irclog2html for blob on 2002.05.28

01:03:30RussI'm just getting Starting kernel ...
01:39:32Russhey Sammy
01:39:49Sammyhi Russ :)
01:41:51Russprpplague: its not making the jump to decompress_kernel
01:48:25prpplagueRuss: bummer
01:48:51prpplagueRuss: time to for mass debug printf's lol
09:04:09KuEdEmmatten: Where are u studying ?
09:05:03KuEdEah soo.. interessant
09:05:15KuEdEfh oder uni ?
09:05:41KuEdE<-- too
09:08:17seletzKuEdE: hi
09:08:37seletzKuEdE: have you had time to test your miniprint port?
09:09:07seletzKuEdE: its in CVS
09:09:24KuEdEyes.. u wrote me in the mail  
09:09:29KuEdEbtw: thx
09:09:41erikmgets more coffee
09:10:13erikmthere ought to be a law that forbids flat bed scanners to break in the morning
09:10:34KuEdE<-- has switched to another linux machine  . so he got to install some things
09:33:45seletzgetting coffee
10:11:32Sammylater all...
11:29:42seletzhi again
12:42:52seletzanyone here knows wether or not the sysctl interface may be used for board-specific things (like controlling on-board serial MUX hw and the like)?
12:43:36seletzasking on kernelnewbies
12:54:54seletzBZFlag: hi
13:51:44prpplaguemoring guys
13:52:25prpplaguemorning ibot
13:53:31mmattenhi prpplague
14:12:53seletzprpplague: hi
15:33:21prpplagueso whats going on today guys?
15:50:39seletzprpplague: well, i'm writing a power-button-switch driver.
15:50:46seletzprpplague: sorta weird.
15:50:51prpplagueseletz: interesting
15:51:08seletzprpplague: have to dig deep.
15:52:49prpplagueseletz: i've got a vr3, a dev yopy, and an inhand ELF
15:53:01seletzprpplague: urgh
15:53:04prpplagueseletz: coooool
15:53:07seletzprpplague: sorry, caps lock
15:53:24prpplagueaccidents happen
15:53:58seletzprpplague: Pic signals on gpio 18: "user wants to pwroff". I have a device driver which listens on that gpio (irq).
15:54:17prpplaguehas successfully gotten his companies email servers removed from blacklists
15:54:24prpplagueseletz: us the pic for anything else?
15:54:34seletzprpplague: nah.
15:55:02seletzprpplague: off-switching has to work _always_
15:55:29prpplagueseletz: i was thinking about using using a pic to poll multiple serial devices
15:55:54prpplagueseletz: our ex-sysadmin (winblows expert) got use blacklisted big time
15:55:54seletzprpplague: the weird part is: users have to have the option "no, i dont really meant it. Dont switch off".
15:57:29seletzprpplague: is the dev yopy different from the "startrek comm style" one?
15:57:34prpplagueseletz: i've got a vr3, a dev yopy, and an inhand ELF that i need to get rid of
15:57:43prpplagueseletz: no its the original
15:57:51seletzprpplague: does it work?
15:57:59prpplagueseletz: ya, works fine
15:58:09seletzprpplague: power consumption?
15:58:22prpplagueseletz: major
15:58:39prpplagueseletz: i'm not sure if its the code or the hardware
15:58:44seletzprpplague: runs 1/2 hour or what? lol...
15:59:04prpplagueseletz: more like 1.5 hour, and only a day or some in suspend
15:59:25seletzprpplague: what about the vr3?
15:59:39prpplagueseletz: works fine
15:59:51prpplagueseletz: i'm just not into the mips side of things
16:00:01prpplagueseletz: it just seem to have the power i need
16:00:33seletzprpplague: ah, ok. Me too :)
16:01:03prpplagueseletz: seen the ELF board?
16:01:15seletzprpplague: you gave me the url once
16:01:29seletzprpplague: do they have some sort of audio chip?
