# 08:57:23 |
smartin |
joins #crosstool-ng |
# 10:04:23 |
smartin |
quits : Ping timeout: 258 seconds |
# 10:10:28 |
sh4rm4 |
joins #crosstool-ng |
# 10:57:12 |
smartin |
joins #crosstool-ng |
# 13:50:33 |
smartin |
quits : Ping timeout: 258 seconds |
# 16:53:12 |
mnt_real |
joins #crosstool-ng |
# 20:53:44 |
kugel |
joins #crosstool-ng |
# 20:53:46 |
kugel |
parts #crosstool-ng |
# 20:53:54 |
kugel |
joins #crosstool-ng |
# 20:54:04 |
kugel |
hello |
# 20:54:14 |
kugel |
can someone help me on how to use a local patch repository? |
# 21:29:59 |
kos_tom |
I can probably give some help if no-one else is around. |
# 21:30:17 |
kos_tom |
I've never used that feature in crosstool-ng, but I've heard of it, so I might try to help. |
# 21:32:31 |
kugel |
kos_tom: just wondering about the naming conventions (path |
# 21:32:38 |
kugel |
paths etc) |
# 21:33:21 |
kugel |
in the meantime I reverse-engineered it from /usr/lib/ct-ng/patches and ./build.log. there's zero information on this feature |
# 21:34:24 |
kos_tom |
Note that you must ensure that the directory containing your custom |
# 21:34:25 |
kos_tom |
patches is arranged the same way the official directory is. |
# 21:34:33 |
kos_tom |
in the help text of the LOCAL_PATCH_DIR option |
# 21:34:46 |
kos_tom |
isn't that sufficient? |
# 21:35:23 |
kugel |
not if the official repositry doesnt have patches for what I need one (eglibc ports) |
# 21:36:07 |
kos_tom |
ah, yes, that's the kind of complicated case probably. |
# 21:36:18 |
kos_tom |
I guess Yann would gladly take patches to improve the help text and/or the documentation |
# 21:36:49 |
kugel |
also I didn't find that help text, I only looked in the top option (Patches origin) |
# 22:00:42 |
y_morin |
joins #crosstool-ng |
# 22:01:24 |
y_morin |
kos_tom: heyda! :-) |
# 22:01:47 |
y_morin |
kugel: every option in the menuconfig has an associated help text. |
# 22:02:21 |
y_morin |
kugel: as for the eglibc patch directory, follow the naming of the glibc structure, that is: |
# 22:02:26 |
kos_tom |
y_morin: I keep receiving some error e-mails from your e-mail address @anciens.enib.fr |
# 22:02:45 |
y_morin |
kugel: /local/patches/eglibc-ports/2.13/ |
# 22:03:01 |
kugel |
it's 2_13 actually |
# 22:03:33 |
y_morin |
kos_tom: yeah, I know that. I'm banging on the admins so they let me tweak the config of the server. It uses some kind of black-list to detect spam, but guesses wrong most of the time... :-( |
# 22:03:38 |
y_morin |
kugel: right. |
# 22:04:31 |
y_morin |
kugel: and I got it wrong... It's: /local/patches/eglibc/ports-2_13 |
# 22:06:27 |
y_morin |
kugel: and the path you must enter in the config is: /local/patches |
# 22:06:47 |
kos_tom |
y_morin: actually, I don't think it tells me that my mail is spam, it simply timeouts. |
# 22:07:01 |
kugel |
y_morin: thanks. valuable information |
# 22:07:10 |
y_morin |
kugel: cheers! :-) |
# 22:07:21 |
y_morin |
kos_tom: gah... :-( |
# 22:07:45 |
y_morin |
fumbles in his memory to remember how to access this fscking server... :-/ |
# 22:09:20 |
kugel |
any way to configure what CFLAGS to use? |
# 22:09:50 |
kugel |
the same way as --with-XXX/--enable-YYY ? |
# 22:11:24 |
y_morin |
