**** BEGIN LOGGING AT Sat Nov 09 02:59:58 2013 Nov 09 10:30:01 lumag: I'm building IrDA Nov 09 11:10:41 lumag: something happens Nov 09 11:10:43 root@collie:~# irattach /dev/ttySA1 Nov 09 11:10:44 root@collie:~# NET: Registered protocol family 23 Nov 09 11:10:44 root@collie:~# lsmod Nov 09 11:10:44 Module Size Used by Nov 09 11:10:44 irda 101096 0 Nov 09 11:10:44 root@collie:~# Nov 09 12:42:07 hi lumag_ Nov 09 12:42:31 Now force MSC0 = 0xfff8fff8 -> no MTD http://pastebin.com/zSpGWPPf Nov 09 15:18:56 same with MSC0 = 0xfff8fffa Nov 09 15:18:58 http://pastebin.com/xqH6PXqk Nov 09 16:17:11 Creating 4 MTD partitions on "sa1100": Nov 09 16:17:11 0x000000000000-0x0000000c0000 : "bootloader" Nov 09 16:17:11 0x0000000c0000-0x0000001c0000 : "kernel" Nov 09 16:17:11 0x0000001c0000-0x000000fe0000 : "rootfs" Nov 09 16:17:11 0x000000fe0000-0x000001000000 : "angel stuff" Nov 09 16:17:15 ^^ Nov 09 16:17:45 maybe call it bootblock Nov 09 16:28:55 0001C000 43 4D 41 44 │ 9D 00 00 00 │ FF FF FF FF │ FF FF FF FF CMAD....�������� Nov 09 16:28:55 0001C010 FF FF FF FF │ FF FF FF FF │ FF FF FF FF │ 54 55 43 48 ������������TUCH Nov 09 16:28:55 0001C020 DF F6 8C 3E │ C4 C3 C3 3E │ EA FF FF FF │ D1 FF FF FF ��.>���>�������� Nov 09 16:30:37 https://github.com/andrea-adami/z-prom Nov 09 17:38:08 I have to show to lumag the values are swapped Nov 09 17:38:25 reading were Nov 09 17:38:27 Nov 01 21:35:21 SHARPSL PARAMS Nov 09 17:38:27 Nov 01 21:35:22 COMADJ 9d Nov 09 17:38:27 Nov 01 21:35:22 PHAD ffffffff Nov 09 17:38:27 Nov 01 21:35:22 UUID ����������������TUCH���>���>���������������������������� Nov 09 17:38:30 Nov 01 21:35:23 TOUCH XP 3e8cf6df Nov 09 17:38:32 Nov 01 21:35:25 TOUCH YP 3ec3c3c4 Nov 09 17:38:34 Nov 01 21:35:27 TOUCH XD ffffffea Nov 09 17:38:36 Nov 01 21:35:29 TOUCH YD ffffffd1d Nov 09 17:38:38 Nov 01 21:35:31 ADADJ ffffffff Nov 09 17:38:58 bluelightning: around? Nov 09 17:39:39 ant_home: about to go out... what's up? Nov 09 17:39:52 np, about couple of patches Nov 09 17:40:07 so lumag will build same kernel Nov 09 17:40:14 later is ok as well, np Nov 09 17:40:58 ok, I'll be back later / around for some of tomorrow Nov 09 17:41:22 enjoy Nov 09 17:51:32 irda seems ok Nov 09 17:51:35 collie login: root Nov 09 17:51:35 root@collie:~# modprobe irda Nov 09 17:51:35 NET: Registered protocol family 23 Nov 09 17:51:35 root@collie:~# modprobe ircomm Nov 09 17:51:35 IrCOMM protocol (Dag Brattli) Nov 09 17:51:35 root@collie:~# modprobe ircomm-tty Nov 09 17:51:37 root@collie:~# irattach /dev/ttySA1 -s Nov 09 17:51:39 root@collie:~# lsmod Nov 09 17:51:41 Module Size Used by Nov 09 17:51:43 ircomm_tty 18877 0 Nov 09 17:51:45 ircomm 9528 1 ircomm_tty Nov 09 17:51:47 irda 101096 2 ircomm_tty,ircomm Nov 09 17:51:49 root@collie:~# /dev/i Nov 09 17:51:51 initctl ircomm11 ircomm16 ircomm20 ircomm25 ircomm3 ircomm6 Nov 09 17:51:53 input/ ircomm12 ircomm17 ircomm21 ircomm26 ircomm30 ircomm7 