leowt changed the topic of #linux-rockchip to: Rockchip development discussion | http://linux-rockchip.info | http://irclog.whitequark.org/linux-rockchip
ssvb has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-rockchip
akaizen_ has joined #linux-rockchip
tonikasch has quit [Quit: Bye!]
akaizen has quit [Ping timeout: 256 seconds]
akaizen_ has quit [Ping timeout: 245 seconds]
akaizen has joined #linux-rockchip
akaizen_ has joined #linux-rockchip
akaizen has quit [Ping timeout: 252 seconds]
akaizen__ has joined #linux-rockchip
akaizen has joined #linux-rockchip
akaizen_ has quit [Ping timeout: 246 seconds]
akaizen__ has quit [Ping timeout: 240 seconds]
hramrach has quit [Remote host closed the connection]
hramrach has joined #linux-rockchip
Astralix has joined #linux-rockchip
Astralix1 has quit [Ping timeout: 264 seconds]
atiti has joined #linux-rockchip
eval-_ has quit [Ping timeout: 248 seconds]
eval- has joined #linux-rockchip
shineworld has joined #linux-rockchip
ganbold_ has joined #linux-rockchip
eebrah|noPC has joined #linux-rockchip
eebrah|noPC has quit [Ping timeout: 260 seconds]
<ganbold_> arokux1: got your radxa board? if so did you look for gpio pwr pin for usb host?
sjoerd has joined #linux-rockchip
shineworld has quit [Ping timeout: 248 seconds]
ganbold has quit [Remote host closed the connection]
<sjoerd> Hey, is there a known mod for a mk802iv board to add a serial port
deviker has joined #linux-rockchip
<deviker> Hi, I've a Mk908 V3 stick and I succeded compiling several kernels (omegamoon, the original one from the sdk and one with marvin). I open rkandroidtool and it seems it sends the new kernel to the device. Using rkflashtool.new I obtain success too. The problem is that when I reboot the stick it shows the old kernel instead of the new one. Any idea why?
ganbold_ has quit [Remote host closed the connection]
ramcq has joined #linux-rockchip
<deviker> (BTW I've got serial port input/output to/from the stick)
ganbold_ has joined #linux-rockchip
Theueip_ has quit [Quit: Leaving...]
arokux2 has joined #linux-rockchip
hramrach has quit [Ping timeout: 240 seconds]
hramrach has joined #linux-rockchip
hramrach has quit [Write error: Broken pipe]
hramrach has joined #linux-rockchip
Galland has joined #linux-rockchip
eebrah_ has quit [Ping timeout: 240 seconds]
eebrah has joined #linux-rockchip
eebrah is now known as Guest60148
pfalcon has joined #linux-rockchip
sarg has quit [Read error: Connection reset by peer]
massi has joined #linux-rockchip
naobsd has joined #linux-rockchip
<pfalcon> naobsd: Hi, from http://irclog.whitequark.org/linux-rockchip/2013-09-28: "<naobsd> logical location = bootloader and some other important information are hidden, only partitions which can be accessed by Linux/Android are visible " - how to get to bootloader in NAND then?
<naobsd> pfalcon: https://github.com/crewrktablets/rkflashtools/blob/master/rkflashtool.c "rkflashtool i" will do the work, but
<naobsd> it's not same format as RK...bin in firmware images
<naobsd> pfalcon: why you need to do it?
<pfalcon> naobsd: any specific args for it? I tried to give at offset 0, and mostly saw junk, definitely didn't start with "BOOT"
<pfalcon> naobsd: for one, picuntu kernel doesn't boot for me from recovery partition, I wanna understand why. otherwise, it all should be reversed and made well-known ;-)
<naobsd> pfalcon: please read my message carefully. format is different
<pfalcon> naobsd: crypted? obfuscated? is there any docs on it?
