ibotlog2html for #crosstool-ng

<< Previous 2012-11-19 Next >>

# 01:19:18 boucman_work quits : Ping timeout: 264 seconds
# 01:23:28 boucman_work joins #crosstool-ng
# 01:25:39 alan_o joins #crosstool-ng
# 01:46:43 mingwandroid quits : Read error: Connection reset by peer
# 03:14:37 alan_o quits : Quit: Leaving
# 04:32:56 boucman_work quits : Ping timeout: 245 seconds
# 04:37:01 vicky joins #crosstool-ng
# 04:51:29 boucman_work joins #crosstool-ng
# 05:01:35 vicky Hi ctng, I am cross-compiling filesystem for arm9 processor using ctng toolchain in buildroot 2012.02.My host is gentoo 12.1 64 bit os.I am facing this problem "http://pastebin.com/NqUVcuLw" when compiling glibc.Any ideas?
# 06:07:32 sspiff joins #crosstool-ng
# 07:30:42 boucman_work quits : Ping timeout: 255 seconds
# 07:37:47 boucman_work joins #crosstool-ng
# 08:13:49 boucman_work quits : Ping timeout: 268 seconds
# 08:18:18 diorcety joins #crosstool-ng
# 08:20:53 boucman_work joins #crosstool-ng
# 08:25:02 boucman_work quits : Client Quit
# 08:49:08 sspiff quits : Read error: Connection reset by peer
# 09:04:31 sspiff joins #crosstool-ng
# 09:10:47 Net147 joins #crosstool-ng
# 09:11:21 codyps quits : Ping timeout: 276 seconds
# 09:44:07 codyps joins #crosstool-ng
# 09:50:33 smartin joins #crosstool-ng
# 11:56:04 vicky quits : Quit: Leaving
# 12:33:35 obiwahn what is the sysroot prefix about?
# 12:33:55 obiwahn is it the location where the cross compiler looks for missing libs?
# 12:34:15 obiwahn like /lib/arm-linux-gnueabihf
# 12:58:10 obiwahn http://paste.debian.net/210683/
# 13:00:13 diorcety paste the end of build.log
# 13:01:08 Net147 quits : Quit: HydraIRC -> http://www.hydrairc.com <- s0 d4Mn l33t |t'z 5c4rY!
# 13:01:31 obiwahn found it
# 13:01:56 obiwahn diorcety: i tried to put my sysroot to the same dir the chain is installed to
# 13:36:22 sspiff quits : Remote host closed the connection
# 13:46:32 obiwahn diorcety: i have some libs on the pi for gls and other things
# 13:46:50 obiwahn what is the best way to make them available with crosstool-ng?
# 13:48:38 obiwahn http://paste.debian.net/210699/ build.log of the failing install
# 13:49:40 obiwahn [ALL ] /bin/mkdir: cannot create directory `/home/janu/public/cross/chains/arm-pi_gcc4.6.3_test_sysr
# 13:49:43 obiwahn oot-linux-gnueabi/arm-pi_gcc4.6.3_test_sysroot-linux-gnueabi/lib': File exists
# 13:49:45 obiwahn [ERROR] make[5]: *** [install-toolexeclibLTLIBRARIES] Error 1
# 13:49:47 obiwahn [ALL ] make[5]: *** Waiting for unfinished jobs....
# 13:50:39 obiwahn (sysroot) sysroot directory name
# 13:50:50 obiwahn (/home/janu/public/cross/sysroots/) sysroot prefix dir (READ HELP)
# 13:51:04 obiwahn /home/janu/public/cross/sysroots/ <- it stoped working when setting this
# 14:44:49 sfan5|OFF is now known as: sfan5
# 15:57:18 alan_o joins #crosstool-ng
# 16:39:24 diorcety quits : Quit: Leaving.
# 16:44:21 thewonderidiot joins #crosstool-ng
# 17:10:09 diorcety joins #crosstool-ng
# 17:50:57 y_morin joins #crosstool-ng
# 17:57:49 sim hi
# 17:58:56 sim I am trying to build a x86_64 cross-toolchain (gcc 4.7.2 eglibc 2.3) with multilib support (32 and 64-bit)
# 17:59:05 sim I don't even know if it is supported
# 17:59:28 sim during the step build-libc-startfiles_32, I got the following error:
# 17:59:35 sim http://pastebin.com/iijv0h48
# 18:01:11 sim it looks like "gcc -m32" is called to compile x86_64 assembly constants
# 18:02:25 sim for example, I can see in the log file:
# 18:02:53 sim "gawk -f ../scripts/gen-as-const.awk ../sysdeps/x86_64/link-defines.sym | x86_64-lacie-linux-gnu-gcc -m32 -S ..."