curses the server... It's down... :-( |
# 22:11:28 |
kugel |
or CT_TARGET_CFLAGS? |
# 22:11:38 |
y_morin |
kugel: depends on what you want to pass in the CFLAGS? |
# 22:11:55 |
kugel |
-U_FORTIFY_SOURCE |
# 22:12:11 |
y_morin |
kugel: then, there is an option in the C library sub-menu for that. |
# 22:12:27 |
kugel |
it should apply to gcc |
# 22:14:05 |
y_morin |
kugel: C-library ---> [ ] Enable fortified build (EXPERIMENTAL) |
# 22:14:26 |
y_morin |
kugel: by default -U_FORTIFY_SOURCE is passed, but if you set the above to 'y', then it is not. |
# 22:14:48 |
y_morin |
kugel: it's a C library build option, it goes into the C library sub-menu. |
# 22:15:40 |
kugel |
y_morin: I didnt enable experimintal features, this it should be off |
# 22:15:46 |
y_morin |
kugel: note that you have to enable the experimental features to see this option. If you don't enable experimental, then you won't see the option, and it default to 'n'. |
# 22:15:51 |
y_morin |
kugel: yes. |
# 22:16:26 |
kugel |
I need to pass it to gcc/binutils as well |
# 22:17:16 |
kugel |
otherwise the toolchain generates calls to symbols that are only there with _FORTIFY_SOURCE |
# 22:17:35 |
y_morin |
kugel: for binutils, it is not possible to pass specific CFLAGS for now. So it needs changes to the code. |
# 22:17:37 |
kugel |
(if I compile the target program with -U_FORTIFY_SOURCE then everything is fine again) |
# 22:18:41 |
y_morin |
kugel: ditto for gcc. |
# 22:18:52 |
kugel |
bummer |
# 22:18:59 |
y_morin |
kugel: for gcc, you may get along with using the option "Flags to pass to --enable-cxx-flags" |
# 22:19:55 |
y_morin |
kugel: you mean gcc e;its calls to 'fortified' symbols? What's the exact error message? |
# 22:20:23 |
kugel |
no error message |
# 22:20:49 |
kugel |
it just emits stuff like __open_2@@GLIBC_2.7 |
# 22:20:58 |
kugel |
(i guess the linker is doing it but anyway) |
# 22:21:27 |
kugel |
this is for a normal open() call which resolves to open@@GLIBC_2.4 with -U_FORTIFY_SOURCE |
# 22:21:56 |
kugel |
btw does ct-ng use ccache in anyway? |
# 22:22:44 |
y_morin |
kugel: ccache is known to break the toolchain build, for sure. The only workaround to fix that is to remove it entirely; I have no fix. |
# 22:23:06 |
y_morin |
kugel: and when running the toolchain proper, I have absolutely no idea on how ccache behaves... |
# 22:23:48 |
y_morin |
kugel: I'd suggest that you try without ccache first, and then enable it only if everything works. |
# 22:47:06 |
kugel |
y_morin: problem is that gcc does "#define _FORTIFY_SOURCE 2" by default |
# 22:47:10 |
kugel |
I need to disable that |
# 22:48:52 |
y_morin |
kugel: indeed, but I have no idea on why it does that. What gcc is that? |
# 22:48:56 |
y_morin |
(version) |
# 22:49:30 |
kugel |
4.4.x |
# 22:51:32 |
y_morin |
kugel: did you enable libssp? |
# 22:52:24 |
kugel |
no |
# 22:52:44 |
y_morin |
kugel: I see no place where gcc would do that. The only occurences of 'FORTIFY_SOURCE' are in libssp. And nowhere else in the code is there any reference to even just 'FORTIFY' |
# 22:53:13 |
y_morin |
kugel: care to pastebin your .config, please? |
# 22:58:46 |
y_morin |
quits : Quit: Night... |