| 00:01.35 | prpplague | CosmicPenguin i'm just not sure if this guy is just too intrenched in the EE style of big companies, or if he's just not that concerned with the quality of his work |
| 00:02.16 | CosmicPenguin | it can be both |
| 00:02.17 | CosmicPenguin | sadly |
| 00:03.41 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 00:03.46 | prpplague | trying to dig around for some conferences within driving distance, but doesn't appear to be too many |
| 00:06.56 | prpplague | hehe, definetly an HR created job posting "10+ years experience with Qt" |
| 00:08.11 | g1powermac_PB | hmm |
| 00:08.26 | g1powermac_PB | could use somebody with that many years of Qt experience ;-) |
| 00:08.36 | g1powermac_PB | dang qtopia/uclibc issue |
| 00:08.48 | prpplague | hehe yet another one for embedded linux engineer - "Have a basic knowledge of DOS window commands and features" |
| 00:08.58 | prpplague | that one made me laugh |
| 00:09.13 | SpeedEvil | tries to remember that bastard dos window thing. |
| 00:09.40 | SpeedEvil | It was shipped with later MSDOS 5 I think. |
| 00:09.50 | SpeedEvil | (though that's not of course what they mean) |
| 00:13.45 | roxfan | embedded development is often done on windows |
| 00:14.23 | roxfan | though developing linux is probably more convenient on a linux host |
| 00:15.38 | prpplague | roxfan what i found funny was that if someone was qualified to do embedded linux dev, surely knowing/learning dos command line would not be a problem |
| 00:15.49 | roxfan | i guess |
| 00:16.08 | roxfan | well, those ads are usually written by HR drones anyway |
| 00:16.19 | prpplague | roxfan imho its like having an architect job posting with "must be able to do addition and subtraction" |
| 00:16.25 | roxfan | heh |
| 00:16.27 | prpplague | roxfan which was my point |
| 00:16.48 | roxfan | like those "must have 5 years C# experience" ads in 2001 |
| 00:17.01 | baked_potatoe | Hey, I think I got my JTAG working. If I am right I just never hooked HFXOUT to HFXIN which were an output and input for a external crystal.. I guess the internal ring oscillator needs to be connected through these too. |
| 00:19.26 | roxfan | i can't get my arm-jtag dongle working :/ |
| 00:20.08 | baked_potatoe | Well, TDO still does not go low exactly when I clock the first bit in (zero) for the bypass thingy, but it waits until I clock a few more times and then it does? Is this a little weird? |
| 00:21.03 | roxfan | do you set ir to bypass first? |
| 00:21.34 | roxfan | the default instruction on many systems is idcode |
| 00:21.45 | baked_potatoe | I just clocked five times with TMS high... ahh you think it has idcode? |
| 00:21.50 | roxfan | dr of which is 32 bits |
| 00:22.07 | baked_potatoe | But, if I try to clock in eight bits at a time it never goes low... |
| 00:22.28 | prpplague | hmm all the good courses are either in new york or san fran |
| 00:23.46 | baked_potatoe | But, if I try to clock in eight bits at a time it never goes low... |
| 00:23.50 | baked_potatoe | crap... |
| 00:23.57 | baked_potatoe | (wrong window, sorry) |
| 00:27.56 | prpplague | amazing, dfw area isn't listed anywhere |
| 00:27.59 | prpplague | http://www.webagesolutions.com/training/unix/LS110/index.html |
| 00:31.05 | prpplague | think i'll open a linux training school |
| 00:31.20 | prpplague | "Dave's embedded linux training and nudie bar" |
| 00:32.00 | Crofton | wants to run a nudie bar |
| 00:32.44 | prpplague | Crofton come on over to dfw, we'll do the nudie bar in the front of the building, embedded dev consultant in the back |
| 00:32.59 | Crofton | heh |
| 00:33.35 | prpplague | "wanted: dancers who are familiar with C and ruby on rails" |
| 00:35.08 | prpplague | goes back to coding |
| 00:35.10 | Crofton | dancers familiar with patent law |
| 00:35.11 | Crofton | http://abclocal.go.com/wabc/story?section=news/local&id=6095987 |
| 00:35.21 | prpplague | looks |
| 00:36.54 | prpplague | jeeze |
| 00:37.32 | prpplague | here is someone who has, for all appearances, a valid patent claim, and a big business steals it |
| 00:37.57 | prpplague | on the other hand you got big business who make up bogus patents and then haras the @@#$% out of small innovators |
| 00:38.38 | prpplague | Crofton i've considered several times of getting organized where just about everything i do i submit a patent |
| 00:39.01 | Crofton | yeah |
| 00:39.28 | Crofton | I'm not up on bra technology, but they had some comments the idea was old |
| 00:40.21 | prpplague | Crofton yea, but that trick isn't new either, meet with someone, tell them the product will never sell, then go and develop your own design |
| 00:40.35 | Crofton | yep |
| 00:40.48 | prpplague | Crofton its exactly what the guy that did the intermiten windshield wipers had done to him |
| 00:41.17 | prpplague | Crofton hey did you figure out your nand flash issue? |
| 00:41.23 | Crofton | no |
| 00:41.31 | prpplague | bummer |
| 00:41.43 | Crofton | I need to check a couple more things |
| 00:41.51 | Crofton | the cycle is time consuming |
| 00:41.54 | prpplague | so after configuring the rootfs section into two partitions the problem went away? |
| 00:42.00 | Crofton | and I ahve several things going at once |
| 00:42.03 | Crofton | no |
| 00:42.05 | prpplague | Crofton np |
| 00:42.11 | Crofton | it depends where the border is |
| 00:42.15 | prpplague | Crofton ahh ok |
| 00:42.24 | prpplague | Crofton this a known good board? |
| 00:42.31 | Crofton | but I can wipe kernel by writing into 3, when writing into 2 does not wipe kerene; |
| 00:42.35 | Crofton | hmmm |
| 00:42.41 | Crofton | probably |
| 00:42.57 | prpplague | almost sounds like a stuck address line |
| 00:43.21 | Crofton | this is NAND |
| 00:43.24 | Crofton | no address lines |
| 00:43.30 | Crofton | jsut data bus |
| 00:44.07 | prpplague | Crofton yea, i know its latched, but you are going through a controller that might be garbaging the address |
| 00:44.25 | Crofton | yeah, but all that is on dv chip |
| 00:44.41 | Crofton | only about ten traces between the chip and flash |
| 00:45.05 | prpplague | Crofton i assume the controller has some config options for the different block sizes and number of address spaces |
| 00:45.06 | Crofton | the board should work |
| 00:45.34 | Crofton | heh |
| 00:45.38 | Crofton | you push my limits |
| 00:45.50 | Crofton | someone else did the NAND driver for DV |
| 00:47.37 | prpplague | Crofton you have a part number for the nand chip/ |
| 00:47.38 | prpplague | ? |
| 00:48.20 | Crofton | I'll have to look later |
| 00:48.30 | Crofton | I am mobile atm watching the talking heads :) |
| 00:48.43 | prpplague | np |
| 00:48.59 | prpplague | Crofton happen to know which dv soc it is? |
| 00:49.05 | Crofton | dm6446 |
| 00:49.13 | Crofton | same as evm, but different hw |
| 00:49.24 | Crofton | big block NAND |
| 00:49.43 | prpplague | digs |
| 00:51.40 | ds2 | I say do a hex compare of the kernel partition |
| 00:51.50 | ds2 | see if the corruption is all over or just at the boundaries |
| 00:52.21 | Crofton | that may be tomorrows effort |
| 00:52.22 | ds2 | and the pattern of corruption would be useful...like getting trashed w/random vs 0xff's |
| 00:52.23 | *** part/#edev LyosNorezel (n=LyosNore@unaffiliated/lyosnorezel) |
| 00:52.27 | Crofton | or I beat on dsplink some |
| 00:52.31 | Crofton | yeah |
| 00:52.55 | ds2 | but, IMO, it looks like a wrapping problem |
| 01:00.32 | prpplague | Crofton interesting nand flash controller |
| 01:09.06 | prpplague | Crofton wow thats one complexe nand controller |
| 01:09.21 | Crofton | heh |
| 01:09.41 | *** join/#edev Appleman1234 (n=Appleman@131.181.99.30) |
| 01:10.12 | ds2 | Crofton: how do you like the Divinci procs so far? |
| 01:11.35 | Crofton | well, I haven't played with the dsp |
| 01:11.53 | Crofton | I am on odd fit for it, because it is focused on video apps |
| 01:12.05 | Crofton | I am interested in software radio |
| 01:12.10 | ds2 | which DSP does the devinci's sport? |
| 01:12.17 | Crofton | c64 |
| 01:12.26 | Crofton | the omap3 stuff is more interesting |
| 01:12.27 | ds2 | isn't that the same as the OMAP3's? |
| 01:12.43 | Crofton | but, I have a DV board designed for SDR |
| 01:12.47 | Crofton | ds2, exactly |
| 01:13.10 | ds2 | so your problem is finding a OMAP3 that mere mortals can purchase? |
| 01:14.53 | ds2 | Crofton: have you seen the OMAP3 EVM's from Minstral? they are suppose to be orderable |
| 01:15.13 | Crofton | no |
| 01:15.20 | Crofton | I am working on getting a beagle though |
| 01:15.32 | prpplague | Crofton you said you could erase mtd3 but not mtd2? |
| 01:15.40 | Crofton | beagel I should be able to attach a USRP |
| 01:15.47 | Crofton | erase works on both |
| 01:15.54 | prpplague | Crofton or could you erase both depending on where the division is |
| 01:16.01 | ds2 | beagle is nice but not being able to get my hands on one now is irritating |
| 01:16.02 | Crofton | prpplague, exactly |
| 01:16.08 | Crofton | heh |
| 01:16.36 | ds2 | for the flash, you have confirmed your erase block size is what is spec'ed in the datasheet? |
| 01:16.52 | Crofton | it prints the right number during boot |
| 01:18.26 | prpplague | Crofton block size is actually what on this chip? 256k ? |
| 01:18.44 | Crofton | 128K |
| 01:18.57 | prpplague | Crofton ahh ok |
| 01:19.02 | ds2 | and the datasheet says block size is the same as the erase size? |
| 01:19.18 | *** join/#edev CP|home (n=nobody@www.crouse-house.com) |
| 01:19.18 | ds2 | I think some chips might have a larger erase size then block size |
| 01:19.21 | Crofton | I forget |
| 01:19.30 | Crofton | that is likely true for this .... |
| 01:20.02 | ds2 | I much rather be looking at flash problems this the ______ stuff I am looking at |
| 01:20.55 | prpplague | Crofton hmm, if the nand gets a request to erase 128k and it thinks that each block is 64k it might be actually erasing 2 blocks of 128k by mistake |
| 01:21.20 | prpplague | Crofton but that wouldn't explain your problem |
| 01:21.48 | prpplague | i think it might be more likely that your software thinks it is a 256k per block and its actually only 128k per block |
| 01:22.16 | Crofton | let me grab the chip number |
| 01:22.17 | prpplague | which would explain why the kernel was getting hosed |
| 01:23.38 | prpplague | ds2 that makes two of us |
| 01:25.15 | ds2 | prpplague: sigh |
| 01:25.47 | Crofton | NAND device: Manufacturer ID: 0x20, Chip ID: 0xf1 (ST Micro NAND 128MiB 3,3V 8-bit) |
| 01:26.15 | prpplague | Crofton no chip number? |
| 01:26.37 | Crofton | NAND01GW3B2AZA6E |
| 01:26.48 | prpplague | 128megabyte doesn't sound right for big block |
| 01:27.36 | prpplague | locates datasheet |
| 01:27.46 | ds2 | the digikey links is broq |
| 01:28.37 | ds2 | grrr ST sold off NAND |
| 01:30.51 | prpplague | hmm, i'm showing that it has 2k page size |
| 01:31.02 | Crofton | erase size is bigger |
| 01:32.26 | prpplague | right, just locating info |
| 01:34.16 | prpplague | http://www.numonyx.com/en-US/MemoryProducts/NAND/Pages/MLCLargePageTechicalDocuments.aspx |
| 01:36.18 | prpplague | yea block size is 256k |
| 01:37.17 | prpplague | interesting there is a multiblock erase command |
| 01:40.29 | prpplague | ahh i think i've found your issue |