Nov 09 17:51:55 ircomm0 ircomm13 ircomm18 ircomm22 ircomm27 ircomm31 ircomm8 Nov 09 17:51:57 ircomm1 ircomm14 ircomm19 ircomm23 ircomm28 ircomm4 ircomm9 Nov 09 17:51:59 ircomm10 ircomm15 ircomm2 ircomm24 ircomm29 ircomm5 Nov 09 17:52:11 I'll send patches now Nov 09 17:52:13 bbl Nov 09 22:48:55 root@collie:~# ubiformat /dev/mtd2 Nov 09 22:48:55 ubiformat: mtd2 (nor), size 14811136 bytes (14.1 MiB), 113 eraseblocks of 131072 bytes (128.0 KiB), min. I/O size 1 bytes Nov 09 22:48:55 libscan: scanning eraseblock 112 -- 100 % complete Nov 09 22:48:55 ubiformat: warning!: 113 of 113 eraseblocks contain non-ubifs data Nov 09 22:48:55 ubiformat: continue? (yes/no) y Nov 09 22:48:57 ubiformat: warning!: only 0 of 113 eraseblocks have valid erase counter Nov 09 22:48:59 ubiformat: erase counter 0 will be used for all eraseblocks Nov 09 22:49:01 ubiformat: note, arbitrary erase counter value may be specified using -e option Nov 09 22:49:05 ubiformat: continue? (yes/no) y Nov 09 22:49:07 ubiformat: use erase counter 0 for all eraseblocks Nov 09 22:49:09 ubiformat: formatting eraseblock 112 -- 100 % complete Nov 09 22:49:11 root@collie:~# ubiattach -p /dev/mtd2 Nov 09 22:49:13 UBI device number 0, total 113 LEBs (14796672 bytes, 14.1 MiB), available 109 LEBs (14272896 bytes, 13.6 MiB), LEB size 130944 bytes (127.9 KiB) Nov 09 22:49:17 root@collie:~# ubimkvol /dev/ubi Nov 09 22:49:18 ubi0 ubi_ctrl Nov 09 22:49:20 root@collie:~# ubimkvol /dev/ubi0 -N rootfs -m Nov 09 22:49:22 Set volume size to 14272896 Nov 09 22:49:24 Volume ID 0, size 109 LEBs (14272896 bytes, 13.6 MiB), LEB size 130944 bytes (127.9 KiB), dynamic, name "rootfs", alignment 1 Nov 09 22:49:27 root@collie:~# ubiupdatevol /dev/ubi0_0 /rootfs.ubifs Nov 09 22:49:29 sa1100: Chip not ready for buffer write. Xstatus = 80808080, status = 80808080 Nov 09 22:49:31 UBI error: ubi_io_write: error -62 while writing 64 bytes to PEB 87:64, written 0 bytes Nov 09 22:49:35 UBI error: vol_cdev_write: cannot accept more 13824 bytes of data, error -62 Nov 09 22:49:37 ubiupdatevol: error!: cannot write 13824 bytes to volume "/dev/ubi0_0" Nov 09 22:49:40 error 62 (Timer expired) Nov 09 22:49:41 root@collie:~# Nov 09 22:54:40 lumag: sa1100: Chip not ready for buffer write. Xstatus = 80808080, status = 80808080 Nov 09 22:55:41 What does that mean? Nov 09 22:55:55 it cannot complete flashing Nov 09 22:56:16 fwiw I've tried jffs2 before Nov 09 22:56:36 jffs2 padded images can be flashed by bootloader or with flashcpy Nov 09 22:56:54 but I have still issues with jffs2... Nov 09 22:57:14 anyway, I can format the flash Nov 09 22:57:22 I'll try the mtd-tests Nov 09 22:57:40 oh, I'm booting from flash ;) Nov 09 22:58:16 lumag: seems some timing Nov 09 22:58:50 Any progress with timings vs QRY reading Nov 09 23:00:04 not tested more Nov 09 23:00:12 I'll do that now Nov 09 23:00:39 changing MSC0 didn't change Nov 09 23:00:56 ah, collie did