# 00:17:40 |
y_morin |
quits : Quit: Nightty Night! |
# 01:38:04 |
mingwandroid |
quits : Remote host closed the connection |
# 02:09:41 |
diorcety1 |
joins #crosstool-ng |
# 02:12:39 |
diorcety |
quits : Ping timeout: 272 seconds |
# 03:55:06 |
djerome |
quits : Remote host closed the connection |
# 04:02:05 |
mayron |
joins #crosstool-ng |
# 04:14:05 |
mayron |
quits : Remote host closed the connection |
# 07:55:17 |
diorcety1 |
quits : Quit: Leaving. |
# 08:58:34 |
philenotfound |
quits : Quit: leaving |
# 09:09:41 |
philenotfound |
joins #crosstool-ng |
# 10:02:47 |
aalv |
joins #crosstool-ng |
# 10:26:41 |
mingwandroid |
joins #crosstool-ng |
# 10:54:09 |
dzan__ |
quits : Ping timeout: 272 seconds |
# 12:08:07 |
dzan__ |
joins #crosstool-ng |
# 12:39:34 |
Net147 |
joins #crosstool-ng |
# 12:43:31 |
loide |
joins #crosstool-ng |
# 12:56:38 |
dzan__ |
mingwandroid, hi :) |
# 12:57:03 |
dzan__ |
mingwandroid, i was debugging the issue a bit and it's at scripts/build/libc/glibc-eglibc.sh-common: linescripts/build/libc/glibc-eglibc.sh-common: line 106scripts/build/libc/glibc-eglibc.sh-common: line 106 106 |
# 12:57:38 |
dzan__ |
it looks for gcc multilibs=( $("${CT_TARGET}-gcc" -print-multi-lib) ) do build libc |
# 12:57:43 |
dzan__ |
but shouldn't it look for clang? |
# 13:04:42 |
mingwandroid |
dzan__: well, I don't know if you can build *glibc with clang easily, when I build a clang-y toolchain I also build GCC. |
# 13:04:55 |
mingwandroid |
dzan__: did you get my build.log? |
# 13:05:26 |
dzan__ |
mingwandroid, downloading as we speak :) |
# 13:05:33 |
dzan__ |
i'll try building gcc with |
# 13:05:50 |
dzan__ |
mingwandroid, when should CT_TARGET be set? it's not in the config rigth? |
# 13:06:16 |
mingwandroid |
CT_ARCH, CT_VENDOR, CT_OS AFAIR |
# 13:14:29 |
Net147 |
quits : Quit: HydraIRC -> http://www.hydrairc.com <- IRC with a difference |
# 13:58:46 |
An_Ony_Moose |
joins #crosstool-ng |
# 13:58:59 |
An_Ony_Moose |
oh! There's been a new release? |
# 13:59:37 |
fauno |
and my patch couldn't make it :( |
# 14:00:35 |
An_Ony_Moose |
I've intended to submit a patch for quite a while now but I can't remember how x) |
# 14:18:08 |
fauno |
An_Ony_Moose: wait a sec |
# 14:18:30 |
fauno |
An_Ony_Moose: https://sourceware.org/ml/crossgcc/2014-01/msg00042.html |
# 14:44:16 |
Martell |
joins #crosstool-ng |
# 14:50:41 |
dzan__ |
mingwandroid, you were rigth, when building gcc along it works |
# 14:51:00 |
dzan__ |
mingwandroid, should be technically able to do without but ok :) |
# 14:51:16 |
An_Ony_Moose |
weeeeeelll that failed! |
# 14:51:24 |
dzan__ |
mingwandroid, when compiling with the resulting toolchain: hello.c:1:10: fatal error: 'stdio.h' file not found |
# 14:51:27 |
An_Ony_Moose |
segfault on "Installing final compiler" |
# 14:51:28 |
An_Ony_Moose |
:( |
# 14:51:44 |
dzan__ |
gheh An_Ony_Moose what segfaulted? |
# 14:52:24 |
mingwandroid |
dzan__: hmm, ok. I'll need to do some patching for that then :-( |
# 14:52:46 |
dzan__ |
mingwandroid, you rock! :-) |
# 15:01:31 |
dzan__ |
mingwandroid, if you point me to the issue maybe I can solve it for you |
# 15:03:20 |
mingwandroid |
dzan__: thanks ;-) dirocety did most of the clang bits though to be honest. |
# 15:03:25 |
Martell |
quits : Ping timeout: 272 seconds |
# 15:04:18 |
dzan__ |
