00:01.03 | rwhitby | I would put a couple of ~1GB rootfs partitions and small swap partitions on the disk (so you can easily swap between different rootfs's), and then use the rest as data storage |
00:01.18 | rwhitby | no real need for the usb stick if you have a disk permanently attached |
00:02.21 | CoreDump|home | rwhitby: ok |
00:02.36 | CoreDump|home | hey upslug2 is in debian testing already |
00:03.00 | rwhitby | yep, nslu2 is well supported in debian now |
00:12.39 | CoreDump|home | rwhitby: --rootfs=slugosle-4.3-beta-nslu2.bin" and _not_ the jffs2 file right? |
00:13.20 | rwhitby | nope, you want -i *.bin |
00:13.38 | rwhitby | (since you need to flash kernel, rootfs, and fis dir) |
00:14.07 | CoreDump|home | hmmm |
00:15.39 | CoreDump|home | ah I get it. *.bin has rootfs and fis dir, *rootfs* is the rootfs partition only |
00:15.50 | rwhitby | yep |
00:16.06 | CoreDump|home | but the kernel is still extra? |
00:16.08 | rwhitby | *.bin is actaully a full 8MB image, including redboot, sysconf, ... |
00:16.21 | rwhitby | *.bin has everything |
00:16.30 | CoreDump|home | ok |
00:17.18 | CoreDump|home | --image=blahh.bin it is then |
00:17.19 | rwhitby | (and some of the things in it, like redboot and sysconf, don't actually get written by default, unless you use the dangerous rewrite everything switch in upslug2, or the Linksys EraseAll tool (which is known to brick slugs) |
00:17.22 | rwhitby | yes |
00:17.48 | CoreDump|home | rwhitby: yeah, rewriting redboot on each flash would suck bad =D |
00:18.30 | rwhitby | oh, yeah. we *never* overwrite the bootloader. |
00:19.09 | rwhitby | the *only* $USER cases of bricks are when people used (against our strong public warnings) the Linksys EraseAll tool, and had a power failure (or cat incident) while flashing. |
00:19.46 | CoreDump|home | !praise JTAg |
00:20.27 | rwhitby | yeah, I've only needed to use JTAG once in over two years, when I was playing around with Apex bootloader replacement. |
00:20.47 | CoreDump|home | =) |
00:20.55 | CoreDump|home | slugos is booting |
00:21.10 | rwhitby | and the only other developer brick I know of that needed Apex was when jacques used a very early prototype version of slugimage that had a bad bug in it. |
00:21.14 | rwhitby | s/Apex/JTAG/ |
00:21.26 | rwhitby | CoreDump|home: do you have serial on it? |
00:21.44 | CoreDump|home | rwhitby: not yet, had to order parts first |
00:21.55 | CoreDump|home | what's the root p/w? |
00:22.01 | rwhitby | ok, so you're getting the real $USER experience |
00:22.09 | CoreDump|home | righto |
00:22.13 | rwhitby | opeNSLUg |
00:22.38 | CoreDump|home | excellent |
00:22.46 | rwhitby | now turnup init |
00:22.47 | CoreDump|home | "Free Your Slug" |
00:24.55 | CoreDump|home | it sais I need to reboot. The settings are kept on flash right? |
00:28.03 | rwhitby | yep |
00:29.34 | *** join/#openslug hillct (n=H@client200-5.dsl.intrex.net) |
00:30.30 | *** part/#openslug hillct (n=H@client200-5.dsl.intrex.net) |
00:36.30 | rwhitby | CoreDump|home: ready for turnup disk? |
00:38.23 | CoreDump|home | not yet |
00:44.22 | rwhitby | bbiab |
00:53.28 | CoreDump|home | rwhitby: I turnup disk -i a partition |
00:56.38 | CoreDump|home | yay, that was easy |
01:33.33 | rwhitby | sweet |
01:35.46 | CoreDump|home | I forgot how much I hated the .dev branch |
01:36.12 | rwhitby | CoreDump|home: test out "turnup ram" and then scp up a new image into /tmp, then use 'reflash' to flash the new image from inside linux |
01:36.56 | CoreDump|home | wow, nice trick |
01:37.36 | CoreDump|home | I'll try that another time =) |
01:40.03 | rwhitby | of course when altboot supports kexec on nslu2, then we'll just boot into the new kernel from the external disk ... |
