00:29.05 | *** join/#htc-linux sess (n=kamui@ip70-171-82-42.no.no.cox.net) |
01:02.24 | *** join/#htc-linux gnif (n=gnif@203.36.89.23) |
01:02.30 | gnif | hi all |
01:03.12 | gnif | after stuffing around with my toolchain I have decided to try the prebuilt one on hh.org |
01:03.47 | gnif | the file seems very old though, anyone know of the state of this toolchain? |
01:18.30 | gnif | hey, kernel 2.6.18 is out :D |
01:18.39 | gnif | today |
01:28.51 | gnif | well, installed the toolchain, so far it seems to be good :D |
01:58.09 | gnif | compiled a kernel fine... now to see if it can boot |
02:02.30 | gnif | kernel boots :D |
02:02.48 | gnif | only problem is, the old toolchain is pre gcc 4.0 |
02:02.55 | gnif | thus no abi support |
02:03.01 | gnif | but thats ok for my needs :D |
02:33.10 | gnif | woo, got dropbear compiled :D |
02:33.32 | gnif | had to compile zlib, then point dropbear to zlib |
02:33.52 | gnif | and modify the makefile to include the zlib dir in the CFLAGS |
02:34.05 | gnif | seems that ./configure doesnt do that |
02:35.05 | gnif | now to get busybox |
03:11.50 | *** join/#htc-linux g3gg0__ (n=g3gg0@ppp-82-135-5-25.dynamic.mnet-online.de) |
05:28.36 | gnif | it seems that a kernel compiled with pre gcc 4.0 wont boot an initrd image on the magician |
05:29.05 | gnif | i dont have a card reader/writer with me at the moment, so I cant test booting directly from a SD card |
05:53.25 | gnif | my initrd image is good tho |
05:53.36 | gnif | it built a RSA and DSS key, and started dropbear |
05:53.53 | gnif | now to just figure out how to configure the ethernet |
06:01.56 | *** join/#htc-linux awelux (n=awelux__@dslb-084-058-144-093.pools.arcor-ip.net) |
06:06.02 | gnif | changed a few settings, trying to get the kernel to boot the initrd |
06:18.24 | gnif | no go, need a new toolchain |
06:33.37 | *** join/#htc-linux rob_w (n=rob_w@p549B9365.dip0.t-ipconnect.de) |
06:42.47 | gnif | ho rob |
06:42.52 | gnif | *hi |
06:52.09 | rob_w | hi |
06:58.18 | gnif | sorry, using screen, remoting into my pc @ homt |
06:58.20 | gnif | *home |
06:58.37 | gnif | have you had any problems building kernels with the old gcc-3 toolchain? |
07:03.48 | gnif | i m trying to build a new toolchain |
07:03.51 | gnif | all is good so far |
07:04.06 | gnif | but... I cant compile glibc-2.3.6 |
07:04.58 | rob_w | whats the error |
07:06.51 | gnif | it complains that linuxthreads doesnt exist |
07:07.11 | gnif | from what I can tell, its a seperate file to dload that needs decompressing into the glibc dir |
07:07.20 | gnif | is this correct? |
07:07.33 | *** join/#htc-linux rno (n=rno@host-84-9-36-155.bulldogdsl.com) |
07:08.39 | gnif | wait.. i think i got it |
07:08.56 | gnif | didnt realise there was 2 dirs in the linuxthreads tar |
07:09.10 | gnif | seems to be working now LD |
07:09.12 | gnif | *:D |
07:11.53 | rob_w | hmm |
07:12.25 | gnif | I am following the tutoral from hh.org |
07:12.48 | gnif | needs a few steps added |
07:13.42 | gnif | asm-generic also needs copying to the include dir |
07:14.43 | gnif | once my build has finished, and I make sure it works, do you want me to tar it up as a new pre-build? |
07:14.53 | rob_w | people mostly use oe to build also standalone toolchains |
07:14.56 | gnif | using gcc-4 |
07:15.13 | gnif | oe? |
07:15.39 | rob_w | check openembedded.org |
07:16.02 | rob_w | thats a big toolchain in that sense in order to build whole images for gpe or opie or such |
07:16.14 | gnif | cool |
07:16.19 | rob_w | it also has image targets which equal a tool chain .. |
07:16.40 | rob_w | yeah get into that , that will make live easier for you and all others |
07:16.50 | gnif | i am not too keen on prebuilts, i like to know that i made everything properly myself |
07:17.00 | gnif | and how to do it if somthing breaks in the future |
07:17.36 | rob_w | there is no prebuild , it sucks all sources freshly , and it is higly configurable |
07:17.46 | gnif | hrmm, new error while building glibc |
07:18.11 | gnif | BUS_ISA undeclared |
07:18.29 | gnif | sounds like a missing include |
07:18.54 | rob_w | or configure switch ,, see this is all well handled with oe |
07:19.18 | gnif | have to have a look at it |
07:19.39 | gnif | sounds a little like scratchbox |
07:19.59 | rob_w | it is in that sense, but not little ;-) |
07:20.08 | gnif | lol |
07:20.39 | gnif | well, once I have built this one... will look at oe |
07:22.09 | gnif | ahhh, BUS_ISA has been renamed to CTL_BUS_ISA |
07:22.24 | gnif | this ver of glibc has not been patched yet |
07:25.21 | gnif | yup, that did it... |
07:32.14 | gnif | heh, more probs... these were fixed ages ago (2005-11-17), makes me wonder about the freshness of this tar |
08:15.24 | rob_w | hehe see ,, you dont have that with oe either |
08:19.54 | *** join/#htc-linux florian_kc (n=fuchs@dns.kernelconcepts.de) |
08:22.08 | gnif | hutup :P |
08:22.11 | gnif | *shutup |
08:22.24 | gnif | Hit more problems... gonna try out oe |
08:27.08 | florian_kc | good morning |
08:27.13 | gnif | hello |
08:28.06 | gnif | crap, the OE db is large |
08:28.09 | rob_w | gnif, florian_kc is a good point to get oe knowledge |
08:28.28 | rob_w | gnif, as i said " .. not little " , hehe |
08:28.56 | gnif | no, its larger then a large pile of poo |
08:29.33 | gnif | meh, i am on 1.5mbit... 10 mins, max |
08:31.07 | gnif | why does everyone have a "_" in their nick? |
08:31.28 | gnif_wtf | :P |
