IRC log of #cubox of Sun 13 Jul 2014. All times are in CEST < Back to index

09:45 esin Hi
09:46 esin Is it safer to run `shutdown -h now` and then turn cubox off or should I just terminate all running processes and then turn off cubox?
09:46 esin I need to turn off cubox, so that it recognizes my esata hdds
09:47 cbxbiker61_ shutdown forces a sync to disc, that's important
09:59 esin Good to know, thanks. So I'll definetly run shutdown before
12:03 _rd Hi, the standard Debian u-boot works with cubox-i4pro, but it does not even start SPL on a cubox-i2ultra.
12:04 _rd The same sdcards start on a cubox-i4pro without any issue.
12:04 _rd Does anybody know, how the cubox i2ultra and cubox i4pro differ in the u-boot IPL/SPL?
12:15 rabeeh _rd: which image?
12:15 rabeeh from the forums? if so then that image didn't even had SPL u-boot
12:16 rabeeh i can send you a link of an image i'v been working on; the same debian image but supports all CuBox-i and HummingBoard version
12:16 rabeeh s/version/versions
12:16 rabeeh it's not public since i was hoping we (we as a community) could get the kernel and drivers as debian packages
12:17 rabeeh http://dl.dropboxusercontent.com/u/72661517/tbr/debian-jessi.img.xz
12:18 _rd Hi rabeeh
12:19 auke- rabeeh: why are the yocto layers for the CuBox-I based on a 3.0.35 kernel and not 3.10 or even 3.15?
12:19 _rd I worked purely from Debian sources, this is what I did: http://bokomoko.de/~rd/Debian/cubox-i-notes.txt
12:20 _rd The sdcard boots flawless on an i4pro, but not on an i2ultra.
12:20 _rd I am wondering how they differ? Is the u-boot IPL on the i2ultra different?
12:22 _rd i.e. I suspect that there is a limitation or bug in Debian's u-boot SPL (not the one in your image) and I would like to get that fixed in Debian upstream.
12:23 _rd I assume you also follow https://lists.debian.org/debian-arm/
12:23 deniska --include=...emacs --exclude=nano
12:23 deniska I see what you did there =)
12:23 rabeeh auke-: kernel 3.10 in general has gpu and vpu memory leaks
12:23 rabeeh that's in general; jnettlet has been banging his head on all sort of surfaces to figure out how to fix
12:24 rabeeh kernel mainline is still not complete when it comes to support all i.mx6 features
12:25 auke- rabeeh: cubox specific leaks, or does this apply to all imx6 devices?
12:25 rabeeh auke-: applies to all imx6
12:26 rabeeh so; that's the main reason we don't have a more modern kernel support on CuBox-i
12:26 _rd When I look at the hexdump of the SPL from debian and the SPL from e.g. geexbox-devel-20140707-r9600c65.cuboxi.img.xz (which starts SPL on the i2ultra), I see a few differences at the beginning: http://paste.kde.org/pikryxhuv
12:26 auke- ah.. okay. thanks
12:26 _rd Does that ring any bell?
12:27 rabeeh auke-: that's really unfortunate but if you start digging into the details of fsl 3.10 then all sort of memory leaks starts to happen
12:27 rabeeh for instance when recording from multiple cameras
12:27 rabeeh _rd: is that u-boot spl with or without cubox-i support?
12:27 rabeeh cubox-i SPL support works from day one on all cubox-i models
12:28 rabeeh _rd: nop; doesn't ring a bell
12:28 rabeeh where are the sources of debian u-boot?
12:28 _rd With cubox-i support: http://paste.kde.org/pnl8np8pp
12:28 rabeeh do you have a url?
12:29 _rd I dig out the sources
12:31 _rd These are the sources for the package I use: http://sources.debian.net/src/u-boot/2014.04+dfsg1-2
12:32 _rd I just found there is a newer package in Debian experimental: http://sources.debian.net/src/u-boot/2014.07~rc4+dfsg1-1
12:32 rabeeh _rd: look at the patch; it looks ok
12:32 rabeeh this patch - ./debian/patches/cubox-support.diff
12:33 rabeeh although it's really old and doesn't have hummingboard support; but it's ok
12:33 _rd in the 2014.04 version or in the 2014.07 version?
