# 00:09:14 |
wayne |
joins #crosstool-ng |
# 00:10:31 |
wayne |
is mpfr.org down. tried to do a build and the site keeps timing out. |
# 00:13:24 |
wayne |
quits : Client Quit |
# 00:19:13 |
tiredbones |
joins #crosstool-ng |
# 00:43:38 |
Net147 |
joins #crosstool-ng |
# 01:03:36 |
diorcety |
quits : Read error: No route to host |
# 01:11:16 |
al` |
quits : Ping timeout: 248 seconds |
# 01:19:43 |
Net147 |
quits : Ping timeout: 245 seconds |
# 01:20:56 |
Net147 |
joins #crosstool-ng |
# 01:37:46 |
tkil |
wayne -- it's often called "subversion" in package managers. |
# 01:38:16 |
tkil |
e.g., $ rpm -qf `which svn` => subversion-1.6.18-1.fc16.x86_64 |
# 02:16:53 |
mingwandroid |
quits : Quit: Leaving. |
# 02:19:42 |
smartin |
quits : Quit: leaving |
# 02:49:26 |
tiredbones |
thanks tkil |
# 04:02:37 |
tiredbones |
quits : Quit: Leaving |
# 05:21:22 |
ubergeek42 |
quits : Read error: Operation timed out |
# 06:40:39 |
alan_o |
quits : Quit: Leaving |
# 07:38:19 |
Net147 |
quits : Quit: HydraIRC -> http://www.hydrairc.com <- Go on, try it! |
# 07:43:48 |
Net147 |
joins #crosstool-ng |
# 08:18:12 |
mingwandroid |
joins #crosstool-ng |
# 08:22:54 |
mingwandroid |
quits : Ping timeout: 240 seconds |
# 08:28:44 |
smartin |
joins #crosstool-ng |
# 08:29:38 |
plfiorini |
joins #crosstool-ng |
# 08:29:42 |
plfiorini |
hi |
# 09:15:01 |
Net147 |
quits : Read error: Connection reset by peer |
# 09:16:50 |
sfan5|OFF |
is now known as: sfan5 |
# 09:37:22 |
diorcety |
joins #crosstool-ng |
# 10:07:40 |
Net147 |
joins #crosstool-ng |
# 10:45:40 |
y_morin |
joins #crosstool-ng |
# 12:10:29 |
mingwandroid |
joins #crosstool-ng |
# 14:01:19 |
plfiorini |
quits : Ping timeout: 272 seconds |
# 14:07:47 |
plfiorini |
joins #crosstool-ng |
# 14:11:19 |
Net147 |
quits : Quit: HydraIRC -> http://www.hydrairc.com <- Chicks dig it |
# 15:13:24 |
ubergeek42 |
joins #crosstool-ng |
# 15:27:08 |
sh4rm4 |
quits : Remote host closed the connection |
# 15:28:51 |
sh4rm4 |
joins #crosstool-ng |
# 16:20:28 |
diorcety |
quits : Quit: Leaving. |
# 16:42:11 |
mingwandroid |
y_morin: hello. |
# 16:42:32 |
y_morin |
mingwandroid: howdy? |
# 16:42:51 |
mingwandroid |
not bad. I'm hacking away at this cctools stuff again... |
# 16:43:17 |
mingwandroid |
I've got the patch sizes down a lot and am splitting them up into a sensible series. |
# 16:43:19 |
y_morin |
hehe! ;-) |
# 16:44:15 |
mingwandroid |
one thing... you remember that we agreed that it'd be better if I find the SDK headers that we need on opensource.apple.com? |
# 16:45:16 |
y_morin |
mingwandroid: yes. The idea would be to download them at runtime. |
# 16:45:40 |
y_morin |
mingwandroid: Then your scripts would do, well, whatever they ahve to do to use those headers. |
# 16:46:07 |
herdingcat |
joins #crosstool-ng |
# 16:46:15 |
herdingcat |
anyone can have a look? http://fpaste.org/nv8H/ ? |
# 16:47:28 |
y_morin |
herdingcat: what version of crosstool-NG? |
# 16:48:12 |
herdingcat |
y_morin, the latest released version |
# 16:48:22 |
y_morin |
herdingcat: which is? |
