**** BEGIN LOGGING AT Thu Jun 11 02:59:57 2009 Jun 11 03:00:09 And the commit should start with 935626 Jun 11 03:00:18 <9 +!*(70.0€ Jun 11 03:00:30 errks. no backup.py Jun 11 03:01:52 shr_backup.py? Jun 11 03:02:27 none Jun 11 03:02:58 can you past the line given by: `opkg list_installed | grep "shr-settings "` ? Jun 11 03:03:35 * Blu3 just runs tar over ssh Jun 11 03:04:16 paste from term to pidgin doesn't work. So ... sorry Jun 11 03:06:38 Blu3: that's basically what I do, but there are white/blacklists and it's setup so standard bits can be loaded from a future shr-settings-config package, and user defined lists are also available. Jun 11 03:07:19 Blu3: http://trac.shr-project.org/trac/ticket/495. Jun 11 03:07:21 how doesn't it work? Jun 11 03:07:22 ...652500-r5 Jun 11 03:07:31 shift-ctrl-c, to copy Jun 11 03:07:43 DocScrutinizer: that should be it... Jun 11 03:07:45 odd Jun 11 03:08:03 ahh, without the config files it failes... makes sense I sppose Jun 11 03:08:13 s/sppose/suppose/ Jun 11 03:08:13 Toaster` meant: ahh, without the config files it failes... makes sense I suppose Jun 11 03:08:46 for me, $root is on flash, $home on sd Jun 11 03:09:10 Blu3: so? you can define where the archive file goes. Jun 11 03:09:23 eh? Jun 11 03:09:54 what are you talking about? Jun 11 03:10:38 you were commenting about $ROOT is NAND and $HOME is on uSD Jun 11 03:10:49 Blu3: have you ever tried shift-ctrl-c? Jun 11 03:10:50 right..? Jun 11 03:10:56 doc, yep Jun 11 03:11:00 that shouldn't matter with the way I have it set up Jun 11 03:11:17 ctrl-c,ctrl-v is the standard method of cut 'n paste Jun 11 03:11:39 in a terminal those are often used for other things so prepend them with shift Jun 11 03:11:57 if you are using a mouse, shift-right click Jun 11 03:12:13 hmm, shift doesn't change ctrl-c here Jun 11 03:12:22 since right click may be designated already Jun 11 03:12:37 aaah yes. mouse ;-> Jun 11 03:12:45 have you tried simple mouse highlight and middle nouse paste? Jun 11 03:13:18 pardon my typoes. i sliced up my fingertips last night =) Jun 11 03:13:22 -e Jun 11 03:13:23 Blu3: tell me: how am I supposed to hit the middle mouse button on FR? Jun 11 03:13:35 hmm Jun 11 03:13:55 i'd figured it out sometime past Jun 11 03:14:45 well, I could try to hook a mouse to USB ;-) Jun 11 03:15:01 these days i just ssh in Jun 11 03:15:50 if I were ssh in, I wouldn't mention problems c+p from term to pidgin Jun 11 03:16:52 DocScrutinizer: no ssh? Jun 11 03:17:22 no PC next to FR to run a ssh Jun 11 03:27:11 ok folks, I'm off. Good night Jun 11 03:33:17 muuuhhaahAAAA: illume settings, key bindings, add key -> "please press key sequence, or *escape* to abort". Seems there's NO way to remove this on-top-requester :-S Jun 11 03:36:23 well, "enligthenment error, SEGV" Jun 11 03:38:00 but finally I closed this "key sequence" requester :-D Jun 11 04:55:44 re Jun 11 05:38:06 morning Jun 11 05:38:45 deep night ;-) Jun 11 05:42:32 wpwrak: :)) hey, btw, what do you think about enabling ramconsole unconditionally and by default on OM kernels? Jun 11 05:44:03 sounds okay to me. just make sure there are no transition issues, like the system going berserk if you run an old kernel Jun 11 05:46:11 PaulFertser: there's probably a way to fix it to work without mem=127M: #debian-fi.06-06.log:17:30 < p2-mate> lindi-: you could hack reserve_node_zero Jun 11 05:47:51 lindi-: oh, that's even better. And enabling panic timeout or watchdog unconditionally would be beneficial too, as not everybody always runs latest Qi or tweaks u-boot cmdline properly. Jun 11 05:49:32 PaulFertser: that i don't want. you can't use debug board easily if watchdog runs Jun 11 05:49:58 hmm, no severe issues with SHR-unstable current so far Jun 11 05:50:50 and moooooooo Jun 11 05:54:18 what's the details of this watchdog/panic-timeout thing? Jun 11 05:54:41 just to reboot without power-cycling? Jun 11 05:55:16 tracfeed: Ticket #514 (Feedback *needed* when sending an sms fails) created Jun 11 05:55:56 DocScrutinizer: yes Jun 11 05:56:04 DocScrutinizer: so that ram contents does not get lost so easily Jun 11 05:56:23 