<bashintosh>
karlp: the RK3288 has (1) a dedicated EHCI controller which can only work as host controller and (2) an OTG controller (DRD) which provides both OTG and HC functionalities. Internally in the SoC, (2) is a 2-in-1 controller which uses a single driver (DWC_OTG_310 on RK kernels or DWC2 on mainline). (1) has dedicated pins on the SoC and also uses a different driver (EHCI only, non OTG related). (2) has
<bashintosh>
several issues working as host controller when using the DWC_OTG_310 driver (observed on multiple boards, e.g. Firefly) while (1) works flawlessly in all situations. Radxa Rock 2 has its USB ports (behind the on-board hub) interfaced to the SoC EHCI controller, while other boards have their USB ports interfaced to the OTG controller. DWC is short for DesignWareController, developed by Synopsys.
<naobsd>
akaizen: sector 64 is important
<akaizen>
hmm I wrote zeros to it
<akaizen>
maybe in ubuntu /dev/mmcblk0 is remapped to only the useraccessible portion of the eMMC
<naobsd>
akaizen: then you'll get brick. you have to write file build with u-boot
<akaizen>
cant i bridge pins and go to MASK ROM and then just flash another image?
<naobsd>
ubuntu, userland, doesn't do anything in kernel driver. which kernel are you using?
<akaizen>
Linux firefly 3.10.0 #40 SMP PREEMPT Tue Jan 27 16:12:04 CST 2015 armv7l armv7l armv7l GNU/Linux
<akaizen>
Firefly ubuntu image
<naobsd>
hmm... I think I could rewrite idb area via mmcblk0 with Android based kernel...
<akaizen>
is there a usb tool that allows rewrite of that area?
<akaizen>
oh what about AndroidTool
<naobsd>
AndroidTool.exe/upgrade_tool
<akaizen>
ah ok I can select the images and set the offsets
<naobsd>
no
<naobsd>
loader image is handled differently (compared to other images for partitioned area)
<naobsd>
well
<naobsd>
please dump sector 64-68 on mmcblk0
<naobsd>
it should have some data
<akaizen>
it does I can see SPL stuff there
<akaizen>
that I wrote
<naobsd>
ah, what I asked is original state....
<naobsd>
well
<akaizen>
ah ok let me reflash to the base image
<naobsd>
which firefly image are you using specifically?
<naobsd>
94 firefly-rk3288/spl/u-boot-spl-dtb.bin out && \
<akaizen>
yep
<naobsd>
this is what you need
<akaizen>
yes i got the out
<akaizen>
I used -T rkimage
<akaizen>
should I still use rksd?
<naobsd>
sd and emmc is same in this context
<akaizen>
ah ok
<naobsd>
to wipe all special area on emmc/sd, dd if=/dev/zero of=/dev/XXX bs=4M count=1 will be enough
<akaizen>
is this necessry?
<naobsd>
but it erase every thing includes e.g. serialno
<akaizen>
Ok I'll back it up first
<naobsd>
I think you don't need to wipe it (get mask rom mode) for now
<naobsd>
just write rksd out with seek=64 and u-boot-dtb.img with seek=256
<naobsd>
"dd if=/dev/zero of=/dev/mmcblk0 bs=1M" will wipe everything includes parameter, kernel, rootfs, etc
<akaizen>
k i'll try that
<naobsd>
it's too much and may take time. if you really want to wipe whole emmc, you can use blkdiscard command which emit DISCARD command, not "write tons of zero)
<akaizen>
I dont have ubuntu vm right now
<akaizen>
so no upgrade_tool
<akaizen>
I made the rksd out
<akaizen>
i'll try that
<naobsd>
VM? you have to run dd on ubuntu on firefly
<naobsd>
ah,
<naobsd>
anyway
<naobsd>
I totally forgot that there are idb backups ;)
<naobsd>
I usually don't prepare backups while prepareing bootable SD card ;)
<naobsd>
(same for parameter backups)
<naobsd>
sorry ;)
<akaizen>
no worries :) appreciate the help
<naobsd>
I think I explained there is no idb backup on here very recently...