12:33 rabeeh 2014.04
12:33 _rd Hmm....ok.
12:34 _rd Do you have any idea why that sdcard boots only on an i4pro and not on an i2ultra into SPL?
12:34 rabeeh oh
12:34 rabeeh 4GB micro SD?
12:34 rabeeh which brand? Transcend?
12:34 rabeeh we had a bug like that; hold on
12:35 _rd It is a 32 GB card Transcend
12:35 rabeeh https://github.com/SolidRun/u-boot-imx6/commit/dfb384fd31a103078b1e35dc83e53267367d4cf2
12:35 rabeeh that one created overflow on pll calculation
12:36 rabeeh _rd: i think the best way to achieve things here is -
12:36 rabeeh 1. clone - https://github.com/SolidRun/u-boot-imx6
12:36 rabeeh 2. compare board/solidrun of our tree with the patched debian tree
12:36 rabeeh 3. create an additional patch
12:36 rabeeh notice that the master of https://github.com/SolidRun/u-boot-imx6 already supports ALL of CuBox-i and HummingBoard boards
12:37 rabeeh jnettlet mentioned to me that u-boot finally accepted i.mx6 SPL patchset; due to that the CuBox-i and HummingBoard patches can be upstreamed
12:41 _rd Thanks rabeeh, I will check the latest 2014.07 version first and compare the patches after that.
12:42 rabeeh _rd: thanks. i think your main issue is that divisor thing
12:42 rabeeh we have seen it mainly on the solo / dual lite chipset with 4GB Transcend micro SD cards
12:43 rabeeh MikeSeth actually pointed on that for a long time until we finally reproduced and fixed.
12:43 davorin rabeeh? did you receive the pcie wlan adapters?
12:43 rabeeh nop
12:43 davorin weird....
12:44 davorin otoh it also too 1 week until my cbi-4p left your country from shipping
12:44 davorin s/too/took/
12:45 MikeSet 12:45 * MikeSeth appears briefly, pets rabeeh
12:47 rabeeh _rd: maybe we can get assistance here then
12:48 rabeeh we were long looking for getting the kernel, gpu, vpu and all drivers debianized
12:48 rabeeh and the idea was really to get fully fledged working desktop with accelerated gpu, gst-1.0 etc...
12:49 _rd rabeeh: does the i2ultra have a solo/dual lite chipset?
12:50 _rd I thought it is rather close to an i4pro.
12:50 rabeeh i2ultra has the dual chipset
12:50 rabeeh (i.e. the more higher end)
12:50 _rd Then the divisor bug should not apply, right.
12:50 _rd ?
12:50 rabeeh hmm... i'm not sure
12:50 _rd I also checked the 2014.07 image....
12:51 _rd ....does not work either, but has the patch applied: see
12:51 rabeeh we haven't seen that on the dual; but in certain configurations it might surface up
12:51 _rd http://sources.debian.net/src/u-boot/2014.07~rc4+dfsg1-1/arch/arm/cpu/armv7/mx6/clock.c
12:52 rabeeh yeah; looks better
12:52 _rd yes, I would love the fully fledged desktop running fully out of debian sources.....
12:53 _rd ....but it is ok for me, if the situation impoves over time. Right now the cubox i4p serves well as server when running from pure Debian.
12:54 _rd Just the i2ultra seems to cause some trouble.
13:00 rabeeh _rd: and the ones from geexbox works without issues - with?
13:00 rabeeh _rd: and the ones from geexbox works without issues - right?
13:01 rabeeh or whatever u-boot/SPL you have
13:04 _rd yes, the geexboox works without any issues.
13:04 _rd I try the debian image you supplied on dropbox just to cross check that I have no hardware issue, but I do not think so.
13:09 rabeeh _rd: that image is based on LK 3.0.35; but has wifi working without issues
13:10 rabeeh it's an accelerated xorg with xfce; based on image from jas-hacks
13:12 _rd Your comments on the kernel versions before for auke- have been very useful. Do you have that somewhere documented for others?
