# 13:12:51 |
i7 |
joins #crosstool-ng |
# 13:13:46 |
i7 |
hello, I've a problem with this crosstool. I can buil my cross compiler.. in target folder I can see some progam.. but not *-ng |
# 13:13:52 |
i7 |
can someone help me ? |
# 13:14:29 |
i7 |
for example i can see arm-cortex_a9-linux-gnueabi-gcc, arm-cortex_a9-linux-gnueabi-ar ecc |
# 13:14:48 |
kos_tom |
yes |
# 13:14:53 |
i7 |
but not arm-cortex_a9-linux-gnueabi-nm |
# 13:14:54 |
kos_tom |
those are your cross-compilation tools |
# 13:14:59 |
i7 |
sorry, *-nm |
# 13:15:33 |
i7 |
ok, i would buil -nm also, for compiling kernel, compilation fail because -nm is not found, not buiold |
# 13:15:46 |
i7 |
*not build |
# 13:16:05 |
i7 |
have I to enable something ? |
# 13:18:11 |
i7 |
when I try to compile the linux kernel, the message, after a time, is: |
# 13:18:12 |
i7 |
LD init/built-in.o |
# 13:18:13 |
i7 |
LD .tmp_vmlinux1 |
# 13:18:13 |
i7 |
KSYM .tmp_kallsyms1.S |
# 13:18:13 |
i7 |
/bin/sh: 1: arm-cortex_a9-linux-gnueabi-nm: not found |
# 13:18:13 |
i7 |
No valid symbol. |
# 13:18:13 |
i7 |
make: *** [.tmp_kallsyms1.S] Errore 1 |
# 13:29:37 |
i7 |
? any help ? |
# 13:31:46 |
i7 |
anywat, this is a very good crosstool :) |
# 13:31:51 |
i7 |
*anyway |
# 13:43:30 |
i7 |
bye |
# 13:44:09 |
i7 |
quits : Quit: Sto andando via |
# 13:51:18 |
i7 |
joins #crosstool-ng |
# 13:51:53 |
i7 |
hello everybody !! |
# 13:52:19 |
i7 |
I 've found how to create crosscompile *-nm !! |
# 13:52:30 |
i7 |
by myself :) |
# 13:52:36 |
i7 |
I'm so happy !! |
# 13:54:57 |
i7 |
I've to use gcc 4.6.3-linaro ! with gcc 4.6.3, the toolchain do not create *-nm tool |
# 13:57:13 |
kos_tom |
well, since nm is part of the binutils, I hardly see how changing the gcc version would make a difference, but.. |
# 13:58:50 |
i7 |
mm.. ok, I try to reset 4.6.3 to check if i made only this change... |
# 14:03:51 |
i7 |
I need wait 11 min each time to recompile my toolchain.. I'm doing some tests |
# 14:04:13 |
sh4rm4 |
quits : Ping timeout: 276 seconds |
# 14:04:30 |
chffelix |
joins #crosstool-ng |
# 14:06:21 |
i7 |
i'm using binutils 2.22, marked experimental.. |
# 14:09:28 |
sh4rm4 |
joins #crosstool-ng |
# 14:15:07 |
i7 |
one update, gcc 4.6.3 and linaro 4.6 do not create *-nm , gcc 4.7.0 create *-nm but fail at one final step of the buil... |
# 14:15:22 |
i7 |
*build |
# 14:16:18 |
i7 |
I'm tryng with previous version of binutils now.. |
# 15:54:46 |
claudio_ |
joins #crosstool-ng |
# 15:56:15 |
tkil |
joins #crosstool-ng |
# 15:58:23 |
chffelix |
quits : Ping timeout: 260 seconds |
# 16:03:16 |
tkil |