mingwandroid, he's not on irc then? If I can help to get this issue fixed i'm all set |
# 15:05:06 |
An_Ony_Moose |
dzan__: gcc I think. I was stupid enough to relaunch it before recovering build.log though >_> |
# 15:05:12 |
mingwandroid |
dzan__: Well, there's various patches to aid the finding of the headers. AFAIR, Clang is never told of the actual sysroot folders that we specify to GCC/*glibc. |
# 15:05:54 |
mingwandroid |
dzan__: there's e.g. crosstool-ng/patches/clang/head/140-default-gcc-paths.patch |
# 15:06:00 |
An_Ony_Moose |
anyway, trying again with 1.19.0 (was on 1.18.0) and all packages updated |
# 15:06:06 |
dzan__ |
An_Ony_Moose, enable build step saving :-) and debug output then just execute what ct did manually with vallgrind |
# 15:06:48 |
dzan__ |
mingwandroid, i'l have a look at that patch |
# 15:07:01 |
dzan__ |
mingwandroid, iirc there is a flag to make clang show where he's looking right? |
# 15:07:10 |
mingwandroid |
just -v |
# 15:07:15 |
An_Ony_Moose |
dzan__: where are those options? |
# 15:07:26 |
mingwandroid |
compare and contract with gcc -v hello-world.c for example. |
# 15:08:21 |
dzan__ |
An_Ony_Moose, if you use nconfig on your config file the first entry (paths & misc ) has debug options you can enable |
# 15:08:22 |
mingwandroid |
dzan__: other thing to do is to use CT_DEBUGGABLE_TOOLCHAIN=y .. I added that. |
# 15:08:44 |
An_Ony_Moose |
dzan__: thanks |
# 15:08:46 |
dzan__ |
mingwandroid, what does that do/enable? |
# 15:08:54 |
mingwandroid |
"-O0 -ggdb", no stripping. |
# 15:08:57 |
An_Ony_Moose |
ugh why must compiling all the tools take so long though :( |
# 15:09:18 |
dzan__ |
An_Ony_Moose, that why those steps are useful :) you can restart from where it crashed |
# 15:12:48 |
dzan__ |
mingwandroid, http://paste.debian.net/77399/ guess line 12 & 13 are the issue? :p |
# 15:13:15 |
mingwandroid |
dzan__: yeah, now it's ringing some bells! |
# 15:14:15 |
mingwandroid |
dzan__: so we added Prefix calculation patches (search for "Driver::Prefix" or "Prefix" in the src or patches) but they don't get hit when doing C Preprocessor AFAIR .. |
# 15:14:47 |
mingwandroid |
dzan__: now, it's been a week or two since I looked at it and will be a few days before I can get back to it. |
# 15:16:40 |
mingwandroid |
dzan__: you may be able to pass -isysroot on the clang commandline (but I don't recommend that as it's a Darwin GCC 4.2.1 option that we ported across to Clang) |
# 15:17:39 |
dzan__ |
mingwandroid, ok i'll try the isysroot for now, ping me if there would be a solution :-) |
# 15:17:49 |
dzan__ |
mingwandroid, is it a problem I have a custom 'triple' name? the vendor is set |
# 15:21:09 |
dzan__ |
mingwandroid, passed the prefixdir or the ct build as isysroot without result ( just letting you know ) |
# 15:29:31 |
An_Ony_Moose |
dzan__: I don't want "Debug crosstool-ng", right? |
# 15:30:16 |
dzan__ |
An_Ony_Moose, if you want to save the intermediate steps you do |
# 15:30:23 |
dzan__ |
iirc |
# 15:30:33 |
geiser |
joins #crosstool-ng |
# 15:30:33 |
dzan__ |
enabling it will show you the other option |
# 15:30:37 |
An_Ony_Moose |
aah yes |
# 15:30:37 |
An_Ony_Moose |
thanks |
# 15:31:21 |
tagpin |
joins #crosstool-ng |
# 15:33:43 |
An_Ony_Moose |
Is there a way to display all the config options that have been changed from their initial values? |
