**** BEGIN LOGGING AT Sun Jan 03 02:59:58 2016 Jan 03 07:15:37 xes, readonly idea would be nice, is there simple way to setup readonly copy of the talk/wiki pages that use the same db (but with ro access only) then redirect/offer a link to offenders Jan 03 07:16:27 xes, it could even be done via some simple script that would disallow all POSTs and allow only GETs Jan 03 08:08:49 hi Jan 03 08:14:19 jonwil: hi Jan 03 08:14:41 jonwil: can you document somewhere process or state of pulseaudio-nokia? Jan 03 08:14:51 I would like to know what is already done Jan 03 08:15:02 what is in progress and what needs to be done Jan 03 08:15:40 e.g. which modules can be reuses from meego/harmattan and which needs to be RE Jan 03 08:16:44 We cant reuse anything from meego/harmattan Jan 03 08:16:57 really? Jan 03 08:17:03 yeah Jan 03 08:17:13 it is different or what is reason? Jan 03 08:17:20 Too different Jan 03 08:17:40 I see that e.g. music module is copied from meego Jan 03 08:17:57 yeah we are reusing FOSS parts from Meego Jan 03 08:18:05 but we cant reuse any Meego or Harmattan blobs Jan 03 08:18:14 and the FOSS bits have been modified heavily to match Maemo Jan 03 08:24:05 In terms of whats still todo, there are //todo markers covering which functions still need to be cloned (plus there are functions yet to be cloned that are called by ones marked //todo) Jan 03 08:24:23 I have done all the cloning of pulseaudio-nokia that I can do Jan 03 08:30:13 why the interest in pulseaudio-nokia anyway? Jan 03 08:48:44 I just like to know current state Jan 03 08:49:11 freemangordon wrote that some parts after RE are now faster Jan 03 08:49:28 so maybe they could be good stuff for CSSU Jan 03 08:51:31 What we have isn't good enough for CSSU yet Jan 03 08:57:59 Hm. Played around 7 hourse, 28 minutes and 42 seconds of music today, 33% battery remaining. Jan 03 08:58:10 That seems like my normal daily drain. Jan 03 08:58:34 Maybe I normally have stuff running that keep the device busy somehow. Jan 03 08:59:09 most likely Jan 03 08:59:13 powertop etc Jan 03 08:59:46 * Maxdamantus doesn't understand how powertop works. Jan 03 08:59:56 * Maxdamantus will probably just strace everything. Jan 03 09:00:14 or maybe make a special program that ptraces everything in a controlled fashion. Jan 03 09:00:19 maxd, you basically look at the sources of wakeups Jan 03 09:00:40 How low-level are the wakeup sources? Jan 03 09:00:49 by device/driver Jan 03 09:00:57 (interrupts) Jan 03 09:01:39 So what does it say when there's a particular task that's just doing a loop over some sleep system call. Jan 03 09:01:42 s/.$/?/ Jan 03 09:01:54 task as in LWP Jan 03 09:02:49 maxd, write such program (should be a quickie), run and see the powertop output before/after Jan 03 09:05:32 Mk, will do soon. Jan 03 09:06:19 i've found out when writing oscp that keeping audio device open (in pause, not writing to it) keeps pulseaudio churning Jan 03 09:06:42 That could explain it. Jan 03 09:06:45 I use mpd. Jan 03 09:09:58 fix was simple, but ugly Jan 03 09:10:14 Why ugly? Jan 03 09:10:57 because i dont like it, and because it's PA fault Jan 03 09:12:34 they probably do that because of timing issues Jan 03 09:12:59 i bet dmix in alsa doesnt do that Jan 03 09:14:39 My pulseaudio process doesn't seem to be doing anything while idle. Jan 03 09:15:06 unless you're talking about the library part (which would be part of the music player process) Jan 03 09:15:32 maxd: i was talking about 'pause', when stopped apps