**** BEGIN LOGGING AT Mon Nov 11 02:59:58 2013 Nov 11 09:26:40 lumag_: hi. Also even cpu_freq sa11x0-fb sa11x0-fb: dma period = 1372168 ps, clock = 0 kHz Nov 11 09:26:53 no changes with or without Nov 11 09:27:08 Yes, that part really has a bug Nov 11 09:27:19 It still calls cpufreq_getrate() Nov 11 09:27:26 even w/o cpu freq Nov 11 09:27:31 so thesing was right with that hack Nov 11 09:27:52 ...undocumented like the rest... Nov 11 09:28:12 If I have time today, I will reimplement this (and maybe PCMCIA needs changes) as a proper clk_* calls Nov 11 09:28:21 for the reference: pxa uses clk_* Nov 11 11:12:58 lumag_: I know what means Status Register 8080 :) Nov 11 11:13:34 appendix page 24: 0x80 = 01010000 Nov 11 11:13:42 the bits are SR4 and SR6 Nov 11 11:14:07 SR4 = Erroro in page buffer program Nov 11 11:14:22 SR6 = Block erase suspended Nov 11 11:14:36 So there is an error buffer erase suspension? Nov 11 11:15:59 there is a command 0x50 to Clear Status Register Nov 11 11:16:50 but this command is invalid during block erase on suspend or (page buffer) program suspend Nov 11 11:17:14 noe, maybe the code sends reset it in the wrong place for this chip Nov 11 11:19:10 "when erasing multiple blocks or programming several words in sequence, clear this bits before starting each operation" Nov 11 11:53:55 ant_work, did you ask dwmw2? Nov 11 11:54:24 Because it's really over my knowledge of mtd Nov 11 11:55:55 at least seems is not a problem with vpp (scoop gpio 35) Nov 11 11:56:17 yes. Nov 11 11:57:20 this would be bit 4 SR.3 Nov 11 11:57:49 I'll try /* #define CMDSET0001_DISABLE_ERASE_SUSPEND_ON_WRITE */ Nov 11 11:58:05 not sure it is the same thing Nov 11 11:58:27 or even Nov 11 11:58:29 // debugging, turns off buffer write mode if set to 1 Nov 11 11:58:29 43 #define FORCE_WORD_WRITE 0 Nov 11 11:58:40 so we get out page-mode Nov 11 11:59:55 /* Some Intel Strata Flash prior to FPO revision C has bugs in this area */ Nov 11 12:00:12 maybe we have the same bug Nov 11 12:00:51 I'll try to catch dwmw2 ;) Nov 11 12:12:40 For the ERASE_SUSPEND_ON_WRITE you will also need to enable fixup_intel_strataflash for our flashes Nov 11 12:14:29 yes..it's not a startaflash..is a dual-page, is another generation, though Nov 11 12:15:22 there are nasty notes about suspending interleabed chips in cmdset_0001... Nov 11 12:39:08 hmm, I found som emore traces Nov 11 12:39:09 I suppose you don't have CONFIG_MTD_XIP enabled. Nov 11 12:39:10 So in order to completely disable the suspend, Nov 11 12:39:10 please in chip_ready() function just after "case FL_ERASING:" Nov 11 12:39:10 (line 775) put the line "goto sleep;" and repeat your test. Nov 11 13:07:53 lumag_: linusw sent the patches to oe-devel ML Nov 11 13:09:46 and I just merged them :) Nov 11 13:10:00 :) thx Nov 11 13:19:04 ant_work: so I should be able to build an image and boot it on collie now? Nov 11 13:19:36 yes but display is still ko so only serial console Nov 11 13:19:42 ah ok Nov 11 13:19:58 freq = 0 kHz ;) Nov 11 13:20:03 heh Nov 11 13:20:40 fwiw you can blindly login as root and type reboot. It does Nov 11 13:21:03 console is alive Nov 11 13:21:50 and it would be interesting if you'd try to flash a core-image-minimal ubifs Nov 11 13:22:12 maybe this particular chips are dying Nov 11 13:22:56 (I doubt, old ROM works ok) Nov 11 13:25:01 bluelightning: by chance, try the steps at the bottom of http://lists.infradead.org/pipermail/linux-mtd/2013-November/049867.html Nov 11 13:26:24 I do that copying rootfs.ubifs on the CF card and booting from a core-image-base on it Nov 11 13:27:34 * ant_work must absolutely shrink the kernel so core-image-base.ubifs fits on collie..just 1-2 MB Nov 11 17:48:23 http://www.youtube.com/watch?v=1zQrY0NDsTg kexecboot on the s7 