# 15:37:26 |
mingwandroid |
dzan__: I think someone is going to have to roll their sleves up and get stuck into solving the bug properly to be honest with you. |
# 15:38:09 |
dzan__ |
mingwandroid, that sucks, I don't have time for that :p maybe have to switch to building without ct then for now.. |
# 15:38:22 |
dzan__ |
An_Ony_Moose, yes do a diff of the .config against .config.old |
# 15:38:24 |
mingwandroid |
dzan__: we checked as far as cross Darwin working correctly and maybe native Linux and that it compiles on Windows but not really much beyond that. |
# 15:38:49 |
dzan__ |
mingwandroid, ok I see :) |
# 15:39:02 |
mingwandroid |
dzan__: I think you'll probably run into the same issue, that Clang isn't very good at searching for headers. |
# 15:39:45 |
dzan__ |
mingwandroid, so clang/llvm isn't used a lot atm for cross compiling? but I thought google's ndk for android used it by default now |
# 15:40:50 |
An_Ony_Moose |
dzan__: and for the config you get when you simply run ct-ng menuconfig and exit? |
# 15:41:18 |
dzan__ |
An_Ony_Moose, yes it should create an .config.old |
# 15:41:26 |
dzan__ |
but i'm always using nconfig instead of menuconfig so not sure |
# 15:41:40 |
mingwandroid |
yeah, but they don't build with weird sysroot folder that ctng uses in general: |
# 15:42:04 |
mingwandroid |
e.g. I have prefix/x86_64-apple-darwin10/sysroot/include .. |
# 15:42:08 |
An_Ony_Moose |
dzan__: but if I go into menuconfig multiple times it will replace the config.old won't it? |
# 15:42:22 |
mingwandroid |
now if I had: prefix/x86_64-apple-darwin10/include then it'd probably be fine. |
# 15:42:33 |
An_Ony_Moose |
Oh wait I can use ct-ng saveconfig and just look at the file that creates |
# 15:42:35 |
mingwandroid |
you may want to try to specify CT_SYSROOT as "." instead of sysroot, maybe. |
# 15:42:52 |
mingwandroid |
.. it's a bit of a hack |
# 15:43:27 |
mingwandroid |
CT_FORCE_SYSROOT=y |
# 15:43:27 |
mingwandroid |
CT_USE_SYSROOT=y |
# 15:43:27 |
mingwandroid |
CT_SYSROOT_NAME="sysroot" |
# 15:46:13 |
shnyap |
joins #crosstool-ng |
# 15:51:33 |
tagpin |
quits : Ping timeout: 272 seconds |
# 15:51:33 |
geiser |
quits : Ping timeout: 272 seconds |
# 15:53:07 |
shnyap |
quits : Ping timeout: 260 seconds |
# 15:53:56 |
dzan__ |
mingwandroid, thanks i'll try that but I won't get the results today anymore :p took 2 hours :p |
# 15:55:19 |
Martell |
joins #crosstool-ng |
# 15:56:41 |
mingwandroid |
maybe CT_SYSROOT_NAME="./." or something, i seem to remember that some specific values are disallowed! |
# 15:56:53 |
mingwandroid |
but those restrictions could be removed too. |
# 16:07:49 |
dzan__ |
mingwandroid, trying as we speak i'll let you know |
# 16:08:42 |
dzan__ |
Sysroot name contains forbidden slash(es): './.' ;-) |
# 16:09:51 |
dzan__ |
Sysroot name is set to '.' which is forbidden |
# 16:10:14 |
mingwandroid |
urgh, remove those checks I reckon! |
# 16:11:07 |
dzan__ |
doing so :) |
# 16:40:36 |
An_Ony_Moose |
aaaaand the error didn't happen this time OH WELL |
# 16:57:22 |
jopaweb |
joins #crosstool-ng |
# 17:08:27 |
jopaweb |
quits : Ping timeout: 252 seconds |
# 17:17:46 |
y_morin |
joins #crosstool-ng |
# 17:33:24 |
memleak |
joins #crosstool-ng |
# 17:34:31 |
memleak |
