**** BEGIN LOGGING AT Sun Apr 25 02:59:57 2010
Apr 25 06:28:12 <CIA-39> acoul * r21151 /trunk/package/busybox/ (Makefile patches/000-upstream-sed.patch): package/busybox: add upstream sed patch, bump release number
Apr 25 07:26:06 <bodksi1> strange, all mirrors for quilt-0.48.tar.gz failed during build of latest trunk
Apr 25 07:35:02 <acoul1> mirrors probably aren't updated yet
Apr 25 07:42:36 <bodksi1> ok, mirrors.aixtools.net does not resolve at all by the way
Apr 25 07:51:53 <CIA-39> acoul * r21152 /packages/utils/sed/ (. Makefile): [utils/sed] add GNU streams editor
Apr 25 09:41:10 <CIA-39> ralph * r21153 /trunk/package/fuse/patches/ (3 files): [package] fuse: remove empty patches
Apr 25 09:43:47 <CIA-39> ralph * r21154 /branches/backfire/package/fuse/patches/ (3 files): [backfire] merge r21153
Apr 25 09:50:36 <CIA-39> ralph * r21155 /trunk/target/linux/ifxmips/patches-2.6.32/ (5 files): [ifxmips] refresh .32 platform patches
Apr 25 11:49:47 <CIA-39> mirko * r21156 /branches/backfire/Config.in:
Apr 25 11:49:47 <CIA-39> merge commit 21050 ("add "patented"-build flag to allow packages being with and without patented technology")
Apr 25 11:49:47 <CIA-39> to backfire branch, as changes made on packages/ require that value set
Apr 25 11:50:22 <xMff> ^ ty ;)
Apr 25 11:56:26 <mirko> :)
Apr 25 12:19:51 <xMff> nunojpg: can you document your wifitoggle package here: http://wiki.openwrt.org/doc/howto/wifitoggle ?
Apr 25 13:41:43 <ali1234> [florian]: what is this: cmd |= (0 << SPI_CMD_PREPEND_BYTE_CNT_SHIFT);
Apr 25 13:50:52 <ali1234> also why read in the registers before writing them? especially, why use readb to read the command register, which is 16 bits?
Apr 25 14:00:19 <ali1234> it looks like you are constructing the command register all wrong, at least it doesn't match what is defined in regs.h
Apr 25 15:57:09 <CIA-39> ralph * r21157 /trunk/package/ifxos/patches/100-portability.patch: [ifxos] fixed autoconf detection for 2.6.33 kernel
Apr 25 16:18:23 <CIA-39> ralph * r21158 /trunk/package/ifxos/patches/100-portability.patch: [ifxos] (oops - second try) fixed autoconf detection for 2.6.33 kernel
Apr 25 16:20:18 <CIA-39> ralph * r21159 /trunk/target/linux/ifxmips/ (28 files in 9 dirs): [ifxmips] refresh 2.6.33 patches, add more mach support, add crypto support, tbd. mach/board detection
Apr 25 19:13:36 <[florian]> ali1234: this is to indicate there are no prepend bytes to the command we are sendig
Apr 25 19:13:50 <[florian]> ali1234: meaning that everything that we sent/receive is valid payload data
Apr 25 19:14:11 <[florian]> ali1234: the spi core has a special feature where it allows you to fill the unused bytes with something custom
Apr 25 19:14:30 <[florian]> ali1234: this is particularly useful for spi flashes, when sending the address for instance
Apr 25 19:14:42 <[florian]> ali1234: doing a readb/writeb on 16-bits width register is certainly wrong
Apr 25 19:19:33 <swalker> updated openwrt/upstream, http://home.comcast.net/~sdwalker/uscan/uscan.shtml
Apr 25 19:27:00 <ali1234> [florian]: i'm now comparing with what i guess is an old version from neufbox
Apr 25 19:27:06 <ali1234> i see two key differences
Apr 25 19:27:23 <ali1234> one is that the interrupt handler clears the interrupt reg instead of the mask reg
Apr 25 19:27:33 <ali1234> the other is the neufbox version sends a hard reset command
Apr 25 19:28:11 <ali1234> i also looked through cfe machine ccode
Apr 25 19:28:28 <ali1234> saw something interesting, it looks like it is reading the interrupt status from a different place
Apr 25 19:28:43 <ali1234> it could be that cfe supports both 6338 and 6348 spi versions
Apr 25 19:28:59 <ali1234> but i don't think so
Apr 25 19:29:40 <ali1234> check this out: http://pastebin.com/cXgBkNdG
Apr 25 19:32:34 <[florian]> ali1234: I have never really tested this driver to be frank, wrote it after checking what the consumer releae did in spi_read/write
Apr 25 19:33:28 <[florian]> ali1234: your dissassembly obviously looks the correct thing to do