not reboot setting the value you gave me Nov 09 23:03:05 did not reboot or did not boot with them? Nov 09 23:03:29 from serial console, at prompt, reboot - enter Nov 09 23:03:34 frozen Nov 09 23:03:54 not happens whithout touching MSC0 Nov 09 23:04:39 have you found the links? Nov 09 23:05:15 MSC0 = 0xfff8fff8 -> no MTD http://pastebin.com/zSpGWPPf Nov 09 23:05:27 same with MSC0 = 0xfff8fffa http://pastebin.com/xqH6PXqk Nov 09 23:07:52 even with your QRY fixup/ Nov 09 23:07:53 ? Nov 09 23:07:55 Strange. Nov 09 23:08:08 no, w/out hack Nov 09 23:08:10 Maybe I' misunderstanding MSC0 settings. Nov 09 23:08:12 ah. Nov 09 23:08:15 ok then. Nov 09 23:08:42 What about adding delays to cfi probe code? Nov 09 23:08:48 I'm really puzzled here. Nov 09 23:09:07 I've read a strange sentence..maybe outdated... Nov 09 23:09:09 see: Nov 09 23:09:10 We have one chip responding properly and second _same_ chip connected in parallel not responding Nov 09 23:09:28 it woul need 3x 0xFF Nov 09 23:09:41 to be sure the cycle is finished Nov 09 23:09:53 Hmm. Nov 09 23:09:59 One can try. Nov 09 23:10:32 We can also ask Pavel Machek (IIRC he 'fixed' the old sharp.c, didn't he?) Nov 09 23:11:32 http://www.datasheetarchive.com/dl/Datasheet-019/DSA00339271.pdf Nov 09 23:11:35 page 9 Nov 09 23:12:02 I suspect ubi as well. Need to update to 3.12 Nov 09 23:15:30 Try that. Nov 09 23:16:31 it's unclear if it applies to this chips...don't think so, however... Nov 09 23:20:05 hm.. linux-mtd "nor flash bug in UBIFS" Nov 09 23:20:24 don't tell me I'm so lucky... Nov 10 00:05:33 root@collie:~# modprobe mtd_readtest dev=0 Nov 10 00:05:33 root@collie:~# modprobe mtd_readtest dev=1 Nov 10 00:05:33 root@collie:~# modprobe mtd_readtest dev=2 Nov 10 00:05:33 root@collie:~# modprobe mtd_readtest dev=3 Nov 10 00:05:38 read is ok Nov 10 00:07:51 same for mtd_pagetest, exit clean Nov 10 00:08:13 (I rmmod btw each other ofc) Nov 10 00:08:54 Mostly finished with with locomo. locomo-backlight is left on a plate. It is a real beast. Nov 10 00:09:17 well, sweet and sour Nov 10 00:09:47 running now modprobe mtd_speedtest dev=2 Nov 10 00:10:17 this is destructive Nov 10 00:10:51 I really fear that one day you will 'test' the bootloader. Nov 10 00:11:11 heh, read tests on all, the others only on 2 Nov 10 00:11:32 I'm reading the sources of the mtd-tests before Nov 10 00:11:44 :) Nov 10 00:12:47 I've uploaded the partition today on github ;) Nov 10 00:13:09 btw in mtd3 there is only few bytes Nov 10 00:13:47 the source of the bootloader is in mtd0 then Nov 10 00:14:01 s/the source/the bootloader Nov 10 00:14:28 mtd3 is that CMAD... etc? Nov 10 00:14:33 yes Nov 10 00:14:35 I think those are parameters. Nov 10 00:15:01 the bootloader copies those in ram? Nov 10 00:15:08 No. Nov 10 00:15:17 sharpsl_param reads them from flash Nov 10 00:15:26 Using virual address. Nov 10 00:15:41 We can fix sharpsl_param to always use