my "populate" script takes a fair bit of time (say, 20-30s) for a 500MB source and it only adds about 11MB to that. i experimented with using "cp -a -l" (instead of just "cp -a") but it still takes quite a while. before I go profiling it, does anyone have ideas? |
# 16:17:50 |
tkil |
ah. the cp is fast, it must be something downstream. sorry for the noise. |
# 17:02:19 |
i7 |
quits : Ping timeout: 276 seconds |
# 17:18:25 |
claudio_ |
is now known as: chffelix |
# 17:24:47 |
i7 |
joins #crosstool-ng |
# 17:28:55 |
chffelix |
hm |
# 17:29:09 |
chffelix |
maybe someone can help me in this one.. |
# 17:29:35 |
chffelix |
i just built a cross-compiler for the ep9312 |
# 17:29:56 |
chffelix |
it went fine and all.. |
# 17:30:07 |
chffelix |
but the application i compiled is complaining: |
# 17:30:08 |
chffelix |
/usr/lib/libstdc++.so.6: version `GLIBCXX_3.4.15' not found |
# 17:30:46 |
chffelix |
when i built the cross-compiler i chose the same glibc version i have on my target root filesystem |
# 17:30:58 |
chffelix |
so i expected it to be fine.. |
# 17:31:38 |
chffelix |
why is it still complaing about glibcxx mismatch? |
# 17:33:02 |
i7 |
hello, I'm not a guro, I'm known this crosstool only iesterday, by the way, |
# 17:33:30 |
i7 |
I read you must use the same compiler build with toolchian for kernel and for your app... |
# 17:33:31 |
chffelix |
great tool isn't ? |
# 17:33:48 |
i7 |
yes, I think it is a very good tool. |
# 17:34:03 |
i7 |
I have a problem, I can't crosscompile nm for arm.. |
# 17:34:25 |
i7 |
but it is also my first time |
# 17:35:14 |
i7 |
my buil crosscompile gcc, ar,as,cc etcetra but not nm and I do not understand why.. |
# 17:35:33 |
chffelix |
nm? |
# 17:35:40 |
chffelix |
i never used it.. |
# 17:35:59 |
chffelix |
was it supposed to be part of the gcc toolchain? |
# 17:36:01 |
i7 |
i need it for kernel, the kernel say me this: |
# 17:36:47 |
chffelix |
hmm. indeed it is part |
# 17:36:52 |
chffelix |
my build includes it |
# 17:36:53 |
i7 |
LD init/built-in.o |
# 17:36:53 |
i7 |
LD .tmp_vmlinux1 |
# 17:36:53 |
i7 |
KSYM .tmp_kallsyms1.S |
# 17:36:53 |
i7 |
/bin/sh: 1: arm-cortex_a9-linux-gnueabi-nm: not found |
# 17:36:53 |
i7 |
No valid symbol. |
# 17:36:53 |
i7 |
make: *** [.tmp_kallsyms1.S] Errore 1 |
# 17:37:25 |
chffelix |
did you choose a linux build as opposed to bare-metal? |
# 17:37:59 |
i7 |
but i cant configure this tool to build nm... if i choose gcc 4.7.0, it create it, but fail at the one step |
# 17:38:03 |
i7 |
linux |
# 17:38:16 |
chffelix |
gcc-4.7 is having some trouble yet |
# 17:38:35 |
chffelix |
y_morin is looking into it |
# 17:38:37 |
i7 |
yes but it is only the one that create this executable |
# 17:38:45 |
chffelix |
try gcc-4.6.3 |
