IRC log of #cubox of Tue 15 Apr 2014. All times are in CEST < Back to index

03:51 _dab_ I now have an i2ultra.
03:54 _dab_ is geexbox daily build a viable event yet?
08:28 _dab_ anyone know the trick to getting a geexbox nightly to boot?
13:52 _dab_ tomlohave: Is there some special trick to getting a nightly to boot on and i2ultra?
15:19 mk01 jnettlet: do you have a minute ?
16:21 davorin quite quiet these days (o;
16:22 davorin even no email responds from solidrun for almost a week now...
16:26 mal- perhaps any pre holiday
16:27 jnettlet mk01, just got back. shoot
16:29 mk01 wanted to ask again about the kernel cmd line problems I had (with setenv in uboot)
16:29 mk01 but already recompiled uboot with adapted sizes for cmd line paraters #nr and buffer
16:29 mk01 will test later
16:30 mk01 this is one side of the story - when you told me maybe it is too large to fit into variable size
16:30 jnettlet mk01, is this for a 3.0.35 kernel?
16:30 mk01 3.10
16:31 mk01 then you told me I should try using cmd line defined in kernel
16:31 jnettlet you know you can add kernel commandline parameters to your .dts file and just recompile that?
16:31 mk01 problem is I need some parameters extras - user definable
16:32 mk01 so I tried defining in .config with setting it should be as addition to the one parsed from boot
16:32 jnettlet can you give me examples? Most of these should be exposed through sysfs to change in userspace
16:32 mk01 but in that case it is completely ignored
16:32 jnettlet you will need to change your kernel config to append kernel config options to the bootloader options
16:33 mk01 but when it is going about root= or rootfstype or rootflags needed to mount root ?
16:33 mk01 I did that CONFIG_CMDLINE_EXTEND=y
16:33 mk01 but nothing
16:34 mk01 the one from kernel is "ignored"
16:34 mk01 and then such "stupid" issue like putting "-" to variable
16:35 mk01 from uboot command line works single quote
16:35 mk01 '
16:35 mk01 from bootscript I get error
16:35 mk01 on the same command
16:35 mk01 it's driving me crazy
16:36 jnettlet mk01, oh right with device-tree that is broken. I thought I had merged the patch to fix that.
16:36 mk01 If you share the patch I will manage with this "combined" approach
16:38 mk01 otherwise I need to dig into u-boot code :-/
16:40 jnettlet why are the uenv.txt options not working?
16:41 mk01 I don't use uenv.txt as it doesn't support initramfs
16:42 mk01 i would not be hard to adapt, but still the "setenv" problems will be the same, not ?
16:42 mk01 anyhow I tried putting into env file
16:42 mk01 loading it
16:42 mk01 then "set import -t to load"
16:42 mk01 and run XXX to process
16:43 mk01 same error
16:44 mk01 if I try to set env variable from uboot varieble -> same error as in case of setenv on command line. and I don't know about other tricks how to process this
16:44 jnettlet are you using the latest u-boot?
16:46 jnettlet I overhauled SPL quite a bit, and then rabeeh and i fixed another bug that was causing problems because of an overflow
16:47 mk01 today as I was changing the config options I pulled : 3fd848b..ed888a1
16:47 mk01 but again, didn't test yet
16:49 jnettlet if you can post an Image I can test again. I have my hardware back up to full capacity
16:52 mk01 this MY last option (when you told me last time to try expand the sizes). and definitely then I will send you my /boot and 1M of SD if you want. or if you want to look, grab http://xbian.brantje.com/pool/devel/main/x/xbian-package-kernel-6q/xbian-package-kernel-6q_1.3.8-16_armhf.deb, extract and look at /boot/boot.scr.txt
16:52 mk01 this is still working ok
16:53 mk01 if you add even "ahci_imx.hotplug=1 " to "baseconfig"
16:53 mk01 it will not parse it anymore
17:00 mk01 jnettlet: to be exact, setenv baseconfig still works , then I need to add some others ... it is "setenv bootargs $baseconfig ......" and that fails. you will see it when you check the file.
17:01 mk01 but really don't waste the time until I try to boot the new recompiled uboot
17:01 mk01 maybe as you say it will work
17:02 mk01 btw: could be new .bin is 70kb smaller ?
17:03 mk01 220kb vs. almost 300kb before
17:06 jnettlet mk01, yep I switched over to thumb2 mode to get a smaller binary for SPL.
17:06 jnettlet I had more code I needed to fit in SRAM
17:09 mk01 ok
17:19 davorin jnettlet: got a successful boot on the hb3?
17:24 jnettlet davorin, yep everything seems okay
17:25 davorin with som-i4p?
17:25 jnettlet nope just single and dual-core
17:25 davorin ah okay..doesn't help me (o;
17:25 davorin m yhb2 also boots normal with single core
17:26 davorin have to read more into tds and see how i can modify the cbi4 to a som-i4p version..
17:38 jnettlet davorin, do none of your dual-core soms work?
17:48 dv_ does anything about the HB3 change with regards to kernel, dtb etc.?
17:48 dv_ in other words, if I build a yocto rootfs with meta-fsl-arm and "cubox-i" as machine, will it boot?
17:48 jnettlet dv_, yeah they support more stuff
17:49 jnettlet we need to add a imx6q-hummingboard.dts file
17:49 dv_ so the current 3.0.35 solidrun kernel will not work
17:49 jnettlet no that should work because it doesn't use dts
17:49 jnettlet it should pick up most the cubox stuff.
17:49 dv_ yeah but any board related differences?
