**** BEGIN LOGGING AT Fri Nov 21 02:59:58 2014
Nov 21 08:37:43 <lumag> ant_work, hello
Nov 21 08:37:49 <lumag> the -517 error is normal.
Nov 21 08:38:02 <lumag> It's the gpio not being available at the time the console starts.
Nov 21 08:40:11 <ant_work> gm
Nov 21 08:40:37 <ant_work> is disturbing the bootlogo
Nov 21 08:40:56 <ant_work> even with 'quiet'
Nov 21 08:41:57 <ant_work> about PM, maybe some clock change broke resume?
Nov 21 08:42:17 <ant_work> I'll retest 3.17 but we have logs, it was ok
Nov 21 08:43:29 <ant_work> http://pastebin.com/XeqZnNgq
Nov 21 08:58:23 <lumag> ant_work, I will recheck with 3.18-rc5 right now.
Nov 21 08:58:34 <lumag> The worst thing that we can get is 'works for me'
Nov 21 08:58:48 <ant_work> test on qemu or collie?
Nov 21 09:04:12 <lumag> collie, of course
Nov 21 09:05:02 <ant_work> I don't trust much the hw, mostly the batteries can be a problem
Nov 21 09:05:15 <ant_work> but thei skind of repeatable fault is strange...
Nov 21 09:05:43 <lumag> ok. The collie booted, CF is mounted.
Nov 21 09:05:47 <lumag> What should I do next?
Nov 21 09:05:52 <lumag> poweroff or echo mem > ?
Nov 21 09:05:53 <ant_work> # poweroff
Nov 21 09:06:03 <ant_work> same (bad) effects ;)
Nov 21 09:06:13 <lumag> It halted. of some kind
Nov 21 09:06:20 <lumag> Now resetting unit
Nov 21 09:06:31 <ant_work> or even # shutdown -h now
Nov 21 09:06:35 <lumag> (in halt I see the penguin and vertical line on a screen).
Nov 21 09:06:59 <lumag> Something is very bad in "poweroff". The collie sounds!
Nov 21 09:07:04 <ant_work> wow
Nov 21 09:07:04 <lumag> ok, rebooted.
Nov 21 09:07:18 <lumag> It boots fine.
Nov 21 09:07:22 <lumag> Sot "it works for me" :(
Nov 21 09:07:30 <ant_work> first kernel boots
Nov 21 09:07:34 <ant_work> try to kexec
Nov 21 09:07:36 <lumag> Yep.
Nov 21 09:07:42 <lumag> it works.
Nov 21 09:08:14 <lumag> What was the issue with suspend?
Nov 21 09:08:21 <lumag> echo mem , no resume
Nov 21 09:08:38 <lumag> or echo mem, resume, no kexec?
Nov 21 09:09:11 <lumag> Yes. resume is br0ken.
Nov 21 09:09:13 <ant_work> no resume. have to press reset. then linux-kexecboot boots but cannot kexec
Nov 21 09:09:15 <lumag> I will check.
Nov 21 09:09:33 <lumag> kexecboot..
Nov 21 09:09:43 <lumag> kexec...
Nov 21 09:09:45 <lumag> boots
Nov 21 09:09:53 <lumag> Is you kernel lying on CF or on SD?
Nov 21 09:10:12 <ant_work> on CF, I have the kernel without SD
Nov 21 09:10:20 <ant_work> kexecboot v.06
Nov 21 09:10:21 <lumag> Ah, what is your 3.18 config/
Nov 21 09:10:45 <ant_work> the one in meta-handheld master
Nov 21 09:10:59 <ant_work> on my fork I added the locomo stuff
Nov 21 09:11:38 <ant_work> Jay7: can you upload the test-kernels somewhere?
Nov 21 09:11:41 <lumag> Rebuilding with your config
Nov 21 09:12:13 <ant_work> dsame defconfig + patches of 3.17 can be used for 3.18-rc5
Nov 21 09:13:15 <ant_work> can you pls check in D+P menu the value of the B/U battery? Mine is very low
Nov 21 09:13:29 <ant_work> manual says it recharges in 30 hours
Nov 21 09:14:49 <lumag> 1
Nov 21 09:14:59 <lumag> main 648
Nov 21 09:15:01 <lumag> AIN0 464
Nov 21 09:15:16 <ant_work> 1 should ve o.0v
Nov 21 09:15:40 <ant_work> I got it up to 8 (0.1v)
Nov 21 09:16:23 <ant_work> how rapidly does 648 decreases w/out AC?