08:40.54 | *** join/#htc-linux toi (n=pleemans@d54C2F583.access.telenet.be) |
08:41.03 | rob_w | well my name is robert wxxx so i go rob_W |
08:41.05 | rob_w | well my name is robert wxxx so i go rob_w |
08:45.47 | gnif | fair enough |
08:47.20 | gnif | hey, in the OE conf, it wants "MACHINE" |
08:47.29 | gnif | I assume "ipaq-pxa270.conf" would be correct |
08:47.31 | gnif | ? |
08:47.35 | gnif | I am using a magician |
08:47.50 | gnif | wait... |
08:47.53 | gnif | htcuniversal |
08:48.19 | rob_w | no it goes like BLUEANGEL or such .. either we do have MAGICIAN already or u need to define it |
08:48.40 | gnif | there is no magician |
08:50.03 | gnif | wow, how much does this thing build? its got defines for xserver/modules/patches |
08:50.27 | rob_w | it can build a full fledged distribution ... |
08:50.41 | gnif | wow |
08:50.56 | rob_w | ok go check the following dirs to get an idea oe/conf oe/packages/*/*.bb |
08:51.11 | gnif | already looking at em |
08:51.39 | gnif | jsut following the "getting started" |
08:52.01 | rob_w | go inside oe/conf/machine and copy a magician.conf from blueangel.conf ideally |
08:53.14 | rob_w | then see fe, oe/conf/distro where you see major distro targets .. then in oe/packages/meta you will find meta targets ... either full gpe or fe a dev sdk image |
08:53.42 | gnif | sorry if I sound reall noobish, but this is confusing as hell |
08:54.06 | rob_w | you can bitbake gpe-image , which will do the whole gpe for you in a image .. or you can then build single packages via bitbake -b path/to/oe/packgages/myapp/myapp.bb |
08:54.24 | rob_w | in the beginnning yes but then it gets quite hot !! i promise |
08:54.46 | gnif | sounds cool... |
08:55.07 | rob_w | and fe. bitbake -c clean -b path/to/oe/packgages/myapp/myapp.bb clean that builds for the packages |
08:55.49 | rob_w | for a full gpe image calculate some 2 -4 gb disk space ( as it gets all sources and cross compiles them ) |
08:56.16 | rob_w | with this chain , you can easily port to many architectures , either arm or mips or ppc or even x86 |
08:56.32 | gnif | its got in the conf "BA_MODULES_kernel26" and a stack of blueangel stuff... I have patches for the 2.6.18 kernel, what do I do here? |
08:57.07 | gnif | i am already impressed btw |
08:57.47 | rob_w | that tells what modules should be included in the image .. you can change them .. in order to get your kernel+patch combination you most probably make your on oe/packages/linux/mykernel.bb file |
08:58.08 | rob_w | yeah it is impressive and even feeds me and my family now for almost a year ;- ) |
08:58.20 | rob_w | ~hail florian_kc and oe |
08:58.24 | apt | ACTION bows down to florian_kc and oe and chants, "I'M NOT WORTHY!!" |
08:58.41 | gnif | hehe |
08:58.53 | gnif | ~hail me |
08:58.54 | apt | ACTION bows down to gnif and chants, "I'M NOT WORTHY!!" |
08:58.58 | gnif | lol |
08:59.13 | gnif | i wanted it to say "bows down to me" |
08:59.24 | rob_w | ~massacre gnif |
08:59.26 | apt | ACTION sets seeds of discontent and difference among the gnif; years later two factions have formed; history will take its course |
08:59.36 | gnif | heh |
08:59.39 | gnif | whos bot is that? |
08:59.53 | rob_w | ~apt |
08:59.56 | apt | extra, extra, read all about it, apt is the Advanced Package Tool, an advanced frontend for dpkg and rpm (depending on your system). APT is also known as the best package management system in existence today. |
09:00.09 | rob_w | well it is that apt bot here |
09:00.15 | gnif | yes, i know what apt is... |
09:00.17 | gnif | :P |
09:00.17 | rob_w | aka ibot |
09:00.21 | gnif | i couldnt live without it |
09:00.33 | rob_w | yeah he is so cool if you feel lonely |
09:00.45 | gnif | i cant understand why people still use rpm |
09:01.10 | rob_w | damn i do need to create a 5 dimensonal array .. but i cant imagine how and if even i need it really |
09:01.45 | gnif | char[][][][][] |
09:01.47 | gnif | :P |
09:02.28 | gnif | what the hell do you need a 5 dimn array for? |
09:02.40 | rob_w | i ve a log fucntion which is called with 4 parameters ( 3 integers 1 char string) so now i do need to flag each combination so in order to know if that exaxt error log was there before |
09:02.56 | rob_w | see what i mean ? |
09:03.10 | gnif | create a hashtable |
09:03.21 | rob_w | so i can fill my logs only when the error starts and when it goes away .. |
09:03.22 | gnif | generate a md4/5 sum from the 3 ints and string |
09:03.34 | gnif | then look in the hashtable for the md5 |
09:03.46 | gnif | if it exists, then dont add it |
09:03.59 | gnif | hashtable/vector |
09:04.40 | gnif | if that makes any sence? |
09:04.44 | rob_w | hmm sounds interessting .. so i have a 2dimension array with the flag (0|1) and the md5 |
09:04.53 | gnif | nah, not event that |
09:05.02 | gnif | jsut make a dynamic array |
09:05.19 | gnif | if the md5 exists in it, then you know the log has happend before |
09:05.28 | gnif | 1 dimention |
09:05.33 | rob_w | i see |
09:05.53 | gnif | should be fast too, no need to do string comparison |
09:06.45 | rob_w | hmm but i do need to repeat that error when it goes away so i do need the combinations of the 4 parameters |
09:07.17 | gnif | well then create a 2 dimention array, 1 dimn = the md5, 2nd = a struct with the details |
09:08.24 | gnif | btw, doesnt have to be md5, just get a nice fast hashing alg that will generate unique values |