usually release the device Jan 03 09:16:51 Yeah, so I'm guessing mpd is releasing the device, whatever that means. Jan 03 09:17:39 since all I see is a bunch of threads blocked in some SYS_336 call. Jan 03 09:18:18 Apparently that's ppoll Jan 03 09:18:29 silly maemo version of strace. Jan 03 09:22:36 do you know how to import C header file (with function prototypes and structs) into IDA? Jan 03 09:24:30 Is it a header file from the Maemo SDK? Jan 03 09:24:37 Or a header file from somewhere else? Jan 03 09:35:15 LoadHeaderFile ? Jan 03 09:35:57 https://www.hex-rays.com/products/ida/support/idadoc/1367.shtml Jan 03 09:36:59 also: http://reverseengineering.stackexchange.com/questions/6651/importing-c-header-file-with-memory-location-definitions Jan 03 10:07:33 Pali: File->Load File->Parse additional header file Jan 03 10:07:48 keep in mind you might have to set the include paths Jan 03 10:08:19 Pali: in Options->Compiler Jan 03 10:08:44 so the parser to be able to find the #includes in the file it is parsing Jan 03 10:15:18 parse C header file did not imported structs... Jan 03 10:15:57 Pali: check in views->open subviews->local types Jan 03 10:16:23 a ok, they are Jan 03 10:16:30 also, maybe it is good idea to load tlb that jonwil created Jan 03 10:16:50 thats if you are working with Maemo ARM binaries Jan 03 10:17:00 if you are working with x86 or something that isn't Maemo, you dont need it Jan 03 10:17:14 yeah, sure Jan 03 10:19:35 Pali: sorry about pestering, but could you upgrade kernel to -rc7 Jan 03 10:19:43 to see if net oops still exists Jan 03 10:21:20 ok will at it Jan 03 10:24:13 Pali: also, I'll look into that mmc slot name, to see how much of a work it is Jan 03 10:24:24 to implement it in mmc core that is Jan 03 10:24:39 ok Jan 03 10:26:35 are you talking about the fact that mmcblk0 and mmcblk1 are swapped on mainline? Jan 03 10:26:49 (swapped compared to standard maemo kernel) Jan 03 10:26:58 or am I confused again :) Jan 03 10:34:56 Wizzup: in lagacy boot they have the so-called "slot names" Jan 03 10:35:00 *legacy Jan 03 10:35:24 "external" and "internal" Jan 03 10:36:02 ah Jan 03 10:41:32 the slot swapping is very silly :\ Jan 03 10:42:00 yeah Jan 03 10:50:15 how about probing all mmc devices during boot for some signature? Jan 03 10:50:24 that way it wouldnt matter where/what Jan 03 10:58:46 freemangordon: kernel rebased Jan 03 10:58:50 now I'm testing compilation Jan 03 10:58:56 and will push changes to github Jan 03 10:59:21 freemangordon: can you look at that omapfb bug? infinite loop of division by zero? Jan 03 10:59:44 it is reproducable in qemu Jan 03 10:59:52 and also on real n900 Jan 03 11:03:36 hit by: echo 0 > /sys/devices/platform/omapdss/overlay0/enabled && echo 0 > /sys/class/graphics/fb0/size Jan 03 11:05:55 what is echo 0 > /sys/class/graphics/fb0/size supposed to do? Jan 03 11:06:13 I mean, setting size to 0 ... how is that supposed to behave? Jan 03 11:06:41 bencoh, maybe disabling the fb? Jan 03 11:06:55 maybe auto detect something? Jan 03 11:07:17 freemangordon: code pushed to github Jan 03 11:07:38 bencoh: 0 is special value for omapfb to reset Jan 03 11:07:49 ah Jan 03 11:09:56 nonstandard interfaces ahoy! Jan 03 12:45:30 Pali: ok, will do Jan 03 12:55:27 Pali: "ARM: OMAPFB: panel-sony-acx565akm: fix missing mutex unlocks" is still in the tree Jan 03 12:55:41 forgot to drop :-( Jan 03 14:13:22 Wizzup: ping Jan 03 14:15:26 freemangordon: pong Jan 03 14:15:48 Wizzup: http://sprunge.us/jBCN?c, line 52 Jan 03 