**** BEGIN LOGGING AT Sun Apr 01 02:59:57 2007
Apr 01 06:27:54 <sn9__> anyone around right now? i can't seem to find a place to add on to the CFLAGS for kernel compilation specific to a target
Apr 01 07:26:43 <sn9__> never mind; i think i found it
Apr 01 09:11:08 <[florian]> sn9__: ping ?
Apr 01 09:11:21 <sn9__> pong
Apr 01 09:11:27 <[florian]> great
Apr 01 09:11:50 <[florian]> actually, there really seems to be a kernel size limitation for rdc
Apr 01 09:12:01 <[florian]> which is not triggered with 2.4 builds, but with 2.6 it is
Apr 01 09:12:23 <sn9__> i'm trying to recompile redboot, because otherwise, a reflash is needed on each boot, and squashfs cannot be used
Apr 01 09:12:44 <[florian]> ok, this would be good for testing
Apr 01 09:12:55 <sn9__> i will see about that -- you have not seen what i have done with the flash map driver yet
Apr 01 09:12:58 <[florian]> once we have something stabilised, could be great to deal with the original redboot anyway
Apr 01 09:13:22 <[florian]> sn9__: I saw what you did but the flash map driver seems to close to the original one
Apr 01 09:13:40 <sn9__> no, you didn't
Apr 01 09:14:01 <sn9__> i have not submitted it yet
Apr 01 09:14:16 <[florian]> ok, so what I downloaded was nothing you modified ?
Apr 01 09:37:37 <sn9__> crap, looks like linksys further modified the sources after building the binay after all
Apr 01 09:38:01 <sn9__> no idea what they did
Apr 01 09:42:02 <sn9__> [florian]: i get the same error without ccache
Apr 01 09:42:49 <[florian]> sn9__: weird, which distribution do you use ?
Apr 01 09:43:24 <sn9__> you think buildroot generated a bad toolchain?
Apr 01 09:43:52 <CIA-3> florian * r6777 /trunk/target/linux/rdc-2.6/ (2 files in 2 dirs): Code cleanup
Apr 01 09:44:13 <[florian]> sn9__: I doubt so, it might come from a bad configuration of uclibc
Apr 01 09:44:53 <sn9__> btw, the line before the symbol error is: /sbin/init: symbKernel panic: Attempted to kill init!
Apr 01 09:45:12 <sn9__> oh, now i see
Apr 01 09:45:23 <sn9__> the bad symbol is in init
Apr 01 09:46:43 <sn9__> you modified the rdc3210.c file, and so did i. now "svn up" shows it as "C"
Apr 01 09:47:28 <CIA-3> florian * r6778 /trunk/target/linux/rdc-2.6/patches/001-rdc3210_flash_map.patch: the redboot parsing function needs to be exported
Apr 01 09:47:35 <[florian]> sn9__: in conflict, yes
Apr 01 09:47:42 <sn9__> very much
Apr 01 09:48:07 <sn9__> i did not back up my copy, and i forgot how to undo it
Apr 01 09:48:26 <[florian]> this is something I never understand with svn :/
Apr 01 09:49:04 <armijn> http://svnbook.red-bean.com/
Apr 01 09:49:31 <armijn> invaluable reference :)
Apr 01 09:49:38 <[florian]> yes it is well explained, but I am stupid, so I do not understand
Apr 01 09:55:48 <sn9__> ok, i got it back
Apr 01 09:56:21 <sn9__> i can't make sense of some of your cleanups yet
Apr 01 09:56:51 <[florian]> I first try to parse the redboot partition table if it fails, I use the static mapping defining
Apr 01 09:56:59 <[florian]> s/defining/defined/
Apr 01 09:57:11 <sn9__> but there is no redboot partition table
Apr 01 09:57:33 <[florian]> I wanted to test if there was one before
Apr 01 09:57:44 <[florian]> this is something I should have comitted yesterday
Apr 01 09:57:51 <sn9__> and if you use the static mapping table, no wonder you have no room for 2.6
Apr 01 09:58:40 <sn9__> i came up with a simpler approach
Apr 01 09:58:56 <[florian]> ok
Apr 01 09:59:27 <sn9__> you'll see it after i have a successful boot into 2.4
Apr 01 10:00:12 <nbd> xrg_: hi
Apr 01 10:00:21 <xrg_> Hi..
