**** BEGIN LOGGING AT Mon Jun 04 02:59:59 2012 Jun 04 03:43:59 gta04 boots, yay, good morning Jun 04 03:44:25 pabs3: one other guy on the ML did. Jun 04 03:46:37 PaulFertser: hmm, I can't find anything in the mail archives, do you have a pointer? Jun 04 03:48:13 Well, not really fully boots but at least the bootloader and the kernel worked. But only once. Jun 04 03:49:00 pabs3: Alex Samorukov that was Jun 04 03:50:21 ah, thanks. Harald said no-one was interested Jun 04 03:50:28 will forward him a link Jun 04 03:50:58 Thanks Jun 04 03:51:15 are you interested in helping btw? Jun 04 03:52:24 Yes, but i'm not a professional sysadmin, i only configured stuff i used myself, though it included apache etc on remote servers. Jun 04 03:54:27 I think thats ok, I guess you would be able to learn anything you need to Jun 04 03:58:20 Indeed Jun 04 03:59:25 But some tricky stuff like proper configuration of fail2ban, iptables throttling rules and other security-critical tweaks are best to be known in advance :) Jun 04 04:01:14 sure Jun 04 04:08:55 freesmartphone.org: 03morphis 07aurora * rdf031e44a33d 10/aurora-daemon/applications/phone/Dialpad.qml: aurora-daemon: phone: cleanup the dial-pad implementation Jun 04 04:08:55 freesmartphone.org: 03morphis 07aurora * r21734b2d6459 10/aurora-daemon/components/Button.qml: aurora-daemon: components: remove white border from the button component Jun 04 04:09:02 freesmartphone.org: 03morphis 07aurora * r104e17032adf 10/aurora-daemon/components/ (CharacterKey.qml KeyPopper.qml): aurora-daemon: components: don't provide any radius for the corner of our buttons Jun 04 04:09:02 freesmartphone.org: 03morphis 07aurora * r9af7270ba500 10/aurora-daemon/README: aurora-daemon: update needed dependencies in README file Jun 04 04:09:05 freesmartphone.org: 03morphis 07aurora * r0694821e7544 10/aurora-daemon/ (17 files in 3 dirs): aurora-daemon: drop settings application Jun 04 04:09:06 freesmartphone.org: 03morphis 07aurora * r21fbcd2ab324 10/aurora-daemon/ (configure.ac runtime/Makefile.am): aurora-daemon: we need libfso-qt Jun 04 04:09:07 freesmartphone.org: 03morphis 07aurora * r0cc887d33e35 10/aurora-daemon/runtime/ (applicationinfo.cpp applicationstorage.cpp): aurora-daemon: we don't use libfsobasics anymore Jun 04 04:09:08 freesmartphone.org: 03morphis 07aurora * r94d693e37a8e 10/aurora-daemon/ (4 files in 2 dirs): aurora-daemon: remove a lot of unneeded stuff Jun 04 04:17:54 freesmartphone.org: 03morphis 07cornucopia * r86840520aed1 10/ChangeLog: Add mising ChangeLog file Jun 04 04:17:55 freesmartphone.org: 03morphis 07cornucopia * re805b5be005f 10/configure.ac: BUmp version to be in sync with fso-specs Jun 04 04:17:55 freesmartphone.org: 03morphis 07cornucopia * r8dfa22af5bf8 10/src/Makefile.am: Don't use namespaces Jun 04 04:17:56 freesmartphone.org: 03morphis 07cornucopia * r0d3d393061e4 10/src/Makefile.am: Add new source file for org.freesmartphone.GSM.CallForwarding interface Jun 04 04:36:34 using gta04 as a normal debian user is hard: no Neil's kernel in repos it seems. Jun 04 04:38:16 same with gta02 Jun 04 04:38:34 what about http://packages.debian.org/sid/linux-image-3.2.0-2-omap ? Jun 04 04:39:58 I do not think it has everything's needed for gta04 Jun 04 04:40:08 Well, time to go to work :( Jun 04 06:41:27 how could I disable reboot when systemd has a permission issue with dbus ? I had a bad crashes ... restored most of the files via chroot + qemu user static Jun 04 06:42:01 the final step would be easier if I had more than 1 millisecond to read the message before reboot Jun 04 06:47:49 so no one knows if the palm pre wifi drivers are free software compatible? Jun 04 06:50:29 which vendor/product are they ? Jun 