<Nyuutwo>
ZetaNeta: or when you want train your muscles you can get lead acid battery like from UPSes (cheaper, charger is simpler and not so much fire hazard)
<Nyuutwo>
ZetaNeta: when you connect paraell lead acid batteries then can be high current when you connect
<ZetaNeta>
yeah... but my walks are like... ~9 hours long
<Nyuutwo>
get a backpack ;p
<spv>
you can put the batteries in parallel if you're super careful
<ZetaNeta>
"Awesome"
<ZetaNeta>
Exactly in time i was going to grow a mustage and a beard XD
<ZetaNeta>
Gonna get a muscle mountain... No one is going to recognize me
<Nyuutwo>
spv: nah, you need only automotive bulb for connecting it first time and don't use them independly
<ZetaNeta>
so, when 1 runs out... Do i need to "init 0" the cubie? Or there is a very smart way to do this?
<Nyuutwo>
ZetaNeta: connect them parrael and don't give a shit
<ZetaNeta>
"Dats mah job"
<ZetaNeta>
:3
<spv>
smart way to do what, exactly?
<spv>
you want to reboot the machine?
<ZetaNeta>
no
<ZetaNeta>
I mean, i am going to get 2 of those batteries
<Nyuutwo>
make enclosure for them with fuses and connectors for your electronics and charger
<ZetaNeta>
spv, and when 1 runs out... do i shutdown the cubie and connect the next 1, or there is a "way"
naobsd has quit [Ping timeout: 272 seconds]
<Nyuutwo>
you then charge each one independly
<spv>
as per Nyuutwo's suggestion, you can put the two batteries in parallel, and run for ~13 hours
<ZetaNeta>
spv, You been "whispering" about the "super careful"
<Nyuutwo>
then connect them by resistor (bulb) and let them to share charge
<spv>
Nyuutwo is right, you can make it safe by ensuring that when you put the batteries together you do it right
<spv>
as long as you always keep the two batteries together
<Nyuutwo>
when voltage on rsistor is 0 then just remove it
<spv>
and never let them get out-of-sync, voltage-wise
<Nyuutwo>
and you have your 12V 20AH pack
<Nyuutwo>
you can (dis)charge them altogether
<spv>
more succinctly, don't connect the batteries in parallel unless their voltages are exactly the same
<spv>
and we don't mean the advertised voltage
<spv>
the _actual_ voltage
<ZetaNeta>
spv, Can you gime the example of the sound, and probably screams around, if il do it "wrong"?
<Nyuutwo>
ZetaNeta: lead acid batteries are hard to kill
<Nyuutwo>
ZetaNeta: probably your connecting cables will melt
<spv>
the batteries would get hot if you connected them without the correct voltages
<spv>
and so would the cables, yes
<spv>
Then again, there's a reason why lead acid batteries are used in cars -- they dish out a lot of current
<ZetaNeta>
spv, Except turning dead and droping melted metal on my precious floor, anything else?
<spv>
so the cables will probably go first
<spv>
maybe small explosion?
<spv>
lead acid is pretty nasty
<Nyuutwo>
in this batteries acid is in some sponge
<spv>
explosion in the sense of lead acid leaking out
<ZetaNeta>
spv, define small. I saw alot of small "OHMYFACEYOUBASTARD" explosions
<spv>
if you're that worried about it, just connect a light bulb in between the two batteries any time you go out, and if it lights up, don't connect them.
<spv>
(between the two +'s)
<Nyuutwo>
I suspect just that enclosure wouldn't be square anymore
<Nyuutwo>
they have vent holes
<ZetaNeta>
spv, So basicaly it cant shoot a hot piece of its metal shell straight in my face?
<spv>
probably not
<spv>
If they have vent holes like Nytuutwo suggests, you need only worry about the enclosure and cables
<ZetaNeta>
specing, I scrolling down with TAB to spv's nick.... Specing, what you dirty bastard forgot on this holy channel?
<Nyuutwo>
i have once shorted out this kind of battery with small wire, wire was red
<ZetaNeta>
s/I/I been
<spv>
Nyuutwo: did it arc at all?
<Nyuutwo>
nah, no inductance to do so
<Nyuutwo>
it was battery which was weeery badly treated
<Nyuutwo>
3-4 years
<spv>
ZetaNeta: ?
<Nyuutwo>
and every winter was left discharged
<ZetaNeta>
spv, ?
<Nyuutwo>
and for 2-3 hours was it charged
<Nyuutwo>
after full dead
<Nyuutwo>
afair 6V @ ~5Ah
<spv>
ZetaNeta: I thought you were asking about what someone forgot on this channel?
<ZetaNeta>
spv, I just toldn my "friend" whom i suddenly found trying to send you a message, that he is not been expected here by me
<ZetaNeta>
s/is not/have not
<ZetaNeta>
specing, Still i want to call you a dirty bastard who likes to stalk people.
<ZetaNeta>
so, back to our <del>bananas</del> batteries. I guess i found what i need. So again, how do i need to connect them to "share" voltage?
<Nyuutwo>
negative to negative pole
<Nyuutwo>
firstly positive to positive trough bulb
<ZetaNeta>
just a normal light bulb?
