mnemoc changed the topic of #arm-netbook to: EOMA: Embedded Open Modular Architecture - Don't ask to ask. Just ask! - http://elinux.org/Embedded_Open_Modular_Architecture/EOMA-68 - ML arm-netbook@lists.phcomp.co.uk - Logs http://ibot.rikers.org/%23arm-netbook or http://irclog.whitequark.org/arm-netbook/ - http://rhombus-tech.net/
<WarheadsSE> hno :) happy uEnv.txt usage
<WarheadsSE> I kicked my saved env, and off it swooshed (once I remembered to add a rootdelay/rootwait in extraargs
<mnemoc> RaYmAn: what did you find about card_boot0_para ? is it setup by boot1?
<RaYmAn> mnemoc: nothing really
<WarheadsSE> mnemoc: suppose I should maybe add the build sd-card instructions for linux? in the vent you are _not_ using an image
<Turl> hno: you did have prebuilt uboot images right?
<mnemoc> WarheadsSE: the official documentation is in https://github.com/hno/uboot-allwinner/wiki but http://linux-sunxi.org/u-boot is begging for some attention
<hno> Turl, I have some at http://www.hno.se/code/A10/ but not up to date.
<Turl> a long as it works I'm okay
* Turl is feeling lazy today
<mnemoc> you can use the "nightly"
<mnemoc> but hno's old image is known to work
<mnemoc> no idea about cnxsoft's
<hno> but mine is a messy snapshot image with mismatched SPL and main u-boot and where sources have been lost..
<mnemoc> :)
<WarheadsSE> lol
<hno> it does work. But have absolutely no idea how to rebuild that exact version.
<hno> should probably build another one and delete that one.
<hno> and a matching tag in git.
<WarheadsSE> fixed the typo in the github uboot wiki hno
<hno> Hmm.. how do I push a tag?
<hno> WarheadsSE, what typo?
<hno> and thanks!
<WarheadsSE> FAT /boot.scr should have been FAT /uEnv.txt in the uEnv section
dyoung has quit [Quit: ZNC - http://znc.sourceforge.net]
dyoung-away has joined #arm-netbook
dyoung-away is now known as dyoung
dyoung has quit [Changing host]
dyoung has joined #arm-netbook
<Turl> hno: you don't dd these to sector 0 do you? I think I just foobared my sdcard :P
<hno> That u-boot image seems to boot fine on my mele.
<mnemoc> Turl: just blank the first 1M
<mnemoc> Turl: then the spl to 8*1024 and uboot to 32*1024
<Turl> mnemoc: hno's image is a single img though
<hno> Turl, you only need to restore your partition table if you had data on it.
<mnemoc> ah, ok
<Turl> hno: well I had a partition :P
<hno> Turl, just recreate the partition table and you'll be fine.
<hno> the partition contents is still there.
<Turl> hno: part table is first 512B right?
<hno> yes.
<hno> just partition the card as usual.
<hno> and next time skip=1 seek=1 if writing the image to an used card.
<Turl> good tip, didn't know about seek&skip
<hno> "Unknown command 'echo' - try 'help'"....
<WarheadsSE> yeah
<hno> Kinf of expected that to be built-in to hush.
<hno> Fixed.
<hno> new image uploaded, and tag pushed to github.
<hno> RaYmAn, can you please detail how to get JTAG running on A1x in the wiki?
<WarheadsSE> hno: has anyone tired this on the tablets? with success
<hno> WarheadsSE, tried what?
<WarheadsSE> this uboot/spl on the tablets
<hno> mnemoc 6 RaYmAn yes. thanks
<WarheadsSE> Did I just get a POS, or i it not just me?
<hno> POS?
<WarheadsSE> piece of shit
<hno> u-boot-mmc have been used on tablets.
<WarheadsSE> well then the 2 "gooseberry" A721 tablets we have .. don't work
<hno> what dram configuration do they have?
<Turl> so my magic sdcard is not working :< my backlight turns on then device reboots
<WarheadsSE> 4 hynix chips
<hno> then memory setup needs adjustment
<Turl> [ 2.000000] Kernel panic - not syncing: No init found. Try passing init= option to kernel. See Linux Documentation/init.txt for guidance.
<Turl> :<
<WarheadsSE> hno: in.. uboot?
<mnemoc> Turl: passed root= ?
<Turl> yeah
<Turl> and root is mounted
<hno> WarheadsSE, yes,
<Turl> my init is on /init
<WarheadsSE> *wheee
<WarheadsSE> they are 2x 2x128
<mnemoc> Turl: told the kernel to use /init instead of /sbin/init ?
<Turl> no, but my android has init on /init and it boots fine :<
<WarheadsSE> probabl has to do with the phys dram 1 defines
<WarheadsSE> Turl: odd.. what distro is that?
<Turl> my own
<WarheadsSE> ok, whats it based on then
<Turl> if you can call an unpacked android recovery a distro :P
<hno> Turl, krenel do not look for /init by default.
<WarheadsSE> ahm, android ..
<hno> run_init_process("/sbin/init");
<hno> run_init_process("/etc/init");
<hno> run_init_process("/bin/init");
<hno> run_init_process("/bin/sh");
<hno> anything else need an init=...
<Turl> yay works now :)
<Turl> got CWM on a sdcard :P
<hno> WarheadsSE, see u-boot/arch/arm/cpu/armv7/sunxi/dram.c
<mnemoc> Turl: wiki wiki :D
<Turl> mnemoc: I'll do a page when I fix a couple of stuff it has :P like it's unable to flash anything
<hno> many have asked for that
<Turl> mnemoc: did you upgrade -v2 with mainline?
<mnemoc> yes
<WarheadsSE> hno: yup
<Turl> what's the new nr?
<mnemoc> 3.0.39+
<Turl> thx
<Turl> mnemoc: what kind of black magic do you need to build nand-part?
<Turl> nand-part.h:26:19: fatal error: types.h: No such file or directory
<mnemoc> o.o
<mnemoc> -Iinclude/ not been passed?
<mnemoc> Turl: are you using `make` ?
<Turl> got it compiling with make nand-part CFLAGS="-static -Iinclude/" GCC=arm-none-linux-gnueabi-gcc
<Turl> with static alone it explodes
<mnemoc> Turl: why don't you just `make` ?
<Turl> because it builds all the things?
<mnemoc> `make nand-part` ?
tuliom has quit [Quit: Konversation terminated!]
<Turl> I need a static, arm binary
<Turl> and it's CC, not GCC :p
<Turl> make nand-part CFLAGS="-static -Iinclude/" CC=arm-none-linux-gnueabi-gcc
<mnemoc> so Makefile needs ifeq ($(STATIC),yes)
<mnemoc> that repo is a mess :<
<Turl> 'env 4096' 'boot 65536' 'system 524288' 'data 2457600' 'misc 2048' 'recovery 65536' 'cache 262144' 'UDISK 27713536'
<mnemoc> b c d e ....
<Turl> what unit is it?
<mnemoc> smells like 512
<Turl> system is ~252M
<mnemoc> sure?
<Turl> so it's 512B sectors I guess
<Turl> yeah I'm checking with df
<mnemoc> df tells you fs size, not partition size
<Turl> yeah but 512B sector count makes sense :)
<mnemoc> isn't that what I suggested? :<
hipboi has joined #arm-netbook
<mnemoc> 3:30 .... should really go to sleep
<hipboi> aha
<mnemoc> moin hipboi
<mnemoc> and good night :)
<hipboi> i am just got to the office and start to work
<hipboi> good nite
<Turl> mnemoc: ah I thought you suggested 512M :P
<Turl> my bad :)
<mnemoc> Turl: no, simply 512
<Turl> yeah 512 it must be :)
<mnemoc> 512M is kind of large even for Google's filesystem
<mnemoc> iirc they use 4M
<mnemoc> I give up in trying to get serial output for the a13-olinuxino tonight
<mnemoc> good night
<Turl> night mnemoc
<hipboi> what?
<hipboi> a13-olinuxino has no serial?
<Turl> hipboi: I think his SD slot was playing games
<mnemoc> UART0 shared with the uSD and UART1 in a 10pin header
<hipboi> i see...
<mnemoc> but nothing works in my board :<
<hipboi> hehe
<mnemoc> no SD, no usb, no uart, no power button, no vga, ....
<mnemoc> snif snif snif
<mnemoc> good night
<hipboi> good night
<hno> A13 is a little short on uart pins compared to a10
<WarheadsSE> to say the leat
<hno> one muxed with sd, one muxed with lcd iirc.
<mnemoc> VGA_DIS took UART3_RX
<mnemoc> but UART1 is supposed to be there, just can't make it work :<
<hno> VGA_DIS?