<akaizen>
cool it worked
<akaizen>
problem was using rkimage instead of rksd
<naobsd>
akaizen about Rockchip official tools, AndroidTool.exe for Windows and upgrade_tool for Linux
cosm has quit [Quit: Leaving]
<naobsd>
akaizen: well, sorry, what's "this"?
<akaizen>
naobsd: do you know the commands for upgrade_tool or Androidtool?
<akaizen>
Ah "this" is the u-boot and the spl from sjoerd branch / u-boot-rockchip
<naobsd>
to write something to idb area?
<akaizen>
vs flashing ubuntu, rsyncing images over and then dd
<naobsd>
you have to make another file like RKxxLoader.bin
<naobsd>
usually that bin contains DDR init blob and loader blob
<akaizen>
DDR init blob makes sense
<akaizen>
but theres still RK propritary laoder blob too?
<naobsd>
well
<akaizen>
ah i must run. back in an hour!
<naobsd>
you have to prepare RKxxLoader.bin with SPL(-T rkimage in this case) and u-boot-dtb.img
<naobsd>
later
<akaizen>
oh and create my own RKxxLoader.bin?
<naobsd>
that's what AndroidTool/upgrade_tool can handle
<naobsd>
as loader
<naobsd>
or you can implement new function in rkflashtool ;)
<akaizen>
so we need a opensource tool that can write to emmc is what you are telling me :)
<akaizen>
haha yes :)
<akaizen>
will discuss more with you later
<naobsd>
well, mmc driver is working in u-boot
<naobsd>
I see
<akaizen>
bye for now
<naobsd>
currently, "go mask rom mode" -> "load SPL from OTG" -> "load u-boot from SD/eMMC (or OTG, but how?)" -> "update idb area on SD/eMMC from u-boot" might be only possible way to install open source loader to on-board flash without proprietary tool
<naobsd>
if SPL can handle OTG and some protocol to receive data from PC, "load u-boot from OTG" is possible
<naobsd>
I want to know how proprietary DDR init blob loads next loader code from OTG... I guess it's done by mask rom
cnxsoft has joined #linux-rockchip
<akaizen>
naobsd: "Future priorities: Boot U-Boot proper over USB OTG (at present only SPL works)"
<akaizen>
sjg == sjoerd?
<akaizen>
Ah I think no
<akaizen>
How do make partitions in u-boot on the emmc?
<akaizen>
ah switch to mmc dev 1
<akaizen>
and then thats the eMMC and then I can part it
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
amstan_ has joined #linux-rockchip
amstan_ has joined #linux-rockchip
amstan has quit [Ping timeout: 264 seconds]
akaizen has quit [Remote host closed the connection]
akaizen_ has joined #linux-rockchip
sunilmohan_w_ has joined #linux-rockchip
lautriv_ has joined #linux-rockchip
lautriv__ has quit [Ping timeout: 240 seconds]
amstan_ has quit [Ping timeout: 264 seconds]
nighty^ has joined #linux-rockchip
<sjoerd>
akaizen_: no i'm not sjg :)
amstan has joined #linux-rockchip
amstan has joined #linux-rockchip
rory096 has quit [Ping timeout: 246 seconds]
<rperier>
hey all
markm_ has quit [Ping timeout: 252 seconds]
markm_ has joined #linux-rockchip
<lautriv_>
got anyone more details about such tvbox called ZY_X6_V21 (someone said the hardware is CS908) ?
<akaizen_>
sjoerd: do you use SD card or eMMC? How can I format the eMMC from u-boot so I can load up mmind kernel and get gentoo running ;)
<akaizen_>
sjoerd: Hehe, names initalis were similar but then I saw the full names on the commits :)
<sjoerd>
akaizen_: I use both
<sjoerd>
I never formatted the emmc from u-boot
<akaizen_>
How do you normally format emmc then?
<akaizen_>
Also with this mainline u-boot what proprietary parts are left?
