| # 06:30:48 | smartin | joins #crosstool-ng |
| # 06:35:50 | mnt_real | quits : Remote host closed the connection |
| # 07:13:17 | kos_tom | joins #crosstool-ng |
| # 11:23:59 | y_morin | joins #crosstool-ng |
| # 12:52:56 | mnt_real | joins #crosstool-ng |
| # 16:00:36 | y_morin_ | joins #crosstool-ng |
| # 16:18:22 | smartin | quits : Ping timeout: 252 seconds |
| # 16:23:08 | y_morin | quits : Ping timeout: 252 seconds |
| # 16:25:51 | y_morin_ | is now known as: y_morin |
| # 16:35:35 | kos_tom | quits : Ping timeout: 260 seconds |
| # 16:43:24 | y_morin | is now known as: y_morin|away |
| # 17:19:49 | kos_tom | joins #crosstool-ng |
| # 18:39:09 | y_morin|away | is now known as: y_morin |
| # 18:58:07 | smartin | joins #crosstool-ng |
| # 18:58:28 | y_morin | is now known as: y_morin|away |
| # 19:19:36 | y_morin|away | is now known as: y_morin |
| # 19:30:34 | y_morin | is now known as: y_morin|away |
| # 19:36:00 | smartin | quits : Ping timeout: 252 seconds |
| # 20:20:35 | y_morin|away | is now known as: y_morin |
| # 20:28:26 | sh4rm4 | when i tried to build a toolchain with musl libc, it failed because of a missing "link.h". that one seems to come with glibc |
| # 20:29:33 | sh4rm4 | i wonder if whatever gcc is trying to build there can be avoided / circumvented |
| # 20:34:08 | kos_tom | sh4rm4, could you post your .config and put the full build log somewhere and post a mail on the mailing list ? |
| # 20:34:17 | kos_tom | or at least pastebin your build log and config, and post the URL here |
| # 20:35:49 | sh4rm4 | kos_tom, i wasnt using ct-ng |
| # 20:35:57 | sh4rm4 | since there is no template for musl |
| # 20:36:29 | kos_tom | ah, since this channel is named #crosstool-ng, I was assuming that all discussions are related to ct-ng |
| # 20:36:40 | sh4rm4 | also i prefer gcc 3.4.6 |
| # 20:37:08 | kos_tom | ouch, gcc 3.4.6 is a prehistoric thing |
| # 20:37:16 | sh4rm4 | because i can build it on 64 mb ram machine |
| # 20:37:32 | sh4rm4 | while gcc 4.5 eats up as much as 1.1 gb while linking |
| # 20:38:48 | sh4rm4 | and takes 4 times as long to compile ... |
| # 20:39:20 | kos_tom | I guess you should do cross-compiling. |
| # 20:40:03 | sh4rm4 | just to use gcc 4 ? |
| # 20:43:43 | sh4rm4 | i ended up compiling it in a chroot using --with-sysroot |
| # 20:43:59 | sh4rm4 | all statically linked |
| # 20:44:07 | sh4rm4 | on 32bits |
| # 20:44:14 | sh4rm4 | works well on 64bit as well |
| # 20:44:48 | sh4rm4 | the --target option was causing the xgcc invocation, which caused the problem |
| # 20:45:14 | sh4rm4 | btw: the entire toolchain is only 6 MB .xz archive |
| # 20:46:17 | sh4rm4 | a simple select server compiles to a 2.5 KB statically linked binary |
| # 20:46:36 | sh4rm4 | while with gcc 4.5 it's 9 KB |
| # 20:48:07 | y_morin | sh4rm4: as kos_tom said, this channel is dedicated to crosstool-NG discussions. If you are trying to add musl support in crostool-NG, then you are welcome to discuss here. Otherwise, there are better channels to ask for help. I'm afraid people here do not have much musl knowledge (except that it is an umpteenth C library). |
| # 20:49:13 | sh4rm4 | y_morin, i would like to add musl support |
| # 20:49:26 | sh4rm4 | thats why i am trying to understand the issue i was having |
| # 20:49:39 | y_morin | sh4rm4: Good! |
| # 20:50:12 | y_morin | sh4rm4: right, but I'm afraid none here will be able to help you get musl built. When you know that, we can help integrating in crosstool-NG, of course! :-) |
| # 20:50:38 | sh4rm4 | i read something about that on the gcc bugtracker... apinski was basically telling that "glibc is required to cross-compile" |
| # 20:50:49 | sh4rm4 | however that doesnt make much sense to me |
| # 20:51:01 | sh4rm4 | (talking about the "link.h" issue) |
| # 20:51:46 | y_morin | sh4rm4: whatever link.h is, if gcc needs it, then the C library should porovide it. If it does not, this is a bug in the C library. |
| # 20:51:48 | sh4rm4 | my guess was that you guys ran into that issue as well and solved it somehow |
| # 20:52:43 | y_morin | sh4rm4: from what you said earlier, link.h seems to be provided by the C library. gcc must be using that to build target libraries (eg. libstdc++) |
| # 20:53:28 | y_morin | sh4rm4: so link.h must be installed by the C library when doing the headers+start_files pass. |
| # 20:55:01 | sh4rm4 | it was building some "unwind" stuff |
| # 20:56:06 | sh4rm4 | which is obviously not required normally, because it builds fine with --disable-bootstrap but without --target |
| # 20:59:42 | sh4rm4 | maybe it could work with this patch ... http://old.nabble.com/PATCH%3A-Include-%3Clink.h%3E-only-if-USE_PT_GNU_EH_FRAME-is-defined-td29645209.html |
| # 21:00:54 | y_morin | sh4rm4: try with gcc-4.6, it already has bionic support, so the link.h issue could go away with musl as well... |
| # 21:01:26 | sh4rm4 | y_morin, yeah, but i dont want to use that slow bloated beast |
| # 22:40:52 | kos_tom | quits : Remote host closed the connection |
| # 23:36:38 | y_morin | quits : Quit: See ya |