<Nyuutwo>
they are 10Ah each and they are normally 5C so max current is 2A
<Nyuutwo>
yup
<Nyuutwo>
to be safe side we get bulb which draws 2A when connected to 12V
<Nyuutwo>
so get 12V ~24W bulb
<Nyuutwo>
afair headlight bulb is this power
<ZetaNeta>
okay
<Nyuutwo>
connect, wait and when voltmeter shows 0V (or let's say 0.1V) just remove it
<Nyuutwo>
P21W bulb for example
<ZetaNeta>
okay
<ZetaNeta>
gonna try then
<Nyuutwo>
they are dirt cheap
<ZetaNeta>
actualy, i am planning my "for new year" list
<Nyuutwo>
batteries this kind use normally 6,3mm "car" plugs
<ZetaNeta>
i am going to get a fujitsu t902, a replacement TS for a A10 tablet, and replacement screen for a A13, and a cubie or 2
<ZetaNeta>
and the last thing is what i am going to use as a "camera-mp3-computer"
<ZetaNeta>
and what yea guys having for newyear?
<Nyuutwo>
heh, I shoud get working on connecting 7" display with resistive TS to my cubie2 (just buy stanoffs, make some mounting plate for LCD to cubie baseboard and solder connector)
<spv>
heh, cubieboards still have a way to go in terms of full Linux support
<ZetaNeta>
anyone would actually buy a allwinner a10 phone?
<Nyuutwo>
but most of it works
<Nyuutwo>
but firstly I want run my tablet on arch not a bloody android (A20 based)
<spv>
heh given what I've seen thus far of the A20 I would vote against a A10 phone.
<spv>
Nyuutwo you will want to compile your own kernel, at least version 3.13
<ZetaNeta>
i knew about openmoko project, but didnt know what "dead" condition it is in before wanting it for newyear
<spv>
(which technically doesn't exist right now, only 3.13-rc5 does)
<Nyuutwo>
spv: I was thinking about 3.4 sunxi
<spv>
Well, careful about using the nand
<ZetaNeta>
and there are no allwinner smartphones which are actualy "phones"
<spv>
they re-implemented a lot of things that they should not have
<Nyuutwo>
spv: I know
<ZetaNeta>
so i thinked about making out own "with blackjack and hookers"
<Nyuutwo>
i just leave nand alone (for sometime)
<ZetaNeta>
s/out/our
<spv>
Also maybe the gigabit ethernet
<spv>
I found a cheap resistor that might also get what you want out of that light bulb
<ZetaNeta>
Nyuutwo, Hey, you are a arch user. Can you tell me, a experienced gentoo user... what the reason of the arch fetish taking more and more linuxoids everyday?
<Nyuutwo>
ZetaNeta: lol
<Nyuutwo>
I use gentoo on my laptop and on arm boards, arch
<ZetaNeta>
Nyuutwo, Really, its like a disease
bsdfox has quit [Ping timeout: 246 seconds]
<ZetaNeta>
Nyuutwo, Well, i am seing more and more "arch zombies" using it on everything
<Nyuutwo>
got zyxel nsa320, run script from archlinuxarm site and works
<Nyuutwo>
I was thinking about gentoo but compiling on SD card, setting up distcc, .... naah
<spv>
compiling on arm is still a ways off
<ZetaNeta>
Nyuutwo, I am using gentoo on a armv5, with 64 mb of ram
<spv>
the performance / watt of arm is still well below x86
<ZetaNeta>
No distcc
<ZetaNeta>
Nyuutwo, Xorg+GPE+everythingwhatahumanbeingneeds compiles in just a night
<Nyuutwo>
i don't want kill flash so fast
<ZetaNeta>
I am using a 4gb cf card
<Nyuutwo>
and I have used gentoo on 300MHz Celeron with 128MB ram
<ZetaNeta>
If il get a 8gb one... it will be awesome.
<Nyuutwo>
just forget on update for month, and you loss 2 days
<spv>
I wonder what gentoo would be like if one only used clang ...
<spv>
beyond the "can't compile Linux"
<ZetaNeta>
Nyuutwo, I am using gentoo on a Pentium 3 with 512 mb. Its my main virtualization machine
<Nyuutwo>
woh
<ZetaNeta>
It virtualizes sh*tloads of AT&T Unix instances
bsdfox has joined #linux-sunxi
<ZetaNeta>
but i am going to find a replacement OS
<ZetaNeta>
soon
<ZetaNeta>
Maybe early versions of minix. Maybe 386BSD....
<Nyuutwo>
spv: resistor is more expensive (yeah doesn't matter) and bulb has lower resistance when it is cold (it works wery well for stabilizing current)
<ZetaNeta>
Nyuutwo, So its better to cool down a bulb first?
<Nyuutwo>
ZetaNeta: nah, you just connect, and it does job
<ZetaNeta>
ok
<spv>
he's saying there's feedback that helps prevent over-current situations
<spv>
which makes it superior to bare resistors
<spv>
I like resistors just because I feel they're more useful in this context,
<spv>
for example, if you wanted to get an idea of how much current was going through
<Nyuutwo>
bulbs can be used for limiting mains current when you test unknown equipment
juanfont has quit [Read error: Operation timed out]
<Nyuutwo>
I will get soon 60MHz scope hackable to 200MHz and get info on this bloody LVDS display in tablet
Gerwin_J has quit [Quit: Gerwin_J]
Gerwin_J has joined #linux-sunxi
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
<spv>
Nyuutwo: How are you going about increasing the bandwidth of the scope?