# 17:38:48 |
chffelix |
it should work |
# 17:39:06 |
chffelix |
(i also tried gcc-4.7 yesterday) |
# 17:39:22 |
tkil |
chffelix -- libstdc++ is a part of g++ (hence a part of gcc), not of glibc. |
# 17:39:49 |
chffelix |
hmm.. ok tkil.. |
# 17:39:52 |
i7 |
no... I tray binutils 2.22, 2.21.53, 2.21.1a, & gcc 4.6.3 but it not create nm... |
# 17:40:10 |
chffelix |
so i should match gcc also to the version I should have in my distro? |
# 17:40:11 |
tkil |
so it'd be something wrong with your cross-compiler build (or the sysroot debacle) |
# 17:40:24 |
tkil |
no, cross toolchain has nothing to do with your distro. |
# 17:40:32 |
chffelix |
or just copy the libstdc++.so from the crosschain to the target? |
# 17:40:32 |
tkil |
let me look at my config... |
# 17:40:45 |
tkil |
it should do that for you. are you using the ...-populate script? |
# 17:40:54 |
chffelix |
(i meant my target's distro) |
# 17:41:38 |
tkil |
hm. not really sure what that means in this context. i'm using CT_CC_STATIC_LIBSTDCXX=y and successfully building a big C++ project. |
# 17:41:43 |
chffelix |
because i checked my targest's libstdc++.so |
# 17:42:10 |
chffelix |
and it has up to GLIBCXX-3.4.14 |
# 17:42:50 |
chffelix |
and when i compile with recently built GCC-4.6.3 it goes requiring GLIBCXX-3.4.15 |
# 17:42:54 |
tkil |
.... so the versions are mismatched. shouldn't happen, not if you build your app with the same compiler you build your target distribution with. |
# 17:43:02 |
tkil |
are you using any other meta-tools (yocto, buildroot, etc)? |
# 17:43:24 |
tkil |
you'll have to rebuild your target root fs with that new compiler, i suspect. |
# 17:43:38 |
chffelix |
as a matter of fact i was using a ready made root fs |
# 17:43:43 |
chffelix |
from armedslack |
# 17:43:45 |
tkil |
when i change compilers, i rebuild everything on my target root. |
# 17:44:05 |
chffelix |
yeah.. that would be really failsafe indeed |
# 17:44:07 |
tkil |
ah. that's the problem. either rebuild your root, or use the toolchain that they used to build that root. |
# 17:44:24 |
tkil |
c++ is extra sensitive to versioning like that (due to mangling etc). |
# 17:44:31 |
chffelix |
the funny thing is.. |
# 17:44:41 |
chffelix |
i'm using cmake |
# 17:44:45 |
chffelix |
for building |
# 17:44:52 |
tkil |
and c++ is still maturing, so the libstdc++ is usually tied to the compiler version. |
# 17:44:55 |
chffelix |
and i specify the sysroot to be that one |
# 17:45:00 |
chffelix |
(the same of my target) |
# 17:45:09 |
tkil |
*shrug* can't help with that, i use raw makefiles and shell scripts etc. |
# 17:45:30 |
tkil |
