<RaYmAn>
mnemoc: are there any reasons for not merging sun3i, sun4i and sun5i? I mean, other than the devices available, they seem pretty identical? - the minor differences could easily be #ifdef'ed
nibb__ has quit [Read error: Connection reset by peer]
ZaEarl has joined #arm-netbook
<mnemoc>
RaYmAn: their reason is that they provide only partial sources to each manufacturer
<RaYmAn>
mnemoc: I more meant as far as we are concerned :)
<RaYmAn>
e.g. => 3.4 branch
<mnemoc>
RaYmAn: our goal is to merge them, first step is plat-sunxi, but human resources are VERY limited
<RaYmAn>
yeah, I'm aware
<RaYmAn>
I'm obviously asking these questions because I do want to try and find time to pitch in =P
<mnemoc>
in 3.4 I'm trying to clean the script handling, then gpio, plat-wide
<mnemoc>
RaYmAn: many drivers also need to be unified, specially the disp beast
<RaYmAn>
ok, I might take a stab at comparing mach-sun5i to mach-sun4i and merging sun5i specific stuff (with suitable #ifdefs).. Even if it's -> mach-sun4i, it's still a step in the right direction
<RaYmAn>
yeah
<mnemoc>
sun4i and sun5i are easier to merge than sun3i, which was abandoned by allwinner before 2.6.36 (our older code available)
<RaYmAn>
yeah :/
<mnemoc>
but looking carefully sun3i is also very similar to it's bigger brothers
<mnemoc>
RaYmAn: the greatest help at the moment would be a new sunxi kms driver
<mnemoc>
libv suggested using omapdrm as template^Winspiration
<RaYmAn>
yeah - I suspect I'm nowhere near familiar enough with sunxi to do that
<RaYmAn>
:P
<mnemoc>
RaYmAn: no one is :)
<RaYmAn>
I suppose :P
<mnemoc>
a less agressive step would be to flatten the sun[45]i drivers
<mnemoc>
because they currently use layer after layer of some OS-agnotisc fybar
<mnemoc>
fubar
<mnemoc>
which we need to peel off
<RaYmAn>
sounds like tegra in 2.6.32 variant :P
<mnemoc>
if you can remove the layers and wrapper it will be much simpler to add fbcon support
<mnemoc>
and to merge sun4i and sun5i video dirvers
<mnemoc>
and there is no special knowledge needed for that
<mnemoc>
at the end, you'll know enough to implement the sunxi kms driver :)
<mnemoc>
deal? :)
<RaYmAn>
lol
<RaYmAn>
sure, i'll be done in around 3 years
<RaYmAn>
:P
<mnemoc>
sooner someone begins, sooner we'll get it done :)
<RaYmAn>
yeah - I guess I should start with actually making current source work on my a13 devices
<mnemoc>
we have a sun5i branch awaiting for such work
<mnemoc>
it un-removes sun[34]i, and un-removes the code Tsvetan got binary-only
<mnemoc>
also turned random chinese encodings into utf8, and `dos2unix`
<RaYmAn>
lol
<mnemoc>
it's 3.0.8, but much cleaner and complete that the mess on their SDK
arete74 has quit [Quit: leaving]
<mnemoc>
but needs to be merged into our main branch
<mnemoc>
we really need to merge wip/lichee3-sunxi/import-sun5i :)
<RaYmAn>
I guess
<mnemoc>
all yours :)
<RaYmAn>
I'm just wondering if it makes more sense to figure out differences, then just adapt sun4i to work?
<mnemoc>
i'm more fond to "Baby steps" (easier to debug breakages), but it's up to you
<mnemoc>
the reason to keep the lichee3 branch alive is because it mostly remains compatible with the close tree, and so people can use it build modules without replacing the whole world
<mnemoc>
but I'm probably over-caring about that use case
<RaYmAn>
ah, I guess that sorta makes sense
<RaYmAn>
but I'm not sure it warrants further development in that branch =P
<mnemoc>
been a different mach- it can be shared
<RaYmAn>
it just seems a bit silly to have to maintain two branches that are mostly identical
<mnemoc>
also as that's our starting point for sun4i support, it's easier to diff with whatever allwinner releases
<mnemoc>
no development goes there, only importing releases/leaks to then forward port
<mnemoc>
but as I said, I don't have an strong defense for it, and if it's an obstacle to get help, f* lichee3
<RaYmAn>
My issue is probably not having a clue how different sun4i and sun5i is
<mnemoc>
at lichee3 level you can diff straight away
<mnemoc>
in the main branch it's harder because sun4i has been improved
<mnemoc>
at 3.4 level, it's even harder
<mnemoc>
because of api changes, etc
<mnemoc>
and plat-sunxi work
<mnemoc>
that's why I find value on the lichee3-sunxi branch
<mnemoc>
while the official 2.6.36 helps us find the diffs between sun3i and sun4i
<mnemoc>
lichee3 does the same for sun4i vs sun5i
lundman has joined #arm-netbook
<mnemoc>
even if development happens in 3.0-v2 or 3.4
* rm
compiled a kernel with no CPUFREQ and got an entirely new SD card
<rm>
another evening of MK802 "hacking"
<RaYmAn>
mnemoc: oh, definitely - it's useful for comparing of course :)
<mnemoc>
RaYmAn: but it's also important to consider the A13 SDK doesn't only include A13 related improvements, it also includes improvemnts that affect A10 but were made after Qware's release
<mnemoc>
then didn't destroy all sun4i related stuff there
<RaYmAn>
is stuff actually *improved*?
