**** BEGIN LOGGING AT Wed Apr 04 02:59:57 2007
Apr 04 08:06:01 <XorA> morning
Apr 04 08:08:15 <koen|lab> hey XorA & pH5 
Apr 04 08:08:32 <pH5> good morning
Apr 04 09:02:34 <koen|lab> drath: tried using crunch instruction yet on EABI?
Apr 04 09:06:26 <drath> koen|lab: i only have the E9301, which lacks the crunch part
Apr 04 09:10:37 <koen|lab> which they say lacks the crnch part iirc
Apr 04 09:11:18 <drath> heh, yeah
Apr 04 09:11:37 <drath> is there an easy way to test for crunch? some memmapped register?
Apr 04 09:19:11 <drath> hum, i'm recompiling a kernel with crunch support... let's see what happens
Apr 04 09:39:16 <drath> heh, nice - looks like crunch works fine on my EP9301
Apr 04 09:39:56 <koen|lab> :D
Apr 04 09:52:27 <XorA> ~lart ppp
Apr 04 09:54:03 <XorA> its not making LCP link, and gives no useful debug :-(
Apr 04 10:19:54 <drath> ok, using softfloat i get 4 whetstone mips, using crunch i get 20 mips - there might be a problem though, as I had one segfault - rerunning the same binary didn't fault
Apr 04 10:20:21 <koen|lab> gcc4 doesn't implement all crunch errata
Apr 04 10:20:39 <koen|lab> lennert made a remark about that a while back
Apr 04 10:20:47 <drath> ah, ok
Apr 04 10:21:02 <drath> anyway, it's nice being able to mix various fp formats on eabi
Apr 04 10:21:09 <koen|lab> cirrus is still refusing to forward port their stuff to gcc4
Apr 04 10:21:33 <koen|lab> drath: and use an FPU on a cpu that claims to lack one :)
Apr 04 10:21:58 <drath> hehehe, yeah, thanks for telling me about this :)
Apr 04 11:13:54 <XorA> anyone a ppp expert?
Apr 04 11:23:16 <hrw> I never had a modem
Apr 04 11:37:22 <XorA> hrw: needed for GPRS on GSM as well
Apr 04 11:41:06 <hrw> XorA: I used my cellphone as modem only with nokia770
Apr 04 11:48:52 <XorA> Im stuck in LCP negotiation
Apr 04 11:53:12 <hrw> XorA: moment..
Apr 04 11:57:30 <hrw> http://42.pl/palm+gprs.txt
Apr 04 11:57:45 <hrw> its in Polish and for palmos but maybe will help
Apr 04 11:59:51 <XorA> hrw: it works zaurus -> phone but not zaurus -> CF/GPRS
Apr 04 12:06:13 <hrw> ;(
Apr 04 12:09:03 <koen|lab> I had the same problem with my a780
Apr 04 12:09:28 <koen|lab> it doesn't work under linux, but...
Apr 04 12:09:29 <koen|lab> if you issue the *exact same* ppp command in windows it works
Apr 04 12:09:35 <koen|lab> commands*
Apr 04 12:10:32 <XorA> koen|lab: you get stuck in a loop of LCP recd/sent?
Apr 04 12:11:07 <koen|lab> the phone plays mute while pppd tries to do LCP recs
Apr 04 12:11:52 <XorA> loads of other people have had it working, I wonder if this is a kenrel 2.6.* bug
Apr 04 12:12:45 <XorA> http://www.oesf.org/forums/lofiversion/index.php/t23148.html is what I see
Apr 04 12:17:47 <koen|lab> looks like my motorola phone
Apr 04 12:18:11 <koen|lab> I 'solved' it by configuring a ip address and telling ppp to reset that after connect
Apr 04 12:18:29 <XorA> koen|lab: got an example config?
Apr 04 12:18:51 <koen|lab> no, I removed it because it broke every other ppp config
Apr 04 12:18:56 <koen|lab> (e.g. my vpn)
Apr 04 12:20:02 <koen|lab> a combo if <ip>:<ip> and ipcp-accept-remote
Apr 04 12:24:39 <XorA> that wont help though as you still need to complete LCP negotiation
Apr 04 12:29:24 <koen|lab> it helped with my a780......
Apr 04 12:29:37 <koen|lab> but the a780 is completely screwed up anyway
Apr 04 12:29:47 <XorA> IPCP is after LCP though :-(
Apr 04 12:29:59 <XorA> I think the audiovox just isnt receiving data
Apr 04 12:51:30 <XorA> http://www.linuxforums.org/forum/linux-networking/84030-pppd-negotiation-problem.html damn another report of same problem with no solution
Apr 04 12:55:45 <XorA> and I find a report of someone where it works on 2.4.20 zaurus but fails on 2.6.10 laptop the same way, do I start to get feeling 2.6.10 serial is broken
Apr 04 13:18:15 * XorA crys
Apr 04 13:24:44 <lardman|afk> XorA: It's not that you need chap rather than pap is it?
