**** BEGIN LOGGING AT Wed Jan 20 02:59:57 2010
Jan 20 03:19:31 <cshore> cp: cannot stat `/Users/marca56/openwrt/staging_dir/toolchain-mips_gcc-4.3.3+cs_uClibc-0.9.30.2/lib/libc-0.9.30.2.so': No such file or directory
Jan 20 03:19:32 <cshore> cp: cannot stat `/Users/marca56/openwrt/staging_dir/toolchain-mips_gcc-4.3.3+cs_uClibc-0.9.30.2/lib/libuClibc.so.*': No such file or directory
Jan 20 03:19:41 <cshore> on Mac...any clues?
Jan 20 03:57:31 <swalker> updated openwrt/upstream, http://pastehtml.com/view/5t1aevv.html
Jan 20 06:22:09 <swalker> thepeople, Bartman007: is buildbot.openwrt.org making a comeback?
Jan 20 06:43:54 <CIA-88> swalker * r19236 /packages/utils/nano/Makefile: [packages] nano: update to 2.2.2
Jan 20 07:06:11 <CIA-88> juhosg * r19237 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: fix port->flags calculation in rtl8366_get_ports
Jan 20 07:06:14 <CIA-88> juhosg * r19238 /trunk/package/swconfig/ (Makefile src/cli.c): swconfig: show tagged port flag
Jan 20 07:06:17 <CIA-88> juhosg * r19239 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: move variable initialization out from the loop
Jan 20 07:06:19 <CIA-88> juhosg * r19240 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: fix return codes in rtl8366s_reset_vlan
Jan 20 09:30:27 <CIA-88> juhosg * r19241 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: fix parameter checking in rtl8366_attr_get_vlan_info
Jan 20 09:30:33 <CIA-88> juhosg * r19242 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: fix error checking in rtl8366s_set_vlan_4k_entry
Jan 20 09:30:37 <CIA-88> juhosg * r19243 /trunk/target/linux/ar71xx/ (7 files in 5 dirs): rtl8366_smi: add Kconfig option for the debugfs stuff
Jan 20 12:06:37 <Lalloso> I'm still looking for information about the best way to manage the filesystem on an access gateway with openwrt
Jan 20 12:06:51 <Lalloso> if i have let's say 64mb flash and 64mb ram
Jan 20 12:07:01 <Lalloso> what is the best practice?
Jan 20 12:07:15 <Lalloso> to use squashfs on the flash for /
Jan 20 12:07:29 <Lalloso> and tmpfs on ram with tmpfs for some other directories?
Jan 20 12:08:12 <Lalloso> is there a tutorial somewhere on the forum concerning how mini_fo fits into place?
Jan 20 12:08:28 <Lalloso> I have found this up to now http://wiki.openwrt.org/oldwiki/faq#how.do.i.edit.files.on.the.squashfs.image
Jan 20 13:26:23 <CIA-88> juhosg * r19244 /trunk/target/linux/ar71xx/image/Makefile: ar71xx: rename initramfs u-boot images
Jan 20 13:26:25 <CIA-88> juhosg * r19245 /trunk/target/linux/ar71xx/image/Makefile: ar71xx: create special initramfs images for the boards which does not accept standard uImages
Jan 20 13:46:05 <Lalloso^> does anybody know whisch script create the writable jffs2 partition to be used in overlay together with squashfs minifo?
Jan 20 13:59:32 <rtz> Lalloso: there is mount_rout and firstboot in /sbin and preinit in /etc
Jan 20 13:59:51 <rtz> Lalloso: or do you mean during image generation?
Jan 20 14:02:23 <Lalloso^> rtz thanks a lot for your support
Jan 20 14:02:47 <Lalloso^> i mean during image generation i usually create a filename in squasfs compressed lzma format
Jan 20 14:02:54 <Lalloso^> i flash the flash memory with this image
Jan 20 14:03:00 <Lalloso^> then what happens? :)
Jan 20 14:03:38 <rtz> the jffs2 partition isn't created during image creation
Jan 20 14:03:40 <Lalloso^> when i boot there is the firstboot and mount_root scripts which create the jffs2 partition at the and of squashfs partition on the flash?
Jan 20 14:03:50 <rtz> it works like this:
Jan 20 14:04:30 <rtz> during image generation, there are placed some special markers at the end of the image at some special places
Jan 20 14:05:44 <Lalloso^> and then? :-)
Jan 20 14:05:50 <rtz> and the firstboot script (or maybe mount_root, I'm not sure), scans for those markers and if it finds them, starts formatting the flash from the jffs2 start marker to the end of the partition
Jan 20 14:06:10 <rtz> take a look at <platfrom>/image/Makefile
Jan 20 14:06:22 <Lalloso^> okay wonderful
Jan 20 14:06:32 <Lalloso^> then if i edit a 1000 lines long file
Jan 20 14:06:38 <Lalloso^> and change 1 line only
Jan 20 14:06:50 <Lalloso^> on the jffs2 only the "delta" would be written or the whole file?