http://threader.zapto.org/experimental/s7/kexecboot-recovery/SDC11407-.png video is processing though, i elected to stabilise it, apparently im unstable Nov 11 21:31:43 hello ant_home Nov 11 21:33:30 hi lumag Nov 11 21:34:30 from bikerer: http://www.youtube.com/watch?v=1zQrY0NDsTg Nov 11 21:40:25 I think on big screen kexecboot looks better when resized Nov 11 21:45:26 lumag: all the usual 3 patches + sa1100 fb debug or only the 2 patches? Nov 11 21:46:07 2 patches I've sent + m62332 + sa1100 fb debug (to be sure that freq is correct) Nov 11 21:46:22 k Nov 11 21:47:36 no locomolcd? no backlight to the end? Nov 11 21:47:54 no. I hope they will be not necessary Nov 11 21:48:04 got it, building now Nov 11 21:50:59 hm.. drivers/video/sa1100fb.c: patch does not apply Nov 11 21:51:10 are you on 3.10? Nov 11 21:52:10 3.12 Nov 11 21:52:16 Will rebase to 3.10 Nov 11 21:54:14 sent updated second patch Nov 11 22:04:10 thanks Nov 11 22:06:19 if yocto waits longer we'll move to linux-yocto-dev (3.12-rc5 today) Nov 11 22:28:54 lumag: http://pastebin.com/L9wW22iC Nov 11 22:29:06 still 0kHz Nov 11 22:29:29 Hmm. Nov 11 22:30:34 Will recheck locally with 3.10 in a few minutes Nov 11 22:30:53 I'm sure the patch are applied... Nov 11 22:31:22 I'll try to catch dwmw2 now ;) Nov 11 22:46:36 ant_home, Strange Nov 11 22:46:43 In qemu I got sa11x0-fb sa11x0-fb: dma period = 1372168 ps, clock = 59000 kHz Nov 11 22:47:21 Moreover I think it should be impossible to get clock = 0 Nov 11 22:48:24 well, I see the patches applied in the linux-yocot .meta dir Nov 11 22:48:45 wait Nov 11 22:48:50 now I have fb... Nov 11 22:48:58 maybe I took an old image? Nov 11 22:49:03 \o/ Nov 11 22:49:29 so you have an fb? Nov 11 22:49:37 I don't hve sa debug but screen is ok Nov 11 22:49:39 yes Nov 11 22:49:45 Great Nov 11 22:50:00 I'm testing flash quirks now... Nov 11 22:50:04 Any lcd/backlight issues? Nov 11 22:50:13 how to check? Nov 11 22:50:25 visually :) Nov 11 22:50:38 Either everything works or it does not Nov 11 22:50:46 well, poodle is much better quality.. I think for collie this 'poor' fb isok ;) Nov 11 22:51:03 send a pic Nov 11 22:54:01 I'll send linux-yocto update right now Nov 11 22:55:53 great Nov 11 22:55:54 seems ok Nov 11 22:56:17 I wonder, why didn't you see any issues with pcmcia. Nov 11 22:56:18 the transition is ok, kexecboot-kernel.logo, kernel.logo, psplash Nov 11 22:56:31 Because it also had the very same issue Nov 11 22:56:34 I'm booting pcmcia Nov 11 22:57:03 Yes, I remember Nov 11 22:57:15 Maybe it was more robust. Nov 11 22:57:36 I will sent a pcmcia patch somewhere soon. Nov 11 22:58:05 ah..lcd stays black after suspend Nov 11 22:58:57 black, but do you see picture on it? Nov 11 22:59:08 seems off Nov 11 22:59:32 try using some flashlight, if you have one. Nov 11 22:59:57 So it might be a problem of framebuffer, backlight or an lcd locomo part Nov 11 23:00:18 I see the lcd characters so is on Nov 11 23:00:33 so just no light. Nov 11 23:00:57 yes Nov 11 23:01:17 echo 1 > /sys/class/graphics/fb0/blank ; echo 0 > /sys/class/graphics/fb0/blank Nov 11 23:01:23 strange, yesterday's logs show backlight is turned on Nov 11 23:02:06 this sequence: Nov 11 23:02:11 http://pastebin.com/Ui76AT5y Nov 11 23:02:22 (yesterday's log) Nov 11 23:02:50 blank/unblank does not help ? Nov 11 23:03:03 (use the commandline above) Nov 11 23:03:25 you can ignore sa11x0-fb talking about backlight, Nov 11 23:03:34 locomolcd uses different approach. Nov 11 23:03:43 ahh Nov 11 23:04:19 hm.. seems backlight is off even after cmdline reboot... Nov 11 23:04:30 logo and psplash ok but dark Nov 11 23:04:45 suspend messed with some register Nov 11 23:04:48 Hmm. Really strange. Nov 11 23:05:01 or maybe some power saving Nov 11 23:06:22 hard reboot now, all is ok