IRC log of #maemo-ssu for Saturday, 2011-12-03

*** scoobertron has quit IRC00:01
*** bsdmaniak has quit IRC00:10
*** rd has quit IRC00:33
*** andre__ has quit IRC00:57
*** Pali has quit IRC01:27
*** scoobertron has joined #maemo-ssu01:37
*** jonwil has joined #maemo-ssu01:53
MohammadAGmerlin1991: Ping02:02
merlin1991MohammadAG: pong02:04
MohammadAGmerlin1991: Email still formal02:04
merlin1991rofl02:04
*** mase76 has quit IRC02:17
merlin1991FFS kubuntu stuck at bootup02:20
MohammadAGTakes 30s here02:29
MohammadAGBut hey, thats qt for you02:29
MohammadAGI really wish i could get gnome back02:29
*** mase76 has joined #maemo-ssu02:30
merlin1991yeah, but unity, never for me02:31
merlin1991btw any dependency of camer-ui which I should know of, seems the control file is missing some (I've got undefined stuff errors when compiling)02:32
*** mase_76 has joined #maemo-ssu02:40
*** NIN101 has quit IRC02:40
*** mase76 has quit IRC02:43
*** scoobertron has quit IRC02:46
* merlin1991 needs to find his usb -> micro-sd adapter02:51
*** mase76 has joined #maemo-ssu02:52
*** mase_76 has quit IRC02:55
*** M4rtinK has quit IRC02:56
*** mase_76 has joined #maemo-ssu03:05
*** mase76 has quit IRC03:09
*** arcean has quit IRC03:12
*** mase_76 has quit IRC03:21
jonwilMohammadAG: ping03:34
*** mase_76 has joined #maemo-ssu03:34
merlin1991jonwil: he's afk for today03:35
jonwilok03:35
merlin1991as in went to bed :D03:36
jonwil:)03:36
merlin1991it's crazy how big the nemo images end up03:37
merlin1991you dl a 165 mb bz2 file03:38
merlin1991and it uncompresses to 3.6 GB03:38
merlin1991well partly due to the fact that it's a raw diskimage and it contains loads of 0s :D03:38
*** mase76 has joined #maemo-ssu03:39
jonwilI think operator-name-cbs-widget is looking good, the only remaining problem seems to be someone who's network is sending garbage Cell Broadcasts03:42
jonwilAnd short of being able to capture the raw packets from the cell modem to the cellular services daemon, I cant see any way to identify just whats going on or what I can do to solve it03:43
*** mase_76 has quit IRC03:43
jonwilI think its looking good as something that can go into cssu-t03:43
merlin1991hm thinking about it, it makes no sense to load a kernel image into memory just to reboot a second later, what was I thinking adding -R to that flasher cmd? :D03:49
*** mase_76 has joined #maemo-ssu03:50
merlin1991sadly I can't test it properly03:50
merlin1991I've got 2 n900s, one with a completely broken modem, and the other one only does 3G :D03:51
*** mase76 has quit IRC03:53
*** mase76 has joined #maemo-ssu04:00
*** mase_76 has quit IRC04:04
*** mase_76 has joined #maemo-ssu04:11
*** mase76 has quit IRC04:14
*** freemangordon has quit IRC04:17
*** mase76 has joined #maemo-ssu04:21
*** mase_76 has quit IRC04:24
*** mase_76 has joined #maemo-ssu04:31
*** mase76 has quit IRC04:35
*** mase76 has joined #maemo-ssu04:42
*** mase_76 has quit IRC04:45
*** mase76 has quit IRC04:56
*** mase76 has joined #maemo-ssu05:09
*** amiconn has quit IRC05:10
*** amiconn_ has joined #maemo-ssu05:10
*** amiconn_ is now known as amiconn05:11
*** mase_76 has joined #maemo-ssu05:13
*** mase76 has quit IRC05:17
*** jonwil has quit IRC05:19
*** mase76 has joined #maemo-ssu05:23
*** mase_76 has quit IRC05:27
*** mase_76 has joined #maemo-ssu05:34
*** mase76 has quit IRC05:37
*** mase76 has joined #maemo-ssu05:44
*** mase_76 has quit IRC05:48
*** mase_76 has joined #maemo-ssu05:55
*** mase76 has quit IRC05:59
*** mase76 has joined #maemo-ssu06:05
*** amiconn has quit IRC06:07
*** amiconn_ has joined #maemo-ssu06:07
*** amiconn_ is now known as amiconn06:07
*** mase_76 has quit IRC06:09
*** nox- has quit IRC06:10
*** mase76 has quit IRC06:19
*** mase76 has joined #maemo-ssu06:32
*** mase76 has quit IRC06:40
*** mase76 has joined #maemo-ssu06:53
*** mase_76 has joined #maemo-ssu06:57
*** mase76 has quit IRC07:01
*** mase76 has joined #maemo-ssu07:08
*** mase_76 has quit IRC07:11
*** mase_76 has joined #maemo-ssu07:18
*** mase76 has quit IRC07:22
*** mase76 has joined #maemo-ssu07:29
*** DocScrutinizer has quit IRC07:29
*** DocScrutinizer has joined #maemo-ssu07:29
*** mase_76 has quit IRC07:33
*** mase_76 has joined #maemo-ssu07:39
*** mase76 has quit