<naobsd> pfalcon: you can read RK...bin in firmware, but it will not gives answer
<naobsd> well
<pfalcon> naobsd: ok, thanks for hints so far, will keep reading logs
<naobsd> e.g. do you think it's possible to get original .iso image from HDD which is installed by .iso image? logically both have same information, but format must be different.
<naobsd> pfalcon: if you flashed something into recovery partition, it must be kernel+ramdisk image, not kernel-only image
<naobsd> please tell more detail about "picuntu kernel doesn't boot"
<pfalcon> naobsd: I'd assume it would be still possible to grep for contents on hdd ;-)
<pfalcon> naobsd: after "adb reboot recovery" lying android bot is shown
<naobsd> you can see something from "rkflashtool i" other than "BOOT" :)
<naobsd> so it's still Android recovery... hmm...
<pfalcon> naobsd: I found mentions of that on forums. Suggestion was "install finless rom first". but no exact version reference, and it's still suggestion for kiddies. I'd like to udnerstand what's wrong with my stick. Bootloader is obvious suspect
<pfalcon> naobsd: it seems my stick lacks any sensible recovery
<pfalcon> naobsd: so, I wanted to extract bootloader from my device and compare with one shipped by finless
<naobsd> which tool and command line you used to make image for recovery?
<naobsd> kernel+ramdisk image is almost same as standard android boot image, but different a little. bootloader checks it's valid, and if not
<pfalcon> naobsd: old january stuff from picuntu: http://code.google.com/p/rk3066-linux/downloads/list . it has ANDROID header
<naobsd> well
<naobsd> you just used ug802recovkernel.img as is?
<naobsd> w/o modification?
<naobsd> or you did something to replace kernel?
<pfalcon> naobsd: yup, flashed with rkflashtool w
<pfalcon> no replacements
<naobsd> I see. offset and size for "rkflashtool w" were correct?
<Galland> do a "rkflashtool r" to verify you effectively overwrote it
<naobsd> partition layout may be different by device/firmware(it can be changed), you have to use correct value for your device/firmware that you're currently using
<Galland> do: ./rkflashtool r 0x0 0x2000 > parm.bin
<naobsd> to get parameter, "rkflashtool p" does the work
<Galland> open the parm.bin and check the size@offset (recovery) numbers
<pfalcon> naobsd: offsets were correct
<pfalcon> Galland: I did readback
<Galland> gotta update my rkflashtool to latest version
<pfalcon> Galland, naobsd: I did "rkflashtool p" of course
<Galland> what stick do u have pfalcon?
<pfalcon> SO, my hypothesis is that bootloader doesn't support recovery partition well enough
<pfalcon> Galland: mk809/rk3066
<pfalcon> mk809 appear to be the same board as ug802, just in different case
<Galland> please paste here the result of "rkflashtool p" and your recovery writing command
<naobsd> if you can see robot, recovery works well. it also is needed to do "factory reset" in android setting menu
<naobsd> I never heard "special bootloader" for rockchip devices
<naobsd> as galland suggested, check parameter and command line, and read & verify after flashed
<pfalcon> Galland: http://pastebin.ca/2461636 , pretty standard
<Galland> and your full commandline for writing the recovery image?
<pfalcon> Galland: sudo ./rkflashtool w 0x10000 0x8000 <ug802recovkernel.img
<pfalcon> Galland, naobsd: same issues as I experience: https://www.miniand.com/forums/forums/picuntu-linux/topics/can-t-reboot-recovery
<Galland> and since you then do a read and binary compare to verify it...
<sjoerd> hmm, i've build the kernel.img from github.com/aloksinha2001/Linux3188.git, flashed it into the kerenl partitions on the nand but it doesn't look like my mk802iv is actulaly booting that kernel
<Galland> have you tried what TecKnight suggested about flashing it from windows preceded by an Erase IDB?
<pfalcon> Galland: I did, as I told
<pfalcon> Galland: I don't have windows, put of prob. also, would like to understand where problem lies, like extract current bootlaoder, compare it with shipped by finless
<Galland> if you don't have windows how did u use RK's tool to do an Erase IDB?