Apr 25 19:34:53 <ali1234> thing is CFE runs with interrupts disabled
Apr 25 19:35:08 <[florian]> this does not really matter, the basic idea is there
Apr 25 19:35:20 <[florian]> what you do in polling mode will get done in the interrupt handler
Apr 25 19:35:36 <ali1234> yeah but... the main problem is the interrupt does not fire
Apr 25 19:35:44 <ali1234> hum
Apr 25 19:35:54 <[florian]> that needs fixing
Apr 25 19:36:02 <ali1234> also, the difference between 6338 and 6348 regs looks like they are endian swapped
Apr 25 19:36:08 <ali1234> at least the low ones
Apr 25 19:36:24 <ali1234> i did try using the 6348 regs but no dice
Apr 25 19:36:53 <[florian]> I will take a closer look
Apr 25 19:37:00 <[florian]> got some device with a SPI switch here
Apr 25 19:37:19 <ali1234> i am rather certain that the switch is SPI here
Apr 25 19:37:24 <ali1234> in fact i think CFE is configuring it
Apr 25 19:37:44 <ali1234> i'm sure it is getting a reply from something anyway
Apr 25 19:38:16 <[florian]> the "answer" is in the rx fifo
Apr 25 19:38:55 <[florian]> broadcom does not use interrupts as well in their original code
Apr 25 19:39:01 <ali1234> yeah, that is blank at power on but gets filled by the time cfe has loaded up to prompt
Apr 25 19:39:03 <[florian]> maybe that was on purpose
Apr 25 19:39:54 <ali1234> stupid question... where can i get the broadcom code?
Apr 25 19:40:07 <[florian]> hold on
Apr 25 19:41:19 <[florian]> http://www.efixo.net/gpl/bcm963xx_4.02L.01_consumer_release.tar.gz
Apr 25 19:41:33 <ali1234> thanks
Apr 25 19:42:17 <[florian]> I got to go now, see you
Apr 25 19:42:24 <ali1234> cya
Apr 25 19:46:58 <KanjiMonster> I have received an old speedport w500v (bcm63xx), would there be any gain for openwrt if I try to reverse engineer the adsl/voice drivers of it (e.g. are there current devices that have similar chips)?
Apr 25 19:48:23 <KanjiMonster> also, would I be able to damage the dslam through experimenting? ;)
Apr 25 19:49:09 <ali1234> yes there would be gain
Apr 25 19:49:19 <ali1234> 63xx dsl is not supported right now
Apr 25 19:50:13 <KanjiMonster> okay. my biggest concern is the dslam thing; and "new" ones on ebay I only see for 2k euro upwards (which is a bit too much for me ;)
Apr 25 19:50:32 <ali1234> dslam thing?
Apr 25 19:50:48 <ali1234> isn't that the ISP end?
Apr 25 19:50:53 <KanjiMonster> yes
Apr 25 19:51:18 <KanjiMonster> I don't know how much control I have other the connection
Apr 25 19:51:19 <ali1234> well i don't think you'll damage it without say, connecting a small generator to your phone line or something
Apr 25 19:51:36 <KanjiMonster> *over
Apr 25 19:52:12 <maligor> probably not even then
Apr 25 19:52:44 <maligor> and it'd be incredible feat of reasoning if they even knew it was you, if you manage to do it
Apr 25 19:53:10 <KanjiMonster> well, if I damage only the port its quite easy to tell ;)
Apr 25 19:53:28 <ali1234> just be like "huh? what's a dslam?"
Apr 25 19:54:00 <KanjiMonster> heh
Apr 25 19:54:07 <ali1234> anyway you don't have very much control over the connection, usually there is a DSP chip running a firmware, and you won't be rewriting that
Apr 25 19:54:08 <KanjiMonster> ok, sounds fine to test
Apr 25 19:54:31 <KanjiMonster> but first I'll have to get familiar with the platform
Apr 25 19:54:55 <ali1234> well that broadcom source file is probably a good place to start
Apr 25 19:55:07 <ali1234> although i didn't look in it yet
Apr 25 19:55:24 <KanjiMonster> there seems to be a firmware for the dsl, atleast the tarball contains a somthing_phy.bin for the dsl part which is almost the same size as the dsl dram size
Apr 25 19:55:26 <ali1234> aparently there is a dsl module but it needs some really old kernel and closed source
Apr 25 19:55:36 <ali1234> yeah that will be it
Apr 25 19:56:28 <KanjiMonster> at least its already a 2.6 line kernel ;)
Apr 25 21:13:34 <CIA-39> axel * r21160 /packages/net/bmxd/ (5 files in 4 dirs): [packages] /net/bmxd: update to revision 1636, adapt Makefile, do not overwrite /etc/config/bmxd
**** ENDING LOGGING AT Mon Apr 26 02:59:56 2010