Boulet has joined #arm-netbook
<mnemoc> hno: a pin used to disable the LCD/VGA voodoo
<hno> is it uart3 that is on lcd pins?
<Boulet> hi
<hno> ok.
<mnemoc> uart3 is peacefully standing in PG9-PG12
<mnemoc> but got RX abused
<mnemoc> normally only shares pins with SPI1
<mnemoc> LCD used PD*... and no uart is documented there
<mnemoc> uses*
<Boulet> on A13 i see UART0_TX on PF2 and UART0_RX on PF4, which is not on the datasheet
<mnemoc> right
<mnemoc> the datasheet of the A13 is "sightly" incomplete :)
<Boulet> oh there are other things missing ? wouldn't be surprising hehe
<mnemoc> ^^
<mnemoc> help documenting pins at http://linux-sunxi.org/A13-OLinuXino and http://linux-sunxi.org/A13/PIO is highly welcomed
<Boulet> i am still trying to initialize A13's DDR3 :(
<mnemoc> Laurent Ovaert?
<Boulet> yup
<mnemoc> welcome :)
<Boulet> :)
<mnemoc> Boulet: hurry and catch hipboi, before he is abducted by his boss
<Boulet> yeah isn't he working for allwinner ?
<mnemoc> Boulet: yes
<Boulet> i have thought to visit them in Zhuhai since I am in Shenzhen
<hipboi> mnemoc: you need to go to bed now
<mnemoc> hipboi: indeed
<mnemoc> good night, for real
<mnemoc> f* A13
<Boulet> A13 rocks !
kaspter has quit [Quit: Instantbird 1.1]
<WarheadsSE> hipboi: can you do me a favor and fix this naming on the A100 page on your aliexpress store? "Arch Linux ARM" not "ARCH arm linux"
<hno> mnemoc, that pin is just one resistor "R19" away from being used from what i can see in schematics. Would have expected that to be a jumper, not gpio controlled by default
<hipboi> WarheadsSE: ok
<hipboi> WarheadsSE: sorry
<Turl> yay A10 repartitioning works :)
<hipboi> good
<WarheadsSE> ty hipboi
<hipboi> hno: i just confirmed, a13 dram controller is different from a10
<Boulet> glad to hear that now ;)
<hno> how different?
<Boulet> what about the BROMs ?
<hno> BROM do not know about DRAM.
<hno> but A13 BROM do differ from A10. Have not looked at details yet
<hipboi> i will look at the difference
<hipboi> but now i am assigned to study the linux schedule
<hipboi> far from these
<hipboi> sigh
<Boulet> oh
<hno> linux scheduler?
<hipboi> yes
<hipboi> for the multi core
<Turl> linux scheduler is good on SMP :) doesn't need allwinerization haha
<hipboi> i think so
<hipboi> maybe we can take advantage of cgroup etc..
<Turl> android uses cgroups
<hno> cgroups and process groups both do wonders with scheduler.
<Turl> is the nand layout the same on all devices?
<hno> hipboi, A13 ddr controller looks marginally different from the sources i have.
<hno> s/from/based on/
<ibot> hno meant: hipboi, A13 ddr controller looks marginally different based on the sources i have.
<hipboi> is marginally different very big different or small
<hipboi> sorry for my english
<hno> small
<hipboi> ok
<hno> hostport settings, ccm interfacing, zq settings. Maybe some other small detail.
<hno> and a couple bugs less to account for
<Turl> hipboi: maybe interesting for you http://linux-sunxi.org/CWM_on_a_SDCard
<hipboi> nice
<hipboi> with one picture would be better
<Turl> a picture of what?
<hipboi> of the cwm screen
<ZaEarl> Turl, have you seen kk's recovery? the onscreen buttons work great. https://github.com/killerkink/android_bootable_recovery
<Turl> nope, I haven't
<Turl> although I'm not a fan of recovery with on screen buttons, it's like a halfway there solution
<Turl> ZaEarl: btw, I repartitioned my zatab to fit JB better
rsalveti has quit [Remote host closed the connection]
lerc has joined #arm-netbook
MMlosh has quit [Ping timeout: 248 seconds]
MMlosh has joined #arm-netbook
<ZaEarl> Turl, so put together a LiveSuit img that repartitions and installs JB all in one go.
QingPei has joined #arm-netbook
ZaEarl has quit [Quit: Ex-Chat]
ZaEarl has joined #arm-netbook
rsalveti has joined #arm-netbook
dyoung is now known as dyoung-away
ZaEarl has quit [Quit: Ex-Chat]
ZaEarl has joined #arm-netbook
Quarx has joined #arm-netbook
Gujs_ has joined #arm-netbook
tzafrir_laptop has quit [Ping timeout: 240 seconds]
Boulet has quit [Read error: Connection reset by peer]
Boulet has joined #arm-netbook
RITRedbeard has quit [Read error: Connection reset by peer]
RITRedbeard has joined #arm-netbook
<Mazon> am I the only one that is unimpressed by the ouya? - quad core tegra, 1 GB ram, 8GB flash etc etc - for 99$, pretty nice! - HOWEVER, with a release date in 2013, early 2014 - thats actually quite shitty specs?
<Mazon> ok, probably not shitty - but at least not so impressive
<Mazon> if the ouya was ready soon - then it'd be something entirely different
<Mazon> btw, anyone know what the expected price area for the cubieboard will be?
<Mazon> too early in the morning for a rant I guess :)
<hno> 1GB RAM is a little on the low end for such a SoC. And it's Tegra
<Mazon> my beef is, that these specs are great NOW, but not so great in early 2014
<CIA-121> rhombus-tech: sc15 master * reea65d9a768c /allwinner_a10/orders/sc15.mdwn:
<Mazon> at that time, we should be running armv8, quad/octo with yet a new graphics core and 2GB ram and 32GB flash :)
<Mazon> the odroid-x is 129, for a somewhat similar board (I know, bigger, no wifi etc etc) - but in a years time...
<rm> it's not going to be released anyway
<CIA-121> rhombus-tech: sc15 master * r21f5b0e6c2da /allwinner_a10/orders/sc152.mdwn:
<Mazon> I guess to have it at the 99$ pricepoint with controller and package and a margin, they need to have a bit "old" hardware
<rm> there's an opinion that I tend to agree with, that their release timeframe prediction is completely unrealistic
<rm> and afaik they don't even have a working demo yet
<rm> or show it to anyone
<Boulet> they have $6M or so, now they can hire some good people :)
<Boulet> the hardware is probably the easy part
destinal has quit [Ping timeout: 246 seconds]
<orly_owl> is there any arm hardware that supports usb3?
destinal has joined #arm-netbook
<hipboi> there are the price
<rm> Kickstarter has no accountability system
<rm> they can say "ok thanks bye"
<orly_owl> are they an american company?
<rm> get $6M USD and enjoy the rest of their life
<orly_owl> there are similar websites
Gujs_ has quit [Quit: Ex-Chat]
<rz2k> why do we have spam links on frontpage? :/ http://linux-sunxi.org/Main_Page
tzafrir_laptop has joined #arm-netbook
<Mazon> hipboi, nice - will definitely get one when its ready!
<hipboi> rz2k: what spam links?
<hipboi> Mazon: D
<Mazon> yeah, that coupon stuff ..
<Mazon> so just revert that and kick user
<Mazon> probably just automated account creation and spam
Boulet has quit [Remote host closed the connection]
Boulet has joined #arm-netbook
QingPei has left #arm-netbook [#arm-netbook]
<hno> mnemoc, you got mail.
adinn1 has joined #arm-netbook
adinn1 has left #arm-netbook [#arm-netbook]
arokux has quit [Remote host closed the connection]
arokux has joined #arm-netbook
<RaYmAn> hipboi: hi. got jtag somewhat working. You wouldn't happen to have an openocd config file for it or simular? :)
<hipboi> no...
<RaYmAn> damn. so what do you guys use for jtag then?
<mnemoc> moin
<mnemoc> hno: can I write/exe the spl using fel?
<mnemoc> hno: any address (e.g. 0x2000) or there is a preferred location?
<hno> RaYmAn, I would expect they use ARM Keil tools.
<hno> Or.. that's the microcontroller tools. Something similar but for Cortex A
Vayu has joined #arm-netbook
<RaYmAn> i know nvidia uses lauterbach
RITRedbeard has quit [Read error: Connection reset by peer]
RITRedbeard has joined #arm-netbook
<RaYmAn> its just hard to guess a lot of jtag stuff :O
<Turl> literally 1 cent card reader if anyone needs one :P
ssspiff has quit [Remote host closed the connection]
<Turl> hipboi: at 49$ and that small size, cubieboard looks like a good competitor for raspberry pi :)
<Boulet> via fel it should now be able to upload any .elf file to A10/A13 memory and have it executed automatically
<mnemoc> Boulet: hi, hno forwarded me your (?) test1.zip. is it standalone or it's supposed to be dropped somewhere in the u-boot tree?