| 01:40.42 | prpplague | Crofton what offset are you copying your kernel to? |
| 01:41.29 | Crofton | 0x20a0000 length 0x160000 |
| 01:41.50 | *** join/#edev chouimat (n=dieu@r2351064.cidc.net) |
| 01:43.53 | prpplague | Crofton where did you get the numbers for the length? |
| 01:45.45 | Crofton | kernel length? |
| 01:45.54 | Crofton | I converted decimal to hex ... |
| 01:46.05 | prpplague | Crofton 0x160000 appears to fall in the middle of a 256k block, i'm guessing that needs to be 0x180000 so that it marks the end of the block |
| 01:46.14 | Crofton | hmmm |
| 01:46.33 | prpplague | Crofton have you got you physmap mtd mapping available again? |
| 01:46.33 | Crofton | are you really certain the erase block is that larger\, I swear it is 128 ... |
| 01:46.40 | Crofton | Creating 4 MTD partitions on "NAND 128MiB 3,3V 8-bit": |
| 01:46.40 | Crofton | 0x00000000-0x000a0000 : "Bootloader" |
| 01:46.40 | Crofton | 0x000a0000-0x002a0000 : "Kernel" |
| 01:46.40 | Crofton | 0x002a0000-0x01ea0000 : "File System" |
| 01:46.40 | Crofton | 0x01ea0000-0x08000000 : "File System2" |
| 01:46.42 | Crofton | ? |
| 01:46.50 | prpplague | yea that one, one sec let me check |
| 01:48.32 | prpplague | the datasheet i'm looking at clearly says 256k |
| 01:48.57 | prpplague | Crofton did you adjust the physmap values yourself, or were those some default values copied from somewhere? |
| 01:48.58 | Crofton | hmmm |
| 01:49.07 | Crofton | I made them up |
| 01:49.10 | prpplague | ahh |
| 01:49.14 | prpplague | one sec |
| 01:49.15 | Crofton | based on some reading |
| 01:50.02 | prpplague | Crofton yea, my calcs match what i'm seeing on this page http://wiki.davincidsp.com/index.php?title=DM355_NAND_Partitions |
| 01:50.42 | prpplague | Crofton you need the bootloader as 0x40000, kernel as 0x180000, and the rest starting after that |
| 01:51.00 | prpplague | at 0x1c0000 |
| 01:51.21 | prpplague | Crofton you are using the MLC chipset |
| 01:51.50 | Crofton | ? |
| 01:52.38 | Crofton | what is the MLC chipset? |
| 01:52.53 | prpplague | short and sweet its multiplane nand flash |
| 01:53.15 | prpplague | http://www.numonyx.com/en-US/MemoryProducts/NAND/Pages/MLCLargePage.aspx |
| 01:53.32 | prpplague | you have to make sure that your partition sizes fall on 256k boundries |
| 01:53.58 | prpplague | else if you do an erase when it extends into a block from another partition, that gets erased |
| 01:54.25 | prpplague | so all your partition values need to be multiples of 256k |
| 01:54.42 | Crofton | ok |
| 01:54.44 | g1powermac_PB | anyone here have any scratchbox experience? |
| 01:54.52 | Crofton | I need to do some double checking |
| 01:55.00 | Crofton | I swear I came up with half that |
| 01:55.04 | g1powermac_PB | is going insane |
| 01:55.16 | Crofton | I think you are thinking of the 2G version |
| 01:55.17 | prpplague | Crofton right you were probably looking at the info for the SLC |
| 01:55.26 | Crofton | nand02 versus nand01 |
| 01:55.56 | prpplague | Crofton gotta make sure you got the correct one, the part number you gave me was for a MLC chipset |
| 01:56.03 | Crofton | ok, I am going to double check the PN |
| 01:56.14 | prpplague | Crofton you may have mistakenly gotten the datasheet for a SLC part which would indeed be 128k |
| 01:56.21 | Crofton | hunts for reading glasses |
| 01:56.36 | Crofton | the size should be autodetected from the chip id's |
| 01:57.46 | Crofton | thanks for your help, I'm going to double check this in the AM |
| 01:57.49 | Crofton | when I can think |
| 01:59.18 | prpplague | np, i'll continue digging, but so far this fits with your description of the problem |
| 02:01.54 | Crofton | The chip id, 0xF1 is for a 1Gigabit, 128 erase block |
| 02:02.40 | Crofton | from drivers/mtd/nand/nand_ids.c |
| 02:02.53 | Crofton | NAND device: Manufacturer ID: 0x20, Chip ID: 0xf1 (ST Micro NAND 128MiB 3,3V 8-bit) |
| 02:04.21 | Crofton | gn |
| 02:06.11 | prpplague | http://datasheets.weeqoo.com/88888/2006122394956724.pdf |
| 02:06.15 | prpplague | double checing |
| 02:06.17 | prpplague | Crofton later |
| 02:06.47 | Crofton | check the id the chip returns |
| 02:07.11 | prpplague | looking now |
| 02:09.58 | prpplague | Crofton yep that matches |
| 02:10.58 | *** join/#edev chouimat (n=dieu@r2351064.cidc.net) |
| 02:11.41 | prpplague | Crofton that shows it to be 128k block size |
| 02:12.22 | prpplague | Crofton yea its 128k block not 256k |
| 02:15.17 | prpplague | Crofton hmm, so much for that idea |
| 02:15.31 | prpplague | Crofton might try configuring it for 256k block size just for kicks |
| 02:19.27 | prpplague | solar no worries, i was swampped today and didn't get the package shipped |
| 02:33.54 | prpplague | Crofton man i wish i could my jtag on that device |
| 02:53.28 | CP|home | Arctic Ice More Vulnerable To Sunny Weather, New Study Shows |
| 02:53.34 | CP|home | Your tax dollars at work |
| 02:54.30 | prpplague | CP|home indeed |
| 03:04.07 | prpplague | Crofton give me a shout in the morning, i have a couple printk's for you to insert in the nand_base.c to verify that its reading the correct extended bits |
| 03:30.22 | *** join/#edev linac (n=lin@121.41.35.231) |
| 04:01.55 | *** join/#edev chouimat (n=dieu@r2351064.cidc.net) |
| 04:06.49 | *** join/#edev DrZimmerman (n=theo@77-56-246-76.dclient.hispeed.ch) |
| 04:23.30 | *** join/#edev rustye (n=rusty@pool-71-123-136-184.dllstx.fios.verizon.net) |
| 04:25.31 | *** part/#edev rustye (n=rusty@pool-71-123-136-184.dllstx.fios.verizon.net) |
| 04:38.38 | *** join/#edev Appleman1234 (n=Appleman@131.181.47.33) |
| 05:32.06 | *** join/#edev harpal (n=harpal@121.246.75.165) |
| 05:53.35 | *** join/#edev chouimat (n=dieu@r2351064.cidc.net) |
| 06:07.14 | *** join/#edev drath_ (i=vmaster@p5B07D421.dip.t-dialin.net) |
| 06:17.24 | *** join/#edev emil|k (n=emil@p548CDD2B.dip.t-dialin.net) |
| 07:21.39 | *** join/#edev Appleman1234 (n=Appleman@131.181.99.21) |
| 07:26.09 | *** join/#edev [AD]Turbo (n=turbo@host24-241-dynamic.22-79-r.retail.telecomitalia.it) |
| 07:29.05 | [AD]Turbo | yo |
| 07:46.20 | *** join/#edev hw_ (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 07:47.40 | *** join/#edev linac (n=lin@222.76.82.137) |
| 08:36.33 | *** join/#edev howtoo (n=howtoo@195.159.62.66) |
| 08:39.56 | *** join/#edev lyakh (n=lyakh@p57BD2117.dip0.t-ipconnect.de) |
| 09:47.04 | *** join/#edev emil|k (n=emil@i59F63659.versanet.de) |
| 09:50.57 | *** join/#edev ali_as_ (n=as@ambix.plus.com) |
| 10:56.34 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 11:32.33 | *** join/#edev ali_as (n=as@ambix.plus.com) |
| 11:34.26 | *** join/#edev KazaLite (n=kazaLite@203.215.180.205) |
| 11:45.06 | *** join/#edev linac (n=lin@121.41.35.231) |
| 12:13.32 | *** join/#edev howtoo (n=howtoo@195.159.62.66) |
| 12:46.48 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 12:46.48 | *** mode/#edev [+o GPSFan] by ChanServ |
| 13:16.51 | *** join/#edev sjhill (n=sjhill@real.realitydiluted.com) |
| 13:16.51 | *** mode/#edev [+o sjhill] by ChanServ |
| 13:22.40 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 13:25.27 | prpplague | Crofton: you alive yet this morning? |
| 13:25.51 | Crofton | yeah |
| 13:26.28 | prpplague | Crofton: got a suggest for you on your nand issue |
| 13:26.46 | Crofton | ok |
| 13:27.04 | prpplague | Crofton: the erase block size for that chip is calculated from the extended ID bytes |
| 13:27.12 | Crofton | I can build now |
| 13:27.17 | Crofton | the f1? |
| 13:27.23 | prpplague | Crofton: yea |
| 13:27.54 | prpplague | Crofton: let me pull up the relevant code |
| 13:27.58 | Crofton | you really believe it is 256K? I'm thinking that is for the 2G part and I have the 1G part .... |
| 13:29.14 | prpplague | Crofton: sorry no, its a 128k part, but i do think its getting the wrong block |
| 13:29.34 | prpplague | Crofton: one way to find out is to add a printk to the calculation code |
| 13:29.37 | prpplague | Crofton: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=blob;f=drivers/mtd/nand/nand_base.c;h=7acb1a0e7409fbf950a06e97f8e58b3a1be38559;hb=HEAD |
| 13:29.39 | Crofton | ok |
| 13:29.47 | Crofton | right |
| 13:29.47 | prpplague | Crofton: starting at line 2280 is where that is calculated |
| 13:29.54 | Crofton | glad I didn't get it wrong :) |
| 13:30.20 | prpplague | Crofton: yea my bad, i had the wrong datasheet last night |
| 13:30.39 | Crofton | np |
| 13:30.49 | prpplague | Crofton: at 2287 you can add a printk that prints the page size and the block size |
| 13:31.04 | prpplague | Crofton: that should give you the exact info to make sure it is calculated properly |
| 13:31.08 | Crofton | ok |
| 13:31.24 | Crofton | that url is truncating the file at around line 550 |
| 13:31.36 | prpplague | odd |
| 13:32.29 | prpplague | Crofton: hmm, that comes up properly for me |
| 13:32.38 | prpplague | makes a tiny url |
| 13:32.40 | Crofton | got it |
| 13:33.05 | prpplague | http://tinyurl.com/6jd3ng |
| 13:33.27 | Crofton | I found the spot with vim |
| 13:34.05 | Crofton | thanks, I'll give this a try |
| 13:34.56 | prpplague | Crofton: anyway about 2287 add a printk: printk("NAND: pagesize 0x%lx blocksize 0x%lx", type->pagesize, mtd->erasesize); |
| 13:35.34 | prpplague | Crofton: i'm willing to be a dollar that the erasesize is incorrect |
| 13:35.41 | Crofton | yeah |
| 13:36.06 | Crofton | do I need a KERN_INFO in front of that? |
| 13:36.23 | *** join/#edev chouimat|work (n=dieu@209.217.106.98) |
| 13:36.46 | prpplague | Crofton: naw, if don't use anything it forces it to the main console |
| 13:36.52 | Crofton | ok |
| 13:37.26 | Crofton | and I need to add a \n? |
| 13:39.21 | prpplague | Crofton: you can if you like |
| 13:39.41 | prpplague | Crofton: usually the console routines add a \n anyway, but you might add it to make sure |
| 13:39.52 | Crofton | ok |
| 13:39.53 | prpplague | is a printk freak |
| 13:54.03 | prpplague | Crofton: i just hate a mystery and you've got my attention on this problem |
| 13:54.18 | Crofton | heh |
| 13:54.33 | Crofton | image ready, will flash when the bike race is over :) |
| 13:55.02 | chouimat|work | morning |
| 13:56.06 | prpplague | Crofton: bike race? |
| 13:56.09 | prpplague | chouimat|work: greetings |
| 13:56.18 | Crofton | Fleche Wallon |
| 13:56.25 | Crofton | curse cycling.tv |
| 13:57.04 | Crofton | 8Km to go |
| 13:58.29 | prpplague | ahh |
| 14:06.39 | *** join/#edev g1powermac (n=g1powerm@unaffiliated/g1powermac) |
| 14:15.57 | Crofton | NAND: pagesize 0x0 blocksize 0x20000 |
| 14:16.03 | Crofton | that pagesize looks suspicious |
| 14:17.51 | prpplague | Crofton: actually thats correct, that just means that the code needs to get the info from the extended id |
| 14:18.16 | prpplague | Crofton: for the calculated pages size use mtd->writesize |
| 14:18.26 | prpplague | Crofton: well the blocksize is correct |
| 14:18.32 | prpplague | Crofton: so it must be something else |
| 14:18.51 | Crofton | bummer |
| 14:19.49 | prpplague | Crofton: i'll bang away on some my nand flash to see if can come up with any ideas |
| 14:19.54 | Crofton | thanks |
| 14:51.03 | g1powermac | prpplague and GPSFan, got my zipit in today :-) |
| 14:51.23 | g1powermac | also got the cable connectors for the expansion port |
| 14:51.39 | g1powermac | and the seller even gave me a 1gb minisd card :-) |
