**** BEGIN LOGGING AT Sat Nov 15 03:00:01 2014 Nov 15 10:28:49 gm lumag_ Nov 15 10:28:56 hello ant__ Nov 15 10:29:40 I'll try to repartition with SanDisk tool from Windows Nov 15 10:30:29 Let's check if that helps. Nov 15 10:34:23 heh..3.18-rc4 won't boot anymore Nov 15 10:34:36 wtf... Nov 15 10:34:56 from CF I mean Nov 15 10:39:15 let's try on poodle... Nov 15 10:40:06 ? Nov 15 10:40:24 Any dmesg for 3.18-rc4? Nov 15 10:40:38 3.18-rc4 + 16patches + collie gpio-charger Nov 15 10:40:43 dead Nov 15 10:40:56 Strange. Nov 15 10:41:13 no kexec even Nov 15 10:41:21 i.e. no fb refresh Nov 15 10:41:31 do you have dmesg? Nov 15 10:42:14 yes, from 2 days ago Nov 15 10:43:05 http://pastebin.com/tbu1KF1x Nov 15 10:43:38 So the kernel boots, but doesn't touch lcd? Nov 15 10:44:01 I don't have hardware till Monday. Then I will take a look anyway. Nov 15 10:48:53 http://pastebin.com/ktMN01HT Nov 15 10:48:56 poodle Nov 15 10:50:43 * lumag_ plans to papare pxa drivers for COMMON_CLK patches by Robert. Nov 15 10:50:45 and still no response from rmk. Nov 15 10:54:30 I have to recheck poodle Nov 15 10:54:34 did not add CONFIG_SPI_LOCOMO=y Nov 15 10:54:34 CONFIG_GPIO_LOCOMO=y Nov 15 10:54:41 ant__, can you send me ls -lR /sys/devices on collie. Nov 15 10:54:50 for poodle there is no need to enable SPI_LOCOMO Nov 15 10:54:52 if it boots... Nov 15 10:54:55 pxamci is used instead Nov 15 10:55:13 ant__, so it remains silent on serial console? Nov 15 10:55:33 brb in 5 min Nov 15 10:55:36 yes, it seems it doesn't even kexec Nov 15 11:10:22 very strange. Nov 15 11:21:00 ant__, for the reference, could you please send me your kernel image. Nov 15 11:42:50 ok Nov 15 11:43:16 I'll push on github Nov 15 12:02:14 lumag_, uploading on https://github.com/andrea-adami/locomo Nov 15 12:02:42 ant__, just send via e-mail. Nov 15 12:03:14 here the patches (look also under ./test Nov 15 12:03:19 https://github.com/andrea-adami/meta-handheld/tree/master/recipes-kernel/linux/linux-3.18-rc4 Nov 15 12:03:33 will take a look in the evening. Nov 15 12:03:49 ok, thx Nov 15 12:11:04 this is poodle with adjusted defconfig Nov 15 12:11:06 http://pastebin.com/zHdRKMfv Nov 15 12:12:15 locomo locomo.0: LoCoMo Chip: 7649 Nov 15 12:13:14 white display Nov 15 12:32:18 ok, will take a look Nov 15 12:32:40 poodle has CONFIG_KEYBOARD_GPIO=y but I don't see it in the log.. Nov 15 12:36:10 ant__, I will take a look. Nov 15 12:36:30 I don't have poodle configuration in qemu, so I can not check that regularly. Nov 15 12:37:43 ok, I'd say one of the locomo patches blanks display: happens on both devices Nov 15 12:38:54 is that with s/BLANK_NORMAL/BLANK_UNBLANK/ ? Nov 15 12:39:50 refreshed/reposted all my pending/ready patches. Nov 15 12:40:33 it happens almost immediately after kexec Nov 15 12:41:40 no changes after susp/resume Nov 15 12:41:57 Maybe I missed something on poodle. I will recheck 2.4 kernel sources. Nov 15 12:42:17 I still have old version? Nov 15 12:44:37 hm.. + lcd_set_power(lcd_dev, FB_BLANK_NORMAL); Nov 15 12:45:07 on probe Nov 15 12:45:17 can be this Nov 15 12:45:43 yes Nov 15 12:45:56 bbl Nov 15 15:21:54 ge ant__ Nov 15 15:23:08 ge Jay7 Nov 15 15:52:49 wb Nov 15 15:57:12 ok, some progress... Nov 15 15:57:48 