[00:13] <lilo> verbal should be fine
[00:13] <lilo> thank you wrt :)
[00:13] <nbd> yeah, we are
[00:14] <lilo> okay, sounds good
[00:14] <lilo> Kaloz: can I phone you now?
[00:15] <groz> lilo, just the guy i wanna talk to
[00:15] <groz> did you know, you dont properly figure out disk geometry for usb sticks all the time ?
[00:16] <Kaloz> lilo: sure, but my spoken english wouldn't be as good as my written one :)
[00:16] <lilo> Kaloz: it's okay
[00:16] <lilo> groz: hehe
[00:16] <lilo> I'm named after a favorite science fiction character, in John Varley's THE OPHIUCHI HOTLINE. I am *not* named after the LInux LOader, and I didn't write it! :) -- lilo, winter 1993 || Nor am I named after the little girl in the movie, nor the air mattress, nor Last In Last Out. ;) -- lilo, summer 2003
[00:16] <lilo> groz: ^^
[00:16] <groz> LOL
[00:16] Action: lilo needed a FAQ
[00:16] Action: groz found lotsa faq for lilo, none of em right
[00:17] <lilo> groz: btw, I use grub mostly :)
[00:17] <groz> yah, well, grub needs target device mounted to properly install
[00:17] Action: groz dont wanna mount you
[00:18] <groz> ok, this could get real disgusting real fast
[00:18] <lilo> erm, yah
[00:18] Action: lilo hides
[00:18] <groz> hehe
[00:22] <Kaloz> :p
[00:39] <frop> mmmh
[00:40] <frop> dsp_6.02.01.00_acorp
[00:40] <frop> could i use this DSP annex A into my Dlink?
[00:41] <frop> http://submax.altervista.org/raccolta_DSP_Annex_A.tar.gz
[00:47] <Kaloz> [florian], [mbm] *poke*
[03:04] <CIA-17> mbm * r4673 /branches/buildroot-ng/openwrt/package/base-files/default/etc/hotplug.d/ (block/ block/01-mount usb/01-ln usb/01-mount): clean up hotplug mounting scripts
[03:05] <mbm> :)
[06:13] <CIA-17> groz * r4674 /packages/utils/lilo/ (bdlilo.conf lilo.conf): Update lilo example conf files for devfs fixes
[09:17] <[florian]> Kaloz: ping
[09:29] <CIA-17> florian * r4675 /packages/net/wiviz/ (. Config.in Makefile ipkg/): Port wiviz to -ng
[09:45] <CIA-17> florian * r4676 /packages/net/wifidog/ (. Config.in Makefile ipkg/): Port wifidog to -ng
[10:03] <CIA-17> florian * r4677 /packages/net/updatedd/ (. Config.in Makefile ipkg/): Port updatedd to -ng
[10:29] <[florian]> what do you think about this : http://downloads.openwrt.org/people/florian/rcconf/rcconf (shell script)
[10:31] <mbm> it's a nice script, but two questions - why do we need it and what standards is it based on?
[10:32] <[florian]> I don't know if we need it actually, I wrote it for myself. What do you mean by standard ? it relys on the fact that rcS will try to execute all init script matching S*
[10:32] <mbm> I really don't like inventing more languages or configuration formats which are only used in openwrt
[10:32] <[florian]> debian's rcconf sound big for the same thing
[10:33] <mbm> does your script copy the output of debian's rcconf?
[10:34] <[florian]> no, debian's rcconf is ncurses based, this one is pure shell
[10:34] <[florian]> so with debian's rcconf you have a select liste where you choose which service you want to start at boot time
[10:34] <[florian]> here you show which services are enabled, which are not .. and make your choice
[10:35] <murb> don't you just use update-rc.d ??
[10:35] <[florian]> are we talking about the same thing ? this script is intended to run on an openwrt box
[10:37] <[florian]> don't be afraid, it was just a small script to let me configure which services I wanted at boot time from which I don't want
[10:38] <mbm> fyi, that is_protected_service could probably be cleaned up using ${PROTECTED_SERVICES} != ${PROTECTED##*$1*}
[10:39] <mbm> just to avoid the overhead of executing grep
[10:44] <[florian]> thanks
[10:44] <[florian]> by the way, do you think it could be useful to anyone but me ;) ?