| 14:51.41 | GPSFan | g1powermac: hey, that's great. |
| 14:51.52 | g1powermac | of course it would help if my card reader took minisd. . . |
| 14:52.03 | g1powermac | need to get one of those sd to minisd things |
| 14:53.20 | g1powermac | thought it did minisd, oh well |
| 14:54.12 | g1powermac | but good news, the connectors fit on the expansion port :-) |
| 14:54.42 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 14:54.42 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 14:54.42 | GPSFan | g1powermac: I'll be interested to see if you have an aug board (yellow wires) or an oct board (no wires). |
| 14:55.11 | g1powermac | yea, I'll crack it open prolly later today or tomorrow |
| 14:55.21 | chouimat|work | hi CosmicPenguin g1powermac GPSFan and * |
| 14:55.27 | CosmicPenguin | chouimat|work: greets |
| 14:55.32 | g1powermac | heya chouimat|work |
| 14:55.36 | GPSFan | chouimat: morning.. |
| 14:55.47 | CosmicPenguin | little email this morning |
| 14:55.55 | CosmicPenguin | I can never tell if that is good or bad |
| 14:56.01 | chouimat|work | the more I read about perl and perl code the more I want to hurt a certain Larry |
| 14:56.17 | g1powermac | haha |
| 14:56.30 | prpplague | CosmicPenguin: its bad if the only email you get is a pink slip |
| 14:57.17 | CosmicPenguin | this is why I don't accept html mail |
| 14:57.24 | prpplague | hehe |
| 14:57.29 | CosmicPenguin | if you want to give me something pink, you have to bring it to me in person |
| 15:00.06 | g1powermac | wow, the connector works quite nicely |
| 15:00.17 | prpplague | wasn't it radio shack that fired all those people via email? |
| 15:01.05 | GPSFan | g1powermac: could post a picture of that connector? |
| 15:01.24 | g1powermac | dang it, I snapped it together and now I can't get it apart |
| 15:01.31 | prpplague | man i wonder when the celf presentations will be available online |
| 15:01.39 | prpplague | i'd really like to see that jtag session |
| 15:07.28 | g1powermac | woot, ok, need a tack or pin to get it apart |
| 15:07.42 | Crofton | prpplague, http://rafb.net/p/wxLeOH31.html |
| 15:07.52 | prpplague | looks |
| 15:07.54 | Crofton | a reply to my incoherent whining on the DV list :) |
| 15:08.41 | prpplague | ahh that would make sense |
| 15:09.04 | prpplague | but i didn't think the chip part you had was MLC |
| 15:09.13 | g1powermac | fully assembled, this connector would be really tough to get apart |
| 15:09.19 | prpplague | checks the datasheet |
| 15:10.56 | prpplague | Crofton: hmm, the part you referenced appears to be SLC |
| 15:13.59 | prpplague | Crofton: don't think thats the issue |
| 15:14.09 | Crofton | something that bothers me is the controller setup does not seem to take into account the part attached |
| 15:14.29 | prpplague | Crofton: i was wondering about that myself |
| 15:14.45 | prpplague | but since i don;t have one of those boards......... |
| 15:14.57 | Crofton | well, they cost too much for me to get you one :( |
| 15:15.25 | prpplague | Crofton: how much do they run ? about $2k i would guess |
| 15:15.29 | Crofton | sounds like it is time for me to read about the controller and go through the setup process |
| 15:15.30 | Crofton | no |
| 15:15.32 | Crofton | 10K |
| 15:15.42 | prpplague | Crofton: wow |
| 15:15.51 | Crofton | there are a couple fo FPGA's (big ones) and some RF hw |
| 15:15.59 | prpplague | ahh |
| 15:16.00 | Crofton | I managed to beg the DV board |
| 15:16.17 | prpplague | i really really want to find an A8 board |
| 15:16.22 | Crofton | yeah |
| 15:16.36 | Crofton | the would be a much better solution for this task |
| 15:16.52 | Crofton | bbl |
| 15:26.49 | *** join/#edev ico2 (n=ico2@modem-3539.chimpanzee.dialup.pol.co.uk) |
| 16:00.53 | prpplague | ~seen bbradley |
| 16:00.55 | purl | prpplague: i haven't seen 'bbradley' |
| 16:01.15 | g1powermac | hey I got a question for all you buildroot using people |
| 16:01.50 | g1powermac | hmm, nm |
| 16:09.19 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 17:15.42 | *** join/#edev ico2_ (n=ico2@modem-3540.cougar.dialup.pol.co.uk) |
| 17:24.58 | prpplague | ~seen landley |
| 17:25.01 | purl | landley <n=landley@207.47.11.2.static.nextweb.net> was last seen on IRC in channel #edev, 5d 18h 59m 29s ago, saying: 'Probably http://www.oszoo.org/wiki/index.php/Category:OS_images is what you're looking for.'. |
| 17:31.24 | prpplague | hmm anyone got experience with gpio/eint based keyboard matrix? |
| 17:32.33 | prpplague | i'm wondering if you have a nonsymetrical matrix, i.e. 4rowx5col , which makes more sense, to have the eint on the 4 rows or on the 5 cols |
| 17:42.08 | prpplague | robbat2: greetings |
| 17:42.18 | prpplague | whats cookin today? |
| 17:42.44 | chouimat|work | <PROTECTED> |
| 17:43.04 | prpplague | looks |
| 17:44.20 | robbat2 | prpplague, annoyed at fetchmail in the morning, and I have a painful headache :-( |
| 17:44.34 | prpplague | joy joy |
| 17:46.13 | prpplague | chouimat|work: oh my |
| 17:46.38 | chouimat|work | prpplague: I told you it was "amazing" ;) |
| 17:47.23 | solar | prpplague: by chance can you whip up a list of the most common programs you see in use? I've started on that cross-binary repo. |
| 17:49.12 | prpplague | solar: most common in what aspect? |
| 17:49.42 | solar | prpplague: as in stuff that should be added to http://tinderbox.dev.gentoo.org/cross-pkgs/arm-softfloat-linux-uclibc/All/ |
| 17:51.11 | solar | I figure. I'd provide a repo which would allow quick deployments of whatever somebody wants to peice together. |
| 17:51.43 | prpplague | solar: yea that list is pretty accurate |
| 17:51.57 | prpplague | solar: i think the only think i don't see that i'd want is fbi |
| 17:52.02 | *** join/#edev landley (n=landley@adsl-68-93-84-18.dsl.hstntx.swbell.net) |
| 17:52.02 | *** mode/#edev [+o landley] by ChanServ |
| 17:52.24 | prpplague | landley: greetings |
| 17:52.30 | prpplague | solar: and probably a package for devmem |
| 17:52.41 | prpplague | i consider devmem pretty essential |
| 17:52.55 | solar | yeah I noticed devmem in the Root.ext2 |
| 17:53.08 | solar | was thinking that should be made into a busybox-applet |
| 17:53.15 | prpplague | solar: don't get me started |
| 17:53.23 | prpplague | solar: i've had that battle litterly for years |
| 17:53.26 | solar | on devmem as an applet? |
| 17:53.32 | prpplague | solar: indeed |
| 17:53.41 | solar | you pro for it or anti it? |
| 17:53.42 | *** join/#edev bbradley (n=bbradley@92-235-133-146.cable.ubr17.azte.blueyonder.co.uk) |
| 17:53.46 | prpplague | solar: the busybox crew have fought it every time i've brought it up |
| 17:53.51 | prpplague | bbradley: greetings |
| 17:54.03 | prpplague | solar: pro |
| 17:54.34 | prpplague | solar: their argument always that it was too specific an application for inclusion |
| 17:54.35 | bbradley | hi prpplague |
| 17:54.43 | solar | landley: unless you say otherwise. I see no reason for me not to make a quick applet and commit it to bb-svn |
| 17:54.53 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 17:54.53 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 17:55.03 | prpplague | bbradley: hey i saw where you posted that you were interested in a hammer kit but you thought the shipping was too much |
| 17:55.20 | prpplague | bbradley: is that correct? |
| 17:56.02 | prpplague | solar: i don't think landley cares too much about busybox maintence at the momment |
| 17:56.31 | solar | I think he has some mixed feelings about it. |
| 17:56.36 | prpplague | since the great GPL2 vs GPL3 war on the mailing list |
| 17:56.43 | bbradley | prpplague: yeah, when I priced it up it was about 50% of the price of the hammer kit |
| 17:56.47 | solar | he still depends upon it for his firmware linux project |
| 17:57.09 | solar | prpplague: did you ever get around to making it an applet? |
| 17:57.30 | prpplague | bbradley: were you using the UPS or FedEx? USPS is around 10 quid for the kit to the UK |
| 17:58.09 | prpplague | solar: yea landley still depends on it for now until his other app toybox is more robust |
| 17:58.23 | prpplague | solar: ages ago, i'll have to dig it up |
| 17:58.36 | landley | Hiya prpplague. |
| 17:58.45 | prpplague | landley: how was pencon |
| 17:58.52 | landley | solar: you're hanging out on #edev now? :) |
| 17:59.15 | landley | I'll happily add devmem to toybox, but I forget what it is... |
| 17:59.24 | solar | landley: seemed fitting to not reject the invite from prpplague |
| 17:59.40 | prpplague | solar: fbi and lynx are the only ones that i don;t see off hand in your list |
| 17:59.41 | solar | I'll skip on the #elinux till i have a reason to head over there :) |
| 17:59.51 | prpplague | hehe |
| 17:59.54 | landley | solar: I walked away from busybox after my last encounter with Bruce #*%&#* Perens. Denis Vlasenko's the guy you want to ping now. |
| 18:00.02 | landley | I see nothing wrong with a devmem applet, though. |
| 18:00.15 | landley | solar: #elinux is mostly dead. |
| 18:00.27 | solar | landley: I might need you to update my ssh key on uclibc.org however. |
| 18:00.29 | landley | Only reason to hang out there is to ping Tim Riker. |
| 18:00.35 | bbradley | prpplague: interesting, I can't remember which I tried before but £10 is pretty good, I'll have to think about that |
| 18:00.39 | landley | I can probably do that. |
| 18:00.51 | bbradley | prpplague: it all comes down to having enough time for yet another project :) |
| 18:00.56 | landley | prpplague: it's not a question of more robust, it's a question of more complete. |
| 18:01.01 | solar | or just give me root :p |
| 18:01.04 | landley | Toybox currently implements a dozen or so commands, busybox has 100 applets. |
| 18:01.05 | prpplague | bbradley: indeed, just wanted to make sure there wasn't a problem |
| 18:01.42 | landley | prpplague: did the flyers I handed out at CELF and penguicon have any effect? :) |
| 18:01.43 | Crofton | prpplague, there is a function nand_davinci_select_chip with no code in it ..... |
| 18:01.49 | landley | Penguicon was fun. Still sleep deprived. |
| 18:02.02 | landley | Doing one more week at the houston job to disentangle gracefully, but I should be out of here by the end of the month. |
| 18:02.04 | prpplague | Crofton: thats interesting |
| 18:02.15 | prpplague | landley: seems so |
| 18:02.23 | prpplague | landley: we've gotten a number of orders |
| 18:02.26 | Crofton | needs to find his copy of the NAND controller |
| 18:02.30 | Crofton | doces |
| 18:02.44 | prpplague | landley: leaving early or are you done? |
| 18:02.52 | landley | prpplague: I never started. |
| 18:03.01 | solar | prpplague: Any of these sound like the fbi you are talking about? http://rafb.net/p/QRvB1g53.html |
| 18:03.02 | prpplague | landley: oh? |
| 18:03.04 | landley | I've been here two months, out of a five month contract, and I have yet to be asked to write or modify a single line of code. |
| 18:03.23 | landley | At this point, it's pretty much impossible for me to actually accomplish anything in the remaining three months, and I don't want to extend it. |
| 18:03.29 | landley | The money's great, but I'm wasting my time otherwise. |
| 18:03.30 | bbradley | ok, thanks prpplague |
| 18:03.40 | prpplague | solar: yea, fbida |