14:16:00 are you sure this is the correct way to match? Jan 03 14:16:26 ev_keys[p][q] gives us a keycode Jan 03 14:17:44 Let me look Jan 03 14:18:40 freemangordon: Are you sure that is called keycode? Those constants are defined in linux/input.h and iirc are the right way to compare to/with Jan 03 14:19:16 not sure how's that called Jan 03 14:19:29 however , could you look at http://pastebin.com/GhY0KDLL Jan 03 14:19:33 I mean, a stupid answer would be: "it works", but now I want to double check what I did Jan 03 14:19:53 if you tested it, then it is me that screwed something Jan 03 14:20:06 look at the pastebin Jan 03 14:20:46 that one gives me error on mce: match_event_file_by_caps: ioctl(EVIOCGBIT) failed on `/dev/input/eventN', where N = 0,1,2,4,5,6 Jan 03 14:21:21 On maemo? Jan 03 14:21:22 for event3 (which is our gpio switch) it doesn;t give error, but does not find a match either Jan 03 14:21:24 yes Jan 03 14:21:36 linux 4.4-rc7 Jan 03 14:21:56 but evtest works just fine there Jan 03 14:22:26 oh, no, event3 is accell :) Jan 03 14:23:19 does it give that error in the loop or outside of the loop? (when getting events or when getting keys) Jan 03 14:23:48 freemangordon: btw, I think line 41 is incorrect in your paste Jan 03 14:23:50 no idea, have to tweak the code, gimme a minute Jan 03 14:23:56 freemangordon: continue should really be return -1; Jan 03 14:24:11 if one of the request event types is not export, it cannot be the device you're looking fo Jan 03 14:24:15 r Jan 03 14:24:16 is not exported* Jan 03 14:24:28 that is what my code did on line 41 (same line, heh) Jan 03 14:24:30 no, I want to match at least one of the request types :) Jan 03 14:24:42 the same for the keycodes Jan 03 14:24:59 Ah. I thought you wanted to match all. Jan 03 14:25:10 no, match at least one Jan 03 14:25:23 in the meanwhile, this is what evtest gives me http://pastebin.com/Z1NAvLVX Jan 03 14:25:27 fun of third party code writing Jan 03 14:25:38 and specs/reqs not written Jan 03 14:25:44 yeah :) Jan 03 14:26:12 Let me add error checks to my original ioctl code and see what happens Jan 03 14:26:31 Wizzup: wait a second, building Jan 03 14:27:19 Wizzup: mce: match_event_file_by_caps: ioctl(EVIOCGBIT, EV_MAX) failed on `/dev/input/event6' Jan 03 14:27:21 Are you sure all the ioctl's use the return value for error indication? Jan 03 14:27:24 perror gives me 'success' Jan 03 14:27:32 match_device: can't get events: Success Jan 03 14:27:42 not all ioctl's return nonzero on failure Jan 03 14:27:47 according to man ioctl, it should return 0 on success Jan 03 14:27:54 not always Jan 03 14:28:02 Usually, on success zero is returned. A few ioctl() requests use the Jan 03 14:28:02 return value as an output parameter and return a nonnegative value on suc‐ Jan 03 14:28:06 cess. On error, -1 is returned, and errno is set appropriately. Jan 03 14:28:09 Let me check - because errno is not set here Jan 03 14:28:13 ah, I see Jan 03 14:28:15 Which makes me assume that there is in fact, 'no error' Jan 03 14:28:17 but let me double check. Jan 03 14:28:20 I should check for -1 Jan 03 14:28:53 yeah, my bad, let me fix that and test again Jan 03 14:29:54 Yeah, it both line 21 and 45 return -1 on error Jan 03 14:29:59 I think both* Jan 03 14:30:08 http://sprunge.us/hgJP?c Jan 03 14:30:31 oh - indent Jan 03 14:30:39 http://sprunge.us/ejSV?c Jan 03 14:30:39 yep, it was that one Jan 03 14:30:46 sorry for the noise Jan 03 14:30:55 np :) Jan 03 14:31:25 Wizzup: http://pastebin.com/45SgLeRE