[10:46] <mbm> seems to be an overly complex way to do a chmod
[10:46] <[florian]> that's right
[10:47] <mbm> and if we follow debian's trick of using a source instead of an execute then that trick won't work
[10:49] <[florian]> ok, by the way you have lost your brackets ?
[10:49] <CIA-17> florian * r4678 /packages/net/ulogd/ (. Config.in Makefile ipkg/): Port ulogd to -ng
[10:50] <mbm> yep. long story, involves lilo and #openwrt
[10:54] <[florian]> ok, I don't really want to know actually
[10:55] <mbm> good.
[10:56] <groz> mbm, i did eventually get that devfs patch in, seems to work here
[10:56] <groz> but i haven't gotten to th efun part of exercising it yet, doubt i will tonite
[10:56] <groz> getting to late
[10:56] <mbm> did you try the hotplug patches that I did a bit later?
[10:56] <groz> but, now i wanna install openwrt onto a raid5 of 3 usb sticks on my front router hehe
[10:56] <groz> no, not yet, coming
[10:57] <groz> there's only so much time in a day, and today it's alllll been soaked up
[10:58] <groz> i have been playing a bunch tonite with the ppp stuff too
[11:27] <[florian]> lol groz what for ?
[11:28] <groz> because with kernel 2.6 and the ppp 2.4.3 in openwrt, i'm getting problems on the data link
[11:28] <groz> but now i've put in the 2.4.4 without any patches
[11:28] <groz> and all 7 of my links are staying up ok tonite
[11:30] <groz> i use pptp links a lot here
[11:30] <[florian]> ok
[11:30] <[florian]> ppp 2.4.4 is out ?
[11:30] <groz> yah, but thats not where i think the issues lie
[11:30] <groz> it lies here
[11:31] <groz> 2.4 kernel has numerous variants of mppe patches
[11:31] <groz> the 2.6.17 now has an mppe implementation in the mainline
[11:31] <groz> that implementation is incompatible with the patches we put in 2.4.3
[11:31] <groz> that go along with the kernel patches we use
[11:31] <nbd> it shouldn't be hard to backport to 2.4, right?
[11:32] <groz> i haven't looked at the backport yet nbd, but, it should be quite simpmle yes
[11:32] <groz> i'm scrambling to get a couple more things done for my clients here today or tomorrow, before i disappear off flying for a couple days again
[11:33] <groz> that's kind of how i bumped into the pptp / ppp issues
[11:33] <groz> since i converted my main router to be the via running openwrt, i started having problems on the pptp connections to thier offices
[11:34] <[florian]> groz: by the way, do you use ip_conntrack_pptp ?
[11:34] <groz> yes, i am using it
[11:34] <groz> and, it seems to work quite well, what i did to test it was this
[11:34] <[florian]> groz: can you think of reproducing #571 ?
[11:34] <groz> i have my front router with a few full time connections
[11:34] <groz> then i started up another one behind it
[11:34] <groz> and made some pptp connections to the same places
[11:35] <groz> and that's NEVER worked before
[11:35] <groz> worked fine tonite
[11:35] <groz> lemme go see what it is
[11:36] <CIA-17> florian * r4679 /packages/libs/ncurses/Makefile: Add missing include files and libraries (needed for tmsnc)
[11:36] <CIA-17> florian * r4680 /packages/net/tmsnc/ (. Config.in Makefile Makefile ipkg/): Port tmsnc to -ng
[11:37] <groz> hmm, florian, the conntrack may be a red herring on that one, if the router itself is doing the pptp connection
[11:37] <groz> it doesn't need contracks
[11:37] <groz> oh, wait, i'm misreading that
[11:40] <groz> florian, as an fyi, i have that setup now, but, the router is via epia board with linux 2.6 booting from usb key
[11:41] <groz> i have one wrt behind it, with a full time pptp, and, the router itself right now has (checking) 5 pptp sessions up and active
[11:41] <groz> I'll have to shuffle things a bit here, to get a client behind the wrt, and start yet another one frm there
[11:43] <[florian]> ok, it just came to me as you were talking about pptp and I knew there was such issue with it
[11:44] <[florian]> so if you have time to reproduce the bug, thanks, if not, we will probably close it as resolved
[11:44] <groz> give me a few minutes, and i'll get a notebook associated with that wrt
[11:44] <groz> and then get a pptp going thru it
[11:44] <groz> it's using a relatively recent rc5 build
[11:48] <[florian]> groz: a rc5 with the updated kmod should suffice
[11:52] <groz> well, this is interesting