Jan 20 14:06:59 <jow_laptop> whole file
Jan 20 14:07:27 <Lalloso^> I see
Jan 20 14:08:00 <Lalloso^> is there in theory the possibility to improve this mini_fo writing only the "delta"
Jan 20 14:08:18 <Lalloso^> or the gain in write cycle would be minimized by the loss in reading?
Jan 20 14:08:46 <rtz> Lalloso: it should be possible
Jan 20 14:09:03 <Lalloso^> yeah i think so
Jan 20 14:09:18 <Lalloso^> i was wondering if this is a good idea but very complex to realize in practice
Jan 20 14:09:26 <Lalloso^> or a bad idea in the first place
Jan 20 14:09:35 <jow_laptop> it is complex to realize
Jan 20 14:09:41 <jow_laptop> what about binary files?
Jan 20 14:09:49 <rtz> but this will be lot's of work, because you will get lots of very small chunks to store and flash is not the best solution for this
Jan 20 14:09:55 <Lalloso^> reading a modified file would imply to read some part of that file from the squashfs and some part from the jffs2
Jan 20 14:10:01 <rtz> jow_laptop: there are binary xdelta algorithms
Jan 20 14:10:11 <jow_laptop> sure there are
Jan 20 14:10:14 <rtz> Lalloso: that's not the problem
Jan 20 14:10:21 <Lalloso^> jow_laptop i think it's similiar to the problem of binary files on svn?
Jan 20 14:10:27 <Lalloso^> i'm sorry i'm not expert at all
Jan 20 14:10:38 <rtz> storing the stuff efficiently on flash is much harder
Jan 20 14:10:43 <Lalloso^> openwrt is a treasure of knowledge for me and you have a lot of patience :-)
Jan 20 14:13:52 <Lalloso^> i know have an evanluation bord with a nor flash
Jan 20 14:14:04 <Lalloso^> in the end i'm going to have a nand flash
Jan 20 16:19:54 <Lalloso^> rtz: i'm looking at several platform/image/Makefile examples
Jan 20 16:20:05 <Lalloso^> what should i look for concerning mini_fo ?
Jan 20 16:23:41 <rtz> Lalloso: look at this: https://dev.openwrt.org/browser/trunk/target/linux/rdc/image/Makefile
Jan 20 16:24:12 <rtz> Lalloso: the section define Image/Postprocess/sitecom/squashfs
Jan 20 16:24:55 <Lalloso^> it's just 4 lines
Jan 20 16:25:37 <rtz> Lalloso: yes ;)
Jan 20 16:25:43 <rtz> and there: https://dev.openwrt.org/browser/trunk/include/image.mk
Jan 20 16:25:49 <rtz> it adds the markers
Jan 20 16:27:36 <rtz> the jffs2 formatting code picks one of the markers, depending on the block size
Jan 20 16:29:43 <Lalloso^> okay thx!
Jan 20 16:30:55 <rtz> quite easy actually, at least if the image is written 1:1 to flash
Jan 20 16:31:09 <Lalloso^> what do you mean 1:1?
Jan 20 16:33:25 <rtz> for sitecom, the image is written to the flash as it is and the bootloader is really stupid and has no idea about partitions and stuff
Jan 20 16:33:59 <rtz> sometimes you have headers which are stripped or partition tables handled by the bootloader for example
Jan 20 16:36:17 <Lalloso^> ah i see
Jan 20 16:36:22 <Lalloso^> but just to be sure i've understood
Jan 20 16:36:41 <Lalloso^> if i have a partitioning scheme like mtd0 for u-boot
Jan 20 16:36:57 <Lalloso^> mtd1 for kernel, filesystem and environment
Jan 20 16:37:10 <Lalloso^> this system won't work right?
Jan 20 16:37:36 <rtz> depends
Jan 20 16:37:39 <Lalloso^> i need a dedicated mtd partition for just the filesystem otherwise other things in the partition besides the image will be erased
Jan 20 16:38:33 <rtz> 1. what's enviromenent?
