01:05.29 | *** join/#uclibc sjhill (~sjhill@pool-151-201-232-6.pitt.east.verizon.net) |
01:49.21 | DaveM_ | quit |
02:45.27 | *** join/#uclibc mjn3 (~mjn3@cdm-68-226-136-114.laft.cox-internet.com) |
03:02.13 | *** join/#uclibc YogSothoth (~knoppix@adsl-18-27.swiftdsl.com.au) |
03:03.40 | *** join/#uclibc YogSothoth (~knoppix@adsl-18-27.swiftdsl.com.au) |
07:49.22 | *** join/#uclibc tahorg (~loic@perax.net1.nerim.net) |
14:32.28 | *** join/#uclibc dsmith (~dsmith@mail.actron.com) |
14:34.34 | *** join/#uclibc dsmith (~dsmith@mail.actron.com) |
16:05.45 | *** join/#uclibc Qui_Gon (~fox@lns-p19-8-82-65-147-85.adsl.proxad.net) |
16:29.37 | *** join/#uclibc Qui_Gonn (~fox@lns-p19-2-62-147-131-251.adsl.proxad.net) |
17:24.56 | *** join/#uclibc DaveM_ (~david@h24-207-7-221.dlt.dccnet.com) |
17:26.44 | DaveM_ | mjn3: I did a little more testing searching for duplicate files on my entire target system, not just the core |
17:26.57 | DaveM_ | mjn3: I found one interesting thing with uClibc |
17:27.04 | mjn3 | DaveM_: hey. oh? |
17:27.56 | DaveM_ | uClibc: I have an option to cross compile a native toolchain. When it gets to the stage of copying the headers i copies them out of th uClibc/include directory with make headers |
17:28.15 | DaveM_ | uClibc: This has 2 links back to the linux kernel: linux and asm |
17:28.46 | DaveM_ | linux has now has hundreds of 0 byte files from make dep, that now get copied to my target system |
17:30.22 | DaveM_ | maybe when uClibc first sets up headers it should copy them into it's directroy instead of making a link |
17:31.06 | DaveM_ | also ext2 programs are full of hard links on install, but I don't know if you have those in buildroot or not |
17:33.32 | mjn3 | DaveM_: what kernel are you using? |
17:34.22 | DaveM_ | 2.4.21 |
17:35.27 | DaveM_ | the directory is linux/modules/ , I think let me check |
17:36.05 | DaveM_ | wait I did a make clean last night, this is going to take a few min. |
17:36.20 | mjn3 | ok |
17:36.50 | mjn3 | i didn't get any work done yestderay. i took a day off and went to see www.heartofspain.com |
17:38.30 | DaveM_ | nice |
17:39.35 | DaveM_ | I keep looking at this http://www.whistlerblackcomb.com/snowreport/, waiting to take a few days off when it looks really good |
17:40.45 | DaveM_ | A little offtopic, but I thought Lousina was purchased by the US from France. When did France get it from Spain? |
17:41.02 | *** join/#uclibc TheMasterMind1 (~aman@198.38.18.22) |
17:42.10 | DaveM_ | that's the right dir. include/linux/modules/ |
17:43.05 | mjn3 | i've forgotten a lot of the history, but it was originally french, then spanish. then france got it again and sold it to the us |
17:45.12 | DaveM_ | ah, well I never got too much US history in school, just picked up little bits and pieces here and there |
17:45.46 | mjn3 | DaveM_: ok re include/linux/modules. i never noticed since i always build with kernel headers. i don't even have an include/linux/modules dir |
17:46.37 | DaveM_ | mjn3: make dep builds it |
17:47.57 | mjn3 | yeah.. i see it in a kernel tree. i don't have it in my root_fs images since i always build with buildroot using kernel-headers |
17:48.49 | mjn3 | DaveM_: i think napolean got the lousiana territory back from spain and immediately turned around and sold it to the us since he was short of cash |
17:49.59 | DaveM_ | And Cananda gave Michigan back to the US, so we could fight Napolean. It seems like the US won the Napoleonic wars, even though it was never involved |