01:41.48 | rwhitby | CoreDump|home: an altboot ram target would be useful for turnup :-) |
01:41.59 | rwhitby | s/turnup/reflash/ |
01:45.02 | CoreDump|home | rwhitby: I found a problem with your namechange. MACHINE specific ipk's will be blahh_nslu2le.ipk, however, ipkg's arch.conf doesn't list nslu2* as valid ARCH |
01:46.10 | rwhitby | Hmm. I thought I added nslu2* as a PACKAGE_EXTRA_ARCH |
01:47.22 | rwhitby | ipkgarchs='all any noarch arm armv4 armv4t armv5e armv5te ixp4xxle nslu2le' in log.do_rootfs |
01:47.35 | rwhitby | CoreDump|home: where are you experiencing the problem? |
01:47.44 | CoreDump|home | altboot_0.0.0.bb ;) |
01:47.58 | CoreDump|home | I pulled yesterday the last time tho |
01:48.56 | rwhitby | hmm - you're right - the /etc/ipkg.conf doesn't have nslu2le in it. |
01:49.49 | rwhitby | I thought that file would get PACKAGE_EXTRA_ARCHS and MACHINE added automatically |
01:49.53 | CoreDump|home | s/ipkg.conf/arch.conf |
01:50.01 | rwhitby | right |
01:50.19 | CoreDump|home | DEBUG: Couldn't read keyboard ints! |
01:50.45 | CoreDump|home | this is altboot on the slug heh |
01:50.56 | rwhitby | CoreDump|home: do you know where /etc/ipkg/arch.conf is generated? |
01:51.09 | CoreDump|home | not off-hand, no |
01:51.32 | rwhitby | looks like rootfs_ipk.bbclass |
01:51.54 | CoreDump|home | somewhere along these lines, yes |
01:51.58 | rwhitby | from PACKAGE_ARCHS |
01:52.58 | rwhitby | PACKAGE_ARCHS = "all any noarch ${TARGET_ARCH} ${PACKAGE_EXTRA_ARCHS} ${MACHINE}" in bitbake.conf |
01:53.14 | rwhitby | so it should get nslu2le from MACHINE |
01:54.28 | rwhitby | Doh! why is slugos-image removing it? |
01:54.40 | CoreDump|home | he good question |
01:55.20 | rwhitby | ah. that was from when MACHINE was "nslu2", not "nslu2le". Now it is correct it should not be removed. Thanks for finding that! |
01:56.05 | CoreDump|home | np |
01:56.41 | CoreDump|home | which bastardized version of "ps" is slugos using? |
01:57.49 | rwhitby | probably busybox |
01:59.37 | CoreDump|home | oh well, an error message that nobody (expect folks w/ serial port) can see can probably ignored |
02:00.36 | rwhitby | heh - if I see it I won't be ignoring it :-) |
02:00.52 | CoreDump|home | =D |
02:01.08 | rwhitby | OK, arch.conf fixed and pushed. |
02:01.52 | rwhitby | BTW, I always push directly to monotone.nslu2-linux.org using the Master Makefile, and that sync periodically to monotone.openembedded.org, so there is sometimes a delay before CIA reports it in #oe |
02:43.23 | *** join/#openslug jacques (n=jacques@nslu2-linux/jacques) |
02:57.08 | VoodooZ | I was reading your discussion w/ CoreDump|home: what does reflash -i update exactly? kernel, rootfs, fis? |
03:00.14 | rwhitby | kernel and rootfs at least |
03:00.57 | rwhitby | doesn't change FIS |
03:01.32 | VoodooZ | ok, that's what I need now.. I guess I could also write the whole darn image to flash using apex but I'm afraid... |
03:01.47 | VoodooZ | I'd rather do it from linux like you said... |
03:02.40 | rwhitby | just do hexdump compares like I said, and then write it from linux |
03:03.43 | VoodooZ | hexdump |
03:03.46 | VoodooZ | hexdump |
03:03.50 | VoodooZ | oops. |
03:04.10 | CoreDump|home | rwhitby: can we make the buzzer beep at will? |
03:05.14 | rwhitby | sure |
03:05.35 | CoreDump|home | what's the trick from the shell? |
03:05.56 | rwhitby | run /bin/beep |
03:06.23 | CoreDump|home | err, I didn't even try "beep" *cry* |
03:06.51 | rwhitby | someone even got it to play jingle bells last christmas I think |
03:07.12 | CoreDump|home | HEHEHE |
03:07.25 | VoodooZ | rwhitby: ok, I've tried hexdump /dev/mtdblock5 and mtd5 but those are small compared to hexbump'ing FIS. yes I'm confused. |
03:08.13 | rwhitby | VoodooZ: is the difference the new microcode added to FIS ? |
