**** BEGIN LOGGING AT Thu Jan 03 02:59:59 2013 Jan 03 11:24:03 All the way back in the 90s, Qualcomm approached Apple to suggest that it might want to put a radio in its Newton PDA. It could have led to the first ever iPhone two decades early but instead, Apple told Qualcomm to stick its radio chip somewhere else. Jan 03 11:24:11 * RST38h laughs satanically Jan 03 11:27:10 Hi. Jan 03 11:27:15 Is there an Android VM for N900? Jan 03 11:27:20 I want to play Ingress Jan 03 11:28:00 not a VM, but you can try NITDroid Jan 03 11:28:27 I'll look at it, thanks! Jan 03 11:28:44 Is it in extras-devel or do I need to browse? Jan 03 11:30:28 not sure. I think there is an installer in one of the repos Jan 03 11:30:46 Ok Jan 03 12:10:20 Hi all, can anyone offer some advice what to look out for when buying a used N900 ? Jan 03 12:14:51 broken USB, or not connecting to cellular network seem to be main issues Jan 03 12:15:07 +1 Jan 03 12:15:37 also check that the sensors for the camera lens, the backcover and the keyboard slider work correctly Jan 03 12:15:40 do i need to worry about hardware revision or is there only one? Jan 03 12:15:53 there's plenty, but they're all good Jan 03 12:16:10 ok Jan 03 12:16:25 at least, none is specifically bad Jan 03 12:20:33 for hardware side look for usb,slide to unlock key,camera for scratches and also the battery Jan 03 12:22:37 it'll be from ebay but i can ask the seller. Jan 03 12:24:08 initially i need it primarilly for VoIP/SIP over wifi. but with the idea of ultimately running debian on a pocket device. Jan 03 12:24:55 afaict, VoIP/SIP should work right out of the box. Jan 03 13:08:49 grummund: you're going to be disappointed Jan 03 13:09:35 grummund: VoIP/SIP is notoriously bad on the N900. A lot of packet loss, poor codec implementation, with essentially no support in hardware. Jan 03 13:09:59 grummund: even skype is dodgy at best---if it even works at all ever since microsoft bought skype. Jan 03 13:17:22 teotwaki: that's sad. i thought it was going to be my dream phone Jan 03 13:19:21 grummund: N900 isn't a phone. If you expect it to be one, you will be disappointed. Jan 03 13:20:04 well yes i should say pocket computer with softphone Jan 03 13:21:33 grummund, you could always use a non-native SIP implementation Jan 03 13:24:19 it's just for WLAN use, i probably won't put a SIM in it Jan 03 13:29:03 anyway thanks all for the advice. i might give it a go and if it fails on SIP usability then i still have a cool device to play and hack on ;) Jan 03 13:35:09 works for me for so Jan 03 13:35:11 sip Jan 03 13:36:03 teotwaki: afaik skype still works Jan 03 13:36:38 and SIP is only notoriously bad if you try to use it with a wifi router that has poor powersaving Jan 03 13:37:34 or if you try to use it over wifi while you're using a bluetooth headset Jan 03 13:37:41 (because wifi+bluetooth at the same time is bad) Jan 03 13:38:00 uhm...yeah skype has worked ok for me too, but I'd just like to dismiss the comment that "N900 isn't a phone", since I use it as a phone just fine :p Jan 03 13:38:30 (although probably 99% of my use is not related to calling or receiving calls, but anyway) Jan 03 13:41:45 I've been quite happy with N900 as a phone too Jan 03 13:43:00 * grummund read of someone running asterisk on the N900 itself Jan 03 13:55:33 I have been using the N900 as my primary phone since I bought it Jan 03 13:55:40 and for that purpose (making phonecalls) it does great :) Jan 03 14:02:16 :-) Jan 03 14:17:50 apt-get install libqtm-12-* doesn't work. Anyone knows why? Jan 03 14:18:09 http://talk.maemo.org/showpost.php?p=1310197&postcount=122 Jan 03 14:25:13 Is there Voice Guided Navigation yet for N900? Jan 03 14:25:15 grummund: and running asterisk on the N900 is about as useful as growing dead kittens in the basement. Jan 03 14:25:33 grummund: it's technologically cool, but it really doesn't provide anything in the end. Jan 03 14:28:09 teotwaki: i don't know much about asterisk i just read that Jan 03 14:28:55 anyway i bought the N900 so no going back now :-) Jan 03 14:29:06 JOIN USSSSSSSS Jan 03 14:31:40 teotwaki, grummund: yeah, that's the question I wanted to ask too. What's the point of Asterisk on N900? Jan 03 14:32:56 sip gateway? Jan 03 15:07:55 grummund: welcome aboard, you'll be spending some time here from now on :) Jan 03 15:14:58 indeed :) Jan 03 15:25:36 btw, off topic, but just for pure awesomeness: http://www.youtube.com/watch?v=MsoRSI7ei4E Jan 03 15:43:02 meat Jan 03 15:43:04 neat Jan 03 15:43:59 Pork Jan 03 15:43:59 Dork Jan 03 15:47:07 anyone help me in setting up ubuntu lucid desktop VM in virtualbox Jan 03 16:11:48 asterisk on N900 is as silly as a load balancer on your DSL-modemrouter Jan 03 16:12:07 there's no point in it whatsoever Jan 03 16:14:01 welcome grummund in this small but nice club :-) Jan 03 16:15:22 btw I'm using built-in SIP via WRT54 AP all the time, sipgate account. Usually after the initial 2..3s to establish connection it works absolutely fine Jan 03 16:18:20 I adoped the general method of answering like "CHRSHDHSCRCK *cough* *cough* errrr YES?" which gives far end indication they should repeat their initial one or 2 seconds, or at least anticipate I didn't hear them Jan 03 16:19:06 Invariably works like shit here. Skype too. Jan 03 16:19:23 Skype is a bit better though, have ot give it a credit Jan 03 16:19:28 I have used a'loadbalancer on my DSL Jan 03 16:19:30 well Jan 03 16:19:34 ISDN Jan 03 16:20:35 skype on audio calls only is pretty nice; on video though audio transmitted is sometimes slow and the video res is pretty horrible... Jan 03 16:21:10 but then at least it has skype video calls; something taken out of the N9 :( Jan 03 16:21:47 SIP much depends on far end client. All my SIP calls are via sipgate gateway to landline/GSM, and it seems sipgate has pretty good b2bua on their gateways Jan 03 16:22:58 for a plain SIP call to peer with arbitrary crappy SIP client mileage may vary massively Jan 03 16:23:52 my only use for sip is to my isps sip server