<mnemoc>
i think so
<mnemoc>
but I don't remember the diff in detail
<lundman>
sup
Almamuetya has joined #arm-netbook
<mnemoc>
wb lundman
arokux has quit [Remote host closed the connection]
lerc has quit [Read error: Connection reset by peer]
Almamuerta has quit [Quit: Page closed]
IEF has quit [Read error: Operation timed out]
IEF has joined #arm-netbook
arokux has joined #arm-netbook
Almamuerta has joined #arm-netbook
Vayun has joined #arm-netbook
<rz2k>
rm: I have ly-f1 that doesnt boot from SD card anymore after couple of reflows, I can provide logs from u-boot with signs of mmc controller fail
<rz2k>
just saw your message at mailing list about mk802 that doesnt boot from sd
<rm>
yep
<rm>
but
<rm>
does SD work in android for you?
<rz2k>
yes
<rm>
I wonder how that's possible
<rz2k>
uboot fails absolutely randomly: xx1 mmc fail, setenv is undefined, fatload is undefined and so on. sometimes uboot makes it to kernel load and then I have freeze right after "starting kernel: ok"
<rz2k>
I've tried 3 sd cards
<rz2k>
(android from nand works)
<rz2k>
maybe I need to test memory, this ly-f1 didnt work at all until I reflowed DDR
piezo has quit [Quit: quit]
piezo has joined #arm-netbook
nibb has joined #arm-netbook
piezo has quit [Quit: quit]
piezo has joined #arm-netbook
croys has joined #arm-netbook
<rm>
sooo with this one the built in android kernel only sees 512MB RAM
<rm>
despite it having that "DDR3/1GB" sticker on the bottom
<DonkeyHotei>
[Wed 2012-07-25 11:54:53 PM PDT] <DonkeyHotei> this line is never executed, and i want to know why
<mnemoc>
DonkeyHotei: confirmed, I only get drm's greeting on `modprobe mali_drm`. can you please open an issue so I don't forget to look at it tonight?
<DonkeyHotei>
i do modprobe and still don't get the greeting
<mnemoc>
<6>mali: use config clk_div 3
<mnemoc>
<6>mali: clk_div 3
<mnemoc>
Mali: mali clock set completed, clock is 320000000 Mhz
<mnemoc>
Mali: Mali device driver loaded
<mnemoc>
<6>[drm] Initialized drm 1.1.0 20060810
<mnemoc>
please create an issue with the details
<WarheadsSE>
^ i get that
<DonkeyHotei>
that line i get on boot
<DonkeyHotei>
i don't get the mali drm line
<mnemoc>
DonkeyHotei: yes, I agreed. please file a ticket. can't look at it right now. will do tonight... if you file the ticket
gimli has joined #arm-netbook
<DonkeyHotei>
ticket on your github, you mean?
<mnemoc>
yes
arete74 has quit [Quit: leaving]
arete74 has joined #arm-netbook
Almamuerta has joined #arm-netbook
<DonkeyHotei>
i can't seem to do anything on github without signing up
<RaYmAn>
that is decidedly correct
<RaYmAn>
:P
<mnemoc>
DonkeyHotei: you can remind me in 5h too
<DonkeyHotei>
the name github implies the use of git is required; if they also offer svn access, i'm there
<lukestuts>
The u-boot seems to work fine, it gets to
<lukestuts>
Starting kernel ...
<rm>
soooo
<lukestuts>
But then hangs.
<rm>
I am fighting here with an MK802 that doesn't boot from an SD card...
<lukestuts>
At first, I thought it was a 512MB vs 1G ram thing
<lukestuts>
But now I am not so sure
<lukestuts>
Ah, that sounds like a similar thing
<lukestuts>
We have a Large Number of A2000s here
<DonkeyHotei>
they all have 512MB
<rm>
I have 2 MK802, 512MB and (supposedly)1024MB
<lukestuts>
Approximately 1/3 of them have this problem
<rm>
2nd one doesn't boot from SD
<rz2k>
post boot0/1 output
<rm>
rz2k, maybe I am too stupid or LiveSuit is too chinese for me
<lukestuts>
Is boot0/1 output the output from UART?
