**** BEGIN LOGGING AT Sun May 05 02:59:58 2013 May 05 19:03:38 GNUtoo-x60: hi May 05 19:03:55 hi May 05 19:04:00 can you repost your latest log with debug? May 05 19:04:05 yes May 05 19:04:06 seems paste is expired May 05 19:04:13 ah ok May 05 19:04:19 sorry I'm not at home.... May 05 19:04:27 ok May 05 19:04:37 I'm at home now so I'll read chat May 05 19:04:39 and the kernel is on my build laptop May 05 19:04:54 np, paste it when you will have time and possibility May 05 19:04:58 ok May 05 19:05:02 thanks a lot for the help May 05 19:05:12 I remember it a bit May 05 19:05:33 I'm interested what happened there :) May 05 19:05:33 basically it said that init (kexecboot) exited with return value 0000ff00 May 05 19:05:49 looks like exit(-1) or something like May 05 19:05:51 the target device is a lenovo X60 with coreboot May 05 19:05:57 yes, exactly May 05 19:06:20 I suspect GUI initialization code May 05 19:06:21 I've also a very bad patch for vfscanf May 05 19:06:26 me too May 05 19:06:33 I've the i915 driver May 05 19:07:08 do you have other framebuffer apps working? May 05 19:07:16 I didn't try May 05 19:07:23 e.g. zgv (image viewer) May 05 19:07:36 I tried only framebuffer console May 05 19:07:43 successfull? May 05 19:07:46 yes May 05 19:07:49 ok May 05 19:08:57 I may ask you for fbset -v -s -i output May 05 19:09:41 if problem is with GUI init code May 05 19:09:50 I can't do it now May 05 19:09:57 I booted that way: May 05 19:10:10 coreboot->seaBIOS->ipxe May 05 19:10:19 which requires my build laptop May 05 19:10:24 I'll get home soon May 05 19:10:26 and try it May 05 19:10:27 ok, np May 05 19:17:09 I'll go May 05 19:17:10 bye May 05 20:34:30 hi Jay7 ant_home May 05 20:34:45 Jay7, http://paste.debian.net/hidden/6e24b30b/ May 05 21:05:12 hi GNUtoo May 05 21:06:22 ant_home, hi, do you have any idea of why it returns? May 05 21:06:27 graphics can't be displayed? May 05 21:06:47 2) btw I've made a patch for baseof, so it's easy to replace May 05 21:06:57 I'll try to replace vfscanf in a good way May 05 21:07:03 instead of an hackish way May 05 21:07:31 about vfscanf maybe we get it upstream in the next release of klibc May 05 21:07:38 maybe... May 05 21:07:51 ok May 05 21:08:00 all the rest is replaced cleanly May 05 21:08:08 it's only vfscanf that is unclean in my tree May 05 21:08:16 in th emeantime I've disabled kexec-klibc for x86 and x86_64 May 05 21:08:22 yes I saw May 05 21:08:37 and I don't like that patch: it can be fixed by implementing vfscanf May 05 21:08:38 I've now to filter May 05 21:09:01 NOTE: Resolving any missing task queue dependencies May 05 21:09:01 ERROR: Nothing PROVIDES 'initramfs-kexecboot-klibc-image' May 05 21:09:01 ERROR: initramfs-kexecboot-klibc-image was skipped: incompatible with machine qemux86 (not in COMPATIBLE_MACHINE) May 05 21:09:16 that's linux-yocto-tiny-kexecboot now... May 05 21:09:26 yes forget about theses patches May 05 21:09:34 and let's fix vfscanf May 05 21:09:36 it's an unfortunate chain of incompatibilities May 05 21:09:51 because once it's fixed we would have to revert all your patches anyway May 05 21:10:03 well, do you have an idea about that EDD stuff? May 05 21:10:10 can we revert that patch? May 05 21:10:36 I don't have any vague idea about the matter... May 05 21:10:55 afair it was for some laptop May 05 21:11:03 we implement vfscanf like it was advised to do in klibc mailing list.... May 05 21:11:24 it's really easy May 05 21:12:07 hm..remember there are no