Jan 03 16:24:04 to a real phone number Jan 03 16:29:48 same here, and it seems that's what usually works flawlessly, more or less Jan 03 16:30:16 except sipgate.de is not my ISP Jan 03 16:50:10 DocScrutinizer05: I've run a few tests with stock (pr1.3) kernel as discussed and couldn't reproduce the sd-card issues experienced on pk51/thumb. do we have a driver issue after all? Jan 03 16:50:54 seems likely Jan 03 16:51:48 wirr: :( Jan 03 16:51:57 i mean, it's pretty good for you Jan 03 16:52:02 we'd need to bisect it now, to find what's causing the problem in pk51/thumb combo Jan 03 16:53:44 kerio: not especially... i'm very interested in having high quality community-driven maemo systems. i really don't care about stock. Jan 03 16:53:49 if it turns out some valid bugfix in kp51 made a sleeping bug in your card raise its head, we might need a kernel module load option or a sysfs switch to select between two conflicting mutally exclusive paradigms regarding operation of that driver Jan 03 16:54:25 so - what can i contribute? i propose running the same tests on kp51 first on cssu-testing then on cssu-thumb Jan 03 16:54:49 yes, exactly Jan 03 16:55:03 afk, bbl. sorry Jan 03 16:55:10 okay... Jan 03 16:55:18 wirr: also having the full logs would be good Jan 03 16:55:36 kerio: which ones? dmesg output? Jan 03 16:56:06 yep, and syslog too Jan 03 16:57:15 hmm... so I should set up a syslog daemon and log to a remote system? which one do you propose? Jan 03 16:57:36 why a remote system? Jan 03 16:57:49 just install sysklogd Jan 03 16:58:08 out of habbit... i usually don't write logs locally on embedded systems... Jan 03 16:58:33 you don't necessarily have to keep the daemon running all the time Jan 03 16:58:37 ok Jan 03 16:59:08 btw, i forgot to mention: the transmission errors occur as well if I try to start the newest nemo kernel Jan 03 16:59:35 especially because the syslogd in the repos doesn't rotate the logs except when it starts Jan 03 16:59:48 so eventually it fills up the rootfs, as DocScrutinizer can tell you Jan 03 17:01:21 sound like hell alot of uptime :-) Jan 03 17:01:22 DocScrutinizer05: i was also going to go with sipgate (in UK). Jan 03 17:01:49 wirr: it was a sad day for IroN900, yes :( Jan 03 17:01:56 lost ~200 days of uptime iirc Jan 03 17:02:02 ouch Jan 03 17:16:13 kerio: so the test with stock userland and kp51 already fails... Jan 03 17:16:28 well, that's good Jan 03 17:16:31 makes things easier :) Jan 03 17:16:34 wirr: try KP52 though Jan 03 17:17:15 wirr: http://talk.maemo.org/showpost.php?p=1305042&postcount=288 Jan 03 17:17:39 it'll probably not change anything, but it's still better to stay on the latest version Jan 03 17:17:52 http://pastebin.com/bW6bvGt7 Jan 03 17:18:48 wirr: what's in syslog regarding that? Jan 03 17:23:24 just a sec Jan 03 17:28:46 http://pastebin.com/FMVcntyd Jan 03 17:29:37 kerio: i've just noticed: while getting an i/o error on writing to mmcblk1, the logs show errors for mmcblk0 Jan 03 17:30:06 ugh, that might be because the uSD is actually mmcblk0 Jan 03 17:30:13 they are swapped by udev Jan 03 17:30:50 sorry, don't understand that... Jan 03 17:31:14 at boot time, the uSD is mmcblk0 and the emmc is mmcblk1 Jan 03 17:31:23 if you have a uSD inserted Jan 03 17:31:59 ah... so the kernel logging still sees it this way, while userspace addresses it the other way round? Jan 03 17:32:06 i don't know, maybe Jan 03 17:32:19 are you having trouble accessing your home or MyDocs? Jan 03 17:32:25 nope Jan 03 17:34:01 kerio: what about alignment? has data to be aligned on 4k blocks? Jan 03 17:34:36 meh Jan 03 17:34:51 maybe something like aligning to 4M would help? Jan 03 17:34:58 it would be a performance thing, though Jan 03 17:35:22 it can't cause errors, microsds are supposed to be ideal block devices Jan 03 17:36:27 hmm... it just seems to me the error occurs much more often when testing using this cmd: 'dd if=/dev/zero of=/dev/mmcblk1' as when using 'dd if=/dev/zero of=/dev/mmcblk1 bs=4096' Jan 03 17:36:48 wirr: hm Jan 03 17:37:15 echo 8192 > /sys/block/mmcblk1/queue/nr_requests Jan 03 17:37:19 echo 0 > /sys/block/mmcblk1/queue/iosched/slice_idle Jan 03 17:37:20 as root Jan 03 17:37:26 and then try again with the standard block size Jan 03 17:39:42 didn't you say before, the addresses mmcblk0/1 are swapped? Jan 03 17:40:17 good point, i don't know which would be the right one Jan 03 17:40:20 do it to mmcblk0 too Jan 03 17:40:22 ok Jan 03 17:40:31 (it doesn't hurt, anyway - it helps, in fact) Jan 03 17:41:18 test is running... Jan 03 17:43:22 writing the full 32gig will obviously take a while - so far no errors Jan 03 17:52:30 wirr: no error yet? Jan 03 18:27:27 kerio: still writing - no error yet Jan 03 18:35:08 is extras-cauldron-builds down or is it just me missing something? Jan 03 18:35:28 last emails in the archive are from december last year Jan 03 18:42:04 best media player for n900 Jan 03 18:42:05 GO ! Jan 03 18:47:44 hooray, internet connection once more fecked on 2G, while 3G works Jan 03 18:48:20 never heard of that mediaplayer app Jan 03 18:48:20 and no, I'm sure it's not the carrier's fault Jan 03 18:50:04 I honestly should do some forensiocs this time, since this is a FSCKNG ANNOYING bug Jan 03 18:52:13 seems kinda reproducable by going to no-signal area for half an hour while connected via 2.5G Jan 03 18:52:43 so it's somewhat in line with similar behavior on WLAN Jan 03 18:55:33 MSameer: december last year is like three days ago dude Jan 03 18:55:58 f3bruary: openmediaplayer, probably Jan 03 18:56:06 it's basically a drop-in replacement for the stock media player Jan 03 18:56:09 as if icd2 or whatever refuses to start a new association sequence while aborted 'dangling' timedout re-association for same connection still lingering on somewhere in stack Jan 03 18:56:13 cool, thanks Jan 03 18:56:37 DocScrutinizer05: disconnect, kill sscd, kill icd2, start everything up again? Jan 03 18:56:50 kerio: true but I uploaded a package and the archive page didn't get updated so I was