:28sammyif I want download blob in CVS with diag , which modulename should I use ?
17:02:00erikmit's in blob
17:03:57sammyok, it's download finish ...
17:04:27erikmsammy: right, diag lives in blob/src/diag/
17:06:15sammyerikm : you saw that blob can do load kernel image from ram , but I don't see any code doing about this ?
17:06:49erikmsammy: diag is relocated at the same position as the kernel, that's why it works for diag
17:08:33sammyohh.. one question about the word "diag" , maybe call now the english teach time , what is that diag reall mean ?
17:08:58Russ|werkerikm: nexus is a bit weird
17:10:06erikmRuss|werk: hmm? I didn't follow his complete conversation, just picked up the last question
17:11:40prpplaguelooks like intersil is using a large selection of arm9 based products
17:12:52erikmand why not, arm makes sense :)
17:14:40prpplaguebummer, looks like they all don't use mmu
17:14:58ed__what are they using them for ?
17:15:21prpplaguewireless lan and cable modems
17:15:52ed__ahh
17:16:22prpplaguei really had hopes for this board, its a shame that it has no real mmu
17:16:56erikmprpplague: well, if it only has to do one thing uc-linux makes sense
17:18:04ed__i want to see arm microcontrollers with linux built in
17:18:18Russ|werk?
17:18:28Russ|werkah, microcontrollers
17:18:32Russ|werkthat would be a lart
17:18:38erikmor nesa
17:18:41Russ|werkmmuless processors run linux
17:18:43ed__nah something smaller
17:18:52Russ|werkjust no fork(), only vfork()
17:19:01ed__that costs < 30 bux
17:19:12ed__no mmu makes it a pain to port stuff though
17:19:22erikmRuss|werk: that means no multitasking
17:24:55Russ|werkno it doesn't
17:25:06Russ|werkit just means unsafe multitasking
17:25:28Russ|werkand no shared memory afaik
17:28:33erikmBZFlag: good news, I can maybe get bootp/tftp code
17:29:06Russ|werkbootp/tftp code would be simple from scratch
17:29:44prpplaguewe'll i'll put some info on my site about, maybe someone can use it
17:29:58erikmRuss|werk: rmk has that code lying around. it's tested on arm and it works
17:30:47Russ|werkits the network stack and driver that is a bitch
17:31:22Russ|werkI read through the tftp rfc
17:31:57erikmRuss|werk: his complete bootp/tftp loader *including* ethernet stack and driver is 5664 bytes
17:32:24Russ|werkshweet
17:41:07erikmRuss|werk: rmk needs to clean up the licensing issues. there's someone else that wants to use it, he'll probably dual license it
17:45:38erikmRuss|werk: 834 lines of C code including ethernet driver and AIF conversion utility
17:46:15erikmRuss|werk: (according to rmk)
17:57:41prpplague_lunchgeneral speaking, you'd think that a device that provides firewalling/nat would need a mmu right?
17:58:19erikmprpplague: not really
18:06:09BZFlagthat includes the arm pcmcia init code?
18:06:42erikmBZFlag: no, it's for an ebsa110
18:07:29BZFlagah. the tarball I uploaded has arm pcmcia init code in it as well.
18:07:45Russ|werkpcmcia init code?
18:07:51Russ|werknot exactly complex
18:07:52BZFlagand netboot is also GPLed, do doing a clean merge of the two should work.
18:08:10Russ|werkmaybe if you want to init an actual card
18:08:15Russ|werkbut just the slot-s
18:08:17BZFlagRuss|werk: nope, but needed for devices without onboard eth.
18:08:21erikmBZFlag: rmk's code only needs ethernet drivers
18:08:53BZFlagI'm happy with simple pcmcia and ne2k eth card support.
18:08:54Russ|werkI don't think you'd need to handle insertion events
18:09:08erikmBZFlag: there currently only is an am79c961a pnp ether driver
18:09:08BZFlagnope. just detect and configure existing cards.
18:09:33BZFlagne2k cf cards should be common.
18:10:24BZFlagCan we add Russ' splash code as requiring --enable-splash or some such?
18:10:44erikmBZFlag: I've been thinking about it, and I think we can
18:11:08erikmBZFlag: just make it an __init() call, and it will show up immediately
18:11:16BZFlagcool. I need to get jflash compiling against the new kernel/blob too.
18:11:44BZFlagerikm: sounds good. it should add a "splash" command too.
18:11:54erikmBZFlag: btw, I have a very small ramdisk/jffs2 image on the LART site. should work on TuxScreen as well
18:11:57Russ|werkerikm: would need to handle it just right
18:12:05erikmBZFlag: what would a "splash" command do?
18:12:17Russ|werkerikm: the best place to put the actual image you be in a paramater block
18:12:22Russ|werker, would be
18:12:24BZFlagerikm: have you tried the tux buildroot andersee built? it's nice.
18:13:02erikmBZFlag: no, haven't seen it yet. I've been trying to chase down a 220-->110 converter somewhere so I actually power the tuxscreen
18:13:34BZFlagerikm: I assume at some point we will also add conditional lcd testing code and load/flash code for the splash screen. the "splash" command just displays the current splash image.
18:14:06erikmBZFlag: seletz has been working on LCD code in diag. if we move it to the library blob can use it as well
18:14:10BZFlagheh, your comment goes both ways, the tuc uclibc buildroot should work on the lart too. ;-)
18:14:22erikmBZFlag: ok, mine is glibc based
18:15:12erikmBZFlag: but still, I don't see a point in having a splash command when the only thing we show on the LCD is the splash screen
18:15:39BZFlagour current tuxscreen-image.jffs2 is about 640k and has busybox/tinylogin/pcmcia-cs/lrzsz/udhcpc
18:16:14erikmBZFlag: ok, my jffs2 image is about 1MB and has busybox/tinylogin/lrzsz
18:16:41BZFlagerikm: splash should be disabled by default, and require --enable to turn it on, but that same section of code will have the lcd test features in the future.