Nov 21 09:17:05 <ant_work> oafter one min one of my batts does 667->621
Nov 21 09:17:28 <ant_work> the other does 658->618
Nov 21 09:17:46 <ant_work> then it stays around that value
Nov 21 09:18:21 <ant_work> this is the initial discharge isn't? means batt cannot load 100% anymore
Nov 21 09:19:01 <lumag> Looks so,
Nov 21 09:19:05 <lumag> Doing the test right now.
Nov 21 09:19:15 <lumag> I'm not sure about B/U.
Nov 21 09:19:30 <lumag> If it charges with stock rom and discharges with our kernels, we did something wrong.
Nov 21 09:19:50 <lumag> 1 min - 634
Nov 21 09:20:07 <ant_work> good
Nov 21 09:20:26 <ant_work> in battery menu you can do more checks
Nov 21 09:20:48 <lumag> 2 min - 633
Nov 21 09:21:21 <lumag> battery?
Nov 21 09:21:27 <ant_work> yes, the curve goes flat
Nov 21 09:21:29 <lumag> Do you mean 'charge' menu item?
Nov 21 09:21:38 <lumag> 632, 631 now
Nov 21 09:21:56 <ant_work> no, 'battery check', on the top
Nov 21 09:22:00 <ant_work> maybe page 2?
Nov 21 09:22:21 <ant_work> charge does indeed switch the led on
Nov 21 09:24:43 <lumag> It said that 631 is 4.0 V
Nov 21 09:25:40 <lumag> Ok, built 3.18 + OE patches (from meta-hh only, not sure if there are any additional patches) + your defconfig
Nov 21 09:28:10 <lumag> Resume is indeed broken.
Nov 21 09:28:48 <ant_work> hmm
Nov 21 09:29:41 <lumag> Was able to resume by replugging into the cradle
Nov 21 09:30:14 <lumag> kexec still works
Nov 21 09:30:28 <lumag> kexecboot 0.5.8
Nov 21 09:30:29 <ant_work> which version do you have? does it see sd?
Nov 21 09:30:32 <lumag> * 0.5.9
Nov 21 09:30:34 <ant_work> ah, ok, 2.6.31
Nov 21 09:30:36 <lumag> Yes, it sees
Nov 21 09:30:39 <ant_work> I have 3.10
Nov 21 09:30:50 <ant_work> v. 0.6
Nov 21 09:30:54 <ant_work> fwiw
Nov 21 09:31:33 <ant_work> funny that the same rootfs on CF can be booted after C+D reflashing treatment
Nov 21 09:32:01 <ant_work> in fact after flashing it 'suspends' the unit and reset switch is needed
Nov 21 09:32:02 <lumag> It should not be a problem of rootfs.
Nov 21 09:32:15 <ant_work> same collie, same battery...
Nov 21 09:33:01 <lumag> after pressing power button the unit freezes
Nov 21 09:33:17 <ant_work> my wild guess is, some gpio is wrong (high/low)
Nov 21 09:33:29 <ant_work> or some mysterious fw register
Nov 21 09:34:03 <ant_work> touched by 3.18
Nov 21 09:34:10 <lumag> Nothing should have changed.
Nov 21 09:34:18 <lumag> of things that matter.
Nov 21 09:34:51 <ant_work> I'll retry 3.17 on the same collie
Nov 21 09:35:15 <ant_work> if it also fails I'll trash the unit and use the spare one
Nov 21 09:36:27 <ant_work> if there is an hw issue, then is very well handled by sharp rom anyway...
Nov 21 09:36:38 <lumag> What is your gcc version (inside OE)?
Nov 21 09:36:49 <ant_work> 4.9.x
Nov 21 09:38:07 <ant_work> 4.9.1
Nov 21 09:39:22 <ant_work> but the same kernel/rootfs boots after treatment so toolchain should be ok
Nov 21 09:39:32 <lumag> Could you please check if you can resume by pressing the "Sync" button on cradle?
