**** BEGIN LOGGING AT Fri Jun 10 02:59:57 2011
Jun 10 10:29:32 <ant_work> Jay7: fwiw today on klibc-dev
Jun 10 10:29:35 <ant_work> [klibc] [PATCH] limits: Define LLONG_MIN, LLONG_MAX and ULLONG_MAX
Jun 10 10:30:29 <ant_work> I could live with ULONGLONG_MAX || ULLONG_MAX
Jun 10 10:34:50 <ant_work> good that maximilian is tracking kexecboot
Jun 10 11:32:43 <ant_work> Jay7: pb_ says glibc/uclibc/eglibc have ULLONG_MAX
Jun 10 11:33:35 <ant_work> "so I think you should be safe assuming that they exist"
Jun 10 11:34:00 <ant_work> BUG STATUS: RESOLVED: INVALID
Jun 10 11:34:05 <ant_work> :p
Jun 10 11:35:12 <ant_work> dietlibc too
Jun 10 12:16:24 <ant_work> atiti: welcome
Jun 10 12:16:47 <ant_work> atiti: we are a bit stalling with reference to fstype
Jun 10 12:17:06 <ant_work> jffs2 and fat have been somehow added
Jun 10 12:17:18 <ant_work> we still lack ubi/ubifs support
Jun 10 12:18:44 <atiti> hi
Jun 10 12:18:57 <atiti> logfs
Jun 10 12:19:21 <atiti> its a flash aware fs which is to replace jffs2
Jun 10 12:19:35 <atiti> jffs2 has some really long mounting times for large fs sizes
Jun 10 12:19:49 <ant_work> sadly, yes
Jun 10 12:21:14 <ant_work> if we want more users (OMAP) we still need ubi/ubifs, though
Jun 10 12:21:56 <ant_work> uImage support was the first obstacle, now solved
Jun 10 12:22:52 <atiti> ;)
Jun 10 12:23:18 <atiti> what devices are you targetting nowadays
Jun 10 12:23:47 <atiti> i moved away from my zaurus now and entered the world of androids (tablet + phone) :D
Jun 10 12:26:47 <ant_work> well, some dev have n900
Jun 10 12:26:59 <ant_work> other are playing with beagleboard
Jun 10 12:27:44 <ant_work> basically armv6/7 are now using ubi
Jun 10 12:28:34 <ant_work> anyway, logfs seems interesting
Jun 10 12:37:14 <atiti> :)
Jun 10 13:39:09 <ant_work> hi whse-work
Jun 10 13:39:22 <whse-work> hey
Jun 10 13:39:43 <whse-work> I'd just like to say, there are days I hate the OXNAS platform, outright..
Jun 10 13:46:50 <whse-work> I really do want to sort out the use of kexecboot, but this kernel is so frustrating....
Jun 10 13:46:58 <whse-work> and NDA's all over the place.. UHG.
Jun 10 13:50:02 <ant_work> :/
Jun 10 15:20:53 <Jay7> evening
Jun 10 15:21:35 <Jay7> ant_work: can you suggest how to check that we are compiling with klibc?
Jun 10 15:21:50 <Jay7> what #ifdef to check? :)
Jun 10 15:21:56 <ant_work> no need
Jun 10 15:22:00 <ant_work> klibc is fixed
Jun 10 15:22:08 <Jay7> do we have this fix already?
Jun 10 15:22:12 <ant_work> I'll add the patch until 1.523
Jun 10 15:22:17 <Jay7> ok
Jun 10 15:22:31 <Jay7> hope no one will use it with previous versions :)
Jun 10 15:22:33 <ant_work> next week or before
Jun 10 15:22:44 <Jay7> I've done lot of fixes
Jun 10 15:22:55 <ant_work> I'll resend to maximilian som eklibc patches we have
Jun 10 15:23:04 <Jay7> but that menu->current->current->id is still unresolved
Jun 10 15:23:10 <ant_work> yes
Jun 10 15:23:26 <Jay7> but now we have frame around active item :)
Jun 10 15:23:30 <ant_work> but imagine that somehow *once* poodle was not freezing
Jun 10 15:23:35 <Jay7> but have no separating lines
Jun 10 15:23:42 <Jay7> like before with old UI
Jun 10 15:23:55 <Jay7> I'll push changes today
Jun 10 15:24:20 <ant_work> I mean, I entered System Menu
Jun 10 15:24:48 <ant_work> I could do it only one time out of 10 :/
Jun 10 15:25:05 <ant_work> tried inserting/removing CF and SD
Jun 10 15:26:53 <ant_work> you mean there is same background color?