# 16:48:26 |
herdingcat |
y_morin, 1.17.0 |
# 16:48:43 |
y_morin |
OK. Lemme see one thing... |
# 16:49:08 |
herdingcat |
y_morin, thank you mate |
# 16:51:30 |
y_morin |
herdingcat: could you try with the latest version in the repository? There has been a few enhancement to that 'staic build failing' message. |
# 16:51:47 |
y_morin |
herdingcat: it should now tell you exactly what is failing (rather than this generic message). |
# 16:51:57 |
y_morin |
staic -> static |
# 16:52:11 |
herdingcat |
y_morin, sure, no problem. BTW, do I have to be root to do that? |
# 16:52:19 |
herdingcat |
y_morin, I mean ct-ng build |
# 16:52:19 |
y_morin |
mingwandroid: Any problem with the above statements? |
# 16:52:33 |
y_morin |
herdingcat: No, you should *not* be root. |
# 16:52:46 |
y_morin |
herdingcat: being root is alomost always a bad idea. |
# 16:53:14 |
y_morin |
herdingcat: And if you even have to think "Should I be root to do that?" then the answer is almost invariably "No." ;-) |
# 16:53:56 |
y_morin |
thinks 'sudo' was a very, very bad idea, after all. Especially password-less sudo... Sigh... |
# 16:54:33 |
herdingcat |
y_morin, okay, I didn;t use sudo/root |
# 16:54:46 |
mingwandroid |
y_morin: sorry, got distracted... |
# 16:55:06 |
y_morin |
herdingcat: logging in as root is also a Very Bad Idea (TM) ;-) |
# 16:55:09 |
mingwandroid |
y_morin: no that's fine and I agree with it 100% and I think we can make that work for |
# 16:55:50 |
mingwandroid |
y_morin: cctools. however, to build libgcc and crt*.o files, dylib files from Mac OSX SDK are needed at present. |
# 16:56:31 |
y_morin |
mingwandroid: Hmm, repeat that for me. What does that mean? |
# 16:57:25 |
mingwandroid |
y_morin: so cctools == apple's binutils. I'm mostly sure that these can be built fine without the Mac OSX SDK, downloading, at build time, components from opensource.apple.com.. |
# 16:57:45 |
y_morin |
mingwandroid: Very good! |
# 16:57:57 |
herdingcat |
huh? |
# 16:58:18 |
herdingcat |
After ct-ng build failed, there's no build.log in my work dir |
# 16:58:26 |
herdingcat |
It makes me confused, |
# 16:58:26 |
mingwandroid |
y_morin: however, during the build of gcc or llvmgcc, the crt files and libgcc need to link against libSystem.dylib (which shared links to 25 other dylibs) which are part of the Mac OSX SDK. |
# 16:58:56 |
y_morin |
mingwandroid: those are libs for the target, right? |
# 16:59:03 |
mingwandroid |
y_morin: yes. |
# 16:59:14 |
y_morin |
mingwandroid: and we do not have the sources for those libs |
# 16:59:31 |
y_morin |
right? |
# 16:59:47 |
mingwandroid |
y_morin: AFAIK that is the case yes. |
# 17:00:23 |
y_morin |
mingwandroid: then we have to download those pre-built libs to be able to build crt*.o and libgcc* and some other stuff? |
# 17:01:10 |
mingwandroid |
y_morin: we'd require the user to point us at their Mac OSX SDK to build gcc/llvmgcc (I'm not sure about clang/llvm as I've not looked at them yet) |
# 17:01:22 |
herdingcat |
y_morin, can you help me w/ that? After ct-ng build failed, there's no build.log in my work dir |
