<drachensun>
didn't seem like good mailing list material, its an ugly hack and I think there is probably some userspace way to avoid all this, but I couldn't figure out how
simosx has quit [Quit: Αποχώρησε]
TestModule has quit [Quit: Page closed]
stefanro has joined #arm-netbook
stefanro1 has quit [Ping timeout: 252 seconds]
dyoung-away is now known as dyoung
freakazoid0223 has joined #arm-netbook
<WarheadsSE>
quick ?
<WarheadsSE>
can I compile the uboot for the a13 olinuxino from v5
<cholo>
como puedo instalar ubuntu 12.04 desde ubuntu 10.04 desde pen desde la bios a q boton le tengo que dar
<cholo>
para q se instale
<cholo>
ya tengo la iso y descomprimida de universal usb en un pen de 8g
cholo has quit [Client Quit]
<rellla>
cholo: try english please. you'll get more help then.
<specing>
blablabla
gimli has joined #arm-netbook
eFfeM has joined #arm-netbook
<eFfeM>
gm
gimli has quit [Ping timeout: 245 seconds]
gimli has joined #arm-netbook
gimli has quit [Ping timeout: 251 seconds]
tinti has joined #arm-netbook
gimli has joined #arm-netbook
rz2k has joined #arm-netbook
discopig has quit [Read error: Connection reset by peer]
Sv has joined #arm-netbook
<Turl>
morning
<mnemoc>
moin
tinti_ has joined #arm-netbook
tinti_ has quit [Quit: Leaving]
tinti_ has joined #arm-netbook
alcides has joined #arm-netbook
hansg has joined #arm-netbook
<hansg>
Hi guys, I'm working on getting an A10s to boot "our" u-boot and uImage from sdcard. So far I'm moving carefully, a10-meminfo shows the same settings as we've for the a13_mid / a13_olinuxino
<hansg>
Except that the io_width / bus_width for the 2 a13 boards is 8 / 16, and for my a10s it is 16 / 32.
<mnemoc>
hansg: any thought regarding the led-ata patch?
<mnemoc>
hansg: if you have a cubieboard hno documented the way of getting the nand replaced with a second mmc and so been able to breakout jtag
<mnemoc>
err
<mnemoc>
ignore that
<mnemoc>
stupid comment. you need jtag on your A10S stick
<hansg>
mnemoc, no not really, sorry I don't have any background info / knowledge on it, so no opinion, other then that I like the functionality it offers, but it seems that the discussion is about some renaming going on, not the functionality ?
<mnemoc>
f* migrane
<mnemoc>
hansg: in general I don't feel confortable adding non-sunxi patches to our tree
<mnemoc>
so i would like to know what other devs feel about crossing the line for this particular feature
<hansg>
I understand, so maybe ask the submitter if it is possible to do a patch which gives the hdd led without doing the rename stuff ?
<mnemoc>
we will still be patching libata
<mnemoc>
sounds like a reasonable patch to be mainlined
<hansg>
Right, I would like to suggest the following: 1) ask the submitter to do a minimal patch, so without renaming etc, just the minimum necessary to get the functionality. 2) then I'll happily give it a quick review to see if it does not impact things speedwise
<hansg>
WRT mainlining, I agree. But this is a chicken and egg problem, mainlining likely won't happen until there is a user for it, and that user would be sunxi ata code which is far from upstream
<mnemoc>
:)
<Turl>
why not make it work with generic ata devices on mainline?
<Turl>
mainline that
<Turl>
then backport to sunxi tree
<hansg>
My main worry is the patch may end up doing too much work in a hot path in libata slowing things down, which is a reason why getting it mainlined would be really good, because then people who actually know the code will look at it
* mnemoc
is alergic to lkml
<hansg>
Turl, I'm not sure if they will take a patch adding a led-trigger without also adding at least 1 consumer of that trigger in the same patch-set
<Turl>
hansg: maybe I got the patch wrong
<Turl>
but I supposed there was a led trigger framework of sorts, where devices might "trigger_led()" of some such, and that would generate an user specified action?
<Turl>
so the consumer would be the ata code
popolon has joined #arm-netbook
<hansg>
Turl, no the ata code is the producer of the trigger...