Apr 01 10:00:25 <[florian]> sn9__: can't wait, show me :p
Apr 01 10:00:37 <sn9__> ok, one moment
Apr 01 10:00:41 <nbd> xrg_: i wanted to start looking into your changes a bit more, but the website is down
Apr 01 10:00:50 <xrg_> Hm..
Apr 01 10:01:00 <sn9__> i'll re-diff against current svn
Apr 01 10:01:07 <[florian]> sn9__: I am bitching, take your time
Apr 01 10:01:10 <xrg_> Can you try once more, my dsl is unstable and changes IP all the time.
Apr 01 10:01:19 <sn9__> ok, time it is
Apr 01 10:01:58 <nbd> xrg_: actually the ip is correct, i just noticed i get connection refused ...
Apr 01 10:02:33 <xrg_> Nope, point is that IP's are given to other customers. You'll probably end up seeing an XP-machine there!
Apr 01 10:03:04 <nbd> xrg_: the ip that i get matches the one from your irc hostmask :P
Apr 01 10:03:05 <xrg_> Now it should read: ppp248-100.dsl.hol.gr
Apr 01 10:03:20 <nbd> and i get connection refused on that one
Apr 01 10:03:22 <xrg_> let me check then..
Apr 01 10:04:04 <[florian]> sn9__: I understand the kernel could be bigger than the space left for it, but it would fail while booting, the size limitation should not appear after flashing via redboot
Apr 01 10:04:13 <xrg_> D'OH!
Apr 01 10:04:34 * xrg_ has messed his apache..
Apr 01 10:04:36 <sn9__> [florian]: exactly; that's how i managed it in my modified driver
Apr 01 10:05:45 <nbd> [florian], sn9__: we should work on an lzma loader for x86
Apr 01 10:05:51 <nbd> if there isn't such a thing already...
Apr 01 10:05:59 <nbd> probably easy to do
Apr 01 10:06:03 <[florian]> nbd: yes, thought about that
Apr 01 10:06:14 <[florian]> nbd: I am not that confortable with x86 assembly :p
Apr 01 10:06:22 <nbd> just copy the bzImage code and replace the zlib compression with lzma
Apr 01 10:06:26 <sn9__> nbd: figuring out the bootloader issues are necessary anyway
Apr 01 10:06:31 <nbd> no need to write x86 assembly
Apr 01 10:06:39 <[florian]> nbd: ok
Apr 01 10:07:07 <nbd> sn9__: well, if it has static mapping and the kernel is too big, then it's rather clear why it's failing
Apr 01 10:07:07 <sn9__> i now understand why there was a space in the symbol name
Apr 01 10:07:31 <xrg_> "It works! "
Apr 01 10:07:37 <nbd> xrg_: thanks
Apr 01 10:07:42 <sn9__> stdout and stderr were getting interspersed
Apr 01 10:07:47 <xrg_> Thanks for noticing!
Apr 01 10:08:13 <nbd> xrg_: uh, i get an index page when i try to access openwrt.git
Apr 01 10:08:27 <xrg_> w8 ..
Apr 01 10:08:27 <sn9__> apparently busybox is simply using a symbol that uclibc is not exporting
Apr 01 10:13:52 <[florian]> sn9__: ok, strange
Apr 01 10:14:19 <sn9__> oh, wait
Apr 01 10:14:37 <sn9__> i have CONFIG_BUSYBOX_CONFIG_FEATURE_DEVFS set
Apr 01 10:14:43 <sn9__> is that bad?
Apr 01 10:15:54 <[florian]> it is not set here, but it should raise this symbol error
Apr 01 10:18:41 <[florian]> sn9__: do you by chance have the rdc3211 jtag pinout ?
Apr 01 10:18:48 <sn9__> nope
Apr 01 10:18:59 <[florian]> sn9__: so you will replace the bootloader with mtd ?
Apr 01 10:19:13 <sn9__> eventually
Apr 01 10:19:23 <[florian]> sounds pretty risky
Apr 01 10:21:03 <sn9__> oh, my mistake; it's not me who had that option set, but you
Apr 01 10:21:09 <sn9__> i had the option off
Apr 01 10:21:35 <[florian]> ah ah
Apr 01 10:23:09 <sn9__> i don't see the option in menuconfig
Apr 01 10:23:31 <[florian]> Base system -> Busybox / Configuration ->
Apr 01 10:23:54 <sn9__> and then?
Apr 01 10:24:26 <sn9__> which page from there?