Jan 03 14:31:58 that result in "mce: Match found on `/dev/input/event6'" Jan 03 14:32:35 That is what you wanted, right? (I don't know what event6 is) Jan 03 14:32:39 Ah, I see the evtest paste Jan 03 14:32:40 yes Jan 03 14:32:46 So that's good, then :) Jan 03 14:32:57 yep, thanks Jan 03 14:33:18 p.s. maybe you can somehow use errno in the mce_og Jan 03 14:33:36 mce_og? Jan 03 14:33:45 mce_log Jan 03 14:33:59 hang on - phone Jan 03 14:34:10 in theory yes, but I don't really think it makes sense Jan 03 14:34:25 I don;t see errno used in the other parts of the code Jan 03 14:35:52 I just meant that if ioctl's fail, it is sometimes useful to know why Jan 03 14:35:54 but this is good enough :) Jan 03 14:36:20 yes, got the idea Jan 03 14:36:21 also ioctl's should not return anything other than -1 on error, but I guess < 0 is fine too (according to the man page) Jan 03 14:36:29 :nod: Jan 03 14:47:29 freemangordon: wrt the test_bit, it turns the 'key code' into a proper index in the bit array and checks if that bit is set Jan 03 14:47:34 so I think it is correct Jan 03 14:48:08 at least I saw no false matches - but I may be wrong Jan 03 14:48:47 ok, makes sense Jan 03 16:50:18 so Jan 03 16:50:22 someone knows something about qt? Jan 03 16:50:27 if i compile qt4 against new libssl Jan 03 16:50:31 will it support new protocols Jan 03 16:51:03 or it will simply use new libssl but only supporting those on http://doc.qt.io/qt-4.8/qssl.html#SslProtocol-enum? Jan 03 16:52:10 and on the other hand... are there lots of changes between upstream qt and maemo's? Jan 03 16:52:59 http://deriv.debian.net/Maemo/patches/q/qt4-x11/ Jan 03 16:53:02 should these be all? Jan 03 16:53:39 although those are not very useful Jan 03 16:53:48 http://deriv.debian.net/Maemo/patches/q/qt4-x11/qt4-x11_4:4.8.2-1_qt4-x11_4.7.0~git20100909-0maemo1+0m5.patch content is just 'This patch was too large to be useful so it was removed.' Jan 03 16:54:01 i'd think that an empty patch is less useful than a very large one... Jan 03 19:49:34 When will voting start on camera competition? Jan is here, already... And, people are looking for vouchers on the marketplace :-) Jan 03 21:19:55 hi Jan 03 21:21:44 jonwil: hi Jan 03 21:25:26 jonwil: Wizzup: please review the last 5 commits here https://github.com/community-ssu/mce/commits/master if you have time and will :) Jan 03 21:26:54 All of this stuff is meant to still work on stock kernel as well as upstream? Jan 03 21:27:26 yes, it is supposed to still work Jan 03 21:27:37 didn;t test it though Jan 03 21:27:43 ok Jan 03 21:27:46 * freemangordon reboots in KP Jan 03 21:29:18 I have to learn a bit more about what mce does exactly, I'll probably clone the repo later - but the commits looks sane Jan 03 21:29:48 https://github.com/community-ssu/mce/commit/c0648e5a567682fd9302c3225011264af1341ec4 line 382 has an underindented } though I think? Jan 03 21:29:55 first file Jan 03 21:31:00 looks correct to me there Jan 03 21:31:17 hm Jan 03 21:32:32 Just installed the new MCE on my device :) Jan 03 21:32:32 jonwil: isn't it part of the switch Jan 03 21:51:10 Wizzup: github shos it like that Jan 03 21:51:14 *shows Jan 03 21:51:25 jonwil: no damages on KP Jan 03 21:51:42 unfortunately on 4.4 udev hogs the cpu Jan 03 21:51:45 working just fine on stock kernel here Jan 03 21:52:35 I guess udev tries to do something with /dev/input/event_gpio_keys Jan 03 21:55:52 you want me guys to review something? Jan 03 21:56:06 or to test, if you guide me through and confirm that it won't need flashing Jan 03 21:56:09 i got data in there Jan 03 21:56:23 i'm searching for another n900 just for development Jan 03 21:56:29 ceene: feel free to review latest patches to mce, if you feel like so Jan 03 22:04:53 https://github.com/community-ssu/mce/commit/09e11e180126c5a2cffe1d05a37c6c3ed6409f43#diff-119974261001e1b4fa526ad05946cbe6R598 Jan 03 22:05:03 if close() return value is not going to be checked Jan 03 22:05:12 does it make much sense the test for fd!