hello! i'm using make 3.82 trying to compile glibc 2.18 for arm and i'm getting a "cannot remove /usr/include/limits.h: permission denied" y_morin i believe you found a fix for this? |
# 17:35:09 |
memleak |
using the latest hg checkout of "default" branch, head is at "adding patch for glibc 2.10.1 for make 3.82 and later" |
# 17:38:47 |
memleak |
glibc log is at pastebin.com/pRPPWac2 |
# 17:44:44 |
y_morin |
memleak: No, that's a different issue. I don't know why/how it happens, I never had it on my machine... |
# 17:44:50 |
y_morin |
memleak: what's your host distro? |
# 17:46:22 |
memleak |
sadly it's archlinux |
# 17:46:33 |
memleak |
i should really move back to gentoo |
# 17:47:18 |
y_morin |
memleak: Yes, IIRC, the other reports about this /usr/include failure all occured on Arch. |
# 17:47:41 |
y_morin |
But as I do not have Arch here (only Debian or Ubuntu), I can't test. |
# 17:49:46 |
memleak |
ok back to gentoo! thanks y_morin not a problem, ill make the switch |
# 17:49:54 |
memleak |
this distro sucks |
# 17:50:26 |
fauno |
haha |
# 17:50:45 |
fauno |
i us ct-ng on parabola with the same toolchain from arch and it doesn't throw this error |
# 17:50:49 |
fauno |
use* |
# 17:51:11 |
fauno |
not for arm though |
# 18:00:18 |
memleak |
is crosstool-ng stable enough to be used for a complete distribution?? |
# 18:00:46 |
memleak |
i have an ARM board and i rather cross compile everything to it than compile natively and i could see crosstool-ng saving me a lot of time |
# 18:01:50 |
y_morin |
memleak: Yes, I use it all the time to generate ARM (mostly) and MIPS (a bit) compelte setups (from kernel to userland) with Buildroot. |
# 18:01:55 |
y_morin |
memleak: works like a charm. |
# 18:02:10 |
y_morin |
memleak: What board is it you're using? |
# 18:02:37 |
memleak |
cubieboard2 with an allwinner a20 cortex a7 MP dual core processor |
# 18:03:48 |
memleak |
cross compiling with an AMD FX-4350 4.2 GHz quad core so its a lot faster :P |
# 18:03:53 |
y_morin |
memleak: That should be pretty easy to setup. |
# 18:04:02 |
memleak |
it is :) |
# 18:04:44 |
memleak |
ive got u-boot a custom kernel and all kinds of stuff running on it right now, just wanted to make sure i can build more than firmware and kernels |
# 18:05:54 |
y_morin |
memleak: I know some guys even go as far as building Qt, Xorg and other large userspace with toolchains built with ct-ng. :-) |
# 18:06:26 |
memleak |
perfect that's what i need! |
# 18:12:17 |
diorcety |
joins #crosstool-ng |
# 18:16:08 |
memleak |
y_morin, for the cortex-a7 optimization you think i should put it under mtune or mcpu? |
# 18:17:07 |
Martell |
quits : Ping timeout: 272 seconds |
# 18:17:16 |
y_morin |
memleak: Both. :-p |
# 18:17:30 |
memleak |
you can do that? |
# 18:17:56 |
y_morin |
memleak: -mcpu tells what actual CPU to emit instructions for; -mtune tells what CPU to order intructions for. |
# 18:18:08 |
y_morin |
memleak: man gcc <- will tell you you can. ;-) |
# 18:18:57 |
memleak |
i read the official gcc documentation actually but it said mtune can have better performance than mcpu under some circumstances, didn't know you could use both though |
# 18:20:01 |
memleak |
one other thing it doesn't talk about is the vfpv4-neon option. the vfpv4 unit it has is d16, if i use vfpv4-neon will it use vfpv4-d32 or d16 with neon instructions? |
# 18:20:11 |
y_morin |