wondering Jan 03 18:57:21 kerio: I was expecting an email with the result of the build to be added to the page Jan 03 18:57:53 kerio: wilco when @home Jan 03 18:59:21 so it might be an issue with the builder Jan 03 19:00:28 last I heard was autobuilder has issues with web frontend Jan 03 19:00:35 iirc Jan 03 19:02:00 ~ping Jan 03 19:02:02 ~pong Jan 03 19:02:17 bwahahaha 2G Jan 03 19:02:35 DocScrutinizer51: so how do i check the results ? :) Jan 03 19:03:05 sorry nfc Jan 03 19:03:44 bbl Jan 03 19:05:16 kerio: do you know of a audio file size filter ? Jan 03 19:06:10 I got a folder with sms tones (1-2 sec) and a folder with audio books (+-1hr) and dont want them included in the media library Jan 03 19:06:28 I looked around in the settings but didnt find it Jan 03 19:06:34 which suks Jan 03 19:06:37 sucks even Jan 03 19:16:43 f3bruary: check trackerd config Jan 03 19:17:22 f3bruary: I heard in CSSU it even works like supposed to, to exclude dirs from trackerd indexing Jan 03 19:17:50 stock trackerd has a bug in this function to exclude dirs from indexing Jan 03 19:34:25 thanks Jan 03 19:46:08 wirr: did the bug *always* occur when doing that, before? Jan 03 19:49:03 What is the newest firmware for maemo? Jan 03 19:49:42 lnxmen: 21.2011.38-1Smaemo5 Jan 03 19:50:08 latest stable Jan 03 19:50:40 kerio: under power51/cssu-thumb: yes; under pk51/stock I just ran the tests i've told you about... Jan 03 19:56:52 and the bug didn't occur? Jan 03 19:57:02 try restoring your cssu-thumb backup Jan 03 19:57:06 and do it on that, to be sure Jan 03 19:57:10 it shouldn't change anything Jan 03 19:57:39 my guess is that the I/O scheduler on the n900 is so borked that eventually even the kernel stops giving a shit Jan 03 19:57:51 yes, i'll do that as soon as the full 32gig sequential write test on kp51/stock is finished... Jan 03 19:59:03 ...running since 138 min... Jan 03 19:59:12 kerio: 21.2011.38-1Smaemo5 is not exactly a firmware, if anything it's a SSU upgrade revision to the latest "firmware" Jan 03 19:59:27 what's a "firmware" on the n900? Jan 03 19:59:29 seriously Jan 03 19:59:40 there's nothing firm about the ubifs rootfs Jan 03 19:59:44 pretty much an undefined term Jan 03 19:59:58 combo of kernel/rootfs and emmc filesystem? Jan 03 20:00:03 if anything, then I'd call fiasco images "firmware" Jan 03 20:00:04 hell, on the first boot it copies and deletes a huge chunk of data Jan 03 20:00:26 wirr: a firmware can't really be the maemo rootfs Jan 03 20:00:33 it's not firm! Jan 03 20:00:42 technically you're correct Jan 03 20:01:16 in common sense however people refer to images as firmware Jan 03 20:01:44 "disk" images Jan 03 20:01:45 clearly that means that we should start working on a fiasco image for CSSU stable Jan 03 20:02:10 kerio: Can I find it on Nokia site? Jan 03 20:02:15 I always supported that idea Jan 03 20:02:26 lnxmen: no, you have to flash PR1.3 and then do an upgrade Jan 03 20:02:32 okay Jan 03 20:02:40 thankyou Jan 03 20:02:42 kerio: see? ^^^ :-) Jan 03 20:02:52 DocScrutinizer05: >:c Jan 03 20:02:57 lnxmen: latest global image on nokia is X-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin Jan 03 20:03:11 s/X/RX/ Jan 03 20:03:11 wirr meant: lnxmen: latest global image on nokia is RX-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin Jan 03 20:03:18 DocScrutinizer05: how do we do it, though? it's not trivial by any means Jan 03 20:03:27 wirr: are you sure? Jan 03 20:03:39 pretty... Jan 03 20:03:41 well, maybe it's correct Jan 03 20:03:50 ~phone-control Jan 03 20:03:54 ~phonecontrol Jan 03 20:03:55 i heard phonecontrol is http://wiki.maemo.org/Phone_control Jan 03 20:03:58 I'm downloading latest global Jan 03 20:03:58 2011.38 is already SSU-PR1.3.1 Jan 03 20:04:03 but speed is really slow Jan 03 20:04:17 ~35 kb/s Jan 03 20:04:34 Doc, yep Jan 03 20:04:52 DocScrutinizer05: he said 36 Jan 03 20:05:01 yes Jan 03 20:05:05 oh you mean that it's too low Jan 03 20:05:06 I know to read Jan 03 20:05:08 well, nokia is nokia Jan 03 20:05:33 RX-51_2009SE_20.2010.36-2_PR_COMBINED_MR0_ARM.bin PR 1.3 version 20.2010.36-2 Jan 03 20:05:49 lnxmen: currently downloading at 1.3MB/s Jan 03 20:06:21 DocScrutinizer05: is "OS 2009" the PR 1.0? Jan 03 20:06:32 OS 2009 is fremantle Jan 03 20:06:32 not sure Jan 03 20:06:40 oh, there's two of those Jan 03 20:06:48 one is marked as "second" Jan 03 20:08:24 kerio: i'm starting to doubt that this write test will ever finish - the time it took so far amounts to a write speed of approx 3.5MB/s Jan 03 20:08:32 it's all pretty precisely explained on Jan 03 20:08:34 but on the positive side: no errors so far ;-) Jan 03 20:08:34 ~cssu Jan 03 20:08:35 i guess cssu is http://wiki.maemo.org/Community_SSU Jan 03 20:08:45 wirr: 3.5MB/s is about right, yes Jan 03 20:09:00 but that's only because i told you the magic invocation rituals to get a decent I/O Jan 03 20:09:11 * wirr bows before kerio Jan 03 20:09:41 so this is just what the n900's card reader/bus can deliver? Jan 03 20:09:49 ~tell lnxmen about cssu Jan 03 20:10:03 Pali: wirr has been having I/O problems on a uSD that have been "solved", maybe, with a couple of tweaks on the I/O scheduler - could you think about setting those as default? Jan 03 20:10:17 ask ShadowJK for the details, they're his Jan 03 20:11:13 kerio, I do know what linux sheduler doing and changing default configuration because XYZ wrote "it is cool" is not very good... Jan 03 20:11:44 I do not know sideeffects... Jan 03 20:12:17 kerio: I agree we should do something for WD kicking, but very carefully ;) Jan 03 20:12:37 do sth about WD kicking?? WTF?! Jan 03 20:12:40 it's not WD kicking, he was getting IO errors Jan 03 20:13:14 Pali, kerio: Anyway I'll try the tweaks for a week or two on pk51/cssu-thumb Jan 03 20:13:25 if kernel gets so busy so it can't even kick WD, then we have other problems to solve than change the WD timeout Jan 03 20:13:47 the WD wasn't even considered here! Jan 03 20:14:59 kerio: could swappolube set the parameters accordingly? Jan 03 20:15:04 yes Jan 03 