Apr 01 10:25:24 <[florian]> Miscellaneous options
Apr 01 10:26:26 <sn9__> no such page
Apr 01 10:26:55 <[florian]> the configuration submenu is just under busybox in menuconfig
Apr 01 10:28:06 <sn9__> there is no "misc options" submenu there
Apr 01 10:29:22 <sn9__> there is "general options" but this option is not there
Apr 01 10:29:59 <[florian]> between Linux system utilities and Networking utilities ?
Apr 01 10:31:26 <sn9__> that's CONFIG_BUSYBOX_CONFIG_DEVFSD, not CONFIG_BUSYBOX_CONFIG_FEATURE_DEVFS
Apr 01 10:32:01 <sn9__> never mind; it's the next one
Apr 01 10:33:25 <[florian]> ...
Apr 01 10:38:09 <sn9__> btw, feel free to commit your change to the 2.6 kernel config regarding pci
Apr 01 10:39:13 <[florian]> will do in few seconds
Apr 01 10:42:31 <sn9__> ...
Apr 01 10:43:58 <[florian]> I do not remember if the rdc6040 ethernet driver implements napi polling
Apr 01 10:45:16 <CIA-3> florian * r6779 /trunk/target/linux/rdc-2.6/config/default: Working configuration, minimalistic
Apr 01 11:06:34 <CIA-3> florian * r6780 /trunk/target/linux/rdc-2.6/image/Makefile: Restore image creation for rdc boards
Apr 01 11:08:27 <CIA-3> florian * r6781 /trunk/tools/firmware-utils/src/airlink.c: Fix image generation (sn9)
Apr 01 11:11:16 <sn9__> i wasn't finished with that file, either. you'll get a bigger patch
Apr 01 11:12:24 <[florian]> ok
Apr 01 11:12:39 <[florian]> I currently boot, but I am not able to mount the rootfs :/
Apr 01 11:17:48 <sn9__> i can
Apr 01 11:21:58 <[florian]> I will probably need your flash map changes
Apr 01 11:22:25 <sn9__> my flash map changes require makefile changes
Apr 01 11:22:53 <sn9__> otherwise i'd have already given them to you
Apr 01 11:23:18 <[florian]> sn9__: which makefile ?
Apr 01 11:23:49 <sn9__> top target makefile and image makefile
Apr 01 11:24:15 <[florian]> why modify the top target ?
Apr 01 11:26:07 <sn9__> you'll see
Apr 01 11:30:40 <CIA-3> mbm * r6782 /x/: (null)
Apr 01 11:30:55 <[mbm]> hmm
Apr 01 11:31:53 <CIA-3> mbm * r6783 /x/:
Apr 01 11:43:13 <CIA-3> nbd * r6784 /trunk/toolchain/uClibc/patches/150-portability.patch:
Apr 01 11:43:13 <CIA-3> Fix portability patch for uClibc.
Apr 01 11:43:13 <CIA-3> Certain versions of GNU cp treat -P as --parent, which is wrong here and -P should be implied by -R already.
Apr 01 11:43:13 <CIA-3> Fixes build on OSX with fink's fileutils installed.
Apr 01 11:57:26 <sn9__> nbd: you know, you could have just said "/bin/cp" and bypassed fink altogether
Apr 01 11:57:51 <nbd> that would be a hack
Apr 01 11:59:33 <nbd> since there might be other distros that ship a /bin/cp that behaves the same way
Apr 01 12:00:22 <CIA-3> nico * r6785 /packages/net/dircproxy/ (. Makefile): add dircproxy (closes: #1436, thanks to Martin Matuska)
Apr 01 12:02:06 <[florian]> {Nico}: ping ?
Apr 01 12:02:14 <{Nico}> pong
Apr 01 12:02:27 <[florian]> {Nico}: how are you ?
Apr 01 12:03:49 <{Nico}> i'm fine, but really tired (all work and no play)
Apr 01 12:03:55 <{Nico}> whate about you?
Apr 01 12:04:09 <[florian]> {Nico}: I am fine, had more time to play those last weeks
Apr 01 12:04:27 <[florian]> {Nico}: have you tested the ar7 target on conceptronic c54apr{a,b} ?
Apr 01 12:05:28 <{Nico}> [florian]: not yet, using a dlink dsl-524t atm
Apr 01 12:05:37 <[florian]> ok
Apr 01 12:05:42 <[florian]> is it adam2 based ?