<hansg>
OK, so I've just taken a look, and the rename although sensible is not something which we should carry without it going through upstream, the rest is basically a 1 line libata change which although outside of the sunxi realm is something which we could carry IMHO, just the 1 line change is worth the added functionality. And we can then also ask the submitter to send the full version with rename upstream ...
<hansg>
So back to the a10s: I would like to get the boot0 from there any ideas for doing that ?
<Turl>
now that I've read the code, looks like a sensible patch for upstream
<Turl>
I doubt there would be much, if any, opposition to it
<Turl>
it could be separated I guess (rename and move in one, add the call on the ata path in a second patch)
<hansg>
Turl, agreed. So we will need to have someone to submit it upstream, we could ask the original submitter, but I'm not sure he groks the process, so any takers ?
<hansg>
In the mean time we could carry a minimal version which just adds the 1 line call of ledtrig_ide_activity(); to libata-core.c
<mnemoc>
hansg: see hno's allwinner-info repo. he has toys to grab raw nand stuff from a running system iirc
<Turl>
hansg: I can submit it if I get permission from the author
<mnemoc>
hansg: also, the lichee-dev branch in u-boot-sunxi has nand support. you can hack something up to extract the first raw block from the nand using it
<mnemoc>
hansg: it's intended as a drop-in replacement for nanda's linux/u-boot.bin
<hansg>
Turl, the author is unknown, so probably best to just redo from scratch, all it is a git mv + s/ide/disk/ then you're the author :)
<hansg>
Turl, I've a feature request for the Fedora A10 images to make the green led light on boot "like the build-in android does", so it definitely has a green led :)
* hansg
will have his own cubie soon(ish)
<Turl>
well, my cubie didn't come with android but with some factory test image
<mnemoc>
that post on the cb list includes .fex stuff for a blue and a green led
<Turl>
btw, I saw the patch missed a couple of renames on the leds.h, nobody seems to have noticed :P
<mnemoc>
back in 20m
<Turl>
I think I found the leds, they're labelled led1 and led2
<Turl>
needed to look more carefully :)
rellla has joined #arm-netbook
<WarheadsSE>
anyone know the reasoning behind not accepting the patches for the a13_onlinuxino_micro from a while back?
<Turl>
WarheadsSE: patches to what?
<WarheadsSE>
linux-sunxi kernel
<WarheadsSE>
supposedly they were sent
<WarheadsSE>
and someone rejected them, ffrom what I understand from olimex
<Turl>
what kind of patches were them? and to what tree?
<Turl>
disregard the 2nd question
SouL_ has joined #arm-netbook
<Turl>
WarheadsSE: I don't see any olinuxino patches for linux-sunxi
<WarheadsSE>
apparently the -micro has no AXP, and a few other differences from the a13-olinuxino
<WarheadsSE>
and yeah, Turl, I know
<WarheadsSE>
idk if this what when it was under the amery repos
<Turl>
it was probably lost/forgotten then I guess
<WarheadsSE>
quick ? then
<WarheadsSE>
I modified a local copy of the uboot, adding the dram.c differences for the -micro (256M), but my eventual SPL always comes up 512. Thoughts?
<Turl>
did you configure uboot with the right board (the new one you added)?
<WarheadsSE>
yup
<Turl>
and did you make clean or equivalent?
<WarheadsSE>
copy & duplicate a13_olinuxino to a13_alinuxino_micro, boards.cfg, baords/allwinner/a13_olinuxino to a13_olinuxino_micro, make clean, make mrproper, make a13_olinuxino_micro
<Turl>
boards.cfg with correct data?
<WarheadsSE>
I can't right now, but I will get a diff for you to review
<WarheadsSE>
(but yes)
<WarheadsSE>
it comes up with the modified name, just 512, and SPL loop
<br->
hmms. what is the name for the action of converting something into a set of pronouncable words. in this case, say mapping the bits of an md5sum on to english words
<Turl>
encoding?