09:08.50 | rob_w | hmm sounds more interessting .. so but the struct does need to carry ints and char strings . so i need a union ?!? |
09:09.07 | gnif | nope |
09:09.43 | gnif | struct LogEntry { |
09:09.43 | gnif | int your_int1 |
09:09.43 | gnif | int your_int2 |
09:09.43 | gnif | int your_int3 |
09:09.43 | gnif | char *errormessage |
09:10.53 | gnif | i would just use a vector |
09:12.25 | gnif | struct LogEntry { |
09:12.26 | gnif | int your_int1 |
09:12.26 | gnif | int your_int2 |
09:12.26 | gnif | int your_int3 |
09:12.26 | gnif | char *errormessage |
09:12.36 | gnif | and put the md5 in that struct |
09:12.47 | gnif | whoops |
09:12.56 | gnif | ... let me start over... |
09:13.10 | rob_w | its ok |
09:13.29 | rob_w | so i will go for a simple XOR |
09:13.30 | gnif | create a struct that contains, the MD5, your ints, and the char... then just create a vector<yourstruct>. |
09:13.45 | gnif | ok |
09:14.05 | gnif | then itterate the vector and look for entries that the checksum matches |
09:14.15 | gnif | crc32 would probarbly be better |
09:14.36 | gnif | XOR is going to just encrypt your data |
09:14.50 | gnif | you want a lossy alg that gives you just a hash |
09:15.00 | gnif | otherwise too much data to store |
09:15.04 | gnif | and compatre |
09:15.06 | gnif | *compare |
09:15.27 | rob_w | struct logarray { |
09:15.27 | rob_w | int source; |
09:15.27 | rob_w | int id; |
09:15.27 | rob_w | int class; |
09:15.27 | rob_w | char *message; |
09:15.28 | rob_w | int checksum; |
09:15.30 | rob_w | }; |
09:15.32 | rob_w | struct logarray logmessage[50]; |
09:15.43 | gnif | is 50 big enough? |
09:15.51 | gnif | c/c++? |
09:15.52 | rob_w | hmm for the start |
09:15.59 | rob_w | c |
09:16.04 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.216) |
09:16.10 | gnif | dont know too much c |
09:16.16 | gnif | do vetors exist? |
09:17.07 | gnif | then you could have: |
09:17.11 | rob_w | vectors ,, hmm u dont mean pointers don tyou |
09:17.14 | gnif | nope |
09:17.26 | gnif | a vector is a dynamic array |
09:17.55 | gnif | vector<logarray> logmessage; |
09:18.09 | gnif | to add an entry: |
09:18.18 | rob_w | right i might want the iterations of my messages tobe dynamic .. then i can add members as i wish |
09:18.21 | gnif | logmessage.push_back(entry); |
09:18.34 | gnif | exaclty |
09:18.37 | gnif | *exactly |
09:18.47 | gnif | you also then dont risk a buffer overflow |
09:19.08 | gnif | to itterate a vector: |
09:19.21 | gnif | for (int i = 0; i < logmessage.size(); i++ |
09:19.22 | gnif | ) |
09:20.27 | rob_w | hmm that would be really nice ,., but i do need to look up for dynamic array`s .. let me start with some fix values and then i see |
09:20.33 | gnif | fprintf(stdout, "%d, %d, %d = %s\n", logmessage.at(i).source, logmessage.at(i).id, logmessage.at(i).class, logmessage.at(i).message); |
09:21.07 | gnif | the question remains, does C have vectors? |
09:21.49 | gnif | anyway, tis time for me to go home... will jump back online when i get there |
09:23.05 | gnif | btw: I have a background in delphi, and have been learning c++ for the last 3-4 months... so if there is a better way, go for it. |
09:23.21 | gnif | ok, gone now :P |
09:24.11 | cr2 | gnif: C is the portable assembler, there are no arrays as you know them. |
09:24.42 | rob_w | hi cr2 |
09:24.49 | cr2 | hi rob_w |
09:28.47 | rob_w | what if i do struct logarray logmessage[] ; isnt that dynamic then ?? but how do i iterate through or load ? |
09:30.41 | cr2 | you need to allocate the space for this array yourself, so you will know where it is in memory, and what is its size. |
09:32.31 | cr2 | malloc() and free() make things dynamic. everything else is static, but you have an "advantage" of being able to point to any address in your process' address space. |
09:33.54 | cr2 | if it is in the kernel, then to everywhere ;) |
09:43.49 | rob_w | i see |
10:40.19 | *** join/#htc-linux toi (n=pleemans@d54C2F583.access.telenet.be) |
10:55.11 | *** join/#htc-linux gnif (n=gnif@203.36.89.23) |
10:55.25 | gnif | yay, finally got back :D |
10:56.05 | gnif | I re-installed my pc last night due to making a really stupid mistake (sudo rm /v* -R) |
10:56.23 | gnif | debian unstable is really nice |
10:57.50 | gnif | took me about 10 mins to install a base system |
10:58.04 | gnif | then 20-25 mins to setup x.org and install all the gnome packages |
10:58.23 | gnif | and just then, good old module-assistant built my video drivers for me :D |
10:58.57 | gnif | rob, you still there? |
10:59.14 | gnif | rob_w |
11:05.48 | rob_w | yes |
11:05.53 | rob_w | sorry was on lunch |
11:06.16 | rob_w | whats up |
11:06.37 | gnif | just wanted to know how you went |
11:07.02 | rob_w | i am doing ok , i think .. i went for the structure and a 1dimens array of that struct .. |
11:07.12 | gnif | cool |
11:07.24 | gnif | i am getting my oe thing setup |
11:07.32 | rob_w | then i now load the error into the struct if it is not existant yet .. still i need to find a point in my code where to check if the errors are gone .. |
11:07.51 | rob_w | so then i can go one calling my log functions as i like |
11:08.06 | gnif | nice |
11:08.16 | gnif | whats the app for (if you dont mind me asking)? |
11:08.43 | rob_w | sure , it is a deamon to talk to some elektronic hardware of my boss, where we monitor and control a big biogas engine |
11:09.01 | gnif | sounds interesting |
11:09.06 | gnif | how is it interfaced? |