Jun 10 15:27:28 <ant_work> ah, I see
Jun 10 15:27:53 <ant_work> the line on the bottom was invisible once slot was selected, anyhow
Jun 10 15:27:54 <Jay7> ant_work: no, active item have white bg with frame around
Jun 10 15:28:15 <Jay7> need to eat something
Jun 10 15:28:19 <Jay7> will back later
Jun 10 15:28:20 <ant_work> ok, border+color
Jun 10 15:28:23 <ant_work> top
Jun 10 15:28:26 <ant_work> ;)
Jun 10 15:28:32 <ant_work> bbl
Jun 10 16:56:09 <Jay7> http://tinylab.org/index.php/projects/tinylinux/
Jun 10 16:56:18 <Jay7> "Kexec based bootloader is a classic application of Tiny Linux Kernel or we can say Tiny Linux Kernel is a requirement of Kexec based bootloader. This task tries to implement the demos for X86, MIPS, ARM and PPC. There are related resources in the kexec-tools and kexecboot projects."
Jun 10 16:56:43 <Jay7> http://elinux.org/Work_on_Tiny_Linux_Kernel
Jun 10 16:56:50 <Jay7> by Wu Zhangjin
Jun 10 16:57:54 <whse-work> interesting
Jun 10 16:59:39 <Jay7> whse-work: btw, I've mainlined fix for large partitions
Jun 10 16:59:58 <whse-work> :) yw
Jun 10 18:17:37 <whse-work> Jay7: ping
Jun 10 18:17:46 <Jay7> pong
Jun 10 18:18:05 <whse-work> K, so, what are the actual requirements from the uInitrd
Jun 10 18:18:18 <whse-work> kexec.. kexecboot
Jun 10 18:18:20 <whse-work> wtc
Jun 10 18:18:23 <whse-work> *etc
Jun 10 18:18:39 <Jay7> some devices in /dev, /usr/sbin/kexec and kexecboot somewhere
Jun 10 18:18:42 * whse-work has some free time for the first time in 2 weeks
Jun 10 18:19:01 <whse-work> nothing else required ?
Jun 10 18:19:07 <Jay7> iirc, no
Jun 10 18:19:13 <whse-work> hrm..
Jun 10 18:19:50 <Jay7> we are starting kexecboot as init by kernel
Jun 10 18:21:36 <whse-work> interesting
Jun 10 18:22:17 <whse-work> have to figure out how to get a uInitrd running..
Jun 10 18:23:14 <whse-work> never used it with these boards..
Jun 10 18:23:32 <Jay7> why not just compile it into kernel?
Jun 10 18:23:39 <whse-work> that is what I was thinking
Jun 10 18:24:03 <whse-work> strip out everything that isn't necessary, and then compile that in..
Jun 10 18:24:40 <Jay7> ant may show our initramfs directory layout
Jun 10 18:25:26 <whse-work> ant is ?
Jun 10 18:28:42 <whse-work> like, the user ant that was here earlier ?
Jun 10 18:57:23 <whse-work> this kind of what I am looking for ? http://www.microsuncn.com/index.php?title=Analysis_linux_initrd
Jun 10 20:02:51 <whse-work> u said /usr/sbin/kexec
Jun 10 20:03:08 <whse-work> my package in the arch-upstream is /sbin/kexec
Jun 10 20:05:59 <Jay7> whse-work: use --with-kexec-binary
Jun 10 20:06:04 <Jay7> kexecboot's configure option
Jun 10 20:06:08 <whse-work> ah
Jun 10 20:06:24 <Jay7> or just symlink it :)
Jun 10 20:06:55 <whse-work> yeah, I just copied it to /usr/sbin/kexec in the initramfs tree
Jun 10 20:07:16 <Jay7> choose way you like more :)
Jun 10 20:23:05 <whse-work> is thinking I understand this stuff...