Jan 20 16:38:36 <Lalloso^> mini_fo seems a bit dated http://www.denx.de/wiki/Know/MiniFOHome
Jan 20 16:38:46 <Lalloso^> it's a linux_ifx_easy if i have understood well
Jan 20 16:38:50 <Lalloso^> something from infineon
Jan 20 16:39:08 <Lalloso^> it has a proprietary toolchain used inside openwrt
Jan 20 16:40:01 <Lalloso^> I think i'll end up customizing the image.mk to my needs
Jan 20 16:40:36 <rtz> I meant the enviroment partition ;)
Jan 20 16:40:45 <Lalloso^> at the moment the kernel is loaded in to ram from a specifc location of the flash which is outside the image file but on the same partition
Jan 20 16:41:03 <Lalloso^> you mean the filesystem? what do you mean environment?
Jan 20 16:41:06 <Lalloso^> i feel so noob
Jan 20 16:41:18 <rtz> mtd1 for kernel, filesystem and environment
Jan 20 16:41:21 <Lalloso^> it's an 8MB NOR flash
Jan 20 16:41:25 <Lalloso^> ah i see
Jan 20 16:41:25 <Lalloso^> sorry
Jan 20 16:41:39 <Lalloso^> environment are the environment variables used by the boot loader
Jan 20 16:41:52 <rtz> ok
Jan 20 16:41:52 <Lalloso^> such as "bootargs" and things like those if i have understood well
Jan 20 16:43:11 <rtz> 2. find the your platform setup code
Jan 20 16:43:36 <rtz> it's somewhere in the kernel files or patches directory for you platform
Jan 20 16:44:00 <rtz> I need especially the code handeling partitions
Jan 20 16:44:53 <Lalloso^> you mean some mtd.c somewhere?
Jan 20 16:45:09 <rtz> that's one possibility
Jan 20 16:45:37 <rtz> for sitecom it's in arch/x86/mach-rdc/platform.c
Jan 20 16:45:59 <Lalloso^> but what's at the end of the tunnel? i think about repartitioning the flash to have the image on a stand alone partition in order to avoid erasing of kernel and environment
Jan 20 16:46:11 <rtz> 3. you have to understand a) the image format and b) what exactly u-boot does with it
Jan 20 16:46:22 <Lalloso^> now it works with squashfs which being read-only has no cohexistence problem with kernel and environment
Jan 20 16:46:27 <rtz> Lalloso: why don't you want to erase the kernel?
Jan 20 16:46:39 <Lalloso^> because if i erase the kernel
Jan 20 16:46:56 <Lalloso^> it seems that upon next boot u-boot will complain something about CRC error and will ask for a new kernel via TFTP
Jan 20 16:47:12 <rtz> that's right
Jan 20 16:47:14 <Lalloso^> i tried setting verify n as environment variable but it did not work
Jan 20 16:47:26 <rtz> so you have to flash a new kernel
Jan 20 16:47:26 <Lalloso^> it's very interesting to me maybe i'm too nerd :)
Jan 20 16:47:46 <Lalloso^> yes afaik it's the kernel that after being loaded into ram via "bootm"
Jan 20 16:47:52 <Lalloso^> will mount the root filesystem
Jan 20 16:48:08 <Lalloso^> and mess with the partition erasing a bit of itself also :D
Jan 20 16:49:02 <rtz> this happens, if you flash the image?
Jan 20 16:54:00 <Lalloso^> yes and no
Jan 20 16:54:07 <Lalloso^> i have not the mini_fo intoplace
Jan 20 16:54:27 <Lalloso^> i've just a jffs2 image which it seems to write randomly in the partition
Jan 20 16:54:32 <Lalloso^> upon startup
Jan 20 16:54:49 <rtz> hmmm
Jan 20 16:55:02 <rtz> what device do you have? (and which platform)
Jan 20 16:55:04 <Lalloso^> it complains about the magic of the filesystem and things like those but I think it's looking into sectors not assigned to her
Jan 20 16:55:12 <Lalloso^> it's an evaluation board mips
Jan 20 16:55:19 <Lalloso^> infineon afaik
Jan 20 16:55:26 <Lalloso^> without fpu :-\
Jan 20 16:55:38 <rtz> yes, but what are you using?
Jan 20 16:55:55 <rtz> there are a number of mips platforms in openwrt
Jan 20 16:56:03 <Lalloso^> good question
Jan 20 16:56:13 <rtz> ifxmips?
Jan 20 16:56:20 <Lalloso^> linux-ifx-easy ?
Jan 20 16:56:34 <rtz> that's not in openwrt :/
Jan 20 16:56:45 <Lalloso^> ah ha!
Jan 20 16:56:51 <Lalloso^> so it's something proprietary!!!