[11:52] <groz> to get the notebook actually behind the wrt, it ends up with 2 nats ahead of it before it goes out to the internet
[11:53] <groz> i brought down the pptp links on both of them, just to be sure
[11:53] <groz> and validated the pptp conntracks are in on both
[11:53] <groz> it cant connect
[11:53] <groz> yet the router it's connected to wirelessly
[11:53] <groz> has no problem bringing up that same link
[11:56] <nbd> pptp is so annoying
[11:56] <groz> tell me about it
[11:57] <groz> i have to deal with it regularily, cuz, it's what they use
[11:57] <groz> and it's soooooooo brain dead
[11:58] <nbd> if only they knew that they could get better security with less effort and less problems if they simply used openvpn instead
[11:58] <groz> the real issue is, most notebooks come with a setup for pptp already
[11:58] <groz> so it's far less work to just set up the pptp server
[11:59] <groz> than to install stuff on every salesman's notebook
[12:00] <groz> you look at openvpn doc, and the first thing, it starts going on and on about needing to set up certificates
[12:00] <groz> and mention certificates, typical it guys eyes gloss over
[12:00] <groz> and that's it, no point going any farther
[12:01] <nbd> yeah, i'm lucky that where i set up stuff like that they simply trust me to make the right technical decisions
[12:02] <groz> well, typically i dont set that up, i end up dealing with clients that already have it
[12:02] <groz> so, i need to go with what they have already in place
[12:02] <groz> they dont hire me to criticize thier networks, they hire me to get software working
[12:03] <nbd> yeah
[12:17] <[florian]> groz: so does it work with nbd's updated kmods ;) ?
[12:17] <groz> well i cant put that updated kmod onto the front router, it's x86
[12:18] <groz> and i cant get the wrt onto an exposed ip right now
[12:18] <groz> i can tho in a while, when i'm finished what i'm getting out of the way right now
[12:22] <groz> dumb question tho, if i checked it out in the last week, will the updated kmods be in there already, or, do i need to still install the one listed in that dl on the ticket ?
[12:36] <[florian]> depends if the patch goes upstream or not
[12:36] <groz> what i meant was
[12:36] <groz> is the patch in the svn
[12:37] <groz> or, do i have to go get the 'special' one ?
[12:37] <[florian]> that exactly what I tried to say, the patch is not in svn, that is why you need to download nbd's kmod
[12:37] <groz> ok, i'll get it in a couple minutes
[12:38] <[florian]> no pb
[12:40] <[florian]> nbd: did you make some progress with the ethernet driver ?
[12:40] <[florian]> do you use precompiled binaries for 2.6.8.1 kernel or do you link the modules with the current kernel we use ?
[12:43] <nbd> [florian]: i used the .o_saved and linked it for 2.6.17
[12:43] <nbd> i'm currently stuck at a point where __mod_timer locks up
[12:43] <nbd> probably while trying to acquire a spinlock
[12:43] <[florian]> ok
[12:52] <groz> ok, now my notebook has reached a state where it wont connect to _anything_
[12:52] <groz> gonna tell it to reboot, then go get some shut-eye for a while, been at this computer way to late tonite
[12:53] <groz> i'll recheck that stuff in the morning, when i'm wide awake and not falling asleep at the keyboard
[12:53] <groz> but i do have the updated modules on the wrt now
[12:55] <[florian]> nbd: you told CONFIG_package is the replacement for old booleans variables in menuconfig right ?
[12:56] <nbd> which variables?
[12:57] <[florian]> the boolean variables we use in some packages to enable features at configure time, such as BR2_PACKAGE_COMPILE_WITH_FOO
[12:59] <[florian]> {Nico} advised me produce several packages with this set of features enabled or not, but if we take the example of openvpn, it makes too many packages combinations
[12:59] <[florian]> do you see what I am talking about ?
[13:07] <nbd> yeah, i still don't know how we should handle this
[13:08] <florian> this currently prevents me from restoring the exact layout of packages such as openvpn ...
[13:08] <florian> but no problem
[13:44] <florian> nbd: regarding bcm63xx bin drivers, I know it is not as simple as keeping prototypes and symbols, how could I help you making the ethernet work ?
[14:58] <{Nico}> Ã/win 9
[15:30] <nbd> florian: sorry, no idea.