13:12 rabeeh _rd: you mean documented failures? or how to reproduce?
13:13 _rd No, your comments on 3.0, and 3.10, in particular that 3.10 has issues which 3.0 does not have.
13:13 rabeeh 3.0 the memory management in 3.0 was simpler
13:13 _rd If I would have replaced the pure debian kernel by something different, I would probably have gone for 3.10.
13:13 rabeeh (and also not that succesfull)
13:14 rabeeh simply the kernel allocated fixed memory and uses that
13:14 _rd ah, I see.
13:14 rabeeh _rd: yeah i know
13:14 rabeeh jnettet should have news with regards more modern kernel in few days
13:14 rabeeh he has been working around the clock on that; but i think his solution will be very satisfactory
13:15 _rd I run right now the 3.15 debian kernel....it works in "server mode", i.e. not gpu etc.
13:15 rabeeh the work is presented by jnettlet buy lots of masters has been behind that work
13:15 rabeeh yeah; headless mode is easy :)
13:15 _rd I am looking forward to see that.
13:15 _rd Where will it be announced?
13:16 rabeeh hopefully this week
13:16 rabeeh he should have two branches; production and -next
13:16 rabeeh if the community like this approach and willing to support it then we will fork and place the production part on SolidRun github
13:16 _rd Will it be announced on cubox-i.com ?
13:16 rabeeh that kernel should provide support for CuBox-i, HummingBoard, wifi, faster ethernet driver etc...
13:17 rabeeh yeah; we will have a blog post about it
13:17 _rd cool
13:17 rabeeh that's a major work that jnettlet and other masters have been working on in the background;
13:17 _rd are you also working on getting your patches upstream or is that up to linaro to pick the ones they like?
13:17 _rd yes, I fully understand :-)
13:17 rabeeh showing off and giving them credits is essential
13:18 rabeeh rmk has been taking most of the upstream work
13:18 rabeeh together with xobs (Novena laptop project), dv (i.e. Carlos), jnettlet, pengutronix, shawn guo and others
13:18 rabeeh they all deserve huge credit for their massive work
13:19 _rd ok, that will then ensure that the patches eventually reach the debian kernel via upstream ... .(?)
15:39 _rd rabeeh, the SPL in your debian jessi image works.
15:42 _rd Is there a good way to enable the serial console?
15:42 rabeeh _rd: yeah; works
15:42 quitte Hi. I'm about to order but I want to be sure that I can have JTAG access. Is that documented somewhere? will a j-link v8 clone do the trick with openocd?
15:42 rabeeh quitte: please look at the microsom documentation how to access jtag
15:42 rabeeh those are test points on the MicroSOM
15:43 rabeeh http://download.solid-run.com/pub/solidrun/SR-uSOM-mx6/rev-1.2_1.3/
15:43 quitte thanks
15:45 quitte is the quad ready to ship? it is in the shop, but other places made it seem like it was something of the future
15:46 rabeeh for what? HummingBoard
15:46 rabeeh if you want HummingBoard with the quad then you can order that via the mix and match section
15:46 rabeeh (i.e. choose your MicroSOM and then choose your carrier)
15:49 quitte yes, that's what I meant.
16:06 _rd rabeeh, serial console does not work for me.
16:06 rabeeh are you sure you have the i2ultra?
16:06 _rd console [tty0] enabled, bootconsole disabled
16:06 _rd are the last words
16:06 _rd yes, i2ultra
16:06 rabeeh ok
16:06 rabeeh the boot is on the hdmi port
16:06 _rd I see the u-boot output and shell
16:06 rabeeh ssh is there
16:07 rabeeh debian/debian username/password
16:07 _rd yes, I can login via ssh
16:07 rabeeh you can enable serial port via adding console=ttymxc0,115200 to the uEnv.txt bootargs
16:07 rabeeh that will give you the serial port for booting
16:07 _rd echo 'T0:23:respawn:/sbin/getty -L ttymxc0 115200 vt100' >> /etc/inittab
16:07 rabeeh for login i think you need to hack /etc/init.d
16:07 rabeeh and? works?