physical address Nov 10 00:15:42 hm.. why then 'angel stuff' Nov 10 00:16:19 maybe the bootloader has access to those small eraseblocks on top Nov 10 00:16:29 collie doesn't make use of the feature Nov 10 00:17:22 lumag, have you seen how the endiannes differs? Nov 10 00:17:37 0x1234 vs 0x4321 ;) Nov 10 00:17:39 Kidding Nov 10 00:17:51 What do you mean? Endianness-switching the ARM core? Nov 10 00:18:00 iirc we added printk in sharpsl_param Nov 10 00:18:23 but the result is reversed, not just swapped 2x2 Nov 10 00:18:58 ah, wait, I was reading on my PC Nov 10 00:19:30 well, it is LE Nov 10 00:20:48 0001C000 43 4D 41 44 │ 9D 00 00 00 │ FF FF FF FF │ FF FF FF FF CMAD....�������� Nov 10 00:20:48 0001C010 FF FF FF FF │ FF FF FF FF │ FF FF FF FF │ 54 55 43 48 ������������TUCH Nov 10 00:20:48 0001C020 DF F6 8C 3E │ C4 C3 C3 3E │ EA FF FF FF │ D1 FF FF FF ��.>���>�������� Nov 10 00:21:01 and Nov 10 00:21:03 Nov 01 21:35:23 TOUCH XP 3e8cf6df Nov 10 00:21:03 Nov 01 21:35:25 TOUCH YP 3ec3c3c4 Nov 10 00:21:03 Nov 01 21:35:27 TOUCH XD ffffffea Nov 10 00:21:03 Nov 01 21:35:29 TOUCH YD ffffffd1d Nov 10 00:21:03 Nov 01 21:35:31 ADADJ ffffffff Nov 10 00:23:25 See. sa1100 is in little-endian mode. So u32 is written backwards. For example CMAD (comadj) = 0x9d + 0x00 * 256 + 0x00 * 256 ^ 2 + 0x00 * 256 ^ 3 Nov 10 00:23:31 are we reading it in the right way? Nov 10 00:23:45 wow, dense Nov 10 00:24:01 Just get used to read integers backwards :) Nov 10 00:24:13 ah, like manga ;) Nov 10 00:24:31 so TUCH[0] = 0x3e8cf6df Nov 10 00:24:54 ok, anything scrambled here Nov 10 00:25:23 nothing wrong I mean Nov 10 00:26:04 Now I need to resurrect killed thing - locomo gpio irqs Nov 10 00:26:12 lumag: I'll abuse of your patience and take a video of the lcd not refreshing Nov 10 00:26:21 Go ahead Nov 10 00:38:26 sent Nov 10 00:38:44 I don't see the console but I can log and reboot from collie kb Nov 10 00:38:53 blindly Nov 10 00:39:10 seems old frame is not deleted Nov 10 00:40:39 typically is like in img_126 Nov 10 00:40:51 rarely you distinguish the logo Nov 10 00:41:11 probably the new-kernel bootlogo Nov 10 00:41:31 or even the psplash..are all trhe same Nov 10 00:45:06 lumag: how's the patient, doc ? Nov 10 00:45:44 survived to speedtest, phew Nov 10 00:47:07 Hmm. Nov 10 00:47:20 0124 is nearly normal :( Nov 10 00:47:40 kexecboot it from 2.6.31? Nov 10 00:47:45 yes Nov 10 00:48:05 has a nice bootlogo (the same) Nov 10 00:48:28 is from kexecboot.org Nov 10 00:48:44 built with oe-classic :/ Nov 10 00:57:58 Hey, excellent stuff! got it configured and running nicely on this pad, but im having trouble when doing init=/kexecboot Nov 10 00:58:20 is that because i cannot say init=/kexecboot ? i was sure i read that somewhere Nov 10 01:00:00 hi Nov 10 01:00:06 hey ant Nov 10 01:00:12 we do a symlink in our cpio Nov 10 01:00:15 http://cgit.openembedded.org/meta-openembedded/tree/meta-initramfs/recipes-bsp/kexecboot/kexecboot_0.6.bb#n63 