anyway. gotta get back to my project, if nothing else try tracing through cmake and making sure it uses the same compiler that was used to create the root. |
# 17:45:35 |
chffelix |
so i supposed it would point gcc to use my target's rootfs as sysroot instead |
# 17:45:53 |
chffelix |
ok.. thanks tkil |
# 17:46:33 |
tkil |
cross compiling is hard, let's go drinking. :) |
# 17:46:37 |
chffelix |
yeah |
# 17:46:52 |
chffelix |
hehe |
# 17:47:00 |
i7 |
tkill, do you know what influence the creation of nm tool ? |
# 17:47:12 |
tkil |
i didn't have much luck with any prebuilt distro / root / toolchain, so i just bit the bullet and used ct-ng and some shell scripts to build my own. |
# 17:47:37 |
tkil |
takes about 30min on a pretty beefy machine, but builds ct-ng suite + boost, curses, openssl, used to build apache, some performance tools (oprofile, perf, etc). |
# 17:47:41 |
tkil |
pain in the butt. |
# 17:47:52 |
chffelix |
i can imagine |
# 17:48:00 |
tkil |
i7 -- should be a part of binutils. maybe it's built and it's not copied to your target root? |
# 17:48:25 |
tkil |
oh, wait... are you using it to cross-compile the kernel and it's failing? |
# 17:48:32 |
tkil |
goes back to look at your orig question |
# 17:48:35 |
i7 |
yes yes yes |
# 17:48:41 |
i7 |
for the kernel |
# 17:49:38 |
tkil |
hm. so i have my target-nm built in the same place as gcc, ranlib, objcopy, etc... |
# 17:50:04 |
chffelix |
me also |
# 17:50:24 |
chffelix |
target-nm comes out with everybody else (gcc, ar, as.. etc) |
# 17:50:29 |
tkil |
i7 -- can you cd to where your ctng .config file is, and do: grep BINUTILS .config ? |
# 17:50:36 |
tkil |
put it on fpaste.org or wherever... |
# 17:50:56 |
i7 |
ok wait |
# 17:51:53 |
i7 |
# |
# 17:51:53 |
i7 |
# Automatically generated make config: don't edit |
# 17:51:53 |
i7 |
# crosstool-NG 1.15.0 Configuration |
# 17:51:53 |
i7 |
# Fri May 4 19:48:15 2012 |
# 17:51:53 |
i7 |
# |
# 17:51:54 |
i7 |
CT_CONFIGURE_has_xz=y |
# 17:51:56 |
i7 |
CT_CONFIGURE_has_svn=y |
# 17:51:58 |
i7 |
CT_MODULES=y |
# 17:52:00 |
i7 |
# |
# 17:52:02 |
i7 |
# Paths and misc options |
# 17:52:04 |
i7 |
# |
# 17:52:08 |
i7 |
# crosstool-NG behavior |
# 17:52:10 |
i7 |
# |
# 17:52:12 |
i7 |
# CT_OBSOLETE is not set |
# 17:52:14 |
i7 |
CT_EXPERIMENTAL=y |
# 17:52:16 |
i7 |
# CT_DEBUG_CT is not set |
# 17:52:18 |
i7 |
# |
# 17:52:20 |
i7 |
# Paths |
# 17:52:21 |
tkil |
gah, don't spam irc.... |
# 17:52:24 |
i7 |
# |
# 17:52:26 |
i7 |
CT_LOCAL_TARBALLS_DIR="${CT_TOP_DIR}/tarballs" |
# 17:52:28 |
i7 |
CT_SAVE_TARBALLS=y |
# 17:52:30 |
i7 |
CT_WORK_DIR="${CT_TOP_DIR}/targets" |
# 17:52:32 |
i7 |
CT_PREFIX_DIR="${HOME}/x-tools/${CT_TARGET}" |
# 17:52:33 |
tkil |
and especially not with all of it.... |
