**** BEGIN LOGGING AT Wed Jun 01 02:59:56 2011 Jun 01 07:42:52 testlog Jun 01 07:43:25 great, log did rotate Jun 01 07:45:13 kewl Jun 01 08:02:45 http://logs.nslu2-linux.org/livelogs/kexecboot/ Jun 01 08:02:48 rotated logs Jun 01 08:03:29 ant_work: have you seen mail from Paul Sokolovsky in kexecboot-devel? Jun 01 08:04:04 so all is good? Jun 01 08:04:12 ka6sox-home: yes, thank you Jun 01 08:04:43 now we have logging again :) Jun 01 08:05:32 excellent...do you have autobuilder? Jun 01 08:05:49 I have local testbuilder Jun 01 08:05:57 but I'm starting it by hands Jun 01 08:06:13 okay...we have also provision for build logs too. Jun 01 08:06:49 I have local web-server at home for this :) Jun 01 08:07:03 kk...that works too. Jun 01 08:07:07 http://jay-home.ath.cx/testbuilder/logs/ Jun 01 08:07:46 hm.. I should adjust apache config there to show longer file names Jun 01 08:08:09 ya, would be easier to understand. Jun 01 08:08:57 okay back to prepping for new servers... Jun 01 08:11:23 well, now better Jun 01 08:15:27 Jay7: I've read... Jun 01 08:15:46 well, I tested textui immediatly after commit and was working Jun 01 08:16:02 don't remember if I compiled both ui, probably not Jun 01 08:16:11 together I mean Jun 01 08:16:42 basically pfalcon complains there is no automatic fallback to text-ui Jun 01 08:24:41 hehe Jun 01 08:24:59 we talked about this yesterday iirc :) Jun 01 08:29:16 * kenws is just lurking : ) Jun 01 08:38:03 kenws: hello :) Jun 01 08:50:42 ant_work: did you tested some recent opie builds? Jun 01 08:51:08 yes, couple of weeks ago Jun 01 08:51:42 I need something usable on Tosa or Akita.. Jun 01 08:52:12 seems I'll build special kernel with right NAND partitioning compiled Jun 01 08:52:26 simple patch in my private branch Jun 01 08:52:56 about this,iirc kexecboot did indeed find the first mtd, even if resized Jun 01 08:53:00 hm.. or may be just compile cmdline with right mtdparams Jun 01 08:53:27 fstype just scans first chunk (2 bytes) Jun 01 08:53:32 sure, it is always after first 7Mb :) Jun 01 08:53:47 but jffs recovery may easily break it Jun 01 08:54:00 why the hell should you have it with 2.6 ? Jun 01 08:54:10 was a 2.4 cmdline arg Jun 01 08:54:10 good question Jun 01 08:54:53 anyway right formatting is not straightforward Jun 01 08:55:26 if you dont' add -j the cleanmarkers are written on first boot / recovery Jun 01 08:55:28 I wish to try SHR one time.. Jun 01 08:56:07 I think I'll add a shell script to handle formatting and maybe flashing of mtd Jun 01 08:56:25 in zaurus-utils Jun 01 08:56:49 installer? Jun 01 08:57:34 you can always use updater.sh if you want but seems to me it's easier from console Jun 01 08:57:57 console-image with mtd utils + shell script Jun 01 08:58:12 yes, maybe in sbin Jun 01 08:58:12 and name it recovery image like pdaxrom :) Jun 01 08:58:49 actually we could deploy it in all images as we do with nandlogical Jun 01 08:58:58 even better - busybox image with shell script as /etc/rc :) Jun 01 08:58:58 1-2 kb I guess Jun 01 08:59:13 dialog + mtd utils + nandlogical Jun 01 08:59:20 something like cacko installer :) Jun 01 08:59:30 or hack this Jun 01 08:59:58 I have a basic script but w/out ncurses Jun 01 09:00:03 or dialog Jun 01 09:00:15 well.. we should release some kexecboot images with classic OE Jun 01 09:00:23 yes, 0.6? Jun 01 09:00:25 yes.. Jun 01 09:00:31 w/o other boot methods Jun 01 09:00:38 classic OE + classic kexecboot :) Jun 01 09:00:42 heh