Apr 01 12:08:35 <{Nico}> yep
Apr 01 12:10:57 <[florian]> I can't remember what I flashed last time to the c54aprb :p
Apr 01 12:28:58 <CIA-3> nbd * r6786 /trunk/package/iptables/Makefile: fix iptables extension issue with mac os x builds
Apr 01 12:32:59 <sn9__> well, i've determined that the linksys redboot really is a later version, with only cosmetic changes
Apr 01 12:33:47 <sn9__> busybox still has that unresolved symbol in libc
Apr 01 12:57:23 <sn9__> get your red-hot replacement redboot at http://homepage.mac.com/danielg4/redboot.patch and http://homepage.mac.com/danielg4/redboot.bin !
Apr 01 13:04:48 <armijn> :o
Apr 01 13:04:49 <armijn> spam!
Apr 01 13:05:35 <CIA-3> nico * r6787 /packages/libs/libdnet/Makefile: disable check (and python while we're at it) in configure (closes: #1536)
Apr 01 13:24:52 <CIA-3> nbd * r6788 /packages/net/batman/Makefile: fix batman build on osx
Apr 01 13:36:05 <CIA-3> nbd * r6789 /packages/net/bind/Makefile: fix bind compile on osx
Apr 01 13:37:36 <[florian]> sn9__: I have overriden the kernel command line here
Apr 01 13:43:26 <[florian]> armijn: wanna see something fun ?
Apr 01 13:44:13 <[florian]> armijn: http://rafb.net/p/HTSLVu17.html
Apr 01 13:44:16 <[florian]> armijn: c54aprb
Apr 01 13:44:26 <[florian]> c54apra sorr
Apr 01 13:44:29 <[florian]> y
Apr 01 13:46:36 <armijn> [florian]: what's funny about it?
Apr 01 13:47:07 <[florian]> armijn: because it works, and you just asked what was the status of ar7 on those devices
Apr 01 13:47:12 <armijn> ok
Apr 01 13:47:13 <[florian]> armijn: it just works with squashfs
Apr 01 13:47:18 <armijn> sorry, am a bit slow today
Apr 01 13:47:31 <[florian]> armijn: and with jffs2 actually, but the kernel command line needs to be changed for that to work
Apr 01 13:48:17 <armijn> [florian]: so, can I install it via the webinterface already? :P
Apr 01 13:50:18 <[florian]> armijn: not sure about that, but it should not be that hard
Apr 01 13:51:30 <armijn> well, I can always try
Apr 01 13:52:54 <[florian]> flash the generic openwrt-ar7-2.6-squashfs.bin file
Apr 01 13:53:06 <[florian]> and wait a minute for me to add wifi by default
Apr 01 13:53:55 <armijn> oh, not doing it now
Apr 01 13:54:17 <armijn> power supply for the unit I have here is at my office
Apr 01 14:03:40 <[florian]> nbd: have you planned on extending kmod-swtich to ic+ 175x ?
Apr 01 14:05:33 <CIA-3> nico * r6790 /packages/net/dovecot/ (7 files in 3 dirs): add dovecot (closes: #1353)
Apr 01 14:07:01 <CIA-3> nbd * r6791 /trunk/tools/firmware-utils/src/mksyshdr.c: convert from dos to unix format
Apr 01 14:09:05 <ejka> [florian]: ic+ 175 can only do port-based vlans
Apr 01 14:11:17 <CIA-3> florian * r6792 /trunk/target/linux/ar7-2.6/profiles/ (100-Texas.mk 110-None.mk): Add wifi profiles for ar7-2.6
Apr 01 14:11:50 <[florian]> ejka: humm ok, so not really interesting
Apr 01 14:12:43 <CIA-3> nbd * r6793 /trunk/include/scan.mk: add check for openwrt build system makefiles when scanning for metadata
Apr 01 14:14:54 <[florian]> ejka: which firmware files do you load for acx ?
Apr 01 14:15:33 <[florian]> ejka: I renamed ACX100.bin to tiacx111c16, seems to block the device
Apr 01 14:17:34 <nbd> ejka: does that chip require nasty hacks to the ethernet driver like the marvell chip?
Apr 01 14:18:19 <ejka> [florian]: http://195.66.192.167/linux/acx_patches/fw/
Apr 01 14:18:27 <nbd> the marvell chip requires a non-standard trailer to outgoing packets that specifies the port
Apr 01 14:18:39 <ejka> I definitely tested 1.2.x and 2.1.x
Apr 01 14:18:53 <ejka> 2.3.x even
Apr 01 14:18:56 <ejka> nbd: no
Apr 01 14:19:14 <nbd> ah, good
Apr 01 14:19:17 <ejka> but port-based only, so all traffic goes to cpu, once you enable vlans
Apr 01 14:19:24 <nbd> hm
Apr 01 14:19:30 <nbd> sucks
Apr 01 14:20:00 <ejka> I also have ar7 device with adm6996, which has it's registers mapped to phy-space
Apr 01 14:20:35 <ejka> so probably will add support for it to adm6996 switch driver one day
Apr 01 14:20:51 <[florian]> ejka: any major difference between ic+174a/c ?