11:09.10 | rob_w | and i did the image with oe , and the gui with gtk , only the deamon will be "closed" ion that sensce |
11:09.40 | rob_w | it talks to some device via canbus and to the controlling unit via rs485 |
11:09.45 | gnif | i love it when i hear of commercial applicationw with linux |
11:10.04 | gnif | whats rs485? just a variation of 232? |
11:10.42 | rob_w | gnif, yes it is serial connection but with some respect to its electronics .. so it is 2 wire diffrential .. |
11:11.15 | gnif | do you do much with the hardware side of things? or mainly software |
11:11.48 | rob_w | well i try to get into the electronics amap but yet i did drivers for the display device .. |
11:12.16 | rob_w | my boss is the electronic engineer , he will show you the running 1 and 0`s on a oszi with a second |
11:12.44 | rob_w | its quite cool , but yet i could need more money , ,bla anyway i need to finish that log thing today |
11:12.54 | gnif | cool... I wish I had that level of experiance... I have been trying to learn alot of the hardware site... but as soon as it gets more complex then a simple 16f type pic i get lost |
11:13.00 | gnif | *side |
11:14.08 | rob_w | i learned alot about those intefaces since i am here ,, so i can talk canbus quite good now ( at least enough to survive) |
11:14.43 | gnif | question: "Psyco JIT Compiler (http://psyco.sf.net) not available. Install it to increase performance" |
11:14.46 | gnif | should i do it? |
11:14.49 | gnif | and if so, how |
11:14.58 | rob_w | hmmi think i need another flag bit inside the struct to indicate it is already in |
11:15.16 | rob_w | pscyo is quite good if you have lots of ram . |
11:15.23 | gnif | 2GB enough? |
11:15.25 | gnif | :P |
11:15.29 | rob_w | should |
11:15.56 | rob_w | parsing the packages the 2nd and son time (despite you changed a bb file ) will get much quicker |
11:16.01 | gnif | its soooo nice to come home and sit down to a linux pc... i hate my work computer |
11:16.03 | rob_w | but your ram will fill like hell |
11:16.13 | rob_w | your aussie , right ? |
11:16.16 | gnif | as long as it cleans up afterwards |
11:16.19 | gnif | yes |
11:16.31 | rob_w | cool , would love to get a foot there someday |
11:16.57 | gnif | well, I am government, so cant do that much for ya unfortunatly |
11:18.01 | rob_w | i am just interested in some landscape s you have down there ,, rocky costs , outback , salt crockodiles n didgeridoo`s .( which i play btw ) |
11:18.18 | gnif | heh, ive never touched one :P |
11:18.30 | rob_w | i love them |
11:18.46 | rob_w | the clean my head quite fast and clean |
11:18.58 | gnif | i play guitar a bit |
11:19.04 | gnif | but nothing serious |
11:19.26 | gnif | sick!! "bitbake nano" |
11:19.32 | gnif | its like doing an emerge |
11:19.35 | gnif | in gentoo |
11:20.08 | gnif | what country are you in (i know your in the UK) |
11:23.55 | rob_w | wrong germany is my one |
11:24.29 | gnif | cool, my gradma is from over there |
11:24.30 | rob_w | gnif, yeah i bet you know #oe .. for your future detialed questions on the chain itself |
11:24.44 | rob_w | so your a quarter german ,, interesting |
11:24.50 | gnif | yup |
11:24.55 | gnif | she was a "klenke" |
11:25.04 | rob_w | sounds familiar |
11:25.26 | gnif | dunno, the only think I can say in german is "i love you" |
11:25.30 | gnif | lol |
11:25.42 | WizMaui | erm...can I use openembedded out of the box for building sth for htc wizard? |
11:25.48 | gnif | and I only learnt it because its such a rough word |
11:26.11 | gnif | :D |
11:26.27 | gnif | Wiz: I only discovered it today, ask rob |
11:26.42 | rob_w | WizMaui, wizard might also lag it as MACHINE , but so you only need to create a wizard.conf inside oe/conf/machine and there you go .. |
11:26.57 | WizMaui | sounds good ;) |
11:27.03 | WizMaui | hey i'm from germany too ;) |
11:27.10 | rob_w | most diffreence might be the kernel for all devices .. but we aim to catch that with different .configs |
11:27.17 | rob_w | WizMaui, geil |
11:27.17 | gnif | i feel special, theo only auzzie :D |
11:27.39 | WizMaui | hehe ;) |
11:27.48 | rob_w | as you are government you should be best to bring us into auzzie land , right ! ? |
11:28.00 | gnif | haha, just a developer |
11:28.14 | gnif | ich liebe dich |
11:28.17 | gnif | :P |
11:29.03 | gnif | just point out while I am at it... i am not gay... nor do I really love anyone |
11:29.16 | gnif | just the only words in german I know |
11:30.10 | rob_w | now i come to the point to rip apart my good old log functions .. brb |
11:30.18 | rob_w | ich dich auch ! |
11:30.41 | gnif | so what does OE do when its finished building a package? make a .tar? |
11:30.49 | gnif | hold on, google translator time |
11:31.20 | gnif | haha |
11:32.42 | gnif | debian unstable is worlds ahead of where it was before |
11:32.55 | gnif | system responsiveness is awesome |
11:38.04 | rob_w | IMAGE_FSTYPES = "jffs2 tar" |
11:38.04 | rob_w | <PROTECTED> |
11:38.43 | gnif | whats jffs2? |
11:39.05 | WizMaui | as you know htc wizard currently only boots via haret an cc.ext2.tar.gz ...which fs type would i need in oe |
11:39.26 | gnif | actually, haret just boots the kernel |
11:39.37 | gnif | you can use whatever fs you like |
11:40.08 | gnif | but because of the nature of the device, and limited space, I stick to ext2 |
11:42.09 | WizMaui | hmmm...i think we should get the sd card driver working... |
11:42.43 | rob_w | i would recommend ext3 on your sdcard , but you are right |