04 06:50:43 i actually do not know Jun 04 06:50:54 cenobyte_: the best way to check if wifi is working is to do "ip l s wlan0 up; iw dev wlan0 scan" Jun 04 06:50:54 cenobyte_: if that gives you an error, then pastebin dmesg. Jun 04 06:51:58 from wikipedia Jun 04 06:51:59 It uses a Marvell sd8686 chip as a Wireless LAN SDIO Adapter for the WiFi Jun 04 06:52:14 brb Jun 04 06:52:40 by the way marvell for the gta04 is the worst foss choice ... I bought one though as I have spend some time hacking on their computer chipsets I feel I will not have wifi with free drivers easily Jun 04 06:53:14 gasp ... the hardware of marvel chip is easy to support but is so dumb ... most of the stuff is done in firmware Jun 04 06:53:39 which would not be a big issue was it for the fact they do not document how to talk to the firmware at all Jun 04 06:53:48 cenobyte_: looks like it needs non-free firmware: Jun 04 06:53:49 http://www.webos-internals.org/wiki/WLAN Jun 04 06:54:02 prahal: he is talking about palm pre, not gta04 Jun 04 06:54:12 palm pre also use marvell Jun 04 06:54:57 I didn't know gta04 used marvel Jun 04 06:58:55 I believe it will take less time to reverse engineer their firmware then write a driver ... I was so angry at their policy about how to interface with their firmware (as the hardware does nothing one even have to really know how to talk to the firmware) that I did not share my findings and code ... they are so close , not even proprietary ... plain paranoid Jun 04 07:00:37 pabs3: oh yes, it does. Jun 04 07:01:01 pabs3: and "hardware validation" kernel driver libertas doesn't work with nl80211 properly even. Jun 04 07:01:11 fun times Jun 04 07:01:48 prahal: but libertas is there and mainline, what's wrong with it? Jun 04 07:02:49 prahal: is anyone working on the firmware? Jun 04 07:02:50 they have different products Jun 04 07:03:15 a line is small firmware , the other is fully managed by firmware Jun 04 07:03:28 to protect their IP Jun 04 07:04:13 they do not want to even tell how to reset it Jun 04 07:04:56 most of their firmware are not able to do AP Jun 04 07:05:27 in fact even finding a proper firmware is a pain Jun 04 07:06:14 doing hardmac for embedded wifi might also be beneficial power-saving-wise. Jun 04 07:09:59 this is fw Jun 04 07:11:45 I meant screwing the users is not the sole reason to go hardmac. Jun 04 07:11:50 in fact the only fw one earth I bother about Jun 04 07:13:14 someone said on the SHR blog that wifi should work on the palm pre now... Jun 04 07:14:19 cenobyte_: did you try "ip l s wlan0 up; iw dev wlan0 scan" ? Jun 04 07:14:59 this post is from april 2 2011 Jun 04 07:15:00 Palm Pre: Jun 04 07:15:00 WiFi is now correctly integrated and can be used with liwi and other wireless tools. Jun 04 07:16:32 pabs3: he ignores me on purpose, i guess he wants to troll instead of making wifi work :) Jun 04 07:17:10 no i'm not Jun 04 07:17:11 sorry Jun 04 07:17:23 i am way taller than a troll Jun 04 07:17:50 do you think i should try a less-than-current image from 2011? Jun 04 07:18:31 no, try the command PaulFertser suggested Jun 04 07:18:47 if its a topdog marvell it might work Jun 04 07:20:58 prahal: at least he'll know exactly what he's missing. Jun 04 07:21:29 If e.g. he has no wlan0 at all it means he needs some platform-specific magic to activate it first (as is needed on e.g. gta04). Jun 04 07:23:05 cenobyte_: maybe your are seeing http://shr-project.org/trac/ticket/1872 Jun 04 07:24:24 the reason i hesitate is because i am reflashing as we speak Jun 04 07:24:55 cenobyte_: what image are you flashing? Jun 04 07:26:08 the 2012.1 one for palmpre Jun 04 07:26:19 http://linuxwireless.org/en/users/Drivers/libertas Jun 04 07:27:00 