<Nyuutwo>
just producent sells 3 kinds (60/100/200MHz) which has the same hardware
<Nyuutwo>
change one thing in rootfs (based on linux)
<Nyuutwo>
and it thinks it is 200MHz kind
<spv>
nice!!!
<Nyuutwo>
there are 2 versions of input cicut, which one has some problems with stable gain vs freq
<Nyuutwo>
but they differ in <10 resistors
<Nyuutwo>
so I need only check the response to step
<Nyuutwo>
I think to make 10MHz gen on 74HC00 switchable with some 555 in kHz range generator with 74LVC00 as driver, and get 50Ohm impedance
<Nyuutwo>
so firstly I get high slope output, and also I get a way to measure coax length
ZetaNeta has quit [Ping timeout: 245 seconds]
bsdfox has quit [Excess Flood]
<Nyuutwo>
and I read about that factory cal procedure needs 10MHz low jitter clock with high slew rate
bsdfox has joined #linux-sunxi
<Nyuutwo>
but probably I would need it firstly get somewhere calibrated
<Nyuutwo>
*checked
<Nyuutwo>
but for measuring lvds at 55MHz even without correcting it should work enough to get what isn't working
Gerwin_J has quit [Quit: Gerwin_J]
<spv>
clearly you can correct for the gain vs freq relationship in software :P
<Nyuutwo>
yup, but I don't think they made this possibility
<Nyuutwo>
and when you do this in hardware you don't loose ADC range
<spv>
depends on the input signal
<Nyuutwo>
yup, and also probes (which I will get probably 100MHz max)
<Nyuutwo>
but i have passtrough 50Ohm dummy load so I can do many other things
wolfy has joined #linux-sunxi
bsdfox has quit [Ping timeout: 246 seconds]
bsdfox has joined #linux-sunxi
t3st3r has quit [Remote host closed the connection]
t3st3r has joined #linux-sunxi
bsdfox has quit [Ping timeout: 246 seconds]
bsdfox has joined #linux-sunxi
bsdfox has joined #linux-sunxi
bsdfox has quit [Changing host]
rings_IIV has quit []
bsdfox has quit [Ping timeout: 252 seconds]
bsdfox has joined #linux-sunxi
penghb has quit [Read error: Connection reset by peer]
hramrach has joined #linux-sunxi
penghb has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
bsdfox has quit [Ping timeout: 252 seconds]
bsdfox has joined #linux-sunxi
wolfy has quit [Ping timeout: 252 seconds]
Quarx has joined #linux-sunxi
wolfy has joined #linux-sunxi
wolfy has quit [Ping timeout: 252 seconds]
n01 has joined #linux-sunxi
deasy has joined #linux-sunxi
wolfy has joined #linux-sunxi
<gzamboni>
oliv3r, is Nginx a good software ? i heard you talking about it. it seems its getting more and more used. I saw you're configuring the mail server, if you need i have a step by step documentation i did when i did set up in my server, the difference is my server is debian and not ubuntu.
<gzamboni>
oliv3r, if youre planning to move all the boxes to the cominity server pls tell me when it will be done so i can remove the dns zone and configs from my server.
bsdfox has quit [Ping timeout: 272 seconds]
bsdfox has joined #linux-sunxi
naobsd has joined #linux-sunxi
jinzo has joined #linux-sunxi
maz_ has joined #linux-sunxi
AreaScout has joined #linux-sunxi
t3st3r has quit [Remote host closed the connection]
t3st3r has joined #linux-sunxi
FR^2 has joined #linux-sunxi
<mnemoc>
gzamboni: I personally find nginx the best http server, but that can become a religious topic
<mnemoc>
gzamboni: oliv3r is working on setting up mailing lists support on the community server, and eventually that will imply moving the forwards and smtp-auth too.... but we aren't there yet!
<gzamboni>
mnemoc, nice to know about nginx. i will test it.
<gzamboni>
mnemoc, about the mail service it doenst bother me to let them in my server. anyway once you will eventually get there just let me informed :)
<mnemoc>
of course you will :)
emx has joined #linux-sunxi
bsdfox has quit [Ping timeout: 252 seconds]
bsdfox has joined #linux-sunxi
Quarx has quit []
<libv>
oliv3r: do you already know when the sunxi talk is on?
<mnemoc>
hopefully not parallel to lima's :(
<gzamboni>
oliv3r, mnemoc i supose you do backup the server content regularly
<mnemoc>
yes
<gzamboni>
nice :)
<gzamboni>
well i will try to make my soon ride a bike without the security extra whells
<gzamboni>
i have a nice sun here shining
maz_ has quit [Ping timeout: 272 seconds]
<libv>
mnemoc: not that lima is going to be much of a big spectacle like the last two years
<mnemoc>
basic mesa and android's libhardware support would be a big spectacle :)
<mnemoc>
Turl promised to help with the second.... Turl!!!