| 18:04.01 | prpplague | landley: understood |
| 18:04.24 | *** join/#edev emil|k (i=emil@e179234252.adsl.alicedsl.de) |
| 18:04.25 | prpplague | bbradley: if you have problems ordering, give me a shout |
| 18:05.09 | prpplague | solar: boa is missing too |
| 18:05.49 | solar | we dumped boa from gentoo for sec reasons. |
| 18:06.04 | landley | solar: you porting gentoo embedded to the nail? :) |
| 18:06.16 | solar | busybox-httpd works pretty ok for basic needs. thttpd if you need a mini httpd daemon that can work with php. |
| 18:06.41 | solar | landley: just giving a whack at things that will x-compile. Posting those online. Waiting for a new unit to ship. |
| 18:07.09 | prpplague | solar: ahh ok, i always use thttpd, but some people prefer boa |
| 18:07.11 | solar | making a list of things I need to fix along the way also. |
| 18:07.23 | landley | Cool. |
| 18:07.37 | prpplague | solar: maybe i missed it but i don't see pppd or chat |
| 18:08.05 | landley | I've found that any serious attempt to get a distro doing something new spawns 8 zillion new tangents that keep me occupied for a month until I remember what the original goal was. |
| 18:08.44 | solar | it's nothing new for us really. My cross-compile maddness comes and goes in spurts. |
| 18:09.01 | solar | sadly there are not enough ppl that test/submit fixes for gentoo stuff. |
| 18:09.02 | prpplague | landley: new project for the hammer board, taser landley every 7 days to remind him of the original goal |
| 18:09.19 | solar | I was hoping to find some new talent at the ELF-08.. But no such luck. |
| 18:09.40 | prpplague | solar: that list looks pretty darn functional |
| 18:10.49 | landley | Which list? |
| 18:11.04 | RobotGuy | When I fix things, I don't mess with submitting patches because it's usually too political. |
| 18:11.05 | solar | prpplague: don't suppose you have an arm workhouse around there anywhere? Like an iyonix but softfloat |
| 18:11.27 | prpplague | landley: http://tinderbox.dev.gentoo.org/cross-pkgs/arm-softfloat-linux-uclibc/All/ |
| 18:12.25 | solar | thats a very basic cross build env. The target root_fs i wanted to test is here http://tinderbox.dev.gentoo.org/portage/local/misc/arm-softfloat-linux-uclibc.ext2 |
| 18:12.36 | prpplague | solar: sorry not sure i understand your question |
| 18:12.36 | solar | but I could not figure out the correct damned qemu options |
| 18:12.59 | solar | prpplague: IE.. whats the fastest arm devel board you have sitting around? |
| 18:13.22 | solar | the iyonix for example is an arm @ 1Ghz or so. |
| 18:13.23 | prpplague | solar: ahh sorry, not enough coffee today, 600mhz |
| 18:13.31 | prpplague | solar: arm9 |
| 18:14.11 | prpplague | solar: what i'd like to do is get a nice set of nail boards working with a distcc style operations |
| 18:14.23 | prpplague | solar: plug about 10 or 12 nail boards into a hub |
| 18:14.29 | solar | one of these years some vendor is going to be really good and put 4 of those on a single board and give me something with a few gigs of ram. |
| 18:15.07 | prpplague | yea i rarely do native builds for arm |
| 18:15.15 | prpplague | pretty much always cross-compile |
| 18:15.31 | landley | solar: next FWL release I'll have the distcc trick packaged up with an easy user interface. |
| 18:15.34 | landley | Possibly this weekend. |
| 18:15.41 | prpplague | landley: oh? |
| 18:15.51 | prpplague | landley: so we can use it with the nail? |
| 18:15.53 | solar | thats how it goes. But when one can. It's nice to pull from proper native builds. It sorta ensures ./configure got all the proper results. |
| 18:16.01 | landley | prpplague: yeah, I got it working a while back, but the UI sucked. |
| 18:16.09 | prpplague | landley: oh nice |
| 18:16.18 | solar | oh I can do a distcc cluster of a few hundred blades. |
| 18:16.22 | prpplague | i'll have to spend some time testing 4 or 5 nails together |
| 18:16.36 | landley | I took it apart again and never quite made the new pieces line up. It's about 30 minutes of work to finish it, but finding 30 minutes when I'm not exhausted or distracted has bit a bit hard this month. |
| 18:16.38 | solar | but linking itself takes a good bit of time. |
| 18:16.50 | landley | Speaking of which, I need to finish my darn OLS paper. This is getting embarassing. |
| 18:16.58 | landley | They've been VERY NICE about it. |
| 18:17.05 | prpplague | ~lart landley for not finishing his OLS paper |
| 18:17.05 | purl | eats landley's liver with some fava beans and a nice chianti for not finishing his OLS paper |
| 18:17.30 | landley | solar: is linking I/O bound, or CPU bound? |
| 18:17.45 | landley | I've found what takes a lot of time is the ./configure step. |
| 18:18.24 | prpplague | landley: heard anything from morton with regards to the nail board? |
| 18:18.29 | landley | solar: qemu does ./configure natively. Unfortunately, that step doesn't benefit from distcc. :( |
| 18:18.39 | landley | prpplague: I need to email him proper documentation. That's half done. |
| 18:18.57 | prpplague | landley: okie dokie |
| 18:18.58 | landley | Going to Penguicon right after CELF was... "interesting". |
| 18:18.59 | solar | landley: You get qemu booting the default image on the wiki page? |
| 18:19.04 | prpplague | landley: let me know if i can help any |
| 18:19.10 | landley | There's a default image on the wiki page? |
| 18:19.16 | solar | yes and kernel |
| 18:19.29 | prpplague | landley: yea, just the generic kernel and rootfs |
| 18:19.30 | landley | Ah. I built my own kernel for qemu, but the same root filesystem should work. |
| 18:19.57 | prpplague | landley: don't suppose you could post your qemu kernel |
| 18:20.23 | prpplague | landley: what did you have to change/enable to get it to boot from qemu? |
| 18:20.37 | landley | prpplague: please put a link from the tincantools nail page to the edev nail page. |
| 18:20.43 | solar | I want to see those qemu options :) |
| 18:21.16 | prpplague | landley: suppose to be one there |
| 18:21.16 | prpplague | looks |
| 18:21.41 | g1powermac | anyone have an issue with buildroot where it seems to compile everything ok but then there's no toolchain in the spot where its suppose to be? |
| 18:21.52 | prpplague | darn looks like it got removed, i'll get that up there again |
| 18:22.03 | g1powermac | seems like it didn't create any binaries |
| 18:22.11 | prpplague | g1powermac: not i |
| 18:22.27 | g1powermac | prpplague, k, its the strangest thing |
| 18:22.45 | prpplague | landley: hammer support is queued for 2.6.25-rc1 |
| 18:22.55 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 18:22.55 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 18:22.59 | landley | prpplague: woot! |
| 18:23.14 | g1powermac | go to the staging_dir to take a look at the toolchain and not a one binary in it |
| 18:23.18 | landley | lemme copy my linux-miniconfig up to my website... |
| 18:24.21 | landley | Ok, try http://landley.net/miniconfig-linux |
| 18:24.36 | landley | To use that, "make allnoconfig ARCH=arm KCONFIG_ALLCONFIG=miniconfig-linux" |
| 18:24.52 | landley | Then "make ARCH=arm menuconfig" to snoop around in it, if you like. Or just read the file. :) |
| 18:25.40 | landley | I run it with: qemu-system-arm -M versatilepb -nographic -no-reboot -hda image-armv4l.ext2 -kernel zImage-armv4l -append "blah blah blah" |
| 18:26.29 | prpplague | landley: http://tinderbox.dev.gentoo.org/cross-pkgs/arm-softfloat-linux-uclibc/All/ |
| 18:26.32 | landley | The blah blah part should include "rw init=/bin/sh panic=1 root=/dev/sda console=ttyAMA0" |
| 18:26.57 | landley | Oh, and "-net nic,model=rtl8139 -net user" on the end helps too, if you want ethernet. |
| 18:27.17 | landley | But that's not in the append, that's just more qemu options. :) |
| 18:27.37 | landley | prpplague: looks useful. |
| 18:27.54 | prpplague | landley: oops sorry wrong url |
| 18:28.04 | landley | Got a lot of repeats, though. |
| 18:28.05 | prpplague | landley: http://ftp.arm.linux.org.uk/pub/armlinux/source/kernel-patches/git-cur/arm:for-linus.diff |
| 18:29.28 | landley | prpplague: coolness. |
| 18:29.31 | landley | Muchly coolness. |
| 18:30.47 | prpplague | landley: ahh so you are running your qemu as a versatile and not as a 2410 device |
| 18:31.01 | landley | prpplague: it runs the same root filesystem. |
| 18:31.12 | landley | I got that working a year ago and mostly haven't touched it since. |
| 18:31.25 | landley | Although I'll need to when I get around to poking at flash and jffs2 for qemu... |
| 18:31.42 | landley | Mostly I'm focusing on making userspace work on different targets. |
| 18:32.01 | landley | The kernel varies so much per board that what qemu runs and what real hardware runs are unlikely to be on speaking terms most of the time anyway. |
| 18:32.16 | solar | yeah I attempted to use some of the options from your firmware project with the Root.ext2 and the Zimage. No luck. |
| 18:32.43 | landley | What exactly did you try? |
| 18:33.25 | landley | prpplague: keep in mind that circa 2006 arm support was brand new and there weren't a whole lot of boards to choose from. |
| 18:33.36 | landley | That was the only one I could attach a block device to. |
| 18:34.18 | solar | landley: dunno right now it's out of my scroll buffer from this chan. |
| 18:34.31 | landley | shrugs. |
| 18:34.42 | landley | I'm happy to answer questions, but all I can tell you is the answers _I_ found. |
| 18:34.49 | solar | Looking on the elinux site now for that default kernel image = rootfs. |
| 18:34.50 | landley | By trial and error. By poking it with a stick. |
| 18:35.06 | solar | s/=/+ |
| 18:35.31 | landley | solar: in theory you can use their rootfs image with my kernel .config under qemu. |
| 18:35.52 | landley | I haven't tried them, but their kernel with my rootfs worked on actual nail hardware. |
| 18:36.28 | solar | in theory we should be able to use the rootfs + Zimage they have on the site. |
| 18:36.37 | prpplague | solar: i'm gonna install qemu and do some tests |
| 18:36.58 | prpplague | solar: i suspect that the qemu needs to be configured specifically for the s3c24xx |
| 18:38.09 | prpplague | solar: i have a design review meeting at 2 and a conference call at 3:30, but i'll be looking into getting it working |
| 18:38.10 | solar | this is the rootfs I wanted to test. (more or less a rebuild of what I think is the default firmware on the hammer/nail setup) http://tinderbox.dev.gentoo.org/portage/local/misc/arm-softfloat-linux-uclibc.ext2 |
| 18:38.14 | RobotGuy | prpplague: What does it take to gain commit access for buildroot? |
| 18:38.29 | prpplague | RobotGuy: not much these days |
| 18:38.41 | prpplague | solar: grabbing a copy |
| 18:38.43 | RobotGuy | prpplague: Would it be workable for me to gain commit access? |
| 18:39.15 | landley | RobotGuy: ask on #uclibc, not here. |
| 18:39.15 | prpplague | RobotGuy: i'm sure you could, just gotta follow the procedures |
| 18:40.02 | landley | RobotGuy: http://busybox.net/developer.html |
| 18:40.10 | landley | The access is the same for uclibc/busybox/buildroot. |
| 18:40.15 | RobotGuy | looks |
| 18:40.16 | RobotGuy | OK |
| 18:40.37 | landley | P.S. The server is named "morris". That's why I wrote that topic heading. :P |