floats in klibc and va_list .. May 05 21:12:21 could contain floats in theory May 05 21:12:36 http://www.zytor.com/pipermail/klibc/2013-April/003419.html May 05 21:12:52 it compiles if I implement it the dirty way May 05 21:13:00 I'll implement it in a more clean way May 05 21:13:01 ah, yes, you mean rewriting the patch May 05 21:13:17 I'v done thatfor ubi-utils May 05 21:13:17 I mean replacing vfscanf with: May 05 21:13:32 fgets May 05 21:13:34 vsscanf May 05 21:13:38 in kexecboot's code May 05 21:13:42 oops May 05 21:13:46 in kexec's code May 05 21:14:01 then we put the patch in oe May 05 21:14:04 and that's all May 05 21:14:06 we'd appreciate very much that patch May 05 21:14:18 yes but I did it when it was so late.... May 05 21:14:20 can we test it on qemu? May 05 21:14:25 so I'll rewrite it cleanly May 05 21:14:29 because I did that: May 05 21:14:30 yes, pls do May 05 21:15:06 fgets(str, LINE_MAX, fp); May 05 21:15:13 retno = vsscanf(str, scanf_line, argptr); May 05 21:15:23 the LINE_MAX is plain wrong May 05 21:15:38 I really don't know what problem that patch is resolving and why it used vfscanf :) May 05 21:15:41 xmalloc and friends have to be used May 05 21:15:52 ..and I'm too lazy to learn about x86 / EDD now ;) May 05 21:15:54 what patch May 05 21:15:59 the EDD one? May 05 21:16:02 the damned May 05 21:16:06 yes May 05 21:16:07 no idea May 05 21:16:17 but we just replace vfscanf and we're done May 05 21:16:25 it's way easier than blacklisting the package May 05 21:16:37 about the rest of patching, I have already some minor fixes May 05 21:16:45 ok May 05 21:16:51 so you looked at the patch I sent you May 05 21:17:03 yes May 05 21:17:06 ok good May 05 21:17:13 also we need patch headers May 05 21:17:29 sure May 05 21:18:03 I come clenly to the two errors about basename and vfscanf with just a few of header edits May 05 21:18:32 I think it's lighter than your patchest but please keep an eye on it May 05 21:18:58 any suggestion is welcome..as you see there are unexpected issues compiling with klibc May 05 21:19:21 basename is replaced in my patches May 05 21:19:28 yes, strr May 05 21:23:23 GNUtoo: about the kernel panic now May 05 21:23:44 I see devtmpfs loaded so console should be created May 05 21:23:59 fb is detected as well May 05 21:25:40 imagine that on qemux86 I get May 05 21:25:43 We have no tty May 05 21:26:03 Kernel cmdline: vga=0 uvesafb.mode_option=640x480-32 root=/dev/hda rw mem=128M ip=192.168.7.2::192.168.7.1:255.255.255.0 oprofile.timer=1 console=ttyS0 May 05 21:27:23 ant_home, ok May 05 21:27:35 ant_home, I want to test my patch May 05 21:27:41 how do I do that if it fails for me? May 05 21:27:52 tried to enable text ui ? May 05 21:27:56 nope May 05 21:27:59 ahh May 05 21:28:05 and I've no vesa May 05 21:28:15 I used oe's defaults for the kernel May 05 21:28:20 I've i915 instead of vesa May 05 21:30:14 recompile kexecboot with Kernel cmdline: vga=0 uvesafb.mode_option=640x480-32 root=/dev/hda rw mem=128M ip=192.168.7.2::192.168.7.1:255.255.255.0 oprofile.timer=1 console=ttyS0 May 05 21:30:18 oops May 05 21:30:25 recompile with May 05 21:30:30 --enable-textui May 05 21:30:55 in theory it should fallback automatically May 05 21:31:25 ok May 05 21:31:25 if you want, do it text-only --enable-fbui=no May 05 21:31:47 see kexecboot.inc for more options May 05 21:33:04 I've tested text-ui only on zaurus May 05 21:33:13 ..long ago... May 05 21:33:40 ah, maybe you want to test my fork of kexecboot with latest