<br->
pgpphone used to do it. short of irc channels full of nerds, so asking here :)
<br->
wouldn't it be nice if there was a totally legal GPL'd android market app that you could bootstrap with a useful software list using only a handful of english words (aka. md5sum reference to bittorrent DHT)
<br->
so the key encodes a .torrent containing a package respository, device fetches that, provides nice UI to user
<Turl>
there's fdroid
<br->
aah :) yes.. i'm more interested in interesting ways to make it convenient to get google apps and suchlike installed on AOSP builds without undue suffering
<mnemoc>
beware the a13-micro doesn't have pmu so it's .fex needs to be adjusted for that
SouL_ has left #arm-netbook [#arm-netbook]
rsalveti_ has joined #arm-netbook
rsalveti has quit [Ping timeout: 245 seconds]
rsalveti_ is now known as rsalveti
popolon has quit [Quit: Quitte]
ZaEarl_ has joined #arm-netbook
<hansg>
I've just finished modifying my A10s stick, so now it has a serial console "connector" on it. Lrarning some interesting stuff, IE the kernel outputs messages about axp15_xxx and indeed upon examing the pmu it is an axp152, for which we don't have a driver yet ...
<hansg>
Are there any good pictures of A13 pcb-s out there ? Good enough to see the pmu they use ?
<hansg>
I guess I'dd better disable the axp driver for now, leaving the axp run at its defaults ...
* hansg
guesses the axp152 is a stripped 202 to match the stripped A10
<hansg>
mnemoc, does anyone have a linux-sunxi kernel running successfully on a13 based machines ? (just wondering)
pwhalen has joined #arm-netbook
ganbold__ has joined #arm-netbook
ganbold_ has quit [Ping timeout: 272 seconds]
<ZaEarl_>
my a13 tablet uses axp20
<Quarx>
Someone know, which function can be replace this : dma_cache_maint ? Driver from 2.6.32 use this, but since .35+ this func is static and can't be used in drivers.
<Turl>
Quarx: /* * Make an area consistent for devices. * Note: Drivers should NOT use this function directly, as it will break * platforms with CONFIG_DMABOUNCE. * Use the driver DMA support - see dma-mapping.h (dma_sync_*) */ void dma_cache_maint(const void *start, size_t size, int direction)
<provel_>
I suppose is /usr/lib/xorg/modules/drivers
<provel_>
thanks bye
provel_ has quit [Remote host closed the connection]
<WarheadsSE>
hansg: yup. running stable on an a13-olinuxino-micro atm.
fragmint has joined #arm-netbook
<WarheadsSE>
mnemoc: Okay, re: micro/pmu I'm aware of that one, but I cheated and snagged the one out of the debian prelim, bin2fex'd it
* fragmint
raises his hand
provel_ has joined #arm-netbook
<provel_>
hum, ssvb driver give me issue : 61.186] (EE) FBDEV(0): FBIOPUTCMAP: Invalid argument
<provel_>
repeated over and over ..... framebuffer mali r3p1
<fragmint>
is this channel for those wm8650 netbooks?
<ssvb>
provel_: it's not an issue
<Turl>
fragmint: it wouldn't be off topic, but I haven't heard of many people with those in here
<ssvb>
provel_: btw, framebuffer mali will not work
<fragmint>
chromebooks?
<rz2k>
fragmint: for everything ARM-linux-embedded related
<fragmint>
yaaay!
<Turl>
fragmint: most people in here work on allwinner devices, but the channel isn't really limited to them
<ssvb>
provel_: I don't think you even need X server for framebuffer mali
* WarheadsSE
waves to fragmint
<fragmint>
I got two literati ereaders from my local microcenter for $9/ea and am fiddling with them. they have specs similiar to most wm8650 netbooks. (800x480 color display, 600-800mhz armv6, 64mb/ram, 256mb internal SD + ext SD slot, built in keyboard/wifi)
<WarheadsSE>
armv6l w/ vfp, btw
<Turl>
9USD each?
<fragmint>
yep
<Turl>
talk about cheap :O
alcides has quit [Quit: To be continue...]
<provel_>
ssvb : I must remove all the maly drivers, even fb, for your driver to work?
<provel_>
I get that error repeated multiple times and then no X appears
<fragmint>
the updates for them are just a .tgz file with the rootfs. It just flashes those to internal SD on the update process. I slapped netcat from an old version of arm angstrom in there and got a remote shell going.