IRC07:42
*** mase76 has joined #maemo-ssu07:53
*** mase_76 has quit IRC07:56
*** mase_76 has joined #maemo-ssu08:03
*** mase76 has quit IRC08:07
*** mase76 has joined #maemo-ssu08:14
*** mase_76 has quit IRC08:17
*** mase_76 has joined #maemo-ssu08:24
*** mase76 has quit IRC08:28
*** mase76 has joined #maemo-ssu08:34
*** mase_76 has quit IRC08:38
*** mase_76 has joined #maemo-ssu08:45
*** mase76 has quit IRC08:48
*** mase76 has joined #maemo-ssu08:55
*** mase_76 has quit IRC08:59
*** psycho_oreos has quit IRC09:18
*** jonwil has joined #maemo-ssu10:07
*** psycho_oreos has joined #maemo-ssu12:02
*** scoobertron has joined #maemo-ssu12:16
*** mirandir has joined #maemo-ssu12:24
*** freemangordon has joined #maemo-ssu12:33
*** Pali has joined #maemo-ssu12:54
*** arcean has joined #maemo-ssu13:00
*** M4rtinK has joined #maemo-ssu13:01
*** psycho_oreos has quit IRC13:06
*** scoobertron has quit IRC14:02
*** scoobert1on has joined #maemo-ssu14:02
*** andre__ has joined #maemo-ssu14:14
*** andre__ has joined #maemo-ssu14:14
*** mirandir has left #maemo-ssu14:50
*** jonwil has quit IRC15:20
*** jonwil has joined #maemo-ssu15:20
*** m0use has joined #maemo-ssu15:44
*** KaziKluBey has quit IRC15:55
*** rd has joined #maemo-ssu16:48
DocScrutinizersomebody seen freemangordon answering my comments regarding thumb?16:58
DocScrutinizeror is he still busy with funeral party for his plans to get thumb-Qt?16:59
*** psycho_oreos has joined #maemo-ssu17:00
RaimuI don't think he's been around for a little over 24 hours.17:01
DocScrutinizeras soon as there's an evidence for a potential race or other statistically happening glitch, there's no more use at all in doing stress tests to show X hours of flawless operation under any particular circumstances - just to mention that fact17:03
DocScrutinizerlast pst: [2011-12-01 23:47:21] <freemangordon> Yeah, I know17:06
DocScrutinizerso I honestly hope he noticed thumb being thoroughly fsckdup in OMAP3430 on N900, and no use at all in him further investigating thumb errors in gcc17:08
DocScrutinizereven when he came up with a whole Qt thumb build that uses 40% less mem footprint and runs fine on his device and system for days, we wouldn't want to touch it with a 10 ft pole17:09
DocScrutinizerunless he came up with a *comprehensive* list of *all* silicon errata in 3430, and analyzed each single one of them and explained how his workaround is proven to deal with it so it won't cause anymore problems17:11
freemangordondoc, hi17:20
freemangordonseems i missed the thumb party17:20
freemangordonDocScrutinizer ^^^17:21
DocScrutinizerfreemangordon: hey, we got chanlog :-D17:21
freemangordonok, will look at it17:21
freemangordonjust to finish gcc 4.6.2 arm toolchain :D17:22
DocScrutinizerbottom line: there ARE SiERR in thumb, and they aren't even fixable via workarounds as those are mutually exclusive for different aspects of thumb borkedness17:22
freemangordonok, lets check the logs17:23
*** rd has quit IRC17:27
freemangordonok, seems we are talking about errata 430973 and 687067, correct?17:31
freemangordonDocScrutinizer ^^^17:31
DocScrutinizerfreemangordon: I'm not going to redo the hours and days of work stskeeps and dm8tbr did17:32
DocScrutinizerplus some more weeks of investigations of the guys that those two pals based on17:33
jonwilDo we know if there is official info from TI about broken thumb on OMAP3430?17:33
DocScrutinizergoogle is your friend17:34
jonwilok17:34
freemangordonwell, as it seems again that we are talking nonsense I will and thumb discussion until i have working gcc 4.6.2 toolchain. WHEN i test Qt built with gcc 4.6.2 then i will continue comments re thumb. Just for info kernel patch fot errata  430973  is enabled in both stock kernel and KP. And afaik there is a patch in u-boot for errata 687067.17:37
DocScrutinizerso I honestly hope he noticed thumb being thoroughly fsckdup in OMAP3430 on N900, and no use at all in him further investigating thumb errors in gcc17:38
DocScrutinizereven when he came up with a whole Qt thumb build that uses 40% less mem footprint and runs fine on his device and system for days, we wouldn't want to touch it with a 10 ft pole17:38