<sjoerd>
none
<akaizen_>
I was talking with naobsd and he said the DDR init code, and some rock-chip loader
<sjoerd>
I just have a gpt partition talbe and one rootfs
<sjoerd>
the only thing that's still there from rockchip is hardwared in the SOM
<sjoerd>
sorry SOC
<sjoerd>
akaizen_: did you get working network in u-boot on your firefly
<akaizen_>
so I'd be fine if I zero'd out the first 4MB and then wrote just the spl and uboot?
<naobsd>
I didn't say there is proprietary things in u-boot mainline
<akaizen_>
sjoerd: yep
<sjoerd>
akaizen_: correct
<akaizen_>
naobsd: not in mainline, but still in the loader
<naobsd>
proprietary loader has proprietary thing, yes of course
<naobsd>
but you're talking about u-boot mainline
<akaizen_>
yes
<naobsd>
u-boot mainline is not proprietary loader
<akaizen_>
How does DDR get init in mainline then?
<akaizen_>
sjoerd: So basically spl -> u-boot -> network boot kernel -> get a simple shell, format emmc and then load kernel and rootfs?
<sjoerd>
ack
<akaizen_>
then I can just dd if=/dev/mmcblk0 and compress it to have a reproducable image
<akaizen_>
and need to update rktools to support writing it :)
<akaizen_>
and with mmind kernel is good starting point to port in the VPU code and then hopefully ARM can help out with mali GPU and bam
<akaizen_>
fully accelerated, fairly open platform :)
<naobsd>
I think mali is already working on mmind's branch
<akaizen_>
I dont plan to use X11
<akaizen_>
just fbdev as we wrote our own display manager
<naobsd>
ARM already released libMali for fb
<akaizen_>
yep, someone here mentioend performance wasn't all that great
<akaizen_>
it should be at least as good as the binaries from Android 5.1
<akaizen_>
~24 fps in Open GL ES 3 demo at 1080p
<akaizen_>
bnaries from 4.4 were giving ~12 fps
<sjoerd>
fbdev :'(
<akaizen_>
In mmc in u-boot there are dev {0,1}. 1 = eMMC, what is dev 0? and What is SPI flash?
<sjoerd>
0 is SD card
<sjoerd>
does firefly have spi flash?
<sjoerd>
spi flash is normally accessed through the sf command iirc
<lautriv_>
yes SPI is not MMC
<sjoerd>
sure i just wasn't sure about the u-boot command to access it, not whether it's mmc ;)
<akaizen_>
Sorry I'm asking questions about different things as they enter my mind
<lautriv_>
does a stock android rom usually contain .config somewhere ?
<akaizen_>
not generally :(
<naobsd>
generally SDK (!rom) contains defconfig
<akaizen_>
if you are luckly and someone compiled the kernel with config option enabled then check /proc/config.*
<lautriv_>
naobsd, i guess both is not sufficient since i appear to have issues with some GPIO resp. devices to enable :(
<akaizen_>
oh cool sjg also did the u-boot verfied boot code
<akaizen_>
lautriv_: what device? what kernel?
<akaizen_>
I may not have the right one but might be able to get you a starting point
<akaizen_>
(had to recompile kernel for rk3188 to enable i2c)
<akaizen_>
in android
<lautriv_>
akaizen_, SoC is 3188T, device has 8G flash/2G RAM but when i try to flash from radxa, it crashes on bootup and when i replace the kernel for rabian image extracted from stock ROM, it complains about I2C-2
<naobsd>
if "device" is not Radxa Rock series, don't flash anything from radxa
<akaizen_>
lautriv_: Hmm.. we ended up using stock kernel or a updated working kernel and radxa android sdk
<lautriv_>
naobsd, i used only the SD-Images to see how far i can use it but each boot kills my flash :(
<naobsd>
crash/complain/any problem is very expected result...
<lautriv_>
sure but still the same SoC and since there is no other source ...
<naobsd>
don't try any image for another device. 1. learn about device 2. modify source to match device 3. build image
<lautriv_>
naobsd, i don't expect to get any matching config since the init is most likely different but there are no tools suitable for androshit.