<provel_>
and xorg is not anymore running
<Turl>
fragmint: similar to how kindle does it
<RaYmAn>
and kobo
<provel_>
configuration has shadowFB-false row commented
<rz2k>
provel_: you might step into the same issue I have right now on my device
<rz2k>
I cant launch xorg at all
<provel_>
ok same here
<ssvb>
provel_: just use whatever works for you, framebuffer mali should be already fast, it is not like X server is going to accelerate anything
<rz2k>
it just segfaults or locksup harder
penguin42 has joined #arm-netbook
<provel_>
mine freezes ... and log state only that issue
penguin42 has quit [Client Quit]
<fragmint>
I'd like to get another distro running on there but at the moment I'm just fiddling with it until I can get a serial cable with level shifter so I can image the darn thing
<fragmint>
I'm trying to get a shell to display on the ereader itself but chvt/openvt seems to have no effect. I don't see any traces of X on there. was wondering if anyone had any suggestions
<provel_>
removed
<rz2k>
also we have r3p0 drivers in our kernel, so you need x11 mali libraries with same version
<Turl>
application writing directly to frambuffer?
<provel_>
also addedd 99-fbdev ... I can remove it too?
<fragmint>
was that to me or rz2k?
<Turl>
to you fragmint
<fragmint>
believe so. I see directfb/qt on there and thats about the only thing related to graphics I can find
<Turl>
well, if you see directfb then it's likely using it
<penguin42>
lkcl: Have you looked at 'electric' for schematic wrangling; I know it's primarily designed for chip stuff - don't know what it's PCB skills are?
<ssvb>
rz2k: does X work for you in any other configuration (different resolution, VGA output, ...)?
<provel_>
ssvb: maybe this helps, my log latest row is: (==) FBDEV(0): DPMS enabled
eFfeM has quit [Quit: Leaving.]
<ssvb>
provel_: first try to get the standard xf86-video-fbdev driver up and running, if it fails, then the problem is not in my code
<provel_>
ssvb: I was trying to get it up and running
<provel_>
ah standard one... ok
<ssvb>
provel_: xf86-video-fbdev should be normally compatible with pretty much any system unless something is really wrong with it
<aexl1>
lo. is there a way to boot android from sd card on a10?
<Turl>
it should be doable, yes
<aexl1>
Turl: any links?
<ssvb>
provel_: yes, the standard one, and right now I'm preparing a proper release of the driver with its own name (in order to avoid any confusion with the standard xf86-video-fbdev), so you might actually want to wait until it is ready
<Turl>
not that I know of, but it shouldn't be any much different that a normal distro
<Turl>
adjust the mountpoints on the init.sun*i.rc file, repackage ramdisk and then have uboot boot kernel with ramdisk
<rz2k>
ssvb: nope
<rz2k>
I tried vga and hdmi 720p
<ssvb>
rz2k: did bisecting reveal anything useful?
provel_ has quit [Remote host closed the connection]
<aexl1>
Turl: how do i repackage ramdisk?
<rz2k>
need to setup NFS and testing scripts
<Turl>
the normal way, cpio + gzip
<rz2k>
I will do it later, someday :)
<aexl1>
Turl: thx. sounds good. will take a look!
<aexl1>
Turl: there should already be a ramdisk in ICS right?
<Turl>
yes, android usually contains the ramdisk inside the boot image (boot.img)
<Turl>
you can extract it using the 'unpackbootimg' tool
<Turl>
that will get you the kernel and ramdisk out of the boot image
<aexl1>
Turl: thx. brb.
<ssvb>
rz2k: btw, this is how I'm building my kernel for Mele A2000 - https://gist.github.com/4573851 (if you don't have NFS, you can make a tarball with the modules, scp them to the device and use ssh to extract, all from the same script)
<ssvb>
rz2k: no SD card swapping necessary, that is until the system fails to boot :)
<ssvb>
rz2k: bisecting is reasonably fast in this setup
<rz2k>
I will try it
<rz2k>
want to finally get NFS since we have u-boot with network
<aexl1>
Turl: can't find boot.img on nanda. but i will look into this further. thx again!
<Turl>
aexl1: boot.img is on a separate partition
<Turl>
nandc iirc
<Turl>
cat /dev/block/nandc > boot.img
<aexl1>
Turl: oh then i have to connect again. :D
<ssvb>
rz2k: do you mean u-boot can load the kernel over ethernet now? sounds good
<rz2k>
it should, what stops that if we have network :p
<Turl>
it can
<Turl>
I do it :D
<rz2k>
If I remember u-boot functions correctly
<ssvb>
Turl: great, is there any wiki page explaining the details?