Nov 11 23:06:55 ok, I really want to nest now on a real hw. Nov 11 23:07:39 copy your 2.4 kernel and put it on a sd so you'll keep the old ROM in mtd Nov 11 23:07:46 I will try tomorrow to flash 2.6.31 kexecboot Nov 11 23:07:52 yes Nov 11 23:08:21 you'll edit boot.cfg to reboot the old rom Nov 11 23:08:43 being the mtd partitions are the same this should just work Nov 11 23:09:05 ok Nov 11 23:09:11 cramfs is detected Nov 11 23:09:56 How should I flash kexecboot? Nov 11 23:11:18 press reset when pressing C+D, zImage on FAT16 card Nov 11 23:11:36 fat16. fat16 Nov 11 23:11:43 I will have to remember Nov 11 23:12:37 or flashcpy from shell Nov 11 23:12:54 (like nandwrite but for NOR) Nov 11 23:13:33 cfi_cmdset_0001: Suspend erase on write disabled. Nov 11 23:13:34 Using buffer write method Nov 11 23:13:34 Using auto-unlock on power-up/resume Nov 11 23:13:38 now... Nov 11 23:14:49 Will it blink anyhow? Nov 11 23:17:22 I don't have the commands on shell Nov 11 23:17:45 now lcd it is off by itself, power man Nov 11 23:18:00 pressing cursor resumes it. ok Nov 11 23:18:09 is only after apm --suspend Nov 11 23:19:42 hmm..interesting Nov 11 23:19:44 Bs (14272896 bytes, 13.6 MiB), LEB size 130944 bytes (127.9 KiB) Nov 11 23:19:45 root@collie:~# ubimkvol /dev/ubi0 -N rootfs -m Nov 11 23:19:45 Set volume size to 14272896 Nov 11 23:19:45 SR.4 or SR.5 bits set in buffer write (status 23494255). Clearing. Nov 11 23:19:45 Volume ID 0, size 109 LEBs (14272896 bytes, 13.6 MiB), LEB size 130944 bytes (127.9 KiB), dynamic, name "rootfs", alignment 1 Nov 11 23:19:46 root@collie:~# Nov 11 23:20:07 seems it corrects the status register Nov 11 23:20:29 maybe will be a bit slower..I'll do some more tests Nov 11 23:22:13 So just "suspend erase" disabled, or any other fixups? Nov 11 23:22:20 seems we fight an errata... Nov 11 23:22:43 yes#define CMDSET0001_DISABLE_ERASE_SUSPEND_ON_WRITE Nov 11 23:22:58 uncommenting it adds unconditionally the fixup Nov 11 23:23:14 needs proper patch but let see if I can boot ubifs finally... Nov 11 23:28:06 root@collie:~# ubiupdatevol -t /dev/ubi0_0 Nov 11 23:28:06 SR.4 or SR.5 bits set in buffer write (status ffffffff). Clearing. Nov 11 23:28:06 root@collie:~# Nov 11 23:28:59 yes! Nov 11 23:29:00 root@collie:~# ubiupdatevol /dev/ubi0_0 /rootfs.ubifs Nov 11 23:29:01 root@collie:~# Nov 11 23:30:55 hm.. i not detected.. Nov 11 23:30:57 is Nov 11 23:35:51 ubi seems ok Nov 11 23:35:51 http://pastebin.com/NyuczPry Nov 11 23:36:40 maybe UBI API has changed btw 2.6.31 and 3.10 Nov 11 23:37:08 anyway, collie needs just a new linux-kexecboot now Nov 11 23:37:12 ;) Nov 11 23:44:49 ok, Nov 11 23:44:50 collie login: root Nov 11 23:44:50 root@collie:~# flashcp Nov 11 23:44:50 Flash Copy - Written by Abraham van der Merwe Nov 11 23:44:50 usage: flashcp [ -v | --verbose ] Nov 11 23:44:51 flashcp -h | --help Nov 11 23:44:53 -h | --help Show this help message Nov 11 23:44:55 -v | --verbose Show progress reports Nov 11 23:44:57 File which you want to copy to flash Nov 11 23:44:59 Flash device to write to (e.g. /dev/mtd0, /dev/mtd1, etc.) Nov 11 23:45:01 root@collie:~# flashcp -v /initrd.bin /dev/mtd2 Nov 11 23:45:03 Erasing blocks: 50/50 (100%) Nov 11 23:45:05 Writing data: 6400k/0k (100%)) Nov 11 23:45:07 Verifying data: 6400k/0k (100%)) Nov 11 23:45:09 root@collie:~# Nov 11 23:45:11 no failures Nov 11 23:45:41 (I'm doing tests with unpadded images with flashcp) Nov 11 23:47:24 root@collie:~# reboot Nov 11 23:47:24 sa1100: Chip not ready for buffer write. Xstatus = 80808080, status = 80808080 Nov 11 23:47:24 -sh: reboot: Input/output error Nov 11 23:47:55 this was the old kernel kexecbooted by the new one Nov 11 23:48:15 I'l lrebuild from scratch now Nov 12 00:43:46 gn **** ENDING LOGGING AT Tue Nov 12 02:59:58 2013