Apr 04 13:24:57 * lardman|afk thinks back to his problems when he first bought a 5500
Apr 04 13:25:16 <XorA> lardman|afk: no, Im not even at that level of a PPP connection yet
Apr 04 13:25:42 <XorA> lardman|afk: RTM8000 is not seeing the LCP ConfReq or LCP ConfAck that is being sent
Apr 04 13:25:56 <XorA> lardman|afk: I am beginning to suspect serial layer in linux is just fucked
Apr 04 13:27:34 <lardman|afk> XorA: Hmm, it looks familiar though. I'll see of I can find the old log files I had for my 5500, I dug around a lot (prompted me to learn to use Linux :) )
Apr 04 13:28:06 <XorA> you'll notice remote end requests pap auth so it cant be that
Apr 04 13:29:22 <lardman|afk> But you're expecting the log file to make sense!
Apr 04 13:29:37 <XorA> Apr  4 14:16:48 p3n3tr4t0r daemon.debug pppd[2095]: rcvd [LCP ConfReq id=0x8 <auth pap> <pcomp> <accomp>]
Apr 04 13:29:48 <XorA> Apr  4 14:16:48 p3n3tr4t0r daemon.debug pppd[2095]: sent [LCP ConfAck id=0x8 <auth pap> <pcomp> <accomp>]
Apr 04 13:30:14 <XorA> at which point we should go onto next stage having sucessfully negotiated LCP, but no, we just keep on trying
Apr 04 13:31:12 <lardman|afk> Indeed, I had the same thing, and was also pulling my hair out, until I tried setting chap-secrets. I'm not sure if it was a flaw in the Sharp implementation, but it worked (iirc - long time ago)
Apr 04 13:31:39 <XorA> # Secrets for authentication using CHAP
Apr 04 13:31:39 <XorA> # client        server  secret                  IP addresses
Apr 04 13:31:39 <XorA> username        t-mobile        password
Apr 04 13:31:57 <lardman|afk> not that then :)
Apr 04 13:33:03 <lardman|afk> right, I really am afk now
Apr 04 13:33:08 <XorA> as I said we havent even tried to auth yet
Apr 04 13:34:32 <koen|lab> hmmm
Apr 04 13:34:44 <koen|lab> simulation time ~ 2.3 cups of coffee now
Apr 04 13:59:56 <psokolovsky> Hi!
Apr 04 14:00:32 <psokolovsky> koen|lab: I'm tagging 2.6.20-hh6 now, would like to update angstrom to it. it fixes "ts driver spams syslog" issue ;-)
Apr 04 14:05:21 <koen|lab> cool, go ahead
Apr 04 14:05:34 <psokolovsky> ok
Apr 04 14:07:17 <goxboxlive> psokolvsky: while you are changing angstrom.conf. Change htcuniversal kernel source to hh4. So people are bale to use their image they build .
Apr 04 14:07:29 <goxboxlive> s/bale/able/
Apr 04 14:07:56 <mr_nice> goxboxlive: just disable evbug
Apr 04 14:08:01 <psokolovsky> koen|away: in this regard, I also suggest bumping h5000's kernel version to it. h5000 recently getting lots of revamping/development, thanks to bd2, and yesterday his suspend/resume patches were committed, so h5000 should be now pretty usable device.
Apr 04 14:08:44 <psokolovsky> goxboxlive: that's to koen. I suggest fixing universal bugs and following the party.
Apr 04 14:09:11 <koen|lab> psokolovsky: h5000 sounds good
Apr 04 14:09:21 <psokolovsky> ok
Apr 04 14:09:35 <koen|lab> and I would *really* *really* wish that the next htcuniversal kernel upgrade doesn't breaks builds all over the place
Apr 04 14:09:46 <goxboxlive> psokolovsky: I would liked that my self, but you know our situation. There are only one person developing drivers for the HTC devices, and he is kind of over worked at the moment.
Apr 04 14:10:18 <goxboxlive> mr_nice: In kernel source? What is evbug?
Apr 04 14:10:59 <psokolovsky> goxboxlive: that's not true. the whole community develops drivers for all devices, and for specific machine to leverage that, you need to write ~5 lines. You could try to do that.
Apr 04 14:12:01 <goxboxlive> evbug wherent compiled.
Apr 04 14:12:25 <goxboxlive> koen|lab: Why do you say that? We have never made any changes that breaks other devices. 
Apr 04 14:13:42 <koen|lab> where did I mention other machines?
Apr 04 14:14:22 <goxboxlive> kone|lab: Then wich places do you mean we breaks builds? 