11:42.43 | gnif | for which device? |
11:42.43 | WizMaui | opie and co. are to big for my builtin memory |
11:42.43 | WizMaui | htc wizard ;) |
11:42.43 | gnif | i thought mmc was working |
11:42.43 | rob_w | jffs2 is a block fs for flash |
11:43.22 | gnif | ok, so OE builds a package to install into a final image... or what? |
11:43.45 | WizMaui | nope, the driver commands all timeout when initialize the mmc card |
11:44.31 | rob_w | oe builds a rootfs in that sense but it is able to pack it also into a jffs2 image in order tobe flashed as total |
11:45.24 | rob_w | and yes it does build packages in first place , socalled ipk`s those then in the finally stage it installes those ipk`s into its chroot rootfs |
11:45.44 | rob_w | its like baking as you see ! |
11:45.55 | gnif | :D |
11:46.51 | gnif | package qemu-qop-nogfx-native-0.8.0+svn20060921-r0: task do_compile: failed |
11:47.08 | gnif | block-qcow.c:26:18: error: zlib.h: No such file or directory |
11:47.31 | gnif | host needs it, or its a broken dependency? |
11:48.18 | cr2 | WizMaui: oe for zaurus fits into 16MB, even 14. |
11:48.39 | WizMaui | mhhh might be worth a try |
11:48.52 | WizMaui | but i would rather put that stuff on my mmc card ;) |
11:49.04 | WizMaui | erm...i meant minisd |
11:49.16 | cr2 | it is slower, and needs more power. |
11:49.20 | gnif | that fixed it, host needed the headers for zlib |
11:52.05 | gnif | WizMaui: MMC is much slower then system ram... if you can run the main stuff out of system ram it would be much better anyway |
11:52.43 | WizMaui | ok, agree with you |
11:53.28 | WizMaui | but i would appreciate it working on mmc too, because someone could want linux on a spare sd card and see nothing of it when it's not plugged in |
11:53.40 | gnif | yeah, i know what you mean |
11:54.12 | WizMaui | first i have to fix the kerneltree in our mercurial repo |
11:54.24 | WizMaui | don't know why but it lacks a lot of files ;) |
11:55.17 | rob_w | gnif , all packages having "native" on them are need on your host .. so yes you will need zlib.h reachable on the host |
11:58.15 | WizMaui | does someone of you knwo where to get monotone 0.27 or newer for debian sarge amd64? |
11:59.12 | gnif | dunno, I was running alioth64... but stopped after getting so frustrated that packages such as shockwave didnt work |
12:01.18 | cr2 | WizMaui: compile yourself. the only weird dependency is boost lib. |
12:01.47 | WizMaui | yep...but now i need a more current version of libboost? |
12:02.01 | WizMaui | i assume it hase no deb control files ;) |
12:02.11 | cr2 | try ./configure. it will tell you :) |
12:02.40 | cr2 | i compile oe on a machine with 8GB RAM. |
12:03.48 | WizMaui | the unstable packages work as well ;) |
12:04.22 | cr2 | bitbake opie-image works |
12:04.42 | cr2 | i also routinely compile zroadmap |
12:08.45 | gnif | 8GB, showoff :P |
12:08.52 | gnif | i thought 2GB was lots |
12:09.08 | gnif | but then again, this pc is almost 2 years old |
12:09.54 | WizMaui | my server runs a core 2 duo :p |
12:10.36 | WizMaui | but it has only 512mb ram :( |
12:11.03 | gnif | oh yeah, well my pc runs on a Z80 |
12:11.11 | gnif | overclocked to 14mhz |
12:11.24 | WizMaui | hot ;) |
12:11.44 | gnif | i wonder how long gcc will take to compile |
12:11.51 | gnif | :P |
12:12.28 | WizMaui | hey, was one of you ever annoyed of these "hello i'm a mac" adverst by apple? |
12:12.54 | gnif | we dont get adverts like that over here |
12:13.16 | gnif | and if we did, i would have dug my eyeballs out with a spoon |
12:13.51 | WizMaui | http://www.apple.com/getamac/ :( |
12:14.05 | cr2 | WizMaui: i have linux-ppc running on a powerbook :) |
12:14.21 | gnif | tis ok if your running linux on it... |
12:14.32 | WizMaui | i just don't like that most mak users think they are cooler than pc users ;) |
12:14.39 | gnif | but you should have just bought a pc and saved 100000 dollars |
12:14.47 | WizMaui | this is crazy, even since macs are now built with the same hw |
12:14.52 | gnif | yup |
12:15.00 | gnif | you can run XP on em |
12:15.22 | cr2 | who cares. i also have a dec alpha with linux. 64bit since 1992. cpu diversity is good. |
12:15.56 | gnif | yes, but big endieness is not a sign of performance |
12:16.04 | gnif | at the time yes |
12:16.15 | cr2 | alpha is littleendian, btw. |
12:16.26 | gnif | whatever :P |
12:16.34 | gnif | i mean, 64bit |
12:16.49 | gnif | looking at opie screenshots, looks very cool |
12:17.16 | gnif | does the phone work with it? |
12:17.59 | cr2 | 64bit cpu don't let you do crappy programming. sizeof(*) != sizeof(int) |
12:18.12 | gnif | true |
12:18.14 | cr2 | there is no "real" opie phone app. |
12:19.19 | cr2 | zphone and qomunicator were 2 attempts at it. |
12:19.27 | gnif | but then again, if your doing crappy programming, you need to be slapped about with a dead fish |
12:19.37 | gnif | are they any good? |
12:19.44 | cr2 | not really. |
12:20.13 | gnif | its very pretty |
12:20.35 | cr2 | write one :) |
12:20.42 | gnif | I plan to |
12:20.49 | gnif | gotta make my magician work first tho |
12:21.19 | cr2 | the phone on magician. |
12:21.26 | gnif | yup |
12:21.33 | gnif | i think its broken atm |
12:21.40 | cr2 | polishing the rest can take some time. |
12:22.29 | gnif | yeah, tell me about it... I have been working on a linux based media center for the xbox for months now... tis a big job just to make it useable, let alone pretty |