[16:22] <frop> hallo
[16:22] <frop> insmod: unresolved symbol avalanche_request_intr_pacing
[16:22] <frop> insmod: unresolved symbol __up
[16:22] <frop> insmod: unresolved symbol jiffies_64
[16:22] <florian> frop: ar7 device right ?
[16:22] <frop> insmodding a "new" tiatm.o
[16:22] <frop> florian: yep
[16:23] <frop> i'd like to know, if with a new DSP firmware...i can get my pppd working
[16:26] <CIA-17> florian * r4681 / (12 files in 12 dirs): Remove executable bit from these Makefiles. Closes #728
[16:30] <frop> 003-net_driver_cpmac.patch in this file seems that avalanche_request_intr_pacing is present
[16:30] <frop> +extern int avalanche_request_intr_pacing(int, unsigned int, unsigned int);
[16:33] <florian> frop: extern int means the variable is defined elsewhere, in another source file
[16:34] <florian> so you have to find where it is :)
[16:34] <frop> eheh
[16:34] <frop> ...and then?
[16:34] <frop> :)
[16:35] <florian> then, try to see if the symbol is exported or not I guess
[16:36] <frop> ok
[16:37] <frop> trying...
[17:16] <florian> frop: ok ?
[17:32] <florian> nbd: I added assoclist to wlc, it works
[17:32] <florian> now I will do it for wlcompat
[17:34] <nbd> great
[17:38] <florian> I just don't understand yet, what is the data being used for rssi infos : http://pastebin.ca/150145
[17:43] <nbd> are you sure it's -46 and not -56?
[17:43] <florian> yes it's -56 ;) sorry
[17:44] <florian> according to wlioctl.h rssi value is a int16
[17:44] <nbd> it doesn't matter whether you parse it as 32 bit, 16 bit or 8 bit
[17:45] <nbd> it all comes out as -56 if you interpret it as signed value
[17:45] <florian> ok, strange to see that the mac address is also being outputed
[17:46] <nbd> it's simply not touching that buffer area
[17:46] <florian> fair enough
[17:46] <florian> so if I just get the first byte, it should be ok
[17:46] <florian> no, not the first byte
[17:47] <nbd> doesn't matter if you take the first byte, word or dword
[17:48] <florian> I have to find the structure to cast the buffer
[17:51] <florian> nbd: do you have a sample Makefile to link the .o_save objects into a ko ?
[17:52] <nbd> florian: http://pastebin.ca/150153
[17:53] <florian> thanks
[19:00] <florian> nbd: actually it would be good to implement the maclist management in wlcompat
[19:11] <florian> unfortunately, I don't see how I can easily do it
[20:16] <florian> nbd: which impl version did you try for bcm_enet ?
[21:05] <florian> nbd: when linking bcm_enet with my current kernel I get some undefined symbols, do you know how I can solve this
[21:31] <nbd> i have lots of uncommitted changes
[21:31] <nbd> i'll give you an updated patch soon
[21:34] <florian> ok great
[22:51] <florian> nbd: just to know, how could these unresolved symbols be resolved, I mean a general way
[22:52] <nbd> by adding the missing symbols
[22:52] <nbd> btw. which bcm_enet.o did you use?
[22:59] <florian> impl2
[22:59] <florian> from a comtrend source
[22:59] <nbd> was that source also using 2.6.8.1?
[22:59] <florian> yep
[23:00] <nbd> what symbols are missing?
[23:37] <florian> re, nbd http://pastebin.ca/150400
[23:38] <florian> I think I know where it comes from
[23:41] <florian> I forgot enabling networking support :p
[23:41] <nbd> hehe
[23:42] <florian> sometimes I really wonder how dummy I could be ;)
[23:43] <nbd> shit happens
[23:43] <nbd> :)
[23:43] <florian> yeah
[23:43] <florian> do you think we could write a free driver for the ethernet chip ?
[23:43] <nbd> it's a lot of work
[23:43] <nbd> and i'm not particularly good at reverse engineering mips stuff
[23:44] <florian> on my livebox there is a bcm5325 chip
[23:44] <nbd> yeah, i have one on my board as well
[23:45] <florian> so, if I well understand it is almost the same chip as in the latest wrt's, but the bus and the way to "talk" with it is different right ?
[23:52] <florian> did you make some progress with bcm_enet ?
