<lunra>
Has anyone managed to chain u-boot from the proprietary bootloader?
<ganbold_>
naobsd: yes, nothing much about the loader, I was thinking to put freebsd kernel into nand similar way like linux and see how that works, or if the loader can load u-boot then that opens other possibilities
<naobsd>
I never heard about u-boot for chain-load from RK bootloader. maybe same for kexec.
<naobsd>
ganbold: RK bootloader checks header of images on kernel/boot/recovery partition, it should have proper magic, size for trailing data, and CRC for data at end
<lunra>
Is FreeBSD running on any armv7 let alone RK3188/Cortex-A9?
<ganbold_>
naobsd: do we have such formats and details somewhere?
<naobsd>
oh I forgot to add parameter partition which is also need header
<naobsd>
rkflashtool can communicate bootloader via USB
<ganbold_>
interesting
<naobsd>
probably it will work with FreeBSD's libusb
<naobsd>
or if you need a tool for old ugen(4) interface, my original rkflash tool will work ;)
<ganbold_>
I was thinking to build kernel and then using some of those tools like mkbootimg to make kernel in proper format and put it in nand and try
<naobsd>
rkcrc -k will be enough
<naobsd>
(if you have kernel+ramdisk image built with mkbootimg on boot or recovery partition, it will be preferred)
<ganbold_>
ok and if it works that way then need to experiment other things like usb ehci, sd
<naobsd>
rkcrc -k in out, in is generally vmlinuz, just a data
<ganbold_>
so preferred way is boot to recovery partition?
<naobsd>
ganbold: I'm not sure which is preferred, it may vary by purpose
<naobsd>
you can choose any way which you like :)
<naobsd>
rkflashtool can write any data to any (logical) location of nand
<naobsd>
logical location = bootloader and some other important information are hidden, only partitions which can be accessed by Linux/Android are visible
<ganbold_>
good, but do I need to worry about breaking it? it seems it doesn't have jtag
<naobsd>
ganbold: try rkflash -p, it reads parameter partition (offset 0) from nand
<ganbold_>
naobsd: did you get radxa too?
<naobsd>
banbold: there is mask rom mode, which doesn't use bootloader on nand
<naobsd>
ganbold: yes I have 1 rock too
<ganbold_>
cool
<naobsd>
in mask rom mode, boot code in rom in CPU wait next code from USB
<naobsd>
unfortunately currently rkflashtool can't do anything with mask rom mode.
<ganbold_>
seems lots of complications
<naobsd>
if bootloader or some other important data is broken, you need to reflash some file with RK's Windows tools
<ganbold_>
ok, so I don't need to worry about breaking things :)
<naobsd>
ganbold: unless you are the man who don't like microsoft/windows ;)
<ganbold_>
I can use my son's windows :)
<naobsd>
I'm using my wife's one, sometimes she claims 'unknown program is installed on my PC!'
<ganbold_>
:)
<ganbold_>
naobsd: when you think you will be able to put all these informations on web/wiki?
<naobsd>
'when' is most difficult part for me for long time...
<naobsd>
I tried to write something about them several times...
<ganbold_>
yeah, that is the problem
<naobsd>
but now it's good time to try it again, radxa rock seems very nice and developer community is growing
<ganbold_>
that is true, we need all these detailed informations somewhere and developers can use it later on
<ganbold_>
like memory addresses of devices, registers etc.
<naobsd>
unfortunately there was no device and no vendor for open development community.
<naobsd>
until radxa rock.