12:23.30 | gnif | and no, not based off anything, I hate mythtv/freevo... etc... and it has to be fast because of the limited hardware in the xbox |
12:24.01 | WizMaui | what about xbmc |
12:24.01 | WizMaui | it's rather complete, isn't it? |
12:24.02 | gnif | i am using it already, but... |
12:24.09 | gnif | 40char filename limitation |
12:24.12 | gnif | no samba server |
12:24.22 | gnif | web interface is horrible |
12:24.27 | gnif | and its illegal |
12:25.02 | WizMaui | if ms would'nt be so restrictive on their xdk ;) |
12:25.21 | WizMaui | as with xbox360 on the market i think they should consider releasing it ;) |
12:26.40 | gnif | they wont tho... |
12:26.46 | gnif | you are speaking about microsoft |
12:26.50 | WizMaui | i know ;) |
12:27.28 | gnif | they dont know the meaning of free, and dont care of making people's life easier, just making money |
12:28.36 | rob_w | ok now i can check my logerror magic |
12:30.22 | rob_w | segfault .. hehe not special on the first try |
12:30.31 | rob_w | after one hour of coding blindly |
12:35.23 | gnif | does OE generate the locale's every time, or is this just because its my first package? |
12:38.04 | rob_w | it generates them once ... so if dont change your glibc then its ok and done only once |
12:41.15 | rob_w | hmm i am having problems with the 4th paramter to the log function which is a const char *message |
12:41.38 | gnif | how can it be a const char |
12:41.42 | gnif | its not constant |
12:42.02 | gnif | if it makes it easier, you could use a string |
12:42.57 | WizMaui | backports.org saved my day ;) |
12:43.27 | rob_w | hmm if i take this const away .. wonder if i then still can call my log function like logerror(VALUES, siCounter, PWARN, "überschritten"); |
12:44.10 | gnif | yeah |
12:44.12 | gnif | you can |
12:44.38 | rob_w | no i cant it says warning: passing arg 4 of `logerror' discards qualifiers from pointer target type |
12:44.56 | gnif | ok, mayb thats a c++ thing |
12:45.03 | gnif | coz I do that all the time in c++ |
12:45.15 | rob_w | shit .. that s a problem now |
12:45.33 | gnif | why cant you use a const char |
12:45.35 | gnif | whats the error? |
12:46.13 | rob_w | hang on |
12:47.34 | rob_w | it segfaults on strcpy(printfmessage , message); where char *printfmessage; |
12:47.56 | gnif | ahh, did you allocate the memory in printfmessage? |
12:48.12 | gnif | memalloc(printfmessage, strlen(message)); |
12:48.12 | rob_w | not that i know ?=! |
12:48.17 | gnif | *malloc |
12:50.22 | rob_w | ehm malloc doesnt take 2 arguments !?! |
12:50.27 | gnif | :S |
12:50.47 | gnif | whoops |
12:51.00 | gnif | pfrintfmessage = malloc(strlen(message)); |
12:51.08 | rob_w | ah right |
12:51.15 | gnif | sorry, i work with about 8 languages, its easy to get mixed up |
12:51.16 | rob_w | now i get over it .. hang on |
12:51.23 | rob_w | cool |
12:51.45 | gnif | make sure you clean up after, you need to release that memory when your done with it |
12:52.23 | gnif | free(printfmessage) |
12:52.53 | rob_w | hmm |
12:54.32 | gnif | how long you been using c? |
12:54.53 | rob_w | hehe ,,, |
12:55.09 | rob_w | i never managed to get those string stuff to well |
12:55.20 | rob_w | so i never had to do this things |
12:57.26 | gnif | understandable |
12:57.34 | gnif | you might want to read up on pointers and memory management |
12:57.52 | gnif | or your apps will crash unexpectadly |
12:58.04 | gnif | and you will get gray hairs trying to figure out why |
13:04.44 | gnif | yay, its finished the locales... now itc building gcc-cross |
13:10.53 | gnif | yay, its building nano :D |
13:10.57 | gnif | all done!!!! |
13:11.50 | rob_w | good job .. your first oe image within 2 hours |
13:12.10 | gnif | mind you, an hour of that was eating dinner and driving home :P |
13:13.37 | gnif | so... "bitbake linux" will build me the kernel? |
13:14.11 | *** join/#htc-linux iggy (n=iggy@12.45.184.235) |
13:17.20 | rob_w | bitbake -b oe/packages/linux/someversion.bb |
13:18.21 | gnif | ok, so is that how I configure the kernel? |
13:18.24 | gnif | & patch |
13:19.58 | *** join/#htc-linux pH5 (n=ph5@e178223049.adsl.alicedsl.de) |
13:21.21 | gnif | there is no linux-2.6.18 :( |
13:21.30 | cr2 | gnif: are your patches for the hh.org CVS ? |
13:21.39 | gnif | no |
13:21.40 | cr2 | then i can simply commit them to the tree. |
13:21.48 | gnif | they are not my patches |
13:22.09 | gnif | the guy that made em has done STACKS |
13:22.16 | cr2 | well, then try to adapt them. |
13:22.28 | gnif | problem is, they rely on 2.6.18 stuff |
13:22.45 | cr2 | are they all magician specific ? |
13:22.50 | gnif | yes |
13:22.55 | gnif | well, sorta |
13:23.03 | gnif | lots of the pxa audio stuff |
13:23.15 | cr2 | hmm. tough luck. |
13:23.18 | gnif | and he has re-written the usb gadgets |
13:23.29 | cr2 | i hate patches. |
13:23.37 | gnif | couple thousand lines of diff |
13:23.55 | gnif | http://userpage.fu-berlin.de/~zabel/magician/patches-2.6.18-rc7-htc.tar |
13:24.05 | gnif | wait, he has updated those |
13:24.22 | gnif | http://userpage.fu-berlin.de/~zabel/magician/magician-modules-2.6.18-20060920.tar.bz2 |
13:24.29 | cr2 | you know the procedure: it has zero chance to be ever applied to something mainstream, if not split into pieces. |
13:24.42 | gnif | oh, the patches are |
13:24.56 | gnif | there is about 8-9 patches |
13:25.00 | gnif | i think |
13:25.14 | cr2 | these are binary modules. |
13:25.20 | gnif | nope |
