IRC log of #maemo-ssu for Monday, 2012-10-01

*** Guest57704 has quit IRC00:09
*** Guest57704 has joined #maemo-ssu00:09
*** int_ua has joined #maemo-ssu00:22
SpacedOutDocScrutinizer05: I missed getting in on the USB charging discussion.  I've tracked down a problem where plugging into a wall charger works, but plugging into a computer usb port doesn't charge and the computer doesn't see the N900.  musb_core.c  struct musb_ctx holds hardware register values, it can get it into that state.00:25
SpacedOut`echo mem > /sys/power/state` is the most reliable way to get it into that state. If anyone is wanting to debug it I have a patch for compiling the USB drivers as modules which also fixes some concurrency problems in the charger detection routines, a whole bunch of debugging print messages.00:29
*** ekze has quit IRC00:36
*** NIN101 has quit IRC00:43
*** M13 is now known as Guest4036700:48
*** arcean_ has joined #maemo-ssu01:06
*** arcean has quit IRC01:09
DocScrutinizer51SpacedOut: we concluded during 18months of h-e-n development that musb-hdrc has to be monolithic.01:09
SpacedOutI'm fine with it being compiled in the kernel with the default configuration, but I have it working as module, and that is very handy if anyone is trying to do development or debugging on it.01:11
SpacedOutWhich I was.01:12
DocScrutinizer51and setting device to suspend-to-ram is kinda useless since it either wakes up again after 10s due to signal-level change message from cmt, or you block that and it won't even wake up on inbound call01:12
SpacedOutI was wanting to point out that some of the problems you were having with it detecting the usb port and charging sounded familiar with what I was debugging, depending on the specifics and it all gets pretty fuzzy trying to track down all the variables.01:13
DocScrutinizer51or you have very secrit wisdom how to tell bb5 cmt to stop spamming unsol msgs01:13
SpacedOutI think suspend to ram is useless unless you are in offline mode, and that is exactly how I use it, not very often, but I want it near off, but not have to boot it up when I'm ready to use it again.01:14
DocScrutinizer51yep, this works. speedevil done this01:15
DocScrutinizer51sorry for terse answers01:15
DocScrutinizer51N90001:15
SpacedOutI didn't get the reference or what part works, the current suspend to ram works for me after my patches for one to let it work with a computer USB port after resuming.01:18
*** arcean_ has quit IRC01:27
int_uamaemo.org down02:30
int_uaisup.me/maemo.org02:30
int_uaI've just installed CSSU-thumb and U-Boot disappeared, is that ok?02:51
merlin1991it's to be expected02:52
merlin1991cssu-thumb flashes a kernel02:52
merlin1991but you can install the kerne-thumb (or how it's called) bootimg and install uboot again02:52
int_uaok :)03:04
*** int_ua has quit IRC03:05
*** M4rtinK has quit IRC04:08
*** M4rtinK has joined #maemo-ssu04:09
*** M4rtinK has quit IRC04:17
*** dafox has quit IRC04:44
*** Guest57704 has quit IRC04:48
*** Guest57704 has joined #maemo-ssu04:49
*** amiconn has quit IRC05:14
*** amiconn_ has joined #maemo-ssu05:14
*** amiconn_ is now known as amiconn05:14
*** Guest40367 has quit IRC05:59
*** DocScrutinizer05 has quit IRC06:03
*** DocScrutinizer05 has joined #maemo-ssu06:03
*** nox- has quit IRC06:58
*** luf has joined #maemo-ssu09:16
*** fredrinLap has quit IRC09:20
*** fredrinLap has joined #maemo-ssu09:23
*** andre__ has joined #maemo-ssu09:51
*** andre__ has joined #maemo-ssu09:51
*** BCMM has joined #maemo-ssu11:15
*** NIN101 has joined #maemo-ssu11:33
*** dhbiker has joined #maemo-ssu11:42
*** arcean has joined #maemo-ssu12:00
*** CodenameStrike has joined #maemo-ssu12:18
*** dhbiker has quit IRC12:23
*** Guest57704 has quit IRC12:26
*** int_ua has joined #maemo-ssu12:49
*** M4rtinK has