IRC log of #cubox of Sun 11 Feb 2018. All times are in CET < Back to index

12:44 heap_> Ke: and how can i copy out on the new card that 1k of 3rd sectro?
12:48 Ke> if you have 4k card, I have no idea, if it's supported, all SD cards are using 512 sectors
12:50 heap_> means ? i cant do it?
12:51 heap_> so can i install new uboot on new card using my old /boot structure ?
13:42 Ke> what mass storage media are you trying to goot from
13:43 Ke> it is unlikely that it would be 4K media
13:43 Ke> UFS isa and some rare hard disks
13:43 Ke> s/isa/is/
14:00 heap_> i dontget it
14:00 heap_> i have sd card where bootloader + /boot folder was.
14:00 heap_> and rest of the files were on 2.5" hdd connected via usb.
14:35 Ke> heap_: if question was how to write to 1K it,s just like in the link given by jnettlet[m]
14:36 Ke> dd if=SPL of=/dev/mmcblk0 bs=1K seek=1
14:36 Ke> actually the offset dd if=u-boot.img of=/dev/mmcblk0 bs=1K seek=42
14:37 Ke> you know these instructions are intentionally obfuscated, because SoC vendors hate you
14:40 Ke> wonder how wandboard would have a different boot offset as cubox-i, if the have the same SoC
14:43 heap_> Ke: maybe too much complicated description for me
14:43 heap_> all i wanted is to use my old /boot + kernel image and move it to new card...
14:43 heap_> sd *
14:44 heap_> simple things are explained here very complicated... i think... sometimes.
14:44 Ke> sure
14:45 heap_> im not expert.
14:45 Ke> anyway you know how to manage, what is on the SD card
14:45 Ke> after learning that, you can try things out
14:46 Ke> most of the time you do not break anythign
14:46 Ke> it is the way you learn
14:46 heap_> i know i need to migrate /boot directory whcih i already copied out
14:46 heap_> all i need to do is write/migrate boot loader to new sd card.
14:47 Ke> do you have the bootloader image?
14:48 Ke> I think the nasty part that the bootloader configuration is not unified anyway between platforms
14:51 heap_> i dont knwo
14:51 heap_> do you read what i was requestiong?
14:51 heap_> im not changin platfor hw/sw
14:51 heap_> all is the same.
14:51 heap_> just old sd card got corrupted partially
14:51 heap_> so whats is the connection with different platforms?
14:52 heap_> i dont understand
14:55 Ke> no I was just complaining in general
14:56 Ke> I don't have a cubox so no personal experience so I can just try to help you interpret, what I can read
14:56 Ke> if there was a standard way of doing things, I could just tell you
14:57 Ke> is the problem, you don't have a u-boot image?
15:00 Ke> do you know how to copy parts of the SD-card to another SD card?
15:00 Ke> using dd?
15:08 heap_> Ke: i can mount broken sd card and copy out /boot folder without problems
15:09 heap_> and create new fs on new sd card and copy /boot there
15:09 heap_> the problem is that u-boot or boot loader on the old card is i think broken
15:09 heap_> so all i need to do is install new boot loader on new card ; copy /boot on new card and hopefully it should boot and read rest of the FS from 2.5' drive connected via usb
15:10 heap_> am i wrong or right?
15:12 Ke> I don't actually know, ovbiously it works if the old and new bootloader configurations are exactly the same
15:13 Ke> sometimes they may not be
15:37 Ke> on mcbin: Opening serial port /dev/ttyS1 failed: Input/output error
15:37 Ke> suihkulokki: does one of the serial ports on mcbin work for you?
15:38 Ke> suihkulokki: notably the IO error is even with nothing connected
15:38 Ke> openat(AT_FDCWD, "/dev/ttyS1", O_RDWR|O_NOCTTY|O_NONBLOCK) = 3
15:38 Ke> ioctl(3, TCGETS, 0xffffd69bd0f0) = -1 EIO (Input/output error)
15:51 vpeter> Ke: Isn't 1kB offset standard for imx6?
15:52 vpeter> For SPL at least. Location for u-boot.img is programmed in u-boot.
15:53 Ke> vpeter: perhaps, I mean between platforms
15:54 Ke> like non i.MX6
15:54 Ke> and I don't mean just the offset, but all the u-boot things
15:56 Ke> like where do you start loading the other stuff, or whether you have separate SPL
15:56 Ke> sometimes there is a reason for the differences, but very often it's just a vendor preference for no reason at all
15:57 Ke> I think my watchdog works outside of the serial thing
15:57 vpeter> Loader in SoC knows where to look for next part. So here is 1kB. SPL then decide where to look for img. And it loads and execute. And then load kernel, ...
15:57 Ke> quite
16:28 Ke> at least the SoC and the phy heatsinks feel quite cool, when touching them
16:28 Ke> but right now the ethernet is not connected
16:36 vpeter> One problem changed for another :)
16:37 Ke> well it's physically not connected, so not actually unexpected
16:37 Ke> I physically moved the unit to make the watchdog connections
16:37 Ke> now the watchdog is outside the case with reset button, power button and power led connected
16:39 vpeter> Make a picture of whole setup.
16:42 Ke> right now there is so much wiring that it's not really doable
16:42 Ke> I'll get shorter sata cables soon
16:48 Ke> now I'll get to see, whether the watchdog start race conditions with the appereance of ttyUSB0
16:51 Ke> I'll also publish the sources for the pain of everyone else
17:15 Ke> hmm. actually I am drawing the power of the arduino from USB that's kind of stupid
17:22 Ke> now I am running on the spectre-proofish fw
17:28 suihkulokki> Ke: never connected anything to the second serial so no idea
17:54 Ke> vpeter: this is about the way it's connected now https://users.aalto.fi/~jkarlson/files/arduino.svg
19:00 vpeter> jnettlet[m]: building openwrt sr and I'm getting
19:00 vpeter> DTC arch/arm/boot/dts/armada-385-naeba.dtb
19:00 vpeter> ERROR (phandle_references): Reference to non-existent node or label "tdm_pins"
19:23 vpeter> Maybe related to my changes. Rebuilding...
20:13 heap_> vpeter: ?
21:38 GArik> the top usb port doesn't work reliably on cubox-i in arch linux. Is this a known issue? I found a topic with a similar issue in arch forums, but there is no answers
21:51 GArik> i think it used to work better with 4.14.x kernel