**** BEGIN LOGGING AT Mon Oct 08 02:59:59 2012 Oct 08 08:26:11 gm bluelightning Oct 08 08:35:08 hi ant_work Oct 08 08:49:28 I've sent an e-mail to Bruce about a couple of minor issues I've found Oct 08 08:50:50 this evening I'll push on github a modified meta-handheld layer for testing Oct 08 08:52:44 yocto-tiny is at 3.4 as well now, so we can finally update the kexecboot kernel Oct 08 08:52:51 awesome Oct 08 08:52:57 plans to keep 3.2? Oct 08 08:53:12 btw, bruce's latest email said he won't have access to his WR email until wednesday I think Oct 08 08:53:30 heh, yes, Canadian Thanksgiving .... Oct 08 08:53:46 depends how well 3.4 works I guess; what's your impression so far? Oct 08 08:53:56 hm.. kernel is a bit too big Oct 08 08:54:45 note I still have to test runtime... Oct 08 08:55:54 in ktipe/standard there are tons of stuff built-in which we don't need Oct 08 08:56:17 it is more tailored for pc's Oct 08 08:57:12 the problem is that 'subtractive synthesys' is very boring...we have explicitely to add # CONFIG_blah is not set Oct 08 08:57:42 on the other side the ktype/tiny is really minimal Oct 08 08:58:02 would we not just start from ktype/tiny though? Oct 08 08:58:35 my idea is to refactor linux-yocto-tiny now, based on that and see Oct 08 08:58:51 linux-yocto-tiny-kexecboot I mean Oct 08 08:59:21 probably the 'additive synthesys' is easier Oct 08 09:00:08 I've started with the full linux-yocto because it offers support for many extras like wifi, usb, ... Oct 08 09:05:52 we should be using an additive approach, that's what OE-Core is about... Oct 08 09:09:14 well, then we have to rework the base Oct 08 09:09:42 standard adds lo of stuff as default Oct 08 09:09:44 http://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-cache/tree/ktypes/standard/standard-nocfg.scc Oct 08 09:46:46 bluelightning: I've found a new 'small' kernel patches, seems the off-tree Con Kolivas work. Better ask Bruce/Darren about this Oct 08 09:47:14 can't hurt to raise it Oct 08 09:47:20 is not under /ktypes Oct 08 09:47:32 http://git.yoctoproject.org/cgit/cgit.cgi/yocto-kernel-cache/tree/small Oct 08 09:48:05 seems more toward a responsive desktop Oct 08 14:57:41 well, I'll have to rebase after linux-yocto/3.4: fix kconfig warnings and unnecessary options Oct 08 15:50:28 --- Oct 08 15:50:37 oh, finally we have 3.6.1 Oct 08 15:51:16 ah, dromede is away Oct 08 15:52:13 bluelightning: you tried 3.6 on hx4700 isn't? Oct 08 15:53:39 ant_work: I haven't yet no... Oct 08 15:54:21 well, we can't create a .bbappend without 3.6 in oe-core anyway... Oct 08 15:55:45 ok, I'm heading home now Oct 08 15:56:49 do you think it is better/easier for testing if I send a patch orshould I fork the repo on github and add my top-commit? If Bruce has to look at this, github seems the easy way... Oct 08 15:57:32 I'd prefer to clean it out before committing it, or even with D_P = -1 Oct 08 15:58:17 btw, we should start creating multiple linux-yocto-3.X directorys in meta-hh :/ Oct 08 15:58:31 s/ys/ie/ Oct 08 15:58:59 there are already too many imho :) Oct 08 16:00:11 do you not have access to oe-core-contrib? Oct 08 16:00:38 don't know Oct 08 16:00:47 if not, we should fix that Oct 08 16:00:51 but it is rather WIP Oct 08 16:01:21 forking on github should be fine, but ideally RP prefers pull requests to use oe-core-contrib Oct 08 16:01:42 ok but here the scope is limited to meta-hh Oct 08 16:02:15 oh sorry I misunderstood Oct 08 16:02:42 about adding 3.6? We need the corresponding kernel-tools c/o Bruce+Darren Oct 08 16:02:45 :) Oct 08 16:02:56 recipe isn't enough Oct 08 16:02:59 I guess that's not going to happen until after the release Oct 08 16:03:05 no way Oct 08 16:03:24 anyway, I hope you guys will help cleaning the fragments Oct 08 16:03:38 I will try to for sure Oct 08 16:03:42 now all is chaotically copy & pasted :/ Oct 08 16:03:53 I should maybe sor alphabetically Oct 08 16:04:33 I *think* the pxa-common fragment I've isolated could be shared with ipaqs. It's the big one Oct 08 16:05:00 same for the usb-host frag shared by akita-spitz-tosa Oct 08 16:05:22 the machine config are just a ew hundred bytes... Oct 08 16:05:26 *few Oct 08 16:05:45 ..but the resulting kernel is probably toolarge Oct 08 16:06:00 we have to clean that Oct 08 16:11:51 bluelightning: btw that usb-host .scc should be related to the MACHINE_FEATURES_append_spitz = " usbhost" Oct 08 16:12:18 we could make it conditional in the recipe maybe...I was lazy and listed all in SRC_URI Oct 08 16:12:49 (was machine->kernel features) Oct 08 16:13:29 ...it' just an extra append Oct 08 16:19:00 bbl Oct 08 22:24:15 bluelightning: instead of $P I have to use FILESEXTRAPATHS_prepend := "${THISDIR}/${PN}-3.4: because $PV is a git hash :/ Oct 08 22:25:47 hmm, that's still fine though Oct 08 22:30:04 I'm about committing but I see that for some reasons some fragments are included twice or even 3 imes... Oct 08 22:30:15 just some wanings... Oct 08 22:30:57 ouch... Oct 08 22:30:58 Value of CONFIG_CC_OPTIMIZE_FOR_SIZE is redefined by fragment meta/cfg/kernel-cache/ktypes/standard/standard.cfg.sanitized: Oct 08 22:30:59 Previous value: CONFIG_CC_OPTIMIZE_FOR_SIZE=y Oct 08 22:30:59 New value: # CONFIG_CC_OPTIMIZE_FOR_SIZE is not set Oct 08 22:31:08 why that? Oct 08 22:31:49 this setting tends to go away...I set it then months later dromede noticed it was missing again :p Oct 08 23:29:23 bluelightning: it's done Oct 08 23:29:24 https://github.com/andrea-adami/meta-handheld Oct 08 23:29:41 note DEFAUL_PREFERENCE = "-1" Oct 08 23:29:49 +T Oct 08 23:31:22 somehow there is extra-work declaring the frags in src_uri *and* including them in the specific .scc. I'll need Bruce's help here Oct 08 23:31:52 Yocto would prefer to have a single copy of each file in the dir, like in a git repo Oct 08 23:32:13 but the point was sharing the frags... Oct 08 23:34:38 great, thanks, I'll have a closer look at it tomorrow Oct 08 23:35:43 zZZ now.. Oct 08 23:35:45 gn **** ENDING LOGGING AT Tue Oct 09 02:59:59 2012