| 18:42.40 | prpplague | solar: booting that rootfs now |
| 18:43.06 | landley | prpplague: same here. |
| 18:45.09 | landley | Well, it booted. |
| 18:45.21 | landley | It said "/etc/mdev.conf: no such file or directory" an awful lot, but it booted and gave me a command shell. |
| 18:45.37 | landley | Using my kernel and that rootfs.ext2 |
| 18:46.51 | prpplague | boot here too |
| 18:46.59 | prpplague | bootted here too on real hardware |
| 18:47.20 | prpplague | solar: from what i'm seeing there needs to be a config on the qemu side for the s3c24xx stuff |
| 18:47.29 | prpplague | solar: i'll dig on that today and have something shortly |
| 18:47.34 | chouimat|work | ???? |
| 18:47.45 | prpplague | chouimat|work: what? |
| 18:47.58 | *** join/#edev MothaWork (n=monmotha@mail.forwardengineering.com) |
| 18:48.05 | prpplague | MothaWork: hey hey |
| 18:48.09 | MothaWork | hello |
| 18:48.21 | prpplague | MothaWork: question you ever do a keyboard matrix using gpios? |
| 18:48.27 | MothaWork | I have my laptop at work, now, so I can do this without a painful forward of xchat to Xming on Vista |
| 18:48.34 | chouimat|work | prpplague: nm |
| 18:48.37 | MothaWork | I've done a 16-position keyboard with GPIOs, yes |
| 18:48.40 | MothaWork | 4x4 matrix |
| 18:48.48 | prpplague | MothaWork: ahh symetrical matrix |
| 18:48.55 | MothaWork | yes, but no reason it has to be |
| 18:49.01 | landley | prpplague: alas, adding new board support to qemu involves a new C file, not a config file. |
| 18:49.03 | chouimat|work | prpplague: I was experimenting with unicode chars :) |
| 18:49.08 | landley | Probably not hard for somebody who knows what they're doing, but I don't. |
| 18:49.31 | prpplague | MothaWork: well my question is if you have a nonsymetrical matrix, i.e. a 4x5, is it better to interrupt on the 4 or interrupt on the 5 |
| 18:49.46 | prpplague | landley: well qemu is suppose to support the s3c24xx |
| 18:49.55 | prpplague | landley: so it should be just an arg somewhere |
| 18:50.10 | *** mode/#edev [+o prpplague] by ChanServ |
| 18:50.13 | MothaWork | I didn't use interrupts, just polling |
| 18:50.28 | prpplague | ahh |
| 18:50.31 | MothaWork | I was doing a system where I aws actually tri-stating lines in order to allow multiple presses |
| 18:50.36 | MothaWork | that doesn't get along so well with interrupts |
| 18:50.38 | solar | prpplague: any custom patches required for kerels .24 or .25 ? |
| 18:50.46 | prpplague | gotta run to a meeting bbiab |
| 18:50.51 | solar | n/m cya |
| 18:51.04 | MothaWork | however, if you have pin-change interrupts for all of them, using the larger side woudl work better since you could more easily see multiple presses, I'd think |
| 18:51.08 | prpplague | solar: yea for basic support you can use the patch on the arm patch system |
| 18:51.12 | prpplague | solar: let me get you the url |
| 18:51.25 | solar | it can wait. I've got lots of things to work on as well |
| 18:51.59 | prpplague | http://www.arm.linux.org.uk/developer/patches/viewpatch.php?id=4963/2 |
| 18:52.03 | prpplague | bbiab |
| 18:52.10 | agaffney | solar! |
| 18:52.14 | agaffney | where'd you come from? |
| 18:52.46 | landley | agaffney: california, apparently. |
| 18:53.01 | agaffney | pfft |
| 18:53.12 | landley | He hung out with me at CELF last week. I gave him one of the three nail boards prpplague's boss sent me, and he's porting gentoo embedded to it. |
| 18:53.22 | agaffney | ah |
| 18:53.25 | solar | was born in DC. Lived in MD for ~20 years. moved to OR for 8. then bounced down to GA for 4 and have been in CA for almost 2 |
| 18:53.27 | agaffney | silly solar |
| 18:53.34 | agaffney | do you really want to do *more* with gentoo? :P |
| 18:54.01 | solar | embedded/hardened/security/infra yes. General #-dev I'll skip. |
| 18:54.27 | agaffney | heh |
| 18:55.04 | solar | poof and there went the rest of my day. Gotta build 30 blades |
| 18:55.09 | ds2 | hmmm |
| 18:56.41 | landley | solar: how does gentoo/portage interact with build clusters? |
| 18:57.20 | chouimat|work | landley: I have no problem using icecream with it :) |
| 18:57.29 | landley | I know that you have to build some things in a certain order, but I'm wondering how much of that can be distributed across ultiple machines (not via distcc but "this package builds here, this package build runs over there." |
| 18:57.39 | landley | chouimat: icecream? |
| 18:59.12 | chouimat|work | landley: http://en.opensuse.org/Icecream ... and oe has a nice patched icecc-create-env.sh http://oe.linuxtogo.org/wiki/IceCC |
| 18:59.35 | solar | landley: it's framework is not like that now. You could parse an existing deptree and send jobs to multi-boxes and have all those boxes pushing back to a single master host if you really wanted. |
| 18:59.58 | solar | normally it's not a problem. Till you get to these tiny arches like arm where you want to speed things up a little |
| 19:00.03 | landley | solar: the reason I ask is the qemu+distcc trick works fine for actual compilation, but not for ./configure. |
| 19:00.17 | solar | I think more time is spent fighting with things that don't x-compile however vs build time itself |
| 19:00.20 | landley | But if you could run multiple qemu instances, you could have one sending stuff out to distcc while the other ran its ./configure... |
| 19:00.56 | solar | what was that other project that emulated the binfmt_elf but on a per arch basis. |
| 19:01.41 | solar | http://www.scratchbox.org/ that one. In theory if you plucked off just the kernel module part. It should be much faster than qemu |
| 19:11.46 | *** join/#edev landley_ (n=landley@adsl-68-93-84-18.dsl.hstntx.swbell.net) |
| 19:35.57 | RobotGuy | I have not been able to cross compile any of the windowing stuff - X, etc. |
| 19:36.44 | ce_geek | :-/ |
| 19:36.52 | RobotGuy | ce_geek: Hi |
| 19:37.40 | ce_geek | I just cross-compiled digitemp, just for fun. |
| 19:37.48 | ce_geek | I should get off my arse and hook up the sensor now to see if it works :-P |
| 19:38.00 | RobotGuy | wants to cross compile the world |
| 19:38.36 | RobotGuy | I need to cross compile the Phidgets library. |
| 19:39.13 | ce_geek | I noticed that for digitemp on a WRT device they added the following to CFLAGS: CFLAGS = -I/tmp/WRT54G_3_01_3_0922/release/src/linux/linux |
| 19:39.39 | ce_geek | Not sure why they had to include that. I didn't add any stuff for the hammer board and it seems to have compiled just fine. |
| 19:39.59 | ce_geek | Whether or not it's going to work may be a different story |
| 19:41.13 | *** join/#edev dijenerate (n=dijenera@72.22.128.157) |
| 19:44.36 | ce_geek | prepares to solder for the first time in a WHILE |
| 19:45.38 | ce_geek | Long overdue to wipe the dust of my WESD51 |
| 19:52.20 | RobotGuy | ce_geek: I need to reconnect the power and grounds to my compass and BT module too. |
| 19:55.13 | landley_ | Hey, it logged me back on. |
| 19:55.59 | landley_ | RobotGuy: build natively under emulation. |
| 19:56.06 | landley_ | The world does not cross compile, and most of it never will. |
| 19:58.16 | RobotGuy | landley_: emulation? |
| 20:06.56 | landley_ | RobotGuy: Yup, emulation. |
| 20:06.59 | *** mode/#edev [+o landley] by ChanServ |
| 20:07.11 | landley | Basically, my Firmware Linux project has three stages: |
| 20:07.16 | landley | The first builds a cross compiler for the target. |
| 20:07.43 | landley | The second uses that cross compiler to build a native development environment, I.E. a root filesystem you can boot into which contains development tools, including a complete native compiler toolchain. |
| 20:07.56 | landley | The third stage packages that up into a form qemu can boot and run. |
| 20:08.07 | landley | Then what you do is boot into qemu and build in there. Native compiling, under emulation. |
| 20:08.41 | RobotGuy | landley: OH! So you can 'boot' into a native development environment for Hammer and build native code? |
| 20:08.42 | landley | This is about 1/20th the speed of cross compiling, but doesn't suffer from the random pervasive breakage of cross compiling. Everything builds easily if you just leave it alone to do its thing for long enough. |
| 20:08.50 | landley | RobotGuy: yup. |
| 20:09.04 | RobotGuy | Hmmmmmmmmmm. |
| 20:09.22 | landley | Technically it's a slightly different armv4l variant I told qemu to emulate, one with a scsi drive and network card attached so I can have a big hard drive and network access in the development environment. |
| 20:09.31 | landley | But it runs the same binaries, and boots the exact same root filesystem. |
| 20:09.38 | RobotGuy | impatiently awaits delivery of his replacement Hammer Board. |
| 20:09.48 | landley | I've booted a subset of this root filesystem on hammer, only reason I haven't booted the whole thing is it's too big to fit on hammer. |
| 20:10.09 | landley | (Actually, it would fit in a squashfs on hammer. It compresses down to about 12 megs. But I'd need space to extract and compile source code in.) |
| 20:10.26 | landley | Then once I've got it working, I accelerate it with the distcc trick. |
| 20:10.31 | RobotGuy | landley: Thumb Drive or MMC/SD card. |
| 20:10.34 | landley | The distcc trick uses the cross compiler I built earlier. |
| 20:10.39 | landley | RobotGuy: in theory I could do that, yeah. |
| 20:10.47 | landley | In practice, trying to run gcc in 32 megs of ram, it gets a little unhappy. |
| 20:10.56 | landley | I'd need to set up swap on the thumb drive, and that's all sorts of no fun. |
| 20:11.22 | landley | Plus, qemu emulating arm on my 2ghz Core 2 duo runs about as fast as the 200mhz hammer cpu anyway. |
| 20:11.30 | landley | Same ballpark. |
| 20:11.47 | RobotGuy | landley: Yeah, I know, which is why I hope SledgeHammer gets a lot more RAM and Flash. |
| 20:11.50 | landley | If I want 500 megs of ram in the qemu image, I just say "-m 500m". The laptop has 2 gigs. |
| 20:12.07 | landley | The distcc trick speeds up compiles by a factor of 7. |
| 20:12.22 | landley | Not quite back to full cross compile speeds, but back into the general ballpark. |
| 20:12.24 | landley | here's how that works: |
| 20:12.30 | RobotGuy | Many nail boards in a distributed network environment. |
| 20:12.42 | landley | The parts of cross compiling that normally screw up aren't actually the _compile_ parts. |
| 20:13.17 | landley | The big one is ./configure, because it's designed to ask questions about the host and use the answers for the target, which is fundamentally wrong for cross compiling. |
| 20:13.33 | landley | But make can also screw up. It uses environment variables and such that are for the host, not the target. |
| 20:13.57 | landley | The compiler's #include paths fall back to the host's /usr/include if they don't find what they're looking for, so if your target hasn't got zlib but your host does, it'll leak. |
| 20:14.05 | landley | The library paths do the same thing during the linking stage. |
| 20:14.34 | landley | And then during the build, it can build and run various native tools that get things wrong too. |
| 20:14.53 | landley | RobotGuy: do you know how distcc works? |
| 20:15.26 | RobotGuy | landley: No, but I know it has to do with distributed compiling. |
| 20:15.44 | solar | heh agaffney wrote this doc on the topic http://www.gentoo.org/doc/en/cross-compiling-distcc.xml |