cenobyte_: eventually you are going to hit this issue though: http://lists.shr-project.org/pipermail/shr-devel/2012-May/004094.html Jun 04 07:28:09 He's already aware of that. Jun 04 07:28:50 don't know if i should mail him Jun 04 07:29:12 Telling what? Jun 04 07:31:04 not sure Jun 04 07:31:23 but i was not aware that the 2012.7 was more recent than the 2012.1 Jun 04 07:38:12 cenobyte_: in 2012.7 image there is no /lib/firmware/ Jun 04 07:38:20 so it will not work out of the box Jun 04 07:39:00 still in the process of flashing 2012.1 but from last time no it did not work out of the box Jun 04 07:39:35 iliwifi showed nothing on the list, rescan did nothing Jun 04 07:41:52 did you try some commands in the console? Jun 04 07:42:48 not yet Jun 04 07:42:59 will try as soon as i get system back Jun 04 07:43:10 :) Jun 04 07:43:26 booting into shr now Jun 04 07:43:49 it's a shame so few palm pre users use shr Jun 04 07:45:05 pabs3: all what PaulFertser said about admin aplies to me as well Jun 04 07:47:13 ok so connman was not installed by default again Jun 04 07:47:33 that doesn't matter Jun 04 07:48:08 we're not using connman since some time Jun 04 07:48:13 DocScrutinizer51: though i probably have a bit more experience with the modern systems. Jun 04 07:48:59 quite possibly Jun 04 07:49:36 in the connection settings i moved the wifi automatic slider to manual and slid wifi power to on Jun 04 07:49:42 it was previously off Jun 04 07:49:45 I'm even struggling to admin my own tiny centOS box Jun 04 07:50:25 and when it comes to cyrrus I'm completely lost Jun 04 07:51:56 honestly I think I have not enough time for such an admin job Jun 04 07:52:27 i tried "ip l s wlan0 up; iw dev wlan0 scan" ? Jun 04 07:52:36 it said invalid argument Jun 04 07:52:48 cenobyte_: what about "ip l", does it have wlan0 Jun 04 07:53:21 you will need to use ifconfig and iwconfig Jun 04 07:54:21 it just has loopback and usb0 Jun 04 07:55:23 iwconfig says no wireless extensions Jun 04 07:56:55 is the sirloin_wifi module loaded? Jun 04 07:57:58 lsmod shows yes Jun 04 07:59:12 grep dmesg for some output of sirloin Jun 04 08:03:21 To view all the interfaces with ifconfig one needs to do "ifconfig -a" Jun 04 08:04:11 http://corysohrakoff.wordpress.com/2011/09/13/enabling-wifi-ap-mode-on-a-gumstix-overo/ <- there seems to be an AP enabled firmware for the gta04/palm pre chip :) Jun 04 10:17:28 It feels like 2.6.34 is really faster on gta02 than 2.6.29 i'm used too. Too bad restoring GSM functionality after this upgrade is a pita. Jun 04 10:17:34 (and still in progress) Jun 04 10:27:55 PaulFertser: did you find out how to check if it's A3 or A4? Jun 04 10:28:03 mrmoku: no Jun 04 10:28:29 until it will be offered by kernel in cpuinfo or somewhere else the only difference IIRC is the existance of an GPIO Jun 04 10:28:47 have to dig probably in mails to see Jun 04 10:29:01 or compare stuff in /sys between A3 and A4 Jun 04 10:29:41 actually I think Nikolaus told me that on the last Stammtisch Jun 04 10:29:47 so nothing where I could dig into :P Jun 04 10:31:53 * mrmoku needs to reboot his server... Jun 04 10:32:03 PaulFertser: I will do that compare a bit later today Jun 04 10:32:06 bbl Jun 04 10:32:12 mrmoku: thanks! Jun 04 10:32:51 PaulFertser: with 2.6.39 on gta02 the only issue is gprs ... and a patch is floating around in the tracker for kernel ppp Jun 04 10:34:05 are udev and devtmpfs supposed to run both at the same time (devtmpfs write it start, udev does so then mmc on which the system is sitting is removed and reboot ensue Jun 04 10:34:13 I have this issue since a crash though Jun 04 10:34:25 likely a file is missing Jun 04 10:35:12 prahal: i'm yet to solve other problems, not