<mnemoc>
9:15 on sunday and still sleeping... that's why argentina doesn't progress
<oliv3r>
how can they have earned that kind of money
<mnemoc>
oliv3r: what did I break? :(
<oliv3r>
mnemoc: your uarts :p
<mnemoc>
probably... they all have the PCB exposed
rah has quit [Ping timeout: 246 seconds]
rah has joined #linux-sunxi
<oliv3r>
likly a broken crystal
captaini- has joined #linux-sunxi
egbert has joined #linux-sunxi
mturquet1e has joined #linux-sunxi
slopez_ has joined #linux-sunxi
tm512` has joined #linux-sunxi
rm___ has joined #linux-sunxi
netlynx has quit [*.net *.split]
egbert_ has quit [*.net *.split]
xeros has quit [*.net *.split]
tm512 has quit [*.net *.split]
mturquette has quit [*.net *.split]
slopez has quit [*.net *.split]
rm has quit [*.net *.split]
captainigloo has quit [*.net *.split]
binaryferret has quit [*.net *.split]
captaini- is now known as captainigloo
rm___ is now known as rm
bsdfox has quit [Ping timeout: 260 seconds]
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
n01 has joined #linux-sunxi
xeros has joined #linux-sunxi
<wens>
oliv3r: i have a friend that spend like 4k setting up a good environment, like power conditioners, pricy cables
binaryferret has joined #linux-sunxi
<oliv3r>
yeah but for digital transfer, getting a 500 MHz USB cable is stupid beyond means
<oliv3r>
USB won't run faster because you have better cables
bsdfox has joined #linux-sunxi
<mnemoc>
but long distance USB does
<mnemoc>
but still stupid
<oliv3r>
USB is limited to 5 meters
<oliv3r>
after that you need repeaters to amplify the signal
<oliv3r>
we're talking loss of amplitude, e.g. the dac/adc can't see if it's 0 or 1
<oliv3r>
not 'loss of quality'
<oliv3r>
e.g. going from CD quality to FM quality
<oliv3r>
anyway, here: Original LightSpeed USB Cable - $699
<oliv3r>
Paul Candy said, "I never thought that a bloody USB cable would have such an enormous impact on my enjoyment of file playback."
<oliv3r>
LightSpeed USB uses a unique topology, which we call Ultra-high Speed Differential Pair (or USDP). USDP gives LightSpeed 20 times the bandwidth of USB 2.0 high speed, which ensures optimal signal clocking. In fact, LightSpeed's astounding 10 Gigabits per second bandwidth is twice as fast as USB 3.0!
<mnemoc>
lot of stupid people lingering in kickstarter and indiegogo
<oliv3r>
now, this cable may be made out of elven piss, and do terrabit speeds
<oliv3r>
it does not matter for USB2, as that won't ever transfer beyong 480 MBit as that's the maximum of the USB2.0 spec and attached controllers
<oliv3r>
yeah, but it baffles me
<oliv3r>
or here
<oliv3r>
The goal is to design a balanced cable that can transport a digital signal from the source to the load with as much accuracy as possible. In digital cable design, we've found that most cables on the market today do reasonably well at low sampling rates. But the higher the sampling rate, the more distortion occurs in the cable because of impedance mismatches.
<oliv3r>
and these are cables that are 0.5m to 1m; 3m costs you almost double that
FR^2 has quit [Quit: und weg...]
<oliv3r>
this is a scam, pure and simple, and they made a million out of it
<oliv3r>
its redicilous how daft people can be
<oliv3r>
and that this is actually allowed to succseed
<mnemoc>
prepare a talk "bullcrap sold in indiegogo" :p
<oliv3r>
lol
<mnemoc>
30c3 has a "bullshit made in Germany" talk :p
<oliv3r>
wens: btw, that doesn't mean a 4k enviroment is complete BS ,good power conditioners, etc does cost money, and your analog cables should be proper. even digital ones should be reasonable
<oliv3r>
but a 5000 USD 1m ethernet cable, or 700 USD USB 2.0 cables, coplete bs
<mnemoc>
if there is people willing to pay it... it's a good business
<mnemoc>
otoh people with such IQ should not have such amount of money to waste... :\
<oliv3r>
lol
<mnemoc>
some somehting is failing in the model
<mnemoc>
stupid people with too much resources, smart people with too little...
<mnemoc>
and indiegogo to funnel some of it from dumbs to "smarts"
tomboy64 has joined #linux-sunxi
<oliv3r>
if you pop the stack, you 'move' data from the stack, right? not copy?
<mnemoc>
stacks are usually a double linked list.... so no moving or copying involved, just two pointers updated
<oliv3r>
in asm?
<mnemoc>
the next of the head, and the prev of the new first
<mnemoc>
no clue :p
<oliv3r>
i thought you have a stack pointer, that points to the top of the stack
<mnemoc>
my arm asm book hasn't been published yet
<oliv3r>
and if you pop, you get the values, and adjust the stack pointer
<oliv3r>
i just noticed a bit of code, where they jump to 'pop' again, so unless you have pushed something twice, you either get random values, or pop doesn't change the SP, which i would think it does
<oliv3r>
but i've documented the rest of this function a few days ago :p
<oliv3r>
not that anybody is reading so I hsould put it on the ML
<mnemoc>
€30.48 on the book depository, with shipping
<oliv3r>
packt publishing asked if i know people who'd wnat to review my book
<oliv3r>
i think they want 1 - 2 reviewers
<mnemoc>
:o
<oliv3r>
any voulenteers?
