IRC log of #cubox of Sat 19 Apr 2014. All times are in CEST < Back to index

02:09 _dab_ geexbox r17066, does anyone know why it defaults to a bluetooth wiiremote?
02:48 Tenkawa greetings all
02:48 Tenkawa whats new in the cubox world?
04:07 nikitis Anyone have issues with their cubox not powering up properly?
04:08 cbxbiker61 there have been some issues with non-solid-run power supplies not working well
04:09 nikitis i ordered one from solid run
04:09 nikitis long slinder one
04:09 nikitis I figured out that I have to power it up, pull it and quickly put it back in to boot
04:10 cbxbiker61 yeah that sounds familiar
04:10 cbxbiker61 the one i've tested seems to work fine
04:22 _dab_ nikitis: I have same problem, independent of the PSU type. It is uboot (I think)
04:45 nikitis i4 pro?
05:06 _dab_ i4p and i2u
05:08 mk01 _dab_: wifi on i4p is 20mhz band only, right (so no 40)?
05:16 _dab_ mk01: can't remember. using a wifi dongle at the moment
07:05 jnettlet nikitis, ping
07:15 jnettlet mk01, sorry this has taken me so long. I am finally getting to sorting out your problems
07:16 mk01 jnettlet no problem. I reverted the one commit & redistributed new kernel package
07:17 mk01 20 minutes.
07:17 jnettlet mk01, uboot commit? np
07:17 mk01 yes
07:18 mk01 what you told me . PFD locking errata
07:19 jnettlet mk01, I actually think we need to back two of my PFD patches out. The problem I was debugging was most likely due to failing hardware
07:19 mk01 jnettlet: I still have the feeling that it's not fully ok
07:19 jnettlet mk01, yeah there is another patch to remove as well.
07:19 jnettlet I am going to revet them and have rabeeh pull the changes into his branch
07:19 mk01 ah ok
07:20 jnettlet mk01, revert this as well 2d0c4f544da73
07:20 mk01 jnettlet: oh funny. how much you spent time on that - to find out it was HW ?
07:21 jnettlet mk01, not too long. probably half a day. It happens
07:23 jnettlet I am much happier having the real answer though, rather than speculation
07:23 mk01 on the last but one Gotham's merge window my fresh compiled version was refusing to start. so I spent 3h looking for problem. 1h after someone "fixed" broken fix
07:24 mk01 i'm going to revert the 2d0... and will report back
07:27 mk01 strange
07:28 mk01 jnettlet: I had this second commit even in the older u-boot version. I was exactly at "cleanup boot auto detection" and this one was one before
07:30 jnettlet mk01, that was my original patch and seemed less problematic. If it isn't needed I would rather revert back to just wholesale gating and ungating the pfd's
07:30 jnettlet ultimately I was using a bit for detection that is marked as for "debug" only
07:34 mk01 jnettlet: when we speak about gating ungating & ID bits. I spent some time on the CEC again. as I wanted to pin point the problem rather that as precaution take down whole CEC and then up again
07:36 jnettlet For anyone looking for a power-switch for the CBi's. http://www.cjemicros.co.uk/blog/?p=82
07:37 mk01 and found it (although there must e second one). problem was at irqhandler. there is shared irq for cec & hdmi and handler is checking own bit. but even in case of not own IRQ it returned IRQ HANDLED . so HDMI lost "tick" and never woke up
07:38 mk01 jnettlet: but to that I have question.
07:39 mk01 jnettlet: shouldn't the IRQ be regenerated again still because the HDMI bit wasn't cleared ?
07:42 jnettlet mk01, depends on the hardware really.
07:43 jnettlet this is the interrupt for the hdmi hotplug detection?
07:43 mk01 yes
07:43 jnettlet yeah that doesn't need to be cleared. It just sends out the interrupt when the even happens and that is it
07:44 mk01 jnettlet: ok then it makes sense
07:45 jnettlet yep. good find
07:45 jnettlet I am finally getting around to kernel patches today. I am not going to wait on the upstream rebase any longer.
07:45 jnettlet So I will patch on 3.10.30 and then do a rebase later on.
07:46 cbxbiker61 why not 3.10.37?
07:47 jnettlet cbxbiker61, that will happen when I do a rebase
07:48 jnettlet but I am still waiting on some patches from rmk to get finished up
07:48 jnettlet namely a rework of the ethernet driver for performance, and some pl310 cache cleanup
07:49 nikitis jnettlet: pong
07:50 cbxbiker61 i pulled a couple of old 3TB drives out of my main server, i should get a chance to test cubox-i as an nfs server on 3.14 next week or so
07:50 jnettlet nikitis, when you have to unplug and plug you CBi to get it to boot. Is this during u-boot init, or does it get to the kernel
07:50 jnettlet what are the symptoms of it not booting? Front power led on or off?