Jun 10 20:23:38 <whse-work> i can just symlink init to kexeboot, yets?
Jun 10 21:15:02 <Jay7> whse-work: yes
Jun 10 21:15:11 <whse-work> :)
Jun 10 21:15:15 <whse-work> building now..
Jun 10 21:33:54 <ant__> hello
Jun 10 21:34:25 <ant__> Jay7: looking at initramfs right now, I see it has been polluted by some empty dirs
Jun 10 21:34:52 <ant__> now we have even /etc/device_table and /etc/version  :/
Jun 10 21:35:16 <ant__> I'll clean it out (again) soon
Jun 10 21:35:42 <ant__> now and then some metadata changes forcefully populate the images
Jun 10 21:35:54 <ant__> with their cruf
Jun 10 21:38:22 <ant__> happily size is still under control
Jun 10 21:38:23 <ant__> -rw-r--r-- 1 andrea users  81K Jun 10 23:01 Angstrom-initramfs-kexecboot-klibc-image-eglibc-ipk-v2011.06-poodle.rootfs.cpio.gz
Jun 10 21:38:23 <ant__> -rw-r--r-- 1 andrea users  57K Jun 10 23:01 Angstrom-initramfs-kexecboot-klibc-image-eglibc-ipk-v2011.06-poodle.rootfs.cpio.lzma
Jun 10 21:52:03 <ant__> Jay7: udev cleanings pushed. Now Zaurus uses default 1.71
Jun 10 21:52:40 <ant__> (I still aim to get rid of udev sooner or later...)
Jun 10 21:53:20 <ant__> ouch.. 'aim at getting rid' sounds more english
Jun 10 21:59:26 <whse-work> lemme guess, I needed udev huh :p
Jun 10 21:59:59 <ant__> no no, only for the rootfs
Jun 10 22:00:22 <ant__> devtmpfs + mdev sounds promising
Jun 10 22:00:41 <ant__> and lightweight
Jun 10 22:01:16 <ant__> how would you pack udev in the 57k cpio ? :p
Jun 10 22:04:55 <Jay7> ant__: nice
Jun 10 22:05:36 <Jay7> well..
Jun 10 22:05:59 <Jay7> ant__: should I push changes I made to master or may be to another branch?
Jun 10 22:06:12 <ant__> go for master
Jun 10 22:07:36 <Jay7> pushing now
Jun 10 22:07:39 <CIA-10> kexecboot: jay4mail master * rv0.5-172-g854aa92 / (fb.c fb.h res/theme.h rgb.c rgb.h rgbtab.h xpm.c):
Jun 10 22:07:39 <CIA-10> kexecboot: Rewrite colors and xpm images processing code
Jun 10 22:07:39 <CIA-10> kexecboot: Image storage format was changed to use uint32_t packed RGBA.
Jun 10 22:07:39 <CIA-10> kexecboot: FB and XPM processing code was adopted and optimized.
Jun 10 22:07:39 <CIA-10> kexecboot: jay4mail master * rv0.5-175-g2e6c040 /res/theme.h: Make icon pad larger
Jun 10 22:07:39 <CIA-10> kexecboot: jay4mail master * rv0.5-171-g4ee0416 /kexecboot.c: Remove mount point from menu's kernelpath (label)
Jun 10 22:07:40 <CIA-10> kexecboot: jay4mail master * rv0.5-170-g4686a76 /kexecboot.c:
Jun 10 22:07:40 <CIA-10> kexecboot: Rewind log to top on exit from log view
Jun 10 22:07:41 <CIA-10> kexecboot: This should make log view usable on devices with 2 buttons only (DOWN
Jun 10 22:07:41 <CIA-10> kexecboot: and SELECT).