[23:55] <nbd> not yet
[23:59] <florian> my kernel crashes very rapidly when it comes to loading the networking stack
[00:00] --- Sun Aug 27 2006
[00:13] <lilo> thank you wrt :)
[00:13] <nbd> yeah, we are
[00:14] <lilo> okay, sounds good
[00:14] <lilo> Kaloz: can I phone you now?
[00:15] <groz> lilo, just the guy i wanna talk to
[00:15] <groz> did you know, you dont properly figure out disk geometry for usb sticks all the time ?
[00:16] <Kaloz> lilo: sure, but my spoken english wouldn't be as good as my written one :)
[00:16] <lilo> Kaloz: it's okay
[00:16] <lilo> groz: hehe
[00:16] <lilo> I'm named after a favorite science fiction character, in John Varley's THE OPHIUCHI HOTLINE. I am *not* named after the LInux LOader, and I didn't write it! :) -- lilo, winter 1993 || Nor am I named after the little girl in the movie, nor the air mattress, nor Last In Last Out. ;) -- lilo, summer 2003
[00:16] <lilo> groz: ^^
[00:16] <groz> LOL
[00:16] Action: lilo needed a FAQ
[00:16] Action: groz found lotsa faq for lilo, none of em right
[00:17] <lilo> groz: btw, I use grub mostly :)
[00:17] <groz> yah, well, grub needs target device mounted to properly install
[00:17] Action: groz dont wanna mount you
[00:18] <groz> ok, this could get real disgusting real fast
[00:18] <lilo> erm, yah
[00:18] Action: lilo hides
[00:18] <groz> hehe
[00:22] <Kaloz> :p
[00:39] <frop> mmmh
[00:40] <frop> dsp_6.02.01.00_acorp
[00:40] <frop> could i use this DSP annex A into my Dlink?
[00:41] <frop> http://submax.altervista.org/raccolta_DSP_Annex_A.tar.gz
[00:47] <Kaloz> [florian], [mbm] *poke*
[03:04] <CIA-17> mbm * r4673 /branches/buildroot-ng/openwrt/package/base-files/default/etc/hotplug.d/ (block/ block/01-mount usb/01-ln usb/01-mount): clean up hotplug mounting scripts
[03:05] <mbm> :)
[06:13] <CIA-17> groz * r4674 /packages/utils/lilo/ (bdlilo.conf lilo.conf): Update lilo example conf files for devfs fixes
[09:17] <[florian]> Kaloz: ping
[09:29] <CIA-17> florian * r4675 /packages/net/wiviz/ (. Config.in Makefile ipkg/): Port wiviz to -ng
[09:45] <CIA-17> florian * r4676 /packages/net/wifidog/ (. Config.in Makefile ipkg/): Port wifidog to -ng
[10:03] <CIA-17> florian * r4677 /packages/net/updatedd/ (. Config.in Makefile ipkg/): Port updatedd to -ng
[10:29] <[florian]> what do you think about this : http://downloads.openwrt.org/people/florian/rcconf/rcconf (shell script)
[10:31] <mbm> it's a nice script, but two questions - why do we need it and what standards is it based on?
[10:32] <[florian]> I don't know if we need it actually, I wrote it for myself. What do you mean by standard ? it relys on the fact that rcS will try to execute all init script matching S*
[10:32] <mbm> I really don't like inventing more languages or configuration formats which are only used in openwrt
[10:32] <[florian]> debian's rcconf sound big for the same thing
[10:33] <mbm> does your script copy the output of debian's rcconf?
[10:34] <[florian]> no, debian's rcconf is ncurses based, this one is pure shell
[10:34] <[florian]> so with debian's rcconf you have a select liste where you choose which service you want to start at boot time
[10:34] <[florian]> here you show which services are enabled, which are not .. and make your choice
[10:35] <murb> don't you just use update-rc.d ??
[10:35] <[florian]> are we talking about the same thing ? this script is intended to run on an openwrt box
[10:37] <[florian]> don't be afraid, it was just a small script to let me configure which services I wanted at boot time from which I don't want
[10:38] <mbm> fyi, that is_protected_service could probably be cleaned up using ${PROTECTED_SERVICES} != ${PROTECTED##*$1*}
[10:39] <mbm> just to avoid the overhead of executing grep
[10:44] <[florian]> thanks
[10:44] <[florian]> by the way, do you think it could be useful to anyone but me ;) ?
[10:46] <mbm> seems to be an overly complex way to do a chmod
[10:46] <[florian]> that's right
[10:47] <mbm> and if we follow debian's trick of using a source instead of an execute then that trick won't work
[10:49] <[florian]> ok, by the way you have lost your brackets ?
