| # 00:44:02 | Buglouse | quits : Quit: Love IRC |
| # 00:52:01 | Buglouse | joins #crosstool-ng |
| # 00:59:28 | Buglouse | quits : Ping timeout: 246 seconds |
| # 01:15:49 | Buglouse | joins #crosstool-ng |
| # 01:46:36 | Buglouse | quits : Quit: Love IRC |
| # 02:07:42 | jacques_ | joins #crosstool-ng |
| # 02:07:42 | jacques_ | quits : Changing host |
| # 02:07:42 | jacques_ | joins #crosstool-ng |
| # 02:09:32 | jacques | quits : Ping timeout: 250 seconds |
| # 02:13:31 | Buglouse | joins #crosstool-ng |
| # 03:13:30 | al` | quits : Read error: Connection reset by peer |
| # 03:13:48 | al` | joins #crosstool-ng |
| # 06:46:55 | kos_tom | joins #crosstool-ng |
| # 06:57:10 | Buglouse | quits : Ping timeout: 246 seconds |
| # 07:00:10 | y_morin | joins #crosstool-ng |
| # 07:10:33 | Buglouse | joins #crosstool-ng |
| # 07:43:52 | y_morin | quits : Quit: Page closed |
| # 07:44:37 | y_morin | joins #crosstool-ng |
| # 08:25:55 | kos_tom | quits : Ping timeout: 260 seconds |
| # 08:32:59 | Buglouse | quits : Ping timeout: 252 seconds |
| # 08:41:49 | kos_tom | joins #crosstool-ng |
| # 11:18:58 | Esben | joins #crosstool-ng |
| # 12:21:56 | kos_tom_ | joins #crosstool-ng |
| # 12:23:48 | kos_tom | quits : Ping timeout: 260 seconds |
| # 13:20:45 | y_morin_ | joins #crosstool-ng |
| # 13:21:41 | kos_tom_ | quits : Quit: Leaving |
| # 13:24:01 | y_morin | quits : Ping timeout: 252 seconds |
| # 13:29:02 | daggs- | joins #crosstool-ng |
| # 13:29:10 | daggs- | hell |
| # 13:29:11 | daggs- | o |
| # 13:32:41 | daggs- | I have a question, I've been able to succsefully build a cross tool from i686 linux to x86_64 linux with gcc-4.5.2 and glibc-2.12.1, but when I compile an assembly code I get this error: Error: suffix or operands invalid for `cmpxchg'. asking google suggests incompatibility issue, any ideas? |
| # 13:50:19 | y_morin_ | is now known as: y_morin |
| # 13:54:27 | y_morin | daggs-: what binutils version? |
| # 13:55:16 | daggs- | seec |
| # 13:55:44 | daggs- | 2.20.1 |
| # 13:57:33 | y_morin | daggs-: could you pastebin your .config and a test-case, so I can try to reproduce here? |
| # 13:57:42 | daggs- | y |
| # 13:58:04 | daggs- | y_morin: I can paste the config, but the test case might be a problem... |
| # 13:58:05 | daggs- | sec |
| # 13:58:42 | y_morin | daggs-: at least pastebin the .config. If you are building a FLOSS pacakge, just tell me which one. |
| # 14:01:47 | eial | joins #crosstool-ng |
| # 14:03:06 | kos_tom | joins #crosstool-ng |
| # 14:03:48 | daggs- | quits : Ping timeout: 258 seconds |
| # 14:05:00 | eial | is now known as: daggs- |
| # 14:05:13 | daggs- | y_morin: I've passed it, thanks anyway |
| # 14:05:14 | y_morin | daggs-: from answers on google, it seems it's because your build procedure gets confused between build/host. |
| # 14:05:32 | daggs- | y_morin: actually, it is was a casting issue |
| # 14:05:49 | y_morin | daggs-: in your own code? |
| # 14:06:18 | daggs- | btw, from some reason, after build, the include folder is still empty, I had to copy the entire linux headers to that folder |
| # 14:06:32 | y_morin | daggs-: what include folder? |
| # 14:06:56 | daggs- | yes, I'm converting a product which uses cross tool from the old build |
| # 14:07:00 | daggs- | sec |
| # 14:07:57 | daggs- | y_morin: where the build.log.bz2 is created |
| # 14:08:35 | y_morin | daggs-: that's because the Linux kernel headers do not belong there. They are located in the sysroot. |
| # 14:09:27 | daggs- | y_morin: I've needed autoconf.h and for some reason, it wasn't installed even when it was in the tar |
| # 14:11:20 | daggs- | I'm usingthe hg branch |
| # 14:11:44 | daggs- | anyway, I know you need it, so here is the .config: http://dpaste.com/538046/ |
| # 14:12:08 | daggs- | it is linux 686 to linux x64-gcc-4.5.2 and glibc-2.12.1 |
| # 14:18:29 | mnt_real | joins #crosstool-ng |
| # 14:19:05 | y_morin | daggs-: OK, I'll give it a spin here. |
| # 14:20:49 | daggs- | thanks for a great tool :) |
| # 14:20:59 | y_morin | daggs-: I won't be able to reproduce the exact same toolchain, as you are using custom Linux headers. |
| # 14:21:03 | y_morin | Thx :-) |
| # 14:21:14 | y_morin | What version is it based of? |
| # 14:21:28 | daggs- | kernel version? |
| # 14:21:31 | y_morin | yes. |
| # 14:21:36 | daggs- | 2.6.32 |
| # 14:21:42 | y_morin | So I can be as close to yours as possible... |
| # 14:21:44 | y_morin | OK. |
| # 14:21:51 | daggs- | our product supports that kernel currently |
| # 14:22:06 | y_morin | Any reason why you are not using upstream headers? |
| # 14:22:28 | y_morin | There is 2.6.32.39 in crosstool-NG. |
| # 14:22:50 | daggs- | y_morin: it didn't contained autoconf.h... |