joined #maemo-ssu12:52
*** BCMM has quit IRC13:03
*** fredrinLap has quit IRC13:10
*** BCMM has joined #maemo-ssu13:30
*** dhbiker has joined #maemo-ssu13:32
*** int_ua has quit IRC13:52
*** dafox has joined #maemo-ssu13:55
*** dafox is now known as Guest9936513:55
*** ZogG has joined #maemo-ssu13:57
*** int_ua has joined #maemo-ssu13:58
*** sixwheeledbeast has joined #maemo-ssu14:01
*** ChanServ has quit IRC14:05
*** ChanServ has joined #maemo-ssu14:07
*** brooks.freenode.net sets mode: +o ChanServ14:07
*** Guest99365 is now known as dafox14:13
*** Codename-Strike has joined #maemo-ssu14:14
*** CodenameStrike has quit IRC14:15
*** arcean has quit IRC14:15
*** arcean has joined #maemo-ssu14:15
*** lizardo has joined #maemo-ssu14:17
*** pavi_ has joined #maemo-ssu14:28
*** dafox has quit IRC14:48
*** int_ua has quit IRC14:50
*** int_ua has joined #maemo-ssu15:02
*** sixwheeledbeast has quit IRC15:10
*** sixwheeledbeast has joined #maemo-ssu15:11
*** freemangordon has quit IRC15:11
*** sixwheeledbeast has quit IRC15:15
*** sixwheeledbeast has joined #maemo-ssu15:21
*** int_ua has quit IRC15:24
*** sixwheeledbeast has quit IRC15:26
*** dhbiker has quit IRC15:28
*** sixwheeledbeast has joined #maemo-ssu15:32
*** Codename-Strike has quit IRC15:37
*** fredrinLap has joined #maemo-ssu15:37
*** sixwheeledbeast has quit IRC15:38
*** arcean has quit IRC15:39
*** pavi_ has quit IRC15:49
*** fredrinLap has quit IRC16:06
*** CodenameStrike has joined #maemo-ssu16:09
*** NIN101 has quit IRC16:17
*** arcean has joined #maemo-ssu16:37
*** CodenameStrike has quit IRC16:38
*** CodenameStrike has joined #maemo-ssu16:38
luffreemangordon: I postpone thumb2 testing :( I have to use my dev phone for obexd+MAP bug hunting.16:44
*** CodenameStrike has quit IRC16:50
*** arcean has quit IRC16:55
*** arcean has joined #maemo-ssu16:55
*** luf has quit IRC16:58
*** fredrinLap has joined #maemo-ssu17:33
*** freemangordon has joined #maemo-ssu17:36
*** NIN101 has joined #maemo-ssu17:50
*** M13 has joined #maemo-ssu17:52
merlin1991fredrinLap: we've reached 423 dls of the mp :D18:07
fredrinLapmerlin1991: What mp?18:08
merlin1991cssu-thumb metapackge (latest versions)18:08
kerio:D :D :D18:08
merlin1991*version*18:09
freemangordonmerlin1991: something's wrong here, I refuse to believe there are more that 400 -thumb users18:21
*** sixwheeledbeast has joined #maemo-ssu18:25
merlin1991freemangordon: if I clean it down to 1 dl per ip it's 40518:28
freemangordonwell, if that is true, we should move -thumb to testing immediately :D18:30
peetahhow many people screwed up their thumb installation, reflashed and reinstalled it ?18:33
freemangordonlast version? no more than 518:33
freemangordonhmm, maybe about 1518:34
*** arcean_ has joined #maemo-ssu18:36
*** arcean_ has quit IRC18:39
freemangordonin the meantime: gcc version 4.7.2 20120910 (prerelease) (Linaro GCC 4.7-2012.09)18:40
keriorecompile *all* the packages! _ò/18:43
freemangordonwhy?18:43
freemangordonBTW i would've sticked with zeq's toolchain, but it sees it has some problem with NEON18:44
freemangordonlets see how it will be with the newer18:44
kerioat least, recompile *all* the xserver packages! _ò/18:50
*** sixwheeledbeast has quit IRC18:55
*** m0shbear has joined #maemo-ssu18:58
m0shbearis it possible to force an upgrade of stable without doign PC Suite backup?18:59
m0shbearI already use backupmenu for all imaging purposeds18:59
m0shbear**testing18:59
m0shbears/stable/testing/18:59
freemangordoncheck "problems" tab in application manager18:59
DocScrutinizer05(<freemangordon> well, if that is true, we should move -thumb to testing immediately :D) and alienate the 3266 proper testing users?19:00
freemangordonDocScrutinizer05: where did you get that 3266, I was wondering how big is -testing userbase19:01