16:08 _rd No, that is not sufficient. but I did not change uEnv.txt
16:08 _rd will try that.
16:08 rabeeh i tried that before and it didn't work
16:08 rabeeh _rd: those are regarldess
16:08 rabeeh console=ttymxc0,115200 is only for boot messages
16:08 _rd I have the console running on my pure debian SDcard
16:09 rabeeh _rd: i know; this is exactly the difference between seriously made image and a crappy (sorry) one
16:09 rabeeh that's why i'm reluctent to release that image
16:09 rabeeh _rd: the previous link you have sent; is that the one and only script to debootstrap it?
16:10 rabee 16:10 * rabeeh disappears for 15m
16:10 _rd you mean on bokomoko.de ?
16:11 _rd It is not a script, but my set of notes on how I built the (pure) debian image
16:17 _rd The i2ultra with your image and the 3.0.x kernel crashed .... Is there any useful debug information I could collect?
16:34 rabeeh _rd: what kind of crash?
16:34 rabeeh lock? kernel panic? kernel oops?
17:37 rabeeh _rd: around?
19:10 davorin okay..first cbi-4p repaired (o;
19:13 rabeeh davorin: fuse?
19:13 davorin rabeeh: you still haven't told me the exact brand/model number of the 0603 fuse (o;
19:14 davorin just too a regular 5a in 0603 case
19:14 davorin s/too/took/
19:14 davorin next time...place the fuse and diode on the bottom ;-)
19:14 davorin or make the 4-pin header pluggable...
19:15 davorin just have a dead one now....short circuit vcc to gnd...
19:15 rabeeh can't be made pluggable; it will move
19:15 rabeeh mechanically it won't be stable
19:15 davorin hmm...'cause when plugging in cables?
19:15 rabeeh yes
19:15 rabeeh rj45
19:15 davorin okay..then f1 at the bottom please ;-)
19:16 rabeeh ok
19:17 davorin dunno how often the diode breaks though...
19:17 rabeeh the tvs?
19:17 davorin yep
19:18 rabeeh if someone puts 12v PSU then he will fry that tvs
19:18 rabeeh might take few minutes; hopefully by that time he will notice that something is wrong :)
19:19 davorin hmm...lemme check this on the other cb-i4p
19:21 rabeeh davorin: ?
19:21 davorin yes?
19:21 rabeeh you have a cbi-1
19:21 rabeeh :)
19:22 davorin well that too...
19:23 rabeeh so try nasty things on CBI-i1
19:23 rabeeh :)
19:24 davorin hmm...short circuit on the usom-i4
19:27 rabeeh hmmm... you'v killed it then
19:27 rabeeh which psu are you using?
19:27 rabeeh and fuse?
19:28 davorin came back from customer...
19:33 davorin okay...2nd cbi4p fixed as well (o;
19:34 davorin is there a similar powerful cpu like i.mx6 but without vpu/gpu?
19:35 davorin anyway...feierabend here (o;
20:07 goetz Hi
20:08 goetz I'm having trouble getting my Cubox (the original one, no -i) to boot with Debian.
20:09 goetz tried CuBox installer and netboot, both to no avail
20:09 goetz any hints beside http://www.solid-run.com/archive/mw/index.php/Debian_netboot_installer
20:09 goetz or http://www.solid-run.com/archive/mw/index.php/Debian_netboot_installer
20:09 goetz http://www.solid-run.com/archive/mw/CuBox_Installer ?
20:16 goetz or a prepared Debian or *BSD image for SD?
20:42 goetz Cubox installer hangs: http://pastebin.com/pUSU25eF
20:45 rabeeh goetz: i think your u-boot is way too old
20:45 rabeeh you need to flash it first with newer u-boot
20:52 goetz rabeeh: which file is it here? http://download.solid-run.com/pub/solidrun/cubox/u-boot/
20:52 bewees Is uboot updated automatically by archlinux_arm?