Jun 10 22:07:42 <CIA-10> kexecboot: jay4mail master * rv0.5-174-g945e0d4 /gui.c: Draw frame around selected item in FB UI
Jun 10 22:07:46 <CIA-10> kexecboot: jay4mail master * rv0.5-173-g9023987 / (fb.c gui.c util.h): Remove obsolete uint{8,16,32} definitions from util.h
Jun 10 22:07:46 <CIA-10> kexecboot: jay4mail master * rv0.5-176-gf38f07d / (gui.c gui.h kexecboot.c): Clear screen on quit from FB UI
Jun 10 22:10:15 <Jay7> ant__: I've tested this all on x86
Jun 10 22:10:28 <Jay7> but as you know.. arm may have issues :)
Jun 10 22:10:38 <ant__> I'll try soon
Jun 10 22:11:13 <ant__> I'm answering max mailing (klibc)
Jun 10 22:13:47 <Jay7> ant__: I hope vfat and jffs will go upstream :)
Jun 10 22:14:06 <Jay7> btw, ask him about ubifs detection in fstype ;)
Jun 10 22:18:12 <ant__> heh
Jun 10 22:18:26 <ant__> this time it looks like patch goes upstream
Jun 10 22:18:35 <ant__> (jffs2 + vfat)
Jun 10 22:22:26 <Jay7> btw, I've understand how to deal with ULONGLONG_MAX vs ULLONG_MAX
Jun 10 22:22:45 <Jay7> #ifndef ULLONG_MAX
Jun 10 22:22:55 <Jay7> #define ULLONG_MAX ULONGLONG_MAX
Jun 10 22:22:57 <Jay7> #endif
Jun 10 22:23:17 <Jay7> a bit hacky but should work
Jun 10 22:24:31 <whse-work> hmm.. didnt wait for usb.. http://pastie.org/private/6pactjec8y6exan8lhimra
Jun 10 22:25:05 <Jay7> whse-work: try to increase --enable-delay value
Jun 10 22:25:47 <whse-work> hmm
Jun 10 22:25:57 <Jay7> that's common problem now :(
Jun 10 22:26:02 <whse-work> right
Jun 10 22:26:13 <whse-work> well, USE_DELAY is 1 in the config.status
Jun 10 22:26:22 <whse-work> looking for what it thought to wait..
Jun 10 22:26:43 <whse-work>  ,,,
Jun 10 22:26:46 <Jay7> btw, check for /dev/input/* devices
Jun 10 22:26:49 <whse-work> use delay = 1..
Jun 10 22:26:51 <Jay7> in your initramfs
Jun 10 22:27:02 <whse-work> its not there..
Jun 10 22:27:13 <Jay7> /dev/event*?
Jun 10 22:27:13 <whse-work> remember, headless unit.
Jun 10 22:27:26 <whse-work> hmm, no.. but I an add it.
Jun 10 22:27:28 <Jay7> ah.. but then you can't control it..
Jun 10 22:28:09 <whse-work> 'splain
Jun 10 22:28:36 <Jay7> kexecboot listen for /dev/input/event* or /dev/event*
Jun 10 22:28:47 <whse-work> ah...
Jun 10 22:28:56 <whse-work> and that matters for headless.. why ?
Jun 10 22:28:58 <Jay7> i.e. you can't use menu from ssh
Jun 10 22:29:11 <Jay7> I have placeholders for that code
Jun 10 22:29:16 <Jay7> but no real code now
Jun 10 22:29:18 <whse-work> it doesnt have the stuff in it to use ssh
Jun 10 22:29:47 <Jay7> but if you just with to boot first kernel then you may try to live w/o input devices
Jun 10 22:29:57 <whse-work> should I be using mknod .. ?
Jun 10 22:30:31 <whse-work> either way, it's not having time to actually catch the usb drives..
Jun 10 22:30:35 <Jay7> no, it does checking for event device capabilities
Jun 10 22:30:54 <Jay7> just use --enable-delay=2 or 3
Jun 10 22:31:13 <whse-work> i beleive when i ran the command.. i did --enable-delay=15
Jun 10 22:31:13 <Jay7> one of Zauruses have it set to 2 to wait for microdrive
Jun 10 22:31:26 <whse-work> because most of the atached usb are on the other side of a hub..