changed -> FB_UNBLANK in probe but still white on boot Nov 15 15:58:21 but on resume fb is ok Nov 15 15:58:30 writew(LOCOMO_DAC_SCLOEB | LOCOMO_DAC_SDAOEB, lchip->base + LOCOMO_DAC); Nov 15 15:58:35 ^ must be this Nov 15 16:00:47 Is this on poodle? Nov 15 16:01:07 yes Nov 15 16:01:23 No, DAC is a command-oriented bus (i2c), so that command should not do anything (except bringing bus to 'idle' state). Nov 15 16:01:43 seems it fixes fb after susp/resume Nov 15 16:02:07 one or the other change... Nov 15 16:02:53 s/fb/lcd I guess Nov 15 16:03:15 Try adding lcd->is_on = 1; locomolcd_on(lcd); before devm_lcd_device_register() Nov 15 16:03:19 in locomo_lcd.c Nov 15 16:08:04 Also in keyboard related patch there are two lines like writew(0, locomokbd->base + LOCOMO_KEYBOARD + LOCOMO_KIC); Nov 15 16:10:01 Replace that with writew(0, locomokbd->base + LOCOMO_KSC); writew(0, locomokbd->base + LOCOMO_KIC); Nov 15 16:13:32 replace one of those writew calls and drop the second. Nov 15 16:13:43 I did three errors in two lines :) Nov 15 16:33:19 should I still patch locomo_resume? Nov 15 17:08:15 "good to have" but not strictly required. Nov 15 17:08:56 is ok now Nov 15 17:09:07 poodle boots w/o issues? Nov 15 17:09:08 Good. Nov 15 17:09:17 lcd->is_on = 1; locomolcd_on(lcd); Nov 15 17:09:19 ^ Nov 15 17:09:22 yeah. Nov 15 17:09:25 Great Nov 15 17:09:57 ok even after resume Nov 15 17:10:00 :) Nov 15 17:11:15 I'll update github so you can verify the diff Nov 15 17:11:33 (hacking patch by hand...) Nov 15 17:15:29 what about that DAC? Nov 15 17:16:04 pls check out this patch Nov 15 17:16:06 https://github.com/andrea-adami/meta-handheld Nov 15 17:16:10 top one Nov 15 17:39:04 built and tested Nov 15 17:39:34 survives to multiple suspend/resume Nov 15 17:41:13 yes, that is ok. Nov 15 17:44:51 You can insert DAC after writew to LOCOMO_TADC in locomo_resume function. Nov 15 18:01:08 ok, pls check the second patch hijack Nov 15 18:02:01 I've horribly stolen two blank lines... Nov 15 18:08:52 btw the lcd does a 'flash' before suspending Nov 15 18:09:31 it becomes white for an instant Nov 15 18:10:01 not on resume Nov 15 18:10:44 just before suspending Nov 15 18:10:59 it's short but visible Nov 15 18:11:06 (we can live with it) Nov 15 18:59:45 ant__, regarding display. It is normal, unfortunately. Nov 15 19:00:54 It gets white when the LCD image is disabled, but backlight is not. Nov 15 19:28:48 ok, I suspected Nov 15 19:29:20 can you reverse the order of disbling? Nov 15 20:00:54 The order or disabling depends on the order in which devices get probed. Nov 15 20:01:50 And LCD drivers come firs in drivers/video/backlight/Makefile. Nov 15 20:02:03 So they have higher chances to get probed first. Nov 15 20:02:27 Probably it just does not make sense to may attention to that. Nov 15 20:02:34 We have many more interesting problems. Nov 15 20:06:15 ant__, And LCD drivers come firs in drivers/video/backlight/Makefile. Nov 15 20:06:15 So they have higher chances to get probed first. Nov 15 20:06:15 Probably it just does not make sense to may attention to that. Nov 15 20:06:15 We have many more interesting problems. Nov 15 20:09:16 ah, ok Nov 15 20:09:34 sure **** ENDING LOGGING AT Sun Nov 16 02:59:59 2014