Apr 01 14:21:01 <[florian]> ejka: or is it just a revision letter ?
Apr 01 14:21:06 <ejka> [florian]: shouldn't be
Apr 01 14:21:46 <[florian]> ejka: thanks for the url, it justs works fine :)
Apr 01 14:35:25 <[florian]> sn9__: ping ?
Apr 01 14:36:26 <CIA-3> florian * r6794 /trunk/package/kernel/modules/i2c.mk: Fix i2c module dependencies (#1525)
Apr 01 14:39:35 <CIA-3>  * r6795 /: remove the secret rootkit (requested by the OpenWRT developers)
Apr 01 14:40:04 <[florian]> lol
Apr 01 15:06:18 <CIA-3> nbd * r6796 /packages/net/cups/Makefile: fix cups build on osx
Apr 01 15:26:02 <CIA-3> nbd * r6797 /packages/net/dhcp/ (Makefile patches/ patches/100-portability.patch): fix dhcp build on osx
Apr 01 15:43:41 <CIA-3> nbd * r6798 /packages/libs/libnet-1.0.x/ (4 files in 2 dirs): fix libnet-1.0.x build on osx
Apr 01 16:02:40 <{Nico}> nbd: can you backport https://dev.openwrt.org/changeset/6571 to the WhiteRussian sdk?
Apr 01 16:03:22 <nbd> yeah
Apr 01 16:03:24 <[florian]> sn9__: ping ?
Apr 01 16:48:08 <CIA-3> nbd * r6799 /packages/libs/libnet-1.1.x/ (4 files in 2 dirs): fix libnet build on osx
Apr 01 17:10:05 <CIA-3> florian * r6800 /trunk/target/linux/rdc-2.6/image/Makefile: Generate squashfs images as well
Apr 01 17:18:26 <[florian]> nbd: any idea how to get rid of these jffs2 messages ? : http://rafb.net/p/GgMFGk43.html
Apr 01 17:18:52 <CIA-3> florian * r6801 /trunk/target/linux/adm5120-2.6/config/default: Enable MTD_BLOCK
Apr 01 17:20:04 <nbd> [florian]: are you sure the rootfs got flashed?
Apr 01 17:20:29 <[florian]> nbd: I used the following cfe command : flash -noheader 192.168.2.4:openwrt-adm5120-2.6-jffs2-wp54g-wrt.trx flash1.trx
Apr 01 17:20:32 <[florian]> so I think so
Apr 01 17:21:54 <CIA-3> florian * r6802 /trunk/target/linux/rdc-2.6/config/default: Improve kernel configuration
Apr 01 17:22:34 <[florian]> nbd: I have the exact same problem with rdc
Apr 01 17:23:22 <nbd> is all the code to generate this stuff committed in svn?
Apr 01 17:23:46 <[florian]> yes it is
Apr 01 17:23:52 <nbd> ok, i'll check it out
Apr 01 17:24:07 <[florian]> nbd: focus on adm5120, rdc is a bit broken currently :)
Apr 01 17:26:00 <nbd> what is the erase block size of that device?
Apr 01 17:26:17 <[florian]> 64k since it is a 4Mb flash
Apr 01 17:28:06 <nbd> not every 4mb flash chip has 64k blocksize
Apr 01 17:28:21 <[florian]> this is what compex originally does
Apr 01 17:28:58 <nbd> ok
Apr 01 17:29:49 <nbd> does squashfs work?
Apr 01 17:29:56 <[florian]> have not tested it yet
Apr 01 17:30:04 <nbd> please do
Apr 01 17:30:05 <[florian]> I will have a try
Apr 01 17:40:01 <{Nico}> i have an update for dropbear to 0.49, changelog: http://matt.ucc.asn.au/dropbear/CHANGES
Apr 01 17:40:29 <{Nico}> anyone willing to test it?