<mnemoc>
is there content to review? :p
<oliv3r>
not yet, :p
<oliv3r>
but very soon
<oliv3r>
2-3 weeks from now maybe
<oliv3r>
doing asm for fun now
<mnemoc>
I would gladly do it, but it's better if someone with a nicer CV does it... like hansg, libv or mripard
<oliv3r>
on my free sunday afternoon, go figure
<oliv3r>
it's not that deep technical of a book :p
<mnemoc>
but you still want people with something nice in their CV to support it
<oliv3r>
i suppose; i'll put a ail to the list
<oliv3r>
mailman i'll worry about soon again
<mnemoc>
hansg has libusb and redhat fellow, libv has a lot of graphic related CV and mripard is a free-electrons fellow and the official mach-sunxi maintainer
<oliv3r>
i don't think they care hugly though
<mnemoc>
maybe adding someone from denx to the pot
<oliv3r>
and it's really a bigginners book if anything
<mnemoc>
stefanro maybe
vector80 has quit [Quit: Konversation terminated!]
vector80 has joined #linux-sunxi
<oliv3r>
but it's about things like installing an OS on an SD card, simple stuff like that
<vector80>
oliv3r, mnemoc: Hi, did you be able to get A23 android SDK ? I just connect to my PC in office, it seems %100 has been uploaded succesfully...
<oliv3r>
not about hacking gpio's into a kernel driver :p
<oliv3r>
vector80: yes we did
<vector80>
Cool
<oliv3r>
i haven't unpacked it or verified it yet
<vector80>
Now, do you need others ?
<mnemoc>
can you also upload a sha1 file?
<vector80>
mnemoc: When you tell me such things, I really feel soo stupid myself :)
<mnemoc>
vector80: if they aren't in tsvetan's dir already, sure
<vector80>
tomorrow I have a chance to download 4.4 SDK, I will try my best to get it
<vector80>
Ok wait let me do it
<mnemoc>
oh, yes. upload the 4.4 SDK too!
<mnemoc>
tasty blobs
<oliv3r>
mnemoc: i'm a little supprised tsvetsan doesn't have all those sdk's yet
<oliv3r>
maybe vector80 is a bigger customer ;)
<mnemoc>
probably :)
<mnemoc>
tsvetan can't even buy A20-s at competitive prices to be able to make an A20 lime
<mnemoc>
which is sad :(
<vector80>
no no, I am not a bigger customer, I am just helping those guys, and trying helping to you, to be able to setup a bridge, maybe you also help me to understand why this 3.4.61+ kernel is changing my mmcblk0 and mmcblk0 :(
<mnemoc>
so you work for the bigger customer, ic
<oliv3r>
vector80: i read your log; and you have swapped mmmcblk0 and mmcblk1, so root=/dev/mmcblk1 should make your kernel boot. As to WHY those are swapped, i'm not sure yet
<oliv3r>
vector80: it is possible, that the driver enumerates the devices backwards
<oliv3r>
so if you connect and use all 4 mmc devices, mmc4 = mmcblk0, mmc3 = mcblk1 etc
<oliv3r>
but that's just a hunch, i haven't looked at hte mmc driver to be able o say for sure
<mnemoc>
regarding the mmcblk%u order, it's based on registration order, and different drivers work with script.bin differently... some might do it in script.bin order, others in for(i=0;....) order searching for the corresponding script.bin section
<vector80>
yes, to be able to pass "root=something" , I must use uEnv.txt, is there any other way to do this?
<oliv3r>
also, i'm not certain which kernel you mean when you say 3.4.61+, is that sunxi-3.4 or lichee-3.4
<vector80>
oh, so may be, the best way would be to compile my own kernel ?
<oliv3r>
for testing, it shouldn't matter greatly, ideally however, mmc0 = mmcblk0 and mmc1 is mmcblk1
<oliv3r>
olimex's kernel is probably sunxi kernel
<mnemoc>
vector80: we can't help you with the lichee kernels much... specially not if using a prebuilt
<vector80>
that kernel was from Olimex debian image
<vector80>
I think it is a linux-sunxi kernel
<mnemoc>
show the dmesg output
<mnemoc>
there are certain clear differences there between the sunxi and the lichee code
<vector80>
Ok, I will do that tomorrow, try to give you more clear information.
<vector80>
Just let me know,
<vector80>
except uEnv.txt,
<vector80>
is there a way to pass "root=/dev/mmcblk1" to my kernel?
<vector80>
For example, using TFTP
<vector80>
and passing bootargs manually ?
<mnemoc>
you can hardcode kernel args in .config
<mnemoc>
and decide if they accept "extension" from the bootloader or not, and how
<vector80>
clear, for this, I must checkout sunxi kernel, and compile it by myself, right?
<mnemoc>
yes
<vector80>
what do you mean by "if they accept" ? who ?
<mnemoc>
another (more elegant) way is to use uuid instead of /dev/...
<oliv3r>
that is where mmc init starts, look for those strings to see whats going on
<mnemoc>
vector80: your .sha1 exposes probably too much info
<oliv3r>
still
<oliv3r>
mmc0 should be mmc0 and mmc2 should be mmcblk1
<oliv3r>
strange
<oliv3r>
probably a very logic explanation that I don't know :p
<vector80>
Ok, tomorrow, I will do it one by one
<vector80>
firstly I will try uEnv.txt
<mnemoc>
you wrote the fex guide.... see how the fields in the mmc%d_para section are used ;-)
<vector80>
if it boots, I will need your help, how to completely wipe out my eMMC, because it currently has their 6-7 partitions filled with their linuxBSP image
tomboy64 has quit [Ping timeout: 252 seconds]
<mnemoc>
just use fdisk
<vector80>
meanwhile, I see above code, that for loop is counting one by one, shouldn't swap my mmcs...
<vector80>
how about dd if=/dev/zero of=/dev/mmcblk1 bs=1M
<mnemoc>
that will destroy the bootloader too
<vector80>
the bootloader is in TF card
<vector80>
oh you mean boot0 and boot1 ?