Jun 01 09:01:05 I've pinged pfalcon to join Jun 01 09:01:20 I wish resolve that problem with fb console (latest item) Jun 01 09:01:25 now, about debugging Jun 01 09:01:36 priority is imho: Jun 01 09:01:46 1) pfalcon text-uireport Jun 01 09:01:54 2) qvga sys-menu Jun 01 09:02:01 3) slow input (on spitz) Jun 01 09:02:03 1 == ui fallback Jun 01 09:02:28 3 is WIP slowly Jun 01 09:02:53 I'll try to add more graphical primitives Jun 01 09:03:06 e.g. to draw unfilled frame Jun 01 09:03:14 2 is the hardest one to catch... Jun 01 09:03:19 yeah :( Jun 01 09:03:45 I've re-read source code yesterday evening...anything strange Jun 01 09:03:54 but there must be smthg ;) Jun 01 09:03:54 I fear new fb code is source of.. Jun 01 09:04:30 may be fallback to old method Jun 01 09:04:40 hm.. idea Jun 01 09:04:41 I'm thinking about more verbosity just before entering next menu level Jun 01 09:05:11 I'll write line drawing functions which have no logic inside, just calling plot_pixel as before Jun 01 09:05:43 this should reduce possibility of fb ram overflow Jun 01 09:06:10 with price of reduced speed.. but old code was even slower on drawing Jun 01 09:06:28 well, I'll look today Jun 01 09:09:08 seems I should buy 3G-modem Jun 01 09:09:26 now I have no office and no link when not at home Jun 01 09:10:15 ouch Jun 01 09:12:30 this have pro's and con's :) Jun 01 09:22:05 Jay7: did you test with valgrind the latest fb-ui in qvga? Jun 01 09:22:55 I can't get qvga in qemu-kvm :( Jun 01 09:29:23 ok, I'll retry on c7x0, maybe rotated this time Jun 01 10:02:24 ant_work: about power key Jun 01 10:02:55 try to add case KEY_SUSPEND to switch() where keys are processed Jun 01 10:03:21 e.g. bind it to show log info Jun 01 10:03:52 just to check that kexecboot see that keypress Jun 01 10:04:06 yes, th eidea was to dump its value Jun 01 10:04:40 or disable keymap loading in console-image e.g. Jun 01 10:04:41 but the same happens on poodle...with another kb driver (locomo) as collie... Jun 01 10:04:59 I'm not sure what event device produces that keypress Jun 01 10:05:08 common keyboard evdev? Jun 01 10:05:19 I'm a bit lost too Jun 01 10:05:28 I'll start from kernel hardcoded kbmap :) Jun 01 10:05:36 iirc there was some tool to view events on eventX device.. Jun 01 10:05:41 evtest? Jun 01 10:05:45 may be Jun 01 10:05:49 can't remember Jun 01 10:05:51 worth a try Jun 01 10:06:38 kernel reads 372 KEY(7, 0, CORGI_KEY_OFF), Jun 01 10:06:53 in static const uint32_t corgikbd_keymap[] = { Jun 01 10:07:23 7,0 Jun 01 12:03:55 Jay7: I was reading about LIC_FILES_CHKSUM Jun 01 12:04:24 in kexecboot sources there is a correct COPYING Jun 01 12:04:55 (though gitweb shows strange \f ) Jun 01 12:05:16 but then the copyright is spread around Jun 01 12:05:28 maybe an nclude / common.h Jun 01 12:06:17 with that stuff Jun 01 12:30:36 ant_work: I don't know how to better handle this Jun 01 12:30:43 * Jay7 -> afk Jun 01 12:33:29 https://wiki.yoctoproject.org/wiki/Recipe_%26_Patch_Style_Guide#LICENSE Jun 01 12:33:41 seems we can just take one/two files to track Jun 01 12:34:53 ah, no Jun 01 12:35:45 well, maybe :p Jun 01 12:35:56 > This file lists all the files with the text of licenses for the source code Jun 01 12:36:06 >... Jun 01 12:36:08 > There is no limit on how many files can be specified. But generally every project would need specifying of just one or two files for license tracking. Jun 01 12:36:12 ?1? Jun 01 15:54:29 I think license tracking is