<akaizen_>
If I remember correctly later versions came with RK3188T
apritzel has joined #linux-rockchip
apritzel has left #linux-rockchip [#linux-rockchip]
apritzel has joined #linux-rockchip
<naobsd>
unfortunately Rockchip based products are very similar (or same but different name), roms have some compatibility. so people misunderstand "rom is universal" "it's same SoC, same rom must work" etc
<akaizen_>
Maybe CX-919
<naobsd>
I think board picture is most reliable source to identify device
<akaizen_>
yep .. i have about 10 different RK3188 SDKs
<akaizen_>
platform SDKs
<naobsd>
SoC is same, Wifi is same, RAM/NAND size is same, are not helpful at all
<akaizen_>
naobsd: its sometimes helpful.. it can get you close
<akaizen_>
if all those are the same .. good chance they're the same model, jsut different mfgs / resellers
<naobsd>
many different boards use same SoC, same Wifi, same RAM, etc etc...
<akaizen_>
true, maybe if similar specs 49% chance of it working :)
<akaizen_>
better than 5 - 10% if dissimial specs
<akaizen_>
alright off to sleep so I can get hacking tomorrow.
<akaizen_>
Good night everyone!
<lautriv_>
would be easier if they didn't use raped kernels and propietary blobs anyway.
<naobsd>
good night
<akaizen_>
naobsd: oyasuminasai
<naobsd>
:)
* lautriv_
wonders how often a reflash is neccessary beforethat thingie goes /dev/null
ssvb has quit [Ping timeout: 260 seconds]
ssvb has joined #linux-rockchip
<rperier>
when you have time, can someone ack my patch for clk_peri please ?
xcasex has quit [Read error: Connection reset by peer]
xcasex has joined #linux-rockchip
xcasex has quit [Read error: Connection reset by peer]
xcasex has joined #linux-rockchip
<lautriv_>
bah, i found that rk3188 is rather simple clone but stops to boot proper when UART is connected. Probably a ground-loop o.O
chris2 has quit [Ping timeout: 250 seconds]
chris2 has joined #linux-rockchip
<lautriv_>
do we need some errata on 3188 ?
cnxsoft1 has joined #linux-rockchip
cnxsoft has quit [Ping timeout: 264 seconds]
cnxsoft1 is now known as cnxsoft
ssvb has quit [Ping timeout: 240 seconds]
ssvb has joined #linux-rockchip
SamuelMoraesF has quit [Ping timeout: 244 seconds]
SamuelMoraesF has joined #linux-rockchip
cristian_c has joined #linux-rockchip
cristian_c has quit [Read error: Connection reset by peer]
cristian_c has joined #linux-rockchip
cristian_c has quit [Read error: Connection reset by peer]
cristian_c has joined #linux-rockchip
cristian_c has quit [Read error: Connection reset by peer]
cristian_c has joined #linux-rockchip
cristian_c has quit [Read error: Connection reset by peer]
cnxsoft has quit [Quit: cnxsoft]
rory096 has joined #linux-rockchip
cristian_c has joined #linux-rockchip
ganbold has quit [Ping timeout: 255 seconds]
cristian_c has quit [Excess Flood]
cristian_c has joined #linux-rockchip
JohnDoe_71Rus has joined #linux-rockchip
VargaD has quit [Ping timeout: 264 seconds]
VargaD has joined #linux-rockchip
GriefNorth has joined #linux-rockchip
GriefNorth has quit [Remote host closed the connection]
GriefNorth has joined #linux-rockchip
GriefNorth has quit [Read error: Connection reset by peer]
GriefNorth has joined #linux-rockchip
apritzel has quit [Ping timeout: 264 seconds]
GriefNorth has quit [Client Quit]
GriefNorth has joined #linux-rockchip
GriefNorth has quit [Client Quit]
GriefNorth has joined #linux-rockchip
GriefNorth has quit [Read error: Connection reset by peer]
akaizen_ has quit [Remote host closed the connection]
paowz has joined #linux-rockchip
GriefNorth has quit [Ping timeout: 256 seconds]
paowz has quit [Remote host closed the connection]
cristian_c has quit [Ping timeout: 246 seconds]
gb_master has joined #linux-rockchip
cristian_c has joined #linux-rockchip
akaizen_ has joined #linux-rockchip
akaizen_ is now known as akaizen
<akaizen>
lautriv_: I'd be interested to see how errata for Asian SoC compare to other vendors
<akaizen>
mmind00: Which rk3288 defconfig do you recommend for 'v4.2-rockchip-somewhat-stable' tag?