pfft, feedback?! Jun 11 05:57:07 k, we got the aux-flash hack Andy inplemented. could we augment this code snipplet to reboot on pressing aux (only if panic-flashing)? Jun 11 05:57:50 DocScrutinizer: maybe but that does not help with the case where glamo jams nWAIT Jun 11 05:58:14 ouch, yes that's a nasty one Jun 11 05:59:30 still we could get _most_ of the panic logs this way, even without watchdog enabled Jun 11 05:59:41 DocScrutinizer: panic=20 is enough? Jun 11 05:59:42 lindi-: you could add software delays and disable nWAIT ;-) Jun 11 05:59:49 wpwrak: really? Jun 11 06:00:19 nah, it's late at night in BUE Jun 11 06:00:43 lindi-: you can disable nWAIT. not sure if the software delays would work properly, though. i remember that, in addition to the nWAIT, the glamo also needed some extra delays in some cases. Jun 11 06:01:15 lindi-: in any case, that would need quite a bit of research, as the waits may not be constant Jun 11 06:02:10 lindi-: (disable nWAIT) i did this successfully for some power measurements: disable nwait, than hang the glamo and see what this does to current. you can't imagine how satisfying the hanging part was ;-) Jun 11 06:02:25 wpwrak: can you disable it with software? Jun 11 06:02:41 so wtf?! If cpu can *really* do useful stuff while nWAIT, we could even poll and still get better over-all performance Jun 11 06:03:55 I guess we couldn't poll though Jun 11 06:04:18 or is this a multi-function cpu pin? Jun 11 06:04:28 DocScrutinizer: in any case, you couldn't touch the bus. also, i don't know when the glamo samples. if it doesn't sample at the beginning of the cycle, then we have a problem. Jun 11 06:05:18 lindi-: yup.cpu register BWSCON Jun 11 06:05:55 hmm, intriguing topic. Anyway I'd like to close the lid on ramconsole-for-default ATM Jun 11 06:06:37 DocScrutinizer: if the glamo uses nWAIT only to signal that it's busy after having sampled, you could poll nWAIT in software by reading GSTATUS0 Jun 11 06:06:50 lindi-: so, what do you propose so that "ordinary" users will report kernel crashes properly? Or is it just a dream and they won't do that whatever we include in kernel? Jun 11 06:06:58 DocScrutinizer: yeah, sorry for the disgression. it's a fun topic :) Jun 11 06:07:07 yup Jun 11 06:07:50 PaulFertser: i'd suggest you add an automatic "last dmesg dumper" to the distros Jun 11 06:08:30 PaulFertser: ramconsole or mtdoops Jun 11 06:08:37 wpwrak: yes, but for that to work the kernel should be rebooted anyway. lindi- says he dislikes "watchdog by default" idea. Jun 11 06:08:46 PaulFertser: I suggest deleteing a flag-file on shutdown, and create it on sys-init after checking if it's there and doing the buffer-dump Jun 11 06:09:04 wpwrak: heh Jun 11 06:09:08 PaulFertser: actually, i don't use those distros. maybe you can set watchdog to be on by default :) Jun 11 06:09:45 PaulFertser: so go ahead and make watchdog default. I only control it by enabling or disabling the module in /etc/modules on debian Jun 11 06:10:03 PaulFertser: watchdog sounds good to me. if the kernel is crashed, you may as well try to salvage what you can Jun 11 06:10:09 lindi-: should be easy to disable it if you plan to use debugboard Jun 11 06:10:11 PaulFertser: so make watchdog and panic=20 default Jun 11 06:10:17 PaulFertser: that is, unless the watchdog is known to malfunction Jun 11 06:10:19 PaulFertser: remember to configure watchdog timeout to 42 seconds Jun 11 06:10:26 PaulFertser: the default is imho way too short Jun 11 06:10:37 DocScrutinizer: aah, watchdog vs. jtag. i see Jun 11 06:11:14 PaulFertser: then add forcing mem=127M to the ramconsole patch and you'll have a safe setup that doesn't corrupt data unless user manually passes mem=128M Jun 11 06:11:22 /etc/init.d/watchdog stop Jun 11 06:11:35 DocScrutinizer: does that really stop it? Jun 11 06:11:46 lindi-: sure. Great we reached consensus. Now somebody just needs to actually do that :) Jun 11 06:11:51 nah, no idea Jun 11 06:11:59 * wpwrak wonders if there's a way to run the watchdog in cpu time. probably not ... Jun 11 06:12:00 DocScrutinizer: i think it does not disable