# 18:03:26 sim I wonder if something could be wrong during the configure operation
# 18:05:51 sim in the config.log file, I can see that --host is set to x86_64-lacie-linux-gnu
# 18:06:16 sim is that correct while trying to build a 32-bit libc ?
# 18:07:05 sim any hints are welcome :)
# 18:09:40 thewonderidiot sim: I'm not sure crosstool-ng supports multilib
# 18:10:10 y_morin sim: multilib is marked EXPERIMENTAL, hence be ready for breakage. ;-)
# 18:13:19 sim y_morin: oh sure, I am ready for it
# 18:13:45 sim but does anyone at least succeed to build a x86_64 multilib cross-toolchain
# 18:14:23 sim or am I trying to do something which is not intended to work
# 18:15:43 sim the error I get is quite logic given the configure parameters
# 18:16:13 sim and I don't see how it can work
# 18:16:23 sim but probably I am missing something obvious
# 18:16:59 y_morin sim: IIRC, I once succeeded building a multilib toolchain, but I don't use that, I prefer multiple toolchains. So I don't remember what arch that was... :-(
# 18:18:36 y_morin sim: The issue with multi-lib toolchains is that you must *never*, ever forget to pass the appropriate flags, otherwise the system is broken (sometimes in subtle ways). And passins the correct flags is not always trivial, especially for packages that have weird build systems.
# 18:19:40 y_morin sim: At the same time, having one complete toolchain for each variant you want to addres allows you to be 100% sure that you will generate the correct instrcuctions *by default*, which is a very important point!
# 18:20:45 smartin quits
# 18:29:13 sim y_morin: ok, you convinced
# 18:29:36 sim I will use two toolchains, it makes sense
# 18:31:05 y_morin sim: however, if you really want to use a multi-lib toolchain, I won't stop you. And if you manage to make it work, be sure to share your changes! Hehe! :-)
# 18:32:20 y_morin sim: for some features that I am not interested in, I rely on other parties interested to provide the patches. So, if you have a need for multi-lib, and it works for you, that's OK with me to include changes that have a use-case in real life. :-)
# 18:34:52 sim I am not very familiar with toolchain internals and I am already on my back after this build error
# 18:35:54 sim there is not many chances I succeed to sort this out
# 18:36:24 sim but I will try a little bit harder :)
# 18:57:07 smartin joins #crosstool-ng
# 18:59:07 y_morin heads off for dinner. Groumph! :-)
# 19:08:58 thewonderidiot parts #crosstool-ng
# 19:23:02 sfan5 is now known as: sfan5|OFF
# 19:25:11 thewonderidiot joins #crosstool-ng
# 19:28:43 sfan5|OFF is now known as: sfan5
# 20:20:27 mingwandroid joins #crosstool-ng
# 20:31:24 sfan5 is now known as: sfan5|OFF
# 20:51:52 diorcety y_morin: sorry
# 20:52:12 diorcety these are my first patches to open source stuff
# 20:52:42 y_morin diorcety: hello! Why are you sorry? Those patches are pretty good! It's no problem! :-)
# 20:53:14 y_morin diorcety: don't worry about that! I did my mistakes too in the beginnings! And a lot of them...
# 20:53:37 y_morin sighs when looking back to his first endeavour in sending patches...
# 20:55:04 y_morin diorcety: I think at the beginning, people have to understand the purpose of a VCS: be able to identify a faulty change, and understand it, in the *future*.