17:51.25 | mjn3 | well, it did lead indirectly to the war of 1812 between US and GB. the brits burned down the capital |
17:51.45 | DaveM_ | I guess there are 3 solutions for the header problem. 1) make a patch for uClibc to copy the headers instead of linking 2) use kernel headers instead of a kernel tree 3) just delete the extra headers afer installing |
17:52.06 | DaveM_ | that's when we got Michigan, we gave it back in 1815 |
17:52.51 | mjn3 | DaveM_: i don't think the include/linux/modules subdir should be installed at all |
17:53.10 | DaveM_ | I think #1 is the best idea |
17:53.41 | mjn3 | the brits were defeated at the battle of new orleans in 1816 (i think)... but that was about 3 months after a treaty ending the war had been signed |
17:54.39 | DaveM_ | that was 1812 I remember learning that, but at that time the US had badly lost in the north, but England wanted an end to the war so it could fight France |
17:54.45 | mjn3 | DaveM_: yeah.. for the time being #1 is probably best |
17:55.17 | DaveM_ | I'll make a patch sometime today |
17:55.22 | mjn3 | ok |
18:06.54 | mjn3 | DaveM_: hmm... you know, i'm starting to think #3 would be fine. it would certainly be easier, and we're already (conditionally) deleting installed headers |
18:09.31 | DaveM_ | mjn3: Actually I already generated the patches. I'd just like to rebuild uClibc and test first |
18:09.38 | mjn3 | ok |
18:09.42 | DaveM_ | I also found something else odd |
18:09.54 | DaveM_ | I made a patch for that too |
18:10.39 | DaveM_ | It is accepting badly configured kernels for Arm |
18:11.40 | mjn3 | guess i'll look at your patch to see what you mean |
18:29.04 | DaveM_ | mjn3: you want me to e-mail them to the list or to you? |
18:29.50 | mjn3 | either is fine |
18:32.05 | DaveM_ | done |
18:32.18 | mjn3 | ok |
18:40.05 | mjn3 | DaveM_: the arm config this is erik's. basicly he was configuring the headers for his netwinder |
18:42.37 | mjn3 | DaveM_: am i just being slow today? i don't see how your fix_includes.sh changes will avoid copying the linux/modules stuff |
18:44.45 | DaveM_ | mjn3: it avoids linking |
18:45.44 | mjn3 | but won't it just copy the entire include/linux subtree (including the modules subdir) into uClibc/include? |
18:46.34 | DaveM_ | basicly I figured that the linux/modules is only made after "make dep", which is made after the compiler is built which is done after the uClibc sets up it's headers. At least that is the way it works on my system |
18:48.59 | DaveM_ | speaking of which, I'm trying to remember why do I need headers set up before I build a cross-compiler? |
18:49.48 | mjn3 | ah... ok. so it isn't a general solution. it could still break for, say, system-linux |
18:50.49 | DaveM_ | Guess not, I was selfishly thinking of only my own set-up |
18:51.16 | mjn3 | heh... understandable. i just thought i was missing something |
18:55.20 | mjn3 | DaveM_: i'm not sure if the headers have to be there first or not. but the way the make dependencies are set up, they have to be there before uclibc is configured, and i moved the uclibc configuration up so that i'd get prompted by it as quickly as possible |
18:57.23 | mjn3 | DaveM_: oh... erik did the same arm configuration thing in buildroot/make/kernel-headers.mk |
19:00.18 | *** join/#uclibc jc (~jcw@68.219.167.140) |
19:44.53 | DaveM_ | mjn3: that patch doesn't work very well anyways. I need to also change the rm -f's to rm -rf's now they files are directores and not links |
19:45.51 | mjn3 | i think option #3 is best. just do the install and then blow away include/linux/modules if it exists |