watchdog Jun 11 06:12:07 wpwrak: in cpu time? Jun 11 06:12:26 lindi-: some counter that's stopped when jtag halts the cpu Jun 11 06:13:19 wpwrak: so you wouldn't use external watchdog hardware at all? Jun 11 06:14:58 hmm, external? Jun 11 06:15:55 s/external// Jun 11 06:16:35 lindi-: ah, if we use the hw watchdog, doesn't it stop when jtag halts the cpu ? Jun 11 06:17:08 lindi-: well, let me start over. which watchdog are we talking about ? hw, sw, or both ? :) Jun 11 06:17:28 hw Jun 11 06:17:34 wpwrak: no it won't stop when jtag halts cpu Jun 11 06:17:48 wpwrak: it will kill the system if you halt it for longer than watchdog period Jun 11 06:17:57 wpwrak: hw watchdog Jun 11 06:18:03 ah yes, manual says so. grrr :-( Jun 11 06:18:40 patch openocd to always disable the wathcdig before halting the cpu :) Jun 11 06:18:54 s/before/when/ Jun 11 06:18:54 wpwrak meant: patch openocd to always disable the wathcdig when halting the cpu :) Jun 11 06:19:01 I would suggest to have that "ramconsole=" kernel cmdline parameter Jun 11 06:20:42 =0 means no watchdog, no mem-buffer allocation, no buffering of printk Jun 11 06:21:42 := buffersize in kbyte, default = 512 or 1024 Jun 11 06:23:32 DocScrutinizer: that can be done but it won't help if most people don't even know how to configure their boot loader or have a boot loader that can not be configured Jun 11 06:24:00 DocScrutinizer: we are talking about catching the rare "once in a blue moon" crashes Jun 11 06:24:02 default = 512 Jun 11 06:24:10 ok then maybe Jun 11 06:24:25 there's a variable for total memory size? Jun 11 06:24:34 i think it's too late to decrement it in ramconsole init Jun 11 06:24:40 U know how to uboot-setenv "ramconsole=0" Jun 11 06:24:46 I do Jun 11 06:25:06 but yesterday there was a case when somebody had Qi in nand and distro in nand Jun 11 06:26:28 so he probably has no ramconsole parameter set for kernel cmdline -> default will kick in Jun 11 06:27:49 wpwrak: you worked on the wifi driver, right? Jun 11 06:28:03 * wpwrak ducks Jun 11 06:28:23 haha Jun 11 06:28:24 if you have problems to evaluate the value of ramconsole-parameter early on mem-init, then allocate some 20meg and free most or all of it later during ramconsole-init Jun 11 06:28:47 wpwrak: any thoughts on https://docs.openmoko.org/trac/ticket/2277 ? I found the bug (see comments), but not exactly sure how to fix it Jun 11 06:33:41 i'd like to hear about that too :) Jun 11 06:34:01 it clearly broke somewhere during transition to kernel module? Jun 11 06:34:11 but it did not break in the transition to new sdio stack Jun 11 06:36:05 I would also like to point out that cscope is an amazing tool. If any of you use emacs or vim to read or write C and haven't heard of it, go read about it NOW. Like tags but 1000 times better Jun 11 06:36:51 emacs + cscope pwns Jun 11 06:38:07 i'm looking through the wifi driver right now, and cscope takes all the tedium out Jun 11 06:38:21 budfive: what if i have heard and read about it but never used and already forgot what is it good for? Jun 11 06:38:30 PaulFertser: then you're crazy Jun 11 06:38:30 :) Jun 11 06:38:40 budfive: that's not news Jun 11 06:39:30 lol Jun 11 06:39:37 budfive: i use linux cross-reference online and grep and vimpager offline. Jun 11 06:39:38 Seriously, reading unfamiliar code becomes much simpler. Having it immediately tell you who calls a particular function, and to be able to trace the control flow is invaluable Jun 11 06:40:34 PaulFertser: reading about lxr... Jun 11 06:41:28 budfive, lindi-: yeah, i wonder how many problems are actually in that ticket. the spontaneous failure seems to be different from the ifconfig-triggered one. Jun 11 06:41:38 wpwrak: at least 3 :) Jun 11 06:41:50 * wpwrak sighs Jun 11 06:42:37 wpwrak: there's the "iwconfig power off" bug that is patched in the ticket, lindi's crash-on-rmmod and the ifdown-ifup bug Jun 11 06:42:53 PaulFertser: is lxr just those pages on the web? Jun 11 06:43:07 budfive: you can do that locally too. Jun 11 06:43:33 PaulFertser: "do that" means generate the html, or as some sort of emacs mode? Jun 11 06:44:09 