DocScrutinizer05made it up, based on educated guess and the numbers from -thumb19:03
*** int_ua has joined #maemo-ssu19:03
* DocScrutinizer05 ponders to actually roll that cokie thingie suggested some month ago, every upgrade pings a tranceable IP, just once19:06
DocScrutinizer05trceable*19:06
DocScrutinizer05meh19:06
DocScrutinizer05coul dimplement that as browser invocation to show some "news" of CSSU19:07
*** sixwheeledbeast has joined #maemo-ssu19:08
DocScrutinizer05with an auto-forward to the real site, so noboy will set a bookmark to the tally site and call it repeatedly19:08
DocScrutinizer05typo day \o/19:08
DocScrutinizer05iirc BM already does that. Or was it alarmed?19:09
DocScrutinizer05was X-Fade able to get some statistics from real servers?19:12
DocScrutinizer05merlin1991: would *you* be able now? finally?19:13
DocScrutinizer05X-Fade: ping ^^^19:13
kerioDocScrutinizer05: maybe there are more cssu-thumb users than cssu-testing users!19:15
DocScrutinizer05only very maybe19:16
DocScrutinizer05odds are a lot of lusers that only moderately care to get new leete stuff but simply want to stay on a road with occasional update for bugfixes just intalled T or even S and never again showed up anywhere19:17
DocScrutinizer05compare number of registers users on tmo to number of posts on tmo to number of candidates for council19:19
DocScrutinizer05there seems to be 1% that ever posted anything at all, and 0.3% that do it frequently, while the  huge number of masses never utters a sign of life19:20
DocScrutinizer05still many are busy lurkers19:21
*** sixwheeledbeast has quit IRC19:21
DocScrutinizer05same here, or on IRC in general19:21
*** Guest57704 has joined #maemo-ssu19:23
*** dafox has joined #maemo-ssu19:23
*** dafox is now known as Guest7291919:24
freemangordonthe fuch, latest linaro gcc still issues broke neon instruction :(19:27
freemangordon*broken19:27
m0shbearis anybody else getting http://paste.ee/p/zyFOV ?19:28
m0shbearfor /var/lib/apt/lists/downloads.maemo.nokia.com_fremantle_ssu_mr0_._Packages  ?19:28
*** sixwheeledbeast has joined #maemo-ssu19:31
keriowat19:33
keriono19:33
m0shbeardisregard my point, a cleanout of /var/lib/apt/lists solved the issue19:35
*** Guest57704 has quit IRC19:35
kerioi should keep a lists of "modifications that could screw you over"19:35
kerioi manually optified /var/lib/gcc, /var/lib/apt and /var/lib/dpkg19:36
kerioand i have no emmc swap partition now19:36
kerioi don't remember anything else19:36
*** Guest72919 is now known as dafox19:36
m0shbearit appears that the forced backup is due to the builtin "Maemo update" thingy19:37
m0shbearand there's no mention in hildon-app-manager's logs19:37
m0shbearbut why now instead of when I upgraded factory to PR1.2?19:37
kerioforced backup? what?19:37
m0shbear"Maemo update; To update your device to this version of the operating system you must use Nokia PC Suite on your PC. Connect the device to your PC via a USB cable and launch the application."19:38
*** M13 has quit IRC19:39
keriooic19:41
kerioupgrading cssu-testing?19:41
kerioyou need to upgrade operator-name-cbs-widget first19:41
m0shbearthat app is the newest version, but $sudo apt-get upgrade has 58 queued packages, so i'll get to that before re-attempting the update of cssu-testing19:43
keriohahha19:43
keriothis will *not* end well19:43
*** dafox has quit IRC19:43
m0shbeareh, it's mostly busybox and various libraries19:43
m0shbearand how will this not end well?19:44
merlin1991because of bad stuff (tm) in extras-devel?19:45
*** dafox has joined #maemo-ssu19:45
m0shbeari'm used to things breaking19:45
*** dafox is now known as Guest2423819:45
*** sixwheeledbeast has quit IRC19:46
*** sixwheeledbeast has joined #maemo-ssu19:47
*** M13 has joined #maemo-ssu19:48
keriobecause you're only supposed to update CSSU via ham19:54
