irclog2html for blob on 2001.10.05

00:37:07sammymorning all...
02:05:48sammyRuss : ?
02:07:01sammyis that ramdisk need to match with the kernel version ?
02:08:36sammyor only upper then just can run ?
02:09:49sammyyou know what I am saying , if don't forgive my poor english ..
02:10:31Russif there are modules in the ramdisk, the modules need to have the same version as the kernel to work
02:11:55sammywhat kind of modules ?
02:15:51sammyI am just download the ramdisk with jffs2 by erikm an try to make a ramdisk with jffs2 on LART ...
02:16:01Russkernel modules
02:18:57sammyAmm...about this ? /lib/moduleskmod: failed to exec /sbin/modprobe -s -k char-major-4, errno = 2
02:19:14sammyis that a problem with make kernel ?
02:19:34sammyor just the problem with the kmod ?
02:21:27Russdo the modules match the kernel?
02:24:44sammyI don't know , I use the ramdisk with  
02:25:27sammyI don't know , I use the kernel-2.4.9-ac10-rmk3 and with the ramdisk with ramdisk-lart-minimal-new-tty.gz ...
02:27:19sammybut , I don't know about the module very much ... and the LART at this place still can work
02:27:59sammyis that the problem so ?
09:19:02sammyis home now...
13:14:46erikmhmm, I expected seletz to be here
16:40:39Russhmmm....command history support wouldn't be very usefull unless it remembered things between boots
16:41:58prpplagueRuss: something specific you refering to?
16:42:28Russcvs commits
16:43:30prpplaguesomeone adding command history to blob? i'd agree with you on the usefullness
18:10:36erikmhi Russ
18:13:13Russ|werkhey erikm
18:13:25Russ|werk"<Russ> hmmm....command history support wouldn't be very usefull unless it remembered things between boots"
18:13:30Russ|werk"<prpplague> someone adding command history to blob? i'd agree with you on the usefullness"
18:14:17erikmwhat about my last proposal? (making a separate diagnostics tool)
18:14:49Russ|werkI don't know what you'd put in there
18:15:23erikmthings like the memory tester etc.
18:15:33Russ|werkbtw, resistor value for led = (vpp - forward voltage of led)/forward current of led
18:15:44Russ|werkerikm: just add them as extra commands
18:15:59erikmresistor value is OK
18:16:22Russ|werkit'd be easy enough to make a simple term program specifially for blob diag
18:16:44erikmlike expect, you mean
18:16:50Russ|werk52 ohms just seems a bit small
18:18:01erikm470 ohm is usually enough for 10mA LED current and 0.7V forward voltage over LED
18:18:18erikmand 5V Vpp
18:18:24Russ|werkthis led says 25mA @ 2V
18:18:39Russ|werkand its 3.3Vpp
18:19:13erikmhmm 52 ohm is correct in that case
18:19:29Russ|werkdo I even need one?
18:19:35erikmyou'd better
18:20:29Russ|werknow I just need to find the closest manufactured value
18:20:47erikm47 ohm, or what's the next in the E12 range? 68 ohm, I think
18:22:12Russ|werkis there anything replacing tantalums?
18:22:26Russ|werk68 and 47 sound right
18:22:30Russ|werkprobably go with 68
18:23:45erikmafaik there isn't anything replacing tantalums. jdb might know, but he's currently to a conference in zurich
18:27:05erikmanyway, about the command line history
18:27:59erikmI'm going to remove it, and I'll extend the command line parser so it recognises abbreviations
18:28:22erikmthings like "do blob"
18:28:25erikmfor "download blob"
18:29:47erikmI'll also restructure the code so seletz can put his diagnostics stuff in a separate package that you can download and start as if it is a linux kernel
18:31:12erikmbecause blob and "diag" share quite some code, I'll make a lib directory to put in the serial code, LED code, etc. blob and diag will link against this library
18:31:54erikmblob will be small and simple (like the goal was from the start), and diag can be more full featured
18:32:44Russ|werkhey WeirdArms
18:33:32erikmWeirdArms: eh, what time is it at your place?