budfive: no, install cgi scripts and index the sources. Jun 11 06:45:00 budfive: in fact, i sometimes think that even recursive grep over all kernel source tree takes less time than other tasks needed to investigate the problem. Jun 11 06:45:27 So if it's busy grepping i just take my time thinking. Jun 11 06:45:29 PaulFertser: I see. Then I would argue cscope is at least better for some things. As a part of the editor it can greatly speed up dev, since you can edit the code as you find it Jun 11 06:46:32 budfive: sure, no doubt cscope is a nice tool. Jun 11 06:46:43 * DocScrutinizer waits for wpwrak pointing at git-grep ;-) Jun 11 06:47:30 reading... :) Jun 11 06:49:42 budfive: so the "failure without user action" i had spotted would be #4 :-( Jun 11 06:51:16 wpwrak: what bug is that? is that one of the comments? Jun 11 06:51:16 DocScrutinizer: very good idea :-) PaulFertser: you may like my little "gg" script: invokes git-grep and then fires up vi on what it's found, with a search for the string pre-set Jun 11 06:51:21 DocScrutinizer: thanks for reminding me about git-grep! Jun 11 06:51:42 budfive: #2277, 3rd entry Jun 11 06:52:13 wpwrak: yeah, i looked at that. It could possibly be related to one of the ready or enabled variables going false Jun 11 06:52:33 PaulFertser: it pauses before the vi, so if your search isn't narrow enough or if you don't actually want to poke around, you can kill it there Jun 11 06:52:54 wpwrak: or, nice! :) Jun 11 06:53:00 s/or/oh/ Jun 11 06:53:00 PaulFertser meant: wpwrak: oh, nice! :) Jun 11 06:53:55 wpwrak: so is the ar6k supposed to send itself a ready packet when it's ready? The code appears to do this, but after "ifconfig eth0 up", nothing is sent at all (ar6000_rx is never called) Jun 11 06:54:08 wpwrak: the code appears to EXPECT this, i meant Jun 11 06:56:01 budfive: hmm, i haven't come across a "ready packet" yet, so that would be some yet unexplored twist Jun 11 06:57:16 wpwrak: look at what sets arWmiReady to TRUE, and trace that call path Jun 11 06:57:34 wpwrak: we have docs for the ar6k, right? I can't seem to locate them on the wiki Jun 11 06:58:11 docs ? hihi ;-) Jun 11 06:58:15 wpwrak: it would help if users had the permission to retitle bugs :) Jun 11 06:58:46 lindi-: you should talk to a trac-hater about this ;-) Jun 11 06:59:08 maybe i'll fork bugs.openmoko.org soon Jun 11 06:59:14 i already have a search engine for it :) Jun 11 06:59:33 lindi-: well, trac or any bugtracker that doesn't have a sensible interface, where sensible interface = { mail }. Jun 11 07:00:04 lindi-: yeah, your search engine is great Jun 11 07:00:10 wpwrak: So no docs then? Facial expressions travel poorly over irc :) Jun 11 07:02:05 budfive: the code is pretty much all we have in terms of docs Jun 11 07:02:35 wpwrak: who wrote the code? Jun 11 07:02:44 Motherfuckers Jun 11 07:03:20 Be polite young boy, and wash your mouth! Jun 11 07:04:16 bonjour Ainulindale Jun 11 07:04:19 Ainulindale: shutup Jun 11 07:04:22 Hey o/ Jun 11 07:04:28 Ainulindale: :P Jun 11 07:04:48 PaulFertser: so the bq27000 isn't programmed at all ? Jun 11 07:04:51 good morning Jun 11 07:05:07 * Ainulindale slaps PaulFertser with a ???? Jun 11 07:05:25 wpwrak: i don't think so. I haven't found any default values in the datasheet so i assume they'd be 0xff if left unprogrammed. Jun 11 07:05:30 Anyway I have to leave Jun 11 07:05:32 See you guys Jun 11 07:05:35 Ainulindale: Ur utf-8 still borked Jun 11 07:05:36 Ainulindale: luckily i saw only ????+ Jun 11 07:06:05 Ainulindale: have a nice day Jun 11 07:06:31 PaulFertser: 0xff or 0x00 seem like plausible assumptions, yes Jun 11 07:06:52 wpwrak: but judging by the fact that nobody knows who and what programmed in bq27000 for OM i think the point didn't receive the attention it deserved. Jun 11 07:07:21 PaulFertser: well you'd surprised how it hurts getting slapped with "?????" Jun 11 07:08:15 PaulFertser: i think that should all be done by the battery maker. i wonder what they were told ... Jun 11 07:08:21 PaulFertser: bat is OEM Jun 11 07:08:53 OM does not know what exactly they did at bat fab Jun 11 