ready to get back to kernel-level programming on gta02 yet. Jun 04 10:38:51 err hw-rev of gta04? Jun 04 10:39:05 Yep Jun 04 10:39:30 I seem to recall they use same (not overly clever) GPIO numbering Jun 04 10:39:41 i've seen sth in schematics Jun 04 10:40:01 can't check since my gta04 is dead Jun 04 10:43:07 I'd suggest to ask on #gta04, but it seems rather dead Jun 04 11:22:34 radekp: Hi :) your http://qtmoko.sourceforge.net/debian/linux-image-2.6.34-qtmoko-gta02_v44-1_armel.deb is incompatible with the new dpkg policy of not allowing the version to start with a non-digit :) Jun 04 11:25:44 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=620566 Jun 04 11:30:49 PaulFertser: ahh oki, i wanted to fix it long time ago but probably forgot... Jun 04 11:43:58 mrmoku: are you able to delete users (e.g. iwdy23) from trac? Jun 04 12:00:31 jake42: yup Jun 04 12:01:14 jake42: quite some time ago I always deleted the user too when zeroing out a spam ticket... Jun 04 12:01:19 then I got bored of that :/ Jun 04 12:14:52 mrmoku: that's understandable :) Jun 04 12:16:12 are you also able to see the configuration for the antispam-plugins? Jun 04 12:21:52 jake42: all that is configurable via trac itself Jun 04 12:22:54 jake42: right now creation of new users is disabled Jun 04 12:24:09 generally putting users on 'moderated' is way more effective than deleting the account, as the spammers will simply create a new account Jun 04 12:25:56 mrmoku: I wanted to ask if you are able to turn of the captcha thing as no new users (or spammers) can register Jun 04 12:27:14 I once asked in #bearstech to enable captcha as no one seemd to be able to configure it Jun 04 12:30:57 jake42: I will check when finished with lunch Jun 04 12:32:13 mrmoku: lass dirs schmecken :) Jun 04 12:32:38 danke :) Jun 04 12:39:43 jake42: yup I can disable it... want me to? Jun 04 12:40:09 anybody got a clue why ssh(d) is spitting out a uuencoded key over 5 lines, initially on every login? Jun 04 12:40:40 DocScrutinizer05: without any further message? Jun 04 12:40:45 must be some weird loggong/debug function Jun 04 12:40:50 yup Jun 04 12:41:19 jr@HaleBopp:~/Dokumente/downloads> ssh root@192.168.1.30 Jun 04 12:41:21 key_read: uudecode AAAAB3NzaC1kc3MAAAEBAP8Bo+QhQ3s1qaC6AyzC9Ep4ud5vuiyf9n0o/BAy/IYNgs14svWrpv6HDyrzmUUlXZYz9Lw6nyrSd9h1bNLF53OJGYHSdjDhob9FI33SIych7EbwIuQCTfGHEJH0Gdk5ol6LO/kjjoZIYvcw2OkvcUmKCtRFRFWGyWOyiFiCjgYzOu3x/9QQn6jx768RlMDtZwDxIAJ+ECmCEJo3PnXZWRVKU9ypze4nxIzpsTqzEEKjdfihAMKrxJIGJiL6QMGxnkaSW1yOrwnnCQMeotjntkwlVLZD18FI9X733 Jun 04 12:41:22 ... Jun 04 12:41:35 ZVFtUcItXLast login: Mon Jun 4 13:20:13 2012 from 192.168.1.32 Jun 04 12:41:41 DocScrutinizer05: maybe an alias with -v option? Jun 04 12:42:01 Alex[sp3dev]: alias? Jun 04 12:42:18 where would I look for that? Jun 04 12:43:32 seems it came with this suse12.1 install Jun 04 12:44:03 so I suspect it's in /etc/ssh/ssh?_config Jun 04 12:44:35 DocScrutinizer05: maybe so. I'd try to see what whereis says first. and just type 'alias' to see if it prints crap Jun 04 12:44:49 or it's been there since ages, as I rarely ever logged in via ssh on my old laptop Jun 04 12:45:04 so it could as well be on the server side Jun 04 12:45:34 I dunno of any aliases I've configured though Jun 04 12:46:29 whereis Jun 04 12:46:36 hmm. flashed my gnex with cm9. seems they haven't effed up kernel stuff like on sgs2 Jun 04 12:46:40 DocScrutinizer05: whereis ssh Jun 04 12:47:07 duh, shell alias? Jun 04 12:47:09 no way Jun 04 12:47:23 so it was that? Jun 04 12:47:30 no Jun 04 12:51:40 afk Jun 04 12:55:08 the shell is rather 'virgin' on this fresh install Jun 04 12:55:20 and has no suspicious aliases Jun 04 12:56:02 jr@HaleBopp:~/Dokumente/downloads> which