kerioapt is "unsupported"19:54
*** sixwheeledbeast has quit IRC19:58
*** BCMM has quit IRC19:59
*** sixwheeledbeast has joined #maemo-ssu20:02
*** sixwheeledbeast has quit IRC20:15
*** sixwheeledbeast has joined #maemo-ssu20:17
DocScrutinizer05OMG20:28
m0shbeareverything works20:28
m0shbearand that scares me20:28
m0shbear(this coming right after a successful update)20:28
m0shbearhmm, why do  mp-fremantle-community-pr and dnsutils conflict?20:31
*** fredrinLap has quit IRC20:32
*** sixwheeledbeast has quit IRC20:36
*** pavi has joined #maemo-ssu20:41
*** dhbiker has joined #maemo-ssu20:54
*** luf has joined #maemo-ssu21:12
*** M13 has quit IRC21:13
freemangordonthe fuck:21:29
freemangordonhttp://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=8428e84d42179c2a00f5f6450866e70d802d1d0521:29
freemangordonSep 22 19:26:07 Nokia-N900 kernel: [  572.136657] Alignment trap: not handling instruction ece4 at [<4026d84c>]21:29
keriowat21:29
freemangordonpulseaudio21:30
keriowtf did you do21:30
kerioto my n90021:30
freemangordoncompiled it and seems I am hitting that21:30
keriowell fix the kernel21:31
freemangordonfortunately there is a compiler switch to disabl eunaligned access, lets see if it will work21:31
kerioor tell gcc to only do aligned access21:31
keriohehe21:31
keriomaybe that's the problem with xserver?21:31
freemangordonNFC21:32
*** pavi has quit IRC21:32
int_uafreemangordon: but why is that "__LINUX_ARM_ARCH__ < 6" when the title says "ARMv6+" ?21:36
int_uawhy is it _less_ ?21:36
freemangordonbecause on < 6 it is prohibited21:37
keriobecause it'll trap if CONFIG_ALIGNMENT_TRAP is set, *AND* if arch is 5 or less21:37
kerioon 6+, it'll not trap21:37
freemangordon:nod:21:37
freemangordonCONFIG_ALIGNMENT_TRAP is set in our kernel :(21:43
kerioof course it is21:43
kerioit's probably a good thing21:43
keriojust add that patch21:43
kerioif our CPU can handle unaligned access21:44
freemangordonno, it is not a good thing, as our CPU can handle it at a cost of 1 additional cycle (AIUI)21:44
kerioyeah but you fix that by adding that patch21:44
kerioor something21:44
freemangordonyeah, I can try21:45
int_uaofftopic: is that possible to connect touchscreen on N900 to desktop through network?21:45
keriopaaaaaaaaaaaaaaaaalit21:45
keriodammit, he's not here21:45
kerioint_ua: what does that mean?21:45
int_ua=D21:45
int_uakerio: just an idea. To use touchscreen on N900 on my desktop21:46
int_uaI've seen something like USB over IP in Linux long time ago...21:46
int_uamaybe there is something like this to connect the touchscreen?21:47
keriowell, there's VNC...21:47
freemangordonhmm, I replaced that code with a memset and now I have segmentation fault, something else is wring :D21:48
freemangordon*wrong21:48
kerioyou what21:48
freemangordonit was an init loop that was causing SIGILL, I replaced it with memset and got sigsegv21:49
keriocool, that's easier to diagnose :D21:49
keriofix the damn kernel21:49
freemangordonno, it is not the kernel21:49
keriosure it is, it traps unaligned access21:49
kerioand now you botched the aligned replacement21:50
freemangordonnahh, sigsegv is a different animal21:50
freemangordonhehe, I am stupid :D21:51
freemangordonmemset's second argument is not array size ;)21:51
kerioyou could just recompile the original source with the aligned access option21:53
freemangordonalready did that, don;t help21:53
keriowait, really? D:21:54
keriothen what's wrong?21:54
freemangordonNFC21:54
freemangordonlets see how it will be wit *correctly* called memset21:54
kerioalso, is there some binary on my n900 that could do unaligned access?21:54
keriosome binary that comes from cssu-θ21:55
freemangordonkerio: it is most probably gcc issuing incorrect instruction21:55