Nov 21 09:39:49 <ant_work> oh, good one...never ever pressed ;)
Nov 21 09:39:55 <ant_work> I'll do
Nov 21 09:40:55 <ant_work> btw, why is collie so slow to kexecboot?
Nov 21 09:41:36 <lumag> Don't know.
Nov 21 09:41:39 <ant_work> on other Z kernel is launched 'immediately'
Nov 21 09:41:41 <lumag> Maybe reading is slow.
Nov 21 09:41:52 <lumag> Or maybe uncompressor is not so optimal.
Nov 21 09:42:08 <ant_work> can it be it is slow to load from mtd to ram?
Nov 21 09:42:36 <ant_work> poodle is much faster (10-20x) and has half the ram
Nov 21 09:42:57 <lumag> Maybe cache flashing is not optimal.
Nov 21 09:43:12 <ant_work> is cache enabled?
Nov 21 09:44:10 <ant_work> once loaded it is fast to uncompress
Nov 21 09:44:35 <lumag> I'll take a look
Nov 21 09:44:43 <lumag> Somewhere in the future.
Nov 21 09:44:53 <ant_work> ;)
Nov 21 09:45:02 <lumag> ok. pressing power button ("Off") halts the unit
Nov 21 09:45:10 <ant_work> remember we are using an hacked kexec-klibc
Nov 21 09:45:51 <lumag> I wonder why.
Nov 21 09:45:56 <lumag> (poewr offf)
Nov 21 09:47:40 <ant_work> ok, I see, we have to reenable PM_DEBUG
Nov 21 09:47:45 <ant_work> ^_^
Nov 21 09:48:09 <lumag> no.
Nov 21 09:48:17 <lumag> It looks more like a hardware problem.
Nov 21 09:48:38 <lumag> Maybe the GPIO line is reused somewhere.
Nov 21 09:49:09 <ant_work> iirc there have been changes to the reboot handler
Nov 21 09:49:35 <ant_work> afais poweroff/halt do   reboot: Power down
Nov 21 09:51:42 <ant_work> lumag: can you pls send me linux-kexecboot 0.5.9 ?
Nov 21 09:52:48 <lumag> Hmm. Now "uncompressing linux" takes ages.
Nov 21 09:53:30 <lumag> CONFIG_KERNEL_LZ4=y seems to work better
Nov 21 09:53:54 <lumag> Found the issue with power button :)
Nov 21 09:54:16 <lumag> At least identified.
Nov 21 09:59:44 <lumag> Kt
Nov 21 09:59:49 <lumag> It just gets stuck
Nov 21 10:07:08 <lumag> ok. you asked for kexecboot
Nov 21 10:07:16 <lumag> I'll just dump the kernel partition
Nov 21 10:09:30 <ant_work> yes pls, nandlogical
Nov 21 10:09:34 <ant_work> ;)
Nov 21 10:09:38 <ant_work> joking now
Nov 21 10:10:28 <ant_work> so you have an hw problem with your key or is it a kernel issue?
Nov 21 10:12:41 <lumag> sent
Nov 21 10:12:44 <lumag> It's a kernel issue.
Nov 21 10:13:06 <lumag> btw: I suggest to change kernel compression to lz4. It works better on my side.
Nov 21 10:15:11 <lumag> Grrr.
Nov 21 10:15:16 <lumag> It's domain IRQ handling.
Nov 21 10:15:18 <lumag> Again.
Nov 21 10:17:19 <ant_work> you mean for 'production' kernel?
Nov 21 10:17:40 <lumag> yes
Nov 21 10:17:45 <lumag> It decompresses faster.
Nov 21 10:17:48 <lumag> At least it looks so,
Nov 21 10:19:34 <lumag> Damn.
Nov 21 10:19:39 <lumag> Sorry.
Nov 21 10:31:38 <lumag> ant_work, added you in CC list regarding the power button. It will be a thread regarding IRQ0.