03:09.38 | VoodooZ | sorry, I'm confused about what to diff. |
03:10.11 | VoodooZ | can't I just overwrite the darn thing instead? |
03:12.07 | VoodooZ | keep in mind that I don't know much about the whole mtd thing. Got any good tutorial? |
03:16.04 | rwhitby | VoodooZ: /dev/mtdblock5 is your FIS directory partition |
03:16.26 | rwhitby | it's current contents have the FIS directory entries, and the Trailer. |
03:16.37 | rwhitby | the new contents have the microcode in the middle as well |
03:16.57 | rwhitby | I want you to confirm that the start and the end of the block is unchanged, before you go writing it. |
03:17.30 | rwhitby | Since you don't have RedBoot, if you stuff up the FIS are you will need to reload via serial (very slowly). |
03:18.10 | VoodooZ | yeah, It's real slow! |
03:19.11 | VoodooZ | I guess most people going from 3.10 -> 4.3 didn't have this issue because they used upgrade mode right? |
03:19.28 | VoodooZ | I used reflash which doesn't update everything right? |
03:25.26 | CoreDump|home | how can I boot to the flash rootfs after turnup disk? |
03:25.52 | VoodooZ | turnup flash |
03:26.19 | CoreDump|home | VoodooZ: thanks! |
03:26.36 | VoodooZ | type turnup by itself for a full usage. I know because I keep forgetting! :) |
03:26.51 | CoreDump|home | heh |
03:27.23 | VoodooZ | now if I could to the point you are at.. :( |
03:27.27 | rwhitby | CoreDump|home: altboot will do "turnup nfs" as well, right? |
03:27.39 | CoreDump|home | yeah |
03:27.52 | rwhitby | VoodooZ: where are you blocked? |
03:28.27 | VoodooZ | at the same place as before. The FIS. |
03:28.41 | rwhitby | ok, do you have the current contents in a file? what's the filename? |
03:28.46 | VoodooZ | I want to start in a sane state as reflash doesn't flash everything |
03:29.00 | VoodooZ | the FIS part extracted from the full image? yes. |
03:29.13 | VoodooZ | and it's around 128kbytes. |
03:29.31 | VoodooZ | Unless i screwed up of course. |
03:31.21 | rwhitby | ok, good. |
03:31.40 | rwhitby | now, hexdump -C /dev/mtdblock5 > orig.txt |
03:31.54 | VoodooZ | done. |
03:31.57 | rwhitby | hexdump -C fis-file > new.txt |
03:32.11 | rwhitby | diff orig.txt new.txt |
03:32.15 | rwhitby | pastebin the output |
03:32.39 | VoodooZ | holy smoke! It's long! |
03:32.48 | rwhitby | (both files should be the same size, with similar starting contents and trailer, but with very different middles) |
03:33.16 | VoodooZ | ls -l *.txt |
03:33.20 | VoodooZ | oop. |
03:33.23 | rwhitby | can you recognise the FIS directory and the Trailer (which should be the same between both files) ? |
03:33.45 | CoreDump|home | rwhitby: altboot could replace /linuxrc as well |
03:33.47 | VoodooZ | no, that's the problem. the file are not even close to being the same size |
03:34.01 | rwhitby | CoreDump|home: yes, that's the plan |
03:34.07 | VoodooZ | -rw-r--r-- 1 root root 34485 Jan 1 00:37 new.txt |
03:34.07 | VoodooZ | -rw-r--r-- 1 root root 1763 Jan 1 00:37 orig.txt |
03:34.27 | rwhitby | VoodooZ: that's expected |
03:34.45 | rwhitby | orig.txt has a whole lot of FFs in the middle, wherease new.txt has the microcode |
03:34.57 | VoodooZ | Ah! |
03:35.25 | VoodooZ | and the ff don't show up in hexdump? |
03:35.36 | rwhitby | hexdump elides duplicate lines |
03:37.01 | VoodooZ | I see. |
03:37.08 | VoodooZ | Here's the pastebin: http://www.pastebin.ca/356542 |
03:38.49 | CoreDump|home | time for bed, n8! |
03:39.13 | VoodooZ | night |
03:39.44 | VoodooZ | it's going to be my turn soon. But I'd love to understand this before though.. |
03:39.56 | rwhitby | VoodooZ: that looks fine. |
03:40.14 | VoodooZ | Is that the f/w starting at 0001c000? |
03:40.18 | rwhitby | yes |
03:40.24 | rwhitby | microcode |
03:40.34 | VoodooZ | ok. are those minor diffs at the top normal? |