07:50 nikitis jnettlet: i used a usb cable to telnet into it as it boots, but it just sits there until i do the plug thing
07:51 nikitis front power led off
07:51 jnettlet nikitis, what u-boot version are you using?
07:51 nikitis no clue
07:51 nikitis the one that came with it
07:51 jnettlet we should get you on the latest version. We have fixed a handful of early boot issues
07:52 nikitis can you help me?
07:52 jnettlet sure.
07:55 nikitis so if I power it on and dmesg | grep ttyUSB it just sits there, if I do the unplug/plug method then it boots and I see things in the console
07:57 jnettlet nikitis, is your main machinelinux or windows?
07:59 nikitis linux
08:01 jnettlet 1 sec need to finish rewiring this plug
08:11 _dab_ jnettlet: on my i2u, it takes multiple power on/off to get it to boot.
08:12 jnettlet _dab_, are you on the latest u-boot?
08:12 jnettlet if so I need you to revert two commits
08:12 nikitis i'm at root@cuboxi:
08:13 _dab_ jnettlet: no. the latest never boots. I am using whatever was in the git on March 9
08:13 nikitis how do I tell uboot version?
08:13 jnettlet _dab_, then test the binaries that I am posting now.
08:13 jnettlet nikitis, actually just pull the sdhc card out and put it in your linux machine
08:14 _dab_ jnettlet: no prob. Where?
08:14 jnettlet nikitis, _dab_ https://dl.dropboxusercontent.com/u/736509/u-boot/uboot_fixed.tgz
08:15 jnettlet that has SPL and u-boot.img in it
08:15 _dab_ ok, back in a few minutes
08:16 jnettlet nikitis, instructions for putting those files on your SDHC card are here. http://imx.solid-run.com/wiki/index.php?title=Building_the_kernel_and_u-boot_for_the_CuBox-i_and_the_HummingBoard#U-Boot
08:17 mk01 jnettlet: tested img without the second commit. no change. eth0 can't get dhcp address. but I try again after cca uptime 20-25s, it gets address. during the run when dhcp can't be obtained, tcpdump shows regular reply pakets on cubox
08:17 mk01 on cubox's broadcast requests
08:20 jnettlet mk01, plugged into gigabit or 100bt?
08:20 mk01 plugged into 1g
08:20 nikitis jnettlet: this isn't making much sense to me
08:20 nikitis my card automounts like 5 partitions
08:20 nikitis one being /boot
08:20 jnettlet nikitis, that is fine. We are writing to the raw sdhc card.
08:21 nikitis i've got android on it now
08:21 jnettlet mk01, can you try booting with ethernet not plugged in and then plug it in and see what happens?
08:21 jnettlet nikitis, that is fine.
08:21 jnettlet nikitis, do you see which device your sdhc card is showing up as?
08:22 jnettlet and have you extract the archive I posted. You should have two files SPL and u-boot.img
08:22 nikitis yes, and sdf
08:22 _dab_ that took 3 power off/on s
08:23 _dab_ on the i2u
08:23 jnettlet nikitis, so you need to run this command. sudo dd if=SPL of=/dev/sdf bs=1K seek=1
08:24 jnettlet and then this one. sudo dd if=u-boot.img of=/dev/sdf bs=1K seek=42
08:24 _dab_ anyway I can debug this for you?
08:24 nikitis alrighty, one sec
08:25 jnettlet _dab_, well the problem with debugging this is that no front LED means that it failed to read the initial binary from SDHC
08:26 jnettlet _dab_, have you tested other SDHC cards?
08:27 nikitis done, plug in and try?
08:27 jnettlet nikitis, yes
08:27 _dab_ 5 different SD cards so far
08:28 jnettlet interesting, and all exactly the same behavior?
08:28 nikitis same, still have to plug/unplug/plug
08:28 _dab_ yes. I thought originally that it was a uSD problem until I twigged to the pon/poff
08:29 nikitis jnettlet: for me i've tried class 10 SD and class 4, same results
08:30 _dab_ I have tried Transcend(ultimate 4GB 8GB and noname 4GB) , Integral 16GB class 10, Sandisk ultra 8GB
08:30 jnettlet _dab_, quick test. Once you plug/unplug so u-boot loads. Can you escape into u-boot shell
08:31 jnettlet then run fuse read 0 5
08:31 jnettlet and fuse read 0 6
08:31 _dab_ ok, BTB warm boot always works
08:33 _dab_ Reading bank 0:
08:33 _dab_ Word 0x00000005: 00002840
08:33 _dab_ CuBox-i U-Boot > fuse read 0 6
08:33 _dab_ Reading bank 0:
08:33 _dab_ Word 0x00000006: 00000010
08:34 jnettlet okay that is fine
08:36 nikitis nothing is different for me
08:38 _dab_ jnettlet: I have tried with and without video plugged in. Makes no difference
08:38 jnettlet and you are both using the 3amp plug that can with the CBi right?