patches May 05 21:34:19 hmmm May 05 21:34:25 let me try with what I have first May 05 21:34:37 http://paste.debian.net/2285/ May 05 21:39:38 GNUtoo-N71JQ: btw, that failure with vfat...I've already seen it May 05 21:39:49 I think it's because Yocto patches May 05 21:41:39 ok May 05 21:42:13 I'll debug one day... May 05 21:42:21 well May 05 21:42:28 GUI is initialized ok May 05 21:42:32 ohh May 05 21:42:37 Jay7, May 05 21:42:44 how are you? May 05 21:42:50 but no items found for menu creation May 05 21:42:57 ant_home: fine :) May 05 21:43:11 finished my previous work May 05 21:43:15 have some rest May 05 21:43:30 and lot of new ideas :) May 05 21:43:46 yes I'll put my microsd in May 05 21:45:16 http://paste.debian.net/hidden/2890d908/ May 05 21:45:22 I'm traveling a lot those days but I hope we can fix this kexec issue escalating May 05 21:45:31 if (-1 == fill_menu(¶ms)) { May 05 21:45:33 exit(-1); May 05 21:45:35 } May 05 21:46:05 hm.. May 05 21:46:06 s/microsd/sd May 05 21:46:12 second case is interesting May 05 21:46:23 menu have 1 item May 05 21:47:30 it should not panic even with empty list, I've tested May 05 21:47:31 the kernel config lacks probably some stuff May 05 21:47:46 yes I've tested that too in application(not init) mode May 05 21:47:59 yes, It sadly remembers some epic failures with zaurus collie :/ May 05 21:48:00 GNUtoo-N71JQ: can you show boot.cfg from that partition? May 05 21:48:26 GNUtoo-N71JQ: i.e. it fails even in non-init mode? May 05 21:48:55 http://paste.debian.net/hidden/6c742a5e/ May 05 21:49:03 lelt me try that May 05 21:50:09 yes it does May 05 21:50:12 hm.. May 05 21:50:13 probably my patch May 05 21:50:14 console=tty1 console=ttyS0,115200 fbcon=rotate:0 loglevel=8 ignore_loglevel verbose vga=0 May 05 21:50:44 depending how the kernel is configured you'd need to put this in APPEND as well May 05 21:51:09 I use ipxe for now May 05 21:51:20 before putting that in flahs May 05 21:51:22 *flash May 05 21:51:25 it's safer May 05 21:51:52 GNUtoo-N71JQ: can you build it with debug symbols (not strip) and run under gdb in non-init mode? May 05 21:52:13 and show me backtrace output May 05 21:53:32 hmmm May 05 21:53:34 ah.. that's not crash, so no backtrace May 05 21:53:43 why do we strip May 05 21:53:58 just run it under strace then May 05 21:54:01 because oe should strip automatically May 05 21:54:07 and save the binaries before May 05 21:54:12 to put them in -dbg May 05 21:54:19 check out in workdir May 05 21:54:24 under package-split May 05 21:54:56 packages-split exactly May 05 21:55:05 anyway I need to know where it calling exit().. I suspect kexecboot.c:1124 May 05 21:55:12 if (rc < 0) exit(rc); May 05 21:55:25 may be something wrong with evdev's May 05 21:55:49 ok I've it May 05 21:55:51 b exit May 05 21:55:52 GNUtoo-N71JQ: do you have /dev/event* or /dev/input/event* devices? May 05 21:56:14 before it worked on my build laptop May 05 21:56:22 before adding text ui May 05 21:56:25 ok, run it under gdb May 05 21:56:53 ok I'll do it not under real console May 05 21:57:56 kexecboot.c:1113 May 05 21:58:12 1112 /* rc < 0 indicate error */ May 05 21:58:12 1113 if (rc < 0) exit(rc); May 05 21:58:27 in main May 05 21:58:44 I'll remove the text ui May 05 21:58:45 heh.. May 05 21:58:53 let's go deeper :) May 05 21:59:13 rc = do_main_loop(¶ms, &inputs); May 05 21:59:20 rc is coming from here May 05 21:59:34 ok May 05 21:59:37 1092 May 05 22:00:19 sh sorry May 05 22:00:23 I'll retry as root May 05 