<karlp>
akaizen: you'd probably need to see an errata sheet first right?
<akaizen>
Heh, I've never seen a Rockchip or Allwinner errata sheet
<akaizen>
I've only seen Intel and AMD
<akaizen>
and even those are hidden deep in their site
<mmind00>
akaizen: multi_v7_defconfig :-P ... or if you want a head-start use rk3288_veyron_defconfig that's in that branch/tag
<akaizen>
Will they work for firefly?
<akaizen>
whats multi_v7 (arm7a?)
<mmind00>
akaizen: yep ... there is not su much difference between them, and the firefly regulator drivers are also enabled there ... depending on what bootloader you're using, you might need to override the kernel commandline (CONFIG_CMDLINE and CONFIG_CMDLINE_FORCE)
<akaizen>
mmind00: thanks for the heads up! I'm using sjoerd upstream u-boot branch with ethernet enabled :)
<mmind00>
akaizen: multi_v7 is the generic config that builds an image including basic support for all possible arm-v7 socs
<akaizen>
yea i saw every SoC enabled in there
<mmind00>
akaizen: not sure how the kernel commandline is handled there ... I was talking about the vendor-uboot that always injects the crappy android kernel commandline into the atags
<akaizen>
mmind00: well more opportunity to learn about the boot process - hopefully gentoo or some other distro wiki + u-boot manual will clarify things
<akaizen>
mmind00: sjoerd mentioned that with the upstream u-boot his boot process and partition layout is fairly clean
<mmind00>
akaizen: :-) ... I guess there will be some rather easy way to handle that ... once you get rid of the vendor-stuff most things tend to get easier
<akaizen>
Build (rk3288_veyron_defconfig) OK with latest armhf linaro toolchain! Hopefully all systems will be nominal ;)
akaizen has quit [Read error: Connection reset by peer]
akaizen_ has joined #linux-rockchip
cristian_c has quit [Quit: Bye]
<akaizen_>
hmm how can I create a uImage for u-boot to receive via tftp?
ganbold has joined #linux-rockchip
akaizen_ has quit [Read error: Connection reset by peer]
akaizen has joined #linux-rockchip
nighty^ has quit [Quit: Disappears in a puff of smoke]
gb_master has quit [Remote host closed the connection]
<naobsd>
akaizen: btw, who talked about rockchip based device at mozilla?
<akaizen>
naobsd: This was a while back but I spoke with the b2g folks who were working with the Matchstick folks
<naobsd>
ah
<akaizen>
Mozilla wants a succesfuly b2g device / platform
<akaizen>
successful*
<naobsd>
currently some mozilla guys are working for rk3066 based firefox os board
<akaizen>
rk3066 is soooo old
<akaizen>
Would be nice to see them start with something usable like rk3288
<naobsd>
what they want is platform, not new architechure, performance, etc, etc (of course it's better if possible)
<akaizen>
Well platform is meaningless if it cant run any apps lol
<akaizen>
rk3066 will slow to a crawl on HTML JS apps
<naobsd>
"firefox os with Web APIs"
<naobsd>
platform.
<akaizen>
yep
<akaizen>
HTML5 CSS and JS everything
<akaizen>
Its basically Android base layer - dalvik + web APIs = firefox os = b2g
<akaizen>
and you need to run everyhting in a isolated process / vm since its all JS.. you dont want a third party app to be able to call into your phone dialer
<akaizen>
or start sending SMS
<naobsd>
understanding how to make board is one of their goal. making latest/highest target is not main goal.
<naobsd>
btw I know some about firefox os.
naobsd has quit [Remote host closed the connection]
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-rockchip
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-rockchip
akaizen has quit [Read error: Connection reset by peer]