DocScrutinizerunless he came up with a *comprehensive* list of *all* silicon errata in 3430, and analyzed each single one of them and explained how his workaround is proven to deal with it so it won't cause anymore problems17:39
RST38hsorry but why is thumb fucked up? I am using it it is ok17:39
DocScrutinizer*SIGH*17:39
DocScrutinizerwhy is the sky blue, why does C++ suck?17:40
DocScrutinizerit got Silicon Errata17:40
DocScrutinizerin OMAP343017:40
freemangordonDocScrutinizer, there are fixes for those17:40
DocScrutinizerpfff17:40
DocScrutinizerobviously those doing such fixes rhink they won't work17:41
RST38hEveryone has got errata17:41
freemangordonyes17:41
freemangordonagree17:41
RST38hI know some pretty "respectable" chips that have got errata for their main instruction set ;)17:41
freemangordonOMAP3430 is full of sillicon erratas, it is not only thumb17:41
freemangordonshould we stop using n900?17:42
DocScrutinizerso if you got better "fixes" you should discuss it with the guys that originally dealt with those SiERR and kernel patches etc17:42
RST38hSo, you have to be more specific about THUMB in 343017:42
freemangordonRST38h, more specific as in?17:42
RST38hfreemangordon: What exact issue prevents people from using thumb on 3430, as Doc insists?17:43
freemangordonHave no clue of any besides gcc17:43
DocScrutinizerfreemangordon: I simply don't think you are the fist guy on this globe to look into a way to fix thumb SiERR on OMAP3430, and it's quite discouraging that there's not already been a "THUMB ON TI OMAP IS FIXED NOW" news headline17:43
DocScrutinizerand I'm not joining in to a futile effort17:44
freemangordonQt in simbian is thumb compiled, you know that?17:44
freemangordonsymbian*17:44
DocScrutinizerthe issue been around since at least 24 months now17:44
freemangordonDocScrutinizer, Qt symbian libraries are thumb compiled, besides gui and webkit.17:45
freemangordonare there 3430 symbian phones?17:45
DocScrutinizerpretty please refer to all the fucking quoates I gathered for you, now it's up to you to escalate it to the guys I quoted. I'm simply relying on their judgement, and ETX17:46
DocScrutinizerI already expended >3h to explain to you why I initially and instantly said "beware of thumb" and where you could find additional info. I'm not going to invest further time wasted to discuss the issue on a "but why..." basis17:48
*** mirandir has joined #maemo-ssu17:50
DocScrutinizerwhen you come back with "I have looked into 430973 and 687067, discussed it with the original authors, had an idea how to fix it but they won't listen, here's an URL to a paper explaining my take on it and how you could fix" - come back as I'm interested. Until then, no way17:50
freemangordon430973 is already in n900 kernel17:51
freemangordonok, the patch17:51
freemangordonCONFIG_ARM_ERRATA_430973=y17:52
DocScrutinizerand CONFIG_ARM_ERRATA_687067=y, are mutually exclusive17:53
DocScrutinizerAIUI17:53
freemangordonit should be done in uboot AFAIK, not in kernel17:53
DocScrutinizerso fix that, convince the authors of those patches it's a viable fix, and you're the hero of the week17:53
freemangordonfix what, there is uboot patch already17:54
freemangordonAFAIK17:54
* jonwil hopes he can get the ball rolling on getting operator-name-cbs-widget into cssu before he goes on holidays17:55
DocScrutinizerfreemangordon: and what makes you think an uboot fix would help for kernel space (not saying it doesn't - just asking if it actually does)17:56
freemangordonbecause if the nature of the errata17:57
freemangordon"If the IBE is to be enabled, then the L1 System Array Debug Register 0 should be initialized to a17:57
freemangordonzero value."17:57
freemangordonthat is from the description17:58
DocScrutinizerok, so next step for you to do: check this with dm8tbr and stskeeps as I got my info from them, and probably they will send you to the original authors of those patches so you can ask them if the bug is closed then with applying both patches, one in uBoot one in kernel space17:58
freemangordonThis register is for RAM array debug purposes and is not used as a part of normal17:58
freemangordonfunctionality. It is only accessible in a privileged secure mode. Therefore, it can be statically17:58
freemangordoninitialized as a part of the boot code sequence. If the register is used for debug purposes, the value17:58