ssh Jun 04 12:56:04 /usr/bin/ssh Jun 04 12:57:35 well, ssh localhost does the same, so it's this fresh opensuse21.1 ssh install that brings it Jun 04 12:58:35 I already looked into /etc/ssh/ssh_config but couldn't find any obvious key Jun 04 13:04:56 DocScrutinizer05: that looks like a key being split in two lines. Jun 04 13:05:14 DocScrutinizer05: take a look at ~/.ssh/authorized_keys Jun 04 13:23:50 does shr ship /etc/ld.so.conf.d ? I lost the content of my /etc/ld.so.conf and anything related (I bet I should not have used --reinstall on libc6 ... Jun 04 13:39:51 nvm it looks like /etc/ld.so.conf should stay empty and no /etc/ld.so.conf.d Jun 04 14:37:08 heyho Jun 04 14:38:57 letsgo Jun 04 14:40:48 PaulFertser: /sys/class/gpio/gpio10 Jun 04 14:40:53 this one I only have on A4 Jun 04 15:01:35 hi Jun 04 15:02:25 could and does a module print to standout, so I see printf() in the shell where I invoke modprobe? Jun 04 15:02:34 kernel module Jun 04 15:10:41 PaulFertser: ping ^^^ Jun 04 15:11:17 (sorry for stupid question, but I have no linux machine arround to test) Jun 04 15:19:24 DocScrutinizerST: that goes to klog I think Jun 04 15:19:39 nothing on stdout from the module Jun 04 15:19:59 sure thing modules print to klog *usually* Jun 04 15:20:35 so a module has no handle to stdout of modprobe process? Jun 04 15:20:51 prolly not Jun 04 15:20:52 ah you want to know if it is possible to write a module that does that? hmm Jun 04 15:20:55 what for? Jun 04 15:20:59 probably not yeah Jun 04 15:21:52 I'm about to specify a test module that's menat to run tests in kernel on userland behalf Jun 04 15:22:06 it obviously needs any form of UI Jun 04 15:22:07 anyway... need to shop some food... otherwise I will go to bed hungry later this night Jun 04 15:22:10 ahh ok Jun 04 15:22:40 well I used http://docs.openmoko.org/trac/ticket/2212 to fix the kernel issues with gprs (no changes for 2.6.39) ... Jun 04 15:23:05 well, there are still pty and other devices, there are files of any flavour (pipes), etc Jun 04 15:23:26 bbl, cig break Jun 04 15:24:40 DocScrutinizer: I guess netlink is meant to carry this kind of data between user/kernel space Jun 04 15:26:09 though stderr/stdout makes little sense in kernel space :( the shell which provides those pipes being above the kernel Jun 04 15:27:17 yipee gprs working .... only issue now is that eve crashes on google page :/ which is the default (and it does not save the preferences) Jun 04 15:29:30 and ping the site is fun 10 seconds :/ Jun 04 15:32:57 prahal: sounds about right. Though I could also figure the kernel module opens two filehandles, one for input and one for output Jun 04 15:33:36 so the userland running the test and insmod/rmmod'ing the module could set up those files to its liking Jun 04 15:36:09 ala console (serial or netconsole) ? Jun 04 15:38:19 as netlink is meant for such an interaction ... why avoid them avoid it ? http://en.wikipedia.org/wiki/Netlink . Dislike of the name :) Jun 04 15:38:55 I mean there is no named pipe of such nifty things for in kernel work Jun 04 15:39:09 you will have to work with buffers Jun 04 15:39:35 in fact in the end you might have achieved a netlink light **** BEGIN LOGGING AT Mon Jun 04 15:56:03 2012 Jun 04 15:58:44 prahal: many thanks for the comment Jun 04 15:59:52 DocScrutinizerST: systemtap does what you want nicely. Jun 04 16:00:44 DocScrutinizerST: i.e. it allows you to run arbitrary code in kernelspace and output the result to stdout of the "stap" userspace process. Jun 04 16:01:16 PaulFertser: :-) Jun 04 16:01:35 I guess it's a debugging tool though Jun 04 16:01:45 not meant to build a standard UI to a kernel module Jun 04 16:01:49 right? Jun 04 