<mnemoc> smells stand alone
<Boulet> it is stand alone
<Boulet> look at the header of test.c
<Boulet> i am sorry the code is damn ugly, i just wanted to have something run quickly ;)
<mnemoc> doesn't compile here :p
<Boulet> i grabbed bit and pieces here and there haha
<Boulet> oh hum
<Boulet> i use cygwin
<mnemoc> i'm a userspace guy... and in that world that error means "no main"
<mnemoc> I suppose I need to include start.o
<mnemoc> but then it fails uglier
<Boulet> normally no
<mnemoc> then it has to be about gcc flags
<Boulet> (see privatemsg)
<furan> where is the kernel on my allwinner device?
<furan> I mounted nanda and all I can find is u-boot
<Mazon> anyone seen: http://www.ovalelephant.com/forum/viewtopic.php?f=12&t=114 - aparently oval elephant got cheated by Shenzhen Sunrise Ltd
<Turl> ouch
<Turl> the post doesn't say much though
<Turl> furan: nandc
<rz2k> lol
nibb__ has quit [Ping timeout: 246 seconds]
<furan> thanks
<rz2k> to have a good production line in China you actally need to inspect everything yourself, every line, every stage, not by some hireable guy that will take money from both sides and say that everything is okay.
nibb has joined #arm-netbook
<rz2k> I have a friend that runs OEM for testing equipment, cheap multimeters and such stuff, he was cheated by guys from globalsources/alibaba quite often till he realized that he need to control everything.
<Turl> I could use a cheap multimeter :P
<mnemoc> Mazon: you can get it from Tom
<rz2k> meh, buy a Fluke. one multimeter forever, I had one that was ~7 years old, still works at my old job.
<Turl> rz2k: 100-500USD >.< no thanks :)
<Mazon> mnemoc, ?
<Mazon> mnemoc: the above is about the Round Rhino, custom product afaik
<Turl> from what I know, round rhino is similar to the oval elephant
<Mazon> + IR + UART + BCM Wifi/Bluetooth Module
<Turl> yeah that
<Turl> I was not sure if it was public info :)
<Mazon> its in the forum - so I would assume that it is
<rz2k> bcm...
<Turl> rz2k: BCM wifi kills realtek's crap chips any day :)
<RaYmAn> So, anyone tried out the JTAG stuff yet?
sspiff has joined #arm-netbook
sspiff has quit [Changing host]
sspiff has joined #arm-netbook
<mnemoc> RaYmAn: you :)
<mnemoc> Mazon: didn't know the difference between both
<Mazon> the ovalelephant is similar to toms
<Mazon> but the round rhino was something new
<RaYmAn> mnemoc: yeah, was more thinking whether someone verified it yet? :)
<mnemoc> RaYmAn: sure, I was joking ;-)
<RaYmAn> I know :P
<RaYmAn> I'm wondering the best way to do proper debugging..Maybe replacing either u-boot or kernel with a binary that has a single entrypoint and just loops forever? So one can jtag in, halt, load e.g. u-boot and debug through it
<RaYmAn> since we have no way to start cpu in halt
<mnemoc> `resume` doesn't do it?
<RaYmAn> well, yeah, but like, when you apply power, it'll just run regularly until you jtag in and halt it
<RaYmAn> so it's pretty hard to control how far it's gotten in initialization
<RaYmAn> it mihgt be possible to soft-reset I guess
<mnemoc> sounds like an important feature
<RaYmAn> usually you have a hardware signal to reset, but given how we access it, that's not realy possible
<Turl> RaYmAn: can't you say, stop on the uboot console
<Turl> then chainload another uboot to debug it?
<RaYmAn> probably, but doing it that way means uboot initialization code has already run
<RaYmAn> which may or may not be an issue depending on what you want to debug
<RaYmAn> Perform as hard a reset as possible, using SRST if possible.
<RaYmAn> soft_reset_halt might work I suppose
<RaYmAn> I'm mostly guessing, so I dunno what's possible with software reset
<mnemoc> find it out
<mnemoc> my 22" really dislikes olinuxino's 800x600 :p
<RaYmAn> mnemoc: you had a flyswatter and a microsd breakout board, right? ;)
<mnemoc> RaYmAn: yes, but more concerned about getting my olinuxino hardware working (on the stock android) atm
* RaYmAn tries to come up with a way where JTAG would help with that
<mnemoc> weee... uart1 started working :D
<RaYmAn> cool :)
<Boulet> ah good :)
Boulet has quit [Remote host closed the connection]
<RaYmAn> mnemoc: is it possible to access nand from mmcbooted kernel yet?
<Turl> mnemoc: 512 ram? :P
<Turl> mnemoc: btw, repartitioning went smoothly yesterday :)
Boulet has joined #arm-netbook
<Boulet> mnemoc this is the output of uart0 right ?
<Boulet> does A13 has JTAG pins available ?
<RaYmAn> we don't really know
<RaYmAn> I'm kind of thinking that it's identical to a10 in that matter
<CIA-121> rhombus-tech: marco master * r648ab81b033f /allwinner_a10/ainol_novo_elf_7.mdwn:
<RaYmAn> e.g. we can get jtag over microsd port by default (e.g. PF00,01,03 and 05 is set to JTAG by default))
<Boulet> RaYmAn, not sure, there are less pins, they might have left that out
<Boulet> ah ok
<RaYmAn> I'll try it later - I'd just kind of like a full backup of my a13 device before I start playing with nand stuff (I don't really have a suitable livesuite image)
nibb has quit [Read error: Connection reset by peer]
nibb has joined #arm-netbook
<Boulet> yeah would be nice if you could tell us if jtag works there
<Boulet> UART0 was also not documented on A13 datasheet but it is there at same place as A10
<RaYmAn> yeah, on PF02 and 04?
<RaYmAn> so it makes sense they'd make jtag the same way as well
<Boulet> yes
<RaYmAn> do we have a fel-pio dump on an 13?
<Boulet> yeah
<Boulet> it's on github
<RaYmAn> any hint on where? :P
<Boulet> would be nice to make a jtag to SD adapter
<Boulet> and also have UART0 on that, would be quite convenient
<RaYmAn> yeah, should be possible
<RaYmAn> well, that's what the microsd adapter from hipboi has, heh.
<CIA-121> rhombus-tech: marco master * rd6024050a27f /allwinner_a10/ainol_novo_elf_7.mdwn:
<Boulet> but what about this JTAG_SEL0/JTAG_SEL1 ?
<RaYmAn> No idea
<Boulet> RaYmAn, oh so this exists ?
<Boulet> such adapter ?
<CIA-121> rhombus-tech: marco master * r6a9eeb5cb755 /allwinner_a10/ainol_novo_elf_7.mdwn:
<Boulet> GOSH
<Boulet> hahaha
<Boulet> i want that :-D
<RaYmAn> I don't think it's generally for sale, but hipboi might be able to hook you up with one
<Boulet> i hope i can even go see, him i am in shenzhen ;)
<RaYmAn> heh
<Boulet> this is smart that they placed uart0 and jtag on the sd-card slot !
<RaYmAn> yeah
<RaYmAn> I took that idea and I've successfully used it on ASUS Transformer & Transformer Prime as well (and it should work on all tegra devices) - only UART though.
<Boulet> you mean they also export it on the SD card ? neat
<RaYmAn> well, they don't by default
<RaYmAn> but it so happens to be one the functions those pins can take on through pinmuxing
<Boulet> ah right
<RaYmAn> so it won't give you bootloader output unless you are running a custom u-boot
<RaYmAn> but it'll let you get kernel output
<RaYmAn> (with a custom kernel as well, of course, but that's easier to get :P)
<Boulet> you use openocd for jtag ?