07:09:28 DocScrutinizer: yep, understood :( But probably OM should have controlled the process more tightly. Jun 11 07:10:36 well, bat was ordered, was delivered, got some tests, and seemed to work. So everybody was happy Jun 11 07:13:49 DocScrutinizer: bat was spec'ed by openmoko. the bq27000 is way too expensive that anyone in asia would touch it unless forced to ;-) Jun 11 07:14:37 DocScrutinizer: but i don't know how much else happened after the battery maker (probably reluctantly) put that chip into the package Jun 11 07:14:38 DocScrutinizer: i'm not blaming you or anyone in particular :) It's just a bit unfortunate that we don't get the level of accuracy we could have archived. Jun 11 07:15:53 well, I think we need decent tests to even state any (lack of) accuracy Jun 11 07:16:30 DocScrutinizer: i think your log confirms that. Jun 11 07:16:58 for now we got one batlog, and we see a calibration happening there - and this batlog iwn't even a full cycle Jun 11 07:19:29 DocScrutinizer: and datasheet explicitly states that LMD change is happening when the voltage reaches EDV1 and that should correspond to 6.25%. Jun 11 07:20:37 wpwrak: one kernel versioning question: should device name be included in version string? To be able to boot kernels for different devices (read gta01 and gta02) (with corresponding modules) from the same rootfs? Jun 11 07:20:44 hmm, but we don't know if the chip is really using a fixed EDV1 Jun 11 07:21:30 DocScrutinizer: really? From my understanding of datasheet it's EEPROM constant that doesn't change. Jun 11 07:22:45 well, we need more tests, to show what we expect and what we get. Otherwise it's ( (c)andy) handwaving Jun 11 07:25:24 DocScrutinizer: nah, we don't really need more tests. I don't expect users disassembling the pack to supply 21V to one of the qfn pins to reprogram the chip :) Jun 11 07:25:49 if we say 3V5 is lower limit due to gsm operating range, then the CC seems not that wrong Jun 11 07:26:27 we got a programming TP afaik Jun 11 07:26:45 PaulFertser: hmm, that ability doesn't sound like a bad idea. i wonder if this has to go into the version string or if there are other possibilities, e.g., telling the module loader to use a different tree Jun 11 07:27:03 yeah, but no hole in the plastic :-( Jun 11 07:27:13 wpwrak: afaict the second way is more of a hassle. Jun 11 07:28:27 DocScrutinizer: i wouldn't consider capacity outside of GSM operating range to be irrelevant. I'd better tell users "if you battery is lower than 15%, you can't use gsm". Jun 11 07:29:17 yeah, but maybe tpe EE specd otherwise Jun 11 07:30:08 anyway it's not as simple as "bq27k isn't programmed and totally wrong" Jun 11 07:30:38 Sure, i never said that. Jun 11 07:30:57 that's why I say, we need an idea of what we *expect* and what we actually get Jun 11 07:31:26 PaulFertser: (bq27k) your posting could be interpreted that way Jun 11 07:31:38 maybe bq27k is perfect, just for a spec differing from yours Jun 11 07:32:18 PaulFertser: (version) perhaps, yes. that would be a change to CONFIG_LOCALVERSION ? (as you can tell, i'm not much of a user of automatic module loading) Jun 11 07:32:40 "though probably some eeprom values that should be factory-programmed are not optimal" Jun 11 07:32:57 oh, excellent. more project b guesses :) not enough for having a bowl of pop corn with them, though Jun 11 07:33:16 Got to go (job interview). See ya later! Jun 11 07:33:22 ChristW: hey Jun 11 07:33:35 ChristW: have you seen my random soldering iron? Jun 11 07:33:48 PaulFertser: Nope... Jun 11 07:33:54 ChristW: i did it for you actually Jun 11 07:34:02 I did see someone else's though ;-) Jun 11 07:34:03 (or for my self-esteem) Jun 11 07:34:15 Where can I view your random soldering iron? Jun 11 07:34:20 ChristW: https://paulfertser.is-a-geek.org/files/useless_buzzfix/ Jun 11 07:36:03 s/useless/exercise-hardcore/ Jun 11 07:36:13 :-) That could be named a 'random soldering iron'... The one I saw in action at least had a _name_ on it! Jun 11 07:36:22 BTW, dos1's suggestion to disable occupyresource cpu workaround did wonders, i can suspend again anytime