16:01:57 DocScrutinizerST: not only a debugging, it's supposed to be used for various tasks. Jun 04 16:02:19 DocScrutinizerST: if it fits the purpose, why not? Jun 04 16:03:35 the purpose is to run a kernel driver for a phy IF at maximum speed, I.E. with next to zero overhead for uerland sending next datachunk to transmit Jun 04 16:05:01 for that I schall design a tool than can get installed and run on "normal" embedded system (android, yes I know...) Jun 04 16:05:02 shall* Jun 04 16:05:51 this tool should be easy to use, preferrably "portable" from one platform to next, selfcontained... Jun 04 16:06:46 what you suggest sounds to me like "use gdb!" Jun 04 16:07:10 is it similar? or am I wrong Jun 04 16:08:07 hell, I don't evenknow if systemtap is available or at all feasible on android Jun 04 16:10:03 the kernel module under test has a PHY driver to the bottom and a SOCK and NetDev API to the upper/userland Jun 04 16:10:59 usually IP-stack connects to the SOCK API, we don't want any IP stack in this test, for verious partially obvious reasons Jun 04 16:11:57 so the idea was to implement a test kernel module that mimicks IP and connects to CAIF's SOCK API, flooding it with UDP Jun 04 16:13:03 the heck, LTE can go to 100Mb/s downstream ;-) Jun 04 16:13:35 the PHY (HSI) is not even guaranteed to do that speed Jun 04 16:15:13 I'm counting machine cycles in the test's service routine to that SOCK, we need sub-microsencond latency Jun 04 16:15:48 hi JaMa Jun 04 16:16:55 btw if you wonder - this is not completely OT, as I'm helping to build your next platform's modem ;-) Jun 04 16:18:07 hi, .*ST :) Jun 04 16:18:28 the E at end didn't fit ;-) Jun 04 16:19:17 btw: do you know Vancak from STe Prague? by chance? :) Jun 04 16:22:55 JaMa: nope Jun 04 16:23:59 * JaMa off to meeting Jun 04 16:24:39 JaMa: not even in my company addrbook Jun 04 16:25:08 anyway NETLINK sounds good indeed Jun 04 16:26:17 though I prefer the kernlmodule parameter on modprobe, plus grep results from syslog approach ;-D Jun 04 16:26:52 we need to load the module anyway and don't want to run it all the time Jun 04 16:31:46 could a kernel module do a fopen() ? Jun 04 16:32:19 or sth similar? Jun 04 16:32:31 yeah Jun 04 16:32:36 filp_open it is called iirc Jun 04 16:32:45 thanks Jun 04 16:33:23 samsung even did a firmware loader that mounts ext4 partition and opens file in a kernel module.Oo Jun 04 16:50:27 Alex[sp3dev]: did they do this for them (STE) modem firmware? ;-D Jun 04 16:51:22 DocScrutinizerST: i was unable to parse your question. if you want, you may look at samsung tizen git, there's a piece of this, so to say, _code_ Jun 04 17:02:40 Alex[sp3dev]: I just wondered if Samsung did that kernel firmware was the one their linux transfers to the modem, on boot Jun 04 17:03:13 dang, I seem incapable of writing down oe grammatically correct sentence Jun 04 17:03:15 yeah, that one. i guess modem is intel, not ste though Jun 04 17:03:43 probably no big difference (from an outside POV) Jun 04 17:08:58 Alex[sp3dev]: what year was that? Do you really mean they "reinvented" the standard in-kernel firmware loader? Jun 04 17:11:09 PaulFertser: enjoy your reading https://review.tizen.org/git/?p=kernel/linux-2.6.36.git;a=blob;f=drivers/misc/modem/xmm6260.c;h=6e32ce0c5d065c727d57a0894650428ebf22ac99;hb=refs/heads/master Jun 04 17:11:42 static char *envp[] = { "HOME=/usr/home/root", 492 "TERM=linux", Jun 04 17:13:58 Joonyoung Shim !!! we should ask Raster to go punch him :) Jun 04 19:11:51 lol Jun 04 19:12:44 PaulFertser: you've seen the pointers into ugly code I shared recently? Jun 04 19:13:14 DocScrutinizer05: not yet, i'm afraid Jun 04 19:13:25 DocScrutinizer05: have you seen my suggestion