| # 14:23:28 | y_morin | autoconf.h should not be installed in the toolchain, AFAIK. But I'll check. |
| # 14:24:05 | y_morin | autoconf.h contains the configuration values for a _configured_ kernel. For the toolchain, the kernel shall *not* be configured. |
| # 14:24:15 | y_morin | And userspace should not rely on autoconf.h at all. |
| # 14:24:52 | y_morin | The only stuff that *could* use autoconf.h are out-of-tree kernel modules. |
| # 14:25:11 | y_morin | So if you are userspace that includes autoconf.h, then it is broken. |
| # 14:25:52 | y_morin | If your kernel modules depends on autoconf.h being installed system-wide, it is broken; it should use the Kbuild infrastructure that will make it available at build timl |
| # 14:25:55 | y_morin | *time. |
| # 14:28:22 | daggs- | not sure I follow |
| # 14:28:26 | y_morin | daggs-: no need for CT_CC_EXTRA_CONFIG="--with-long-double-128", there is the <*> Enable 128-bit long doubles |
| # 14:28:42 | y_morin | daggs-: autoconf.h has no place in the toolchain. |
| # 14:28:45 | daggs- | I need that file when I build the kernel of that product |
| # 14:29:03 | y_morin | daggs-: when we build the toolchain, we use the kernel headers without a configured kernel. |
| # 14:29:05 | daggs- | correction |
| # 14:29:11 | daggs- | the modules of the product |
| # 14:29:20 | y_morin | daggs-: then the modules are broken. |
| # 14:29:47 | y_morin | daggs-: the modules should use the autoconf.h from the currently configured kernel. |
| # 14:30:03 | y_morin | (assuming you're speaking _kernel_ modules) |
| # 14:30:10 | daggs- | the one on the compiling system? |
| # 14:30:26 | daggs- | these mddules are ours, we wrote them |
| # 14:31:26 | y_morin | daggs-: I understand that. I mean that when you install the toolchain, the kernel headers are installed without being configured, so that is expected that the autoconf.h file is not installed. |
| # 14:31:47 | y_morin | Then, you configure the kernel for your target, but that's not part of building the toolchain... |
| # 14:31:58 | y_morin | Then you build the kernel with this ocnfiguration... |
| # 14:32:36 | y_morin | Then you build your out-of-tree kernel modules. They shall use the autoconf.h from your kernel build, not from the toolchain (because it is not there in the first place). |
| # 14:36:24 | daggs- | I see, problem is we cannot be sure the location of that file is the same, we use different versions of linux desktop |
| # 14:36:34 | daggs- | we only need the compiler |
| # 14:36:41 | daggs- | no the kernel |
| # 14:37:04 | y_morin | daggs-: it should not depend on the machine you are building on. It should depend on where in the path you build the kernel for your target. |
| # 14:37:54 | daggs- | I understand |
| # 14:38:08 | y_morin | Just imagine that, if you change the configuration for your target kernel, you will have to change the autconf.h in the toolchain every time... |
| # 14:38:14 | daggs- | still the tagret kernel is old |
| # 14:38:31 | daggs- | it won |
| # 14:38:37 | daggs- | t be changed |
| # 14:39:46 | y_morin | daggs-: it should not matter: 2.6.32 is equally capable in this respect as more recent kernel. That specific stuff works equally for 2.6.32 as it does for 2.6.38, and even for stuff as old as at least 2.6.18 (IIRC). |
| # 14:41:59 | daggs- | y_morin: I see, my project now is to upgrade the compiler, first I finish that and then I'll see what I can do |
| # 14:42:14 | daggs- | thanks for a great tool again. |
| # 14:43:15 | y_morin | daggs-: of course, but be very carefull in this respect: adding unecessary stuff in the toolchain could well probably ruin all your efforts some time later because of this. |
| # 14:43:22 | y_morin | You're welcome! ;-) |
| # 14:44:44 | daggs- | if I'll encounter this issue, I'll know what to do :) |
| # 14:48:46 | kos_tom | quits : Ping timeout: 248 seconds |
| # 14:56:48 | daggs- | quits : Remote host closed the connection |
| # 15:05:12 | Esben | quits : Remote host closed the connection |
| # 15:26:58 | Buglouse | joins #crosstool-ng |
| # 15:38:34 | y_morin_ | joins #crosstool-ng |
| # 15:39:58 | y_morin | quits |
| # 15:51:24 | y_morin_ | is now known as: y_morin |
| # 15:52:52 | linuxjacques | joins #crosstool-ng |
| # 16:24:21 | y_morin | hides.. Thunderbolts... |
| # 16:24:45 | y_morin | quits : Quit: Back after the storm... |
| # 16:25:29 | linuxjacques | ooh, new release :-) |
| # 17:14:14 | y_morin | joins #crosstool-ng |
| # 18:20:28 | y_morin | quits : Quit: Back on-line in about 2h... |
| # 19:18:08 | y_morin | joins #crosstool-ng |
| # 20:08:13 | kos_tom | joins #crosstool-ng |
| # 20:23:54 | kos_tom | quits : Quit: Leaving |
| # 22:26:22 | y_morin | quits : Quit: Nite All! |-) |