i need. Jun 11 07:36:29 Anyway, gtg, I'm supposed to be there in 25 minutes... Jun 11 07:36:37 ChristW: sure, good luck Jun 11 07:36:57 PaulFertser: i guess if it's good enough for a MiG, it's good enough for a neo ;-) Jun 11 07:37:12 got no suspend issues at all on yesterdays unstable Jun 11 07:37:27 In fact, almost all soviet stuff has a fixed price written on it. This soldering iron has it pressed in metal part. Jun 11 07:38:01 PaulFertser: nice. stable prices :) Jun 11 07:38:13 wow THAT's price stability ;-D Jun 11 07:38:22 heh Jun 11 07:38:36 :-D Jun 11 07:38:48 * wpwrak wonders if they use stone tablets in the moscow stock market Jun 11 07:39:07 lol Jun 11 07:40:51 wpwrak: it also was the case because it was illegal to resell things (at least if you do it to get profit). It's called "speculation" and was a (probably criminal) offense. (if you sell your old bed because you don't need it it's ok, but if you bought something to resell later, then it's offense). Jun 11 07:41:39 PaulFertser: there was always a healthy black market Jun 11 07:42:15 budfive: sure, but it was illegal. I'm just trying to explain the reason for engraving prices on stuff. Jun 11 07:42:35 PaulFertser: yeah, i could imagine that speculation and 5 year plans don't go along very nicely Jun 11 07:43:24 wpwrak: at one point of time there was a slogan "(do) 5 year plan in 4 years" :) Jun 11 07:43:28 actually speculation would have been the only way to ever fulfill a single 5-year plan ;-) Jun 11 07:44:06 Nah, 5-year plan is about how much iron, wood and other stuff must be produced. Jun 11 07:45:27 Probably budfive can tell you more about it, as i'm too young so i haven't seen all that socialist shit myself. Jun 11 07:45:33 so they floated the vodka production ? not even minimum levels, like 1 l per person and day ? Jun 11 07:45:52 PaulFertser: I suspect we're roughly the same age :) Jun 11 07:46:01 budfive: i was born 1985 Jun 11 07:46:05 1981 Jun 11 07:46:38 wpwrak: (vodka) hm, i need to ask someone about it. Jun 11 07:49:41 wpwrak: they tell me vodka production was included in 5-year plan. Jun 11 07:51:05 PaulFertser: very sensible. so that wasn't the flaw that brought the system down then Jun 11 07:59:31 wpwrak: do you know anything about ar6k power consumption? Many wifi drivers only start and stop the queue on dev.stop and dev.close. Doing that in ar6k makes the driver survive an ifdown/ifup, but it's likely using way too much current when down Jun 11 08:00:23 wpwrak: are you a good person to ask, btw? Jun 11 08:02:55 budfive: i measured current with the device enabled and disabled in general, and it's quite thirsty when up (even if not actively communicating) Jun 11 08:03:34 budfive: but if you have more fine-grained changes, you would have to measure the differences Jun 11 08:05:31 budfive: (person) i did most of the fiddling with the ar6000. just don't have much time for it at the moment ... (doing proper debugging of that beast generally requires setting up a test environment, making repeatable test runs, and so on. so that usually means 1-2 days without looking at anything else.) Jun 11 08:06:56 wpwrak: yeah, your tests and work with ar6000 was really cool. Too bad convincing atheros they have firmware bugs is not enough to gain anything :-/ But it was a nice try. :) Jun 11 08:07:24 wpwrak: ok, no worries. I don't have much time either... I think what I'll do is to muck with ar6000_init to see what it is exactly that triggers the "ready" message to be sent in the first place. If I find that, putting it into ar6000_open() may fix our issue Jun 11 08:08:52 PaulFertser: there are still a few tweaks we can make. but yes, eventually we'll run out of options :-( Jun 11 08:10:49 wpwrak: i wonder why not so many people are working on improving gta02 low-level stuff in general. More than 10k users and so little real kernel hackers. Jun 11 08:12:17 PaulFertser: i'm a little surprised as well. perhaps it's because everyone expected openmoko to take care of everything, so few felt a need to get involved. Jun 11 08:12:57 can someone give me a status of shr, i was offline