03:41.18 | rwhitby | that's the new skip regions |
03:41.42 | VoodooZ | ok. so does that mean I'm safe to flash it? |
03:41.45 | rwhitby | yes |
03:42.32 | VoodooZ | ok, would i do cat FIS_file > /dev/mtdblock5 ? |
03:43.42 | VoodooZ | I'm just curious why there's a microcode in the FIS dir partition? Shouldn't that be in a another partition? |
03:43.58 | VoodooZ | then again, I'm totally new to FIS so... |
03:44.54 | rwhitby | we put it there so as not to waste space. |
03:45.05 | rwhitby | yes you can cat that to /dev/mtdblock5 |
03:45.25 | rwhitby | do you really want to give up another 128Kb of flash space for the microcode? |
03:45.39 | VoodooZ | nope. Good point! |
03:45.40 | rwhitby | (when there's 104Kb spare in the FIS directory block) |
03:46.23 | VoodooZ | ok. because of the 128k block size? |
03:46.34 | rwhitby | yes |
03:47.01 | VoodooZ | ok. understood. at least I learned something out of all this. |
03:47.06 | VoodooZ | here goes.... |
03:47.37 | VoodooZ | oops. it says: cat: Write Error: Operation not permitted |
03:48.02 | *** join/#openslug AwayNAiL (n=repvik@nslu2-linux/pdpc.active.NAiL) |
03:48.11 | VoodooZ | unless i have to use mtd5? |
03:48.37 | rwhitby | aha - FIS directory is write protected |
03:48.51 | rwhitby | ok, you need to go into Apex and write that file into the FIS directory partition |
03:48.58 | VoodooZ | ouch. |
03:49.14 | VoodooZ | oh well, it's better than an 8meg file i guess. |
03:49.41 | VoodooZ | my apex is old and probably doesn't understand FIS right? |
03:50.09 | VoodooZ | I have 1.3.13 |
03:50.26 | VoodooZ | which I wanted to upgrade but was too chicken to do. :( |
03:50.48 | VoodooZ | ok, all I need then is the starting memory address. |
03:51.46 | rwhitby | I'm off to get some lunch - back in about 15 minutes |
03:52.45 | VoodooZ | ok. I might see you tommorow then as it's bed time. |
03:52.50 | VoodooZ | Thanks for your patience. |
03:55.35 | VoodooZ | If I want to update my apex, would the default apex package do or is it configured as a second bootloader? |
04:00.47 | *** join/#openslug AwayNAiL_ (n=repvik@sql.kynisk.com) |
04:14.09 | VoodooZ | time for bed. night. |
04:14.40 | VoodooZ | rwhitby: I'll be reading the logs again tommorow so let me know if you remember the exact address of the FIS dir partition. |
04:14.59 | VoodooZ | meanwhile I'll start looking into updating my apex to latest version with FIS driver. |
04:15.03 | VoodooZ | thanks |
04:15.08 | rwhitby | VoodooZ: exact address is in the wiki |
04:15.34 | VoodooZ | ok. I'll try to find it. |
04:16.17 | rwhitby | http://www.nslu2-linux.org/wiki/Info/LastBlockMemoryMap |
04:16.19 | VoodooZ | is there anything in the apex package config that would prevent me from running it from flash? |
04:16.22 | VoodooZ | thanks |
04:16.41 | rwhitby | 0x507E0000 |
04:16.48 | VoodooZ | some stuff on the wiki is outdated so I'm never sure... |
04:16.59 | rwhitby | VoodooZ: I know nothing about Apex as first stage. You are on your *own* there. |
04:17.05 | VoodooZ | no problem. |
04:17.17 | VoodooZ | do you know who set it up for second stage though? |
04:17.55 | VoodooZ | thanks for the link. it's a good read... |
04:21.37 | rwhitby | I set it up for second stage |
04:22.17 | VoodooZ | ok. no worries. that's for another day.. |
04:22.22 | VoodooZ | thanks and good night. |
04:22.28 | rwhitby | night |
05:04.36 | *** join/#openslug \Td (n=Sinclair@p54A0B8D0.dip0.t-ipconnect.de) |
05:43.36 | *** join/#openslug DrZimmerman (n=theo@217-162-225-50.dclient.hispeed.ch) |
05:59.50 | *** join/#openslug Th^ (n=Sinclair@p54A0A982.dip0.t-ipconnect.de) |
08:20.20 | *** join/#openslug CieD (n=CieD@host86-136-171-177.range86-136.btcentralplus.com) |
10:09.24 | *** join/#openslug kolla (n=kolla@drift1.uninett.no) |