Jun 10 22:31:36 <whse-work> pogoplugs have them built in..
Jun 10 22:31:36 <Jay7> then it should wait 15 seconds before scanning devices
Jun 10 22:31:43 <whse-work> right...
Jun 10 22:32:01 <whse-work> but I can't find any such mention of that face in config.status that configure produced
Jun 10 22:32:15 <Jay7> check config.h
Jun 10 22:32:21 <Jay7> there should be define
Jun 10 22:32:28 <whse-work> So, I need a /dev/event*?
Jun 10 22:32:29 <Jay7> USE_DELAY iirc
Jun 10 22:32:38 <Jay7> no, it should work w/o
Jun 10 22:32:50 <whse-work> use_delay 1
Jun 10 22:32:59 <whse-work> wtf, i sat you to 15..
Jun 10 22:33:04 * whse-work smacks
Jun 10 22:33:27 * whse-work re-makes
Jun 10 22:34:05 <whse-work> although I will have to drive home and power-cycle to try again.
Jun 10 22:35:11 <Jay7> don't forget about --enable-timeout
Jun 10 22:35:24 <Jay7> well, I hope delay will help
Jun 10 22:35:29 <whse-work> ditto
Jun 10 22:35:36 <whse-work> hopefully it actually catchs this time..
Jun 10 22:35:42 <whse-work> how 1=15 idk.
Jun 10 22:36:00 <whse-work> good up til that act though
Jun 10 22:39:53 * Jay7 -> sleep
Jun 10 22:43:45 <ant__> I sent you copy of the mail
Jun 10 22:52:41 <Jay7> got it and read
Jun 10 22:53:10 <Jay7> now actually sleep
Jun 10 23:00:39 <ant__> good night ;)
Jun 10 23:24:16 <WarheadsSE> http://pastie.org/private/pvacceqwj8j2cirbdvoloa
Jun 10 23:24:26 <WarheadsSE> well, i dont think it likes not having a tty?
Jun 10 23:24:32 <WarheadsSE> found the drives though..
Jun 10 23:27:30 <ant__> http://paste.debian.net/119493/
Jun 10 23:27:38 <WarheadsSE> even found the boot.cfg :)
Jun 10 23:27:42 <ant__> ^^ this was the minimalistic one
Jun 10 23:28:24 <ant__> did you enable uinput in kernel?
Jun 10 23:28:44 <WarheadsSE> uinput?
Jun 10 23:29:27 <WarheadsSE> << obvious answer
Jun 10 23:29:55 <WarheadsSE> what section :p
Jun 10 23:30:12 <WarheadsSE> uessing input device support
Jun 10 23:30:23 <ant__> sorry, thi sis for userspace
Jun 10 23:30:39 <WarheadsSE> mk
Jun 10 23:30:54 <ant__> check for CONFIG_INPUT=y
Jun 10 23:30:55 <WarheadsSE> so, what do i need to make it happy, or get it to understand ttyS0 is the tty...
Jun 10 23:31:08 <WarheadsSE> its on..
Jun 10 23:31:17 <WarheadsSE> generic input = on
Jun 10 23:31:23 <WarheadsSE> (in make menuconfig)
Jun 10 23:31:31 <ant__> /dev/tty        c       662     0       5       5       0       -       -       -
Jun 10 23:31:37 <ant__> in your device-table
Jun 10 23:31:48 <WarheadsSE> device.. table..
Jun 10 23:31:53 <ant__> or devtmpfs activated in kernel
Jun 10 23:31:55 <WarheadsSE> sorry, not *that* kernel aware
Jun 10 23:32:08 <WarheadsSE> pretty sure that is activated.. but ill double check
Jun 10 23:32:24 <WarheadsSE> perhaps event interface?
Jun 10 23:32:48 <ant__> this is for kb/touchscreen input
Jun 10 23:32:58 <WarheadsSE> right
Jun 10 23:33:03 <WarheadsSE> so, add keyboard?
