**** BEGIN LOGGING AT Fri Jun 14 02:59:58 2013 Jun 14 14:13:04 I log it here for Gnutoo Jun 14 14:13:05 http://git.kernel.org/cgit/utils/kernel/kexec/kexec-tools.git/commit/purgatory/Makefile?id=b4f9f8599679259a471f3430f11966c6db78ad3d Jun 14 14:13:17 the strip is wrong. I'll test evening Jun 14 22:36:42 hi GNUtoo-x60 Jun 14 22:37:08 I've pushed temporary here Jun 14 22:37:10 https://github.com/andrea-adami/meta-initramfs Jun 14 22:37:59 there are many underlying issues..I'm discussing in #klibc Jun 14 22:38:20 s/discussing/learning Jun 14 22:38:21 ;) Jun 14 22:38:45 ok Jun 14 22:40:17 yeah that's legible ;) Jun 14 22:40:17 I see the problem Jun 14 22:40:18 btw do _not_ pass -Os for klibc Jun 14 22:40:18 not passing any -O at all will do at *least* -Os and maybe even better Jun 14 22:40:18 it uses "-e purgatory_start" Jun 14 22:40:18 so it's a broken binary, one that cannot work with klibc because it initialises of its own Jun 14 22:40:20 it uses -nostdlib anyway Jun 14 22:40:22 so you shouldn't need to use klibc on it, you can just continue to use gcc -nostdlib -e … Jun 14 22:40:24 it's probably a bootloader binary blob or something like that, not a unix c programme Jun 14 22:40:26 bin/bin-to-hex purgatory < purgatory/purgatory.ro > kexec/purgatory.c Jun 14 22:40:27 ok Jun 14 22:40:28 yeah… Jun 14 22:40:30 it's "data payload" Jun 14 22:40:32 also make sure to *not* use klcc for purgatory/*.o Jun 14 22:40:34 also make sure to *not* use klcc for purgatory/*.o Jun 14 22:40:36 for CC too Jun 14 22:40:40 for .c -> .o that is Jun 14 22:40:42 stuff like -o purgatory/purgatory.o purgatory/purgatory.c Jun 14 22:40:44 that cannot work with klcc because it's not targetting unix Jun 14 22:40:46 hm I probably need to adopt dietlibc's hack to prevent that… Jun 14 22:40:48 just do NOT use klcc for it Jun 14 22:40:50 see, it was all ok until 2.0.2 .. the binaries even working (the kexec part) Jun 14 22:40:52 │*kexec │ 84448│Jun 14 23:30 Jun 14 22:40:54 just stop overriding CC for purgatory Jun 14 22:40:56 only override it for things that actually run as unix userspace Jun 14 22:41:23 log I get is http://sprunge.us/GCZT Jun 14 22:55:05 ok Jun 14 22:55:28 ah ok Jun 14 22:55:38 that sounds easier Jun 14 22:55:49 usually gcc without -shared or such thing Jun 14 22:55:56 expects a main Jun 14 22:56:09 so just put the right option to make a .o Jun 14 22:56:14 maybe -static? Jun 14 22:56:37 I don't know well what purgatory means Jun 14 22:56:42 if it has a special gcc meaning Jun 14 22:56:48 like crt.0 Jun 14 22:56:51 or something like that Jun 14 22:56:57 I don't remember the exact name Jun 14 23:02:33 anyway that purgatory_patch is noy yet ok Jun 14 23:02:47 I'll retry letting only LD Jun 14 23:03:03 btw, in the log ^^ Jun 14 23:03:06 gcc -isystem/oe/oe-core/build/tmp-eglibc/sysroots/i686-linux/usr/include -O2 -pipe -o bin/bin-to-hex util/bin-to-hex.c Jun 14 23:03:10 gcc ? Jun 14 23:04:13 hmmm Jun 14 23:04:29 the lack of main is because you don't ask gcc the proper format Jun 14 23:04:37 just try to fix that because it's easy Jun 14 23:05:54 I don't remember how to create .o Jun 14 23:06:02 which are objects Jun 14 23:06:19 and I must go to bed soon Jun 14 23:06:23 sorry Jun 14 23:06:52 I have a feet in the shower... Jun 14 23:07:00 cu Jun 14 23:07:25 ok Jun 14 23:07:35 anyway articles presenting the ELF format should have that info Jun 14 23:07:54 evil ninary, isn't? Jun 14 23:07:58 *binary Jun 14 23:08:16 why evil? Jun 14 23:08:33 elf has various sub-formats Jun 14 23:08:36 like the .so Jun 14 23:08:38 the .o Jun 14 23:08:41 the executable Jun 14 23:08:43 and so on Jun 14 23:09:22 I mean kexec(purgatory) Jun 14 23:09:40 with one side in userspace and the other in kernel Jun 14 23:09:45 see this Jun 14 23:09:47 http://lists.infradead.org/pipermail/kexec/2013-June/009166.html Jun 14 23:10:38 ok Jun 14 23:10:52 imagine... Jun 14 23:11:19 I've never used the crash/kdump part Jun 14 23:11:58 I ahve only tested the kexec part w/out caring about the other kernel loaded in memory Jun 14 23:14:24 hmmm Jun 14 23:14:30 yes Jun 14 23:14:38 maybe I should try crash/kdump Jun 14 23:14:55 because I've a problem with my T60 laptop Jun 14 23:15:07 no serial yet Jun 14 23:15:14 and no logs on usb-serial Jun 14 23:15:31 and USB debug is already gone once booted Jun 14 23:15:39 and the problem appears once fully booted Jun 14 23:18:24 http://www.linuxjournal.com/googlesearch?s=ELF Jun 14 23:18:28 it has the articles on ELF Jun 14 23:18:37 I must find the good one Jun 14 23:19:24 use -c Jun 14 23:19:29 try that Jun 14 23:20:52 from http://www.linuxjournal.com/article/1060 Jun 14 23:27:26 the explanation of -c is also in man gcc and matches Jun 14 23:34:21 I've some little progress Jun 14 23:34:25 pushed a commit Jun 14 23:34:44 ok Jun 14 23:34:49 I'll switch computer Jun 14 23:39:32 now it is this Jun 14 23:39:34 http://sprunge.us/IKFU Jun 14 23:40:30 ok, the main error was fixed, good Jun 14 23:40:34 I have to fix the missing prototypes Jun 14 23:40:38 yes Jun 14 23:40:47 still Jun 14 23:40:50 cannot find entry symbol purgatory_start; defaulting to 00000000 Jun 14 23:40:54 is scary :/ Jun 14 23:40:57 ah ok Jun 14 23:41:06 iirc it's an x86 stuff Jun 14 23:41:25 ok Jun 14 23:41:35 I'll follow tomorrow Jun 14 23:41:41 too late now Jun 14 23:41:53 gn Jun 14 23:42:43 http://buildlogs.pld-linux.org/index.php?dist=th&arch=ppc&ok=1&name=kexec-tools&id=5618d124-83b0-422d-8ac2-8a87997d0e32 Jun 14 23:42:49 seem normal.... Jun 14 23:42:55 ok, good night Jun 14 23:43:00 ouch it's 01:42 **** ENDING LOGGING AT Sat Jun 15 02:59:58 2013