<naobsd>
so information are guessed and/or result of trial&error
<ganbold_>
yeah, that is no good :(
<naobsd>
ganbold: there are some amount of code in linux mainline
<naobsd>
written from scratch
<naobsd>
it may help porting another kernel (but it's GPL of course)
<ganbold_>
yeah, have to look those files definetely
<naobsd>
I forgot detail but uart, clock things, and probably mmc and/or usb is supported
<naobsd>
I heard synopsys dwc is used
<ganbold_>
I see
<ganbold_>
I trying to build linux now, will see how it works on it
<naobsd>
well, no, bootloader is loaded from nand, then it waits commands from OTG
<naobsd>
you can read/write nand via USB with rkflashtool
<naobsd>
(you can use RK's windows tools too)
<ganbold_>
I see
<naobsd>
if you 1: press & hold recovery key 2: plug POWER (w/o OTG-PC connection), bootloader wait OTG connection few secs, then (if no OTG) bootloader try to start recovery mode
<naobsd>
recovery mode: load kernel+ramdisk from recovery partition (or kernel in kernel part and ramdisk in recovery part), and generally Android recovery(upgrade) program will run in this mode
<naobsd>
you may rewrite image in recovery
<naobsd>
you can read/write nand partition on Android/Linux, so you may use recovery mode to flash new image to kernel/boot/etc partition
<naobsd>
it may be useful because you can use shell and commands on Linux :)
<ganbold_>
let me try to build kernel and flash nand :)
<naobsd>
I see :)
<ganbold_>
naobsd: do you have script to build linux kernel?
<naobsd>
did you get source for radxa rock? probably other source for other RK3188 will boot but
<ganbold_>
I'm following wiki
<naobsd>
make ARCH=arm CROSS_COMPILE=/home/fun/linux-sunxi/gcc-linaro-arm-linux-gnueabihf-4.7-2013.04-20130415_linux/bin/arm-linux-gnueabihf- rk3188_steak_defconfig
<naobsd>
make ARCH=arm CROSS_COMPILE=/home/fun/linux-sunxi/gcc-linaro-arm-linux-gnueabihf-4.7-2013.04-20130415_linux/bin/arm-linux-gnueabihf- -j 4
<ganbold_>
I started building linux, let's see how that works
<ganbold_>
make complained about missing KDIR for ct11mp in /home/tsgan/rkutils/rk3188_kernel_3.0.36/drivers/gpu/drm/mali_drm/Makefile, but second make is somehow continuing
<ganbold_>
probably mali_drm isn't built and I don't need it right now
<naobsd>
I'm sorry I have no idea about current status of other's kernel source...
<naobsd>
ganbold: my rock has 4GB memory and RTL8188ETV wifi-only chip but 2GB ram
<naobsd>
I'm not it should be called "lite" or not... probably just a noname ES
<naobsd>
I'm sure rk3188_steak_defconfig is good for mine
<naobsd>
but I don't know all ES are same
<ganbold_>
still building linux :)
<naobsd>
my rock has kernel in kernel, android ramdisk(rootfs) in boot, and kernel+ramdisk in recovery. so I just reflash kernel image by rkcrc -k to kernel
<naobsd>
in this case, my kernel is used when normal boot, but not recovery boot
<ganbold_>
wise
<ganbold_>
it is probably linking now , vm is very slow responding
<naobsd>
hehe
<naobsd>
I want new machine to build!
<naobsd>
something like Ivy Bridge 4core 8GB ram
<naobsd>
lower TDP model is preferred ;)
<ganbold_>
I bought for my son Ivy bridge 4core i7-3820 with 16GB ram, 128GB SSD and 2TB HDD, nvidia GTX770 gaming desktop, I should ask him to put vm for me :)
<naobsd>
you should!!!
<naobsd>
and for me too ;)
<naobsd>
only 60GB is left on my 1TB HDD on build machine
<ganbold_>
:)
<ganbold_>
:) what I'm doing, I should ask you to give some files like zImage, Image etc :) but anyhow I should try myself :)
<naobsd>
ah, I wiped them to make tarball!
<naobsd>
ganbold: your kernel can be booted?
<ganbold_>
still linking I think
<ganbold_>
I see "LD vmlinux.o" line and nothing more yet
Super-noob has quit [Ping timeout: 245 seconds]
<naobsd>
lol
<naobsd>
you should try to build kernel on radxa rock!