<Turl>
I tftp 3.8-rc4 with devicetree
<Turl>
ssvb: not really, but it's not hard
<rz2k>
if it uses standard u-boot commands, it should be simple
<Turl>
first you need to set up a tftp server on your pc
<Turl>
to serve the kernel/ramdisk/whatever you need loaded
<Turl>
then build the latest u-boot-sunxi for your device and flash it on a card
<Turl>
then all you need are proper rules for your environment, I can share you the ones I have on my cubie
<Turl>
I dropped them on uEnv.txt, works like a charm :)
<Turl>
(and well then you need an ethernet cable to plug your device on the same network as your tftp server)
<WarheadsSE>
mnemoc / turl .. would I need to compile without the AXP in the kernel for the micro, since it isn't there, or will it safely ignore it.
<ssvb>
Turl: ok, thanks a lot
<Turl>
WarheadsSE: theorically it should ignore it
<WarheadsSE>
good
<Turl>
WarheadsSE: on practice, however.. you never know :P
<Turl>
WarheadsSE: so try and see if you get random cpu eaters or other weird stuff/crashes
<fragmint>
don't think they predate the mk802 but as soon as it came out those a5 ones started showing up on some sites when you searched for tv sticks on price low to high
<fragmint>
on the subject of tv-sticks. be careful if looking for mk808/mk802v3/ug802/etc. bunch of clones have been popping up everywhere in the past few weeks
Quarx has quit []
<rm>
it is silly to talk about "clones" where there might be not one true "original device" in the first place
pwhalen has quit [Remote host closed the connection]
gimli has quit [Ping timeout: 245 seconds]
pwhalen has joined #arm-netbook
hste_ has joined #arm-netbook
leowt has joined #arm-netbook
<leowt>
hi
<leowt>
what are the current issues with 3.4 branch?
<Turl>
usb gadget is not ported yet
freakazoid0223 has joined #arm-netbook
tinti_ has quit [Quit: Leaving]
* fragmint
raises his hand
<fragmint>
busybox with static libraries/uclibc... waste of mem?
<Turl>
depends on what you're going to do with it, what is it going to run on
tinti has quit [Remote host closed the connection]
tinti has joined #arm-netbook
simosx has joined #arm-netbook
simosx has quit [Changing host]
simosx has joined #arm-netbook
leowt has quit [Quit: leowt]
<aexl_>
gnexus build a 'static' busybox = 2.1MB
<specing>
thats huge?
<specing>
My whole coreutils install is 4M :)
<specing>
(Though I have to admit busybox contains much more functionality)
<fragmint>
oh sorry Turl, that little ereader (666 bogomips ArmV6, 64mb/ram, 256mb storage built in)
<fragmint>
I assume my best bet is to build a loaded busybox/rootfs, throw it in qemu and see if things work then flash it to the thing and spend the next week figuring out why stuff doesn't work on it
gimli has joined #arm-netbook
<Turl>
depending on what use you're going to give to busybox, it might be better to make it static and avoid the headache of mismatching libs and stuff
hansg has quit [Remote host closed the connection]
<Turl>
64mb is not that little
<Turl>
it's not a lot either though
<fragmint>
pretty much what I was thinking
<fragmint>
I would like to see if I couldn't get VNC going on it
<penguin42>
should be plenty for that
<fragmint>
maybe an NES emulator or doom
<penguin42>
since my 1st x86 Linux box had 16MB I think, and I did used to run ARM Linux on 4MB - although that was a long time ago :-)
<fragmint>
busybox has an option to compile doom built in... and it auto-fetches the shareware WAD!
<aexl_>
fragmint: :D
<Turl>
fragmint: lol
<Turl>
aexl_: system partition has init.sun4i.rc? o.O
<aexl_>
well / is system right?