for two weeks ... Jun 11 08:13:41 hi there Jun 11 08:13:45 BluesLee: it kicks ass Jun 11 08:13:53 i only read that there is a hw fix for #1024 which is the biggest news Jun 11 08:14:00 PaulFertser: why? Jun 11 08:14:17 BluesLee: constant improvements Jun 11 08:14:41 PaulFertser: also speed improvements? Jun 11 08:15:04 BluesLee: i don't think so Jun 11 08:15:14 PaulFertser: compared to other distros shr is less responsive Jun 11 08:15:52 BluesLee: i highly doubt that. Jun 11 08:16:01 PaulFertser: did the official release happened? Jun 11 08:16:06 me too Jun 11 08:16:09 BluesLee: debian is slow too. And all other distros are OE based as is SHR. Jun 11 08:16:32 BluesLee: only official press-release about gta02 given to the hands of community. Jun 11 08:16:33 PaulFertser: actually scrolling in debian is usable. Jun 11 08:16:47 KaZeR_mobi: is it distro-specific? Jun 11 08:16:52 starting apps and gui repsonsiveness seems to be faster with qtopia or whatever they call it now Jun 11 08:16:54 or maybe it was in h:1 Jun 11 08:17:18 BluesLee: qtopia use a framebuffer instead of X, makes a huge difference Jun 11 08:17:20 BluesLee: i don't care about q*. Jun 11 08:17:29 KaZeR_mobi: no, that's not the reason. Jun 11 08:17:30 PaulFertser: i don't know. scrolling is just better :) Jun 11 08:17:34 KaZeR_mobi: X is not slow! Jun 11 08:17:53 KaZeR_mobi: it's developers abusing X to do things it's not designed for are slow (read retarded). Jun 11 08:17:57 ;) Jun 11 08:18:07 BluesLee: i agree with PaulFertser : qtopia* or android can't be compared to the other distros Jun 11 08:18:17 for instance the shr settings apps Jun 11 08:18:39 it feels like waiting for christmas Jun 11 08:18:40 BluesLee: so, how does matter if you runn shr settings on SHR or on OM2009? Same OE. Jun 11 08:18:52 PaulFertser: in that case scrolling should be removed. i'd rather select my contacts by typing and having a filtered list rather than some kind of non-working and useless eye candy Jun 11 08:18:57 i never used om200x images Jun 11 08:19:26 BluesLee: probably shr settings are waiting for slow frameworkd reply, so it'll be the same on any distro using frameworkd. Jun 11 08:19:49 KaZeR_mobi: i don't know what scrolling are you talking about, i'm still using Zhone :-/ Jun 11 08:20:06 PaulFertser: no, scrillng with python-elementary actually seems slower than using C-based elementary Jun 11 08:20:12 scrolling... Jun 11 08:20:20 PaulFertser: even loading terminal takes a few seconds Jun 11 08:20:23 PaulFertser: turning on/off things in shr settings is not responsive enough Jun 11 08:20:33 KaZeR_mobi: correct Jun 11 08:20:38 PaulFertser: zhone under shr? Jun 11 08:20:39 frameworkd is going to speed up a lot when vala Jun 11 08:20:40 or a simple calculator Jun 11 08:20:44 turning on/off stuff is due to frameworkd/dbus slowness Jun 11 08:21:01 KaZeR_mobi: i'm debian user Jun 11 08:21:41 what's with a simple calculator blueslee? Jun 11 08:21:55 PaulFertser: ah, ok Jun 11 08:22:44 DocScrutinizer frameworkd is going to speed up a lot when vala : any eta? just curious. /me is always careful with promises (duke nukem forever will kick ass ;) ) Jun 11 08:24:27 aiui the migration will be step by step. Abyss is already there Jun 11 08:28:03 what's abyss? Jun 11 08:28:44 new implementation of the mux daemon I think, ask mickely Jun 11 08:28:47 mickeyl Jun 11 08:31:56 ok. mux daemon is the one reponsible for the gprs buffer getting full ? Jun 11 08:46:54 * mrmobil wonders if we should finally getabyss into our feed Jun 11 08:49:04 pro Jun 11 08:50:59 last time I tried I failed due to some vala probs Jun 11 08:51:12 time to retry Jun 11 08:53:03 if abyss is stable enough, getting it would be nice. this would allow (me :P) to use gprs without fearing of modem "hang" :) Jun 11 08:53:24 is a muxer needed at all if I don't use gprs Jun 11 08:53:26 ? Jun 11 08:54:20 rtp, we can have both installed and swutch via config Jun 11 08:55:52 mrmobil: hmm... that would be nice to have but as gsmmuxd is not really maintained afaik, I'm not sure it worth being able to switch Jun 11 08:59:38 rtp, being able to