# 17:52:34 |
i7 |
CT_INSTALL_DIR="${CT_PREFIX_DIR}" |
# 17:52:36 |
i7 |
CT_RM_RF_PREFIX_DIR=y |
# 17:52:38 |
i7 |
CT_REMOVE_DOCS=y |
# 17:52:40 |
i7 |
CT_INSTALL_DIR_RO=y |
# 17:52:42 |
tkil |
sighs |
# 17:52:42 |
i7 |
CT_STRIP_ALL_TOOLCHAIN_EXECUTABLES=y |
# 17:52:44 |
i7 |
# |
# 17:52:46 |
i7 |
# Downloading |
# 17:52:48 |
i7 |
# |
# 17:52:50 |
i7 |
# CT_FORBID_DOWNLOAD is not set |
# 17:52:54 |
i7 |
# CT_FORCE_DOWNLOAD is not set |
# 17:52:56 |
i7 |
CT_CONNECT_TIMEOUT=10 |
# 17:52:57 |
TheHound |
i7: dude |
# 17:52:58 |
i7 |
# CT_ONLY_DOWNLOAD is not set |
# 17:53:00 |
i7 |
# CT_USE_MIRROR is not set |
# 17:53:02 |
i7 |
# |
# 17:53:04 |
i7 |
# Extracting |
# 17:53:06 |
i7 |
# |
# 17:53:08 |
i7 |
# CT_FORCE_EXTRACT is not set |
# 17:53:09 |
tkil |
anyway. my binutils-relevant bits are: http://fpaste.org/EL1J/ |
# 17:53:10 |
i7 |
CT_OVERIDE_CONFIG_GUESS_SUB=y |
# 17:53:12 |
i7 |
# CT_ONLY_EXTRACT is not set |
# 17:53:14 |
i7 |
CT_PATCH_BUNDLED=y |
# 17:53:15 |
tkil |
.... no flood prot on this channel? :( |
# 17:53:16 |
i7 |
# CT_PATCH_LOCAL is not set |
# 17:53:18 |
i7 |
# CT_PATCH_BUNDLED_LOCAL is not set |
# 17:53:20 |
i7 |
# CT_PATCH_LOCAL_BUNDLED is not set |
# 17:53:22 |
tkil |
wanders off for a while... |
# 17:53:24 |
i7 |
# CT_PATCH_BUNDLED_FALLBACK_LOCAL is not set |
# 17:53:25 |
TheHound |
we need more people with @ here |
# 17:53:27 |
i7 |
# CT_PATCH_LOCAL_FALLBACK_BUNDLED is not set |
# 17:53:28 |
i7 |
# CT_PATCH_NONE is not set |
# 17:53:30 |
i7 |
CT_PATCH_ORDER="bundled" |
# 17:53:32 |
i7 |
# |
# 17:53:33 |
tkil |
someone go wake up yann. :) |
# 17:53:34 |
i7 |
# Build behavior |
# 17:53:36 |
i7 |
# |
# 17:53:38 |
i7 |
CT_PARALLEL_JOBS=0 |
# 17:53:40 |
i7 |
CT_LOAD=0 |
# 17:53:42 |
i7 |
CT_USE_PIPES=y |
# 17:53:43 |
TheHound |
to kick this bloody annoying douche |
# 17:53:44 |
i7 |
CT_EXTRA_FLAGS_FOR_HOST="" |
# 17:53:46 |
i7 |
# CT_CONFIG_SHELL_SH is not set |
# 17:53:48 |
i7 |
# CT_CONFIG_SHELL_ASH is not set |
# 17:53:50 |
i7 |
CT_CONFIG_SHELL_BASH=y |
# 17:53:54 |
i7 |
# CT_CONFIG_SHELL_CUSTOM is not set |
# 17:53:56 |
i7 |
CT_CONFIG_SHELL="${bash}" |
# 17:53:59 |
i7 |
# |
# 17:54:01 |
i7 |
# Logging |
# 17:54:02 |
i7 |
# |
# 17:54:04 |
i7 |
# CT_LOG_ERROR is not set |
# 17:54:06 |
i7 |
# CT_LOG_WARN is not set |
# 17:54:08 |
i7 |
CT_LOG_INFO=y |
# 17:54:10 |
i7 |
# CT_LOG_EXTRA is not set |
# 17:54:12 |
i7 |
# CT_LOG_ALL is not set |
# 17:54:14 |
i7 |
# CT_LOG_DEBUG is not set |
# 17:54:16 |
i7 |
CT_LOG_LEVEL_MAX="INFO" |
# 17:54:18 |
i7 |
# CT_LOG_SEE_TOOLS_WARN is not set |
# 17:54:21 |
i7 |
CT_LOG_PROGRESS_BAR=y |
# 17:54:25 |
i7 |
CT_LOG_TO_FILE=y |
# 17:54:26 |
i7 |