going to be a big PITA... Jun 01 15:58:48 bbl Jun 01 21:35:30 Jay7: anything new wrt debug to test? Jun 01 21:52:30 Jay7: now OE is in sync with upstream...bugs included :p Jun 01 21:58:01 and, about --enable-static-linking compile kexecboot as static executable [default=no] Jun 01 21:58:31 I'm under th eimpression kexecboot-klibc is just overriding with own klcc flags Jun 01 21:59:34 a glibc/eglibc static version doesn't make much sense imho Jun 01 22:17:49 ant__: sorry, no debug today Jun 01 22:18:19 about static - someone should check :) Jun 01 22:18:35 np Jun 01 22:19:34 any ideas about license files? Jun 01 22:19:54 we can collect all copyrights in AUTHORS file e.g. Jun 01 22:20:46 beside you there are some authors listed in some files Jun 01 22:20:57 is it still actual? Jun 01 22:20:57 yes, I mean that Jun 01 22:22:21 almost all was rewritten Jun 01 22:23:13 imho, they all should be respected as authors Jun 01 22:23:31 ok, the list them back in 2009 Jun 01 22:23:43 *then Jun 01 22:23:46 on kexecboot.org we have hall of fame Jun 01 22:24:15 we must add pfalcon :) Jun 01 22:24:35 for the next ver with losetup etc. Jun 01 22:24:47 with ui fallback then :) Jun 01 22:25:17 I hope he will join: we can reuse much of his work in th eold initramfs scripts Jun 01 22:25:57 btw: we shall start to think about devicetree Jun 01 22:26:11 he is spending his time for linaro now Jun 01 22:26:16 it can be appended to the kernel Jun 01 22:26:34 or reside elsewhere Jun 01 22:26:43 I don't know what we should consider wrt devicetree.. Jun 01 22:28:37 btw btw, talking with RP he understands CONFIG_MTDPARTS will be 'yes' by default, just not configurable Jun 01 22:29:22 so we can stick with 'mtdparts' tag for now Jun 01 22:29:45 ok, poodle image is built Jun 01 22:30:16 now, what do you thinkl: printf before entering system menu other immediately after entering it? Jun 01 22:30:41 add both :) Jun 01 22:30:50 :p Jun 01 22:31:28 wow... 'other' -> oder -> or Jun 01 22:31:41 I'm a bit tired... Jun 01 22:32:08 left brain speeks english, right german Jun 01 22:32:16 bah Jun 01 22:32:45 speaks very bad, indeed Jun 01 22:43:17 Jay7: locomo keyboard has big problems! Jun 01 22:43:30 I can't test [0-9] keys Jun 01 22:44:22 r and s ok as before Jun 01 22:44:33 hm.. strange Jun 01 22:44:59 how do you type on collie? Jun 01 22:45:04 check with console image with keymap loading disabled Jun 01 22:45:30 I'm pretty sure it's broken Jun 01 22:45:32 I'm not sure I've tested it before.. Jun 01 22:46:09 with 2.6.3x the central button doesn't work too Jun 01 22:46:17 cursors yes Jun 01 22:46:39 I fear it is like for ben Jun 01 22:47:00 we need a keymap even for initramfs env Jun 01 22:48:00 yeah, that is what I've thought Jun 01 22:48:12 or patch for kernel keymap Jun 01 22:48:20 should be easier Jun 01 22:48:48 in kexecboot I see Jun 01 22:49:00 Unknown event type: 0x0000 Jun 01 22:49:06 Keys or Buttons Jun 01 22:49:11 Repeat Jun 01 22:49:15 Power Jun 01 22:55:10 ok, frozen in System Menu Jun 01 22:55:40 here there are 5 items listed Jun 01 22:56:06 and it looks like there is space for a 6th Jun 01 23:08:21 now testing as standalone on poodle Jun 01 23:10:26 he he Jun 01 23:10:32 Jay7: ;)) Jun 01 23:10:43 with 6 items it does not freeze Jun 01 23:10:49 hm.. Jun 01 23:10:54 (standalone adds Exit) Jun 01 23:11:05 comment it :) Jun 01 23:11:10 wait Jun 01 23:11:23 I can press Bac but then it returns here Jun 01 23:11:31 strange Jun 01 