20:15:09 well, i'm not sure about slice_idle Jan 03 20:15:12 then what'S the problem? Jan 03 20:15:24 ...wirr was getting *errors* Jan 03 20:15:29 yes, sure Jan 03 20:16:10 it's a singulary report though, so far Jan 03 20:16:32 IO-related lockups and general slugginess is far from being unheard of Jan 03 20:16:41 kerio: I am getting errors from my ADATA when it is overheated ;) Jan 03 20:16:51 get a real uSD Jan 03 20:16:55 like, a sandisk one Jan 03 20:17:07 kerio: are you playing smart? Jan 03 20:17:20 maybe a little :3 Jan 03 20:17:30 if those new parameters would evidently improve perfomance for everybody, get thorough tests for side effects, and we also see more of those reports about errors, we might consider making them a kernel or CSSU default Jan 03 20:17:35 the point is that faulty card could lead to io errors Jan 03 20:17:46 DocScrutinizer05: :nod: Jan 03 20:18:02 card works fine on a card reader, fails on two separate N900s, seems to work with those tweaks Jan 03 20:18:11 but we should find a way to test them Jan 03 20:18:25 kerio: i'll label that card faulty then Jan 03 20:19:47 or at least "incompatible" Jan 03 20:20:07 kerio: we're not going to taint kernel to make defect hardware work, as long as that defect hardware is not shipped by the dozens of millions to the general public Jan 03 20:20:12 and even then... Jan 03 20:21:13 basically we assume somebody used his head to find optimum values for IO parameters, for a set of most common usecases Jan 03 20:21:30 that's a *reeeeeeeeeally* bold assumption Jan 03 20:21:44 if we find that's actually not the case, we have to do proper evaluation and verification of our new set of parameters Jan 03 20:22:16 DocScrutinizer05: that is not the case, n900 is using the default kernel scheduler values for HDD Jan 03 20:22:18 that's why i asked Pali if he'd /consider/ that Jan 03 20:22:30 fixing a fringe case for one defect uSD is not a good reason to force new IO parameters on everybody Jan 03 20:22:39 DocScrutinizer05: though I strongly agree we should be very careful with those Jan 03 20:23:10 can somebody check if ELF binary is thub compiled? Jan 03 20:23:18 *thumb Jan 03 20:23:21 Pali: readelf? Jan 03 20:23:43 Pali: there's not any *decent* test Jan 03 20:23:58 i.e. readelf -a $YOUR_COOL_BIN Jan 03 20:24:04 I posted a shell oneliner in #maemo-ssu quite a long time ago Jan 03 20:24:17 DocScrutinizer05: there is, readelf Jan 03 20:24:29 I need to know if this diablo nokia closed lib is thumb or not: http://maemo.org/packages/view/libicd-network-ipv6/ Jan 03 20:24:29 and what will that tell you? Jan 03 20:24:39 freemangordon: well, it doesn't really help with self-generated code Jan 03 20:24:49 ARM_THUMB_ISA Jan 03 20:25:06 it didn't for modest when it been fekked Jan 03 20:25:49 $ readelf -a /usr/lib/icd2/libicd_network_ipv6.so | grep thumb Jan 03 20:25:49 the only way been to disassemble binary and look for jump addr pointing to impair addr Jan 03 20:25:52 no output Jan 03 20:26:08 grep -i --> also no output Jan 03 20:26:15 Pali: heh, i was about to say that Jan 03 20:26:32 Pali: grep THUMB Jan 03 20:26:40 it is in uppercase Jan 03 20:26:45 -i --> case insensitive Jan 03 20:26:50 however, i'll check it for you Jan 03 20:26:52 the only way been to disassemble binary and look for jump addr pointing to impair addr Jan 03 20:27:01 libicd-network-ipv6_0.15-0.1_armel.deb Jan 03 20:27:06 $ readelf -a /usr/lib/icd2/libicd_network_ipv6.so | grep thumb Jan 03 20:27:11 no -i Jan 03 20:27:22 [2013-01-03 21:26:08] grep -i --> also no output Jan 03 20:27:23 [21:26:08] grep -i --> also no output Jan 03 20:27:40 the only way been to disassemble binary and look for jump addr pointing to impair addr Jan 03 20:27:51 hm, the scrutinizer is stuck again Jan 03 20:28:00 * kerio kicks DocScrutinizer05 Jan 03 20:28:08 DocScrutinizer05, do you know if diablo was thumb compiled? Jan 03 20:28:11 that should've fixed it Jan 03 20:28:27 I thought it got stuck on the way from your rhetina to your brain (you got a brain, do you?) Jan 03 20:28:29 Pali: wait, it is not readelf Jan 03 20:28:34 sorry, my bad Jan 03 20:29:08 DocScrutinizer05: there's no need to be rude :< Jan 03 20:29:20 just kidding Jan 03 20:30:17 hm, autobuilder not working... Jan 03 20:30:44 I uploaded wide-dhcpv6 20080615-11.1maemo1 via dput and nothing Jan 03 20:31:26 Pali: well, it is readelf, but firle has no attributes Jan 03 20:31:31 *the fiel Jan 03 20:31:34 file Jan 03 20:31:58 Pali: what was this build with? Jan 03 20:32:08 nokia binary Jan 03 20:32:27 nobody know Jan 03 20:32:37 readelf -A libicd_network_ipv6.so Jan 03 20:32:39 Attribute Section: aeabi Jan 03 20:32:41 File Attributes Jan 03 20:32:48 and then nothing Jan 03 20:32:55 pali / wirr: I remember wirr's read errors from yesterday.. Anyone care to summarize what you've done and current statsus? :) Jan 03 20:33:49 (trying to fish a summary before I scroll bavk and start replying yo things 24h ago) Jan 03 20:34:26 [2013-01-03 21:31:20] http://mg.pov.lt/maemo-ssu-irclog/%23maemo-ssu.2011-11-29.log.html#t2011-11-29T00:04:00 Jan 03 20:34:28 [2013-01-03 21:33:35] ^^^ that's not yet the ultimate check, which more was with a regex checking for [13579bdf] as last digit of an addr Jan 03 20:34:39 Pali: launching IDA Jan 03 20:34:57 ShadowJK: flashed stock pr1.3; seq read/write tests successful Jan 03 20:35:11 eg no errors Jan 03 20:35:17 Pali: do we know the target CPU at least? Jan 03 20:35:18 Pali: I don't know if diablo been thombified, but I honestly doubt it Jan 03 20:35:41 ShadowJK: upgraded to pk51, I/O errors after a few hundred megs of writing Jan 03 20:35:45 freemangordon, something for N8x0 Jan 03 20:35:51 I really have no idea Jan 03 20:36:05 ~wtf eg Jan 03 20:36:07 EG: evil grin Jan 03 20:36:08 I copied it and it worked on n900 with n900's version of icd2 Jan 03 20:36:19 ShadowJK: then applied some IO scheduler tweaks from kerio, which apparently originally came from you Jan 03 20:36:30 but somebody on TMO thread wrote that his icd2 crashed with illegal instruction