| 20:15.46 | landley | Distcc acts as a wrapper around gcc. You stick it in the front of the $PATH. |
| 20:16.23 | landley | solar: I've been telling people about this technique for two years now. It was my idea, but other people got it actually _working_ before I did. :) |
| 20:16.30 | landley | (See "the todo list of doom"...) |
| 20:16.58 | landley | RobotGuy: when you invoke the discc wrapper, it parses the gcc command line you fed it, and runs the local copy of gcc with "-E" to preprocess the source file(s). |
| 20:17.16 | landley | This resolves all the #include paths. |
| 20:17.43 | landley | Then it sends the preprocessed source (with all the #include files inserted and all the macros resolved) across the network to one of the distcc daemons. |
| 20:18.06 | landley | The daemon on the remote system invokes its local copy of distcc to turn the preprocessed C file into a binary .o file. |
| 20:18.23 | landley | It then sends the .o file back through the network to the distcc wrapper, which writes it to the local filesystem as if it was output from gcc. |
| 20:18.49 | landley | When gcc is told to link stuff to produce executables, the distcc wrapper calls the local copy of gcc to do it. |
| 20:19.13 | landley | The reason they do this is that the different compilers on the different machines might have slightly different headers or libraries, so they make sure that the #includes and the linking are all done on the same machine. |
| 20:19.32 | landley | But the .c->.o conversion can be done anywhere, it's very standardized and a little compiler version skew won't hurt it. |
| 20:19.50 | landley | Now imagine that the remote distcc daemons are set up to call the _cross_ compiler. |
| 20:20.29 | landley | And we native compile inside qemu with a distcc daemon running outside of qemu, hooked up to the cross compiler on the host machine. |
| 20:20.50 | landley | Inside qemu, the ./configure stage happens natively, and thus gets all the right answers (very, very slowly). |
| 20:21.11 | landley | Then make runs natively, and again works correctly because there's only one context so it can't get confused. |
| 20:21.36 | landley | Make calls gcc, and the distcc wrapper calls gcc -E locally to handle all the #includes, but this isn't at all CPU intensive (in fact it's I/O bound). |
| 20:21.53 | landley | Then the distcc wrapper inside qemu dials out through the virtual network to talk to the distcc daemon running on the host machine. |
| 20:22.16 | landley | (The way to make that work is to run the distcc daemon on 127.0.0.1 on the host, and tell qemu to talk to 10.0.2.2. |
| 20:22.19 | landley | checks that number... |
| 20:22.45 | solar | landley: a little compiler skew can hurt the end result quite a bit if you are intermixing gcc-3.x and gcc-4.x cross-distcc-hosts. It's really best to make sure the exact same cross-toolchain is installed everywhere. |
| 20:22.53 | landley | Yup, 10.0.2.2. |
| 20:22.59 | landley | solar: only if you're using C++, I think. |
| 20:23.16 | landley | Modulo the STUPID STUPID STUPID exception handling based setjmp/longjmp crap. |
| 20:23.22 | landley | (C is not C++. Grrr.) |
| 20:23.46 | landley | RobotGuy: 10.0.2.2 is a magic address within qemu's "-net user" network that says "connect to loopback on the host". |
| 20:23.49 | solar | not pleased about lzma being pushed on the world for basic things. |
| 20:24.10 | landley | So whatever port you run the distcc daemon on on the host, use that same port on 10.0.2.2 inside qemu. |
| 20:24.20 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 20:24.20 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 20:24.25 | landley | solar: lzma would suck much less if there was anything or "file" to grab onto to identify it. |
| 20:24.35 | landley | That was my main qualm about merging it into busybox. |
| 20:25.00 | landley | And unfortunately, I didn't get that cleared up until _after_ it was merged. Sigh. |
| 20:25.08 | solar | mine is it's written in c++ and is not cross-compile aware at all. And now stuff like libpng is using it by default. |
| 20:25.31 | landley | solar: the lzma in busybox isn't written in c++. Unless I'm thinking of a different compression thingy... |
| 20:26.53 | landley | solar: look at busybox/archival/libunarchive/decompress_unlzma.c |
| 20:26.59 | landley | At least that's where 1.2.2 had it... |
| 20:27.07 | solar | hrmm neat. |
| 20:27.22 | solar | goes hunting for vapier then |
| 20:27.40 | landley | It contains a certain amount of The Suck, but I did have _some_ standards for what I was willing to merge. :) |
| 20:28.23 | landley | RobotGuy: anyway, the distcc daemon on the host compiles .c->.o and then the wrapper writes the .o file into the filesystem on qemu's virtual block device. |
| 20:28.33 | landley | And then the linking stage happens inside qemu, 100% natively. |
| 20:29.07 | RobotGuy | Wow, that's pretty cool. I'm more interested in seeing what a bunch of Hammers can do in a network. |
| 20:29.08 | landley | And _that_ is the distcc trick. Speeds up compiles within qemu by a factor of 7 in my benchmarks, but moving the heavy lifting of compilation outside of the emulator, but keeping all the tricky bits inside of it so they can't get screwed up by cross compiling. |
| 20:29.28 | landley | RobotGuy: in theory, you could use a hammer as the central node to run this, and have it talk to x86 machines with the cross compiler. |
| 20:29.39 | RobotGuy | Now that I know I can WiFi with hammer.. |
| 20:29.53 | landley | In practice, I'm not sure how happy it'll be with the 32 megs of memory limitation (gcc is a _pig_), and I don't know if it would be any faster than qemu. |
| 20:30.10 | landley | I do know that the ./configure stage running natively inside the emulator becomes the bottleneck. It's not helped at all by the distcc trick. :( |
| 20:30.28 | RobotGuy | landley: I wasn't necessarily talking about compiling. I can find lots of uses for a distributed network. |
| 20:30.33 | landley | That's why I was asking Solar earlier about spreading different package compiles out across a cluster. |
| 20:30.39 | landley | RobotGuy; oh sure. |
| 20:31.06 | RobotGuy | landley: I'm thinking a master control robot with a swarm of little sensor robots. |
| 20:31.11 | landley | In theory you could cluster up to three nail boards via their USB interfaces before you're drawing too much current from the root USB socket. |
| 20:31.22 | landley | Or you could wire up your own sockets for the hammer, being a hardware guy. :) |
| 20:31.28 | solar | I've got a few spare clusters of 150+ Dual/Quad Core's 2-16G Ram if you want to build a framework |
| 20:31.55 | landley | solar: my net connection cut out earlier when I was poking you about spreading gentoo package compiles across different machines. |
| 20:32.06 | RobotGuy | solar: I could have lots of fun with power like that. |
| 20:32.19 | landley | I'd like to run multiple qemu instances (possibly on different machines, possibly on the same SMP box), to make the ./configure stage suck slightly less. |
| 20:32.48 | landley | The ability to keep multiple distcc daemons busy isn't much help if you then have to stop for 7 minutes to let the next ./configure stage rumble by. |
| 20:33.08 | solar | landley: i can put you on a Dual Dual Core LS20 with 16g of ram if you want to hack around for a little to try and speed things up. |
| 20:33.09 | landley | Some of the FSF ./configure stages are just _insane_. |
| 20:33.32 | landley | solar: cool, but my laptop is a Core 2 duo. Dual 2ghz processors with 2 gigs of ram. |
| 20:33.41 | landley | not quite as poweful, but enough for experimentation. |
| 20:33.55 | landley | I'm just not nearly familiar enough with the gentoo build process to perform major surgery on it. |
| 20:34.17 | solar | I should grab some lunch. |
| 20:34.19 | landley | My next FWL release should have the distcc trick set up with an easy user interface. |
| 20:34.25 | solar | landley: yeah I'm here to fix that. :) |
| 20:34.33 | landley | just run "./run-with-distcc.sh" instead of "./run-emulator.sh" and it'll set itself up. |
| 20:35.24 | landley | (./run-with-distcc.sh calls ./run-with-home.sh (to set up a second ext2 image on /dev/hdb mounted on /home, so there's plenty of writeable space), which calls run-emulator.sh) |
| 20:35.27 | solar | infact. I want to know more about your tinycc when I get back. Need to know if it can build a small project of mine which could be used to quickly bootstrap any ststem |
| 20:35.33 | landley | That's why there's all those crazy environment variables. |
| 20:35.41 | landley | solar: sure. |
| 20:35.45 | landley | If it doesn't work, I'll fix it. |
| 20:35.57 | landley | Can't currently promise anything about how _long_ fixing it will take, though. |
| 20:36.05 | landley | My schedule's still up in the air this week... |
| 20:36.47 | solar | cvs -d :pserver:anonymous@anoncvs.gentoo.org:/var/cvsroot co gentoo-projects/portage-utils ; cd portage-utils ; make CC=tinycc ? |
| 20:37.09 | solar | if that works. Then we have next best thing to gold (as long as you can build busybox also) |
| 20:37.11 | landley | Possibly? |
| 20:37.23 | landley | Alas, right now it can't sanely build busybox. |
| 20:37.33 | landley | It lacks dead code elimination. |
| 20:37.36 | landley | That's on my todo list. |
| 20:37.44 | solar | need lunch. bbiaf. |
| 20:37.45 | landley | Busybox relies on dead code elimination, unless you're building "make allyesconfig". |
| 20:37.47 | landley | ok. |
| 20:38.51 | landley | It's cd gentoo-projects/portage-utils actually. |
| 20:40.58 | landley | solar: nope, right off the top it needs to be taught about -std=gnu99 and -dM |
| 20:41.08 | landley | Possibly the first is a NOP. |
| 20:50.50 | prpplague | returns from conference call |
| 20:53.56 | prpplague | diggs into qemu |
| 20:59.03 | RobotGuy | I should go back to Gentoo one of these days. |
| 21:02.58 | Dr{Who} | landley.. have you seen this http://www.stud.uni-hannover.de/~daniel/tcc/ |
| 21:03.00 | RobotGuy | Maybe when I get a case and psu for my third PC |
| 21:03.56 | Dr{Who} | btw.. i have yet to successfully build tinycc on arm that will run. I also still cant get gdb to work but im almost setup on doing gdbserver on the hammer. |
| 21:04.30 | prpplague | hmm openmoko guys have s3c2410 working under qemu |
| 21:04.46 | prpplague | but i dont see where its added to main line qemu |
| 21:04.57 | landley | DrWho: yeah, he's the guy who contributed the arm port. |
| 21:05.14 | landley | I merged his code into my fork back when it was just a fork and not a separate project. Circa 2006, I think? |
| 21:05.33 | landley | DrWho: it runs under qemu but not on real arm hardware. |
| 21:05.40 | landley | It has alignment faults, which qemu doesn't catch. |
| 21:05.43 | landley | It's on my todo list. |
| 21:05.54 | landley | I think this is my last week at this job, so next week I should be able to catch up on stuff. |
| 21:06.47 | prpplague | landley: what did they say when you told them you were quitting? |
| 21:06.49 | Dr{Who} | ok i am trying to do what i can but this is very very advanced stuff and i have minimal time into the code base so far.. If i had a debugger going i could do more so thats still a big issue to solve for me. |