<RaYmAn> mnemoc: got a fel-pio dump of the a13 board? I can't seem to find it anywhere
<RaYmAn> yeah
<Boulet> this works well with cortex-a8 ? never tried
<RaYmAn> it works, but given I've had to try and patch together a config file, it doesn't work perfectly :P
<RaYmAn> openocd has pretty good a8 support as far as I can tell
<RaYmAn> ty
<RaYmAn> well, PF0,PF1, PF3 and PF5 has function 4 set by default like on a10 (and on a10, that's jtag)
<RaYmAn> So I guess that does indicate it's the same :)
<CIA-121> rhombus-tech: marco master * raeb57bd5bc4c /allwinner_a10/ainol_novo_elf_7.mdwn:
<RaYmAn> anywyas, gotta go shop a bit, later
<Boulet> mmmh
<Boulet> see ya
<CIA-121> rhombus-tech: marco master * r5f3a68ac48c3 /allwinner_a10/ainol_novo_elf_7.mdwn:
<mnemoc> RaYmAn: in pins the olinuxino only has PF2/PF4 exposed, and not sure what's exactly wrong with my uSD slot
<RaYmAn> mnemoc: did you verify those pins vs uSD slot with a multimeter?
<RaYmAn> it's a bit harder to verify the rest
<mnemoc> RaYmAn: no clue how to open this thing
<mnemoc> the slot I mean
<RaYmAn> I meant like, using the microsd breakout? :P
<RaYmAn> I mean, you know PF2/4 should go to the uart header (the 5-pin one)
<mnemoc> meh... good idea
<mnemoc> I'll test that in 1-2h
<RaYmAn> you could also test for whether the other pins have signals kinda (while the device is on)
<mnemoc> yes, will search for the breakout and test that in 1-2h
<RaYmAn> psh, you take that work thing far too seriously ;)
<mnemoc> :)
<mnemoc> I need to finish some stuff before friday or I won't be allowed to leave :-/
<mnemoc> it's not that much, but I can't take the risk
<Boulet> mnemoc which country are you in ?
<Boulet> spain ?
<mnemoc> yes, but I don't work "here"
pwhalen has quit [Quit: Leaving]
<Boulet> ah
<mnemoc> and as "contractor" I don't really have the right to have vacations at all... so I prefer to not push the luck
pwhalen has joined #arm-netbook
<Triffid_Hunter> hey all, what's the latest on mele A1000 hacking?
<xxiao_> this looks like Mele2000
cubieboard has joined #arm-netbook
<Boulet> cubieboard ;)
<cubieboard> hi
<Boulet> hi A10
<zenitraM> can we expect a sun6i cubieboard sometime? :P
<RaYmAn> I'd imagine sun6i has to come out first? ;)
<Boulet> i bet an FPGA version of it already exists hehe
<RaYmAn> probably
<rz2k> interesting, did allwinner actually fix heating issues in sun6i? a10 is pretty hot, guys with mk802 placing heatsinks for heavy loads. I believe sun6i to be MPCore, so different asic layout will be used, but still..
<rz2k> also, imx6 and tegra3 are in metal housings instead of usuall plastic ones.
<specing> rz2k: it isn't as hot when running normal linux
<specing> I guess mali isn't enabled
<rz2k> cant say that my a10 on lyf1 was cold running linaro, especially when I've tested mplayer for an hour :p
<WarheadsSE> xxiao_: I dont think that is an A10. I beleive it is supposed to be tegra3
<rm> mine was uber hot when in Android
<rm> mk802, that is
<xxiao_> i was told the new mk802 will have axp209 integrated, due end of this month
<rm> I now just run it at 300-1000MHz with the ondemand governor
<rm> well, and no graphics
<rm> 20:50:27 up 13 days, 22:59, 1 user, load average: 1.07, 1.04, 1.05
<rm> cpufreq stats: 300 MHz:82.05%, ............ 1.01 GHz:17.90%,
<traeak> ouya was last said to be tegra3 although i'm not so sure that's the best choice
<traeak> too bad the t6xxx mali isn't going to be out before then
<traeak> probably will kick the living crap out of tegra3's gpu
<WarheadsSE> fair chance
<RaYmAn> with some luck there'll even be cortex a15 out by then :P
<WarheadsSE> not like the drivers out now are open enough though
<WarheadsSE> getting accel working on the Mali is a bitch.
<WarheadsSE> (in linux, with not-out of date X)
<Triffid_Hunter> WarheadsSE: possible?
<WarheadsSE> possible, yeah
<WarheadsSE> just a PITA
<WarheadsSE> Cedar would be nice too.
dfenwick has joined #arm-netbook
<RaYmAn> mnemoc: did you play around with any custom kernels? :)
cubieboard has quit [Quit: Leaving]
<mnemoc> RaYmAn: trying to get wip/lichee3-sunxi/import-sun5i compiling
<RaYmAn> okay
<mnemoc> updated drivers need int sw_get_chip_id(struct sw_chip_id *); implemented
<RaYmAn> annoying
<mnemoc> I'll just fake it for now
<RaYmAn> are the drivers significantly different? Or well, is the hardware really any different? (Other than lacking some stuff)=
<mnemoc> mostly all the same stuff
<RaYmAn> 'k
<mnemoc> the goal should really be to merge sun4i and sun5i
<RaYmAn> yeah
<mnemoc> I've done some merging in 3.4
<RaYmAn> We really need a way to dump full nand from fel :> hehe
Vayu has quit [Ping timeout: 265 seconds]
<mnemoc> probably the best way is to push an small installer/recovery system to ram
<RaYmAn> yeah
<specing> Can't you use uboot for that?
<RaYmAn> hm
<specing> first load and then hexdump to console?
<RaYmAn> over uart? :S
<RaYmAn> won't that take...forever?
<specing> no
<specing> 10kb/s
<specing> I've booted Linux over serial before
<RaYmAn> I suppose you only need to dump the first 10-20MB/s
<specing> yep
<specing> and uboot supports loady
<specing> maybe it supports sendy too?
<RaYmAn> can uboot access the full nand?
<mnemoc> in theory...
<RaYmAn> though,my issue is that I want a full dump before changing e.g. script.bin since I have no livesuite image :S
<Boulet> i think the best is just to do a small program that will dump the nand to DDR3, then use fel to retrieve
<RaYmAn> yeah, that's what I was thinking, but the question is how much is required to writ ethat :P
<RaYmAn> presumably some pinmux setup, some nand setup, then talking to nand?
<mnemoc> writing the full 4GB when you only need a couple of MB instead very nice
<Boulet> RaYmAn basically yes
<Boulet> fel is a good interface actually
<hno> mnemoc, you should be able to hut uSD connector pins on the PCB using a needle
<Boulet> you could pass some parameters to some functions via fel and ask fel to execute the function you want
<mnemoc> hno: pushing up the front or the back?
<mnemoc> don't want to break it
<hno> usually the pcb connections are exposed at the back. But i have not seen what connector they are using.
<mnemoc> ah, sorry... I thought you said I can open the slot
<mnemoc> damn migrane :>
<mnemoc> :<
<mnemoc> can't think
<hno> the slot shield is soldered for stability, no opening without desoldering.
<RaYmAn> hno: does bootrom do different things wrt initialization when booting e.g. sd and fel? (Other than presumably setting up usb)
<mnemoc> i would assume it's fel who sets up usb
<RaYmAn> well, fel is in bootrom, sin't it?: P
Quarx|2 has joined #arm-netbook
sspiff has quit [*.net *.split]
Quarx has quit [*.net *.split]
lundman has quit [*.net *.split]
specing has quit [*.net *.split]
termleech has quit [*.net *.split]
arete74 has quit [*.net *.split]
robws has quit [*.net *.split]
<RaYmAn> I'm mostly thinking about the idea about loading u-boot spl from fel
<mnemoc> RaYmAn: not a new idea :p
* RaYmAn doesn't know enough about a10/13 initialization sequence
<RaYmAn> mnemoc: I know , I read it here! :P
sspiff has joined #arm-netbook
Quarx has joined #arm-netbook
lundman has joined #arm-netbook
specing has joined #arm-netbook
termleech has joined #arm-netbook
arete74 has joined #arm-netbook
robws has joined #arm-netbook
<mnemoc> from Boulet we have a toy that initializes the A13 dram
<mnemoc> so we are very near to the sun5i-mmc uboot
Quarx has quit [Ping timeout: 255 seconds]
tzafrir_laptop has quit [Ping timeout: 248 seconds]
<Boulet> the watchdog code i found on git does not work well
<Boulet> it does reset the chip once, but not more ;)
<Boulet> i modified the sequence and now it works all the time
<Boulet> it's useful to nicely come back to fel mode after your program as finished executed (or on demand)
<RaYmAn> In reality I'm more thinking about manually loading uboot spl from jtag..Either from fel mode, or from a specially crafted sd binary that sets up jtag port on sd. i'm assuming bootrom loads the full "spl" binary to memory, then executes, so there's no dependency on sd as soon as it runs
<RaYmAn> does that make sense? :P
<Boulet> i don't know ;)
sspiff has quit [*.net *.split]
lundman has quit [*.net *.split]
specing has quit [*.net *.split]
termleech has quit [*.net *.split]
arete74 has quit [*.net *.split]
robws has quit [*.net *.split]
sspiff has joined #arm-netbook
termleech has joined #arm-netbook
lundman has joined #arm-netbook
specing has joined #arm-netbook
arete74 has joined #arm-netbook
robws has joined #arm-netbook
<Boulet> on git there is a very small program that you can write on the SD-CARD so that the board boot into fel-mode at power-up
<Boulet> i think for u-boot development, DDR3 initialization must be added to that program
<Boulet> so upon reset, you go to fel mode, and DDR3 is also initialized
<Boulet> from there, you can upload ever code you want anywhere, and execute whatever you want
<Boulet> then if you want JTAG, it is up to you, but i think it is not even necessary
<RaYmAn> probably not necessary, but fun! :)
<Boulet> i admit it would be great to have
<hno> For u-boot development the SPL works fine. The problem is SPL development.