08:39 nikitis whatever I ordered from solid-run yeah
08:39 nikitis it runs fine once it's going
08:39 _dab_ On the i4pro I have a high quality 5v/3a, i2u has 5v/2.5A
08:40 _dab_ Have tried different PSUs
08:40 _dab_ voltage is 5.18 and very stable
08:40 jnettlet one last bit of information.
08:41 jnettlet when you unplug and plug it back in for it to boot. What does Reset cause read out as?
08:42 nikitis interesting
08:42 _dab_ Reset cause: POR
08:42 nikitis now
08:42 nikitis I just found something out
08:43 nikitis if I plug in the back of the cubox-i4pro with the adapter already plugged in, I have to do a pull out and put back in
08:43 nikitis Clarifying: If the adapter is plugged into the wall FIRST, then plug in the other end into the cubox, I have to do the plug/pull/plug trick.
08:44 nikitis but I just tried it with it already plugged into the cubox
08:44 nikitis and THEN plugged into the wall, and it worked
08:44 _dab_ let me try that
08:45 nikitis I think there's a power surge in my adapter because it's already plugged in. And it needs to be drained
08:45 nikitis if I plug it in drained, it seems to boot on first try
08:45 _dab_ most s modes take a while to lose power
08:46 nikitis This looks like adapter design fail
08:46 _dab_ makes no difference for me
08:46 nikitis I've always only plugged into the back of the cubox after it was already in the wall
08:46 _dab_ jnettlet: I have tried 4 different PSU
08:47 jnettlet _dab_, I understand.
08:47 nikitis I just this time did it in reverse because I wanted to check that it was 5v 3a, then plugged it back into the wall, and it booted first time
08:47 nikitis when i plugged it into the wall, it was already plugged into the cubox
08:48 nikitis so 1.) wall => cubox = plug/pull/plug, 2.) cubox => wall = boots first time.
08:49 jnettlet nikitis, you have tested this multiple times?
08:49 nikitis no only once
08:49 nikitis let me try again
08:49 _dab_ nikitis: that means that it boots with a low voltage. depending on the PSU they take varying times to reach a stable 5V
08:50 nikitis just did it again, cubox then wall = first boot
08:52 nikitis yeah but if I do Wall then cubox, it won't boot, but the adapter must already have capaciters charged or something, and deliver instant 5v
08:52 nikitis the cubox then drains it some, but it's incorrect and won't boot, then when i quickly pull and plug back in, it's correct
08:53 nikitis maybe time for adapter redesign
08:56 _dab_ jnettlet: need anything else?
08:57 jnettlet _dab_, you have reported this to rabeeh already?
08:58 jnettlet it doesn't seem to be a software problem at this point, I guess I could give you something that will send out debug output as early as possible
08:58 jnettlet although we have seen some other i2u related weirdness. Maybe we need to re-check the DDR3 setup for that SOM
08:59 jnettlet I will look at that quick.
08:59 _dab_ jnettlet: I have told rabeeh. My gut feel is uboot. It warm boots too well. I will try anything you give me
08:59 nikitis Is there anything I can do for my situation hardware wise to improve? Like a different adapter?
09:00 jnettlet _dab_, but POR and WDOG boots are two separate processes for the MX6
09:00 jnettlet different init steps are taken, and different clock resets are done
09:07 _dab_ jnettlet: On my i4pro, I have no problems with a Mar9 uboot, yet I have problems with current uboot. Therefore methinks software.
09:09 jnettlet _dab_, that is a separate problem. We have figured out what is wrong with the i4p booting.
09:09 _dab_ ok
09:12 jnettlet _dab_, on sec. I have one more u-boot change for you to tes.t
09:13 jnettlet _dab_, same link, has been updated
09:16 _dab_ ok
09:17 jnettlet _dab_, actually, please verify those binaries boot your CBi4p successfully as well if you don't mind.
09:20 _dab_ will do
09:27 _dab_ jnettlet: no luck. but it is better. Only takes one retry to boot (repeated this a few times)
09:27 _dab_ now checking i4
09:32 jnettlet _dab_, hold on. just found one more discrepancy. Let me get you another binary for the i2u
09:33 _dab_ ok
09:33 jnettlet uploaded
09:38 _dab_ ok
09:47 _dab_ no change on the i2u
09:47 _dab_ works on second PO
09:49 _dab_ do you want the last one tested on the i4?
09:50 jnettlet _dab_, nope it won't make a difference
09:50 jnettlet I will talk with rabeeh about it
09:50 _dab_ ok. dinner time
21:31 heap_ hi, not sure what i did in my archarm but i have in systemctl [email protected] running and [email protected] is in failed state