Apr 01 17:45:38 <h3sp4wn> I guess I could test it on brcm24 and ixp4xx (given a Makefile and whatever changes are needed)
Apr 01 17:48:27 <sn9__> [florian]: pong
Apr 01 17:48:46 <[florian]> sn9__: great
Apr 01 17:49:02 <[florian]> nbd: yes squashfs works
Apr 01 17:49:13 <[florian]> nbd: except that I still have this out of memory errors
Apr 01 17:49:27 <sn9__> how did you override the kernel command line? there is no option for it on i386
Apr 01 17:50:13 <[florian]> sn9__: I copied the command line I wanted in saved_command_line
Apr 01 17:50:38 <sn9__> you patched the kernel source?
Apr 01 17:50:55 <[florian]> sn9__: yes
Apr 01 17:51:11 <sn9__> that's a worse hack than patching redboot
Apr 01 17:51:46 <sn9__> besides, redboot needs patching to support jffs2-only images, anyway
Apr 01 17:52:38 <{Nico}> h3sp4wn: http://downloads.openwrt.org/people/nico/patches/package-updates/dropbear-0.49.diff
Apr 01 17:52:48 <[florian]> I am not fond of replacing redboot from a user point of view
Apr 01 17:53:05 <nbd> sn9__: why does redboot need patching?
Apr 01 17:53:46 <sn9__> nbd: look at the patch for yourself: http://homepage.mac.com/danielg4/redboot.patch
Apr 01 17:54:05 <[florian]> sn9__: there is no need to change the kernel command line, this can easily be overriden
Apr 01 17:54:24 <sn9__> [florian]: ok, but for squashfs only
Apr 01 17:55:12 <sn9__> there is still the problem of jffs2 changing the firmware's checksum in place
Apr 01 17:56:54 <[florian]> sn9__: if you use -b 1 while generating images, it seems to work
Apr 01 17:57:04 <nbd> sn9__: where is the source that this patch applies against?
Apr 01 17:57:47 <sn9__> nbd: ftp://ftp.linksys.com/opensourcecode/wrt54gr/ecos_WRT54GR/ecos_WRT54GR_RedbootSRC_GPL.zip
Apr 01 17:58:13 <sn9__> [florian]: only for squashfs
Apr 01 18:00:10 <sn9__> where is your saved_command_line patch?
Apr 01 18:00:59 <[florian]> sn9__: locally only yet
Apr 01 18:01:56 <sn9__> i think that, just like with some adam2's on ar7, a bootloader patch to support jffs2 should be an available option
Apr 01 18:02:28 <[florian]> also consider that most people will not change their bootloaders
Apr 01 18:02:43 <[florian]> and that at least the first flashing process will need to be workarounding this
Apr 01 18:02:55 <sn9__> then those people will use squashfs
Apr 01 18:03:11 <[florian]> that is fine with me
Apr 01 18:03:47 <CIA-3> florian * r6803 /trunk/target/linux/adm5120-2.6/config/default: Small changes in the kernel config
Apr 01 18:03:55 <sn9__> the adam2 patch has been known for some time
Apr 01 18:05:55 <h3sp4wn> {Nico}: works fine on xscale - I will test brcm-2.4 as soon as its built
Apr 01 18:07:03 <nbd> sn9__: it can work without redboot patching
Apr 01 18:07:15 <nbd> sn9__: only for tftp upgrades it checks the whole image checksum
Apr 01 18:07:31 <nbd> sn9__: on boot it checks the first 1024 bytes
Apr 01 18:07:43 <sn9__> nbd: it checks the fast checksum on every boot, wich changes
Apr 01 18:07:49 <sn9__> *which
Apr 01 18:07:56 <nbd> why does it change?
Apr 01 18:08:04 <sn9__> that is not 1k
Apr 01 18:08:17 <nbd> oh, right
Apr 01 18:08:18 <sn9__> look at the algo again
Apr 01 18:08:20 <nbd> i misread something
Apr 01 18:09:30 <nbd> no, wait
Apr 01 18:10:38 <nbd> i can only see it checking 1k
Apr 01 18:11:11 <nbd> the only fwcheck_fast call supplies 1024 as size value
Apr 01 18:11:21 <nbd> and that is used for the call to the crc32 function as well
Apr 01 18:11:33 <sn9__> it makes three checks
Apr 01 18:11:59 <sn9__> only one of them is the first 1k
Apr 01 18:12:39 <nbd> where?