<RaYmAn> well, I do already have jtag on a10, and I'm fairly certain it'll work on a13 as well
<RaYmAn> the hard part is just booting from sd and then suddenly changing to jtag over same pins :S
<hno> Yes, everything indicates A13 have JTAG on PF just as A10 does.
<hno> Booting from SD and then change to JTAG is not difficult.
<RaYmAn> no, but it requires the SPL doing it presumably :)
<RaYmAn> given they use same pins
<Boulet> someone needs to write the code that read and write from NAND ;)
<RaYmAn> well, does nand even work when booted from sd atm? :P
<RaYmAn> Cause in my experience, it doesn't :P So that might be a good start
<mnemoc> sure
<hno> If adding the NAND driver to u-boot yes.
<hno> it adds cleanly.
<Boulet> hno, how big is the SPL ?
<RaYmAn> it has to fit in sram doesn't it? Or am I misundertsanding the boot process?
<hno> ~15 KB with MMC driver.
<hno> yes, SPL have to fit in SRAM.
<Boulet> it seems NAND driver would be quite bigger than MMC driver, is that correct ?
<hno> The full NAND driver with logical layer is way too big. Need to be isolated down to simple physical layer.
<Boulet> right
<hno> and need to figure out why almost any change breaks SPL..
<Boulet> when u-boot loads the linux image from nand, it uses the full logical layer ?
<Boulet> it already has a filesystem actually ?
<mnemoc> by default it: nand read 40007800 boot;boota 40007800
<Boulet> the nand blocks are 128KB if i remember correctly ?
<Boulet> what is uboot's binary size ?
<Boulet> if uboot fits in 1 nand block that will makes things quite easier :D
<mnemoc> 132336
<mnemoc> mmc-only
<Boulet> ah
<mnemoc> an nanda's is.....
<mnemoc> and*
<mnemoc> 245736
<hno> Boulet, yes u-boot by default uses logical layer with partition table, filesystem driver etc.
<Boulet> yeah so i wonder how it is supposed to be done
<Boulet> reserve a few blocks for u-boot at the beginning of the flash ?
<Boulet> and do custom nand-block chaining
<mnemoc> u-boot.bin is a file in nanda
<mnemoc> the kernel itself is written raw on a nand partition
<mnemoc> nanda = vfat
Quarx|2 has quit []
Quarx has joined #arm-netbook
<Boulet> ok
<Boulet> there is no mystery, the BROM looks for a magic string
<Boulet> to load boot0, so our custom boot0 will probably have to do something similar to locate u-boot and load it into SDRAM
<Boulet> and if uboot takes several nand blocks, they will have to be linked, using the physical block index should be fine
<RaYmAn> hno: am I right that you discovered there was some extra, possibly important stuff in the nand before nanda? Or is it safe to dump everything you can from regular booted kernel and then livesuit an a13 image for another (a13) device? Or Did I read stuff wrong? :P
<hno> boot0 and boot1 resides in the first NAND erase blocks outside the logical layer
<RaYmAn> that's the only thing?
<RaYmAn> I'm mostly trying to figure out if I can whip up a livesuit image using the dumped nand* partitions and an a13 livesuite image :P So trying to figure out if those boot0/boot1 are fully generic across a10 or a13 or if there can be device specific stuff encoded in them
<hno> it should be the only pieces in raw nand blocks.
<hno> boot0 & boot1 are cpu specifc
<mnemoc> hipboi's friend confirmed it's all in the first raw block
<RaYmAn> ok
<hno> it's more than one block.
<hno> see src/format/nand_frmat.c
<Boulet> ah cool
<Boulet> that one is easy then
<Boulet> but uboot cannot simply follow boot0/boot1
<Boulet> ah yeah i remember the first block of a NAND is guaranteed to never be a bad block out of the factory
<Boulet> probably for boot reasons
<Boulet> maybe a megabyte or so should be reserved at the beginning of the nand flash
<RaYmAn> on eMMC, it's even encoded in the eMMC protocol as seperate "emmc" boot partitions (not regular partitions)
<Boulet> yeah here, some custom logic has to be done :(
<Boulet> i think using a magic string followed by the number of the next physical nand block
DEAT has quit [Read error: Connection reset by peer]
DEAT has joined #arm-netbook
<hno> RaYmAn, A10 is not using SD boot protocol as far as I know.
<RaYmAn> hno: yeah, it's not using eMMC afaik (and I believe it only applies to eMMC)
<hno> It applies to SD as well.
<RaYmAn> oh
<hno> And A1x can boot from eMMC. A13 even have 8-bit MMC bus capability. A10 only 4 bit bus width.
<hno> A13 also supports cATA.
<hno> CE-ATA I meant.
<Boulet> oh ?
<Boulet> what is that ?
<mnemoc> so it's not merely a mutilated A10
<Boulet> haha
<Boulet> mutilated ;)
<mnemoc> :)
<RaYmAn> so harddisks over mmc? lol
<mnemoc> sounds better than crippled to me...
<mnemoc> WarheadsSE used the term "gimpy" last time
<WarheadsSE> Yeah, american saying
Quarx has quit []
<Boulet> getting late here
<Boulet> good bye all
nibb has quit [Read error: Connection reset by peer]
Boulet has quit [Quit: zzzzzz]
nibb has joined #arm-netbook
tzafrir_laptop has joined #arm-netbook
CIA-121 has quit [*.net *.split]
hark has quit [*.net *.split]
RaYmAn has quit [*.net *.split]
diego_ has quit [*.net *.split]
rmull has quit [*.net *.split]
ajmitch has quit [*.net *.split]
CIA-121 has joined #arm-netbook
diego_ has joined #arm-netbook
hark has joined #arm-netbook
RaYmAn has joined #arm-netbook
ajmitch has joined #arm-netbook
rmull has joined #arm-netbook
traeak has quit [Remote host closed the connection]
traeak has joined #arm-netbook
<hno> mnemoc, have imported the a13 hack in Allwinner-info/a13/test1 and fixed up Makefile a bit making it build with Linaro bare-metal toolchain.
<hno> full toolchain segfaults in linking for some reason.
<Turl> RaYmAn: reeealy old talk but if you want to dump full nand make a SDCard CWM and dd it :)
acfrazier has quit [Ping timeout: 272 seconds]
<Turl> make sure you use latest kernel that has access to full nand
<RaYmAn> Turl: on A13? :P
<Turl> should work :P
<RaYmAn> except the kernel is missing some stuff for a13
<Turl> yeah that :P
acfrazier has joined #arm-netbook
<mnemoc> hno: ok
<mnemoc> Turl: have you tried .39 yet?
<Turl> mnemoc: yeah, works
<mnemoc> not for everyone :<
<Turl> mnemoc: weird that stuff explodes on wemac though :<
<Turl> mnemoc: btw I've been using https://github.com/allwinner-dev-team/linux-allwinner/commit/5cbb716e1ce70192e834e7a3259b1ff0fa9c6b79 for some days on my mele, works fine
<Turl> it's a squash of lundman's wemac patches
eFfeM has joined #arm-netbook
acfrazier has quit [Ping timeout: 260 seconds]
<mnemoc> Turl: that one was "rejected" because it removed too much knowledge
acfrazier has joined #arm-netbook
<Turl> removed too much knowledge? o.O
<mnemoc> without proper documentation those commented blocks of incomplete features are our only way to understand what else the units can do
<mnemoc> and how
<Turl> mnemoc: but how is it 'removing' the knowledge?