# 20:56:02 y_morin diorcety: if one mixes up many changes in a changeset, it is not easy to understand. But if the changes are split into smaller hunks, then suddenly it makes sense!
# 21:06:57 y_morin diorcety: I just want to say that you grabbed the generated-menu concept really OK! Nice! :-)
# 21:13:50 diorcety i have to make small one ... but hard to do after all
# 21:14:46 y_morin diorcety: As you're using git, here's a tip I learned lately:
# 21:15:12 y_morin diorcety: use: "git rebase -i" to make changes in the middle of a series
# 21:15:39 y_morin diorcety: What I suggest is:
# 21:15:48 y_morin diorcety: 1) create a branch: git checkout -b darwin master
# 21:15:59 y_morin diorcety: 2) do some commits on that branch
# 21:16:17 y_morin diorcety: 3) when you want to add a change in the middle, or change an existing changeset: git rebase -i origin
# 21:16:35 y_morin diorcety: then, it spawns an editor, where you can:
# 21:16:44 y_morin diorcety: a) change the order of the changesets
# 21:16:55 y_morin diorcety: b) mark a changeset for edition
# 21:17:05 y_morin diorcety: c) merge changesets
# 21:18:26 y_morin diorcety: sonce I've been told the above (hello kos_tom!), I've seen git with whole new eyes! :-)
# 21:18:30 y_morin *since
# 21:22:31 dalias i want to become a git ninja
# 21:33:38 diorcety yes i know
# 21:33:54 diorcety i know rebase :)
# 21:38:45 diorcety it's just hard to divide stuff when you don't really know it
# 21:39:22 y_morin diorcety: right. Hence the reason I initially asked you to make your changes available for early review.
# 21:39:55 y_morin diorcety: there's another rule of thumb: always publish changes as soon as possible, to get early review, and be sure to not hit a dead-end.
# 21:40:55 y_morin diorcety: can you imagine how disapointed you could have been if, when you finally eventually your Darwin patches, 90% of them had to be rewritten?
# 21:41:59 y_morin diorcety: now that those patches (elf2flt and binutils rework) are (nearly) upstream, you get a better understanding of how things work, and how to submit. It's a net win in the end.
# 21:53:05 diorcety ok
# 21:56:31 smartin quits : Quit: good night
# 22:11:50 mingwandroid I've finally decided to learn autotools properly.
# 22:13:48 diorcety mingwandroid: ?
# 22:14:12 mingwandroid reading a book on it.
# 22:18:02 y_morin mingwandroid: when done properly, autotools are not that nasty.
# 22:18:12 y_morin mingwandroid: it requires carefulness, though...
# 22:19:40 y_morin mingwandroid: I've made a quick intrusion into autotools land when I package kconfig-frontends. It turns out that, when done properly, autotools do indeed ease packaging.
# 22:20:31 y_morin mingwandroid: But they are still lacking things, so we still have to do some things manually.
# 22:20:42 y_morin mingwandroid: What book are you reading?
# 22:21:53 y_morin mingwandroid: if you want to look, I think I have a rather clean use of autotools: http://ymorin.is-a-geek.org/hg/kconfig-frontends/
# 22:22:17 y_morin mingwandroid: Of course, its not perfect, but I believe it to be clean enough.
# 22:32:56 mingwandroid I'm reading No Starch Autotools
# 22:34:26 mingwandroid I just taught myself mostly as I went along, but generally it caters for *most* things.
# 22:35:18 mingwandroid I don't know cmake so autotools wins by default ;-)
# 22:35:38 y_morin mingwandroid: yes, please. No cmake. ;-)
# 22:36:21 y_morin mingwandroid: is that book worth the read (and price) ?
# 22:41:15 obiwahn y_morin: what goes into sysroot?
# 22:41:39 y_morin obiwahn: the sysroot contains:
# 22:41:41 obiwahn and what is the proper way to link against libs that are on the raspberry pi?
# 22:42:03 y_morin obiwahn: 1) system headers: C library headers and kernel headers
# 22:42:04 mingwandroid maybe in paperback (it's £33 for paper, £22 for ebook)
# 22:42:29 y_morin obiwahn: 2) system libraries: mostly from the C library, a few from gcc
# 22:42:55 y_morin obiwahn: crosstool-NG is *not* meant to target existing systems. It's meant to build a toolchain that will built a system *from scratch*.
