**** BEGIN LOGGING AT Mon Dec 10 02:59:58 2012 Dec 10 10:12:00 Jay7: one minor thing in kexecboot.c remove "* TODO Show debug info in special dialog" Dec 10 10:12:05 done Dec 10 10:12:49 hehe Dec 10 10:13:01 what should we do with last 2 patches? Dec 10 10:13:06 revert? Dec 10 10:13:23 or revert only one with ASSEMBLY? Dec 10 10:13:58 or wait for some fix? Dec 10 10:23:44 I'm perplexed Dec 10 10:24:04 quick fix is to force cmdline for arm as well Dec 10 10:24:42 bluelightning: maybe we could ask hpa :) Dec 10 10:26:45 Jay7: it's still there in asm/setup.h Dec 10 10:26:49 in 3.6 Dec 10 10:26:50 19 #define COMMAND_LINE_SIZE 1024 Dec 10 10:27:02 well, maybe do that in kexecboot? Dec 10 10:27:40 .. and remove that include Dec 10 10:27:41 I've built without including asm/setup.h Dec 10 10:27:44 yes Dec 10 10:27:51 may be this is good idea Dec 10 10:27:52 put an IFNDEF around Dec 10 10:28:01 and make another configure option :) Dec 10 10:28:08 we get rid of patch + issue together :) Dec 10 10:28:35 1024 seem senough, though. a config options is perhaps superfluous. dunno Dec 10 10:29:09 Tequila was faced to case when increase was needed iirc Dec 10 10:29:54 we just add less than 1kb memory footprint Dec 10 10:30:29 and I think it's important to have space for exotic framebuffers/timings Dec 10 10:31:19 ok, let's remove include and add our own #define Dec 10 10:31:34 I'll make patch this evening Dec 10 10:32:12 then for the other issue I think we can sanitize the patch and keep it Dec 10 10:32:30 (evdevs.c) Dec 10 10:33:43 I think also here you'd put an IFNDEF around #define BITS_PER_LONG (sizeof(long) * CHAR_BIT) Dec 10 10:34:15 but we need to investigate / test more Dec 10 10:35:49 I've googled and the isue is on PPC it could be 32 or 64 Dec 10 10:35:58 not only on PPC... Dec 10 10:36:39 I had a quick talk on #klibc on that ;) Dec 10 10:36:58 seems about ioctl the only guru is hpa Dec 10 11:07:55 Jay7: there are doubts about th bitshifting in #klibc ight now... Dec 10 11:08:14 sorry for the typos, I'm on two kb Dec 10 11:23:22 ant: I looked at the bitshifting thing, and if the original worked, the result will also work but NOT be endian-neutral like the original Dec 10 11:23:23 so it's a behaviour change on not-little-endian platforms Dec 10 11:23:49 mira|AO> it may be a fix or a breakage, but it's a behaviour change on big-endian and pdp-endian(god beware us) Dec 10 11:41:59 omg Dec 10 11:43:11 ant_work: what was meant under 'the result'? new version by zeldin? or one from headers? Dec 10 12:08:15 I think he's talking about bitshifting Dec 10 12:08:45 the preceding comment was Dec 10 12:08:46 hpa: ant_work was asking for a review of https://github.com/kexecboot/kexecboot/commit/109cf8d8563d7adf4b869421f3e9f6422e3adc34 Dec 10 12:08:47 * maks didn't like this magic bitshifting, so passed on. (; Dec 10 12:09:38 then I said I don't see runtime diffs on arm / 32bit Dec 10 12:10:07 at this point mira did comment Dec 10 12:12:31 Jay7: isn't nanonote big endian? Dec 10 12:12:39 not sure Dec 10 12:21:54 no, seems debian on it is mipsel Dec 10 19:38:20 hi dromede Dec 10 19:39:50 hi there Dec 10 20:04:22 i see that you guys merged both of my fb.c patches Dec 10 20:27:12 dromede_: yes, and Jay7 is merging some zeldin's patches Dec 10 20:27:21 corrections are needed, though Dec 10 20:27:59 hi, btw Dec 10 20:28:48 i'm not surprised, "refactor line and pixel drawing routines" was only properly tested on a 16 bit framebuffer Dec 10 20:28:52 hi jay7 Dec 10 20:29:22 dromede_: I've fixed bug in 24/32bpp btw Dec 10 20:29:31 amended to your commit Dec 10 20:29:50 so it works on 24/32bit? Dec 10 20:30:04 it works on 24bpp Dec 10 20:30:10 not tested on 32bpp Dec 10 20:30:15 great! Dec 10 20:30:21 dromede_: we need zeldin's confirmation that it works on 24bits big-endian Dec 10 20:30:38 there is a pending patch for this Dec 10 20:30:45 well, if it works on 24, i see on reason why it shouldn't work on 32 bit Dec 10 20:31:18 ant_home: so, what would we do with zeldin's patches applied? Dec 10 20:31:45 I'd remove the CMLINE stuff alltogether Dec 10 20:32:04 I agree Dec 10 20:32:06 then, about bitshifting, it happens to work here Dec 10 20:32:22 just a minor warning to fix (for klibc only afais) Dec 10 20:32:42 should we increase cmdline size from 255 to 1024? Dec 10 20:32:46 yes Dec 10 20:32:56 ok, I'll