keriowhich gcc?21:55
freemangordonboth zeq's and mine (4.7.2 linaro 2012.09)21:56
kerioyeah but did you use it already?21:56
keriothis new one21:56
freemangordonit runs YAY21:57
kerioYAY21:57
keriountil it'll segfault again21:57
keriofor instance, when accessing that initialized data21:57
freemangordonoulseaudio enable every damn hacking options, like fast-math and unsafe-optimisation and whatnot21:57
keriohehe21:57
freemangordonI really wonder how this wirks at all (stock one)21:58
freemangordon*works21:58
freemangordonbut now it works, lets see if there is sound21:58
freemangordonyes, there is21:59
freemangordon:D21:59
kerio:D21:59
keriootoh, you should look for a more general fix21:59
freemangordonI know21:59
kerionew kernel-cssu! ^____^21:59
kerioit might be the excuse to install it21:59
freemangordonOK, works after reboot too.22:00
m0shbearkerio: alignment issues usually trigger SIGBUS22:01
m0shbearSEGV is more pagetable protection issues22:02
freemangordonm0shbear: i have a clear message from the kernel22:02
keriom0shbear: no, the SIGSEGV is from his botched replacement22:02
keriothe original issue is SIGILL iirc22:02
freemangordonyep22:02
freemangordon<freemangordon> Sep 22 19:26:07 Nokia-N900 kernel: [  572.136657] Alignment trap: not handling instruction ece4 at [<4026d84c>]22:02
freemangordonand that sigills in gdb22:03
deepyAlignment trap22:03
freemangordonece4 is vstmia22:03
freemangordonVSTMIA          R4!, {D16-D17}22:03
m0shbeari know not arm asm22:03
freemangordonand r4 is not word aligned22:03
m0shbearand zeroing the unaligner bits -> segv?22:04
freemangordonthough it seems like a problem in the toolchan22:04
keriom0shbear: no, not knowing how memset works -> segv22:05
kerio:D22:05
freemangordonm0shbear: i passed memset array size as second parameter nad -1 as third22:05
freemangordonkerio: hehe22:05
freemangordonthe expected result - sigsegv22:06
m0shbearis it possible to trap right before to do regs + callstack dump?22:06
keriom0shbear: it's an already-diagnosed problem though22:07
freemangordonm0shbear: of course i can, i have gdb installed, but why should i do that?22:07
kerionew gcc generates code with unaligned access, and the kernel is configured to trap that22:07
keriobut it works fine in recent ARM archs22:07
m0shbeargcc 4.7?22:07
freemangordonyep22:07
freemangordon4.6 too22:07
keriofreemangordon: then how does the thumb stuff work?22:07
freemangordonkerio: though the patch is for kernel unaligned access22:07
keriooh22:08
freemangordonkerio: that is why i said it is most probably a toolchain problem22:08
lufI'm curious if someone using BT + synchronizing contact and more between phones :)22:08
m0shbearwhy not retain 4.5 to keep gcc from UA code generation?22:08
keriobecause 4.7 is better22:08
freemangordonm0shbear: there is an option to disable it22:08
freemangordonwhich didn't help22:09
m0shbearhow bloated is gcc's arm backend?22:10
freemangordonagain a sign of misbehaving toolchain22:10
keriofreemangordon: did you properly make clean and make?22:10
freemangordonm0shbear: well, it is gcc-linaro, should be the best one available22:10
freemangordonkerio: come on :P22:10
keriono srsly22:10
freemangordoniirc22:10
m0shbearproably a few assert()s here and there will trigger ICEs, making it easier to hunt down the source of the unavoidable UAs22:11
keriom0shbear: how do you test all codepaths though?22:11
m0shbeari want to say lint, but know im prolly wrong22:12
freemangordonm0shbear: pulseaudio source and toolchain are available to be downloaded, I will owe you a beer if you can trigger an ICE22:12
freemangordonm0shbear: http://wiki.maemo.org/CSSU-thumb_toolchain_setup_%28gcc4.7.2-linaro%2922:13
freemangordonand puleaudio in in maemo repos22:13
freemangordon*pulseaudio22:14