Nov 21 10:32:15 <lumag> But this doesn't help us with the other issues,
Nov 21 10:38:19 <ant_work> wow, you got Russels's reply at least for this
Nov 21 10:43:20 <lumag> Yeah. It seems, he responds. If it is not a patch for sa1100
Nov 21 10:43:33 <ant_work> he he
Nov 21 10:49:10 <ant_work> arnd timidly started by taking the sa1100-debug to linux-next
Nov 21 10:49:34 <ant_work> RMK has to answer now
Nov 21 12:20:45 <lumag> Unfortunately rmk wasn't helpful.
Nov 21 12:21:03 <lumag> So 1/3 of ARM arches is somewhat broken. Just a tiny bit.
Nov 21 12:54:53 <lumag> ant_work, Until we have better solution, I'd suggest to pick either of patches from IRQ0 thread as a fix for "wakeup" issue.
Nov 21 12:56:15 <ant_work> ok
Nov 21 13:00:46 <lumag> ant_work, heh. poweroff works.
Nov 21 13:00:58 <lumag> Module display not being switched off.
Nov 21 13:11:52 <ant_work> that irq commit is 3.18, that's why 3.17 was ok
Nov 21 13:23:27 <lumag> yes.
Nov 21 13:23:52 <lumag> And I'm still puzzled, what broke the suspend/kexec for you.
Nov 21 13:24:19 <lumag> And what "sounds" in collie.
Nov 21 13:26:04 <ant_work> I didn't notice any buzz
Nov 21 13:27:23 <ant_work> iirc 2.6.31 had older kb driver but maybe I'm confusing with pxa models
Nov 21 13:27:43 <ant_work> linux-kexecboot 3.10 should be equivalent (but the SD...)
Nov 21 13:28:34 <ant_work> ideally I'd like to be able to 'poweroff' the device while at kexecboot menu, like for the other z
Nov 21 13:29:03 <ant_work> (poodle restarts but this is another issue...)
Nov 21 13:30:50 <ant_work> I have already dumped the partition for a disassembly party ;)
Nov 21 13:53:33 <lumag> ant_work, kb drivers changed for pxa models.
Nov 21 13:53:58 <lumag> Mine collie sounds pretty much. But that might be due to the DAC powering the speaker.
Nov 21 13:54:57 <ant_work> pretty sure audio is not enabled in linux-kexecboot for size reasons
Nov 21 13:56:34 <lumag> Yes. But one of the dacs is.
Nov 21 13:56:48 <ant_work> maybe tosa was needing it
Nov 21 13:56:50 <ant_work> ah ok
Nov 21 13:57:11 <lumag> I'm sketching over defconfig in oe.
Nov 21 13:57:35 <ant_work> yes pls, give a check
Nov 21 13:57:56 <lumag> Well the defconfig resulted from oe-defconfig
Nov 21 13:58:05 <lumag> CONFIG_BSD_PROCESS_ACCT=y - probably not necceary
Nov 21 13:58:14 <lumag> CONFIG_MEMCG=y - what is that?
Nov 21 13:58:45 <lumag> Maybe switch from SLAB to SLOB.
Nov 21 13:58:53 <ant_work> I think they originate from linux-yocto
Nov 21 13:59:01 <lumag> CONFIG_PERF_EVENTS=y - do we need that?
Nov 21 13:59:05 <lumag> Yes, probably.
Nov 21 13:59:12 <ant_work> about SLOB, we changed because pxa_udc was broken
Nov 21 13:59:23 <lumag> Ah. ok.
Nov 21 13:59:25 <lumag> Interesting.
Nov 21 13:59:29 <ant_work> long ago...
Nov 21 13:59:36 <lumag> Because there should be no difference
Nov 21 13:59:42 <lumag> Was it pxa25x or pxa27x?
Nov 21 13:59:52 <ant_work> akita iirc, pxa27x
Nov 21 13:59:53 <lumag> CONFIG_PERF_EVENTS=y - probably of no use
Nov 21 13:59:59 <lumag> ok
Nov 21 14:00:06 <ant_work> this is the perf fw of yocto I guess
Nov 21 14:00:16 <lumag> Those kconfig entries can just come from default 'Y'
Nov 21 14:00:40 <ant_work> ok, production kernel is done with --alldefconfig
Nov 21 14:00:47 <lumag> CONFIG_OABI_COMPAT=y -- is it still useful? Well, maybe.