20:00.13 | *** join/#uclibc ade|desk (~adrian@modem-2009.giraffe.dialup.pol.co.uk) |
20:00.55 | ambassador | uh... |
20:01.00 | ambassador | does gcj work with uclib? |
20:01.14 | DaveM_ | ambassador: yes |
20:01.25 | ambassador | woohoo! |
20:01.45 | ambassador | howto enable in buildroot? |
20:01.47 | DaveM_ | ambassador: well actually for dynamic programs only (as of uclibc 0.9.21, I haven't checked lately) |
20:02.02 | mjn3 | DaveM_: really? wow. i've never tried it |
20:02.36 | DaveM_ | mjn3: works well, kind of slow though, but then again Java is slow in general |
20:02.38 | mjn3 | ambassador: hey |
20:02.46 | ambassador | mjn3: hi |
20:03.12 | ambassador | do I just add to TARGET_LANGUAGES? |
20:03.18 | ambassador | what is the gcj target name? |
20:03.25 | DaveM_ | java |
20:03.30 | ambassador | cool, thx |
20:03.39 | ambassador | here goes :-) |
20:04.17 | ambassador | :-) |
20:04.28 | DaveM_ | ambassador: just a warning, you might want to hit enter, then watch a movie or something, cause you will be waiting a long time |
20:04.57 | ambassador | hmm... currently reading Enders Game |
20:16.35 | mjn3 | ambassador: good book |
20:17.13 | ambassador | a friend sent it to me |
20:22.15 | ambassador | it's building gcj :-) |
20:22.25 | *** join/#uclibc TheMasterMind1 (~aman@h-68-166-69-153.MCLNVA23.dynamic.covad.net) |
20:28.56 | *** join/#uclibc ade|home (~adrian@modem-3246.gorilla.dialup.pol.co.uk) |
20:54.17 | ambassador | bummer: |
20:54.18 | ambassador | .libs/libgcj.la-2.o(.text+0x15c):/home/brad/src/uClib/buildroot-bdl-2/toolchain_build_mipsel/gcc-3.3.2/libjava/javax/swing/GrayFilter.java:71: relocation truncated to fit: R_MIPS_CALL16 _Jv_InitClass |
20:54.34 | ambassador | and: |
20:54.35 | ambassador | .libs/libgcj.la-2.o(.text+0x3e8):/home/brad/src/uClib/buildroot-bdl-2/toolchain_build_mipsel/gcc-3.3.2/libjava/javax/swing/AbstractAction.java:66: relocation truncated to fit: R_MIPS_GOT16 javax::swing::event::SwingPropertyChangeSupport::class$ |
20:55.12 | ambassador | (while building libgcj.so.4.0.0 |
20:55.14 | ambassador | :-( |
20:58.34 | mjn3 | yuk |
21:00.17 | *** part/#uclibc sjhill (~sjhill@65.117.135.105) |
21:01.43 | *** join/#uclibc sjhill (~sjhill@65.117.135.105) |
21:02.07 | *** part/#uclibc sjhill (~sjhill@65.117.135.105) |
21:05.19 | ambassador | I think this is the mips "large got" issue |
21:08.10 | mjn3 | ambassador: there seems to be a config options to deal with it... --with-libgcj-mips-xgot |
21:09.01 | ambassador | oooh... |
21:10.04 | mjn3 | ambassador: gcc.gnu.org/ml/java-patches/2003-q4/msg00191.html |
21:10.39 | ambassador | EXTRA_GCC_CONFIG_OPTIONS=--with-libgcj-mips-xgot ? |
21:11.26 | mjn3 | i don't know what that would do for non-mips. you might want to do add it conditionally |
21:12.39 | ambassador | oh, I only use this buildroot on mips :-) |
21:13.18 | mjn3 | ok then. let me know if it works for you |
21:13.36 | mjn3 | back later |
21:34.54 | *** join/#uclibc bug1 (~bug1@c211-28-81-139.sunsh3.vic.optusnet.com.au) |
21:35.59 | bug1 | a few bugs coming out of the woodwork today |
21:36.22 | bug1 | tar/cpio hardlinks are causing me problems as well |
22:29.10 | *** join/#uclibc ade|desk (~adrian@modem-699.gazelle.dialup.pol.co.uk) |
22:47.33 | mjn3 | ambassador: how'd it go? |
23:12.29 | Qui_Gonn | anybody has ever used linux with i2c ? |
23:12.51 | DaveM_ | Qui_Gonn: yes |
23:12.56 | kergoth | Qui_Gonn: yep |
23:13.13 | Qui_Gonn | i will make a parallel adapter |