# 17:01:49 |
herdingcat |
y_morin, and config file is disappeared as well. |
# 17:02:09 |
y_morin |
herdingcat: just a sec... |
# 17:02:54 |
y_morin |
really can't do two IRC discussions, reg-test build the ct-ng latest code, hack on buildroot, read mails, all at the same time... |
# 17:03:05 |
y_morin |
should buy a second brain... :-] |
# 17:03:21 |
y_morin |
mingwandroid: you mean, we can't download those libs ? |
# 17:03:37 |
y_morin |
herdingcat: Just a sec, I'm preparing a test-case for you to test... |
# 17:04:19 |
herdingcat |
y_morin, okay |
# 17:04:35 |
herdingcat |
y_morin, .config also disappeared. |
# 17:04:40 |
y_morin |
also wonders what on earth is stealing ~100MiB/s bandwidth on his disks... :-( |
# 17:06:02 |
crazedpsyc |
check iotop? |
# 17:06:10 |
y_morin |
herdingcat: do you have the .config somewhere else? |
# 17:06:44 |
y_morin |
crazedpsyc: iotop requires task accounting (or some such) in the kernel, which I do not have... |
# 17:07:00 |
mingwandroid |
y_morin: AFAIK, the SDK only *seems* to be available as part of Xcode with a EULA that states that you'll only use it on "Apple Branded Hardware": |
# 17:07:11 |
mingwandroid |
"You agree not to install, use or run the SDK on any non-Apple branded computer, or to enable others to do so" |
# 17:07:14 |
mingwandroid |
http://images.apple.com/legal/sla/docs/xcode.pdf |
# 17:07:24 |
crazedpsyc |
y_morin: ah, lame |
# 17:07:49 |
herdingcat |
y_morin, no |
# 17:08:03 |
herdingcat |
y_morin, after ct-ng build .config gone |
# 17:08:07 |
herdingcat |
y_morin, what magic? |
# 17:08:30 |
y_morin |
mingwandroid: Ah, this is a PITA... We have no choice but to let the user enter the PATH to his/her SDK, then... |
# 17:09:16 |
y_morin |
herdingcat: Can you check in the install prefix of the toolchain, if there is an executable named *-ct-ng.config (or smthg like that) |
# 17:09:47 |
huli_ |
joins #crosstool-ng |
# 17:10:01 |
mingwandroid |
y_morin: I'd still quite like to allow people who want to to build cctools without forcing an SDK to be identified? I don't know, but I suspect that llvm/clang may be buildable without requiring these dylibs to be present... |
# 17:10:30 |
y_morin |
mingwandroid: Well, if the libs are optional, then we *do* want to build without them. |
# 17:10:44 |
mingwandroid |
y_morin: and llvmgcc/gcc are being deprecated by llvm/clang anyway. |
# 17:10:59 |
y_morin |
mingwandroid: if they are a nice-to-have, we can later add an option to use them. Optionally. |
# 17:11:30 |
mingwandroid |
y_morin: ok, so proceed with cctools downloading headers at build time, then when it comes to gcc/llvmgcc later we'll have to add a configure argument for the user to identify the SDK? |
# 17:11:42 |
y_morin |
herdingcat: the toolchain is by default install in ${HOME}/x-tools/tuple/ |
# 17:12:12 |
y_morin |
herdingcat: tuple being replaced by the toolchain tuple, of course (arm-unknown-eabi in your case, I think) |
# 17:12:30 |
y_morin |
mingwandroid: Yes, sounds good to me. :-) |