13:25.23 | gnif | whoops |
13:25.25 | gnif | wrong link |
13:25.35 | gnif | http://userpage.fu-berlin.de/~zabel/magician/magician-zImage-2.6.18-rc7-20060919 |
13:25.38 | gnif | argh |
13:25.41 | gnif | wrong again |
13:25.47 | gnif | http://userpage.fu-berlin.de/~zabel/magician/magician-patches-2.6.18-20060920.tar.bz2 |
13:25.48 | gnif | there |
13:25.54 | gnif | i should learn to read ey :P |
13:26.24 | gnif | he has called it all by "magician" not "htc-magician" so i dont know what that will affect |
13:26.32 | cr2 | ok. that looks better. |
13:26.40 | gnif | i have a feeling the gadgets stuff is much more advanced then whats @ hh.org |
13:26.51 | gnif | and the alsa driver doesnt oops on the magician |
13:27.14 | cr2 | the machine name is "magician". it has historical roots. |
13:27.37 | gnif | yeah, i know that... i mean, i dont know what problems patching will cause |
13:28.16 | gnif | ie, the arch/arm/ dir is magican, not htcmagician |
13:28.52 | cr2 | it is called htcmagician in hh.org CVS |
13:28.58 | gnif | yeah, i know |
13:29.03 | gnif | thats what I am saying |
13:29.12 | cr2 | 704K patch, hmm. |
13:29.18 | gnif | the hh.org sources wont compile for magician, none of the lcd stuff has been updated |
13:29.28 | gnif | the structs are wrong |
13:29.31 | cr2 | it's easy. |
13:29.42 | cr2 | i will update universal today. |
13:30.09 | gnif | i fixed it all myself, then found this stuff and realised that its being maintained more actively |
13:30.16 | gnif | so dropped the hh cvs version |
13:30.24 | gnif | now i regret that decision |
13:30.56 | gnif | whats the chances of migrating hh to 2.6.18? |
13:33.20 | gnif | I would be happier using hh.org cvs... and would gladly see about moving these patches into it |
13:36.40 | pH5 | gnif: unfortunately I don't have too much time, lately. I'm going to split up magician.patch as a first step (remove the asoc sound stuff). |
13:37.34 | gnif | so do you want me to see what I can do? |
13:37.41 | pH5 | gnif: did you have problems with 2.6.18? |
13:37.45 | gnif | none |
13:38.13 | gnif | even the fast fpe works |
13:38.29 | pH5 | I didn't rewrite pxa audio - that's just the AsoC framework from the openzaurus kernels |
13:38.47 | gnif | ahhh, are you Zabel? |
13:38.49 | pH5 | yup |
13:38.53 | gnif | :D |
13:39.11 | gnif | how come these patches are seperate to hh.org? |
13:39.29 | gnif | i would be happy to help move them in |
13:39.56 | gnif | your usb gadgets code is very different |
13:40.03 | pH5 | I didn't want to depend on too many hh.org things. |
13:40.36 | gnif | well, i must say that the 2.6.18 kernel with your patches runs excellent |
13:40.37 | pH5 | The usb code is from rpsys.net/openzaurus, too. |
13:40.42 | gnif | only detected one problem |
13:41.11 | gnif | the volume/up/down (whatever key that is), up doesnt seem to be mapped |
13:41.15 | gnif | but I can look into that |
13:42.59 | pH5 | gnif: those should be mapped to KEY_VOLUMEUP/VOLUMEDOWN, maybe they don't appear in the xserver due to some 7bit limitation |
13:43.19 | gnif | nah, i havt even got xserver running |
13:43.32 | gnif | I have had real problems with my crosstool |
13:43.42 | gnif | these guys finally pointed me to OE |
13:48.56 | gnif | anyway, i am off to bed, its 11:50pm here |
13:49.34 | gnif | cya's all tmw... |
14:20.08 | WizMaui | | /usr/bin/ld:/stuff/tmp/work/x86_64-linux/qemu-qop-nogfx-native-0.8.0+svn20060921-r0/trunk/x86_64.ld:62: parse error |
14:20.08 | WizMaui | | collect2: ld returned 1 exit status |
14:20.08 | WizMaui | | make[1]: *** [qemu-arm] Fehler 1 |
14:20.10 | WizMaui | ;( |
14:20.28 | WizMaui | NOTE: package qemu-qop-nogfx-native-0.8.0+svn20060921: failed |
14:20.28 | WizMaui | ERROR: Build of opie-image failed |
14:23.02 | WizMaui | does someone have a clue whats wrong with it? |
14:23.24 | cr2 | WizMaui: this is the infamous locale problem |
14:23.29 | cr2 | ENABLE_BINARY_LOCALE_GENERATION = "0" |
14:23.45 | cr2 | in build/conf/local.conf |
14:24.35 | WizMaui | thanks ;) |
15:24.38 | *** join/#htc-linux alge (n=alge@chello080109231226.4.uni-klu.teleweb.at) |
15:32.32 | *** join/#htc-linux goxboxlive (n=jrs@9.80-202-160.nextgentel.com) |
16:02.55 | *** join/#htc-linux skodde (n=skodde@unaffiliated/skodde) |
16:54.30 | *** join/#htc-linux rob__w (n=bob@p213.54.210.106.tisdip.tiscali.de) |
17:06.45 | *** join/#htc-linux lamikr (n=chatzill@aragorn.kortex.jyu.fi) |
18:30.08 | *** join/#htc-linux skodde (n=skodde@213-140-16-178.fastres.net) |
18:38.27 | *** join/#htc-linux florian (n=fuchs@p5083B5BC.dip0.t-ipconnect.de) |
19:14.26 | *** join/#htc-linux pleemans (n=peter@d51A5E421.access.telenet.be) |
19:17.37 | *** join/#htc-linux lamikr (n=chatzill@aragorn.kortex.jyu.fi) |
19:27.43 | *** join/#htc-linux rob__w (n=bob@p213.54.210.106.tisdip.tiscali.de) |
19:58.04 | *** join/#htc-linux youpi_ya (n=ouuoou@82-47-214-98.stb.ubr11.brad.blueyonder.co.uk) |
19:58.19 | youpi_ya | good ev |
20:00.04 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.179) |
21:22.23 | *** join/#htc-linux alexey_ (n=alexey@87.240.9.73) |
21:33.03 | *** join/#htc-linux iggy (n=iggy@gentoo/developer/iggy) [NETSPLIT VICTIM] |
21:33.03 | *** join/#htc-linux lpotter (n=ljp@c210-49-127-23.rochd1.qld.optusnet.com.au) |
21:33.04 | *** join/#htc-linux Kevin2 (n=Kevin@207-237-52-30.c3-0.avec-ubr12.nyr-avec.ny.cable.rcn.com) [NETSPLIT VICTIM] |
21:33.04 | *** join/#htc-linux mndctrl (n=mind@2.81-167-1.customer.lyse.net) [NETSPLIT VICTIM] |