<rz2k>
yes
<rz2k>
first lines before u-boot and etc kicks in
<rm>
"when Boot the PVP or cannot discen whether (power ofF), colse the PVP's power"
<lukestuts>
ok, shall I put it on pastebin or similar?
<rz2k>
lukestuts: then, if it goes to u-boot you will see whats happening there, it should execute initial stuff like loading script.bin to exact address and etc. if this stuff fails u-boot will tell you
<rz2k>
paste that to pastebin
Almamuerta_ has joined #arm-netbook
gimli has quit [Quit: Verlassend]
<rm>
oh, so I had to insert it *while* in livesuit...
<rz2k>
rm: common tactic with livesuit: select firmware, press flash without anything attached, you will see obvious message about that nothing is attached, now press FEL button and attach device
<rz2k>
in 3-6 seconds you will see prompt about mandatory format
<rz2k>
dont close error message, wait 5-8 minutes
<rz2k>
flashing will go background
<lukestuts>
rz2k: Is there a way to use LiveSuite with the A2000?
<rz2k>
lukestuts: now paste whats happening if you boot from SD
<rz2k>
lukestuts: yes, all sunxi devices use livesuit/phoenix protocol called eFex to flash internal nand
<lukestuts>
Will do, just reimaging my card to make sure it's the right image!
<lukestuts>
So I just have to have the A2000 on the same LAN?
<lukestuts>
Got to say, these are great things
<rm>
it's not LAN based, you need to connect it to a (Windows) PC via USB
<lukestuts>
Ah, so I need to make a USB connector for the internal USB inside the A2000?
<rm>
I think one of the external ones will work
<lukestuts>
USB A-A?
* rz2k
pokes mnemoc
<rz2k>
I dont have mele so I cant say anything about what port you need to use to connect it to livesuit
ka6sox-away is now known as ka6sox
<mnemoc>
the OTG of the mele is inside, a header but no connector
<rz2k>
lukestuts: you can reball/reflow small BGAs with air, big ones like northbridges or CPUs will need infra-red and preheaters beneath the board
<rz2k>
(infrared station is $1k + minimum)
<specing>
:O
Quarx has quit []
<lukestuts>
rz2k: thanks :)
<rm>
lukestuts, please keep us updated if you solve this
<rm>
you could be my savior if your problem turns out to be exactly the same as mine :p
<rm>
or the other way round
<rm>
in any case, you should subscribe to the mailing list
<lukestuts>
rm: will do, I have quite a few not-booting boards!
<lukestuts>
Got another not-working one - this stops at <5>Linux version 3.0.36-t1+ (tobyc@adonai) (gcc version 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu5) ) #15 PREEMPT Tue Jul 3 17:45:51 EST 2012 CPU: ARMv7 Processor [413fc082] revision 2 (ARMv7), cr=10c5387d every time over UART
<lukestuts>
but boots Android from nand every time!
<rm>
from the non-sd-booting mk802, got the dmesg from android
<hackandfab2>
I'm thinking on ordering a telechips mini board (like mk802). I've seen they're distributing the sources. Has anyone tried it?
<traeak>
telechips isn't OS friendly
hackandfab2 has quit [Ping timeout: 252 seconds]
hackandfab2 has joined #arm-netbook
RITRedbeard has quit [Read error: Connection reset by peer]
<rm>
A10 at least boots from SD
<rm>
sometimes.
<rm>
so my guess is it's an u-boot level problem
<rm>
related to hardware versions
ka6sox-away is now known as ka6sox
<hackandfab2>
so no one has tried them?
<mnemoc>
20:28:09 < traeak> telechips isn't OS friendly
<hackandfab2>
why?
<traeak>
send them an email and ask them why
<traeak>
their dual cortex a5 looks interesting...that's fine if you only wnt to run android...an android that would probably not get updated
<rm>
I wonder if it's possible to put my own kernel to nanda
<mnemoc>
rm: sure
<rm>
aaaaand then have it mount the SD card as the root fs
<rm>
not nand*
<mnemoc>
sure too
<rz2k>
you can even make up your own img for livesuit
<rm>
maybe that's the (only) way to somehow boot these
<rz2k>
with dual boot
<rm>
lukestuts, ^
<rz2k>
there is kitchen for a10 by sztupy on XDA
<rz2k>
and redscorpio's unpacker/packer on 4pda
<mnemoc>
DonkeyHotei: thanks
<rm>
well, I think I have write access to nanda while booted in Android, no?
