Vasco has quit [Remote host closed the connection]
Vasco has joined #linux-exynos
_whitelogger has joined #linux-exynos
_whitelogger has joined #linux-exynos
LiquidAcid has joined #linux-exynos
miasma has joined #linux-exynos
<miasma>
hi, i wanted to try mainline u-boot with odroid-x (exynos 4412), but these instructions don't quite match with the current version http://linux-exynos.org/wiki/ODROID/u-boot
<Wizzup>
could you updade them?
<Wizzup>
update*
<miasma>
yes, but what needs to be done? i couldn't find the 'fuse' parts
<miasma>
i used 'odroid_defconfig', but it only produces the uboot.bins
<miasma>
and dtb
<miasma>
apparently i need to 'dd' some binary in the boot sector or so, to make the SoC find u-boot
<Wizzup>
those should be provided by hardkernel
<miasma>
ok i see
<miasma>
are those specific to some u-boot version? i guess they're not updating their stuff
<miasma>
last update 4 years ago
<Wizzup>
there is a good chance that whatever signed binaries they had before will still work
<miasma>
ok i'll try
<miasma>
sadly the u-boot is badly broken. got some multiple definition errors with pmic stuff
<miasma>
maybe i'll try with bl1 & bl2 binary blobs
<miasma>
or is it so that bl1 & bl2 require a signed compatible u-boot
<LiquidAcid>
miasma, the u-boot docs already cover all this
<LiquidAcid>
i don't think there is any need to replicate everything to the wiki
<miasma>
OK
<miasma>
LiquidAcid: but the instructions in the wiki are broken and the u-boot's doc/README.odroid covers only 'Odroid X2/U3/XU3/XU4', not X
<miasma>
at least that's what it says
<LiquidAcid>
miasma, X is the same as X2
<LiquidAcid>
iirc it even uses the same blobs
<LiquidAcid>
miasma, i think the only difference is that the X has a regular exynos4412, and the X2 has the prime version, which allows for higher clocks
<miasma>
LiquidAcid: the mainline u-boot gets stuck when displaying ram size, but arch linux provides 2016.11 which seems to work https://pastebin.ca/3872393
<miasma>
also i couldn't find a way to switch from 'x2' to 'x' in u-boot configuration menu
<miasma>
maybe it's guessing the ram speed wrong
<LiquidAcid>
hmm, i think i remember there being some issues
<miasma>
aside from the dram detection, the patch still has some non-offensive parts for just detecting the board. currently u-boot doesn't seem to differentiate between x vs x2 at all
<LiquidAcid>
i'm not aware of differences with respect to the mmc subsystem, but i might be wrong there
<miasma>
i could try with other sd cards too
<miasma>
just assumed a samsung card would co-operate with samsung SoC
<LiquidAcid>
miasma, could you check if this branch (with the odroid-x patch applied) works?
<LiquidAcid>
i really don't see anything in the first patch that would interfere with mmc operation
<LiquidAcid>
also the alarm patch should just do the same
<miasma>
yep. i'll test with different u-boot versions to see if it breaks at some point
_whitelogger has joined #linux-exynos
<miasma>
LiquidAcid: i managed to locate the problem a bit better. patched uboot 2017.1 works, but 2017.7 doesn't. i couldn't compile 2017.3 or 2017.5 since they produced some compile errors
<miasma>
so apparently something in u-boot 2017.3-2017.7 broke the sd/emmc support
libv_ has joined #linux-exynos
libv has quit [Ping timeout: 260 seconds]
libv has joined #linux-exynos
libv_ has quit [Ping timeout: 260 seconds]
<LiquidAcid>
miasma, interesting, strange that my stable branch is not working for you then
<miasma>
it might be too recent
<LiquidAcid>
miasma, no, it's working for me, it's actually the ver. i'm using
<miasma>
i guess my board is just dead. i tested with some prebuilt images from different sites and it seems to crash on boot
<LiquidAcid>
do you have an sdcard inserted at that point?
<miasma>
yes, it boots from the card
<miasma>
i'm pretty much willing to give up. don't want to waste your time. I suspect the board has gone bad during storage. I haven't used it in a year
<LiquidAcid>
can you repeat the boot process a few times and check if it always hangs when initializing the sdhci?
<miasma>
i could. it seems it hangs at exactly the same point with each kernel
<miasma>
well, a different point with different kernels, but it's reproducible
<LiquidAcid>
so never gets further than "s3c-sdhci 12530000.sdhci: clock source 2: mmc_busclk.2 (100000000 Hz)" ?
<miasma>
this is the best I got. it starts to boot userspace, but then shuts down https://pastebin.ca/3872529
<miasma>
this is with exynos_defconfig and kernel module support disabled so that it's easier to test with a single zImage
<miasma>
i also tried if it works better without the serial cable connected, but there's no difference
<LiquidAcid>
and you did the memtest?
<miasma>
not yet. still working on it
<LiquidAcid>
otherwise you might be chasing ghosts
<LiquidAcid>
with all that random behaviour my guess is on either memory or the sdcard
<miasma>
so i built the u-boot with memtest support. how should i start this with uenv.txt ? the u-boot prompt doesn't seem to support usb keyboards or communication via serial
<LiquidAcid>
miasma, huh?
<LiquidAcid>
that sounds wrong, you should be able to control it with the serial/uart
<miasma>
i get the prompt 'odroid #', but nothing happens if i send stuff via serial
<LiquidAcid>
check you settings
<LiquidAcid>
if you get the prompt, then you've aborted the autoboot, so something must go over the serial line
<miasma>
i disabled autoboot from the uboot's config and removed zImage so it wouldn't boot
<LiquidAcid>
ah, ok
<LiquidAcid>
nevertheless this has to work
<miasma>
thanks for your time. i'll see if I get any further tomorrow, but if not, I'll just give up. seems that it's just broken