Apr 01 18:15:25 <sn9__> it only calls fwcheck_fast() once, but look what buffer is passed to it
Apr 01 18:16:00 <sn9__> that is 1k, but not the first 1k
Apr 01 18:16:20 * sn9__ had to look again
Apr 01 18:16:25 <nbd> oh, right
Apr 01 18:16:45 <nbd> but even that can be worked around
Apr 01 18:17:01 <nbd> simply 'correct' the fwsize to move the fs part outside of the image
Apr 01 18:17:06 <nbd> we do the same trick on the broadcom platform
Apr 01 18:17:13 <sn9__> huh?
Apr 01 18:17:25 <sn9__> oh
Apr 01 18:17:30 <nbd> reduce the fwsize to the size of image+kernel
Apr 01 18:17:34 <nbd> and recalculate the crc
Apr 01 18:17:37 <nbd> at the first boot
Apr 01 18:17:39 <nbd> in the flash map driver
Apr 01 18:18:02 <sn9__> i will change airlink.c to compensate
Apr 01 18:18:13 <nbd> will that work with tftpd?
Apr 01 18:18:21 <nbd> at least on broadcom this can't be done ahead of time
Apr 01 18:18:31 <sn9__> oh, right -- it won't work
Apr 01 18:18:32 <nbd> because there the tftpd uses the size specified in the container format
Apr 01 18:18:40 <nbd> this has to be done from the flash map driver
Apr 01 18:18:56 <sn9__> perfect, i'll do that
Apr 01 18:19:05 <nbd> great
Apr 01 18:21:58 <sn9__> no, wait, chicken and egg problem
Apr 01 18:22:42 <nbd> ?
Apr 01 18:23:38 <sn9__> currently, my flash map driver uses the total size from the container for mapping (you have not seen this version yet)
Apr 01 18:24:06 <nbd> where's the chicken and egg problem?
Apr 01 18:25:39 <sn9__> maybe i can use the reserved field to store that
Apr 01 18:26:24 <sn9__> in any case, won't i now have to put all the crc32 code in the driver?
Apr 01 18:27:08 <[florian]> sn9__: how did you solve your symbol problem, actually I get the exact same problem with rdc-2.4
Apr 01 18:28:20 <sn9__> [florian]: i have not solved it yet. i have an idea about that, and i intended to talk about that, but nbd's idea regarding the crc32 takes precedence atm
Apr 01 18:28:40 <[florian]> sure
Apr 01 18:30:34 <CIA-3> florian * r6804 /trunk/ (4 files in 4 dirs): Limit the number of hotplug childrens
Apr 01 18:32:32 <{Nico}> any reasons we keep arptables and ebtables packages in core?
Apr 01 18:34:37 <sn9__> nbd: are you saying i need to reflash from within the driver, and recalculate the crc32 there, too?
Apr 01 18:38:09 <nbd> sn9__: just change the block with the firmware size and the checksum
Apr 01 18:38:23 <sn9__> block?
Apr 01 18:38:30 <nbd> flash block
Apr 01 18:38:39 <nbd> the broadcom mtd map should show you how it's done
Apr 01 18:38:54 <sn9__> one moment...
Apr 01 18:39:52 <sn9__> brcm-2.4?
Apr 01 18:40:58 <CIA-3> florian * r6805 /trunk/target/linux/rdc-2.6/files/drivers/net/r6040.c: Use an updated r6040 ethernet driver which implements NAPI polling
Apr 01 18:41:07 <[florian]> sn9__: brcm-2.4 or brcm47xx-2.6
Apr 01 18:42:42 <sn9__> what's all this sflash_* business?
Apr 01 18:42:57 <nbd> ignore it
Apr 01 18:43:23 <sn9__> then how do i flash the block? does ioremap take care of that?
Apr 01 18:43:45 <nbd> no
Apr 01 18:43:51 <nbd> there is a specific function for that
Apr 01 18:43:55 <nbd> that uses mtd functions
Apr 01 18:44:44 <sn9__> can those mtd functions be used before there are mtd partitions?
Apr 01 18:45:35 <nbd> yes
Apr 01 18:45:44 <nbd> you work on the devices instead of the partitions
Apr 01 18:46:35 <sn9__> i cannot find that in brcm-2.4
Apr 01 18:46:45 <nbd> let me check...
