**** BEGIN LOGGING AT Sun Oct 20 02:59:58 2013 Oct 20 22:35:13 hi lumag Oct 20 22:35:22 bluelightning, last round of patches Oct 20 22:35:34 hi ant_home. Oct 20 22:35:59 Oh, btw: got a problem with kexecboot 0.6 on tosa/tmiofb Oct 20 22:36:01 bluelightning, I've kept low profile this time Oct 20 22:36:18 lumag, I've seen the kb patch btw Oct 20 22:36:33 what was the problem? Oct 20 22:37:05 ant_home, kexecboot changed mmap() argument from width * bpp * height to line_length * height. Oct 20 22:37:41 hm.. I remember it, was zeldin's patch? Oct 20 22:37:43 Unfortunately w/o VT (tty being active) on framebuffer, no mode is set (technically tmiofb_set_par is no called) and thus line_length = 0. Oct 20 22:38:50 My intuitive patch to add set_par() call to tmiofb_probe() was declined by framebuffer maintainer. Oct 20 22:39:02 s/intuitive/naive/ Oct 20 22:39:11 with the reason 'fix kexecboot instead' Oct 20 22:39:23 you mean this one? Oct 20 22:39:24 https://github.com/kexecboot/kexecboot/commit/7d96d9a6d1d00b137416d320a5b00e385513d889 Oct 20 22:47:15 I think proper fix would be first to (try to) set mode and only after that to mmap image. Oct 20 22:47:33 ant_home, no Oct 20 22:49:54 Hmm. Can't find it. Oct 20 22:50:57 Maybe I'm mistaken and it is just a coincidence of kernel and kexecboot simultaneous update. Oct 20 22:52:54 hm.. maybe I've imported the 3.10 recipe too quickly Oct 20 22:53:33 prolly I didn't reconfigure it properly Oct 20 22:54:05 I did long tests on with kernel 3.2 and some with 3.8 Oct 20 23:02:27 http://www.spinics.net/lists/linux-fbdev/msg12696.html Oct 20 23:04:21 ok, we add it to linux-yocto-tiny-kexecboot then? Oct 20 23:16:31 ant_home, I think so Oct 20 23:17:17 it sems fb.c was long declaring Oct 20 23:17:18 fb.base = (char *) mmap((caddr_t) NULL, Oct 20 23:17:18 /*fb_fix.smem_len */ Oct 20 23:17:18 fb.stride * fb.height, Oct 20 23:22:42 lumag: now with 3.12-rc6 all 'known' patches for Zaurus have been merged Oct 20 23:23:14 pending is: locomokbd_hack (maybe we can send upstream ?) Oct 20 23:23:53 is just for kexecboot, normally there are keymaps loaded in userspace Oct 20 23:24:19 then, for collie, I'm trying to revive the SD/MMC driver Oct 20 23:24:43 ant_home, oh :) Oct 20 23:24:52 but the framebuffer has still frequency issues Oct 20 23:25:12 it is an spi mode driver, isn't it? Oct 20 23:25:20 yes Oct 20 23:25:34 ant_home, I have collie, but I don't want to brick it. Oct 20 23:26:05 Do you know of a way to boot original 2.4 and then kexec contnemporary kernels? Oct 20 23:26:09 lumag, oh, and collie MTD is still Read Only :/ Oct 20 23:26:26 yes, you copy the old zImage on CF/SD Oct 20 23:26:35 in /boot Oct 20 23:26:53 and then boot.cfg redirects to another partition Oct 20 23:27:04 i.e. root=/dev/mtd2 Oct 20 23:28:14 ant_home, you mean to flash kexecboot? Oct 20 23:28:27 yes, but the old one Oct 20 23:28:38 the 2.6.31 on kexecboot.org Oct 20 23:28:59 * lumag would like to have original 2.4. Oct 20 23:29:08 as a primary OS Oct 20 23:29:24 yes, wait a min Oct 20 23:29:35 http://brooknet.no-ip.com/~lex/public/zaurus/tosa/ Oct 20 23:29:44 he talks about collie, he did the same Oct 20 23:30:50 what I fear, however, are the differences wrt jffs2 Oct 20 23:31:02 iirc collie original was cramfs 2.4 Oct 20 23:32:16 I have the 'ROM' here if you need Oct 20 23:33:11 lumag, this is today's situation on collie Oct 20 23:33:13 http://paste.debian.net/59988/ Oct 20 23:33:43 * lumag did not turn collie on for about two years. Oct 20 23:34:21 as Read Only is good :p Oct 20 23:34:48 and err.. locomo-spi is in progress... Oct 20 23:35:13 at least is compiling now Oct 20 23:35:52 LinusW and Broonie say that need to be converted to the new buffered interface and do_one_transfer Oct 20 23:36:30 ant_home, I can look on that if/when I have time. But I won't promise that I will have it anytime soon. Oct 20 23:36:48 thx Oct 20 23:36:55 I send all patches to OE Oct 20 23:37:37 Regarding patches: I have sent few fixes for pxa/ASoC. At least tosa does not resume properly w/o some of them. Oct 20 23:38:52 I have marked the patches on LAKML Oct 20 23:39:07 do these apply to 3.10 as well? Oct 20 23:40:53 Yes Oct 20 23:41:16 I did not send a meta-hh pull request/patches yet. Oct 20 23:41:51 btw there is linux-yocto-dev: will soon be the 3.12-rcX Oct 20 23:42:08 is still 3.10 iirc so is nonsense to use it Oct 20 23:42:23 I tend to use self-build kernels. Oct 20 23:42:38 Because it is much easier to tune debug options/features/etc. Oct 20 23:42:43 sure but linux-yocto is ok for future maintainance Oct 20 23:42:56 now it is well integrated with OE Oct 20 23:43:02 Yes. Oct 20 23:43:11 and I can put the common patches back to SRC_URI Oct 20 23:43:20 (it didn't work back then) Oct 20 23:43:53 and the KERNEL_FEATURES fragments as well fwiw Oct 20 23:45:27 the idea is having a modular set of features whit the minimum in .cfg Oct 20 23:46:04 but surely 3MB for a collie kernel is a bit too much ;) Oct 20 23:47:37 * ant_home packing luggages for tomorrow Oct 20 23:47:39 gn **** ENDING LOGGING AT Mon Oct 21 02:59:58 2013