18:34:43WeirdArms5:30 am
18:35:29WeirdArmslate or early
18:35:35WeirdArmswhichever way you want to look at it
18:35:45erikmWeirdArms: woke up early, or not yet to bed?
18:35:50WeirdArmsnot yet bed
18:35:55WeirdArmsWorking on FASS
18:36:06WeirdArmstrying to work out how to debug it
18:36:55WeirdArmsand why Russel does some of the things he does
18:37:07WeirdArmsI sent him a patch
18:37:21WeirdArmshe probably won't have time to look at it though
18:38:20WeirdArmsyou know it's fun when you disable the domain which tags the exception vector table
18:38:22erikmhe's currently at Alan and Telsa
18:38:28WeirdArmsinfinite exception loop
18:38:34WeirdArmsYeah he said
18:40:12erikmhe hoped for good weather, but yesterday evening they had already two "mild showers" as Telsa called it
18:41:14erikmmind you: he had to close the window or otherwise his bed would become soaking wet :)
18:42:36WeirdArmsWhat's he doing there
18:42:54WeirdArmsHe said "I'm working from home at Alan's house, if that makes sense"
18:43:03erikmjust visiting afaik. and some walking, if weather permits
18:43:22WeirdArmsI want to know whats wrong
18:43:29WeirdArmsI get a looping page fault
18:43:40WeirdArmsI thought if a page fault fails you get a segfault
18:44:00erikmthat's what I would think
18:44:01WeirdArmserikm: what you up to?
18:44:12WeirdArmsI'll check if it is a bad fault
18:45:25erikmblob hacking. playing with the assabet. turns out the second assabet I got from Intel has a faulty jtag cable
18:45:57WeirdArmsbut the other cable works?
18:46:08WeirdArmsDo you think you would be able to look at the PLEB flash stuff?
18:46:10erikmtakes some time before you find out
18:46:38erikmI'm not sure. I'm currently completely restructuring the directory layout
18:46:39WeirdArmsIf you interested I can send you a patch for the FASS stuff too ;)
18:47:21WeirdArmstrying to get it done for a mid nov paper deadline
18:47:27erikmseletz submitted command line history to CVS. jdb and I completely disagree with that, so I'll remove it
18:47:31WeirdArmsDon't know how likely that is but damn I'm going to try
18:47:44WeirdArmsyeah fix delete first :)
18:47:55erikmfix delete? what's wrong with delete?
18:47:57WeirdArmsbackspace doesn't work for me
18:48:14erikmnever had any problems with that
18:48:15WeirdArmsmight be my term
18:48:38WeirdArmsit works
18:48:49erikmyou reparied the LART?
18:48:59WeirdArmsusing pleb
18:49:13WeirdArmswith your life saving small ramdisk
18:49:42WeirdArmsyou can probably imagine how much that helps when I can't flash stuff
18:49:54erikmyeah, a *lot* smaller
18:50:22WeirdArmsI probably could work it out, we have an old kernel patched to enable the flash but it would take me a day to track it down off someone and move it to the current kernel
18:53:56WeirdArmsI think the only time you get a repeated fault
18:54:02WeirdArmsis when the page has been mapped in ok
18:54:12WeirdArmscould be a caching issue
18:54:30WeirdArmsdo you know how to turn the cache off in linux erik?
18:55:06erikmyeah, play with the MMU
18:55:28Russ|werkwhat flash do you have?
18:58:11erikmWeirdArms: cache disable code is in arch/arm/boot/compressed/head-sa1100.S
18:58:37WeirdArmsIntels 3v advanced+ boot block flash
18:58:54WeirdArms2 * 2MB chips 32bit wide
19:04:51Russ|werkand blob isn't flashing it?
19:05:44WeirdArmsI haven't written the flash code for it
19:05:51WeirdArmsI tried the tuxscreen code
19:05:54WeirdArmsthat didn't work
19:06:05Russ|werkuse the lart code, but without the plaiting
19:06:07WeirdArmsI thought the flash chips would be sufficiently similar
19:06:14WeirdArmssame chips?