CT_LOG_FILE_COMPRESS=y |
# 17:54:29 |
i7 |
# |
# 17:54:30 |
i7 |
# Target options |
# 17:54:33 |
i7 |
# |
# 17:54:34 |
i7 |
CT_ARCH="arm" |
# 17:54:36 |
i7 |
CT_ARCH_SUPPORTS_BOTH_MMU=y |
# 17:54:38 |
i7 |
CT_ARCH_SUPPORTS_BOTH_ENDIAN=y |
# 17:54:41 |
i7 |
CT_ARCH_SUPPORTS_32=y |
# 17:54:42 |
i7 |
CT_ARCH_SUPPORTS_WITH_ARCH=y |
# 17:54:45 |
i7 |
CT_ARCH_SUPPORTS_WITH_CPU=y |
# 17:54:46 |
i7 |
CT_ARCH_SUPPORTS_WITH_TUNE=y |
# 17:54:48 |
i7 |
CT_ARCH_SUPPORTS_WITH_FLOAT=y |
# 17:54:50 |
i7 |
CT_ARCH_SUPPORTS_WITH_FPU=y |
# 17:54:54 |
i7 |
CT_ARCH_SUPPORTS_SOFTFP=y |
# 17:54:56 |
i7 |
CT_ARCH_DEFAULT_HAS_MMU=y |
# 17:54:58 |
i7 |
CT_ARCH_DEFAULT_LE=y |
# 17:55:00 |
i7 |
CT_ARCH_DEFAULT_32=y |
# 17:55:02 |
i7 |
CT_ARCH_ARCH="" |
# 17:55:04 |
i7 |
CT_ARCH_CPU="cortex-a9" |
# 17:55:06 |
i7 |
CT_ARCH_TUNE |
# 17:55:08 |
i7 |
ah ok sorry, irc have cut parto f it also |
# 17:55:10 |
i7 |
ok wait |
# 17:56:02 |
i7 |
http://fpaste.org/5smD/ |
# 17:56:43 |
tkil |
i7 -- just for future reference, *never* paste more than 1-2 lines to irc. ever. |
# 17:56:48 |
i7 |
i can see you have set to use gold.. |
# 17:56:57 |
tkil |
huh. if i do, i forgot about that. :) |
# 17:56:57 |
i7 |
ok, I'm sorry |
# 17:57:03 |
tkil |
still not sure why nm wouldn't be built. |
# 17:57:39 |
i7 |
I will investigate |
# 17:57:48 |
i7 |
thanks for your time :9 |
# 17:57:57 |
i7 |
:) |
# 17:58:29 |
tkil |
check out the build log, look for errors. |
# 18:01:35 |
i7 |
wow, I've found .config on the web for my processo, I will try it :) |
# 18:05:51 |
tkil |
if you're not willing to really do a lot of plumbing, you might be better off with a prebuilt root + toolchain. codesourcery offers free (but not supported) ones, manufacturers often do as well. |
# 18:06:14 |
tkil |
but yeah, trying to build a kernel with something like that, is about when i started wanting to spin my own. :( |
# 18:06:32 |
tkil |
alternately, look at solutions (like yocto, eldk, buildroot, openembedded) that do more of it for you. |
# 18:06:44 |
tkil |
i do like ctng for a lot of things, but it is just one niche out of many. |
# 18:36:34 |
sh4rm4 |
quits : Ping timeout: 276 seconds |
# 18:49:01 |
sh4rm4 |
joins #crosstool-ng |
# 18:52:03 |
tkil |
quits : Quit: Leaving |
# 21:04:53 |
chffelix |
quits : Quit: Leaving |
# 21:27:53 |
i7 |
Hello, now I'm ready with crosscompiling |
# 21:28:14 |
i7 |
:) |
# 21:28:28 |
i7 |
*-nm is build now :) |
# 21:28:33 |
i7 |
bye |
# 21:30:56 |
i7 |
quits : Quit: Sto andando via |
# 21:30:57 |
sh4rm4 |
alternatively you could have tried to symlink you local nm to gnu-eabi-whatever-nm |
# 22:32:43 |
typn |
quits : Remote host closed the connection |