<Galland> rkflashtool is not windows RK tool
<naobsd> pfalcon: you also can write image to partitions on Android
<pfalcon> Galland: I use only rkflashtool from linux. "I did" == read back and binary-compared
<naobsd> pfalcon: please try, disconnect power, boot into *bootloader* again, then read back once again...
<pfalcon> naobsd: the problem I have is that (some) bootlaoder doesn't boot stuff written to recovery partition ;-)
<naobsd> pfalcon: you can see robot, it must be recovery
<pfalcon> naobsd: did that too ;-). stuff is there, doesn't boot. sucks, I know. spending second day on that, so embarrassing ;-)
<naobsd> sjoerd: "it doesn't look like my mk802iv is actulaly booting that kernel" please explain detail
<pfalcon> naobsd: I see broken robot with exclam sign, nothing more. that can me "bootloader could not found recovery image"
<sjoerd> naobsd: it always boots ``Linux version 3.0.36+ (hotack@hotack-desktop) (gcc version 4.4.3 (GCC) ) #45 SMP PREEMPT Fri May 3 16:58:13 CST 2013''
<naobsd> pfalcon: no, bootloader doesn't show anything on screen
<naobsd> pfalcon: can you still boot Android (boot partition)?
<naobsd> sjoerd: you flashed kernel-only image into kernel partition?
<pfalcon> naobsd: well, there're whole chain of bootloaders... bootrom, primary, then what's in "boot" partition, etc. understanding that mess is what I'd like to do ;-)
<pfalcon> naobsd: I can. After showing that broken droid screen, it hangs for ~1min\, then boots into normal android too. no factory reset happening.
<sjoerd> naobsd: i did, i also boot the zimage into an android image and put that on eitehr boot or recovery but same thing
<Galland> pfalcon: then get the RK3066 Tech. Ref. Manual and read the boot part: http://hwswbits.blogspot.com/2013/06/full-rk3066-technical-reference-manual.html
<Galland> sjoerd: does a pre-compiled recovery work for you (like the ones linuxium published on G+)?
<naobsd> sjoerd: your kernel image has "KRNL" header?
<pfalcon> Galland: I did, not enough info there. Where/how exactly bootrom searches for bootloader in nand? how I can see that nand bootloader (that's original question I asked naobsd)
<naobsd> pfalcon: I see. something must be happen that I never met... :(
<sjoerd> naobsd: yeah
<deviker> Did anyone managed to use a custom kernel on the mk908 v3 ?
<sjoerd> Galland: let me retry that, now that i have a serial console i can actually see what's going on properly :)
<Galland> deviker: I'm booting my mk908 v3 with my kernel Linux3188 (no wifi though): github.com/Galland/Linux3188
<deviker> Galland: what command line do you use to flash it to the stick?
<Galland> deviker: sudo ./rkflashtool w 0x0A000 0x08000 < recovery.img
<Galland> deviker: sudo ./rkflashtool b
<deviker> Galland: I'm not trying to boot linux, I'm trying to use another android kernel on te V3
<sjoerd> t428-3.0.36-linuxium-test+-kernel.img does seem to boot
<Galland> then it's your compilation's fault ;)
<sjoerd> could well be
<sjoerd> but that's a recovery image not a bare kernel
<sjoerd> lets see what linuxium-1080-mk802iv-nand-kernel.img when flashed to the kernel partiation does i guess
<sjoerd> The bootlaoder doesn't seem very talkative when it falls back
<deviker> Galland: if I do rkflashtool.new w 0x00004000 0x00004000 < ./kernel.img on the MK908 V3, it flashes ok but then after rebooting i see the previous kernel again and it seems nothing has changed.
<sjoerd> I guess there is no u-boot for these boards
<Galland> sjoerd: ...-nand-kernel.img just changes the bootcmdline
<Galland> sjoerd: in order to boot rfs from internal nand flash
<sjoerd> Galland: changes it compared to what ?