19:06:22Russ|werkthe lart code is for intel flash, the tuxscreen code is for amd
19:06:32WeirdArmsyeah maybe I shoudl
19:07:04erikmRuss|werk: did I already explain you my plans to rewrite the flash code?
19:07:45prpplaguehowdy all
19:07:48erikmhi prpplague
19:07:49Russ|werkdifferent files for differetn flash?
19:07:59erikmno, not necessary
19:08:28erikmsuppose we want to write an image to the flash, we first start reading the flash
19:09:03erikmif we don't find 0xffffffff, we issue an erase command for that address which will erase the complete block
19:09:37Russ|werkI would say just erase, I wouldn't trust 0xffffffff
19:10:16erikmkeep on scanning and erasing until we have enough room for the image
19:10:28erikmoh, but it works
19:10:47Russ|werkwhat if it was a power loss during an erase
19:11:02Russ|werkor the first part of whatever it is, happens to be 0xffffffff
19:11:12erikmthe code that copied blob from the LART external flash to the internal flash used this trick
19:11:18WeirdArmsthats what I was wondering
19:11:20erikmwe scan the whole block
19:11:40Russ|werkI've watched some mtd discussions on this sort of thing
19:11:46erikmthe flash writer wouldn't even have an idea what a block is
19:11:48Russ|werkand it doesn't seem to be worth it
19:11:51WeirdArmsusing the LART flash code does the same as the TuxScreen one did
19:11:56WeirdArmsit crashes blob
19:12:02Russ|werkWeirdArms: turn off the plaiting
19:12:08WeirdArmsit is off
19:12:28Russ|werksuppose its time to pull out the data sheet and schematics tehn
19:12:49prpplagueerikm: i looked back on the conversation about command line history, i like the final decision
19:13:04WeirdArmsI just don't feel like working out the flash
19:13:04erikmprpplague: the separate diagnostics tool?
19:14:28erikmRuss|werk: so what could go wrong if we use my plan?
19:14:49erikmRuss|werk: especially if I CRC32 or md5sum the contents of the flash
19:15:03Russ|werkif you lose power during an erase
19:15:15erikmRuss|werk: ok, one block toast
19:15:18Russ|werkthen read the area, and notice its all 0xffffffff
19:15:23WeirdArmsyou have a partial erase and you are not sure of the bundary
19:15:35Russ|werkwhen you go abck and read it again, it may not still be 0xffffffff
19:15:41Russ|werkthe bits wobble around
19:15:54erikmRuss|werk: ok, something went wrong. so we program the flash
19:16:09Russ|werkhow do you know something went wrong, you read all 0xffffffff
19:16:21erikmRuss|werk: after programming we md5sum the contents of the flash and detect something went wrong
19:16:46Russ|werkhow do you md5sum an erased blockZ
19:16:48Russ|werker, ?
19:16:48erikmuser gets error "flash failed"
19:17:10erikmlet me start from begin:
19:17:12Russ|werkI'm confused
19:17:18erikm1. erase block
19:17:27erikm2. power fails, block is half erased
19:17:32erikm3. power on again
19:17:40erikm4. download image
19:17:47erikm5. flash image
19:18:01erikm6. half erased block is detected as clean, so not erased again
19:18:06erikm7. write image
19:18:19erikm8. re-read image from flash and md5sum
19:18:33erikm9. compare md5sum with md5sum from the image in memory
19:18:37Russ|werkjust seems simpler to erase than to do all this checking and re-erasing reflashing
19:19:21erikmRuss|werk: no, the advantage of this approach is that the flash code doesn't have to have an idea about block boundaries
19:19:35WeirdArmsis it such a big deal for it to know though?