Jan 20 16:57:22 <Lalloso^> i will repartition the mtd and see what happens if I allocate a full partition to just the jffs2 image
Jan 20 16:57:32 <Lalloso^> then I'll work to have the mini_fo in the picture
Jan 20 16:58:04 <Lalloso^> i also have a router station pro
Jan 20 16:58:08 <Lalloso^> from ubiquity networks
Jan 20 16:58:15 <rtz> Lalloso: start with building an openwrt image, if you want help for openwrt ;)
Jan 20 16:58:16 <Lalloso^> and it has mini_fo on it!!!!!
Jan 20 16:58:26 <rtz> they use openwrt
Jan 20 16:58:30 <Lalloso^> rtz: i already build the image with openwrt
Jan 20 16:58:43 <Lalloso^> it's just that it works with an external toolchain
Jan 20 16:59:05 <rtz> why didn't you use the default internal toolchain?
Jan 20 16:59:09 <Lalloso^> target system in make menuconfig
Jan 20 16:59:24 <Lalloso^> states Infineon EASY50812
Jan 20 16:59:39 <Lalloso^> i think because this "chipset" needs a special toolchain
Jan 20 16:59:58 <Lalloso^> but I don't want to waste your time anymore
Jan 20 17:00:01 <rtz> Lalloso: I doubt it
Jan 20 17:00:10 <Lalloso^> ah
Jan 20 17:00:22 <rtz> and which platform did you use in menufconfig?
Jan 20 17:00:26 <Lalloso^> but I think that the kernel have been patched
Jan 20 17:00:34 <Lalloso^> by Infineon and it's not a vanilla
Jan 20 17:01:03 <Lalloso^> what is "platform" ? :)
Jan 20 17:01:05 <jow_laptop> I'd try using the ifxmips target
Jan 20 17:01:31 <rtz> Lalloso: the very first option in menuconfig
Jan 20 17:01:32 <Lalloso^> jow_laptop: you stay silent but you read somehow in the background eh? :-)
Jan 20 17:01:46 <Lalloso^> it's Infineon EASY50812
Jan 20 17:01:52 <jow_laptop> Lalloso^: I'm at work and only look here from time to time
Jan 20 17:01:54 <rtz> Lalloso: and openwrt is also heavyly patched
Jan 20 17:02:22 <rtz> Lalloso: make a screenshot of your menuconfig screen
Jan 20 17:02:24 <Lalloso^> i trust the openwrt patches more
Jan 20 17:02:59 <Lalloso^> where do i put it?
Jan 20 17:03:19 <rtz> find a free image hoster
Jan 20 17:03:24 <jow_laptop> imagebin.ca
Jan 20 17:03:26 <rtz> imagehack or something
Jan 20 17:04:30 <Lalloso^> however if i've understood well
Jan 20 17:04:38 <Lalloso^> openwrt now has the ability to work with an external toolchain
Jan 20 17:04:45 <Lalloso^> not compiled inside openwrt
Jan 20 17:04:54 <Lalloso^> therefore it works well :-)
Jan 20 17:05:05 <rtz> hmmm
Jan 20 17:05:06 <Lalloso^> It's very interesting to talk with you
Jan 20 17:05:16 <jow_laptop> Lalloso^: Infineon was very supportive regarding OpenWrt, they even opened their driver sources and stuff
Jan 20 17:05:24 <jow_laptop> I see no point in using an external toolchain
Jan 20 17:05:29 <jow_laptop> the ifxmips target should work
Jan 20 17:05:39 <jow_laptop> it's arm-style processors as far as I know, nothing fancy
Jan 20 17:05:46 <Lalloso^> well
Jan 20 17:05:58 <Lalloso^> afaik it has at list some strange conception of endianess
Jan 20 17:06:06 <rtz> jow_laptop: arm != mips
Jan 20 17:06:09 <Lalloso^> it's something in the middle between big-endian and little-endian
Jan 20 17:06:30 <rtz> Lalloso: what version of openwrt are you using?!?