Nov 21 14:00:56 <ant_work> also coming from yocto
Nov 21 14:01:13 <ant_work> it ll starts to when I unified the defconfig under yocto 3.10
Nov 21 14:01:24 <ant_work> I then resplit using the produced artifacts
Nov 21 14:01:38 <lumag> CONFIG_IP_ADVANCED_ROUTER=y - definitely not necessary to small PDAs
Nov 21 14:02:04 <lumag> NAND and UBI can be switched off on collie.
Nov 21 14:02:21 <ant_work> no, UBI is ok (for tests ;)
Nov 21 14:02:37 <ant_work> core-image-minimal fits
Nov 21 14:02:42 <lumag> So does CONFIG_KEYBOARD_ATKBD
Nov 21 14:02:45 <lumag> ok for ubi.
Nov 21 14:02:53 <lumag> But nand is not existing :)
Nov 21 14:03:13 <ant_work> as long as we use 'defconfig' method we cn just purge those entries
Nov 21 14:03:34 <lumag> MAX1111 - not usefull.
Nov 21 14:03:38 <lumag> SENSORS_MAX1111
Nov 21 14:03:42 <ant_work> i.e. the mini-defonfig will be reexpanded and default bvalues applied where missing
Nov 21 14:04:09 <ant_work> so if you want to disable a default you have to do it explicitely
Nov 21 14:04:26 <lumag> ok.
Nov 21 14:04:31 * lumag just sketched over.
Nov 21 14:05:04 <ant_work> linux-yocto is just too big for this kind of handhelds
Nov 21 14:05:20 <lumag> Interesting. In my kernel I had battery_collie switched off.
Nov 21 14:05:29 <lumag> Maybe that is one of the reasons for having B/U = 1
Nov 21 14:05:37 <ant_work> hm..
Nov 21 14:06:56 <ant_work> http://www.manualslib.com/manual/151829/Sharp-Sl-5500.html?page=72
Nov 21 14:06:59 <ant_work> ^ about it
Nov 21 14:16:00 <ant_work> he
Nov 21 14:16:01 <ant_work> Number of charging and discharging of the operating
Nov 21 14:16:01 <ant_work> battery life
Nov 21 14:16:01 <ant_work> Approx. 500 times
Nov 21 14:16:01 <ant_work> Note: The Battery life and storage time may vary depending on the usage.
Nov 21 14:16:07 <lumag> [   12.964379] power_supply backup-battery: prop VOLTAGE_NOW=0
Nov 21 14:16:22 <lumag> Either it got disconnected somehow, or it's dead.
Nov 21 14:16:35 <lumag> Strange.
Nov 21 14:16:37 <lumag> [   63.644133] power_supply main-battery: prop VOLTAGE_NOW=0
Nov 21 14:16:37 <ant_work> it goes to 0 if the main batt is low
Nov 21 14:16:48 <ant_work> I remember similar readings
Nov 21 14:16:56 <ant_work> maybe there is a bug in collie_battery
Nov 21 14:17:14 <ant_work> (I have some 2.6.31 patches for battery around)
Nov 21 14:21:00 <lumag> pushed into the queue.
Nov 21 14:22:29 <lumag> fun. rmk is fast to say no, but doesn't say Y.
Nov 21 14:46:36 <ant_work> bbl
Nov 21 17:32:16 <ant__> lumag, kb-0.5.9 doesn't have that problem...
Nov 21 17:33:51 <ant__> well, not after halt..but it is back after echo mem > ...
Nov 21 17:33:58 <ant__> argh
Nov 21 17:36:58 <ant__> well, it doesn't resume but after reset it does boot from CF
Nov 21 17:37:05 <ant__> it's back to normal
Nov 21 17:41:24 <ant__> btw, mine doesn't squeak ;)
Nov 21 18:04:46 <ant__> root
Nov 21 18:04:59 <ant__> heh..this is not collie console ...
Nov 21 18:05:01 <ant__> sry
**** ENDING LOGGING AT Sat Nov 22 03:00:00 2014