Nov 10 01:01:03 aha, im trying to get it working from an android mkbootfs/img Nov 10 01:01:11 had ln -s Nov 10 01:01:22 and appended init Nov 10 01:02:44 strange thing is it runs in both android and chroot debian , and can kexecboot to debian but will not run as init Nov 10 01:03:05 kexecboot checks itself Nov 10 01:03:05 maybe i missed something Nov 10 01:03:07 https://github.com/kexecboot/kexecboot/blob/master/kexecboot.c#L714 Nov 10 01:03:44 ah, i might have missed some folders you think ? Nov 10 01:04:08 maybe, try to enable debug Nov 10 01:05:08 k thanks ;) will set up a fork on git too, think the button config could be put in a header Nov 10 01:05:33 great, let us know Nov 10 01:05:46 will do , thank you Nov 10 01:11:03 ant_home, what OE commit id are you building for 2.6.31? Nov 10 01:11:21 Or you are using old build Nov 10 01:12:44 Could not find anything particular interesting in oe's patches. Nov 10 01:12:56 And nothing interesting (from my POV) in upstream changes. Nov 10 01:13:56 I built with old oe-classic Nov 10 01:14:27 I have a copy of the changes if you want Nov 10 01:14:42 but really the idea is to use oe-ore ;) Nov 10 01:14:45 +c Nov 10 01:15:15 ant_home, I just can't find 2.6.31 in oe-classic Nov 10 01:15:23 (still running mtd_stresstest fwiw ;) Nov 10 01:15:26 linux-kexecboot_2.6.31 Nov 10 01:15:48 I have the same feature-set for collie like for others Nov 10 01:15:55 I'll send you Nov 10 01:16:04 mom Nov 10 01:16:57 ant_home, do you remember: was 2.6.32 already broken? Nov 10 01:17:10 what is the earliest known-broken kernel? Nov 10 01:17:42 2.632 Nov 10 01:17:58 sent 2.6.31 Nov 10 01:18:11 there were some more patches for 2.6.31 Nov 10 01:18:25 a guy on angstrom-devel in 2011 Nov 10 01:18:37 I think I've sent to you Nov 10 01:18:41 already Nov 10 01:22:07 sent kernel+modules+patches Nov 10 01:23:41 fwiw I ahve 2.6.32 , .38, .39 with patches for collie Nov 10 01:23:52 all have same issue Nov 10 01:24:29 no worth Nov 10 01:26:23 (still running stresstest...) Nov 10 01:28:46 ant_home, Do you have cpu_freq enabled? Nov 10 01:29:45 yes Nov 10 01:30:07 Try disabling that Nov 10 01:30:41 ah Nov 10 01:30:55 that was changed with 2.6.32/2.6.33 iirc Nov 10 01:31:15 thesing patches to enable it Nov 10 01:31:33 I'll do if collie survives the mtd-tests :p Nov 10 01:33:34 I'm worried about you wearing nor completely Nov 10 01:34:42 ;) Nov 10 01:34:57 torturetest for tomorrow (hopefully) Nov 10 01:35:35 sadly the led doesn't signal activity on mtd Nov 10 01:38:11 Is it still alive? Nov 10 01:38:57 yes, my bad I did not add debug to cmdline...vanilla image Nov 10 01:39:08 pr_info("%d operations done\n", op); Nov 10 01:39:16 need to check dmesg later Nov 10 01:39:46 "Number of operations to do (default is 10000)" Nov 10 01:39:55 we were chatting, I left defaults... **** ENDING LOGGING AT Sun Nov 10 02:59:58 2013