| 21:08.50 | landley | prpplague: I'm still waiting to talk to a manager to whom that is relevant. |
| 21:08.58 | landley | My team lead got reassigned. i dont' report to him anymore. |
| 21:09.02 | landley | I currently have _nothing_ to do. |
| 21:09.03 | prpplague | landley: lovely |
| 21:09.19 | landley | The other guy in my office/other contractor from this consulting company quit on friday. |
| 21:09.21 | prpplague | they gonna be pissed? |
| 21:09.31 | landley | No idea. Don't really care either. |
| 21:09.37 | Dr{Who} | if i am correct this lib provides for the machine spacific handeling of large byte operations and floating point etc etc stuff that has to be machine specific as a hmmmm API i guess so that the C compiler can just link to this lib that does the machine specific actual work. |
| 21:09.40 | landley | What are they going to do, fire me? |
| 21:09.56 | landley | DrWho: it's libgcc.a |
| 21:10.15 | landley | It's tinycc's implementation of libgcc.a, and you can substitute gcc's libgcc.a for it if you haven't got the right one lying around. |
| 21:10.21 | Dr{Who} | unfortunatly i have never studied libgcc.a and what it does. |
| 21:10.24 | landley | You can even suck it out of the cross compiler. |
| 21:10.52 | ds2 | what is the purpose of libgcc? |
| 21:10.57 | landley | It contains stuff like soft float, stack unwinding code for the stupid C++ style setjmp/longjmp, 64 bit math functions on 32 bit processors, and so on. |
| 21:11.14 | ce_geek | prpplague: Did you end up with any conclusions about the level shifting and such? |
| 21:11.15 | landley | The shared version of it (libgcc_s.so) exists to screw up C libraries other than glibc. |
| 21:11.19 | ds2 | a helper library? |
| 21:11.21 | prpplague | lovely qemu only compiles with gcc 3 |
| 21:11.25 | *** join/#edev nhoman (n=homann@79.97.106.107) |
| 21:11.27 | landley | prpplague: yeah. |
| 21:11.29 | Dr{Who} | ok thats what i figured. |
| 21:11.36 | landley | They're working on that. |
| 21:11.45 | landley | I need to catch http://landley.net/qemu up to the present. |
| 21:11.47 | ce_geek | prpplague: Just curious. I have been working on mechanical stuff this afternoon, getting a nice ribbon cable to a bread board, getting things mounted on plexi glass, etc. Tonight I'm digging into the schematics. |
| 21:11.53 | prpplague | ce_geek: sorry it must have slipped my feable mind, what level shifting were we discussing? |
| 21:12.00 | ce_geek | prpplague: I got digitemp compiled and apparently running on the hammer |
| 21:12.06 | ce_geek | prpplague; For the DS1820 temp sensor. |
| 21:12.10 | prpplague | ce_geek: oh right |
| 21:12.25 | prpplague | ce_geek: yea, you'll need to connect it via the ttl level not the rs-232 level |
| 21:12.29 | ce_geek | prpplague; Seems like I should be all set, all the hammer pins should be 3.3V right? |
| 21:12.50 | prpplague | ce_geek: you can access it via the rxd2 and txd2 thru hold pads in the protoarea |
| 21:12.58 | ce_geek | prpplague: J6 provides RS232 levels for projects that require it, yes? |
| 21:13.08 | prpplague | ce_geek: that is correct |
| 21:13.39 | prpplague | ce_geek: make sure that JP2 pins 5/6 and 7/8 are not jumpered |
| 21:14.06 | prpplague | landley: bummer guess that means i need to install gcc3 |
| 21:14.10 | ce_geek | prpplague: Okay. Glad I'm understanding that. I'll have a scope tonight to double check stuff. I think I'll have a chance to hook it up tonight and see if it works now that I've got a nice bread board interfaced with those pins. I'll take a picture when I bring my camera back to the lab with me tonight, it's kinda neat. Set up for people to mess with stuff like in a lab. |
| 21:14.20 | ce_geek | An afternoon 3-hour academic lab, that is. |
| 21:14.22 | prpplague | ce_geek: good deal |
| 21:14.42 | prpplague | landley: hmm, guess i need to create a qemu patch for the hammer |
| 21:14.48 | landley | prpplague: or grab the prebuilt 0.9.1 binary. |
| 21:14.48 | prpplague | solar: you still around? |
| 21:14.51 | ce_geek | prpplague: Ooh glad you mentioned JP22, not sure I'd have thought to care about that. Not sure what it does, I'll read up tonight. |
| 21:14.58 | prpplague | landley: i need to hack code |
| 21:15.01 | ce_geek | *JP2 |
| 21:15.12 | landley | One of the criteria for a 1.0.0 release is builds with other compilers. |
| 21:15.16 | landley | prpplague: ah, in that case, yes you do. |
| 21:15.19 | prpplague | ce_geek: JP2 enables the rs-232 levels for the uart2 |
| 21:15.29 | ce_geek | prpplague: But only for the outputs on J6? |
| 21:15.34 | landley | For the moment, anyway. There's a new code generator being put in that breaks the gcc 3.x dependency, but it's not finished yet. |
| 21:15.36 | prpplague | ce_geek: correct |
| 21:15.53 | ce_geek | prpplague: So the status of the JP2 jumpers shouldn't matter if I'm using the thru-hole stuff on the proto area, right? |
| 21:15.56 | *** join/#edev dfarnsworth (n=dfarnswo@65.200.49.156) |
| 21:16.07 | ce_geek | prpplague: (Can I just ignore JP2 then?) |
| 21:16.32 | prpplague | ce_geek: techinaclly no, but its always good to disable any IC's that might create issues |
| 21:16.41 | ce_geek | prpplague: I agree. Good advice. |
| 21:17.21 | ce_geek | prpplague: I'll be back on later tonight. Time for dinner. TTYL. |
| 21:17.27 | prpplague | okie dokie |
| 21:23.09 | prpplague | landley: yea doesn't look to hard to add hammer support to qemu |
| 21:42.22 | prpplague | solar: ping |
| 22:07.00 | *** join/#edev cajaka (n=none@cpe-24-58-193-200.twcny.res.rr.com) |
| 22:12.08 | *** join/#edev MonMotha (n=monmotha@c-69-245-153-193.hsd1.in.comcast.net) |
| 22:12.49 | CosmicPenguin | hmm - does anybody know if there are any good debug applications that can easily figure out the mapping between a processe's virtual memory space and physical memory? |
| 22:13.07 | CosmicPenguin | i.e. some gadget that knows how to do the math without thinking too much? |
| 22:14.33 | Crofton | what are you trying to do? |
| 22:14.41 | Crofton | exmap? |
| 22:15.08 | CosmicPenguin | oh - sexy |
| 22:19.15 | solar | prpplague: sorry be back in a few. I'm stuck in my own meetings |
| 22:21.03 | Crofton | exmap made my first trip to FOSDEM worthwhile |
| 22:26.32 | *** join/#edev prpplague (n=dave123_@ppp-70-244-166-139.dsl.rcsntx.swbell.net) |
| 22:27.58 | prpplague | ho ho hum |
| 22:28.41 | prpplague | Crofton you find any more out about your nand flash issue? |
| 22:34.55 | prpplague | Crofton btw, it took me a minute or two to figure out that it was you that had made the linkedin request |
| 22:35.26 | Crofton | heh |
| 22:35.31 | Crofton | sorry about that |
| 22:35.52 | Crofton | no, I received a beagle today and they will dominate my time now |
| 22:36.19 | Crofton | beagle + USRP will get me more press at this point |
| 22:36.27 | Crofton | since it is a low cost solution |
| 22:36.58 | Crofton | the Lyrtech guys are really nice and bought me dinner once, so when I have time I work on that |
| 22:37.23 | Crofton | I need someone to step up and through money at the problem :) |
| 22:37.25 | prpplague | USRP? |
| 22:37.31 | Crofton | ~usrp |
| 22:37.31 | purl | methinks usrp is a Universal Software Radio Peripheral or http://comsec.com/wiki?UniversalSoftwareRadioPeripheral |
| 22:37.36 | prpplague | oh duh |
| 22:37.45 | prpplague | Crofton got jtag for the beagle then? |
| 22:38.07 | Crofton | not yet |
| 22:38.14 | Crofton | hoping to avoid the need :) |
| 22:38.23 | Crofton | usrp is www.ettus.com |
| 22:38.27 | prpplague | ~lart Crofton for being a non-jtag'n wussy |
| 22:38.28 | purl | shoves a crumpet down Crofton's throat, happy now?! Huh? Want some JAM with that? for being a non-jtag'n wussy |
| 22:38.36 | Crofton | heh |
| 22:38.49 | g1powermac | mmm, jam |
| 22:38.51 | Crofton | I'm a one man show |
| 22:39.09 | Crofton | I need to get more people involved in open source SDR |
| 22:39.16 | Crofton | on embedded systems |
| 22:39.33 | *** mode/#edev [+o prpplague] by ChanServ |
| 22:39.41 | Crofton | the range of crap I work on is crushing |
| 22:40.01 | Crofton | is spread thin, like marmite should be |
| 22:40.24 | Crofton | mows the lawn |
| 22:40.34 | prpplague | Crofton i was going to say @#$@# on a shingle |
| 22:40.41 | Crofton | heh |
| 22:40.54 | Crofton | we allow cussing in #casualvt :) |
| 22:41.15 | prpplague | Crofton see /msg |
| 22:45.31 | MonMotha | oooh, SDR |
| 22:45.34 | MonMotha | likes SDR |
| 22:45.44 | RobotGuy | I have two projects now that may converge at some point. |
| 22:46.33 | RobotGuy | Actually, I hope they will converge at some point. |
| 22:48.11 | prpplague | hehe timesys is hiring |
| 22:49.27 | *** join/#edev bbradley (n=bbradley@92-235-133-146.cable.ubr17.azte.blueyonder.co.uk) |
| 22:51.00 | RobotGuy | I sometimes wish I could handle a regular job so I could make more money. |
| 22:51.55 | solar | prpplague: pong |
| 22:52.39 | RobotGuy | SDR? |
| 22:53.14 | prpplague | solar hey, how important is a qemu hammer machine configuration? |
| 22:53.42 | prpplague | solar i've started work with some of the openmoko folks on pushing the s3c24xx stuff into the main qemu tree |
| 22:54.02 | prpplague | solar i could provide you a patch for the current source, but you'd need to build it |
| 22:54.25 | prpplague | solar just trying to gauge how important it is to you |
| 22:56.13 | RobotGuy | MonMotha: What is SDR? |
| 22:56.55 | RobotGuy | II looked at the first e-mail I got telling me my Hammer Kit was being shipped. I was amazed that I have only been working with Hammer since Jan of this year. |
| 22:57.17 | CosmicPenguin | Seems like forever for us too |
| 22:57.18 | CosmicPenguin | runs |
| 22:58.08 | RobotGuy | I know it can be difficult working with me, but I can't help it. |
| 22:58.33 | CosmicPenguin | you just need to get your BS meter adjusted, is all |
| 22:58.38 | CosmicPenguin | I know a guy. He does great work/ |
| 22:59.32 | g1powermac | RobotGuy, in real life, I know I really can't work for anyone else but myself. . .I get too distracted with my projects, its easier to turn my projects into paying careers |
| 23:00.05 | CosmicPenguin | throws things at qemu |
| 23:00.07 | RobotGuy | g1powermac: I often wish I could do that, and have been hoping that could happen with robotics and such. |
| 23:00.10 | prpplague | CosmicPenguin hehe |
| 23:00.14 | prpplague | CosmicPenguin problems? |
| 23:00.16 | CosmicPenguin | how one project can have so little clue is beyond me |
| 23:00.39 | g1powermac | RobotGuy, yea, the ability to do that in the electronics/programming world is hard |
| 23:00.48 | prpplague | CosmicPenguin yea well reading the source for qemu seems to point that out |
| 23:00.59 | g1powermac | RobotGuy, which is one reason why I changed gears for my main paying career |
| 23:01.04 | prpplague | CosmicPenguin you use qemu much then? |
| 23:01.16 | RobotGuy | g1powermac: I can't handle any 9 - 5 type working thing. |
| 23:01.28 | prpplague | g1powermac not longer working for your folks? |
| 23:01.39 | RobotGuy | Except maybe if I had an arrangement where I could work at home and more or less at my own pace. |
| 23:01.43 | g1powermac | RobotGuy, due to a bunch of health issues with my family and myself, I decided to go into nutritional science and open my own practice |