16:01:36prpplagueseletz: ya its nice, but WAY over priced
16:02:09prpplagueseletz: ya they have an fpga with a ucb1300
16:03:27seletzprpplague: what do you want for it?
16:04:45prpplagueseletz: $800 obo
16:06:06seletzprpplague: urg. ok, thats too much for something just to play around ....
16:06:27seletzprpplague: though i know the're _very_ expensive.
16:06:53prpplagueseletz: ya i payed close to $4000.00 for it
16:07:17seletz_4000_ _$_
16:07:33seletz800$ is a bargain, then ....
16:08:22prpplagueseletz: ya i was just starting out with embedded an i thought it would be cool to learn on
16:08:28prpplaguewas stupid
16:26:56prpplagueseletz: ya, i won't make that mistake again
16:28:34erikmheya, Russ|werk
16:28:47Russ|werkhey erikm
16:28:49Russ|werkI'm having a little kernel trouble
16:28:54prpplagueRuss|werk: howsy
16:28:56Russ|werkI'm chainbooting blob from bootldr, then running a kernel
16:29:19Russ|werkthe kernel fails to make the jump in head.S to decompress_kernel is misc.c
16:29:27erikmRuss|werk: on what platform?
16:29:38Russ|werkcamion, sa1110
16:29:53Russ|werkhey prpplague
16:30:04prpplaguecamion - the pda from hell
16:30:13prpplagueRuss|werk: morning
16:30:21prpplagueRuss|werk: still having fun i see
16:30:23erikmRuss|werk: ok. I thought it was on ipaq
16:30:34Russ|werkits more like an assabet
16:30:53Russ|werkit has a similar bcr and a lot of the same components
16:32:08erikmRuss|werk: how did you fool bootldr in chainbooting blob?
16:32:17Russ|werkload ram 0xc0040000 0x10000
16:32:22Russ|werkjump 0xc0040000
16:32:25erikmRuss|werk: ah, that's wrong
16:32:42erikmRuss|werk: just load blob-chain as if it is a kernel image
16:32:51erikmRuss|werk: then boot it
16:32:58Russ|werkthat doesn't do anything
16:33:10Russ|werkblob doesn't load in that case
16:33:12erikmRuss|werk: in that case bootldr will switch off the MMU, and afaik jump doesn't do that
16:33:33Russ|werkI'm trying to turn off the MMU in chain.S
16:33:42erikmRuss|werk: don't try to
16:34:15erikmRuss|werk: there is a dirty trick in chain.S that puts the correct magic word in the image that fools bootldr to think the blob-chain image is a kernel
16:34:34Russ|werkI tried that
16:34:40erikmRuss|werk: IIRC it is currently only enabled for ipaq
16:34:54Russ|werkit thinks its a netbsd kernel, and then boots it, but nothing happens
16:36:13erikmRuss|werk: well, chainbooting works on ipaq
16:36:51erikmRuss|werk: I'm afraid you'll have to go through bootldr's #ifdef hell to figure out what's different
16:37:21Russ|werk$Id: bootldr.c,v 1.30 2000/11/21 18:18:25 jamey Exp $
16:37:26Russ|werkits a pretty old version
16:37:34Russ|werkand its probably been modified by CIIT
16:37:56Russ|werkso I can't really go through the actual source
16:39:03Russ|werkis there any way of turning off the MMU mapping in chain.S?
16:39:17Russ|werkaparently is a mapping of DRAM to both its original location, and 0x0
16:39:23Russ|werkas well as flash to somewhere else
16:40:02erikmRuss|werk: *nod* IIRC the DRAM  is mapped 1:1, and the flash on two locations (once cached, once uncached)
16:40:22Russ|werkso..can I undo that in chain.S?
16:40:45Russ|werkI only want to do this until I have a working kernel, or until I find the jtag pins
16:40:48erikmRuss|werk: IIRC prpplague made an attempt to do that, prpplague?