freemangordonif you want to play with it I will hand you a couple of patches you'll need22:14
kerioICE?22:14
freemangordoninternal compiler error22:14
*** int_ua has quit IRC22:15
m0shbearthe first trick is to figure out the offensive lines of code22:16
freemangordonfor (c = 0; c < PA_CHANNELS_MAX; c++)22:16
freemangordonm->map[c] = PA_CHANNEL_POSITION_INVALID;22:16
freemangordonpulseaudio-0.9.15\src\pulse\channelmap.c:16522:17
keriowhy does m->map point to an unaligned thing?22:17
freemangordonbecause m is a structure22:17
kerioi said m->map, not m22:17
kerioand i said point to22:17
keriounless the problem is m->map itself22:18
m0shbearand inducing padding may unalign other instances22:18
freemangordontypedef struct pa_channel_map {22:18
freemangordonuint8_t channels;22:18
freemangordon/**< Number of channels */22:18
freemangordonpa_channel_position_t map[PA_CHANNELS_MAX];22:18
freemangordon/**< Channel labels */22:18
freemangordon} pa_channel_map;22:18
freemangordonthat is why22:18
kerio...that's stupid22:19
kerioWHY UINT822:19
m0shbearuint8_t _pad[3]; after .channels22:19
m0shbearwhy or why not22:19
kerio*why*22:20
freemangordonm0shbear: no way, there are closed source PA plugins from Nokia22:20
m0shbearprobably the code was assumed for byte addressable machines22:20
m0shbearnamely x8622:20
freemangordonARM can access bytes, not sure for NEON SIMD22:20
freemangordonthough it is a toolchain problem, if i find time will try to reproduce it with a simple helloworld.c and open a bug22:21
m0shbearapart from x86, which other arches are alignment-agnostic wrt word access?22:22
m0shbearalso, is it possible to do struct translation much like ksplice?22:26
lufShit. obexd redefine one type which prevent phone synchronizing :(22:26
m0shbearin order to not break abi yet work around gcc bug22:26
freemangordonm0shbear: i'll just patch it to use memset and will move on22:28
lufbut it seems nobody is using it as I see no bug report for it.22:28
*** dhbiker has quit IRC22:29
m0shbeari'll get to playing with unaligned struct members after my pm nap and dinner22:30
m0shbeari expect fun times and a few ICEs22:30
*** toxaris has joined #maemo-ssu22:33
freemangordonluf: you still play with obexd? I am eager for new bluetoothd to test, i use BT pretty much22:37
luffreemangordon: unfortunately yes.22:44
lufI stucked with some stupid mistake with pointers but I can't locate it L(22:45
lufBTW bluez-4.99 has more problems - It has some delay during connecting with HF and also it doesn't show the battery level :(22:46
*** macmaN has quit IRC23:03
*** macmaN has joined #maemo-ssu23:06
*** BCMM has joined #maemo-ssu23:16
luffreemangordon: ping23:18
freemangordonpong23:19
freemangordonluf: ^^^23:19
luffor what are you using BT?23:19
freemangordonHF, file transfer,what else... lemme think23:20
freemangordonaah, mouse, keyboard23:20
*** Guest57704 has joined #maemo-ssu23:20
lufHF in car (also phonebook, MAP, ...) or just sound?23:21
freemangordonluf: sending contacts sometimes23:21
freemangordonhf in my ear23:21
freemangordonwell, maybe it is HS profile didn't check it23:21
lufnp. thanks for the info.23:22
freemangordonboth profile supported23:22
freemangordonHFP/HSP23:22
*** M4rtinK has quit IRC23:27
*** M4rtinK has joined #maemo-ssu23:28
*** int_ua has joined #maemo-ssu23:30
*** Sc0rpius_ has quit IRC23:33
lufHmmm valgrind points me to something what seems ok for me. But who knows ...23:40
*** Guest57704 has quit IRC23:47
*** Guest57704 has joined #maemo-ssu23:48
luffreemangordon: can you help me?23:49
*** MrPingu has joined #maemo-ssu23:58
lufSomeone keen with valgrind here?23:58
keriofreemangordon: hmm, i can't seem to get a cursor over buttons in the menus23:59
*** lizardo has quit IRC23:59
keriowhoops, ECHAN23:59

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!