<Turl> the commented stuff will still be there
<mnemoc> no, it's removed by the patch
<mnemoc> and then forgotten in the history
<Turl> just not visible and polluting the driver
<Turl> mnemoc: you can just go to an old revision and see it
<Turl> that's the point of having revisions :<
<mnemoc> we only remove code we *know* we don't need
<mnemoc> in this case the commit mixes both, cleanup of unused code, and fixes
<Turl> :<
<mnemoc> it damages the history
<mnemoc> you don't know what to look
<Turl> I guess I'll just make 3 separate commits then
<Turl> //->/**/, optimize, readability, kill commented stuff with fire :P
<mnemoc> you have 3 parts. the actual fix, style changes, and removal of disabled but valuable knowledge
<mnemoc> then you know what to look at
Vayu has joined #arm-netbook
tahnok has joined #arm-netbook
dfenwick has quit [Ping timeout: 245 seconds]
dfenwick has joined #arm-netbook
<hno> Yes, well defined commits helps a lot in preserving history.
Gujs has quit [Read error: Operation timed out]
<mnemoc> http://pastie.org/4413686 <--- .39 from nand on the a13-olinuxino (3.0-v2 as-is)
ejstacey has quit [Ping timeout: 248 seconds]
ejstacey has joined #arm-netbook
<hno> mnemoc, you got our kernel running on A13?
<mnemoc> yes
<hno> great!
<mnemoc> :)
<mnemoc> still need to merge the code from the SDK, and then apply the rest of the changes made to sun4i to sun5i. but yes, runs :)
<hno> <6>pcf8563 1-0051: low voltage detected, date/time is not reliable.
<hno> [ 3.550000] pcf8563 1-0051: low voltage detected, date/time is not reliable.
dfenwick has quit [Ping timeout: 245 seconds]
<hno> Is there an RTC in A13?
<hno> or is that a separate chip?
ejstacey has quit [Ping timeout: 248 seconds]
* hno need to study schematics better.
<hno> Ah, there is a separate RTC.
ejstacey has joined #arm-netbook
<mnemoc> hno: http://linux-sunxi.org/images/4/4f/A13-olinuxino-rev-a-front.jpg <--- on the bottom of the picture, just above the 10pin header
<mnemoc> looks like the RTC from my ignorant eyes
<hno> But no battery unless you connect a LiPo.
<mnemoc> [mmc]: mmc 0 power off !!
<hno> Nice, it has full OTG including power both direction.
<mnemoc> that was is awesome indeed
<hno> All of PF is available on RM23 & RM20 in schematics. Would expect RM11 & RM12 from the board picture.
<tahnok> hey all
<tahnok> I'm wondering if there's a way to rotate the video output of my mk802
<rm> 1) no 2) which os?
dyoung-away is now known as dyoung
<tahnok> using a lubuntu image from the rikomagic forums
<rm> oh, then it should be
<rm> install 'xrandr'
<rm> then see "xrandr --help" for "--rotate"
<rm> but I am not completely sure if fbdev supports that
<tahnok> yeah, I tried xrandr but it fails
<hno> it does not.
<tahnok> I tried xrandr --output default --rotate left
<tahnok> but it fails with output default cannot use rotation 'left' reflection 'none'
Gujs has joined #arm-netbook
<hno> hardware supports rotation, but we are missing xord drivers. We do have kernel drivers however.
<hno> s/xord/xorg/
<ibot> hno meant: hardware supports rotation, but we are missing xorg drivers. We do have kernel drivers however.
<hno> should also be possible to rotate using mali.
<rm> one would think rotation in 90 degree increments is not a rocket science and something fbdev should support...
<hno> rm, you are right. fbdev do actually support rotation.
<tahnok> is there a configuration option in evb.bin?
<hno> __u32 rotate; /* angle we rotate counter clockwise */
<traeak> uint32?
<hno> no idea if that's used by disp driver.
<hno> traeak, from kernel fb.h header
<traeak> 1, 2, 3 pi that's not very descriptive :-p
<mnemoc> most of that stuff is connected to ioctl()s
<traeak> 1, 2 or 3 radians isn't very descriptive i mean
Quarx has joined #arm-netbook
<CIA-121> rhombus-tech: marco master * r69dff25bb636 /allwinner_a10/ainol_novo_elf_7.mdwn:
<hno> traeak, the FB_ROTATE_.. defines?
<Quarx> anyone know about? http://www.eastpower.com.tw
<hno> Quarx, no, what about them?
<RaYmAn> fwiw, there is some work being done on using Android drivers (in particular opengles and friends) on plain GNU/Linux. Working on gingerbread roms so far, but ics and jb drivers are coming.
<RaYmAn> Combined with wayland, it could work well =P
<traeak> hno that's cool, thanfully documented. it's a pet peeve of mine, peopel give us spec docs and they tend to not tell you what the units are
<Quarx> hno: i have this device http://www.eastpower.com.tw/front/bin/ptdetail.phtml?Part=P17&Rcg=6 And i need kernel sources, but they don't want give me.... i thought, maybe someone can get it:)
<mnemoc> RaYmAn: ndiswrapper style?
<Quarx> from CHina'
<RaYmAn> mnemoc: I dunno...A linker hack of sorts, so you end up with both glibc and bionic in memory, but you can call it all from glibc programs
<mnemoc> Quarx: looks nice
<mnemoc> Quarx: as it's android 2.3
<Quarx> mnemoc: yes, but without kernel this device useless
<mnemoc> sad*
<Quarx> i can't upgrade to 4.1
<hno> Quarx, which SoC is it using?
<Quarx> Qcom msm8260
<Quarx> 1.5ghz 2x
<hno> pretty sure there is kernel sources for that one out there.
<Quarx> for msm8260 yes, but not for this device
<RaYmAn> hno: not all socs are as easy to get working across devices as allwinner ;) *curses tegra*
<Quarx> many drivers missing
<hno> Quarx, that's usually the case, even in "official" sources.
<hno> Quarx, if manufacurer refuses to give you sources you have two options
<hno> a) Forget it
<hno> b) Collect the sources you can find for each piece used in the device.
<hno> But first of all, verify that you have unlocked bootloader access. Otherwise it's kind of pointless.
<Quarx> bootloader unlocked, fastboot available. :)
<hno> and get in touch with others hacking on devices with the same SoC. xda-developers is a good place to look.
<RaYmAn> and perhaps check cyanogenmod github.. I'm fairly certain there are at least a couple of 8260 devices in there and kernel sources for those devices
arokux_h has joined #arm-netbook
<RaYmAn> android of course, but still source =P
<Quarx> yes, many devices with 86(2)60
<RaYmAn> is 8260 just the non-radio version?
<Quarx> 8260 gsm, 8660 cdma
<RaYmAn> ah
<RaYmAn> what device do you have?
<RaYmAn> not bad for a china-copy, heh
<Quarx> yea
eFfeM has quit [Quit: Leaving.]
<RaYmAn> if you have fastboot boot, you should be able to get pretty far with dmesg output, /sys/ dumps and ramconsole (Assuming ramconsole is enabled in stock kernel)
<hno> Quarx, pick the closest match to your device and work from there. Make sure you have full recovery available. UART access to bootloader & kernel is also preferred.
<RaYmAn> hno: what compiler are you using for that a13 test1? codesourcery?
* mnemoc prefer to stick to google phones
<RaYmAn> samsung is pretty good as well
<Quarx> don't buy motorola :)
<RaYmAn> mnemoc: 'k, thanks.
<hno> RaYmAn, linaro bare metal.
<hno> works for me.
<mnemoc> RaYmAn: therefore, galaxy nexus :D
<hno> no idea if the resulting binary works.
<RaYmAn> doesn't work with arm-eabi-4.4.3
<mnemoc> RaYmAn: beware, that test takes like 10M to write and verify the 512M of ram
<mnemoc> 10m*
<mnemoc> but works :)
<hno> Yes, it do not enable full clock on the CPU.
<WarheadsSE> that's our latest glibc/gcc pre-compiled for x86_64
<hno> I think.
<mnemoc> WarheadsSE: https://github.com/hno/Allwinner-Info/tree/master/A13/test1 is very... special
<hno> indeed.
<hno> well, it compiles fine with any gnueabi compiler as well, but ld gets an heartattack on something.
<WarheadsSE> I see it
<WarheadsSE> we do all our ld on ARM devices
<hno> Not sure if it's a compiler flag missing, or if the ld script is barfed somehow.
<WarheadsSE> can't check it atm
<hno> collect2 giving a segmentation fault isn't exactly verbose error message
<hno> don't worry about it. It's just temporary anyway.
<RaYmAn> hno: odd - directly from your source, it doesn't compile with that toolchain you linked :S
<hno> once the pieces is understood it will be turned into an A13 u-boot SPL.