<naobsd>
I/O must be weak, but...well, kernel source + output consumes only 800MB
<naobsd>
we will have 4core 2GB(ram and storage) small build machine!
<naobsd>
there are some developers who pushed source from same taballs, they made/are making own changes on his repo, but I don't merge/pull any of them
<ganbold_>
hm, how to connect this OTG to vm in mac?
<naobsd>
there are some repos for RK30/31 on github, but they didn't share history. same for me, currently linux-rockchip-3.0 is same as my personal repo, it's not central thing at all.
<naobsd>
ganbold: which VM are you using? I guess there is menu item or icon to do it
<naobsd>
ganbold: you may need to change guest setting to include virtual USB host function
<ganbold_>
ah, I had to do reboot bootloader
<ganbold_>
should I paste rkflashtool p output here?
<naobsd>
radxa kernel source should contain newer changes from RK and small RR specific changes
<ganbold_>
which one?
<naobsd>
ganbold: I told you some hours ago
<ganbold_>
ah, ok
<naobsd>
rk3188_steak_defconfig is for RR
<naobsd>
on that source
<ganbold_>
but that is android, right?
<arokux2>
naobsd: is the wire you have shown to me something special or can I find it somewhere else?
<naobsd>
I feel it's special, but I'm not sure about connector/wire world... I gues technically it should not be special, but it's a "nice idea!" thing
<naobsd>
in short: I don't know any other online shop that sells that wire
<tonikasch>
mmm, omegamoon is a developer as well :)
<arokux2>
tonikasch: yeah, but where does he have those RK patches from?
<arokux2>
tonikasch: also, why is he dropping code without history :(
<tonikasch>
I don't know
<tonikasch>
I'll ask
<hipboi>
arokux2, you have the latest code too
<hipboi>
i have created the account for you
<arokux2>
hipboi: where, on nand?
<hipboi>
check your email
<hipboi>
from git@git.linux-rockchip.org
<tonikasch>
I'm still tyding up my environment, have just passed exams and for much time I've been almost away
<hipboi>
omegamoon is merging his 3.0.72 kernel with latest rk kernel
<arokux2>
hipboi: nothing.
<hipboi>
arokux1, :O
<hipboi>
let me check
<hipboi>
arokux2, arokux@gmail.com?
<arokux2>
hipboi: yep
<hipboi>
arokux2, i sent you the information
<hipboi>
good night all
<arokux2>
hipboi: oh, it was from gitlab@linux-rockchip.org and landed in spam!
hipboi has quit [Quit: Leaving]
<tonikasch>
arokux2 oh, yes, latest rk patches were taken from the git.linux.rockchip.org radxa repository
tonikasch has quit [Quit: Bye!]
tonikasch has joined #linux-rockchip
Omegamoon has joined #linux-rockchip
<Omegamoon>
hi there
<Omegamoon>
just reading the IRC logs...
<Omegamoon>
so @arokux2, please explain, what is the problem?
<Omegamoon>
@arokux2 if you say "also, why is he dropping code without history"
<arokux2>
Omegamoon: yes :) it is nice to have history, like 3.0.72 kernel history plus other commits.
<arokux2>
Omegamoon: so that it can be easily see what rk added.
<Omegamoon>
well, then clone the repo and help start making the history
<tonikasch>
Me, I have already cloned the repo and noticed the GPIOs are specifiable now :D As soon as I receive the radxa board I'll begin testing it
<arokux2>
Omegamoon: sure. but didn't you have it?
<Omegamoon>
arokux: what I had was tons of kernels for all kind of devices, which I merged and changed, then I upgraded the kernel to 3.0.72, and after that I merged the result with the radxa git, which contains the latest rockchip changes. It's hard to get this done in the first place, so forgive me if I didn't build a history of each and every change I made
<tonikasch>
:)
<Omegamoon>
That was for arokux2 previously, not arokux... all these personalities ;-)
<arokux2>
Omegamoon: I apologize if said it to blunt.
