<Astralix>
I checked out stable-3.0 branch and I can find
<Astralix>
radxa_rock_pro_linux_defconfig
<Astralix>
and
<Astralix>
radxa_rock_linux_defconfig
<Astralix>
I took the last one as I have a preview-board, not the pro
<naobsd>
I think I didn't recommend 3.10 kernel...
<naobsd>
and I asked about error, and I recommend RK official toolchain
<Astralix>
I have never compiled any 3.0.xx rockchips kernel with less than a 4.8.2 hardfloat gcc and all of them worked. What has gone wrong?
<Astralix>
I thought you recommendet me the origin/wip/lsk-android-14.04-radxa-rock
<Astralix>
And this one compiled till yesterday. Then it started to fail with something like this
<Astralix>
/radxa_linux/kernel_rockchip/drivers/gpu/arm/t6xx/kbase/src/linux/mali_kbase_mem_linux.c:1334:49: error: ‘VM_RESERVED’ undeclared (first use in this function)
<Astralix>
ok, let' check, switching to 4.6 androideabi
<Astralix>
checkout stable-3.0
<naobsd>
well
<Astralix>
make O=../out radxa_rock_defconfig
<Astralix>
cd ../out
<Astralix>
make -j12
<Astralix>
already failed
<naobsd>
you didn't provide about your problem more than "it fails", I cannot answer any "sure" thing :(
<Astralix>
just a second
<Astralix>
radxa_linux/kernel_rockchip/arch/arm/mach-rk3188/../mach-rk30/ddr.c:3682:1: fatal error: opening dependency file arch/arm/mach-rk3188/../mach-rk30/.ddr.o.d: No such file or directory
<naobsd>
about gcc, what I said is arm-eabi-4.6, not androideabi
<naobsd>
please remove "O=../out"
bengal has joined #radxa
<Astralix>
I provided patches to repair the out of tree build for 3 years now to any rk kernel I see
<naobsd>
you said radxa's tree must have fix?
<Astralix>
There are several bugs in the makefiles RK added... and they keep these bus since the first RK2918 kernel I ever saw
<naobsd>
well
<Astralix>
I check again and provide you the fixes
<Astralix>
But this kernel is based on gallads kernel, if I see the comments
<naobsd>
what do you need? how to compile radxa's tree? want to talk how RK tree bad?
<Astralix>
Probably I mixed some of the many information I got in the last days while trying to boot linux on my radxa, without taking one of the prepared images
<Astralix>
I didn't want to offend you. sorry!
<Astralix>
However, take a look at galland, omegamoon or even my kernels. Most of them can be compiled out of tree and they work pretty fine and fast with gcc 4.8.2hf and newer.
<naobsd>
well
<Astralix>
But I didn't want ot be creepy, I just wanted to build ubuntu desktop based on latest greatest things available
<Astralix>
The preview radxa's are not well supported, what I understand, there are only 100 of them.
<naobsd>
which do you want, compile radxa's kernel, or say how good your kernels?
<naobsd>
and you know how to fix, no need to ask me :(
<Astralix>
I want to share information to get the best of what is there. If you have something good for me, I'd be happy to take it and if I find something good for you, I send it to you
<naobsd>
you objected all my answers
<naobsd>
thank you
<Astralix>
No, compiler finished using 4.6.2 with no-out-of tree
<Astralix>
so as soon, as I find the problem why it doesn't build out-of-tree I send you a patch
<naobsd>
Makefile need to be fixed
<Astralix>
Cause compiling in kernel source tree makes work with git so ugly cause of the million of un-watched files.
<Astralix>
yes
<Astralix>
I didn't object anything you told me.
<Astralix>
I was just wondering why some of the problems are still there
<naobsd>
linux-rockchip kernel repository have no relation to any other kernel repository
<naobsd>
and it has full history
<Astralix>
Yes, as I said, could be my fault. I checkt many pages of people, who describe how to install linux on an RK3188 board
<Astralix>
So I may have mixed some info. However it is open source, so taking patches and updates from other people is not forbidden
<Astralix>
[ 0.000000] Linux version 3.0.36+ (astralix@BigBox) (gcc version 4.6 20120106 (prerelease) (GCC) ) #1 SMP PREEMPT Sun Oct 5 19:08:14 CEST 2014
<naobsd>
some "other people"'s repository doesn't keep history. any changes in that repository cannot be reused because author is not clear.
<naobsd>
if you send patches to me/linux-rockchip, please make sure that you're the author(you have copyright), otherwise it cannot be merged
<Astralix>
I am doing open source long enough to keep that in mind
<Astralix>
The problems of the past where ugly.
<naobsd>
sadly it seems some of "other people" doesn't have experience about open source like you
<Astralix>
there where only kernels lekaed by OEMs, and they often where only non-working snap-shots.
<naobsd>
I know leaked source didn't have history
<Astralix>
This was the reason why often kernels popped up in github, that had no history
<naobsd>
what I said is the things after first big commit
<Astralix>
Yes, was a hard time. We didn't know what to do and how to fix. We didn't had any manual or schematics.
<Astralix>
We didn't have any support nor any internal information
<Astralix>
But we even ported JB to RK2918...
<naobsd>
what I said is, some people committed others work as own work
<naobsd>
I don't blame about your great effort
<naobsd>
I'm not talking about your work.
<Astralix>
OK, I missunderstood you
<naobsd>
then
<naobsd>
you compiled stable-3.0
<Astralix>
Yes
<Astralix>
It boots and failes at DDR init.
<naobsd>
it should have issue, what I recommend is radxa-dev-kk
<naobsd>
oops
<naobsd>
DDR should work fine :(
<Astralix>
But it could be as wiki is for radxa Lit or Pro, I have Preview