Jan 03 20:36:30 ShadowJK: queue size and slice_idle Jan 03 20:36:37 since then 32 gig seq writing test still running Jan 03 20:37:27 ~180 min, which amounts to somewhat < 3MB/s Jan 03 20:37:57 but no I/O errors so far Jan 03 20:38:00 Pali: that's pretty much a good indication that it actually been a thumb instruction Jan 03 20:39:11 wirr: what been your original write thruput before it threw IO errors? Jan 03 20:39:45 Doc: unfortunately didn't write it down, I remember sth >4MB/s Jan 03 20:39:45 Pali: yes, it is thumb Jan 03 20:40:13 freemangordon, so for that we need thumb kernel... Jan 03 20:40:14 wirr/kerio: 4M alignment seems common now. I've also seen 8M for a 64g sd Jan 03 20:40:24 so higher? Jan 03 20:41:39 * DocScrutinizer05 wonders how freemangordon finally found out about thombiness of that binary Jan 03 20:41:45 IDA Jan 03 20:41:56 doing what exactly? Jan 03 20:42:13 what? disassembling ;) Jan 03 20:42:19 and then? Jan 03 20:42:30 freemangordon, so what to do? Adding some thumb dependency to package? Jan 03 20:42:32 or not? Jan 03 20:42:45 Pali: for sure Jan 03 20:43:01 what is good solution for pushing thumb compiled application into extras-devel? Jan 03 20:43:10 kernel-feature-.... Jan 03 20:43:12 none Jan 03 20:43:13 Hm, if dd without bs fails, it's likely that the card does a full read-modify-write cycle for every 512 bytes, so 4M read/write each time or even worse, might slow it down so that it blows through timeouts? Jan 03 20:44:05 DocScrutinizer05: and after that looking at the disassembly - instructons size, address being odd, etc Jan 03 20:44:30 >>...address being odd,...<< :-D Jan 03 20:44:53 thankyou :-) Jan 03 20:45:01 ShadowJK: i'm not able to judge that Jan 03 20:45:16 but maybe I can run additional tests? Jan 03 20:45:19 [2013-01-03 21:27:39] the only way been to disassemble binary and look for jump addr pointing to impair addr Jan 03 20:45:24 DocScrutinizer05: but actually I prefer to look at instructions size Jan 03 20:45:32 freemangordon, can you check if "kernel-feature-errata-430973-workaround" is good dependency? Jan 03 20:46:17 Pali: hmm, wait Jan 03 20:47:21 well, usually I use "kernel-feature-errata-430973-workaround[armel]" Jan 03 20:47:36 I have Architecture: armel Jan 03 20:47:39 but having in mind you don't have the real source package Jan 03 20:47:42 yep Jan 03 20:48:03 I adding dependency to metapackage ipv6-support Jan 03 20:48:08 Pali: should work Jan 03 20:48:11 ok Jan 03 20:49:14 I've seen this interesting thing on recent sandisk (and adata), where the card seems atleast for a moment able to keep track of lots of halfwritten blocks.. I wonder if certain IO patterns cause this prepared-work or deferred-work to collapse, forcing the card to do some massive gigabyte-worth of housekeeping tasks all of the sudden, resulting in timeouts Jan 03 20:49:40 freemangordon, but problem is that autobuilder not building package... Jan 03 20:50:03 that it works on stock and not pk makes little sense to me though Jan 03 20:50:31 about 30 minutes ago I uploaded package and still nothing... Jan 03 20:50:42 ShadowJK: that's been my very initial assumption Jan 03 20:50:46 is there any log from autobuilder?ň Jan 03 20:50:53 DocScrutinizer05 ^^^ Jan 03 20:51:07 wirr; btw, get iostat (sysstat in tools repo, iirc), "iostat -x -k -d mmcblk0 mmcblk1 10", the bigger the avgrq-sz, the better 512k and up is especially good Jan 03 20:51:15 Pali: the extras cauldron mailing list ? Jan 03 20:51:28 M4rtinK, can you give me link? Jan 03 20:51:33 Pali: somebody else asked same thing ~2h ago. I still have no idea about autobuilder Jan 03 20:52:47 I found this: https://garage.maemo.org/pipermail/extras-cauldron-builds/ Jan 03 20:52:54 [2013-01-03 19:56:50] kerio: true but I uploaded a package and the archive page didn't get updated so I was wondering Jan 03 20:53:17 [2013-01-03 19:59:21] so it might be an issue with the builder Jan 03 20:53:18 [2013-01-03 20:00:28] last I heard was autobuilder has issues with web frontend Jan 03 20:53:33 so far nothing from autobuilder :( Jan 03 20:53:46 that archive is out of date: Ending: Sun Dec 16 19:26:07 EET 2012 Jan 03 20:53:54 true Jan 03 20:54:12 but last package was builded: 2012-12-27 10:42 Jan 03 20:54:31 isn't that roundabout the date when maemo.org infra been set into r7o mode? Jan 03 20:55:04 (12-16) Jan 03 20:56:00 since 12-27 Nemein might actually be busy migrating autobuilder, I don't know Jan 03 20:56:48 I honestly would appreciate they would update us more frequently about status of migration Jan 03 20:57:09 and in more detail Jan 03 20:59:32 I want to update ipv6 support... Jan 03 21:00:18 finally icd working with ipv6 only networks (radvd or dhcpv6) Jan 03 21:12:07 Pali: why don't ask jonwil to RE that lib :P Jan 03 21:12:29 freemangordon, because it worked on my device :D Jan 03 21:12:49 today I reread tmo posts and found some sigill Jan 03 21:12:58 haha Jan 03 21:13:00 I see Jan 03 21:13:09 freemangordon and now ipv6 support working fine Jan 03 21:13:16 but you need kernel with ipv6 support Jan 03 21:13:23 so kernel-power Jan 03 21:13:31 and last kernel-power has also thumb patch Jan 03 21:13:43 so I do not see any problems... Jan 03 21:13:49 Pali: we need to restart the work on kernel in CSSU Jan 03 21:14:06 freemangordon, I changed some patches in kp52 Jan 03 21:14:11 and I want to release it Jan 03 21:14:12 merlin1991: has setup mediawiki on his server, we can use that Jan 03 21:14:21 Pali: ok Jan 03 21:14:28 after that I merge kernel-power to cssu git kernel repo Jan 03 21:14:34 ok Jan 03 21:14:45 and then we can start discussion about patches for new kernel in cssu Jan 03 21:15:04 freemangordon, btw I can start ikiwiki if needed Jan 03 21:15:27 (good wiki sw which store pages in git :D) Jan 03 21:15:37 Pali: it does not matter for me, as long as we are making progress Jan 03 21:15:47 so you can use it offline :-) Jan 03 21:16:13 but first I need autobuilder! Jan 03 21:16:14 Pali: windoze Xp here, lets stick to HTML :) Jan 03 21:16:34 Pali: afaik