switch is already there in frameworkd Jun 11 08:59:46 rtp: abyss is stable Jun 11 09:00:15 spaetz: yes, muxer is needed anyway, it's good to separate unsolicited responses from commands. Jun 11 09:00:31 spaetz: some modems lack it but support for single-line modems is limeted. Jun 11 09:01:58 what modem? Jun 11 09:02:08 * mrmobil needs power now Jun 11 09:03:06 PaulFertser: great :) Jun 11 09:15:24 re Jun 11 09:23:03 ahh, ok Jun 11 09:24:19 mrmoku|a`: I suggest to not switch until we pushed a new -testing out of the door. I really, really want a stable milestone before the next architecture shift Jun 11 09:24:44 spaetz: i suggest to give user a choice! Jun 11 09:25:12 PaulFertser: nothing wrong with that, but given how few we are I wouldn't want to spent too much effort sto offer choice NOW Jun 11 09:25:26 I'd rather bring what we have in a shape that can be called stable Jun 11 09:25:35 spaetz: imho abyss muxer worths it. Jun 11 09:25:47 and focus on that. If we pushed a -testing out, then choice and progress is good Jun 11 09:25:52 mrmoku|a`: hmm, since dbus is upgrdaed, nytowl corrected the rules.yaml failure, frameworkd was pretty stable. Now it only remains a pretty nasty bug. Only the gsm part fails silently (the phone can be suspended/resumed), just no incoming/outgoing call. Everything looks normal Jun 11 09:26:10 PaulFertser: there is always something that is worth to update. I find it sad that there is still no SHR release Jun 11 09:26:16 mrmoku|a`: are you aware of bugreport like this? Jun 11 09:49:45 Does anyone have an idea if 22uF X5R 20% 6.3� 0805, GRM21BR60J226M is ok for a 1024 fix? I can't find ESR in the datasheet for that so i'm not sure. Sourcing right caps here is not as easy as i wish... Jun 11 09:52:18 ESR is very low in any ceramic cap Jun 11 09:52:30 you only have to worry about ESR in electrolytic caps Jun 11 09:52:48 This is ceramic Jun 11 09:52:56 cer may be sensitive to vibration, though. but that doens't seem to be an issue in this case Jun 11 09:52:57 wpwrak: so, no need to worry, it's ok? Jun 11 09:53:23 (though ti's spec strongly suggests they mean one should use tantalum) Jun 11 09:53:41 yeah, should be good. x5r is a decent material. Jun 11 09:54:18 Cool, then sourcing of that one is easy. Thanks again! Jun 11 09:55:32 Is there a page somewhere with clear instrucitons how to connect through ssh to your moko? For a reason it doesn't work anymore... Jun 11 09:55:40 (yes i did change tot eth2) Jun 11 09:56:21 pieterc: yes, usb networking wiki page Jun 11 09:56:57 well... those instructions i did follow, but when trying to ssh i get a time out Jun 11 09:57:25 I had the same issue with OM2009... now i flashed the latest SHR Jun 11 09:59:28 spaetz, I do not want to switch the default Jun 11 09:59:54 just offer abyss in the feed for people to try it ooout Jun 11 10:02:29 khiraly1, dos1 said something like that yesterday... though for him it was a crashed ophonekitd Jun 11 10:02:37 I guess you don't use it :P Jun 11 10:02:42 nope Jun 11 10:02:46 I dont use it Jun 11 10:03:02 khiraly1, have some frameworkd DEBUG logs? Jun 11 10:03:09 mrmoku: of course Jun 11 10:03:18 trac #? :) Jun 11 10:03:22 can anyone help me with my ssh problem? Jun 11 10:03:23 http://pastebin.com/df1a99fa Jun 11 10:03:51 pieterc, can you ping 192.168.0.202? Jun 11 10:03:57 I just read on bug #435 he was fighting with the same problem what nytowl already fixed two days ago. I also tell him here in irc. Im really sorry he missed it and spend time on it. (I forgot to update #435) Jun 11 10:03:58 nope Jun 11 10:04:24 sendmsg: operation not permitted Jun 11 10:04:28 mrmoku: I clean it up (remove my phone numbers and privacy) and put online. I was mainly interested if there is already a know bug id Jun 11 10:04:34 pieterc, if you open a terminal on the FR and do 'ifconfig usb0' what does it show? Jun 11 10:04:50 khiraly1, don't know... Jun 11 10:04:53 mrmoku: it'll be eth2 Jun 11 10:05:18 Zorkman: ip addr add 192.168.0.200/24 dev eth2 Jun 11 10:05:53 mrmoku: it gives me the right ip adress of the openmoko Jun 11 10:06:01 192.168.0.202 Jun 11 10:06:06