# 17:13:18 |
herdingcat |
quits : Ping timeout: 264 seconds |
# 17:14:02 |
huli_ |
y_morin, noticed the issue, fixed seems |
# 17:14:10 |
huli_ |
y_morin, Retrieving needed toolchain components' tarballs |
# 17:14:21 |
huli_ |
y_morin, I think it should be in the doc |
# 17:14:26 |
y_morin |
huli_ == herdingcat ? |
# 17:14:32 |
huli_ |
y_morin, agreed. |
# 17:14:37 |
mingwandroid |
y_morin: ok. cool. so far, I've found 95% of the headers needed... there are some however, that are procedurally generated from some .defs files... I may need to compile the generator program for those :-( |
# 17:14:39 |
huli_ |
nods |
# 17:15:34 |
y_morin |
Got to go for a little while... Bak in < 10 min |
# 17:20:58 |
y_morin |
huli_: See: http://crosstool-ng.org/download/ibot-logs/ <-- there are few questions for you at the end of the log... |
# 17:23:29 |
y_morin |
huli_: Sorry, will give corect url in a sec... |
# 17:24:06 |
y_morin |
huli_: http://crosstool-ng.org/download/ibot-logs/today.html#quick-17:09:16 |
# 17:24:23 |
y_morin |
huli_: http://crosstool-ng.org/download/ibot-logs/today.html#quick-17:11:42 |
# 17:27:15 |
huli_ |
y_morin, thanks mate. |
# 17:27:51 |
huli_ |
y_morin, btw, I encounter this error when downloading: Not at this location: "http://www.mpfr.org/mpfr-3.1.0//mpfr-3.1.0" |
# 17:27:59 |
huli_ |
y_morin, I cannot access the url |
# 17:29:54 |
y_morin |
huli_: Could that be that MPFR changed their layout?... Checking... |
# 17:30:28 |
y_morin |
Hmm... Looks like mpfr.org is down... :-( |
# 17:31:05 |
y_morin |
huli_: Can you access mpfr.org from a browser? |
# 17:31:20 |
huli_ |
y_morin, cannot |
# 17:31:30 |
huli_ |
y_morin, so can I use Fedora rpm package? |
# 17:31:52 |
y_morin |
huli_: Nope. Distro's packages are already patched. ct-ng wants the upstream versions... |
# 17:31:53 |
mingwandroid |
mpfr.org is often down eh? |
# 17:32:19 |
y_morin |
huli_: I'll put a copy on the crostool-ng.org mirror. Just a sec... |
# 17:32:58 |
huli_ |
y_morin, thanks |
# 17:33:45 |
y_morin |
huli_: actually, it's already there. |
# 17:34:14 |
huli_ |
y_morin, how can I download it? |
# 17:34:24 |
y_morin |
huli_: In the ct-ng menuconfig: Paths and misc options --> [*] Use a mirror |
# 17:34:26 |
huli_ |
y_morin, and after download when I shall put ? |
# 17:34:39 |
y_morin |
huli_: the default location is already setup to use crosstool-ng.org as the mirror. |
# 17:34:39 |
huli_ |
y_morin, reconfigure? |
# 17:35:09 |
y_morin |
huli_: jsut run: ct-ng menuconfig, and navigate to: Paths and misc options, the select [*] Use a mirror |
# 17:35:16 |
y_morin |
*then select... |
# 17:36:03 |
huli_ |
y_morin, done, do I need to select Only use mirror (NEW) ? |
# 17:36:15 |
y_morin |
huli_: Nope. |
# 17:36:31 |
y_morin |
huli_: ct-ng will try the mirror if it can not download from the upstream sources. |
# 17:36:39 |
huli_ |
y_morin, okay, then save config? |
# 17:36:47 |
huli_ |
y_morin, run ct-ng build? |
# 17:37:09 |
y_morin |
huli_: and the mirror (crosstool-ng.org) is hosted gratis so I'd rather not we use too much bandwidth from it. |