18:17:10BZFlagas well as displaying a boot menu of sorts on devices that have some boot selector buttons etc.
18:17:12erikmBZFlag: well, the test features can go into diag, that's what it's for
18:17:21Russ|werkmy splasd screen says BLOB! in big white letters
18:17:43BZFlagIt would be nice to loose 384k and gain pcmcia, no? ;-)
18:18:36Russ|werk384k?
18:18:39erikmBZFlag: well, I'm going to test PCMCIA on LART next week, so I'll keep it in mind
18:18:39BZFlagerikm: true, but I think there will be a need to test the lcd in diag, and showing splash is one possible "test" ;-)
18:19:10BZFlagRuss|werk: erikm said his image was 1M but did not have pcmcia-cs in it.
18:19:51erikmBZFlag: I know tuxscreen needs the LCD for user feedback
18:20:13BZFlagerikm: even without a shannon to test on, I'd suggest you try the shannon buildroot-tux out. I expect it will work fine on lart, just by changing kernel and blob config.
18:21:24erikmBZFlag: I'll try next week
18:22:08BZFlagerikm: tux wants lcd for user menu, true. but I do not want all boxes to have to have it. It's not essential code. I do want to avoid multiple people doing the same hack, so including it but disabling it by default seems the best way to reuse efforts.
18:22:19erikmBZFlag: oh, ok
18:22:32erikmBZFlag: so an --enable-splash would make sense
18:22:39BZFlagagreed.
18:22:53Russ|werkthe splash code will end up being very small
18:22:57erikmBZFlag: otoh, what about this:
18:23:02BZFlagoer perhaps --enable-lcd ? if there will be more lcd test code too.
18:23:12erikmthat's what I wanted to propose
18:23:19BZFlagheh, great minds. ;-)
18:23:40erikmbut really, test code should go into diag, I really don't want to make blob as bloated as bootldr
18:23:59erikm(FYI: a bootldr image for assabet is about 100K)
18:24:00BZFlaghmm... explain the plan for diag?
18:24:12BZFlagsome --enable-diag flag?
18:24:26erikmBZFlag: you should subscribe to the blob-cvs-commit list
18:24:31BZFlagheh
18:24:58erikmseletz wanted some test code for his platform and wanted to put it in blob
18:25:13erikmbut I want blob to be small and simple
18:25:36erikmso what I made (in CVS) is a blob and a diag image, which share a library of common code
18:26:04erikmif you want to run the diagnostics, you just type "download kernel", upload diag, and "boot" into it
18:26:08BZFlagon phone
18:38:48BZFlagthat makes good sense for a lot of diag code.
18:39:45BZFlagI do think that on platforms that only have large flash blocks (like iPAQ) it makes sense to link all that stuff in.
18:40:01BZFlagIf you have 256k of space, you might as well use it.
18:40:12erikmBZFlag: the point is that you almost *never* need it
18:40:25BZFlagnow perhaps there would be a diag image in that same block and a load command to get to it.
18:40:32erikmBZFlag: so if you really need it, just download it over the serial link
18:42:17BZFlagso perhaps diag could live in flash on the ipaq right after blob and be loaded from there if you ever need it?
18:42:32erikmBZFlag: on ipaq it would make sense
18:42:53erikmBZFlag: but on machines with boot block flash I pretty much doubt it
18:43:01BZFlagagreed!
18:45:53erikmjust got rmk's code
18:47:14erikmthis is sweet
18:47:23erikm300 lines for arp+bootp+tftp
18:48:04erikm109 lines for the IP stack
18:48:20erikm220 lines for the ethernet driver
18:48:57BZFlagnice.
18:49:08BZFlagwhat is this code from?
18:49:19erikmrmk wrote it for his ebsa110
18:49:59erikmbenh already ported it to a couple of other architectures
18:50:01BZFlagI meant, was it part of a loader, kernel, stand alone, etc.?
18:50:28erikmit's part of a loader
18:50:43BZFlagcool.
18:51:30erikmthe ebsa110 firmware is pretty brain damaged, could only do 38k4 over the serial port
18:54:45Russ|werkmmm....rmk code
18:55:21erikmRuss|werk: what's wrong with it?
19:01:51sammyneed to get some sleep ...
19:01:58sammynight all
19:02:18prpplaguewhat do you think guys - http://www.abcsinc.com/small-linux/dsc00044.jpg
19:03:07erikmcute, what's it?
19:03:27prpplaguemy thinclient project
19:04:03prpplaguethis is the new version with dual pcmcia or second ethernet
19:04:11prpplaguehttp://www.linuxdevices.com/products/PD4854992533.html
19:04:50prpplaguebasically it is a small x-windows terminal
19:18:21prpplagueno exactly a ground breaking design, but not bad for my first commercial, mass produced product
19:18:48prpplagues/no/not
19:20:38erikmlarge reset button, you expect it to run windows? :)
19:21:19prpplaguethe reset has been removed! that was added by the tawainese sub-contractor, we weren't amused
19:36:36Russ|werkI was saying rmk good
19:39:35erikmRuss|werk: OIC. mmm, as in "mmm, I like it" :)
19:39:50Russ|werkyup
19:50:59erikmgoes zzz
20:21:03prpplagueerikm: goodnight

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with infobot logs, split per channel and by date, etc.