[10:49] <CIA-17> florian * r4678 /packages/net/ulogd/ (. Config.in Makefile ipkg/): Port ulogd to -ng
[10:50] <mbm> yep. long story, involves lilo and #openwrt
[10:54] <[florian]> ok, I don't really want to know actually
[10:55] <mbm> good.
[10:56] <groz> mbm, i did eventually get that devfs patch in, seems to work here
[10:56] <groz> but i haven't gotten to th efun part of exercising it yet, doubt i will tonite
[10:56] <groz> getting to late
[10:56] <mbm> did you try the hotplug patches that I did a bit later?
[10:56] <groz> but, now i wanna install openwrt onto a raid5 of 3 usb sticks on my front router hehe
[10:56] <groz> no, not yet, coming
[10:57] <groz> there's only so much time in a day, and today it's alllll been soaked up
[10:58] <groz> i have been playing a bunch tonite with the ppp stuff too
[11:27] <[florian]> lol groz what for ?
[11:28] <groz> because with kernel 2.6 and the ppp 2.4.3 in openwrt, i'm getting problems on the data link
[11:28] <groz> but now i've put in the 2.4.4 without any patches
[11:28] <groz> and all 7 of my links are staying up ok tonite
[11:30] <groz> i use pptp links a lot here
[11:30] <[florian]> ok
[11:30] <[florian]> ppp 2.4.4 is out ?
[11:30] <groz> yah, but thats not where i think the issues lie
[11:30] <groz> it lies here
[11:31] <groz> 2.4 kernel has numerous variants of mppe patches
[11:31] <groz> the 2.6.17 now has an mppe implementation in the mainline
[11:31] <groz> that implementation is incompatible with the patches we put in 2.4.3
[11:31] <groz> that go along with the kernel patches we use
[11:31] <nbd> it shouldn't be hard to backport to 2.4, right?
[11:32] <groz> i haven't looked at the backport yet nbd, but, it should be quite simpmle yes
[11:32] <groz> i'm scrambling to get a couple more things done for my clients here today or tomorrow, before i disappear off flying for a couple days again
[11:33] <groz> that's kind of how i bumped into the pptp / ppp issues
[11:33] <groz> since i converted my main router to be the via running openwrt, i started having problems on the pptp connections to thier offices
[11:34] <[florian]> groz: by the way, do you use ip_conntrack_pptp ?
[11:34] <groz> yes, i am using it
[11:34] <groz> and, it seems to work quite well, what i did to test it was this
[11:34] <[florian]> groz: can you think of reproducing #571 ?
[11:34] <groz> i have my front router with a few full time connections
[11:34] <groz> then i started up another one behind it
[11:34] <groz> and made some pptp connections to the same places
[11:35] <groz> and that's NEVER worked before
[11:35] <groz> worked fine tonite
[11:35] <groz> lemme go see what it is
[11:36] <CIA-17> florian * r4679 /packages/libs/ncurses/Makefile: Add missing include files and libraries (needed for tmsnc)
[11:36] <CIA-17> florian * r4680 /packages/net/tmsnc/ (. Config.in Makefile Makefile ipkg/): Port tmsnc to -ng
[11:37] <groz> hmm, florian, the conntrack may be a red herring on that one, if the router itself is doing the pptp connection
[11:37] <groz> it doesn't need contracks
[11:37] <groz> oh, wait, i'm misreading that
[11:40] <groz> florian, as an fyi, i have that setup now, but, the router is via epia board with linux 2.6 booting from usb key
[11:41] <groz> i have one wrt behind it, with a full time pptp, and, the router itself right now has (checking) 5 pptp sessions up and active
[11:41] <groz> I'll have to shuffle things a bit here, to get a client behind the wrt, and start yet another one frm there
[11:43] <[florian]> ok, it just came to me as you were talking about pptp and I knew there was such issue with it
[11:44] <[florian]> so if you have time to reproduce the bug, thanks, if not, we will probably close it as resolved
[11:44] <groz> give me a few minutes, and i'll get a notebook associated with that wrt
[11:44] <groz> and then get a pptp going thru it
[11:44] <groz> it's using a relatively recent rc5 build
[11:48] <[florian]> groz: a rc5 with the updated kmod should suffice
[11:52] <groz> well, this is interesting
[11:52] <groz> to get the notebook actually behind the wrt, it ends up with 2 nats ahead of it before it goes out to the internet