=-1? Jan 03 22:06:05 and also, wouldn't fd be anyway checked when open is called? Jan 03 22:06:45 not important anyway Jan 03 22:07:11 ceene: I am not sure it is a good idea to call close(-1) Jan 03 22:07:35 i guess -1 would be returned by close Jan 03 22:07:41 with errno set to EBADF Jan 03 22:07:58 also, see who and how calls register_io_monitor_chunk Jan 03 22:08:04 re open() Jan 03 22:08:56 on that same patch it's only called is the previous call to open wasn't -1 Jan 03 22:09:01 so it should be safe anyhow Jan 03 22:10:18 22:29 < Wizzup> https://github.com/community-ssu/mce/commit/c0648e5a567682fd9302c3225011264af1341ec4 line 382 has an underindented } though I think? Jan 03 22:10:24 i also see that on github Jan 03 22:11:40 ceene: correct, but it is on line 381 :) Jan 03 22:11:44 oh Jan 03 22:11:55 well, that's all i could find :) Jan 03 22:11:57 gotta go now Jan 03 22:11:58 cya! Jan 03 22:12:02 thanks Jan 03 22:16:02 jonwil: please pull mce, as I just fixed a nasty bug I introduced earlier :) Jan 03 22:16:11 ok Jan 03 22:18:30 rebuilding MCE now :) Jan 04 00:16:16 * DocScrutinizer05 spanks freemangordon with a trout Jan 04 00:18:04 you went fishing? Jan 04 00:25:43 So let's pretend I'm a total newbie to maemo devel, but not to programming/devel in general Jan 04 00:25:55 Where do I start so I can deploy/develop on/for a spare n900 that I have Jan 04 00:26:06 Wizzup: SB Jan 04 00:26:26 Right Jan 04 00:26:45 Do you have a Linux system? Jan 04 00:26:55 If not, the Maemo Dev VM is the easiest way to get started Jan 04 00:27:14 even if you have a linux system the dev vm is better Jan 04 00:27:32 since a lot of automagic sb does doesn't work so well anymore with new kernels :/ Jan 04 00:27:54 jonwil: I many linux systems Jan 04 00:27:56 well, depends on how much RAM you can spend on it. Jan 04 00:28:01 I have an arm laptop in fact Jan 04 00:28:10 (not sure if that makes things easier or harder - I also have intel laptop(s)) Jan 04 00:28:25 not sure there is SB for ARM Jan 04 00:29:00 I thought that there was something with qemu on intel that emulates arm to get a SB, but I ... am not into this yet Jan 04 00:29:11 so basically: set up SB on an intel machine, in a vm Jan 04 00:29:15 (the dev vm) Jan 04 00:29:26 no Jan 04 00:29:29 you grab the vm Jan 04 00:29:29 s/set up SB/set up a/the dev VM/ Jan 04 00:29:33 and it has evertyhing set up Jan 04 00:29:37 Yeah, that's what I mean. Jan 04 00:29:40 Cool. Jan 04 00:29:46 yeah you grab the dev VM, run it n vmware and its good to go Jan 04 00:30:00 (optional step convert to vbox or qemu :)) Jan 04 00:31:36 I will do that tomorrow then. Time to hit the sack now. Thanks for the quick responses Jan 04 02:49:41 ~sb Jan 04 02:49:42 i heard scratchbox is a cross-compiling system that uses binfmt_misc, rpc calls, and an nfs mount to make a cross-build appear to be 100% native, and is found at http://www.scratchbox.org/, hosted by maemo now. Also at http://maemo.merlin1991.at/files/SB **** ENDING LOGGING AT Mon Jan 04 02:59:58 2016