<Galland> deviker: if you are targetting "kernel" partition, the you should use "... w 0x4000 0x6000"
<Galland> deviker: slight difference but...
<sjoerd> Anyway, what i'm trying to work out more is whether the bootloader is actually running the kernel from the kernel partition or not
<Galland> sjoerd: changes it compared to getting the rfs from a uSD card
<deviker> Galland, thx I'll try.
<naobsd> if there is kernel+ramdisk(ANDROID!) image in boot(or recovery), it will be preferred
<Galland> sjoerd: maybe naobsd can tell you if you should be flashing an android kernel into "kernel" or "boot" partition
<Galland> that's it
<pfalcon> naobsd, Galland: ok, thanks for your help, guys, will keep figuring it out...
<sjoerd> naobsd: i see
<naobsd> "if there is ANDROID! image with old kernel in boot/recovery ..."
<naobsd> correction
<sjoerd> Is using abootimage to put a zImage into an ANDROID! image the right way of doing that
<naobsd> sjoerd: please specify which "abootimge" you're using. RK's ANDROID! is not same as standard
<sjoerd> naobsd: I guess that's the answer right there then ;)
<sjoerd> I'm using the standard one
<naobsd> there are some "mkbootimg for RK"
<sjoerd> do you have a git tree link handy?
<naobsd> https://github.com/naobsd/cm_system_core/tree/ics-rockchip-naobsd/mkbootimg this is my old one... it should be replaced by one from latest SDK ;)
<naobsd> mine can be built w/o android tree, it uses openssl libs on host OS
<pfalcon> naobsd: so, maybe its should *not* be replaces with one from sdk? ;-)
<naobsd> addresses are different, and tags_addr, page_size, unused[2], name[], and cmdline[] are also used to make SHA1 for hdr.id
<naobsd> sorry, I just wanted to say "there is official one in RK's Android 4.2.2 SDK"
<naobsd> no need to maintan my unofficial one anymore ;)
<naobsd> I'll push it and remove my branch
<naobsd> when I have enough time :(
<naobsd> I need to write some mails etc right now...
<sjoerd> great now it boots the kernel i build
<naobsd> sjoerd: congrat :)
<sjoerd> no to get it to actually start my userspace
<sjoerd> can one only pass kernel commandlines via the kernel config option, or does the loader do the right thing when adding cmdlines to the kernel image
<Galland> there is a .config option to get the cmdline from the bootloader
<Galland> right below the forced cmdline itself in .config
<deviker> which tools do you use to unpack official firmwares? (I've seen weindal tools and they don't work for the one I have fot mk908 V3)
<Galland> if you mean me: I don't do that stuff
<sjoerd> what's the standard serial console on these devices? ttyFIQ0 ?
<Galland> yes
<Galland> and you have to enable anything saying FIQ in the .config
<sjoerd> rockchip-3.0.36-mk802iv from linux-rockchip-3.0 seems a lot better configured in that sense
* sjoerd is a bit lost wrt. all the different ekrenl trees
<Galland> hehe there is an ongoing effort, mainly by omegamoon, to mainline RK kernel with android-3.0 kernel tree
<Galland> as parallel trees
<naobsd> deviker: https://github.com/naobsd/rkutils/blob/master/rkunpack.c I'm using my own tool
<naobsd> it has some minor bug, don't use "update.img" name as a source... it will be overwritten ;)
<naobsd> standard(?) unpackbootimg can be used for ANDROID! boot/recovery image
<naobsd> sjoerd: I'll upload new kernel source which is based on kernel source for Radxa Rock which includes many fixes from RK
<naobsd> I want more time :(
<pfalcon> me too ;'-(
atiti has quit [Read error: Operation timed out]
massi has quit [Quit: Sto andando via]
<sjoerd> so what's the current "suggeted" tree ooi ?