[11:53] <groz> i brought down the pptp links on both of them, just to be sure
[11:53] <groz> and validated the pptp conntracks are in on both
[11:53] <groz> it cant connect
[11:53] <groz> yet the router it's connected to wirelessly
[11:53] <groz> has no problem bringing up that same link
[11:56] <nbd> pptp is so annoying
[11:56] <groz> tell me about it
[11:57] <groz> i have to deal with it regularily, cuz, it's what they use
[11:57] <groz> and it's soooooooo brain dead
[11:58] <nbd> if only they knew that they could get better security with less effort and less problems if they simply used openvpn instead
[11:58] <groz> the real issue is, most notebooks come with a setup for pptp already
[11:58] <groz> so it's far less work to just set up the pptp server
[11:59] <groz> than to install stuff on every salesman's notebook
[12:00] <groz> you look at openvpn doc, and the first thing, it starts going on and on about needing to set up certificates
[12:00] <groz> and mention certificates, typical it guys eyes gloss over
[12:00] <groz> and that's it, no point going any farther
[12:01] <nbd> yeah, i'm lucky that where i set up stuff like that they simply trust me to make the right technical decisions
[12:02] <groz> well, typically i dont set that up, i end up dealing with clients that already have it
[12:02] <groz> so, i need to go with what they have already in place
[12:02] <groz> they dont hire me to criticize thier networks, they hire me to get software working
[12:03] <nbd> yeah
[12:17] <[florian]> groz: so does it work with nbd's updated kmods ;) ?
[12:17] <groz> well i cant put that updated kmod onto the front router, it's x86
[12:18] <groz> and i cant get the wrt onto an exposed ip right now
[12:18] <groz> i can tho in a while, when i'm finished what i'm getting out of the way right now
[12:22] <groz> dumb question tho, if i checked it out in the last week, will the updated kmods be in there already, or, do i need to still install the one listed in that dl on the ticket ?
[12:36] <[florian]> depends if the patch goes upstream or not
[12:36] <groz> what i meant was
[12:36] <groz> is the patch in the svn
[12:37] <groz> or, do i have to go get the 'special' one ?
[12:37] <[florian]> that exactly what I tried to say, the patch is not in svn, that is why you need to download nbd's kmod
[12:37] <groz> ok, i'll get it in a couple minutes
[12:38] <[florian]> no pb
[12:40] <[florian]> nbd: did you make some progress with the ethernet driver ?
[12:40] <[florian]> do you use precompiled binaries for 2.6.8.1 kernel or do you link the modules with the current kernel we use ?
[12:43] <nbd> [florian]: i used the .o_saved and linked it for 2.6.17
[12:43] <nbd> i'm currently stuck at a point where __mod_timer locks up
[12:43] <nbd> probably while trying to acquire a spinlock
[12:43] <[florian]> ok
[12:52] <groz> ok, now my notebook has reached a state where it wont connect to _anything_
[12:52] <groz> gonna tell it to reboot, then go get some shut-eye for a while, been at this computer way to late tonite
[12:53] <groz> i'll recheck that stuff in the morning, when i'm wide awake and not falling asleep at the keyboard
[12:53] <groz> but i do have the updated modules on the wrt now
[12:55] <[florian]> nbd: you told CONFIG_package is the replacement for old booleans variables in menuconfig right ?
[12:56] <nbd> which variables?
[12:57] <[florian]> the boolean variables we use in some packages to enable features at configure time, such as BR2_PACKAGE_COMPILE_WITH_FOO
[12:59] <[florian]> {Nico} advised me produce several packages with this set of features enabled or not, but if we take the example of openvpn, it makes too many packages combinations
[12:59] <[florian]> do you see what I am talking about ?
[13:07] <nbd> yeah, i still don't know how we should handle this
[13:08] <florian> this currently prevents me from restoring the exact layout of packages such as openvpn ...
[13:08] <florian> but no problem
[13:44] <florian> nbd: regarding bcm63xx bin drivers, I know it is not as simple as keeping prototypes and symbols, how could I help you making the ethernet work ?
[14:58] <{Nico}> Ã/win 9
[15:30] <nbd> florian: sorry, no idea.