<vector80>
or you mean uboot ?
<mnemoc>
uboot and uboot spl and uboot env
<oliv3r>
vector80: if you can boot your image, you can dd emmc.bin i gave you a few days ago :)
<oliv3r>
vector80: but it's probably much easier to use an initramfs for your initial root
tomboy64 has joined #linux-sunxi
<oliv3r>
anyway, yes, dd /dev/zero should clear out emmc; once you have the right device
<vector80>
I am getting more and more confused... When I boot from TF card, the system doesn't need a boot0.bin and boot1.bin ?
<oliv3r>
vector80: you will never need boot0 and boot1
<oliv3r>
as you don't need nand
<vector80>
waww, if this is the case, why I need boot0 and boot1 when booting android then ?
<vector80>
I am sorry, asking too many questions, but I really want to understand, why AW created those boot0, boot1 things, and why you don't need these ?
<vector80>
And for sha1 info, how can I give you more correct info ?
<mnemoc>
livesuit uses boot0/boot1 before their hacked u-boot
<ssvb>
is there anything interesting to do in the middle of day in Brussels on Friday before FOSDEM?
<vector80>
"hacked uboot" whats that mean ? Is that history written in wiki ?
<mnemoc>
vector80: just remove the path to the tar.gz
<vector80>
ok wait
<oliv3r>
vector80: boot0 and boot1 is an ancient bootloader, from their days before linux. they simply stuck with it
<vector80>
ok, you mean, if they are ancient, that means I can use sunxi uboot for my android image too
* mnemoc
needs to find where to stay between sat-sun in brussels....
<oliv3r>
vector80: boot0 boot1 is LIKE u-boot-spl and u-boot, you either use boot01 or u-boot. we are workinging on u-boot only
<vector80>
what is SPL meaning ?
<oliv3r>
if you use their built in NAND controller, you are stuck with boot01 as our u-boot only has highly experimental nand support
<oliv3r>
mnemoc: oh i forgot that didn't i
jemk has joined #linux-sunxi
<oliv3r>
vector80: Secondary boot loader
<oliv3r>
vector80: primary is the BROM inside the chip
<oliv3r>
so u-boot is teriaery really
<vector80>
Cool, from this info, let me think:
<vector80>
1) I don't have a NAND
<oliv3r>
so you are much happier with u-boot from sunxi :)
<vector80>
2) sunxi-uboot have TFTP-NFS functionalities, AW uboot does not.
wolfy has quit [Ping timeout: 252 seconds]
<vector80>
3) I just need to find a way, how to copy sunxi-uboot to AW android SDK, remove boot0-1, create pack... Oh, for that one, I also don't need a pack image... I can make a master TF card, right?
<vector80>
One master TF card, which can copy from itself to my eMMC ?
<oliv3r>
vector80: correct
<oliv3r>
well i wouldn't say 'itself' but yes
<vector80>
That master TF card would include some partition images, and write tose images to eMMC when booted ?
<oliv3r>
one master (100 master if you want to do parallel installing) TF card, that dd-es bootloader, makes partition via fdisk script, formats partition or just dd's data, that's your call
<oliv3r>
vector80: exactly
<oliv3r>
vector80: but much much cleaner imo is to have a script that formats, and just extracts files/copies files
<oliv3r>
no need to have images imo
<vector80>
Ok, to be able to create such TF card, do I need weeks or months ?
<oliv3r>
hours
<oliv3r>
it's not rocket science :p
<vector80>
Clear, I will stay with you on monday
<oliv3r>
haha, there goes my vacaction :p
<vector80>
Until when ?
<oliv3r>
but first, we gotta see if thatemmc u-boot works that i sent you
<oliv3r>
then we gotta figure out why emmc is changing orders
<oliv3r>
and finally device a simple installer script that you will have to make perfect for your setup (e.g. reliable, error checking etc) :)
<vector80>
To be able to see it, I must be able to pass init
<vector80>
As you remember, I copied your uboot.bin to TF card, and booting was OK
<vector80>
I also tested the hardware like this: I boot from eMMC ( AW linuxBSP ), and then inserted debian TF card, and it detected that TF card as mmcblk1
<vector80>
This is also very funny, because according to schematic and fex file, TF card is mmc0
<oliv3r>
vector80: but that was u-boot for TF, u-boot for eMMC is slightly different
<vector80>
What is the difference between TF and eMMC uboots ?
<oliv3r>
vector80: so you need 2 u-boots in your archive, 1 that boots from emmc, which will be un the u-boot repo; 1 that is slightly differently configured, that you use on your (master) TF cards
<oliv3r>
vector80: maybe, for some reason I don't understand (yet) if you boot from mmc2 the kernel will always enumerate that one first; then do everything else
<oliv3r>
vector80: but i don't see this in the driver yet
<oliv3r>
vector80: u-boot can only boot from 1 mmc, so you configure it to have its boot device be mmc0 or mmc2
<oliv3r>
it's a very minor modification, that you change in boards.cfg
tomboy64 has quit [Ping timeout: 260 seconds]
<oliv3r>
(CONFIG_MMC=2 vs CONFIG_MMC=0) or just even leave CONFIG_MMC out as it defaults to 0
<vector80>
ok, currently, in TF card, there is your uboot.bin, which is compiled for TF card. That one support uEnv.txt, right?