19:19:51Russ|werkI see what you are trying to do there
19:20:02Russ|werkand it might be helpfull for the user to know that something didn't f lash right
19:20:13erikmRuss|werk: flashing blob >1k on LART already fails
19:20:18erikm16k I mean
19:20:18Russ|werkskipping erases ends up being a side effect
19:20:33prpplagueerikm: ya the seperate diagnostics and the shorten keyword recognition ie "do blob"
19:20:56erikmRuss|werk: *nod*
19:21:10erikmit's a side effect, but it makes the flash code blocksize agnostic
19:21:23erikmso it will work without any modifcation on all platforms
19:21:35erikmexcept the difference for commands, of course
19:22:28prpplagueerikm: i've not finished my patch yet, this week has been crazy
19:22:45erikmprpplague: hehe, over here as well :)
19:23:53erikmprpplague: I've been giving courses in fundamental filter theory, fundamental sampling theory, and fundamental maths to coworkers :)
19:25:36prpplagueerikm: fundametal filter theory? audio filtering or data filtering?
19:26:18erikmprpplague: things like: aliasing has to be dealt with immediately, you can't first put your data through an integrator and try to remove the aliasing effect afterwards
19:26:41erikmprpplague: there is no difference in audio filtering and data filtering for me
19:28:14prpplagueerikm: true, the description just kinda points you in the intended application area
19:28:48erikmprpplague: but in this case it was about filtering data from accelerometers and gyroscopes
19:29:18prpplagueerikm: lol, ok,ok
19:29:43erikmprpplague: sampled at 2000Hz, filtered and decimated to 100Hz (otherwise userland keep up with the speed)
19:30:41WeirdArmserikm: How's the trimedia chip thingy project going? Have they got anywhere? Has JDB got his crazy dual powerPC board done yet to outdo it?
19:31:28erikmWeirdArms: Sony UK made a quad trimedia board, don't know about the current status of the dual PPC board
19:32:04erikmWeirdArms: unfortunately sony wouldn't listen to our advice and put the cheapest PCI bridge on the board they could find
19:32:54erikmWeirdArms: a bloody transparent bridge, which will severely limit you in the amount of memory you can put in your host, especially if you want to put three our four quad trimedia boards in your host
19:36:29WeirdArmsany guesses of the different between pte and ppte?
19:37:02erikmpage table entry and pointer to page table entry?
19:37:19WeirdArmslooks like hardware page table entry and linux page table entry
19:37:47WeirdArmsprintk(", *pte = %08lx", pte_val(*pte));
19:37:55WeirdArmsprintk(", *ppte = %08lx", pte_val(pte[-PTRS_PER_PTE]));
19:38:31WeirdArmsthat looks like an offset the size of the leaf page entry away
19:39:59WeirdArmsthe entry seems set to reserved
19:40:03WeirdArmswhich isn't good
19:41:28erikmWeirdArms: sorry, I don't know
19:45:59erikmdecides between wisky and cognac
19:47:23erikmnope, wisky
19:49:59erikmlife is good: bach's brandenburger concertos on the stereo, glass of single malt whisky within reach, and a new chair :)
19:52:15erikmI bought the chair by accident: I was at Ikea for a new lamp, tried the chair, and fell in love with it
19:56:39WeirdArmsthis doesn't make sense
19:56:48WeirdArmsit shouldn't be faulting
20:02:48erikmproblems with the 'net, Russ|werk?
20:05:29Russ|werkxchat is acting "funny"
20:08:11WeirdArmsI found the bug but I'm too tired to focus
20:08:28erikmWeirdArms: write it down, go to bed
20:08:36erikmWeirdArms: and fix it tomorrow
20:08:45WeirdArmsit is tommorow
20:08:50WeirdArmsNo i must fix it now
20:57:39erikmWeirdArms: bug already fixed?
21:26:16prpplagueerikm: i think he's fallen asleep at the keyboard....
21:27:51erikmprpplague: I think so, yes
21:29:33prpplaguehmm paycheck time...
21:51:37prpplagueerikm: howdy, i figured you were headed for some zzzz's
21:51:52erikmno, ISDN bailed out
21:51:52WeirdArmserikm: i hven't fixed the bug but I know exactly whats wrong and how to fix it ;)
21:59:16erikmgoes zzz
22:45:44BZFlagconsider yourself LARTed needs to be "blob's your uncle" ;-)
22:47:40Russ|werkBZFlag: did you get a chance to test the bootp stuff in 0.9.6 yet?

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