memleak: They are not exclusive. You can specify both. Although if you provide the same value to both, then -mtune is not needed. |
# 18:21:45 |
y_morin |
memleak: Not sure about neon... :-/ |
# 18:23:15 |
aalv |
parts #crosstool-ng |
# 18:25:08 |
memleak |
well if you don't specify -d16 or -fp16 for -mfp will it compile by default 32? |
# 18:25:57 |
memleak |
(hinting that vfpv4 is basically vfpv4-d32 even though vfpv4-d32 isn't a specific flag) |
# 18:27:03 |
memleak |
i've been wondering about all that for years and the docs aren't clear at all about it |
# 18:31:16 |
alan_o |
joins #crosstool-ng |
# 18:32:11 |
y_morin |
memleak: I don't know. I sually let gcc decide, and it has always worked so far... |
# 18:32:16 |
y_morin |
*usually |
# 18:32:51 |
memleak |
ok :) |
# 18:33:23 |
y_morin |
memleak: But if you know your FPU, then you may want to specify it explicitly. YMMV, as they say... ;-) |
# 18:35:00 |
memleak |
wow i had to dig through the gcc source to find this out! |
# 18:36:08 |
memleak |
gcc/config/arm/arm.c about 800 lines down, vfpv4 is d32 (as i assumed) and neon-vfpv4 is d32 (hoping it was d16 but it is rather d32) so for my a7 i really shouldn't be using vfpv4-neon considering the FPU only supports d16 |
# 18:36:46 |
memleak |
yet the kernel loaded... i guess ARM processors can still understand and emulate D32 even if they don't have it? |
# 18:38:17 |
y_morin |
memleak: the kernel does not use floating points. |
# 18:38:44 |
y_morin |
memleak: The real test is with floating point in userland. |
# 18:38:56 |
memleak |
ah so basically my distro would have been broken :P |
# 18:39:54 |
memleak |
i wish there was a vfpv4-neon-d16 xD |
# 18:40:09 |
loide |
quits : Read error: Connection reset by peer |
# 18:40:35 |
y_morin |
memleak: I guess your best bet would be with vfpv4-d16, no? |
# 18:41:39 |
memleak |
you are correct i'm just saying truly ideally |
# 18:46:42 |
An_Ony_Moose |
y_morin: is this useful? http://soundray.org/linus/casio-sample.tar.gz |
# 18:47:41 |
y_morin |
An_Ony_Moose: What's in there? |
# 18:48:27 |
y_morin |
Ah, a sample... |
# 18:48:58 |
y_morin |
An_Ony_Moose: Well, you can submit it, yes! :-) |
# 18:49:22 |
An_Ony_Moose |
How? Can I just say "I hereby submit the sample I just posted a link to"? :P |
# 18:49:50 |
y_morin |
An_Ony_Moose: Send a patch to the list. |
# 18:50:08 |
An_Ony_Moose |
do I have to do mercurial stuff for that? |
# 18:50:22 |
An_Ony_Moose |
Because I don't have mercurial and can't *really* be bothered to install it... |
# 18:51:03 |
loide |
joins #crosstool-ng |
# 18:51:05 |
y_morin |
An_Ony_Moose: On principle, yes. But if you just send the tarball as-is, with a Signed-of-by: Your Name " in the mail body, I'll add it myself. |
# 18:51:40 |
y_morin |
An_Ony_Moose: "Signed-of-by" is described there: http://elinux.org/Developer_Certificate_Of_Origin |
# 18:52:04 |
y_morin |
"Signed-off-by", even... |
# 18:52:23 |
memleak |
y_morin, do you plan on ever switching to git? |
# 18:54:07 |
y_morin |
memleak: Yes, I had hoped to do it before Christmas, but I'm late on it (need to upgrade the VM first) |
# 18:54:24 |
An_Ony_Moose |
Um, the crossgcc mailing list? |
# 18:54:28 |
y_morin |
An_Ony_Moose: Yes. |
# 18:55:05 |
y_morin |
An_Ony_Moose: As instructed on: http://crosstool-ng.org/#contacts |