[16:22] <frop> hallo
[16:22] <frop> insmod: unresolved symbol avalanche_request_intr_pacing
[16:22] <frop> insmod: unresolved symbol __up
[16:22] <frop> insmod: unresolved symbol jiffies_64
[16:22] <florian> frop: ar7 device right ?
[16:22] <frop> insmodding a "new" tiatm.o
[16:22] <frop> florian: yep
[16:23] <frop> i'd like to know, if with a new DSP firmware...i can get my pppd working
[16:26] <CIA-17> florian * r4681 / (12 files in 12 dirs): Remove executable bit from these Makefiles. Closes #728
[16:30] <frop> 003-net_driver_cpmac.patch in this file seems that avalanche_request_intr_pacing is present
[16:30] <frop> +extern int avalanche_request_intr_pacing(int, unsigned int, unsigned int);
[16:33] <florian> frop: extern int means the variable is defined elsewhere, in another source file
[16:34] <florian> so you have to find where it is :)
[16:34] <frop> eheh
[16:34] <frop> ...and then?
[16:34] <frop> :)
[16:35] <florian> then, try to see if the symbol is exported or not I guess
[16:36] <frop> ok
[16:37] <frop> trying...
[17:16] <florian> frop: ok ?
[17:32] <florian> nbd: I added assoclist to wlc, it works
[17:32] <florian> now I will do it for wlcompat
[17:34] <nbd> great
[17:38] <florian> I just don't understand yet, what is the data being used for rssi infos : http://pastebin.ca/150145
[17:43] <nbd> are you sure it's -46 and not -56?
[17:43] <florian> yes it's -56 ;) sorry
[17:44] <florian> according to wlioctl.h rssi value is a int16
[17:44] <nbd> it doesn't matter whether you parse it as 32 bit, 16 bit or 8 bit
[17:45] <nbd> it all comes out as -56 if you interpret it as signed value
[17:45] <florian> ok, strange to see that the mac address is also being outputed
[17:46] <nbd> it's simply not touching that buffer area
[17:46] <florian> fair enough
[17:46] <florian> so if I just get the first byte, it should be ok
[17:46] <florian> no, not the first byte
[17:47] <nbd> doesn't matter if you take the first byte, word or dword
[17:48] <florian> I have to find the structure to cast the buffer
[17:51] <florian> nbd: do you have a sample Makefile to link the .o_save objects into a ko ?
[17:52] <nbd> florian: http://pastebin.ca/150153
[17:53] <florian> thanks
[19:00] <florian> nbd: actually it would be good to implement the maclist management in wlcompat
[19:11] <florian> unfortunately, I don't see how I can easily do it
[20:16] <florian> nbd: which impl version did you try for bcm_enet ?
[21:05] <florian> nbd: when linking bcm_enet with my current kernel I get some undefined symbols, do you know how I can solve this
[21:31] <nbd> i have lots of uncommitted changes
[21:31] <nbd> i'll give you an updated patch soon
[21:34] <florian> ok great
[22:51] <florian> nbd: just to know, how could these unresolved symbols be resolved, I mean a general way
[22:52] <nbd> by adding the missing symbols
[22:52] <nbd> btw. which bcm_enet.o did you use?
[22:59] <florian> impl2
[22:59] <florian> from a comtrend source
[22:59] <nbd> was that source also using 2.6.8.1?
[22:59] <florian> yep
[23:00] <nbd> what symbols are missing?
[23:37] <florian> re, nbd http://pastebin.ca/150400
[23:38] <florian> I think I know where it comes from
[23:41] <florian> I forgot enabling networking support :p
[23:41] <nbd> hehe
[23:42] <florian> sometimes I really wonder how dummy I could be ;)
[23:43] <nbd> shit happens
[23:43] <nbd> :)
[23:43] <florian> yeah
[23:43] <florian> do you think we could write a free driver for the ethernet chip ?
[23:43] <nbd> it's a lot of work
[23:43] <nbd> and i'm not particularly good at reverse engineering mips stuff
[23:44] <florian> on my livebox there is a bcm5325 chip
[23:44] <nbd> yeah, i have one on my board as well
[23:45] <florian> so, if I well understand it is almost the same chip as in the latest wrt's, but the bus and the way to "talk" with it is different right ?
[23:52] <florian> did you make some progress with bcm_enet ?
[23:55] <nbd> not yet
[23:59] <florian> my kernel crashes very rapidly when it comes to loading the networking stack
[00:00] --- Sun Aug 27 2006