dput should work Jan 03 21:16:36 without it I cannot push kernel-power to extras-devel Jan 03 21:16:58 freemangordon 1 hour ago I pushed wide-dhcpv6 package Jan 03 21:17:01 nothing happened Jan 03 21:17:50 I pushed version 2008 Jan 03 21:17:56 but there is only old: http://repository.maemo.org/extras-devel/pool/fremantle/free/w/wide-dhcpv6/ Jan 03 21:18:15 also nothing here: http://maemo.org/packages/ Jan 03 21:19:07 and nothhing here: http://maemo.org/packages/view/wide-dhcpv6-client/ Jan 03 21:19:26 Pali: https://garage.maemo.org/extras-assistant/index.php Jan 03 21:19:55 aaa here is: https://garage.maemo.org/extras-assistant/index.php :-) Jan 03 21:20:00 see the top in the queue :P Jan 03 21:20:03 wide-dhcpv6_20080615-11.1maemo1 2013-01-03T19:51:57Z Jan 03 21:20:21 but why is not buidling? Jan 03 21:20:35 NFC Jan 03 21:20:51 or is that list of waiting packages? Jan 03 21:21:11 then something happened: 2013-01-01T00:03:29Z Jan 03 21:21:17 new year? Jan 03 21:21:35 nokia turned off buidling machine? Jan 03 21:21:43 DocScrutinizer05 ^^^ Jan 03 21:22:34 ~wtf NFC Jan 03 21:22:36 NFC: no fucking clue Jan 03 21:24:19 The stock hsmmc data timeout code looks bizarre :/ Jan 03 21:26:14 Pali: https://garage.maemo.org/builder/fremantle/?C=M;O=D Jan 03 21:27:01 missing also :-( Jan 03 21:27:09 wide-dhcpv6 version 2008 Jan 03 21:27:25 now going offline, bye Jan 03 21:27:50 any rought timeline on the autobuilder btw ? :) Jan 03 21:38:40 M4rtinK: what particularly? (not that I knew of *any* timeline for autobuilder) Jan 03 21:44:56 DocScrutinizer05: provided they don't fix it themselves, I'd guess it is dependent on HIF getting access to the machine running it Jan 03 21:45:28 DocScrutinizer05: when would should that be, cca ? Jan 03 21:45:59 sorry, can't parse your last 2 posts Jan 03 21:46:43 https://garage.maemo.org has a r/o notice Jan 03 21:47:08 I'm more puzzled that autobuilder still doing *anything* Jan 03 21:47:11 I mean when the Hildon Foundation gets access to the machines running the Maemo infrastructure Jan 03 21:47:49 that's unclear, and ATM HiFo dropped the whole migration on council's feet basically Jan 03 21:49:11 HiFo is in negotiations with Nokia about legalese stuff. Once those papers are signed, hiFo will become owner of the domains/URLs Jan 03 21:50:05 well, its at least online :) Jan 03 21:50:19 and TMO is working Jan 03 21:50:29 Nemein is busy moving infra from current servers to some interim vservers (thus the r/o notice), and once Nemein installed the RX300 irons in colo hoster's racks and set up the OS on them, they move the vservers to those irons Jan 03 21:50:39 but IMHO the Wiki & autobuilder outage is harming the community quite a lot Jan 03 21:50:51 sure Jan 03 21:51:45 I hope we end with at least some semi-interested people with root access to those machines Jan 03 21:51:49 TMO will work until 2013-01-20 the least Jan 03 21:52:04 ShadowJK: probably you want me to run iostat while writing to the sd card? Jan 03 21:52:41 so that bit rot & related issues can be handled in some timely manner Jan 03 21:52:48 it was more of a FYI than a request for information :) Jan 03 21:53:08 it's completely unclear when/if Nemein will make those vservers the ones the URLs are pointing at Jan 03 21:53:24 :-) of course... Jan 03 21:53:43 it's also unclear when the current servers are expected to go offline Jan 03 21:53:50 its quite frustrating if something quite crucial stops working and there is no-on to contact :) Jan 03 21:54:04 oh, so its happening in parallel - good to know Jan 03 21:54:04 yes Jan 03 21:55:16 M4rtinK: the idea is to rsync/cp the current servers' content to the new machines, and eventually switch the URL DNS pointers Jan 03 21:55:34 BTW, any info about the COBS (build.pub.meego.com) infrastructure ? Jan 03 21:55:47 obviously until that switch you need to freeze old machines or they will render new machines obsolete Jan 03 21:55:54 last I heard it should run at least until about March/February Jan 03 21:56:23 sorry, I'm not up-to-date regarding the meego stuff Jan 03 21:56:39 so autobuilder is probably also offline Jan 03 21:56:41 np :) Jan 03 21:56:57 Do N900 applications also all steal all contact data like just about all Android applications do? Jan 03 21:57:12 BTW, what about the Fremantle OBS that was put on ice ? :) Jan 03 21:57:14 (autobuilder) as mentioned above I'm puzzled that it's obviously not (completely) offline Jan 03 21:57:20 resp r/o Jan 03 21:57:37 well, it kinda still worked recently Jan 03 21:57:53 just the Web UI was borked since November Jan 03 21:57:58 ~snowboard Jan 03 21:58:05 yes, and that's basically a flaw Jan 03 21:58:07 ~n900snowboard Jan 03 21:58:20 but scping source packages worked for some Jan 03 21:58:47 (not for me - public key error :D ) Jan 03 21:58:48 Does everyone have me on ignore? Jan 03 21:59:11 It shouldn't be a hard question to answer for anyone who knows anything about the device. Jan 03 21:59:13 fasta: no, and no Jan 03 21:59:13 ShadowJK: http://pastebin.com/1KwG0DRw seems not so bad... Jan 03 21:59:28 DocScrutinizer05: thank you Jan 03 21:59:31 yw Jan 03 21:59:44 but unfortunately now i got i/o errors again, this time even using the scheduling tweaks on kp51 Jan 03 21:59:53 I don't understand why the mass buys spyware on a massive scale then. Jan 03 22:00:08 Only because it has a faster processor and a more fancy screen? Jan 03 22:00:46 wirr; weird in emmc to usd case mmcblk1 shows both reads and writes Jan 03 22:00:51 swap? Jan 03 22:01:05 The only good place for an Android device seems to be /dev/null. Jan 03 22:01:52 Speaking of which: someone said I could test harddisk performance by doing dd if=/dev/zero of=/dev/null o.O Jan 03 22:02:05 What has the world come to? Jan 03 22:02:23 wirr; error -110 in dmesg? Jan 03 22:02:23 *cough* Jan 03 22:02:33 ShadowJK: yep Jan 03 22:03:28 Are the Lumia phones spyphones like the Androids and the i-devices? Jan 03 22:03:41 ShadowJK: there is a swap partition on mmcblk1 (default nemo image) but it is not used/active