<rm>
could be the easiest way
<DonkeyHotei>
mnemoc: wow, that was fast
<mnemoc>
DonkeyHotei: would you be kind and add some references to the line that's not running?
<WarheadsSE>
I read arch linux arm ?
<rm>
or maybe
<rm>
u-boot is used in the nand too, right
<rm>
if so, then maybe should try copying that one to SD
<mnemoc>
nand's uboot doesn't know to read mmc
<mnemoc>
mmc's uboot doesn't know to read nand
<WarheadsSE>
that tidbit is annoying
<rz2k>
why allwinner did that, by the way?
<WarheadsSE>
why make it easy?
<rz2k>
lol
<mnemoc>
rz2k: support each thing consumes scarse KBs
<mnemoc>
nand support is too fat to allow anything else
<WarheadsSE>
que? rz2k | thats renesas emma ev2 loading archlinux arm
<hackandfab2>
Ok traeak I've seen that the kernel telechips has put on their web is for android only :(
<rm>
should build a kernel which would hardcode "root=/dev/mmcblk0p2", then put it to nanda, then try booting >_<
<rz2k>
WarheadsSE: I'm just playing with my old tablet, Linaro somehow fails to load there (init crashes), so I've decided to run alarm there. I've used omap rootfs
<WarheadsSE>
oic, whats the chip in that?
<WarheadsSE>
rm, lol, that might work..
<rz2k>
Renesas EMMA EV2 Dual core Cortex-A9 + powervr SGX-something
<rz2k>
512MB DDR2
<WarheadsSE>
n701?
<rz2k>
widely known name of this tablet is Rena3
* WarheadsSE
dig
gimli_ has joined #arm-netbook
<traeak>
another powervr
<traeak>
heh
* hackandfab2
away
gimli has quit [Ping timeout: 272 seconds]
<WarheadsSE>
eww at clock speed, but hey
<WarheadsSE>
ryo kernel
<rz2k>
you will be surprised, but it has really good response time in android
<rz2k>
and I had no slowdowns or stuttering in alarm too
<rz2k>
Smallart sells uHost stick with this SoC, by the way.
<rz2k>
their A10 one is not the only one.
<WarheadsSE>
Well if you ever feellike putting a guide together for it
<WarheadsSE>
you know where to put it
<WarheadsSE>
whats the kernel version?
<WarheadsSE>
and how bastardized is it
<rz2k>
there is three working kernels, 2.6.29, 2.6.37 and...
<WarheadsSE>
k, but i dont have a device I can do anything with it, so I can't put a official package together :p
<WarheadsSE>
ls
<rz2k>
dont worry, I'm just playing with it until I will get something from this era, support for android for ev2 ended on 2.2 :/
<WarheadsSE>
heh
<rz2k>
only interesting thing about this SoC is that it boots 2.2 in 10 seconds (I've measured) and it is ice cold. even in 3d games I barely feel it warming up.
hackandfab2 has quit [Ping timeout: 250 seconds]
P4R4N01D has joined #arm-netbook
<DonkeyHotei>
[Thu 2012-07-26 11:44:41 AM PDT] <mnemoc> DonkeyHotei: would you be kind and add some references to the line that's not running? <----- no lines are running; from what i can tell, nothing in the module is ever executed
<CIA-122>
rhombus-tech: hands master * rbcc803177d24 /sandbox.mdwn: trivial edit to see if git update is now happeneing
ka6sox is now known as ka6sox-away
Turl||webchat has joined #arm-netbook
<Turl||webchat>
mnemoc: ping
lerc has joined #arm-netbook
<Turl||webchat>
mnemoc: latest 3.0-v2 doesn't seem to boot at all
<mnemoc>
Turl||webchat: booted for me this morning...
<WarheadsSE>
hmm, havent tried it yet :p
<Turl||webchat>
mnemoc: hm...
<mnemoc>
WarheadsSE: nothing important has changed
<mnemoc>
Turl||webchat: on your 1GB device booting from nand check you have the ignore atag_mem option enabled
<Turl||webchat>
mnemoc: yeah looks like off >.<
<Turl||webchat>
mnemoc: it shouldn't crash though
<mnemoc>
Turl||webchat: if your evil bootloader doesn't pass atag_mem info you'll be doomed
<Turl||webchat>
it should be like 'if(no atag mem?) try_evil_hax_from_aw()'
<mnemoc>
Turl||webchat: could, assuming we care about people without a decent bootloader :p
<Turl||webchat>
I'll switch the BL if there's one that can boot boot.imgs from nand :)
<mnemoc>
Turl||webchat: file an issue and I'll add the optional fallback this weekend
<mnemoc>
maybe tomorrow morning
<mnemoc>
Turl||webchat: but file the ticket and argue it, many people considers the evil hack should be entirely removed