<Galland> currently I'd suggest mine: https://github.com/Galland/Linux3188
<Galland> but that may change in the next weeks
<sjoerd> hehe
<sjoerd> right got to a serial console on my own userspace \o/
<pfalcon> Galland: https://github.com/Galland/Linux3188: "51 commits". Did you guys consider extracting rk patchset and rebasing that on top of official linux tree? so far I didn't see any tree which would be based on upstream and not on some random code blob.
ganbold_ has quit [Remote host closed the connection]
<Galland> hehe that's exactly what I am doing at this very moment
<pfalcon> cool ;-)
<naobsd> https://github.com/linux-rockchip/linux-rockchip-3.0 this is based on android-3.0, but I'll replace it soon
<Galland> Omegamoon successfully merged a RK kernel with android-3.0 official kernel tree (also incorrectly referred to as 3.0.72)
<Galland> I am doing the git backend work to have the full git history
<Galland> and add the patches we've been doing to these RK kernels
<naobsd> actually I already have new one on my PC, I can push it somewhere if someone want
<Galland> naobsd: based on android-3.0 and which RK tree?
<naobsd> Galland: Release:RK3188_R-BOX_ANDROID4.2.2-SDK_V1.0.0_130514 and following patches from Radxa Rock tree
<naobsd> well
<Galland> that sounds very sweet
<naobsd> kernel/ subtree, actually
<Galland> I'd love to browse it
<naobsd> SDK is single git repo ;)
<Galland> fork it
<Galland> and you merged it with android-3.0 kernel tree, right?
<naobsd> yes, as like as linux-rockchip-3.0 on github. as you know RK doesn't use exact android-3.0 tree, they have some changes from outside of android-3.0
<naobsd> so I chosen some point from android-3.0 which is almost same
<Galland> yup, actually they like to screw well established code...
<naobsd> chose
<ramcq> has anyone seen any newer kernels around for rockchip?
<ramcq> don't newer android 4.x need newer kernels than 3.0?
<Galland> which point in time exactly? :)
<Galland> ramcq: latest from rockchip is 3.0.36+
<pfalcon> naobsd: well, that "sdk" things is not menageable at all with 4Gb+ content. someone should diff that against upstream and extract only rockchip-pertinent bits...
<ramcq> well, newer than 3.0, because there's no more upstream support for security fixes in 3.0, so it's not going to be safe in any products from around now
<Galland> newer androids require 3.4 afaik (don't know about kitkat)
* ramcq nods
<ramcq> that will have long term support for a couple of years yet
<ramcq> but nobody saw a rockchip with a 3.4 kernel in the wild
<naobsd> pfalcon: yes. that is why I made kernel subtree
<naobsd> well
<naobsd> let's delete linux-rockchip-3.0 now... it's just an archive of source tarballs for some specific products
<naobsd> no unique change is committed
<naobsd> I'll do same thing again with Radxa's
<Galland> naobsd: have you tested it? rk31/rk30?
<naobsd> I tested a little on RR and MK802IV-AP6210
<naobsd> I don't have enough time to do everything at once...
<naobsd> branch for RR is pushed
<naobsd> I'll push branch for MK802IV-AP6210 soon...
<Galland> does it change that much?
<naobsd> wip/rockchip-3.0-radxa-rock is for RR, only 1 change from rockchip-3.0
<naobsd> I didn't make any change yet. it's 100% same as Radxa's tree
<naobsd> I'll delete some branches comes from android/kernel_common...
<Galland> but why a branch for each device?
<Galland> :)
<naobsd> I thought I prepare more things before pushing
<naobsd> it doesn't have ifdef's properly
<naobsd> ifdef or own -board.c
<Galland> understood
<naobsd> anyone has RR?