16:41:26Russ|werk        @ disabling MMU and caches
16:41:26Russ|werkmrc     p15, 0, r0, c1, c0, 0   @ read control reg
16:41:26Russ|werkbic     r0, r0, #0x0d           @ clear WB, DC, MMU
16:41:26Russ|werkbic     r0, r0, #0x1000         @ clear Icache
16:41:26Russ|werkmcr     p15, 0, r0, c1, c0, 0
16:41:48prpplagueerikm: yes i was able to do that, however i'm not sure if the code is specific to the bootloader that was used on the ELF
16:42:21mmatten<- goin home, have a nice day all
16:42:30prpplagueRuss|werk: you can get my source at
16:42:32Russ|werkcya matt
16:42:37prpplaguemmatten: later
16:44:11prpplagueRuss|werk: you should be able to do a diff to find the specifics
16:46:43prpplagueRuss|werk: the section i changes was in start.S based on the normal_boot
16:48:15prpplagueRuss|werk: IIRC, it does some gymnastics to return a large block of memory back to 1:1 before shuting down the mmu
16:49:52Russ|werkwell the memory that chain.S is running in is 1:1 with bootldr
16:52:43prpplagueRuss|werk: hmm, i think you'll have to do more the clear and shutdown the cache when chain loading
16:52:56erikmRuss|werk: there were some special prerequisites for turning of the mmu. I don't remember exactly, but it's described in either the SA11x0 manual or the ARM ARM
16:53:21prpplagueerikm: ya thats what i ran into
16:54:28Russ|werkerikm: I do basically whats in head-sa1100.S
16:55:14Russ|werkprobably want to drain the caches first
16:55:23Russ|werkmight be my problem
16:55:25erikm*nod* that's important
16:55:41erikmIIRC you had to flush and invalidate both caches
16:55:44Russ|werk * Pause for a short time so that we give enough time
16:55:45Russ|werk * for the host to start a terminal up.
16:55:55Russ|werkwhat does that mean (head-sa1100.S)
16:56:02prpplagueMCRp15, 0, r0, c1, c0, 0
16:56:02prpplagueMCRp15, 0, r0, c8, c7, 0
16:56:02prpplagueMCRp15, 0, r0, c7, c7, 0
16:56:31prpplagueRuss|werk: IIRC, that is what i needed to clear the mmu properly
16:56:36erikmRuss|werk: IIRC it was for nico so he could switch to another console or so
16:56:46Russ|werkerikm: so why is it still there?
16:57:03erikmRuss|werk: you know nico is blind?
16:57:08Russ|werkI know
16:57:23Russ| embedded systems don't have displays anyway
16:57:57erikmRuss|werk: no, but he runs download tool on one console and terminal program on another or so
16:58:18erikmRuss|werk: and reading back is painful with his braille terminal
16:58:44Russ|werksuppose I'll just disable it in my local copies
16:59:12erikmRuss|werk: makes booting go faster
16:59:59Russ|werkthats the idea
17:00:52seletzok, i call it a day.
17:00:57seletzsee ya all
17:01:02erikmseletz: see ya
17:01:10seletz... and have a nice day!
17:02:58prpplagueseletz: later
17:19:18erikmRuss|werk: I just asked nico about the delay
17:19:47erikmRuss|werk: the delay was added for the case where the kernel is loaded with angelboot via the same port as the serial console
17:20:04erikmRuss|werk: this delay would allow you to start minicom after angelboot terminates and before any console messages are sent through
17:20:25Russ|werkso why not do an ifdef angelboot or something
17:21:53erikmRuss|werk: just remove it. angelboot is considered dead
17:22:06erikmRuss|werk: and it's only cosmetic code
17:22:27erikmRuss|werk: if you have to use angelboot, you could as well chain boot blob
17:24:17erikmRuss|werk: I think rmk will accept a patch that removes the delay, cause IIRC he wasn't too glad with it
17:29:15erikmgoes home before the rain starts
17:50:37KuEdEseletz: I validated my new stuff inCVS ... its ok  
18:12:11mmatten|homemmm seletz home, tough luck :(
21:17:41prpplaguewhy hello there mr.cto

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