# 22:43:38 obiwahn a friend copied libs from the raspberry pi to /lib/arm-linux-... of the computer where the cross compiling is done. it works but i guess it is not the intended way
# 22:43:47 obiwahn ah
# 22:43:52 obiwahn :)
# 22:44:01 obiwahn i told him it is evil
# 22:44:09 y_morin obiwahn: besides, your RPi is most probably running a Debian. On Debian, they have applied very intrusive patches to binutils, gcc and glibc, to introduce the infamous multi-arch thingy of theirs.
# 22:44:29 y_morin obiwahn: that multi-arch stuff means the linker, and the libraries, are not located in the usual places.
# 22:45:13 y_morin obiwahn: thus, if you try to run an executable built with a ct-ng toolchain, on the RPi, it will most probably *not* run, because it will not find the dynamix linker.
# 22:45:30 obiwahn mh
# 22:45:40 y_morin obiwahn: well, I do believe it's evil. A lot of people out-side Debian believe it is evil.
# 22:45:43 obiwahn so i would have to change dists?
# 22:46:25 y_morin obiwahn: however, in case of Debian on the RPi, the best bet is to follow Debian's way. And it's a tortuous way, at that.
# 22:46:50 mingwandroid (install mer on your RPi!)
# 22:46:53 obiwahn uh so i should read on multiarch debian?
# 22:46:55 mingwandroid ...sorry.
# 22:46:55 y_morin obiwahn: Sorry, I can't say for your use-case. I'm running Debian on my RPi, but I do not (yet) cross-compile for it.
# 22:46:56 obiwahn mer?
# 22:47:11 mingwandroid just something I've hacked about with in the past ;-)
# 22:47:31 y_morin mingwandroid: is that something related to Maemo, meego and that stuff?
# 22:47:55 mingwandroid ignore the upside down-ness of these vids:
# 22:47:56 mingwandroid http://www.youtube.com/watch?v=LlF95gj5kxA&feature=plcp
# 22:48:07 mingwandroid http://www.youtube.com/watch?v=zoM9a8oR-m0&feature=context-cha
# 22:48:18 mingwandroid http://www.youtube.com/watch?v=tUkkLaomApc&feature=context-cha
# 22:48:26 mingwandroid I ported mer to tegra2 last year.
# 22:50:02 mingwandroid (mostly based on the hard kernel hacking work of Eduardo José Tagle)
# 22:50:40 mingwandroid y_morin: yes. I did this stuff right when mer was being reborn.
# 22:51:21 y_morin Ok
# 22:51:45 mingwandroid https://plus.google.com/108138837678270193032/posts/SNSBFaY5AEE
# 22:52:22 mingwandroid martin brook was involved in the tegra 2 port, he's still working on Mer, I've not had time. Keep meaning to install it on my RPi though.
# 22:55:14 mingwandroid one final video of it https://plus.google.com/108138837678270193032/posts/SNSBFaY5AEE
# 22:55:42 mingwandroid check out the zoom and rotation towards the end, very cool!
# 23:06:15 mingwandroid y_morin: I don't think the book is really worth it tbh, if you know autotools quite well.. there's not much so far that I've learnt.. it's quite a big book though 364 pages so I'll let you know.
# 23:06:44 y_morin mingwandroid: Whaoo. 364 pages! That's a lot...
# 23:06:50 y_morin mingwandroid: Thanks!
# 23:24:42 y_morin quits : Quit: Nighty Night!
# 23:45:00 mingwandroid quits : Quit: Leaving.
# 23:45:19 alan_o quits : Ping timeout: 246 seconds

Generated by ibotlog2html by Yann E. MORIN