freemangordonshould be reset to zero when the debug sequence completes.17:58
freemangordonDocScrutinizer, for third time I am saying CONFIG_ARM_ERRATA_430973=y17:59
DocScrutinizerI am saying 6456th time: so what17:59
freemangordonnothing, leave it18:00
* jonwil wishes he could find a way to help 654321 find out why he gets garbage cell broadcast SMS messages on his N90018:00
DocScrutinizeryou're assuming it already IS fixed while the original authors claim the patches are mutually exclusive - sorry when I don't buy your story18:00
jonwilbut without getting into GDB or packet logging the cellmo, there is no way to do it18:00
freemangordonok, I will ask them. Who are the original authors?18:02
DocScrutinizerbtw we got no uBoot on N900 by default, not really18:02
freemangordonso?18:03
DocScrutinizer[2011-12-03 16:54:39] <freemangordon> fix what, there is uboot patch already18:04
DocScrutinizerand MEH now, for good18:04
freemangordonI asked so what if we don't have uboot by default, it is easy to be installed18:04
freemangordonwell, that is not productive, let me finish with 4.6.2 toolchain,.18:05
Palifreemangordon: I'm prepairing u-boot rx51 patches for upstreaming...18:05
freemangordongood, is there a patch for 687067 errata?18:06
PaliI do not know, but I think no18:06
freemangordonnot good :D18:06
Paliif it is rx51 specified, sure no18:06
Paliif it is somewhere in omap code, we can try to find it...18:07
DocScrutinizerPali: patches to a chainloaded uBoot might not help anything, as there are things like initializing RAM etc that can *nly* be done by first step bootloader. If that one fails on it, it's pretty hard to do them later _again_18:07
freemangordonno, it is OMAP343018:07
PaliDocScrutinizer: do you mean first step booloader x-loader or NOLO?18:08
DocScrutinizerwhatever is doing the specific initialization18:08
DocScrutinizerwas an random example, not telated to specific patch18:08
DocScrutinizerrelated*18:08
PaliSo we do not know it18:08
DocScrutinizerjust pointing to the difference between original uBoot and chainloaded uBoot18:09
freemangordonwell, if there is a code under our control which runs in secure mode, the errata could be fixed according to docs.18:09
DocScrutinizerform NOLO POV uBoot and linux kernel are no differnece at all18:09
Paliyes, you are right18:10
DocScrutinizerahaaa, well not even NOLO is running under secure mode AFAIK18:10
DocScrutinizerfor sure uBoot is NOT running under secure mode18:10
DocScrutinizerNOW we're talking18:11
Palix-loder is the first booloader - it is signed?18:11
DocScrutinizeryes18:11
DocScrutinizerneeds to be18:11
DocScrutinizerthat's why we can't patch it18:11
Palido we know something how to generate signature?18:11
DocScrutinizerwe however can and did patch NOLO18:12
Paliand is NOLO signed?18:12
DocScrutinizeryes, get the secret key from Nokia, sign shit with it - simple as that18:12
DocScrutinizerNOLO is NOT signed afaik18:12
DocScrutinizeron N90018:12
DocScrutinizeron N9(50) it is18:13
Paliso secret key is in nokia... does omap3 have someting like unsigned boot?18:13
DocScrutinizerthat's why I say N900 NOLO is not running in secure mode18:13
DocScrutinizerPali: not really18:13
Paliso nokia has special omap board with unpublished sign key?18:14
DocScrutinizerwe could ask Nokia for a new xloader incorporating that patch supposed to go into uBoot18:14
DocScrutinizersigning is done on a special buildhost18:15
DocScrutinizerx86 or whatever18:15
DocScrutinizerwith special signing tools from TI, publicly available18:15
freemangordonhttp://www.digipedia.pl/usenet/thread/13019/12482/18:15
freemangordonhttp://git.doredevelopment.dk/?p=mirror/u-boot-omap.git;a=commitdiff;h=dd420b370b209e093bad391dcf0903609732cd0218:17
freemangordonpali, do we have that ^^^ ?18:18
Paliwhere?18:19
freemangordonin uboot18:20
Paliin which version?18:20
freemangordonlatest :D18:20
Palithere are more rx51 u-boot version...18:20
Paliif you mean my last rebased patches on top of master, I can look for it18:20
freemangordonyeah18:21
DocScrutinizeron carefully reading the answer by Måns Rullgård in above linked post, I read this as "don't use thumb on OMAP that has the SiERR as it's not at all fixable"18:21