23:11:47 as like menu-id was not updated on exit Jun 01 23:12:06 so I'm prisoner in system-menu :) Jun 01 23:12:39 btw, two rapid clicks on back do the trick Jun 01 23:12:41 hm.. strange Jun 01 23:12:43 I'm out Jun 01 23:13:05 can you compile w/o optimization? Jun 01 23:13:14 this all looking very strange Jun 01 23:13:19 I fear some gcc bug Jun 01 23:13:37 hm.. maybe input rate is too fast Jun 01 23:13:52 rapid click is ok Jun 01 23:14:01 well, may be.. Jun 01 23:14:26 (it was ok with old kernel) Jun 01 23:15:03 but when I've tried to debug non-working UI on collie time ago I got very strange behaviour Jun 01 23:15:35 ok, same binary, bug is repeatable Jun 01 23:15:37 kexecboot just fails on different places Jun 01 23:17:03 frozen with 5 items, working strangely with 6 Jun 01 23:17:42 this only happens in submenu Jun 01 23:18:04 first level item population doesn't matter Jun 01 23:22:38 Jay7: Alignment trap: kexecboot (429) PC=0x00014438 Instr=0xe5933000 Address=0x401a7c6f FSR 0x013 Jun 01 23:23:50 nice Jun 01 23:24:10 disable thumb and check compiler options.. Jun 01 23:24:21 may be try -O0 Jun 01 23:24:33 this was maybe when exiting from standalone mode Jun 01 23:26:54 ah, no Jun 01 23:27:03 it happens right on entering System Menu Jun 01 23:27:51 this corresponds to those 'long clicks' Jun 01 23:29:00 like double-pressure, exit/enter Jun 01 23:29:19 twilight zone... Jun 01 23:30:37 ok, now we need pb__ and the binary ;) Jun 01 23:30:54 or google and objdump Jun 01 23:32:51 shit.. client's windows infrastructure almost destroyed.. both AD computers are lost SAM Jun 01 23:33:09 one have snapshot but seems it is obsoleted Jun 01 23:33:13 omg Jun 01 23:33:36 one, who know how all that shit works now in USA in vacation Jun 01 23:33:57 employees have only my phone to resolve that Jun 01 23:34:03 FUCK Jun 01 23:34:08 typical scenario :/ Jun 01 23:34:38 how is Windows integrated thgere? Jun 01 23:35:12 is it impossible to move all services to *x? Jun 01 23:38:48 all servers are in vm's under xenserver Jun 01 23:38:54 that's good part Jun 01 23:39:05 so I have snapshots at least Jun 01 23:40:48 that will save your bacon. Jun 01 23:42:02 no, that will save client's business.. Jun 01 23:42:16 I'm only infrastructure admin there Jun 01 23:42:36 i.e. xenserver, storages, network and unix machines support Jun 01 23:42:44 no windows support at all Jun 01 23:42:53 "I don't do windows" Jun 01 23:42:58 but they will ask me at morning.. Jun 01 23:43:11 because they have no one else Jun 01 23:43:45 that is not unique problem in IT world these days. Jun 01 23:51:09 heh, I saw that back in the 80's and choosed the DIYS Jun 01 23:51:21 way Jun 01 23:51:55 ...but I'm still blocked with our own ms-acess frontend Jun 01 23:52:45 people are just used with it and refuse the php webapp :/ Jun 01 23:53:19 so some are running Windows... WIN2000 :] Jun 01 23:53:49 at least I know the risks... Jun 01 23:55:54 ok, almost 2 o' clock Jun 01 23:56:04 good night Jun 01 23:56:55 almost 4 o' clock.. :( Jun 02 00:00:32 Jay7, go to bed...they will call you too early. Jun 02 00:00:47 I'm trying to resolve this somehow.. Jun 02 00:01:16 we have a saying in the US...."Punt" Jun 02 00:07:45 when you can't deal with the problem...punt it over to someone who can...aka find a windoze guy. Jun 02 00:22:21 kind of.. well.. it can't be resolved from home right now, so going to bed Jun 02 00:22:59 nite **** ENDING LOGGING AT Thu Jun 02 02:59:56 2011