23:13.30 | Qui_Gonn | so then how can i communicate with the i2c devices ? |
23:14.56 | DaveM_ | Qui_Gonn: Look in linux/Docu.../i2c/ and linux/drivers/i2c/ for information |
23:15.14 | Qui_Gonn | i've looked at this |
23:15.27 | Qui_Gonn | but i don't understand all the thing |
23:15.42 | Qui_Gonn | my devices will be detected automatically ? |
23:16.08 | kergoth | i'd recommend reading up on how i2c works. google is your friend. |
23:16.53 | DaveM_ | I haven't used it in about a year, but I believe there is a kernel option to have a /proc/bus/i2c that you can examine to see what is on the i2c bus |
23:17.48 | solar | uclibc site down? |
23:18.00 | Qui_Gonn | my question is that how linux knows about the adapter ? |
23:18.03 | Qui_Gonn | and devices ? |
23:18.11 | DaveM_ | solar: just slow for me |
23:18.49 | solar | I'm getting connection refused |
23:19.07 | solar | n/m it's fixed. |
23:21.02 | DaveM_ | Qui_Gonn: I think you should take kergoth's advice, and read up on i2c. It isn't that complicated of a protocol and I know at one point I found some very good docs on the web |
23:21.27 | Qui_Gonn | i've read on the web and in the src |
23:21.34 | Qui_Gonn | but i do not understand |
23:27.35 | DaveM_ | Qui_Gonn: can you ask a specific question? Those are much easier to answer. You might also want to look into a kernel mailing list |
23:28.24 | Qui_Gonn | so i would like to know how can i make the adapter detected and the devices detected |
23:29.58 | DaveM_ | What's in /proc/bus/i2c? |
23:30.22 | Qui_Gonn | for the moment nothing because i do not have the adapter |
23:30.41 | Qui_Gonn | but i would like to know the way to use it |
23:31.15 | DaveM_ | You have to compile support for it into your kernel, or use modules |
23:32.34 | Qui_Gonn | ok for the adapter i think the philips parallel port is right |
23:32.42 | Qui_Gonn | then for the devices ? |
23:35.08 | DaveM_ | if it isn't showing up in that file then there is something wrong with kernel configuration |
23:42.31 | DaveM_ | mjn3: the libgcc_s solution we came up with last week, only works for the simple case |
23:47.07 | mjn3 | DaveM_: hey.. i was gone for a while. a friend dropped by out of the blue |
23:48.57 | mjn3 | guess i'm not surprised about the libgcc_s problem |
23:51.50 | DaveM_ | oh well, I guess there is no abi comptability between 21 and 23. |
23:52.06 | mjn3 | sigh |
23:53.06 | mjn3 | erik will probably want a bug fix release out soon... sometime next week i'd guess. at least things are starting to work smoothly again |
23:54.20 | mjn3 | since i've been fighting with buildroot, the new stdio stuff will likely be in the release after that. that will break the abi again |
23:55.03 | DaveM_ | I was half way done writing a question and you answered it. Thanks |
23:55.10 | DaveM_ | you read my mind |
23:56.11 | mjn3 | i'll be doing locale changes as well, which will likely be abi-incompatible. but that stuff really needs cleaning up |
23:57.15 | mjn3 | hopefully... after the new locale stuff, stdio core, and native gettext + iconv stuff is in place, we might be in a position to start keeping the abi stable |
23:57.29 | DaveM_ | cool |
23:57.47 | mjn3 | actually, in my opinion, we'd be close to a 1.0 release. but there are still some issues |
23:57.57 | mjn3 | the headers really need auditing |
23:58.09 | mjn3 | there are problems with libm... missing functions mainly |
23:58.37 | mjn3 | it would be nice to have more arch-specific str/mem functions |
23:58.56 | mjn3 | the shared lib loader 'weak' bug needs to be stomped once and for all |