# 17:37:31 |
y_morin |
huli_: Yes, "exit and save config", then ct-ng build |
# 17:37:54 |
huli_ |
y_morin, thanks, running now |
# 17:43:01 |
diorcety |
joins #crosstool-ng |
# 17:51:08 |
huli_ |
y_morin, http://fpaste.org/Oc8a/ |
# 17:52:21 |
y_morin |
huli_: care to pastebin your .config, please? |
# 17:52:46 |
huli_ |
y_morin, http://fpaste.org/dqo1/ |
# 17:54:02 |
y_morin |
huli_: OK, I'll give it a spin here. |
# 17:54:41 |
y_morin |
huli_: so, you're using 1.17.0. What was the reason for your 'static build not possible' ? |
# 18:02:12 |
huli_ |
y_morin, I didn't install glibc-static rpm package |
# 18:02:30 |
y_morin |
huli_: So, the mew message is much more helpfull? |
# 18:02:36 |
y_morin |
mew -> new |
# 18:02:46 |
huli_ |
y_morin, sorry? |
# 18:03:04 |
huli_ |
y_morin, you mean I need to install GMP? |
# 18:03:22 |
y_morin |
huli_: no. |
# 18:03:34 |
y_morin |
huli_: how did you find that you needed to install glibc-static? |
# 18:04:51 |
huli_ |
y_morin, http://comments.gmane.org/gmane.comp.gcc.cross-compiling/15226 |
# 18:05:12 |
huli_ |
y_morin, I installed gmp-devel and gmp-static, seems fixed. |
# 18:05:55 |
y_morin |
huli_: No, that's *not* the solution. ct-ng should *not* need the host packages for gmp/mpfr et al. |
# 18:06:32 |
huli_ |
y_morin, hmm |
# 18:06:40 |
y_morin |
huli_: besides, if your distor uses a different version or a patched version, then you risk running into runtime issues when using the toolchain, and those will be much harder to debug. |
# 18:06:58 |
huli_ |
y_morin, okay... |
# 18:07:05 |
huli_ |
y_morin, so what's your idea? mate? |
# 18:07:08 |
y_morin |
huli_: I'm test-running your .config here to see what's wrong |
# 18:07:33 |
y_morin |
But my machine is currently very very loaded, and the build is a bit slow... |
# 18:07:45 |
huli_ |
y_morin, thanks mate |
# 18:09:12 |
y_morin |
huli_: what's your host distro? |
# 18:11:01 |
huli_ |
y_morin, Fedora 16 x86_64 |
# 18:11:31 |
y_morin |
huli_: OK, I'll see if I can setup a VM with an FC16... |
# 18:12:16 |
huli_ |
y_morin, great, thanks mate |
# 18:12:28 |
y_morin |
has found what is causing disk slowdowns: the RAID-5 array is being checked... |
# 18:12:47 |
huli_ |
y_morin, I do really hope Fedora can include this great package |
# 18:12:59 |
y_morin |
huli_: PPL builds just fine here on my Debian... |
# 18:13:14 |
alan_o |
joins #crosstool-ng |
# 18:13:16 |
y_morin |
huli_: so, it has something to do with FedoraCore... :-/ |
# 18:14:11 |
huli_ |
y_morin, okay.. |
# 18:14:35 |
huli_ |
y_morin, I gonna use ct-ng to create toolchian then I can compile many programs for my rpi |
# 18:14:59 |
y_morin |
huli_: Ah, OK! :-) |
# 18:15:23 |
huli_ |
y_morin, it tool about 12 hrs to compile OpenMPI in RPi. |
# 18:15:43 |
y_morin |
huli_: then, you may want to wait a few hours, I'll push a new sample to build a hard-float toolchain for the RPI. ;-) |
# 18:16:06 |
huli_ |
y_morin, the current one is not hfp? |
# 18:16:09 |
y_morin |
huli_: And your current .config won't be able to build for the rpi/ |