Apr 01 18:48:21 <nbd> sn9__: https://dev.openwrt.org/browser/trunk/target/linux/brcm47xx-2.6/files/drivers/mtd/maps/bcm47xx-flash.c#L144
Apr 01 18:49:29 <nbd> the code that uses these functions starts at https://dev.openwrt.org/browser/trunk/target/linux/brcm47xx-2.6/files/drivers/mtd/maps/bcm47xx-flash.c#L259
Apr 01 18:52:21 <CIA-3> nbd * r6806 /packages/libs/openldap/ (Makefile patches/500-cross-compile.patch): simplify openldap makefile and make it build on osx
Apr 01 19:08:01 <CIA-3> nbd * r6807 /trunk/package/fuse/ (Makefile patches/100-cross_compile.patch): fix fuse compile on osx
Apr 01 19:10:58 <h3sp4wn> {Nico}: I would say its fine (and if you reboot it finally closes any open ssh connections which is nice) - tested on bcm 2.4 and xscale 2.6
Apr 01 19:13:14 <sn9__> what should i do in the driver if kmalloc() fails?
Apr 01 19:13:26 <nbd> return and do nothing
Apr 01 19:13:32 <nbd> it shouldn't fail anyway
Apr 01 19:13:59 <sn9__> do nothing, as in, do not create partitions?
Apr 01 19:14:28 <nbd> yep
Apr 01 19:14:53 <sn9__> wouldn't that trigger a panic?
Apr 01 19:23:20 <nbd> if kmalloc fails, then you're in deep shit anyway :)
Apr 01 19:27:58 <sn9__> after i mtd->write, will the changes show up in the ioremapped struct automagically?
Apr 01 21:10:37 <[florian]> sn9__: any progress on your flash map driver ?
Apr 01 21:28:54 <CIA-3> kaloz * r6808 /trunk/target/linux/atheros-2.6/ (config/default files/arch/mips/atheros/prom.c): use CONFIG_CMDLINE on Atheros boards
Apr 01 21:33:44 <CIA-3> florian * r6809 /trunk/target/linux/rdc-2.6/files/drivers/net/r6040.c: Fix the warning
Apr 01 21:48:34 <CIA-3> florian * r6810 /packages/net/ntpclient/files/ntpclient.init: Remove useless -s option
Apr 01 21:50:48 <CIA-3> florian * r6811 /trunk/package/kernel/modules/i2c.mk: Fix i2c kernel configuration dependency (#1511)
Apr 01 21:59:20 <CIA-3> florian * r6812 /trunk/package/kernel/modules/usb.mk: Add usbnet and asix ethernet based adapters support (#1481)
Apr 01 22:04:47 <[florian]> off to bed
Apr 01 23:16:52 <CIA-3> nico * r6813 /packages/net/chillispot/ (Makefile patches/ patches/010-linux_dev_tun.patch): fix /dev/tun path issue (closes: #1541), bump release number
Apr 01 23:17:15 <nbd> {Nico}: ping
Apr 01 23:24:15 <{Nico}> nbd: pong
Apr 01 23:24:47 <nbd> that commit breaks 2.4 and the proper solution to the problem is to fix hotplug2
Apr 01 23:25:28 <{Nico}> nbd: why does it break 2.4?
Apr 01 23:27:40 <nbd> because we keep packages for 2.4 and 2.6 in the same repository
Apr 01 23:27:48 <nbd> it's built by the sdk, so it's per-arch
Apr 01 23:28:02 <nbd> so it must be binary compatible to both
Apr 01 23:29:39 <{Nico}> oh, i thought we had per-target repositories
Apr 01 23:31:07 <{Nico}> there are other packages affected: iodyne and batman
Apr 01 23:31:27 <{Nico}> and vtun and openvpn also i guess
Apr 01 23:37:53 <{Nico}> nbd: how can i fix hotplug2?
Apr 01 23:38:25 <nbd> i'm working on it at the moment
Apr 01 23:38:28 <{Nico}> by adding an exception for tun in the rules files?
Apr 01 23:42:11 <nbd> yep
Apr 01 23:44:41 <CIA-3> nbd * r6814 /trunk/package/base-files/files/etc/hotplug2.rules: move /dev/tun to /dev/net/tun
Apr 01 23:45:59 <CIA-3> nbd * r6815 /packages/net/chillispot/ (Makefile patches/): revert [6813] - no longer necessary
Apr 01 23:46:06 <nbd> {Nico}: fixed now
Apr 01 23:53:48 <{Nico}> nbd: good, that way we have both /dev/tun and /dev/net/tun
Apr 02 00:01:04 <CIA-3> nico * r6816 /packages/net/batman/patches/101-tun_device.patch: revert [6686] - no longer necessary since [6814]
**** ENDING LOGGING AT Mon Apr 02 02:59:56 2007