<oliv3r>
once i can confirm CONFIG_MMC=2 works for you (that's the file i sent you) then i can push my u-boot branch
<vector80>
rz2k: that file is an Android SDK for A23
<vector80>
I am not expert like you, but as I understand from the conversations, aw is trying to unify things
slapin_ has quit [Ping timeout: 245 seconds]
bluse has joined #linux-sunxi
slapin has joined #linux-sunxi
bluse_ has quit [Ping timeout: 264 seconds]
<Turl>
mnemoc: nice trolling attempt ;)
<vector80>
I will try to upload one more file, its name is MeriiLinux.tar, this one support eMMC, and be able to generate LiveSuit compatible images which can be written to eMMC via USB OTG. I don't know if this is useful for you or not.. :(
<mnemoc>
Turl: :p
<mnemoc>
rz2k: I just added a .txt.gz with the file listing
<libv>
gzamboni, oliv3r: yes, i will talk about lima
<Turl>
oliv3r: are you writing a book? :O about what?
<libv>
i currently intend to hold the talk on sunday afternoon
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<mnemoc>
I just uploaded the pdf of the CAN controller to dl..../A20/
<netlynx>
libv, on Fosdem what track/devroom is our talk in?
dlan has quit [Remote host closed the connection]
dlan has joined #linux-sunxi
<libv>
netlynx: my own :p
<libv>
netlynx: graphics devroom, but i haven't done the scheduling yet
tomboy64 has joined #linux-sunxi
<kz1>
can anyone tell me why allwinner chose not to enable "reboot bootloader" in their version of uboot?
Turl has quit [Quit: >.<]
Turl has joined #linux-sunxi
wolfy has quit [Ping timeout: 252 seconds]
<mnemoc>
oliv3r: it seems the problem with my "drunk" usb/uart was grounding. feeding the CT from the same powered usb hub when I connect the ft232rl adapter works fine
<mnemoc>
maybe now the cheap adapters work too
<hramrach>
like grounding the adaptor on the same PSU as the board supply?
HdkR has quit [Ping timeout: 252 seconds]
<hno>
Hmm.. can someone please try sending me an email? Been too quiet today, but I can not find anything obvious wrong..
<hramrach>
didn't have problems like that but my board PSU and PC PSU are probably powered from the same wall circuit
<hramrach>
hno: holidays ;-)
<hno>
hramrach, yes, but still.. not even the usual spam flow..
Gerwin_J has joined #linux-sunxi
<hramrach>
hmm, does not bounce immediately
<hramrach>
but if you did not just get a spam from me it was eaten somewhere along the way
HdkR has joined #linux-sunxi
<mnemoc>
hramrach: yes, I grounding both from the same (good) PSU. before I was using a cheap "2A" PSU for the CT
<mnemoc>
hno: still need a mail?
<mnemoc>
hramrach: but they all come from the same wall outlet anyway... so it's odd
<n01>
mnemoc: is the guy going to mainline the CAN driver?
<mnemoc>
no idea, I didn't know about it until rz2k directed me to the thread
mturquette has quit [Quit: Lost terminal]
parabyte has quit [Quit: Leaving]
boycottg00gle has joined #linux-sunxi
<vector80>
Uploading A20-TVBOX SDK... this one is also android, but it is different than MID SDK. As you already know, Aw provides two android SDK, one is for MID, another is TVBOX. In MID SDK, HDMI resolution is only one, and you can't change it. But TVBOX SDK support multiple resolutions, and also aspect ratio is more correct for 16:9 LCD TVs...
<vector80>
MeriiLinux-eMMC support SDK upload is complete...
<vector80>
I will try to provide date information and sha1sum for these files too..
mturquette has joined #linux-sunxi
<mnemoc>
thanks!
<rz2k>
sun9i mentions in u-boot
<rz2k>
and libnand.o blob again
<Turl>
rz2k: on kernel land they seem to be unifying things, with a mach-sunxi and stuff
<Turl>
they changed the sunNi meanings even :|
<rz2k>
another .o in sunxi-vfe
<Turl>
vfe?
<rz2k>
drivers/media/video/
<rz2k>
video frontend engine probably
<Turl>
cedar?
<rz2k>
camera crap
<Turl>
ah
<rz2k>
csi & other mipi
itdaniher has joined #linux-sunxi
<vector80>
oliv3r: I forget to tell you something... Based on your suggestion, we added a STATUS LED on our board. It is connected to PB15 I think. I will clearify this point, and let you know. Maybe you can add it to boards.cfg ??
<itdaniher>
Does anyone have grievances with the USB host supportin the sunxi kernel? I've spent ~wk fighting DWC_OTG and I need something else.
<rz2k>
Turl: there are omx/stagefrighthw for cedar sources in the android folder, any chances for us to use them in native linux env. ?
kivutar has joined #linux-sunxi
<oliv3r>
mnemoc: i've seen those mails; just not the pdf
<oliv3r>
mnemoc: looks AW style doc
<oliv3r>
we shuld put it in the a10 docs section
<oliv3r>
or A20; thanks :p
<rz2k>
note that there are zero copyrights on it
<rz2k>
it can be easily a confidential crap that can lead to bad things, hope its not.