commit this Dec 10 20:33:03 if you really want, add a config knob, default = 1024 Dec 10 20:33:25 to me it seems pointless today Dec 10 20:37:31 dromede_: any comment about https://github.com/kexecboot/kexecboot/commit/109cf8d8563d7adf4b869421f3e9f6422e3adc34 Dec 10 20:38:12 pushed Dec 10 20:38:18 atm we just need to add an #ifndef around define BITS_PER_LONG (sizeof(long) * CHAR_BIT) Dec 10 20:38:28 which happens to be defined at least in klibc Dec 10 20:38:56 afk Dec 10 20:39:02 some commenter said after this patch the code does not respect endiannes Dec 10 20:39:12 (if I understand the comment) Dec 10 20:39:36 ant_home: no comment Dec 10 20:39:44 Jay7: pls add this #ifndef and bump version when you can Dec 10 20:39:51 0.6.0 ? Dec 10 20:40:27 dromede_: it would indeed be strange, seen patch title Dec 10 21:55:21 Jay7: I've tested this on both klibc and eglibc. No warnings Dec 10 21:55:24 http://paste.debian.net/215469/ Dec 10 21:55:34 ant_home: fine Dec 10 21:59:35 you know, I had again failure on spitz...no ubi (ubi-1) Dec 10 21:59:48 then I remembered the wrong default partitioning Dec 10 21:59:48 :) Dec 10 22:01:06 during holidays I'll start to write the kernel mtd parser, like done for some Broadcom recently Dec 10 22:20:34 Jay7: still there? Dec 10 22:20:41 yep Dec 10 22:21:08 awake enough to push the fix and the version? Dec 10 22:21:54 I'd like to refresh oe version asap, pwgen and others are playing Dec 10 22:22:48 ant_home: do you tested #ifndef-ed version? Dec 10 22:22:53 *did Dec 10 22:23:03 btw, do you think the next round of kernels should be 3.4 ? Dec 10 22:23:13 yes, no warnings Dec 10 22:23:19 ok Dec 10 22:24:06 I'm unsure about 0.6.0 Dec 10 22:24:48 it's for the screenshots :) Dec 10 22:24:50 we have merged big changes but tested them very little Dec 10 22:25:15 then go with 0.5.9.1 Dec 10 22:25:31 0.5.90? Dec 10 22:25:47 otherwise people around will not know which version they are using... Dec 10 22:26:21 v.0.5.10 then Dec 10 22:26:41 really, we are still pre - 1.0 so I don't see the point Dec 10 22:27:11 after 0.5.9 we can boot ubi and this is a big diff imho Dec 10 22:30:33 about testing, we've tested on qemu, poodle (qvga), spitz Dec 10 22:31:01 it would be nice to have feedback from zeldin Dec 10 22:31:17 and pwgen for mips ofc Dec 10 22:31:33 pwgen is building for R Pi now :) Dec 10 22:31:44 ah, ok, then is arm Dec 10 22:31:58 nanonote is mipsel Dec 10 22:32:02 you have the mips ;) Dec 10 22:32:13 heh Dec 10 22:33:40 yes, I have :) Dec 10 22:34:21 btw I bought a collie for 15Eur Dec 10 22:35:04 I hope it's not faulty Dec 10 22:35:57 tosa still priced over 150Eur on ebay Dec 10 22:38:30 Martin Jansa is now WebOS Developer at Hewlett-Packard Dec 10 22:38:32 hehe Dec 10 22:38:37 ah Dec 10 22:38:49 via LinkedIn Dec 10 22:39:09 wow, so you have collie now? waiting for delivery? Dec 10 22:39:28 yes, is in Muenchen Dec 10 22:39:50 should come next week Dec 10 22:40:01 maybe before Dec 10 22:40:33 * Jay7 need to buy Pandaboard ES.. Dec 10 22:40:33 I have around 10-15 kernels for collie..I hope I'll find the good one to upload :p Dec 10 22:40:49 CEWG will refund me it Dec 10 22:41:04 why exactly pandaboard? Dec 10 22:41:26 but it's challenge to deliver something now to Russia.. Dec 10 22:41:35 ant_home: Tim decided so :) Dec 10 22:41:41 I see Dec 10 22:42:00 FedEx and DHL doesn't ship to physical persons Dec 10 22:42:05 only to organizations Dec 10 22:42:51 but almost all shops in EU are shipping by FedEx or DHL Dec 10 22:43:38 seems I should try USA via relative or some China seller Dec 10 22:44:03 sounds very bad Dec 10 22:53:48 hmm Dec 10 22:54:14 Jay7: I think theme-tui.h could need to #include "../config.h" Dec 10 22:54:28 or not.. Dec 10 22:54:43 I'm still can't choose version :) Dec 10 22:54:43 for #ifdef USE_TEXTUI Dec 10 22:55:25 hm, that guard is superfluous now Dec 10 22:56:40 I've pushed #ifndef-ed macro Dec 10 22:59:18 tui.c includes config.h and then theme-tui.h Dec 10 22:59:28 it's all right Dec 10 22:59:55 (theme-tui.h is only included once, there) Dec 10 23:02:53 Jay7: about version, http://www.randomizer.org/form.htm Dec 10 23:02:59 ^_^ Dec 10 23:03:20 ;) Dec 10 23:03:39 well, let's postpone version change to tomorrow Dec 10 23:04:01 * Jay7 falling asleep Dec 10 23:06:34 gn, see you tomorrow **** ENDING LOGGING AT Tue Dec 11 03:00:00 2012