Apr 04 14:14:38 <goxboxlive> We have never commited any stuff that breaks builds, not in the kernel and neither in OE
Apr 04 14:15:03 <koen|lab> the last 2 kernel version bumps made htcuniversal images unbuildable
Apr 04 14:15:48 <mr_nice> sorry all
Apr 04 14:17:17 <goxboxlive> That's not right, they build have been builded, but 19-hh13 didnt build, and that had nothing to do with our htc universal, it was an bug in the common source.
Apr 04 14:20:35 <goxboxlive> hh13 builds, but there where a tagges version in the .19 wich didntbuild, Dont remember wich.
Apr 04 14:21:17 <psokolovsky> goxboxlive: you got the idea, right - test/update defconfig before updating OE to new version. as it's hard to dump defconfig over irc, better use bugzilla.
Apr 04 14:21:57 <psokolovsky> goxboxlive: I also have to say that updating defconfig themselves is boring, that's why we move towards using defconfigman for that. consider adding htcuniversal support to it.
Apr 04 14:23:34 <goxboxlive> psokolvsky: first of all, i am not updating anything in OE. It's you who mostly update our kernel version. 
Apr 04 14:23:54 <goxboxlive> defconfigman never heard of it. I'll take a look at it.
Apr 04 14:24:34 <goxboxlive> wich device use defconfigman? hx4700?
Apr 04 14:25:00 <psokolovsky> goxboxlive: nope, I update (request to update) only versions of devices I'm sure about. I'd be glad to look over htcuniversal too, but you guys should follow latest changes to make that possible, or otherwise it doesn't scale.
Apr 04 14:25:54 <psokolovsky> goxboxlive: with -hh6, h3900, h4000, hx4700 use defconfigman's configs in kernel CVS, with h5000 and magician to follow.
Apr 04 14:26:16 <psokolovsky> Once that stabilizes, I'm going to propose the same for OE.
Apr 04 14:27:12 <goxboxlive> you free to look at htc-universal too, i guess cr2 will just be happy to get some help with it, i guess he is quite fed up with universal.
Apr 04 14:29:48 <psokolovsky> goxboxlive: I've got more stuff to look at already. So, it would be beneficial if more folks follow the changes being made. It's all quite transparent - RFCs are sent to kernel-discuss, and enough time is left for people's acquaintence and discussion.
Apr 04 14:31:38 <psokolovsky> goxboxlive: all in all, if you will try to follow those changes, I'll be glad to help/hint in unclear places. As for myself to do changes w/o being able to test them myself, that's on the bottom of my queue ;-I.
Apr 04 15:23:27 <XorA> hmm, Sandisk ExtremeIII cards are insanely fast
Apr 04 18:18:33 <XorA|gone> ok this sucks, gprs card works fine on 2.4-crappix kernel
Apr 04 18:24:27 <koen> "it works, because of bugs"
Apr 04 18:24:53 <XorA|gone> no, I think 2.6 serial/ppp layer has bugs
Apr 04 18:25:26 <XorA|gone> koen: which would tie in with your A780 problems and people not getting it to work on x86 as well
Apr 04 18:25:45 <koen> it refused to work on osx as well, btw
Apr 04 18:25:57 <koen> and osx is bugfree ;)
Apr 04 18:26:26 <XorA|gone> koen: Oh I thought you were talking about accessing the GSM modem in A780 from linux on the phone
Apr 04 18:26:49 <koen> linux/86 -> phone and osx -> phone
Apr 04 18:26:53 <koen> x86*
Apr 04 18:27:03 <koen> so your theory still holds
Apr 04 18:27:36 <XorA|gone> it seems pppd->uart data doesnt ever get there
Apr 04 18:27:43 <XorA|gone> but picocom works fine
Apr 04 18:27:58 <XorA|gone> I can even dial ppp with picocom and hang up again
Apr 04 18:28:20 <koen> hmmm
Apr 04 18:28:35 <pH5> XorA|gone: does it have multiple serial channels?
Apr 04 18:28:37 * koen is still trying to sober up after handing in the project report
Apr 04 18:29:00 <pH5> the internal modem of the magician is connected via two uarts, one for AT commands, one for data
Apr 04 18:29:21 <XorA|gone> pH5: no, its just one
Apr 04 18:29:46 <XorA|gone> pH5: Audiovox RTM8000 (also same chipset used in other chips)
Apr 04 18:29:47 <koen> XorA|gone: sandisk extremeIV cards are even faster, btw
Apr 04 18:29:58 <koen> RTFM8000
Apr 04 18:30:22 <XorA|gone> s/chips/CF cards/
Apr 04 18:31:23 <XorA|gone> pH5: to get back to AT commands you can +++/ATO
Apr 04 18:31:36 * XorA|gone wonders if anyone tried ppp on Neo
Apr 04 18:32:54 <pH5> XorA|gone: I think there's already something about ppp in the wiki
Apr 04 18:38:53 <XorA|gone> this probably explains why I never got a t-mobile PCMCIA card working in zaurus either
**** ENDING LOGGING AT Thu Apr 05 02:59:56 2007