<Turl>
/ is ramdisk
<Turl>
/system is system
<aexl_>
ooooh :D
<fragmint>
yay, this reminds me of the first time I built a kernel
<Turl>
run 'mount' to see the list
<fragmint>
I have no idea what I'm doing... I'm just hitting options like "oooh, that looks like i might want it someday"
<aexl_>
Turl: right but it does not say ramdisk ;)
<aexl_>
fragmint: :D again
<Turl>
fragmint: haha
<aexl_>
it's a little bit confusing since system is nandc is boot.img is ramdisk
simosx has quit [Quit: Αποχώρησε]
TheLarch has joined #arm-netbook
TheLarch has joined #arm-netbook
TheLarch has quit [Changing host]
wingrime has quit [Ping timeout: 248 seconds]
SouL_ has joined #arm-netbook
<fragmint>
have any of you folks tried using amazon ec2 for building rootfs's?
SouL_ has left #arm-netbook [#arm-netbook]
TheLarch has quit [Quit: <puff of smoke>]
L84Supper has joined #arm-netbook
<techn_>
aexl_: could you send patch to sunxi-media-create.sh? :)
<specing>
fragmint: What, you don't have a real computer?
<fragmint>
no. =(
<fragmint>
I mean I have a computer, its just not horribly quick or anything
newbie has joined #arm-netbook
<specing>
well my 4 years old core2duo is enaugh to keep 4 Gentoo systems updated (2 compiled in a qemu-arm chroot, other in a i586 chroot) while Im playing games
newbie is now known as Guest57742
<specing>
so not an excuse!
<specing>
just let it run overnight
newbie1 has joined #arm-netbook
<specing>
(if your current computer is a pentium 4 then I don't know what is keeping you from buying something newer)
hp__ has quit [Ping timeout: 264 seconds]
TestModule has joined #arm-netbook
Guest57742 has quit [Ping timeout: 264 seconds]
newbie1 is now known as hp__
<aexl_>
techn_: if i'm ever able to complete the mess i started. :D
<WarheadsSE>
The rest of that is, afaict, already done, only that CONFIG_AXP209_POWER needs to be jumped.
<Turl>
no idea if there's any recommended style for that stuff on uboot
<Turl>
WarheadsSE: hno might know
<techn_>
WarheadsSE: why not remove CONFIG_AXP209_POWER from general config and add that to boarch which has it
<WarheadsSE>
mm
<WarheadsSE>
because, AFAICT, I am not seeing one config/board
<WarheadsSE>
eg make a13_olinuxino works, but I see no matching configs/*.h for it, of any kind.
<Turl>
WarheadsSE: is the og mk802 supported on uboot now?
<Turl>
that one didn't have axp I think
<WarheadsSE>
is it an a13?
<techn_>
a10
<techn_>
hmm.. it should auto detect if there is axp
<WarheadsSE>
I see mk802/mk802ii board files
<Turl>
WarheadsSE: there should be a includes/config/sun5i.h
<WarheadsSE>
yeah, i am looking at it.
<Turl>
well, you could make a sun5i-noaxp.h with #..... #include "sun5i."..#undef CONFIG_AXP209_POWER ..#....
<Turl>
then use sun5i-noaxp:...... instead of sun5i:...
<Turl>
but it still feels hacky
<WarheadsSE>
right
<WarheadsSE>
although, that would sort-of solve the issue.
<WarheadsSE>
all the ways seem hackish.
<techn_>
hmm.. I think if (!power_failed) clock_set_pll1(1008000000); partly a bug :/ usually when there is no axp voltages are freezed high enough for higher clocks
<Turl>
yeah but if it there's an axp and it fails you don't want to raise the clock
<techn_>
yeah.. partly :)
<Turl>
WarheadsSE: what's the trouble though? I'm not seeing any problem that could come from the lack of axp
<Turl>
all the axp function would fail, your clock would be left slow
<Turl>
but it shouldn't fail
<Turl>
unless the axp functions are fatal
<WarheadsSE>
idk, I am just looking at the modifications, considering what happened with AXP not being present causing massive kernel failures.
<WarheadsSE>
I've made the minor adjustment for RAM to ro the dram.c & adjusted the name in board.c of board/allwinner/a13_olinuxino_micro/
<WarheadsSE>
and boards.cfg of course.
<techn_>
Turl: mk802 users would propably want this CONFIG_NO_AXP feature
<Turl>
does their spl work as of today?
<WarheadsSE>
I could see anyone who has a board that craps
<WarheadsSE>
(because of no AXP0
<WarheadsSE>
going the route of sun5i_noaxp & include/undef is looking more likely a decent idea.