<WarheadsSE> RaYmAn: give my stack a shot, for giggles
<hno> RaYmAn, try again. Had forgot to push a flags change.
<hno> mnemoc, how did you get it to compile?
<RaYmAn> there we go :)
<Turl> Quarx: newest motorola photon has unlockable bootloader :) I hope they keep it up
<hno> mnemoc, your UART works now?
<mnemoc> hno: UART1, yes
<hno> Turl, do they also publish kernel sources?
<Quarx> Turl: what is price of motorola photon ?
<hno> mnemoc, how did you get test1 to compile?
<RaYmAn> most major non-asian companies do publish source these days (eventually)
<mnemoc> hno: http://dpaste.com/783429/ <--- my changes over the original .zip
<Turl> hno: yeah motorola publishes sources :)
<Turl> hno: although release-to-source time seems to be the output of a random function
<Turl> Quarx: it's CDMA 4G LTE
<mnemoc> hno: and then: http://dpaste.com/783431/
<Turl> and it isn't sold yet
<hno> Turl, crippled by any binary-only modules?
<Quarx> ah cdma:(
<Turl> hno: idk, it isn't sold yet
<Quarx> i want gsm lte
<Turl> HSPA or HSPA+ is more than enough for me, nobody offers LTE on my country
<Turl> next year with any luck they'll start I guess
<Quarx> in my country no cdma at all
<mnemoc> hno: if I use the normal linaro toolchain the linker fails with undifned references to __init_array_start and __init_array_end
<destinal> Turl It
<destinal> Turl: I think the current speculation is that it's going to be snapdragon
<Turl> destinal: snapdragon what?
<Turl> Quarx: yeah no CDMA here either
<Turl> there was CDMA once upon a time, like 10+ years ago :)
<mnemoc> same in .cl
<Turl> movicom I think
<destinal> Turl: all anyone will say is 1.5 ghz dual core (probably snapdragon) .. maybe MSM86xx ?
<Turl> destinal: which phone?
<destinal> Turl: photon Q
* mnemoc cries over his unpowered uSD slot
<Turl> ah
<Turl> yeah pretty sure it's an S4 destinal
<hno> mnemoc, no power in the SD slot?
<Turl> qualcomm is the only one making CDMA LTE radios
<mnemoc> hno: that what the driver claims
<destinal> anyone know whether freedreno is in a usable state?
<mnemoc> hno: hipboi's break out doesn't expose all pins, only the useful ones, and those are alive
<Turl> last I checked they were way behind lima destinal
<Turl> and lima is not usable yet afaik
<mnemoc> hno: and I can't reach the pins inside the socket
<hno> mnemoc, pin 4 should have 3.3v.
mpthompson has quit [Quit: OUCH!!!]
nibb has quit [Read error: Connection reset by peer]
<mnemoc> hno: the board isn't happy with me touching the pins :p .... it has rebooted 3 times already :p
nibb has joined #arm-netbook
<mnemoc> in the 14pin jtag breakout pins 1,5,7,9,11,13 are high
<RaYmAn> that sounds about the same as I'm getting
<mnemoc> uhm
<mnemoc> maybe it's the "card detect"
<mnemoc> enabling the PB3 (card detect) in the GPIO2 header didn't help :<
<RaYmAn> try and see if you can convince the kernel driver to assume always connected (and hence ignore card detect) and boot with it in
<RaYmAn> to see if kernel sees it then
<mnemoc> oh! adb reboot with the pin jumped and got into `fel`
<RaYmAn> heh
* mnemoc tries to remember if it's a fel card or not :p
<RaYmAn> reboot without the card in, but pin jumpered? :P
<mnemoc> hehe
<hno> mnemoc, don't touch. Use the multimeter and isolate it so you can only hit one pin at a time.
<mnemoc> easier said than done :<
<hno> a piece of tape around the multimeter pin should do the trick.
<hno> same when you use a needle. tape the needle, cut the tape just outside the sharp point. That way only the sharp point connects.
<hno> mnemoc, which pin?
<hno> PB3?
<mnemoc> PB3 = GPIO-2/11 = card detect input
<hno> GPIO-2-11 is PIN7 / PC1/NALE/SPI0_MOSI. Messing with that and you block the embedded NAND.
<hno> which would result in FEL mode.
<hno> assuming the SD do no work.
* mnemoc scratches his head and go back to the schematics
<mnemoc> I know I was tired when I did http://linux-sunxi.org/A13-OLinuXino#Expansion_ports
* hno too. PIN7 is not pin 7..
<mnemoc> but I would hate to have done it wrong
<hno> Very confusing numbering.
<mnemoc> that's why I did the wiki thing
<mnemoc> to try to make some sense out of it
Quarx has quit [Quit: KVIrc 4.0.4 Insomnia http://www.kvirc.net/]
<mnemoc> is it fear to destroy a mSD/SD adapter to get a "break out" :p
<mnemoc> fair*
<hno> OK. GPIO-2 PIN numbers are numbered lengthwise, not the normal pair numbering. Schematics have both.
<hno> mnemoc, you have one of those strange SD card to uSD socket adapters?
<mnemoc> at least a dozen
<hno> how come?
<hno> Most people only have uSD card to SD socket adapters.
<mnemoc> sorry. uSD goes into SD
<mnemoc> but inside they have a plastic thing with paths
<mnemoc> which goes from uSD pins to fat SD pads
<hno> Yes, but you can't fit that in the uSD socket.
<RaYmAn> OK, JTAG works on A13 too (through same pins)
<hno> RaYmAn, excellent! script.fex comments seemed to indicate pins was rearranged a bit, but may be just a silent error (no one notices unless using the comments as note on which pin goes where)
<RaYmAn> hno: the pins in the script.fex I've seen are wrong for some reason (checked both A10 and A13) - the pins are swapped around in practice
<RaYmAn> ^ that's what works
<hno> RaYmAn, pin names in script.fex are mostly irrelevant for function. As long as all GPIOs in the group is set right things works.
<RaYmAn> oh, hmm
<mnemoc> "don't trust the comments" (tm)
<RaYmAn> I guess that makes sense
<destinal> Turl: lima looks really promising at the moment. very glad mali seems to be overtaking the ARM SoC GPU world
<RaYmAn> it's just funky they do it wrong in that file still :P
<RaYmAn> lol
<hno> The exception is those few that map to GPIO, like SD card detect and USB ID, VBUS_DETECT, VBUS_ENABLE
dyoung is now known as dyoung-away
<hno> RaYmAn, can you map those to 4 bit SD interface names as well please?
<mnemoc> RaYmAn: drivers are very lazy, after checking foo_enable != 0 the use a helper to request all GPIOs in that section regardless the name
<RaYmAn> hno: should already be in the jtag page
* mnemoc would link the JTAG_x names to JTAG
<mnemoc> the page
<RaYmAn> Good idea
<hno> RaYmAn, just crosschecking the datasheet. We should add pin numbers to the A13 PIO page as well.
<RaYmAn> hno: pin numbers on what? It seems uhm, irrelevant for that particular page that PF is usually brought out to microsd
<hno> Well, A13 do have all pins exposed which is great for hacking. It's not so much for PF but in general.
<mnemoc> RaYmAn: he mains the pin number in the chip
<mnemoc> means*
<RaYmAn> ?
<RaYmAn> I figured PXY would be sufficient
<hno> yes
<mnemoc> e.g. PF0 is 107
<mnemoc> the A10 has all balls hidden, but with A13 people can actually use each directly
<RaYmAn> oh
<mnemoc> so you can eventually use a pin which isn't connected in the board
<mnemoc> for those can solder obviusly :p
<mnemoc> who*
<RaYmAn> Indeed
<RaYmAn> :P
<RaYmAn> I'll let someone with those skills add the pin numbers then ;)
<hno> I said "we".
<mnemoc> RaYmAn: they are in the schematics of the olinuxino
<RaYmAn> :)
<hno> mnemoc, not sure I trust those as your SD port is not working.
<mnemoc> hno: good point :<
<hno> Do we have script.bin from an actual working A13 firmware somwhere?
<arokux_h> hi, how are you normally upload a new kernel to an SD card? 1) mount boot partition 2) cp ?
<hno> arokux_h, yes
<hno> then umount, eject
<arokux_h> hno, to sync?
<arokux_h> (flush buffers?)
<hno> eject syncs and removes the block device when done.
<arokux_h> which block device?!
<arokux_h> of a partition?
<arokux_h> why should it be removed?