about that ssh error? Jun 04 19:13:34 unclear whom to punch for that though, as it claimed to be "copied from Unix" Jun 04 19:13:44 nope Jun 04 19:14:25 split into two lines? Jun 04 19:14:52 well I'll look into it, but I pasted only start and end of that cruft, separated by '...' Jun 04 19:15:18 as original spans 5 or 6 lines on a terminal with ~150char width Jun 04 19:15:59 actually it seems it's not split into separate lines but simply 5 times wrapped around Jun 04 19:16:27 anyway, I copied the key file from the other machine Jun 04 19:16:40 works fine over there Jun 04 19:19:09 PaulFertser: it's exactly the id_dsa.pub it seems Jun 04 19:21:10 or the line in known_hosts for that machine, which looks identical Jun 04 19:24:23 DocScrutinizer05: what happens if you delete it from authorized_keys? Jun 04 19:48:35 PaulFertser: dammit, my authorized keys and id_dsa.pub was borked - both truncated Jun 04 19:49:02 1023 bytes, I guess that's no random length Jun 04 19:49:22 DocScrutinizer05: heh Jun 04 19:49:58 should've been 1115 Jun 04 19:52:14 freesmartphone.org: 03morphis 070.11 * rf036e08780d1 10cornucopia/fsoaudiod/src/lib/Makefile.am: fsoaudiod: lib: add missing fsoaudio-2.0.vapi and fsoaudio.h files to EXTRA_DIST Jun 04 19:58:16 now it works in either direction flawlessly Jun 04 19:59:02 though it seems odd to log in ssh localhost and no password gets asked Jun 04 20:02:14 but that's just a wierd feeling I shall get over Jun 04 20:04:29 when the local id_dsa.pub is identical with the local authorized_keys, I can't see any security threat though it feels strange Jun 04 20:05:31 now I guess I just have to do same shite for root, though I probably don't want to use same keypair for that ;-P Jun 04 20:06:58 dman is that a PITA to migrate a whole system without exactly cloning it Jun 04 20:07:40 this xchat client allone drives me nuts Jun 04 20:08:23 don't get me started about KDE... :-x Jun 04 20:16:15 PaulFertser: so many thanks for the sterling tip about split key line. It was absolutely to the point, though my second half of some 80 chars simply been missing Jun 04 20:16:55 some 90 even Jun 04 20:16:59 DocScrutinizer05: i'm glad it worked :) Jun 04 20:18:39 so much for today with private sysadmin, special nighmare kmail incl mail migration tomorrow... maybe Jun 04 20:28:54 PaulFertser: http://lxr.free-electrons.com/source/net/caif/caif_socket.c#L377 Jun 04 20:28:57 ff Jun 04 20:29:02 :-S Jun 04 20:29:57 346 * Copied from unix_stream_recvmsg, but removed credit checks, Jun 04 20:29:59 347 * changed locking calls, changed address handling. Jun 04 20:30:39 so I'd guess the original evil was by author of unix_stream_recvmsg & siblings Jun 04 20:31:41 * DocScrutinizer05 glares at memcpy_toiovec in line 426 Jun 04 20:31:59 * DocScrutinizer05 tries to figure count of CPU cycles for that Jun 04 20:33:52 http://lxr.free-electrons.com/source/net/core/iovec.c?a=microblaze#L82 *YAAAAAAAAWN* Jun 04 20:34:22 DocScrutinizer05, hi, do you know the maximum speed of the s3c24xx serial ports, in other words in the freerunner is there an external clock for the serial line or not(if there is none it's limited to 115200) Jun 04 20:34:48 I'd guess there's none Jun 04 20:35:55 ok Jun 04 20:35:59 I cloudily recall we "forgot" to connect two pins under SoC, might be related Jun 04 20:36:13 thanks a lot Jun 04 20:36:52 all disclaimer "IMHO & IIRC & AFAIK" Jun 04 20:38:50 probably for the shitty calypso even 19200 was enough ;-D Jun 04 20:41:00 PaulFertser: you might argue each semi-decent compiler will optinize this BS of assigning one value after the other to err, then do a conditional goto or break. But honestly, this makes me vomit by looking at it Jun 04 20:43:21 