10:46.36 | *** join/#openslug rwhitby (n=rwhitby@nslu2-linux/rwhitby) |
11:28.56 | CoreDump|afk | rwhitby: do you have any thoughts how $USER should select the partition to be booted by altboot? |
11:29.39 | CoreDump|afk | nm, I'll hack up something |
11:47.47 | rwhitby | CoreDump|afk: just by /dev/sda1 type name, configuring it from the cmdline or a menu input |
11:47.59 | rwhitby | (and it takes effect on the next boot) |
11:48.36 | rwhitby | there is no way for $USER to select the boot partition in real time - it's always the value they selected whilst configuring altboot during the previous session before they rebooted. |
11:51.58 | CoreDump|afk | yep |
11:52.10 | CoreDump|afk | I'll hack a menu |
11:55.22 | rwhitby | note that we make sure we mount the partition by UUID, in case the usb hub reorders them randomly, or someone plugs another disk in ahead of the one your booting from in the USB order. |
11:55.41 | rwhitby | (see the slugos-init scripts for how we do the uuid boot selection. |
11:57.13 | CoreDump|afk | ok |
11:58.52 | rwhitby | the user selects by /dev/sda1 partition name, but the behind-the-scenes code turns this into a UUID and uses that in the boot process |
12:13.14 | *** join/#openslug caplink811-log (n=caplink8@dslb-088-073-039-237.pools.arcor-ip.net) |
12:27.48 | *** join/#openslug caplink811_log (n=caplink8@dslb-088-073-050-118.pools.arcor-ip.net) |
13:47.06 | VoodooZ | who replaced redboot with apex here? |
13:48.09 | VoodooZ | NAiL: you did too? recently? which version? |
13:48.41 | NAiL | A long time ago |
13:48.47 | NAiL | very long time ago |
13:49.07 | NAiL | I have no idea which version actually |
13:49.54 | VoodooZ | ok. I did too way back so I now need a newer version (w/ FIS support) but I'm afraid my old defconfig might not cut it anymore. |
13:50.19 | VoodooZ | rwhitby: thanks for making the FIS writeable. I'll give that a try... |
13:59.17 | VoodooZ | how can I rebuild slugosbe-image with the latest pulls without deleting tmp. ? |
14:05.09 | *** join/#openslug gerdi__ (n=gerdi@dslb-084-059-053-069.pools.arcor-ip.net) |
15:30.16 | VoodooZ | What's the alias of the apex dude again? |
17:02.09 | *** join/#openslug mwester (n=mwester@nslu2-linux/mwester) |
17:22.18 | *** join/#openslug pleemans (n=peter@d51A5E76A.access.telenet.be) |
17:47.09 | *** join/#openslug apt (i=ibot@pdpc/supporter/active/TimRiker/bot/apt) |
17:47.09 | *** topic/#openslug is Developer (not end-user) discussion about the OpenSlug custom firmware for the Linksys NSLU2 device. |
18:20.19 | *** join/#openslug caplink811-log (n=caplink8@dslb-088-073-012-216.pools.arcor-ip.net) |
19:13.13 | *** join/#openslug mproctor (n=mproctor@netblock-66-245-237-56.dslextreme.com) |
19:53.02 | *** join/#openslug caplink811-log (n=caplink8@dslb-088-073-034-022.pools.arcor-ip.net) |
22:52.04 | *** join/#openslug rwhitby (n=rwhitby@nslu2-linux/rwhitby) |
22:53.51 | CoreDump|home | hi |
22:56.17 | rwhitby | hey CoreDump|home |
23:00.58 | *** part/#openslug CoreDump|home (n=mhentges@hentges.net) |
23:11.15 | *** join/#openslug kapookee (n=kapookee@c-69-243-24-15.hsd1.va.comcast.net) |
23:18.25 | VoodooZ | rwhitby: good. I rebuilt and flashed the latest slugos and was able to cat FIS to mtdblock5 thanks to your change |
23:20.59 | rwhitby | cool |
23:28.31 | VoodooZ | cool! It picked up the firmware too! |
23:29.22 | VoodooZ | Ive got a fresh copy of apex ready to flas too which supports FIS. |
23:30.16 | rwhitby | sweet - is the configuration for first stage bootloader any different from the current slugos-nslu2-armeb_config in OE ? |
23:48.56 | *** join/#openslug saladino (n=saladino@55.Red-80-35-102.staticIP.rima-tde.net) |
23:49.04 | saladino | hi |
23:50.18 | rwhitby | hi saladino |
23:50.35 | saladino | hi rwhitby |