# 18:16:19 |
huli_ |
y_morin, why? |
# 18:16:46 |
huli_ |
y_morin, I have kernels so what I wanna do is just compile programs |
# 18:17:17 |
y_morin |
huli_: Ah, sorry, I misread the .config. |
# 18:17:48 |
y_morin |
huli_: but still, your toolchain will not be optimised for the rpi. You forgot to specify the arch and cpu. |
# 18:18:13 |
huli_ |
y_morin, it's okay, I will just use march and mcpu for gcc |
# 18:18:16 |
y_morin |
huli_: and by default, I think gcc outputs armv5 opcodes, while rpi os armv6 |
# 18:18:34 |
y_morin |
huli_: why do it for every package, when you can setup the toolchain to do it by default? |
# 18:19:20 |
huli_ |
y_morin, sorry? |
# 18:19:25 |
huli_ |
y_morin, I don;t understand. |
# 18:20:02 |
y_morin |
huli_: 1) the default cpu/arch variant for gcc is armv5. |
# 18:20:13 |
huli_ |
y_morin, because cross-compile is quite fast in my i7 laptop |
# 18:20:16 |
y_morin |
huli_: 2) you can configure gcc to use a different default |
# 18:20:37 |
y_morin |
huli_: 3) this can be done with ct-ng by entering appropriate valuse in the menuconfig |
# 18:21:00 |
huli_ |
y_morin, for 3) what I shall add for gcc? |
# 18:21:17 |
huli_ |
y_morin, and is that for hard float? |
# 18:21:27 |
y_morin |
huli_: just a sec... |
# 18:21:36 |
huli_ |
y_morin, sure |
# 18:22:59 |
y_morin |
huli_: http://code.bulix.org/aqeh07-82790 |
# 18:23:15 |
huli_ |
y_morin, what;s that? |
# 18:23:34 |
y_morin |
huli_: that's my .config for a toolchain that targets the RPI. |
# 18:23:49 |
y_morin |
huli_: optimised for armv6, and hardfloat. |
# 18:24:04 |
huli_ |
y_morin, what about mine? |
# 18:24:42 |
y_morin |
huli_: just diff the two files, you'll see the differences... ;-) |
# 18:25:04 |
y_morin |
huli_: and I know mine is working ;-) |
# 18:25:06 |
huli_ |
y_morin, mine is hard-float |
# 18:25:26 |
y_morin |
huli_: diff the two files, you'll see the delta. ;-) |
# 18:26:09 |
y_morin |
huli_: also, I'm using the latest content from the repository. |
# 18:26:55 |
y_morin |
got to go for real life matters. Probably back ~21:00 UTC... |
# 18:27:38 |
huli_ |
y_morin, yes, I noticed that |
# 18:28:00 |
huli_ |
y_morin, mate, just one last question, my toolchain is using hard-float, right? |
# 18:28:26 |
huli_ |
y_morin, CT_ARCH_FLOAT="hard" |
# 19:19:21 |
smartin |
quits : Quit: leaving |
# 19:20:35 |
codyps |
quits : Ping timeout: 248 seconds |
# 20:25:57 |
Net147 |
joins #crosstool-ng |
# 20:26:33 |
Net147 |
quits : Client Quit |
# 20:42:52 |
codyps |
joins #crosstool-ng |
# 20:49:11 |
diabel |
hello all |
# 21:06:37 |
sfan5 |
is now known as: sfan5|OFF |
# 21:11:51 |
devcoder |
joins #crosstool-ng |
# 21:29:42 |
plfiorini |
quits : Ping timeout: 272 seconds |
# 21:35:28 |
plfiorini |
joins #crosstool-ng |
# 23:06:49 |
diabel |
y_morin: I've got new errors |
# 23:12:47 |
diorcety1 |
joins #crosstool-ng |
# 23:12:47 |
diorcety |
quits : Read error: Connection reset by peer |
# 23:14:03 |
diorcety1 |
quits : Read error: Connection reset by peer |
# 23:50:49 |
devcoder |
quits : Quit: devcoder |