Jan 20 17:07:30 <Lalloso^> it's the trunk of sometime ago
Jan 20 17:07:37 <Lalloso^> about 18250 revision
Jan 20 17:07:56 <Lalloso^> however i'll consider using the ifxmips target
Jan 20 17:08:11 <Lalloso^> you have given me a lot of subjects and hints
Jan 20 17:08:22 <rtz> Lalloso: and check out a current version of trunk
Jan 20 17:08:27 <Lalloso^> I think I now need some time to further investigate these possibilities
Jan 20 17:08:38 <rtz> I'm not sure what you have there, but it isn't standard openwrt
Jan 20 17:08:39 <Lalloso^> and come back to you with more questions which make sense
Jan 20 17:09:10 <Lalloso^> okay
Jan 20 17:09:40 <Lalloso^> i'll consider this also but I'm not the only one playing with this project as you may have understood from my very limited understanding :)
Jan 20 17:11:35 <rtz> as I said, first step is using a standard trunk version
Jan 20 17:11:43 <jow_laptop> rtz: right, I confused ifxmips with ixp4xx
Jan 20 17:13:10 * lupine_85 is getting source code for his ADSL2+ card :)
Jan 20 17:13:12 <lupine_85> rock on
Jan 20 17:20:26 <Lalloso^> damn one must be quick with nickserv
Jan 20 17:40:41 <CIA-88> juhosg * r19246 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: coding style cleanup
Jan 20 20:36:23 <CIA-88> juhosg * r19247 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: minor code optimalizations
Jan 20 20:36:26 <CIA-88> juhosg * r19248 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: simplify rtl8366_{get,set}_port_vlan_index functions
Jan 20 20:36:28 <CIA-88> juhosg * r19249 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: make rtl8366_phy_config_{aneg,init} static
Jan 20 20:36:32 <CIA-88> juhosg * r19250 /trunk/target/linux/ar71xx/files/drivers/net/phy/rtl8366_smi.c: rtl8366_smi: simplify rtl8366_attr_get_port_link function
Jan 20 23:30:18 <cshore> I have question about the x86 port.  I planning on doing a VM to test my preinit/firstboot changes on a different architechture than my routers (which are brcm63xx).  Can I just use the generated images as a disk image, or how do I 'flash' the firmware?
Jan 20 23:30:46 <xMff> depends on the environment it should run in
Jan 20 23:31:03 <cshore> qemu
Jan 20 23:31:21 <cshore> well qemu-kvm actually
Jan 20 23:31:23 <xMff> http://wiki.openwrt.org/doc/howto/qemu
Jan 20 23:31:47 <xMff> qemu should be able to use the .image ... image directly
Jan 20 23:32:09 <cshore> ok, so that's generated directly by the buildroot?
Jan 20 23:32:17 <xMff> yes
Jan 20 23:32:27 <cshore> awesome.  That should be easy then
Jan 20 23:32:28 <xMff> and use the ext2 one
Jan 20 23:32:38 <xMff> squashfs is problematic
Jan 20 23:32:49 <xMff> no free space for jffs2
Jan 20 23:33:26 <cshore> oh, that's part of what I need to test .... hmmm....can I dd seek (or is it skip) to extend the size of the image?
Jan 20 23:33:41 <xMff> I am not sure why it happens actually
Jan 20 23:34:00 <xMff> I think the image is capped to its content size somehow leaving no spare space after squashfs
Jan 20 23:34:19 <xMff> so the attempt to format jffs2 will result in attempt to access beyond end of device or so
Jan 20 23:34:51 <xMff> maybe you can convert it to a fixed-size image with qemu-img somehow
Jan 21 00:22:04 <Acinonyx> cshore: to have free space on jffs2, enable the option to pad jffs2 image to root partition size
Jan 21 00:26:40 <cshore> Acinonyx: thanks
Jan 21 00:28:58 <cshore> now for the ext2 version, does it use an initramfs?  That way I can test both jffs2 stuff and totally different boot system from what the routers I've been developing on, so that I can be sure I haven't broken anything
Jan 21 00:31:02 <cshore> xMff: I noticed some messages in my syslog after boot that makes me think there maybe a problem with the firewall patch we did (uci: illegal argument).  They weren't there before.  I haven't had a chance to look at it though
Jan 21 00:31:35 <Acinonyx> cshore: no, it doesn't use initramfs, IIRC
Jan 21 00:32:11 <Acinonyx> also, sysupgrade doesn't work on ext2
Jan 21 00:33:14 <cshore> Acinonyx: ok...do you know what it does use for mounting the rootfs, and how do you upgrade ext2?
Jan 21 00:37:02 <Acinonyx> it doesn't use initramfs
Jan 21 00:37:49 <cshore> ok, it must modify mount_root then
Jan 21 00:38:56 <cshore> or is the ext2 attached to the kernel image?
Jan 21 00:39:11 <cshore> or, wait, does it use grub?
Jan 21 00:39:19 <Acinonyx> the modules for ide and stuff is compiled in the kernel
Jan 21 00:39:45 <Acinonyx> it uses grub. the root fs is a seperate partition
Jan 21 00:40:09 <cshore> ok, that explains it...it'll be a good test for if I've broken non-jffs2 mounts then
**** ENDING LOGGING AT Thu Jan 21 02:59:56 2010