<arokux2>
I've just get used to the way it is done in linux-sunxi.
<arokux2>
Omegamoon: yeah.. I'm logged in from two machines, so that is why two nicks.
<Omegamoon>
no offense taken, maybe it's just me having to deal with too many people acting in a gimme-gimme-gimme fashion, where good is never good enough it seems ;-)
<arokux2>
Omegamoon: you know it can be always better. I'm not working in that fashion really, I'd like to help with mainlining...
<arokux2>
Omegamoon: what are your plans?
<Omegamoon>
one code base to begin with, which is what I tried
<arokux2>
Omegamoon: and then mainline?
<Omegamoon>
that's a huge step!
<arokux2>
Omegamoon: what is the next one then?
<arokux2>
:)
<Omegamoon>
but if you have experience in that field...
<Omegamoon>
feel free to join!
<arokux2>
I have joined already.
<arokux2>
I can add usb host to kernel
<arokux2>
I (almost) added it to sunxi
<Omegamoon>
I was thinking more about getting these annoying closed sourced stuff out
<Omegamoon>
like Mali, wifi etc
<Omegamoon>
proper bootloader
<arokux2>
Omegamoon: what wifi chip is it?
<Omegamoon>
which device?
<tonikasch>
there are many
<arokux2>
the rtl8188cu which you can find with sunxi devices works with mainline kernel driver - rtl8192cu
<Omegamoon>
we have realteks, broadcoms, espressif, marvell...
<arokux2>
oh..
<Omegamoon>
that's cool, but there are some 20 to support accross different devices
<tonikasch>
and the always-present gpios mapped differently for every device
<Omegamoon>
which is not nice in the current situation where these are handled through a few closed source kernel modules
<Omegamoon>
so that has a high priority :-)
<Omegamoon>
Mali support for linux distros like ubuntu... same problem
<tonikasch>
brcmsmac was supposed to support bcm40181 and bcm40183 wifi chips but I don't remember the actual status...
akaizen has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
tonikasch has quit [Quit: Bye!]
Omegamoon has left #linux-rockchip [#linux-rockchip]
tonikasch has joined #linux-rockchip
akaizen has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
<naobsd>
arokux: origin of rockchip's 3.0 kernel source are not sure. they based on AOSP 3.0, but it also contain some changes outside of AOSP 3.0 history.
<naobsd>
arokux: my linux-rockchip-3.0 is based on AOSP 3.0. I chosen most similar point (my guess) to make branches to put RK kernel sources
<naobsd>
arokux: so if you want to merge latest AOSP 3.0, you can do as usual... git merge android-3.0 on linux-rockchip-3.0 (and resolve conflict).
<naobsd>
5310c3c75b0dd301da8c76e0971e82d194df6fdb this is the point I chosen to merge RK 3.0.36 source
<naobsd>
I will push RK kernel source in 4.2.2 SDK on top of this point with RK's history
<naobsd>
so you can see both AOSP and RK history as much as possible
<arokux2>
naobsd: thanks for explanation. no I have no interest in android, only in some clue how RK hardware works so it can be mainlined.
<naobsd>
history contain author info, it's very important
<naobsd>
if you want to cherry-pick
<naobsd>
to mainline
<naobsd>
it's difficult to know who wrote some lines without history... I think "committer"(developer around here) is recorded as "author" other RK kernel source repo
<arokux2>
naobsd: I doubt you can cherry pick something at all. normally a rewrite from scratch is needed.
<naobsd>
arokux: yes, actually I think so too :)
<naobsd>
but I like source with history
<arokux2>
naobsd: this is just from what I've see in sunxi land.
<arokux2>
naobsd: yes, me too, that is why I've asked.
<naobsd>
if I made change somewhere and record history, I can show commit id to tell what I changed
<naobsd>
"please run diff between other-tree and my-tree to get my change" is not good
<naobsd>
ganbold: there is "KERNEL_IMG: 0x60408000" in parameter, but I'm not sure it can be changed