22:01:17 rc = process_ctx_menu(params, action); May 05 22:01:24 :985 May 05 22:01:27 as root it works May 05 22:01:34 sorry I'm really tired now May 05 22:01:39 hm.. May 05 22:01:56 of course in the initramfs it's root May 05 22:01:57 i.e. when running in non-init mode it works now (as root)/ May 05 22:02:10 yes May 05 22:02:12 exactly May 05 22:02:19 interesting.. May 05 22:02:20 under my build laptop May 05 22:02:25 in a full blown distro May 05 22:03:13 have you devtmpfs in your kernel? May 05 22:03:25 yes May 05 22:03:27 or /dev/event* in your initramfs if no devtmpfs May 05 22:03:30 should I have it mounted? May 05 22:03:41 kexecboot mounted it on start May 05 22:03:55 ok May 05 22:04:00 so it should not be mounted May 05 22:04:02 #ifdef USE_DEVTMPFS May 05 22:04:02 if (-1 == mount("devtmpfs", "/dev", "devtmpfs", May 05 22:04:05 and anyway it works May 05 22:04:13 I mean it finds my microsd May 05 22:04:16 it fails after that May 05 22:04:31 yeah.. May 05 22:05:03 what happens if /dev/event is lacking May 05 22:05:04 ? May 05 22:06:04 can't say exactly right now, but I think kexecboot should exit from do_main_loop() with rc = -1 May 05 22:06:10 then exit later May 05 22:06:24 because of no input devices May 05 22:06:43 let me add event interfaces May 05 22:06:44 ok May 05 22:06:48 /dev/event* or /dev/input/event* May 05 22:06:56 it check both places May 05 22:07:15 because I had input May 05 22:07:18 but no event May 05 22:07:46 ok now it works May 05 22:08:17 now let me try klibc May 05 22:09:19 o_O May 05 22:09:39 thanks a lot btw May 05 22:09:45 np May 05 22:09:52 so what fixes issue? May 05 22:10:10 adding events May 05 22:10:14 I had input already May 05 22:10:18 but no event interface May 05 22:10:30 no event* files in /dev/input dir? May 05 22:11:34 yes May 05 22:11:36 exactly May 05 22:11:43 that configuration option was lacking: May 05 22:12:12 Event interface (INPUT_EVDEV) May 05 22:12:16 in input May 05 22:12:19 I had mouse and keyboard May 05 22:12:51 seems we need to add docs about reqiured kernel options May 05 22:12:53 this is one fragment to add to the next linux-yocto-tiny-kexecboot kernel May 05 22:13:26 ok May 05 22:13:29 I plan to make 3.8 version modular May 05 22:13:29 about my patch: May 05 22:13:38 well, good to know about that May 05 22:13:44 too late here May 05 22:13:47 * Jay7 -> sleep May 05 22:13:58 GNUtoo-N71JQ, you see, we test on Zaurus with heavily tested config May 05 22:14:10 kernel is almost the same since years May 05 22:14:17 http://paste.debian.net/hidden/cdda7f26/ May 05 22:14:46 although I heard some stories about the XZ compression parameter May 05 22:14:58 I'm also really tired right now May 05 22:15:02 hm.. there where a couple of vfscanf iirc May 05 22:15:03 sorry if the patch is wrong May 05 22:17:09 (l'm that tired that some minutes ago I tried kexecboot not-as-root) May 05 22:17:27 so I guess I'll go to sleep really soon May 05 22:18:17 np, I'll have still to reorder the patches May 05 22:18:55 we'll fix it in the next days May 05 22:19:07 ok May 05 22:19:21 many thanks for your help! May 05 22:19:59 I'm spending the little time on kernel lately, now it's time to refresh kexec May 05 22:20:40 I'll send out the last temporary fix for parsing May 05 22:23:00 ok May 05 22:23:04 np May 05 22:23:21 many thanks for the help for fixing my kernel config May 05 22:30:25 * ant_home zZZ May 05 22:30:27 gn May 05 22:30:38 good night **** ENDING LOGGING AT Mon May 06 02:59:58 2013