code should be a) efficient and b) readable and clear, whatever comes first. This code is neither of both Jun 04 20:45:21 even a prettyprinter would improve the code ;-P Jun 04 20:49:43 depend on wht you want to do.... Jun 04 20:58:36 gnutoo: look at lines http://lxr.free-electrons.com/source/net/caif/caif_socket.c#L394 plus 2 Jun 04 20:58:44 that's idiotic Jun 04 20:59:07 next 5 lines the same Jun 04 20:59:11 pure mess Jun 04 21:01:17 hmmm Jun 04 21:01:36 recently on linux-wireless there was patches correcting stuff like: Jun 04 21:01:53 if (foo | SOMETHING) Jun 04 21:02:00 instead of & Jun 04 21:02:29 if (sk->sk_state != CAIF_CONNECTED) Jun 04 21:02:30 || (sock_flag(sk, SOCK_DEAD)) Jun 04 21:02:32 { Jun 04 21:02:33 err = -EPIPE; Jun 04 21:02:34 goto unlock; Jun 04 21:02:36 } Jun 04 21:03:33 that's readable clean code Jun 04 21:03:37 in my book Jun 04 21:05:19 lol indeed Jun 04 21:05:25 maybe send a patch.... Jun 04 21:05:27 not this abomination at http://lxr.free-electrons.com/source/net/caif/caif_socket.c#L398 Jun 04 21:05:48 indeed Jun 04 21:07:34 oh, there will be patches "reported by Joerg Reisenweber" Jun 04 21:08:18 but I'm not supposed to do major patching on public domain @ work, not even when it's "our2 own code Jun 04 21:08:26 "our" Jun 04 21:09:29 ok Jun 04 21:09:33 the whole docu is massively out of date and obsolete, e.g. kernel API doesn't exist anymore Jun 04 21:10:06 modinfo will most likely tell nonsense, a c&p error Jun 04 21:11:19 one of the high_watermark module parameters should be low_watermark, or sth like that (don't recall details, it's been last week I found it) Jun 04 21:18:45 is anything known about SHR ticket 2012 (fsogsmd hogging CPU)? I was thinking of investigating but I don't want to reinvent the wheel. Jun 04 21:22:20 jeepingben: excellent approach Jun 04 21:23:41 the damn ticket number is extremely unsuited to search in chanlog ;-D Jun 04 21:24:11 actually every single post since 5 month has it Jun 04 21:24:42 maybe gnutoo knows more Jun 04 21:25:08 gnutoo: ^^^? Jun 04 21:25:52 hi Jun 04 21:26:17 jeepingben, I know exactly why Jun 04 21:26:27 basically we use alsaloop Jun 04 21:26:34 and it works like that Jun 04 21:26:43 there is a call incomming or outgoing Jun 04 21:26:48 fsogsmd knows it Jun 04 21:26:58 so it tells it on dbus Jun 04 21:27:07 fsoaudiod which is listenning on dbug get that message Jun 04 21:27:16 and activate the alsaloop based plugin Jun 04 21:27:18 however.... Jun 04 21:27:29 there is a time between that activation and when the modem is ready Jun 04 21:27:40 because the modem is the one who is clock master Jun 04 21:27:53 ouch Jun 04 21:27:58 so then what happens in alsaloop is that poll() returns too fast Jun 04 21:28:18 instantly with -EAGAIN I guess Jun 04 21:28:49 like the asoc poll() is broken for our case where the clocks master are from the modem Jun 04 21:29:05 in which case alsaloop SHOULD call a usleep(1000) Jun 04 21:29:36 that poll() returning too fast (30us) instead of 30000000us or something big like that causes the 100% CPU usage Jun 04 21:29:52 and then the 100% CPU usage causes other problem once the modem is started Jun 04 21:29:56 because it doesn't sotp Jun 04 21:29:58 *stop Jun 04 21:30:01 yep, that's why alsaloop needs a usleep() Jun 04 21:30:14 ah ok Jun 04 21:30:23 how should I put that? Jun 04 21:30:44 thanks Jun 04 21:30:44 in case that poll returns with the typical error it should throw when master clock not running Jun 04 21:30:45 If ( delay < 100us) Jun 04 21:30:58 sleep(30000000us) Jun 04 21:30:59 ? Jun 04 21:31:12 I probably won't be able to get very far with that Jun 04 21:31:12 it returns no error I think Jun 04 21:31:13 sleep(1000us) Jun 04 21:31:40