21:33.51 | cr2 | hi |
22:59.43 | Kevin2 | Hi |
22:59.50 | Kevin2 | cr2: Are you still around? |
23:02.06 | Kevin2 | Well, if you get this message, I put a new haret patch on svn. I can now compile HaRET with -O again. |
23:03.16 | cr2 | Kevin2: just wake up and wanted to go to bed :) |
23:07.12 | cr2 | hmm. interesting. is wince capable of handling things like that #define __irq __attribute__ ((__section__ (".text.irq"))) ? |
23:09.31 | Kevin2 | It does work. |
23:09.37 | florian | looks scary :-) |
23:09.52 | Kevin2 | It is the linker that handles that stuff. wince never sees it. |
23:10.25 | cr2 | ok. |
23:10.43 | Kevin2 | To be honest, though, I can't guarentee the next ld/gcc version wont croak on it. However, it should be something we can fix up if it does. |
23:11.02 | Kevin2 | The Linux kernel does this stuff too. |
23:13.09 | cr2 | i was worried that the wince dynamic loader does not know what to do with .text.irq section. |
23:13.58 | Kevin2 | I don't think it ever sees it. By the time the linker is done, it just tells wince to load some blob at some address. |
23:14.59 | Kevin2 | Hrmm. The name is in the section headers. But I confirmed that it does work for me. |
23:15.35 | cr2 | objdump -h |
23:16.04 | Kevin2 | Yeah, I just did that (-x). |
23:16.05 | cr2 | <PROTECTED> |
23:16.06 | cr2 | <PROTECTED> |
23:17.19 | cr2 | i have looked deeper into the phone control. |
23:17.33 | Kevin2 | If you're concerned, I might be able to reorg it so it gets merged back into .text. However I did test it without issue. |
23:17.41 | Kevin2 | Cool, anything interesting. |
23:17.58 | cr2 | this is really scary. |
23:18.14 | cr2 | .text.irq is CODE, so it will be properly handled. |
23:19.03 | cr2 | the phone control is a really complex thing. even with a lot of stuff disassembled, there is a lot of things to understand. |
23:20.54 | cr2 | the usb host control just makes it even more compicated. |
23:21.14 | Kevin2 | I'd like to see whatever descriptions you can put together for them. The various phones likely have a lot in common. |
23:21.55 | cr2 | we have 3 different "base" phones. |
23:21.56 | Kevin2 | Hrrm. So you think host is directly involved in control traffic. That sucks. |
23:22.45 | cr2 | it is difficult to separate the gpios that work for the phone and for the usb host. |
23:23.10 | cr2 | because these things happen simultaneously. |
23:23.15 | Kevin2 | BTW, I was thinking of rewriting linboot to use the section stuff also, and rewrite most of the assembler boot code in C. We really only need about 10 assembler lines - just about everything can be done from C code. |
23:23.32 | cr2 | agreed. |
23:24.12 | *** join/#htc-linux gnif (n=gnif@203.36.89.23) |
23:24.17 | gnif | hi all |
23:24.23 | cr2 | and we need much less C code. AllocPhysMem is saving us a lot of trouble. |
23:24.30 | cr2 | hi gnif |
23:24.44 | Kevin2 | I know you and ph5 got in a discussion the other day about phone tracing. wgpio isn't really useful for watching phone gpios, because it only shows a snapshot in time - it can't show ordering of gpio events. |
23:25.23 | cr2 | and my old idea of linking the zimage into haret. |
23:25.53 | Kevin2 | Unfortunately, tracing gpio activity via the "wi" stuff isn't too easy. If you watch the full range of gpio registers, you get a lot of traffic. Further, it isn't easy to tell what was being written (or read) on the gpio registers, because you don't get the register read in some cases. |
23:26.31 | cr2 | well, i have traced the pxa and asic3 gpios and disassembled a lot of wince code. |
23:26.54 | Kevin2 | We may be able to trace host activity by watching writes to the usb host registers, and then dissasembling the areas that write to it, and then setup instruction breakpoints on those areas of the code. |
23:27.15 | cr2 | i can assign most of the (phone) gpios to some wince code, but not all. |
23:27.33 | Kevin2 | I'm reluctant to get into diassembly -- I really don't have the patience for it. |
23:27.46 | cr2 | i'd like to avoid disassembling the usbh.dll maybe there is an easier way to do it ? |
23:28.54 | cr2 | i have learned to find the gpios and addresses really fast. but the full logic behind them needs much more patience. |
23:28.58 | Kevin2 | I was reading the pxa usb specs the other day. It doesn't look easy to catch in the general case. However, they may be using some simplified scheme that we could setup instruction breakpoints on. (And then code up a handler that copies what was being transfered to the "trace buffer".) |
23:29.33 | cr2 | btw, is there a gdb for wince, or something like that ? |
23:34.51 | Kevin2 | Not that I know of. The "wi" stuff is basically a homegrown gdb. |
23:37.35 | cr2 | stepi/dis ? |
23:38.36 | Kevin2 | "set insn 0xaaabbbcc" |
23:41.06 | cr2 | and setting a break address ? |
23:43.41 | Kevin2 | It will set an instruction breakpoint. When it fires, it sets an instruction breakpoint for the next instruction and resumes. When the next trap fires, it resets the break point to the first instruction and resumes. |
23:45.17 | cr2 | i'll learn the opcodes this way :) |
23:52.34 | gnif | sorry to interrupt... just a quick q... there is no support in OE for the magician, the hh.org kernels complain that its an incompatible device... what do i do about this? |
23:54.42 | cr2 | there should be a way to insert a C driver to usbh. like the snoopypro is doing it. |