# 18:56:25 |
memleak |
i take it the mailing list used to get spammed a lot judging from this line here: "You are a nice girl from Russia (or wherever), and you have pictures of your big breast to show me." |
# 18:57:09 |
y_morin |
memleak: Not really, I picked that from somewhere else, and since it made me laugh, I replicated it. :-) |
# 18:57:40 |
memleak |
heh :) |
# 18:58:44 |
An_Ony_Moose |
shent! |
# 18:59:16 |
y_morin |
An_Ony_Moose: Thanks! |
# 19:00:32 |
dvaerghoffmann |
joins #crosstool-ng |
# 19:00:48 |
dvaerghoffmann |
y_morin: Here. I'am |
# 19:00:49 |
dvaerghoffmann |
:) |
# 19:01:25 |
y_morin |
dvaerghoffmann: :-) |
# 19:01:31 |
dvaerghoffmann |
tried a grep -i usleep .config within my ct-ng profile |
# 19:01:34 |
dvaerghoffmann |
no result |
# 19:03:03 |
y_morin |
dvaerghoffmann: That's because it's a uClibc option, not a ct-ng one. |
# 19:03:35 |
y_morin |
dvaerghoffmann: And it's removed if you don't set UCLIBC_SUSV3_LEGACY |
# 19:03:40 |
dvaerghoffmann |
ahhh |
# 19:04:10 |
dvaerghoffmann |
is it the same like 'Use obsolte features'? |
# 19:05:05 |
y_morin |
dvaerghoffmann: No, that is a crostool-NG option. You need to first configure your uClibc manually, then configure crostool-Ng to use the uClibc's .config file you generated. |
# 19:05:12 |
dvaerghoffmann |
i see |
# 19:05:28 |
y_morin |
GTG for dinner. Back later tonight (GMT+1) |
# 19:06:36 |
y_morin |
An_Ony_Moose: Just a quicky before I go for dinner: your sample has not yet arrived on the list. Don;t post HTML mail, it's gonna be rejected by the list server. |
# 19:06:42 |
y_morin |
Now, I'm off... |
# 19:13:38 |
blueness |
quits : Read error: Connection reset by peer |
# 19:24:00 |
dvaerghoffmann |
parts #crosstool-ng |
# 19:39:06 |
mingwandroid |
quits : Remote host closed the connection |
# 19:40:20 |
forsergey |
joins #crosstool-ng |
# 19:45:31 |
forsergey |
quits : Remote host closed the connection |
# 19:59:04 |
sergeypronin |
joins #crosstool-ng |
# 20:00:15 |
mingwandroid |
joins #crosstool-ng |
# 20:00:33 |
borisprivet |
joins #crosstool-ng |
# 20:13:25 |
borisprivet |
quits : Ping timeout: 246 seconds |
# 20:13:26 |
smartin_ |
joins #crosstool-ng |
# 21:22:16 |
loide |
quits : Quit: Saindo |
# 21:25:01 |
loide |
joins #crosstool-ng |
# 21:28:40 |
loide |
quits : Client Quit |
# 21:50:49 |
mingwandroid_ |
joins #crosstool-ng |
# 21:53:09 |
mingwandroid |
quits : Ping timeout: 248 seconds |
# 21:53:49 |
sergeypronin |
quits : K-Lined |
# 22:11:40 |
memleak |
hey y_morin you seem like a really cool guy |
# 22:12:42 |
y_morin |
Thanks! :-) |
# 22:12:55 |
memleak |
that' |
# 22:12:55 |
memleak |
s like the 10th time i saw someone k-lined |
# 22:13:48 |
memleak |
so we're friends? :) |
# 22:14:22 |
y_morin |
K-lined? |
# 22:15:06 |
memleak |
yeah basically just an IP-address ban from freenode |
# 22:15:28 |
memleak |
i'm in a few channels and everyone is getting k-lined like crazy, and now this channel too |
# 22:17:44 |
y_morin |
Ah, wikipedia says what k-lined means. Did not know... |
# 22:25:11 |
blueness |
joins #crosstool-ng |
# 22:30:03 |
memleak |
i got k-lined once for trolling thats why i know |
# 22:30:15 |
memleak |
IRC and thirteen year olds don't mix |
# 22:34:03 |
An_Ony_Moose |
quits : Ping timeout: 252 seconds |
# 23:30:48 |
y_morin |
quits : Quit: nighty Night! |