in maemo. i definitely cannot explain the reads on mmcblk1 Jan 03 22:03:53 That would seem a very Microsoft way of doing things; doing things exactly the same. Jan 03 22:03:55 http://pastebin.com/tQ3WGNaj Jan 03 22:04:04 [ 628.561950] HWRecoveryResetSGX: SGX Hardware Recovery triggered Jan 03 22:04:16 lnxmen: power-cycle your device Jan 03 22:04:43 this is the long-standing bug in SGX firmware Jan 03 22:04:50 s/the/a/ Jan 03 22:04:51 freemangordon meant: this is a long-standing bug in SGX firmware Jan 03 22:04:59 Does the N900 come with cgroups preconfigured? o.O Jan 03 22:05:09 fasta: yes Jan 03 22:05:13 Did Nokia invent them? Jan 03 22:05:25 At least, I thought they weren't _that_ old. Jan 03 22:05:52 NokiaTechTeam++ Jan 03 22:07:28 Merged in 2007. Jan 03 22:10:40 ShadowJK: I think the device descriptors in iostat are swapped (so mmcblk0=sd, mmcblk1=emmc). then the write effectively are page swaps Jan 03 22:10:51 freemangordon: thanks, it works :) Jan 03 22:10:53 right Jan 03 22:11:10 lnxmen: np Jan 03 22:14:13 so far I can say the following in regards to the sandisk card: Jan 03 22:14:19 - works fine in stock pr1.3 Jan 03 22:14:42 - throws a lot fi/o errors in kp51 with default Jan 03 22:15:18 - behaves way better in kp51 using ShadowJKs tweaks but still throws i/o errors from time to time Jan 03 22:15:45 ShadowJK: any advice on how to find optimal alignment/scheduler settings for this card? Jan 03 22:17:06 well the scheduler stuff just tries to make data "easier" for the card to process, it's still possible that the vard might get data patterns it times out on Jan 03 22:17:47 so far couldn't find _any_ documentation on sandisk website regarding recommended settings Jan 03 22:18:18 heh, docs? what docs? Jan 03 22:18:45 ;-) Jan 03 22:18:56 these are highly confidential and trade secret information, yada yada yada Jan 03 22:19:25 * ShadowJK suspects everyone is just ignoring it now Jan 03 22:19:58 ... Jan 03 22:20:28 ignoring timeout stuff Jan 03 22:21:15 hmm... doesn't seem to be a good option if this results in I/O -110 transmission errors :-( Jan 03 22:21:45 could there be any relation to the older -110 problem with palis u-boot bootstrapping the nemo kernel? Jan 03 22:22:17 But basically kp sets timeout to maximum possible, and stock kernel sets it shorter, maybe Jan 03 22:22:23 DocScrutinizer05: it's Jan 03 22:22:25 ~snowboarding Jan 03 22:22:25 methinks snowboarding is a sport that kills your n900 Jan 03 22:22:34 So kp should give the card the longest possible time to do stuff Jan 03 22:22:47 oh no, not you, someone else who was talking at the same time, sorry for the ping Jan 03 22:24:28 Why it works on stock and not pk is mysterious :) Jan 03 22:34:59 ShadowJK: MAXINT==MININT-- Jan 03 22:35:06 ? Jan 03 22:37:35 DocScrutinizer05: not necessarily Jan 03 22:38:57 ~coffee Jan 03 22:38:58 extra, extra, read all about it, coffee is the reason the net exists, the drug of choice for a GNU generation, http://www.chez.com/emarsden/downloads/coffee.el, /usr/share/doc/HOWTO/en-html/mini/Coffee.html, geiseri's favorite beverage Jan 03 22:39:06 kerio: once more, verbose: setting timeout to MAX might trigger a quite common bug in the card driver Jan 03 22:39:49 Doc: is there an easy way for me to verify without to compile a custom kernel (don't have the sdk)? Jan 03 22:39:51 usually called one-off-bug Jan 03 22:39:52 iirc Jan 03 22:40:11 wirr: sorry, I never looked into that driver Jan 03 22:40:59 a modinfo will tell a bit about allowed module parameters, if any Jan 03 22:41:22 I gather that module is no module though, it's probably monolithic Jan 03 22:41:40 aka built-in to kernel Jan 03 22:42:16 if it were a module, you could try unloading it and force-loading the stock kernel's driver module instead Jan 03 22:43:09 mhm... I'll check that out Jan 03 22:43:24 wirr: otoh I'm rather sure pali and other devels here will be happy to compile test kernels for you Jan 03 22:43:56 Doc: cool, that sounds generous :-) Jan 03 22:44:22 at least pali should be interested in finding out complete info about a possible flaw/bug in his powerkernel Jan 03 22:44:55 freemangordon might be willing to help as well Jan 03 22:45:45 wirr: btw doc: is not highlighting here (otherwise I'd get ~500 flase positives per day) Jan 03 22:46:50 xchat should allow tab-completion of nicknames Jan 03 22:47:30 DocScrutinizer05: thanks Jan 03 22:47:36 np :-) Jan 03 22:49:43 DocScrutinizer05: while I'm restoring my crypto-backup could you tell me a little something about what's to be expected from maemo's future? Jan 03 22:51:12 I mean considering the current transition of maemo.org can we expect a boost in activity or is cssu-development rather slowing down due the lack of new hardware? Jan 03 22:54:06 wirr: the transition shouldn't really change things Jan 03 22:54:22 at best we'll get a bit more control over the repos Jan 03 22:54:34 so things should work a bit better, i suppose Jan 03 22:55:28 kerio: so far that sounds goods Jan 03 22:55:49 yep :D Jan 03 22:56:11 the lack of new hardware is annoying, yes Jan 03 22:57:51 whats the stopper in porting maemo to other hardware? i mean the n9 seems ideal considering its omap Jan 03 22:58:13 the closed blobs Jan 03 22:58:19 and all the closed stuff Jan 03 22:59:02 arrgh... it's always the same problem :-( Jan 03 22:59:21 also, maemo kinda sucks Jan 03 22:59:28 something like cordia HD would be better Jan 03 22:59:32 why's that? Jan 03 22:59:35 or, hell, just debian with hildon desktop Jan 03 22:59:43 oh yeah! Jan 03 22:59:50 * wirr agrees strongly Jan 03 23:00:21 ~closed Jan 03 23:00:21 is it already possible to run hildon ontop of debian? Jan 03 23:00:22 from memory, closed is http://wiki.maemo.org/Why_the_closed_packages or https://wiki.maemo.org/Fremantle_closed_packages Jan 03 23:00:33 messybox, upstart, the fucking FUBAR partitioning, the billion ways to run daemons... Jan 03 23:00:41 wirr: well, it's open source Jan 03 23:00:56 and the cordiahd guys patched it to support resolutions other than 800x480 or 480x800 