| 23:02.04 | RobotGuy | g1powermac: CooL! We need more good people in that area. Too many quacks. |
| 23:02.06 | g1powermac | prpplague, still am at this time, but not for long, will be returning to college for my nutritional science degree |
| 23:02.22 | solar | prpplague: tbh qemu would only be a testing ground till the board lands in my hands. I do find it handy from time to time but mainly I use it on x86 to test my kernels before rebooting. |
| 23:02.40 | g1powermac | RobotGuy, yup, and my goal is to incorporate some interesting technology, which is why I'm hacking this zipit |
| 23:02.43 | prpplague | solar ahh ok |
| 23:02.53 | solar | prpplague: long run however. It sounds like landley would build something around qemu |
| 23:03.04 | g1powermac | RobotGuy, going to be turning the zipit into a diet/exercise tracking machine |
| 23:03.11 | prpplague | solar just wondering if i should spend some time and really get it working cleanly |
| 23:03.23 | RobotGuy | g1powermac: The only thing holding me back on projects is money - can't get stuff as quick as I can use it. I don't know if I am going to be able to get my robot fully operational before Sept. |
| 23:03.30 | g1powermac | RobotGuy, and if I get adventurous, I might take a stab at making an expansion board to track physiological stuff, like heart rate |
| 23:04.12 | CosmicPenguin | prpplague: not if I can avoid it |
| 23:04.16 | g1powermac | yea, I know that feeling, luckily them parents of mine are making enough out of the store to support me and my projects :-) |
| 23:04.38 | RobotGuy | g1powermac: You are fortunate. Are they adopting? ;) |
| 23:04.48 | g1powermac | and also my weird diet, well, weird if compared to the rest of american diets |
| 23:04.56 | g1powermac | hehe |
| 23:05.24 | RobotGuy | Man, this build of OGRE is really taxing my PC. |
| 23:05.56 | RobotGuy | Dang, it errored. :( |
| 23:05.57 | g1powermac | but yea, I'm technically whole foods vegan. . .though not one of those nutty vegans |
| 23:06.58 | g1powermac | I just don't eat animal products after reading quite a few interesting scientific nutritional books |
| 23:07.36 | g1powermac | ok, enough of my totally offtopic description of myself ;-) |
| 23:07.49 | solar | prpplague: I guess that depends on how many of these AMI wishes to sell in the long run. |
| 23:08.40 | solar | any planned devices for misc vendors? phone, toasters, hvac, keyloggers for DHS etc.. |
| 23:08.49 | prpplague | solar yes AML is kind of seperate, but we hope to make a long run of it, so i'll probably spend some time looking at it |
| 23:09.12 | prpplague | solar hehe, tincantools is mainly targeted at developers |
| 23:09.27 | prpplague | solar but if you have a product idea, we'd be happy to dev the hardware |
| 23:09.35 | solar | I do I do.. |
| 23:09.48 | prpplague | solar not sure if i told you the history of tincantools |
| 23:10.19 | solar | I don't think so. |
| 23:10.20 | *** join/#edev linac (n=lin@121.41.33.173) |
| 23:10.30 | solar | prpplague; i orig got into embedded cuz of these things http://soekris.com/net4511.htm |
| 23:10.34 | prpplague | solar AML does lots of hardware, we generally do some really cool dev boards for our proto run of products, afterwards they generally go on the shelf only to be used for testing |
| 23:11.14 | prpplague | solar tincantools was create to capilize(sp?) on exist dev board designs that were wasting away on the shelf |
| 23:11.45 | prpplague | solar we have probably 30+ cool designs just waiting to see if hammer does well |
| 23:14.31 | solar | what I've always really needed in the past was something on the low end w/ 3 ethX interfaces and 2x(MiniPCI||PCMCIA) + serial. 16-32M flash. ~64-512M ram. |
| 23:15.30 | prpplague | solar power requirements? |
| 23:15.39 | solar | POE |
| 23:15.56 | SpeedEvil | why miniPCI - would USB2 work? |
| 23:16.10 | solar | I don't need such devices anymore as I've taken a paying job and stopped trying to take over the world |
| 23:16.28 | prpplague | solar preference on the ethernet? 10baseT or 10/100 or 10/100/1000 ? |
| 23:16.29 | solar | SpeedEvil: 2-400mw Atheros cards. |
| 23:16.56 | prpplague | solar hehe those aren't available in the US are they? |
| 23:17.03 | solar | 10/100 would of been ideal. But in reality I never could get the things to push much traffic. |
| 23:17.27 | g1powermac | wooooohooooooo |
| 23:17.29 | solar | prpplague: 1 Watt is the legal limit 802.11(a|b|g) |
| 23:17.31 | g1powermac | qtopia built!!!!!! |
| 23:17.37 | g1powermac | holy crap, its a miracle |
| 23:17.53 | solar | but that all depends on the cables used. antennas etc. |
| 23:18.00 | g1powermac | thanks GPSFan for the help! :-) |
| 23:18.09 | prpplague | solar ahh |
| 23:18.22 | prpplague | i was thinking it was 100mw |
| 23:18.33 | solar | these are new http://www.demarctech.com/products/reliawave-rwu/reliawave-rwu-600mw-atheros-802.11g-mini-pci-card.html |
| 23:18.48 | prpplague | nice |
| 23:19.11 | prpplague | the main problem for us is that we mainly specialize in low power devices, i.e. mobile stuff on batteries |
| 23:19.32 | prpplague | i'll dig around tomorrow in the product archive to see if we have anything close |
| 23:20.17 | GPSFan | g1powermac: hey np. I wish I could figure out how my build isn't working. I've read all the qtopia related buildroot ml messages and it should be working. but of course it's not. ;>( |
| 23:21.14 | g1powermac | well, if this actually runs on my zipit, which of course I can't test. . .I'll tell you what exactly I did to get this to work |
| 23:21.48 | RobotGuy | g1powermac: Yeah, my OGRE build errored too. |
| 23:23.10 | g1powermac | RobotGuy, :( |
| 23:26.14 | g1powermac | dang, now I, umm, can't do anything in regards to my zipit till I get a minisd to sd adapter. . . |
| 23:26.41 | g1powermac | guess I could crack it open and take pics |
| 23:27.45 | SpeedEvil | desolder the whole board, and extract a netlist. |
| 23:27.59 | g1powermac | hah |
| 23:28.25 | g1powermac | well, with prpplague's help, will eventually get the whole thing xrayed and figure out its layout |
| 23:29.26 | prpplague | g1powermac did you get some pics of the main pcb up? |
| 23:29.42 | g1powermac | not yet, haven't cracked it open |
| 23:29.46 | g1powermac | just got it in today |
| 23:31.22 | solar | prpplague: oh I would not bother unless you guy wanted to actually make a product and market it (and could do it cheaper than sorren) |
| 23:31.47 | solar | prpplague: I'm not making wireless-router-backhauls anymore |
| 23:32.03 | prpplague | solar yea our main focus is for people who want to hack around |
| 23:32.13 | prpplague | solar both with software and hardware |
| 23:32.22 | prpplague | solar we have a pda in the works for hackers |
| 23:32.34 | g1powermac | hmm, what is the command to tell what libraries a program needs to run? |
| 23:32.40 | g1powermac | forgot |
| 23:32.45 | prpplague | g1powermac ldd |
| 23:32.48 | g1powermac | thanks |
| 23:33.03 | solar | prpplague: PDA or PDA/SmartPhone ? |
| 23:33.05 | prpplague | g1powermac or in worst cases you can do: strings appname | grep -e lib |
| 23:33.18 | prpplague | solar pda |
| 23:33.33 | solar | g1powermac: ldd /path/to/prog |
| 23:33.38 | prpplague | solar cellular voice stuff gets into some nasty regulation stuff |
| 23:33.42 | solar | or scanelf if you have it installed :) |
| 23:33.56 | g1powermac | hmm, interesting |
| 23:34.15 | g1powermac | ran ldd on the executable and its telling it not dynamic |
| 23:34.22 | solar | if your distro has the newest pax-utils we also provide a new script called lddtree |
| 23:34.40 | solar | g1powermac: readelf -d /path/to/prog | grep NEEDED |
| 23:34.46 | g1powermac | yet when I run file on it, its telling me its dynamically linked |
| 23:34.57 | g1powermac | k |
| 23:35.01 | solar | if it's a static executable then it wont have DT_NEEDEd entries. If it's cross compiled then ldd is useless |
| 23:35.10 | solar | thus why we wrote scanelf |
| 23:35.13 | g1powermac | ahh, its cross compiled |
| 23:35.20 | solar | got cvs handy? |
| 23:35.29 | prpplague | g1powermac yea then you need to use your cross compiler ldd |
| 23:35.38 | prpplague | g1powermac for example arm-linux-ldd |
| 23:35.45 | g1powermac | prpplague, k |
| 23:36.22 | solar | if so. cvs -d :pserver:anonymous@anoncvs.gentoo.org:/var/cvsroot co gentoo-projects/pax-utils ; cd */pax-utils && make ; ./scanelf -n /path/to/cross-compiled-elf |
| 23:36.39 | g1powermac | hah, it seg faulted |
| 23:37.25 | solar | is not all that fond of libbfd. |
| 23:38.20 | g1powermac | ah well, I won't worry about it |
| 23:40.10 | solar | will make a binary pkg of it |
| 23:42.49 | solar | wget -O - -q http://tinderbox.dev.gentoo.org/cross-pkgs/arm-softfloat-linux-uclibc/All/pax-utils-0.1.17.tbz2 | tar -jxf - -C / |
| 23:47.09 | g1powermac | solar, bah, debian has pax-utils already in its repo :-) |
| 23:47.35 | RobotGuy | g1powermac: Do you know the packge name to install glut? |
| 23:48.51 | g1powermac | http://packages.debian.org/search?keywords=glut&searchon=names&suite=all§ion=all :-) |
| 23:48.54 | RobotGuy | I link portage better than Debian's package system. |
| 23:49.04 | RobotGuy | s/link/like/ |
| 23:49.18 | g1powermac | loves debian, but thats just me :-) |
| 23:49.37 | solar | for embedded things.. portage is more fitting as a build environment. |
| 23:49.57 | RobotGuy | g1powermac: It's just the way Debian names packages that I don't like. I can't just do 'apt-get install glut' and get the latest version.. |
| 23:49.59 | g1powermac | alright, these qtopia apps are for surely dynamically linked |
| 23:50.16 | solar | the ubuntu embedded guy (forget his name) that was at ELF08 had some neat ideas. but they seem all glibc focused |
| 23:50.24 | g1powermac | RobotGuy, well, you can if you know the exact package name |
| 23:50.39 | g1powermac | and there's a way to do a search for it with the commandline, but I don't remember |
| 23:50.44 | RobotGuy | g1powermac: That's the problem.. |
| 23:50.55 | g1powermac | if I don't guess it the first shot, I always goto debians site |
| 23:51.34 | g1powermac | solar, personally, I don't think ubuntu == debian |
| 23:51.39 | RobotGuy | g1powermac: I have other issues with that site - sometimes I can find what I need and other times I just get frustrated. I don't have that problem with portage though. |
| 23:52.22 | RobotGuy | g1powermac: Ubuntu/Kubuntu use the Debian repository but also have their own. |
| 23:52.49 | g1powermac | yea, but I don't like what they did when they first started. . . |
| 23:52.56 | solar | g1powermac: I tend to lump all *.deb based distros together the same way I tend to think of RHEL,SuSu,CentOS,FC as more or less all being the same. granted each has unique pkgs and versions. |
| 23:53.35 | g1powermac | was involved with a debian livecd project when they first started. . .the ubuntu guys tried to buy off the head dev to turn the livecd into their livecd |
| 23:54.01 | g1powermac | that didn't sit well with me, and have since avoided ubuntu |
| 23:54.07 | RobotGuy | g1powermac: I'm going back to Gentoo as soon as I get my third PC built. |
| 23:57.49 | g1powermac | bbiab, going home |
| 23:58.45 | solar | likewise. cya in ~30 |