20:52 goetz u-boot-cubox_hynix_cubox_1GB_uart.bin or u-boot-cubox_hynix_cubox_unified_spi.bin
20:53 bewees On my x86/64 desktop system, gummiboot (my efi bootloader) is updated automatically by the package manager
20:56 goetz bewees: that bootloader won't be written to flash, but to hard disk, that seems to be different
20:57 bewees Hm, so uboot is something like a bios?
20:57 rabeeh goestz: there are multiple ways to flash u-boot
20:57 bewees And it is flashed to an eeprom chip on the cubox-i
20:57 rabeeh easiest is get the _spi.bin version on a tftp server; setup ipaddr and serverip and then run 'bubt'
20:58 rabeeh in case you kill the u-boot on the SPI flash then you can recover with the _uart.bin version
21:00 goetz ah, thanks, now I know the difference of the files
21:00 goetz documentation / the wiki is a bit ... for the already advanced.
21:07 goetz rabeeh: thanks a lot. I thought that 5.4.4 SR1 was _older_ than my version 5.4.4 and therefore did not try to update.
21:07 goetz rabeeh: now the installer works
21:07 rabeeh goetz: i know; in CuBox-i we changed the way things; that's why we moved to a fresh new forums
21:07 rabeeh one of the main design goals of both CuBox-i and HummingBoard is that all software is flashed on external SD card; even u-boot is completely booted from the micro SD card
21:07 rabeeh goetz: great !
21:08 rabeeh how old is your machine?
21:08 goetz rabeeh: yes, without googling I would not have found the old pages for original Cubox at all -- via navigation on solid-run I couldn't find it. A link "archived products" or "retired products" would be nice.
21:08 goetz rabeeh: I got one of the first engineering samples
21:08 goetz just a sec
21:08 rabeeh oh; good !
21:09 rabeeh 2 years old?
21:09 goetz rabeeh: February 2012 I got it
21:09 goetz no, sorry, mid of March 2012
21:09 rabeeh wow; almost 2.5 years
21:09 rabeeh ok; 2 years and 4 months :)
21:09 quitte bwees: this is nitpicking - but no a bios provides functions to the os (basically only dos uses those anymore) to access peripherals such as the disks. u-boot does not do that. it only leaves some data behind containing information about the hardware - optionally loads darta to ram and executes code. at least that is my understanding.
21:10 rabeeh goetz: this gives me huge deal of satisfaction about the quality
21:10 goetz rabeeh: I used it with Ubuntu 10.04 til now, but as the functionality of the box should now change (it will be connected to the internet directly now) I need to go for a supported distro. Therefore get rid of old Ubuntu
21:10 goetz rabeeh: yes, I can't complain :-)
21:11 quitte rabeeh: are any of the i.mx6 fuses destroyed prior to shipping?
21:13 rabeeh quitte: MAC address, booting from SD2 fuses are blown priort to production
21:13 rabeeh if you order the microsom as-is without carriers it's intact
21:13 rabeeh that's probably because the customer wants to design his own carrier board and we leave them intact
21:13 goetz rabeeh: thanks a lot again for incredibly quick help, and that on a Sunday!
21:13 quitte rabeeh: great. I do not want the sd fuses blown. I don't care about the mac
21:14 quitte rabeeh i do howerver want a carrier board,too
21:14 rabeeh goestz: Sunday is minor; Germany vs. Argentina is different story :)
21:14 rabeeh quitte: interesting
21:14 rabeeh will PM you
21:15 rabeeh have to leave
21:15 rabeeh ttyl
21:15 goetz ciao
21:15 goetz rabeeh: I'm German :-)
21:15 bewees Both teams play well
21:15 goetz (and not watching it ...)
21:15 bewees Iam watching too ^
21:15 goetz ... must ... setup ... my ... fresh ... cubox ... install
21:29 quitte_ I'm not watchiing... what is all the commotion about ?
21:30 goetz rabeeh: btw: great ide to include a mini USB serial connector onboard, helped me a lot
21:30 goetz s/ide/idea
21:31 quitte 21:31 * quitte_ didn't know that - but agrees
22:43 goetz ciao