<mnemoc> to move it from one device to the other
<RaYmAn> hno: the test1 app seems to work fine - jtag reads of the memory it sets correctly sets it when read back using jtag
<hno> good
<mnemoc> good news, the fel-card works. it's only in linux where the uSD doesn't work
<RaYmAn> so why didn't it work yesterday? :S
<mnemoc> RaYmAn: no clue
<mnemoc> RaYmAn: now it does. SAME card
* hno is getting a bit impatient to play with his board on monday
<mnemoc> :)
<hno> mnemoc, maybe UART0 works also?
<arokux_h> hno, I'm sorry, can you please explain which block device is getting removed?
<mnemoc> hno: having UART1 working, why may I want to disrupt the uSD enabling uart0?
<RaYmAn> hno: I have a script.bin off this a13 tablet if you want
<hno> arokux_h, the one of the removable SD card.
<hno> arokux_h, just try it and you'll notice what I mean.
<arokux_h> hno, I see, but this isn't necessary so that it is removed, or?
<mnemoc> arokux_h: eject will turn the block device to size 0
<hno> arokux_h, if you plan on removing the card you should eject first.
<hno> it's easy and safe.
<arokux_h> I've done only sync up to now, with usb sticks, how ever o-O
<hno> same there.
<hno> sync also works, but not as safe.
<hno> RaYmAn, yes please. Not that I expect anything strange in there.
<RaYmAn> hno: skumler.net/c925.bin
<arokux_h> eject mmcblk0p1
<arokux_h> eject: did not find a device /dev/mmcblk0p1 in /sys/block/
<arokux_h> oh, I should have done: eject mmcblk0
<mnemoc> hno: as I won't be here next week grab the .config (for stock 3.0-v2) I used for the test. https://filetea.me/t1s354b0
<RaYmAn> mnemoc: how are you booting it etc? Just replacing on nandc?
<mnemoc> I copy uImage into nanda and in the uboot prompt:
<mnemoc> setenv bootargs console=${console} root=${nand_root} init=${init} loglevel=${loglevel}
<mnemoc> fatload nand 0 0x43000000 script.bin
<mnemoc> fatload nand 0 0x48000000 uImageX
<RaYmAn> ah, right
<RaYmAn> clever
<mnemoc> bootm 0x48000000
<mnemoc> can't risk to ruin the stock os
<RaYmAn> I guess that doesn't help me then (because of my fear of breaking script.bin...)
<RaYmAn> mnemoc: olimex guys don't have a livesuit image for the preflashed one?
<mnemoc> you can preload a test .bin
<mnemoc> and test it
<hno> RaYmAn, you can name script.bin anything you like for the kernel.
<hno> RaYmAn, they do, but not uploaded anywhere yet.
<mnemoc> also no livesuit for linux "yet"
<RaYmAn> mnemoc: preload? I need it so u-boot puts console on sd uart
<RaYmAn> mnemoc: no, but booting a vm or whatever to flash is possible :P
<mnemoc> requires usb2.0 so qemu/kvm doesn't work
<RaYmAn> virtualbox? :)
<mnemoc> oracle
* RaYmAn gives up
<hno> kvm have usb2 now.
tuliom has joined #arm-netbook
<mnemoc> nice
<hno> but it's not very flashing friendly. Binds to usb id, and do not like reconnects.
<mnemoc> :(
<hno> tried flashing an old phone and it soft-bricked it due to the phone reconnecting.
<hno> binding the whole USB PCI controller to the kvm guest should work, but my boxes are too old for that.
<RaYmAn> mnemoc: you had allwinner-v3.0-android-v2 partially booting, yes?
<Turl> virtualbox works just fine for livesuite
<Turl> that's what I use
<mnemoc> RaYmAn: http://pastie.org/4413686
<hno> mnemoc, if BROM boot from SD works but Linux not then maybe one of the data pins is not working. I think BROM uses one-bit SD mode, maybe even SPI mode.
<hno> i.e. one of pins 1,2,8 maybe isn't working in your board.
<hno> RaYmAn, you should be able to JTAG the UART PIO settings
<hno> but right.. you'll loose JTAG capability if SD gets enabled.
<RaYmAn> indeed
<RaYmAn> and given there's no jtag reset pin to hold it ready so I can jtag halt it early, it's hard to control
<hno> Maybe Pin #99 is JTAG port select pin on A13. It's NC in datasheets which I do not beleive.
<mnemoc> hno: http://dpaste.com/783458/ <--- from dimitar's script.bin
<mnemoc> hno: but no idea who uses card_boot0_para
<hno> boot0
<hno> recorded in the header of boot0
<mnemoc> for the sake of testing I'll remove sdc_detmode and sdc_det from mmc0_para
<hno> in an phoenixcard
<hno> RaYmAn, if I am not mistaken the JTAG port select pins overrides PIO forcing the port to JTAG.
<mnemoc> hno: didn't help :<
<RaYmAn> hno: hmm, interesting if that's the case
<RaYmAn> the hard part will be finding it on this tablet :S
<hno> mnemoc, I think 1 is detect by SD bus probing.
<RaYmAn> I'm not really seeing a whole lot of test points
<hno> RaYmAn, pin 99 on the cpu.
<hno> not big, but accessible.
<mnemoc> dd if=/dev/zero of=/dev/sde bs=1024 count=1 skip=8
<mnemoc> ^---- safe to un-fel a card?
<hno> yes
<hno> err, seek=8
<hno> mnemoc, !!!
<mnemoc> I always confuse those too :<
<RaYmAn> it only wipes out MBR and stuff if you do it wrong ;P
<hno> skip skips data.
<hno> seek positions
<mnemoc> skip = if, seek = of
<hno> yes
* mnemoc remaking the partition table :p
<hno> added PF pin numbers. Hope it's correct.
<hno> Matches datasheet and olimex schematics.
<hno> Wonder if RESET_N pin also resets JTAG. Would be nice if it does not.
<RaYmAn> hm
<RaYmAn> how would I test that
<RaYmAn> I think the board has a button connected to RESET_N
<mnemoc> there is a reset button, yes
<mnemoc> but not sure exactly where it goes
<RaYmAn> I meant on this tablet's board, but yeah :)
<mnemoc> RaYmAn: open it!
<RaYmAn> I have
<RaYmAn> there is a button that resets :>
<mnemoc> :D
<RaYmAn> jtag still works after reset, but it starts up in run mode, so I'd probably have to really quickly halt it
<mnemoc> try to add a breakpoint
<RaYmAn> I suppose that's a good idea
<RaYmAn> probably at 0xffff400?
<mnemoc> try :)
<RaYmAn> usage: bp <address> <length> ['hw']
<RaYmAn> hmm
* RaYmAn ponders what length means in this context
<RaYmAn> google to the rescue? :P
<RaYmAn> nope :(
<hno> ffff4000 is the reset entry point.
<RaYmAn> so I guess that doesn't work
<RaYmAn> hm
<mnemoc> I guess the length is used to set a range (anything between address and address+size-1)
<hno> reset button on olimex board goes to the pmu, not the cpu.
<hno> same in the mele for the reset jumper I found.
<RaYmAn> ah
<RaYmAn> quite possible
<mnemoc> ; sdc_detmode --- 卡检测模式 |
<mnemoc> ; 1-gpio检测,2-data3检测,3-无检测,卡常在(不卡拔插) |
<mnemoc> ; 4 - manual mode(from proc file system node) |
* mnemoc wonders why 1 instead of 2
arokux_h has quit [Remote host closed the connection]
<mnemoc> didn't help :'(
<hno> mnemoc, goot question. if true then 1 should require a gpio pin assigned.
<mnemoc> yes, PB3
<mnemoc> why wasting a pin on that....
<mnemoc> -sdc_det = port:PB03<0><1><default><default>
<mnemoc> but it didn't work anyway, so the problem is not there
<mnemoc> and also no way to enable the usb host ports...
<mnemoc> i suppose I should send him the board back :<
DEAT_ has joined #arm-netbook
<hno> do you see the hub?
<mnemoc> $ adb shell busybox lsusb
<mnemoc> Bus 001 Device 001: ID 1d6b:0001
<mnemoc> Bus 003 Device 001: ID 1d6b:0001
<mnemoc> Bus 002 Device 001: ID 1d6b:0002
<hno> 3 busses?
<hno> but no hub..
<mnemoc> and I have the wifi module, a usb/rs232, a keyboard and usb stick connected
DEAT has quit [Ping timeout: 248 seconds]
<mnemoc> but yes, not even the gl850G is there
<hno> wonder if the bus is enabled. remember wifi hack
<mnemoc> usbc0 was usb_host_init_state = 0, but I changed that to 1
<mnemoc> usbc1 was/is 1
<mnemoc> but this is dimitar's script.fex, which afaik it's what they used to test the boards there
<hno> not sure which i/os they tested