17:49 jnettlet obviously things that won't work are LVDS, mPCIE
17:50 dv_ hmm okay, so that kernel will work, just not fully
17:50 jnettlet should
17:50 jnettlet I am just getting to digging in
17:50 jnettlet so far I have two boards up and running
17:50 dv_ I have patches for meta-fsl-arm ready to use your 3.10.30 kernel already (more specifically, the solidrun fork of your kernel)
17:51 dv_ I cannot submit them before the DVI thing is fixed though :/
17:52 dv_ well, perhaps I can talk to otavio to include the current release and let it work with pure hdmi monitors for now..
17:52 jnettlet yep, well now that I have more hardware to test against I can get back to things.
17:59 davorin only the single works...not the som-i2u and som-i4p
18:03 jnettlet mk01, you are hitting this limitation in u-boot. #define CONFIG_SYS_MAXARGS 16
18:03 jnettlet davorin, oh the i2u is the same class as the i4p. I just have the i2dl
18:04 davorin also no i4p?
18:04 jnettlet that is the difference. Just need to tweak the device-tree file
18:04 jnettlet mk01, we can bump that to 32 easy enough
18:05 davori 18:05 * davorin switching computer
18:05 mk01 ok, then the changed CONFIG will work I changed to 32 before the compilation
18:06 mk01 thanks for the hint! although it looks like I can't count to 20 :-/
18:06 jnettlet well you have to include the initrd and dtb in the arg length
18:07 mk01 hm, right. and initramfs mem address ?
18:07 jnettlet yep
18:07 mk01 thx.
18:21 davorin jnettlet: you have a som-i4, right?
18:57 jnettlet davorin, yes, but it just got fully mounted back into a CBi casing
18:57 jnettlet I need it to be protected for a while, since I was without for a bit.
18:57 davorin well...i can send you some cbi-i4p (o;
18:57 davorin have all models on stock..
18:58 jnettlet I will keep that in mind. But right now I am swimming in dev hardware I need to get caught up on
18:59 davorin i'm still waiting for a feedback frmo rabeeh or atai regarding new hardware...especially hb3
19:01 davorin because i need the som-i2u or som-i4p for own development for a hospital here in .ch
19:19 jnettlet davorin, have you tried to boot the original 3.0.35 kernel on a HB with the i4p som in it?
19:20 davorin nope...
19:20 davorin i just pulled the git versions of the kernel and submarine
19:21 jnettlet I will be sure to include a imx6q-hummingboard.dts in my next rebase
19:21 davorin hope to have more time over this long weekend...
19:21 davorin but there is the other problem of cooling with som-i4p and hb
19:22 jnettlet yes.
19:22 davorin with i1 and i2 is no problem..but the i4 has the metall cap exposed...
19:22 davorin and it needs to be low-profile
19:22 davorin so add-on boards won't get a short circuit
19:23 davorin have to call anyway tomorrow a supplier specialised in heatsinks and casing
19:23 jnettlet and you don't want to block the mPCI and lvds ports
19:23 davorin exactly..
19:23 davorin anyone ever tested if it would fit into the regular rpi case? (o;
19:24 jnettlet of course an aluminum case with a copper spacer down to soc would work
19:24 davorin if it doesn't fit, what's the purpose of the formfactor (o;
19:24 jnettlet the new one was adjusted to fit the RPi perfectly
19:24 davorin even with msata on bottom?
19:25 jnettlet not all cases but most yes.
19:25 davori 19:25 * davorin speaking of the multicomp case
19:27 davorin has there been a shortage in i.mx6 cpus lately?
19:27 jnettlet I will be testing RBPi cameras over the next couple of days
19:27 jnettlet initial tests rabeeh does says they need a color space conversion done in the drivers
19:28 davorin ah cool...have some spare camera on stock...regular and ir...
19:29 jnettlet I have both
19:29 davorin is there already a program to test it?
19:29 jnettlet I need to port the drivers over the RBPi kernel tree. I don't think they are upstream yet
19:29 jnettlet and then merge it into FSL's crazy mem2mem v4l driver setup
19:33 davorin it's csi-2?
19:34 davorin said csi wasn't specified with audio channel (o;
19:35 davorin s/said/sad/
19:36 jnettlet yes csi-2 as best as I can tell
19:36 jnettlet but you need the sensor driver as well
19:40 davori fpga -> csi-i (o;
20:43 mk01 jnettlet: you are right, RBPi cam drivers are not upstream. it was introduced only recently to RBPi
21:44 jnettlet mk01, well they will be in my kernel soon enough. hopefully with IPU integration to do the necessary colorspace conversions
22:00 mk01 jnettlet: RPI needed to solve this as well. they were waiting for broadcom fw update to allow processing in HW
22:06 mk01 jnettlet: one more question I'm curious about. according to dmesg timing, eth0 is started here
22:06 mk01 [ 11.070626] fec 2188000.ethernet eth0: Freescale FEC PHY driver [Atheros 8035 ethernet] (mii_bus:phy_addr=2188000.ethernet:00, irq=-1)
22:06 mk01 and is made UP by "ip" here
22:06 mk01 [ 15.065189] libphy: 2188000.ethernet:00 - Link is Up - 1000/Full
22:07 dv_ quad-videodecoder?
22:07 dv_ the vpu isnt a quadcore :)
22:07 mk01 but pings from another machine gets responded firstly at 50seconds
22:07 mk01 Request timeout for icmp_seq 48
22:07 mk01 64 bytes from 192.168.1.12: icmp_seq=49 ttl=64 time=1.216 ms
22:08 mk01 with 3.0.35 it was also a bit later (like +5+10 seconds after adapter responded UP), but not 40s
22:08 mk01 why is that ?
22:31 davorin you mean reply after it showed as up?
22:32 davorin and when you try to set arp -s manually before ping?
22:32 davorin also long delay?