Jun 10 23:33:11 <ant__> you should have compiled in kernel the driver for your serial console
Jun 10 23:33:20 <WarheadsSE> yeah, its in there
Jun 10 23:33:25 <WarheadsSE> but it doesnt seem to like it.
Jun 10 23:35:27 <ant__> do you have a correct kernel cmdline?
Jun 10 23:35:49 <WarheadsSE> console=ttyS0,115200
Jun 10 23:36:04 <WarheadsSE> what else do i need?
Jun 10 23:36:44 <ant__> does the console work with normal kernel?
Jun 10 23:36:52 <WarheadsSE> yup..
Jun 10 23:36:59 <WarheadsSE> just isnt recognized by kexecboot
Jun 10 23:37:15 <ant__> http://paste.debian.net/119493/
Jun 10 23:37:26 <ant__> those devices are needed
Jun 10 23:38:02 <ant__> are you running stand-alone or in the initramfs now?
Jun 10 23:38:05 <WarheadsSE> so pre-populate node /dev/tty ?
Jun 10 23:38:10 <WarheadsSE> initramfs
Jun 10 23:38:17 <ant__> ok, populate the cpio
Jun 10 23:38:19 <WarheadsSE> it doesnt recognize it in standalone either
Jun 10 23:39:12 <WarheadsSE> so i need /dev/tty
Jun 10 23:39:22 <WarheadsSE> and console.. to be safe..
Jun 10 23:39:36 <ant__> either you populate it by hand or you let devtmpfs do this job
Jun 10 23:39:41 <WarheadsSE> thats
Jun 10 23:39:43 <WarheadsSE> hmm
Jun 10 23:39:54 <WarheadsSE> well it's auto recognizing all the drives, they arent populated..
Jun 10 23:40:02 <ant__> I've tested both and there is no speed diff
Jun 10 23:41:06 <WarheadsSE> yeah, i'm just saying
Jun 10 23:41:18 <WarheadsSE> added tty to the initramfs
Jun 10 23:41:21 <WarheadsSE> ttyS0 was already there
Jun 10 23:42:00 <WarheadsSE> re-gen uImage.
Jun 10 23:42:56 <WarheadsSE> gah. have to go for now.
Jun 10 23:42:58 <WarheadsSE> shall return
Jun 10 23:43:57 <ant__> try this
Jun 10 23:43:58 <ant__> http://paste.debian.net/119494/
Jun 10 23:44:21 <ant__> kexecboot will destroy and recreate the device nodes, though
Jun 10 23:44:54 <WarheadsSE> + can't open directory '/dev/input': No such file or directory
Jun 10 23:44:54 <WarheadsSE>  and nothing.. again
Jun 10 23:45:00 <ant__> remember devtmpfs *must* be mounted by hand in the case of initramfs
Jun 10 23:45:08 <WarheadsSE> hrmm
Jun 10 23:45:10 <WarheadsSE> g2g
Jun 10 23:45:22 <ant__> we trested bith way and there was no significant advantage
Jun 10 23:45:27 <ant__> bye
Jun 10 23:46:04 <ant__> /dev/input      d       755     0       0       -       -       -       -       -
Jun 10 23:46:04 <ant__> /dev/input/event c      660     0       0       13      64      0       1       8
Jun 10 23:46:09 <ant__> is what you lack
Jun 10 23:47:10 <ant__> and   FS ext4dev is not supported by kernel
Jun 10 23:47:30 <ant__> again, compile all in kernel, not as module
Jun 10 23:48:08 <ant__> good night
Jun 11 01:49:10 <WarheadsSE_> hmmm....
Jun 11 01:49:23 <WarheadsSE_> even with input/event it still blech
Jun 11 02:19:35 <WarheadsSE_> cfg.parser.c, doesnt account for ttySX in set_ttydev
Jun 11 02:19:41 <WarheadsSE_> THAT is why it fails.
**** ENDING LOGGING AT Sat Jun 11 02:59:57 2011