Palifreemangordon: There is none commit with text "687067" on master: http://git.denx.de/?p=u-boot.git&a=search&h=HEAD&st=grep&s=68706718:23
Paliso no18:23
PaliDocScrutinizer: where is that special signing TI tool?18:25
DocScrutinizerand knowing XorA I don't think he " just screwed up something when testing gcc 4.5.x"18:25
DocScrutinizerTI documented the algo, I think signing tools are commonly available based on that18:25
DocScrutinizerbut they all need the "private" key matching the "public" key burnt to ROM mask or config flash storage of OMAP18:27
DocScrutinizeryou won't get any such private key18:27
DocScrutinizermatching the SOC in N90018:28
Palican we get public?18:28
DocScrutinizerMr Flop is sleeping on that one, it's under his pillow18:28
DocScrutinizernot even that one is available, as you can't read it out from SOC18:28
DocScrutinizerAFAIK18:29
freemangordonwell, i really want to build new toolchain and test with it. gcc 4.2.1 has too many bugs to be a reliable thumb working or not benchmark.18:29
DocScrutinizerthere's no such benchmark, ever18:29
freemangordonso bbl18:29
Paliubuntu has arm-linux-gnueabi-gcc 4.6.118:30
DocScrutinizerunless you construct a special code sequence to specifically invoke the bug18:30
DocScrutinizerthus proving it doesn't show up under conditions of that test, and explaining why the conditions are not any special so the result is generally valid for all situations of code and engine state18:31
DocScrutinizeraccording to axiomatic rule of "you never can prove that something doesn't ever fail"18:33
DocScrutinizerstatistics like "runs for me since 6 months now, without the error showing up" are inherently useless18:34
DocScrutinizerthis is not like science and engineering work18:34
*** mirandir has left #maemo-ssu18:36
* jonwil would give ANYTHING for a copy of connui-cellular source right now :P18:40
DocScrutinizerbtw you know you SHOULD NOT mix binaries built on different versions of gcc, on same system?18:40
DocScrutinizer~211918:40
DocScrutinizerI've seen problems introduced by such mixing several times already18:42
freemangordonPali, it is most-probably build against newr glibc18:42
freemangordonnewer*18:42
freemangordonwell, libc18:42
freemangordonDocScrutinizer: why is that?18:43
freemangordonas far as libc version remains the same there should be no problem18:43
DocScrutinizerI don't exactly recall the details, but most probably because of subtle ABI changes anywhere18:43
Paliyou can try to rebuild new gcc in scratchbox18:43
freemangordonthat is what i am doing18:44
Paliok18:44
freemangordonbut stupid linker "ld: cannot find crt1.o: No such file or directory"18:44
freemangordoncannot figure what is the problem any help?18:44
DocScrutinizersee, you're building new gcc against new libc, now you might have libc idiosyncrasies in gcc that may cause isssues with binaries later using an older libc as found on your tatget - sounds somewhat non-conclusive to me but maybe it's along that line18:46
freemangordonno, I am builbing gcc statically linked with correct target libc18:47
freemangordonand ld is part of binutils18:47
DocScrutinizeror maybe I'm plain wrong this time, and mixing things up.18:47
*** jonwil has quit IRC18:47
freemangordonseems like ./configure switch is missing, but i cannot find which18:48
DocScrutinizersrc code incompatibilities are so obvious that we don't need to discuss them here I guess18:48
DocScrutinizerbasically new compilers create new different binaries, even on same source code. Sometimes the new binary code behaves different to former code, and sometimes this is causing problems18:50
*** rd has joined #maemo-ssu19:02
*** rd has quit IRC19:46
*** nox- has joined #maemo-ssu19:48
*** rd has joined #maemo-ssu19:52
*** rd has quit IRC20:25
*** bsdmaniak has joined #maemo-ssu21:10
*** rd has joined #maemo-ssu21:14
*** bsdmaniak has quit IRC21:18
*** Pali has quit IRC21:19
freemangordonis compiling Qt with gcc 4.6.2 and binutils 2.22 :D. lets see how it will be when installed on the device21:34
*** rd has quit IRC21:49
*** rd has joined #maemo-ssu22:14
*** rd has quit IRC22:33
*** rd has joined #maemo-ssu23:07
*** rd has quit IRC23:15

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