Jan 03 23:00:58 kerio: ok, I rephrase: is it already possible for non-developer? Jan 03 23:01:21 it *should* be just a x11 window manager Jan 03 23:02:05 kerio: the other sucking parts could get fixed easily in a spinoff, since any maemo on N9 wouldn't be fremantle anyway Jan 03 23:02:41 my point is that we don't really want maemo, we want a linux distro with hildon-desktop Jan 03 23:02:58 well, no Jan 03 23:03:08 I also want ICD for example Jan 03 23:03:10 i wonder why the openpandora guys didn't go with hildon-desktop too Jan 03 23:03:15 the status bar Jan 03 23:03:18 the dialer Jan 03 23:03:38 ICD is kind of a piece of crap though Jan 03 23:03:51 and can't manage more than one connection at a time, at least with the current UI Jan 03 23:03:54 *GUI Jan 03 23:03:55 there's a lot more than HD that defines awesomness of fremantle Jan 03 23:04:10 the status bar is part of hildon Jan 03 23:04:21 not just hildon-desktop though, that's true Jan 03 23:04:23 not of HD though Jan 03 23:04:31 why the dialer? Jan 03 23:04:43 because it works Jan 03 23:04:46 it makes phone calls, what's so good about it? Jan 03 23:05:01 the address book system is really good, that's true Jan 03 23:05:02 that it makes phonecalls Jan 03 23:05:18 meh, phone calls are overrated :P Jan 03 23:05:27 besides, couldn't we grab the openmoko stuff for that? Jan 03 23:05:43 *can't Jan 03 23:05:58 DocScrutinizer05: it's 4 bits in a 32 bit register. 15 (1111) is reserved, 14 is longest timeout, and pk hard codes that, while the original code gets a timeout requested from higher level mmc code Jan 03 23:06:19 And then sets timeout to closest available of the 15 Jan 03 23:07:24 Another patch I found online sets it to twice the requested timeout (but still limited to 14 max, which is max the omap hw supports) Jan 03 23:13:45 ShadowJK: so what do you suggest for my case? Jan 03 23:14:16 I have no idea :( Jan 03 23:44:58 ShadowJK: SiErr? Jan 03 23:45:26 . Jan 03 23:45:38 ? Jan 03 23:45:58 14 defective in silicon Jan 03 23:46:57 * DocScrutinizer05 also rises eyebrows on "15: reserved" Jan 03 23:48:14 I'd build a PK module that hardcodes 7 instead 14 Jan 03 23:48:29 another one with 13 Jan 03 23:49:27 hell, if the 'patch' in PK that makes this adata card work is that it uses an hardcoded 14 instead a value it gets from card... Jan 03 23:50:50 could we enable some debug mode in that driver, to see what wirr's 'defective' card requests for that timeout value? Jan 03 23:51:16 raw value, not after trimming Jan 03 23:52:23 maybe the card's controller depends on that timeout? to save some internal timers for whatever Jan 03 23:53:13 after all flash-erasure of a page is some timed process Jan 03 23:53:14 DocScrutinizer05, ShadowJK may this be related: http://e2e.ti.com/support/dsp/omap_applications_processors/f/42/t/29521.aspx Jan 03 23:54:54 That's a circuit board design thing... if Nokia did it wrong we'd see issues on many many devices and cards Jan 03 23:55:14 wirr: excellent find Jan 04 00:00:10 DocScrutinizer05, ShadowJK: I assume you're aware of the TI OMAP errata document of 2010? There's something bout "MMC: Multiple Block Read Operation Issue" Jan 04 00:00:22 Nope Jan 04 00:00:33 nope Jan 04 00:01:47 http://www.ti.com/product/omap3530 Jan 04 00:01:48 anyway N18 doesn't show up as a pin name on OMAP3530 of N900 Jan 04 00:01:55 DocScrutinizer05, ok Jan 04 00:02:23 seems to indicate it's actually NC Jan 04 00:03:21 * DocScrutinizer05 bookmarks this gem anyway, it's an excellent example of the daily WTF you're facing when building hw Jan 04 00:03:59 DocScrutinizer05, you're an electronics/hw guy? Jan 04 00:04:25 ~joerg Jan 04 00:04:26 somebody said joerg was a HW-developer and engineer of Openmoko, usually known as DocScrutinizer Jan 04 00:06:29 ~docscrutinizer Jan 04 00:06:29 rumour has it, docscrutinizer is jOERG, a HW-developer and engineer of Openmoko Jan 04 00:06:47 I see. :-) Jan 04 00:07:05 I'd actully say senior EE Jan 04 00:07:42 so why don't you build a new device for us ;-) Jan 04 00:07:53 does this ring a bell? "The maximum block size supported by the host controller is 1024 bytes. This value is hard-coded in the Jan 04 00:07:53 register MMCi.MMCHS_CAPA[17:16] MBL field and cannot be changed. Jan 04 00:07:53 " Jan 04 00:08:18 because it costs "a significan slice of a million", as SpeedEvil likes to put it Jan 04 00:08:44 unsurprisingly... Jan 04 00:09:28 (ring a bell) well, it for sure should show up in driver somehow Jan 04 00:09:59 something that's hardcoded in hw needs also to get hardcoded to the driver Jan 04 00:10:05 hmmm... so that's not related to /sys/block/mmcblk1/queue/nr_requests? Jan 04 00:10:35 err nope, that's number of entries in IO scheduler queue afaik Jan 04 00:10:52 IOW number of pending IO Jan 04 00:11:02 ok... Jan 04 00:11:11 should be unrelated to any blocksize Jan 04 00:11:38 * DocScrutinizer05 waves and heads out for a nice beer Jan 04 00:11:47 next pub Jan 04 00:12:30 DocScrutinizer05, bye then and cheers Jan 04 00:12:37 wirr: (EE) I'm available btw, if you got some interesting project and the money to pay me :-) Jan 04 00:13:15 DocScrutinizer05, I'm in network security, but who knows... Jan 04 00:14:07 :-) Jan 04 00:17:12 wirr: btw I contributed a bit to gta04 Jan 04 00:17:36 gta04.org Jan 04 00:18:22 that's what i thought when you said openmoko... Jan 04 00:19:51 gta04 basically unrelated to OM Jan 04 00:20:00 didn't know that Jan 04 00:21:20 it's an awesome incredible "one man project" courtesy Nikolaus Schaller, and recycles a few components of openmoko phones Jan 04 00:21:58 I never felt like the latter was an extremely brilliant idea Jan 04 00:22:16 could as well have used screen and case of N900 Jan 04 00:22:44 would have been a more practical form factor... Jan 04 00:22:52 or arbitrary other case and LCD Jan 04 00:23:46 but if he's still on it this might happen for a next release? Jan 04 00:24:07 i'm not involved in that project anymore Jan 04 00:24:14 I see Jan 04 00:24:19 o/ **** ENDING LOGGING AT Fri Jan 04 02:59:58 2013