<Galland> omegamoon's took a more recent android-3.0 (just some 4 commits from head) and made it work
<Galland> do you plan on merging latest android-3.0?
mrueg has quit [Remote host closed the connection]
<naobsd> latest android-3.0 is merged in wip/rockchip-3.0.72
<naobsd> but I can't remember I tested it or not ;)
<naobsd> ah, I remembered, it worked, I need to revert ion to see youtube ;)
mrueg has joined #linux-rockchip
mrueg has joined #linux-rockchip
<Galland> :)
<naobsd> latest android image may be better to used with latest kernel, there are some changes both kernel and userland
<naobsd> I built id w/o any change, and I noticed there are some google apps. I think I can't redistribute it ;)
<naobsd> built it from Radxa's SDK
<Galland> my RR is still on the mail
<naobsd> I'll make one another Android image w/o G apps for redistribute
<naobsd> I wanted to renew my build machine before starting work...
unexpekt0r has joined #linux-rockchip
eebrah|noPC has joined #linux-rockchip
pfalcon has quit [Ping timeout: 240 seconds]
eebrah|noPC has quit [Remote host closed the connection]
pfalcon has joined #linux-rockchip
<unexpekt0r> hi all
<unexpekt0r> i get this error when connecting my usb serial device to mk808b
<unexpekt0r> seems there is a problem with cdc-acm ?
<unexpekt0r> but im kinda cluess on how to compile the driver, and not use the standard driver that is shipped with picuntu
pfalcon has quit [Ping timeout: 245 seconds]
<Galland> is that cdc-acm a module?
<Galland> it seems a module compiled for a different kernel than yours
<Galland> one with tracing disabled
deviker has quit [Ping timeout: 246 seconds]
tonikasch has joined #linux-rockchip
pfalcon has joined #linux-rockchip
<pfalcon> naobsd, Galland, Astralix, sarg: I put up a page with bootloading questions I find interesting - hopefully not just for me: http://linux-rockchip.info/mw/index.php?title=User:Pfalcon#Bootloading_Questions . If you have any answers/info/hints, please share them. If answers will get filled in, I'll move that page as normal wiki page later. Thanks!
<tonikasch> pfalcon dissasembly is a step forward illegality :(
<pfalcon> tonikasch: ehm, who said that? GPL violators?
<pfalcon> if they come by, they can kindly provide detailed spec instead ;-)
<tonikasch> :p
<tonikasch> I know :)
<tonikasch> but some people might not feel comfortable with disassembling bootloader as it is illegal, even if in this case is a completely ethical thing to be done
<sjoerd> tonikasch: Depends on where you live
<tonikasch> I don't know if it's illegal here in Spain, I've heard it is but I'm not really aware
<sjoerd> In europe it's legal if done for interop
<pfalcon> all in all, that's why I posted those q's on personal wiki page - they're personal request to another persons for education. may need polishing for general wiki content
<tonikasch> interop?
unexpekt0r has quit [Remote host closed the connection]
_whitelogger has joined #linux-rockchip
<Galland> chinese pdf technology ;)
<pfalcon> Galland: if only pdf ;-)
<pfalcon> Ok, some bootloading info collected at http://linux-rockchip.info/mw/index.php?title=User:Pfalcon#Some_answers.2Fpointers
<pfalcon> And Is started great rockchip kernel dump at http://linux-rockchip.info/mw/index.php?title=User:Pfalcon#Rockchip-related_kernel_trees . How many kernel trees are around? dozen? fifty? make your bets!
tonikasch has quit [Ping timeout: 248 seconds]
tonikasch has joined #linux-rockchip
pfalcon has quit [Quit: Leaving]
tonikasch has quit [Quit: Bye!]
akaizen has joined #linux-rockchip
Galland has quit [Quit: Saliendo]
akaizen has quit [Ping timeout: 245 seconds]
akaizen has joined #linux-rockchip
akaizen has quit [Remote host closed the connection]
tonikasch has joined #linux-rockchip
akaizen has joined #linux-rockchip
tonikasch has quit [Quit: Bye!]
ganbold has joined #linux-rockchip