<Turl>
rz2k: they've been around for quite a bit now; nobody tried them out of android though
<Turl>
they're linked with a blob so iirc, so you'd need libhybris or sth
<rz2k>
[00:18:59] <vector80> MeriiLinux-eMMC support SDK upload is complete... - wrong chmod on the file
<rz2k>
mnemoc: ^
<rz2k>
403 from here :(
<Turl>
let me fix that
<oliv3r>
vector80: can do yes; PB15; ideally you should run the gpio toggle commands frmo u-boot to get me the decimal value
<oliv3r>
Turl: yes, packt publishing asked me to write a book about sunxi (well they wanna do it like raspberry pi, about cubieboard, but i wanna make it universal and even lean towards LIME more
<oliv3r>
Turl: but i need 1 - 2 people to revie wit
<oliv3r>
review*
<oliv3r>
i need to fix this W here
<Turl>
oliv3r: tech review?
tomboy64 has quit [Read error: Connection reset by peer]
<Turl>
oliv3r: I assume they'll have their own language reviewers?
<Turl>
mripard: ping
<mripard>
pong
<hramrach>
oliv3r: I guess I could read through it to see if there is something too obvious missing
<hramrach>
vector80: are you making a board? What kind?
<oliv3r>
Turl: no clue; they just asked for someone to review; but this is entry level stuff; beginner like content
<hramrach>
itdaniher: there were people who had grievances with the EHCI controller not reporting disconnects with the 3.4 kernel on a20.
<oliv3r>
hramrach: i guess they want someone to read if it is technically accurate
paulk-collins has joined #linux-sunxi
<Turl>
oliv3r: I can give it a read if you need then
<hramrach>
oliv3r: I guess I can verify beginner stuff. Managed to build a working image after a few weeks of effort ;-)
<Turl>
mripard: hi :)
<Turl>
mripard: I sent the clk pull to mike and I've got a tag with the dt bits if you want to pull them
<oliv3r>
then i will send your e-mail addresses to packt publishing as formal reviewers :) thanks you two
<hramrach>
oliv3r: it probably also depends on the kind of PSU. The small ones without should be doubly-insulated with no connection between the mains and the output but some have some sort of feedback which puts the output ground at a rather high potential
<hramrach>
I have a moderately sized grounded 5V metal brick which probably does not suffer from these issues
<vector80>
hramrach: Yes, it is a new board with A20 + 1Gbyte DRAM + 8Gbyte Sandisk eMMC ( no NAND ). unfortunately it incorporates some proprietary circuits, so it will not be open hardware :(
NightShade has joined #linux-sunxi
<vector80>
oliv3r: thank you, but the question is, can I toggle a GPIO pin from uboot console ? Can I set its direction and value just like /sys/class/gpio/... interface ?
<oliv3r>
vector80: sure, just use the try 'gpio<enter> to see some help
<vector80>
cool !!!
<vector80>
you are no:1
<oliv3r>
i think you can enter the PB15 value and it will return the decimal value
<vector80>
I will try it tomorrow
<vector80>
Why PB15's value is not constant for each board ?
<vector80>
I mean, I have seen this before, gpio number for android and linux images are different for the same board even..
<vector80>
Who makes the enum ?
vector80 has quit [Quit: Konversation terminated!]
vector80 has joined #linux-sunxi
kivutar_ has joined #linux-sunxi
<oliv3r>
vector80: it is; but my memory is bad
kivutar has quit [Ping timeout: 272 seconds]
<oliv3r>
vector80: and in boards.cfg we put decimal values :)
<oliv3r>
so probably you do gpio PB15 on
<Turl>
oliv3r: wanna help a noob get his cubietruck booting? :P
<oliv3r>
Turl: that's hat my book should be about
<oliv3r>
Turl: what's the problem
<vector80>
seems pb15 is 47
<oliv3r>
vector80: quite plausible
<Turl>
oliv3r: half of the time it boots from nand, the other half it prints this and hangs http://sprunge.us/UYRU
<oliv3r>
Turl: looks like bad PSU or flaky mmc
<oliv3r>
Turl: doesn't look noob issue
<vector80>
oliv3r: I checked again, it is PB16, so gpio is 48 , sorry again
<oliv3r>
ok once we get emmc u-boot working i'll build you one with gpio 48 and we'll test that :)
<vector80>
clear
<vector80>
have a good night
<vector80>
bye
vector80 has quit [Quit: Konversation terminated!]
kivutar_ has quit [Quit: Ex-Chat]
<oliv3r>
he isn't the greatest hacker around :)
<Turl>
oliv3r: I remade the card and it works better now o.O
<Turl>
# tftpboot
<Turl>
Waiting for PHY auto negotiation to complete.... TIMEOUT!
<Turl>
ENET Speed is 100 Mbps - FULL duplex connection
<Turl>
does the gmac stuff not work yet? :p
<oliv3r>
Turl: it does
<oliv3r>
i haven't pushed it
<oliv3r>
i have a few other patches stacked and wanted to push everything in one go
<oliv3r>
but vector80 needed a bit of handholding
<oliv3r>
you can use wens-v3 tree
<Turl>
I am using the v2 tree
paulk-collins has joined #linux-sunxi
<oliv3r>
Need v3
lynxis has quit [Ping timeout: 260 seconds]
lynxis has joined #linux-sunxi
<Turl>
oliv3r: v2 is broken? meh :p
cajg_ has joined #linux-sunxi
cajg has quit [Ping timeout: 265 seconds]
<Turl>
oliv3r: with v3 the autoneg still times out
<Turl>
wens: ^
AreaScout has quit []
<spv>
Anyone know what the hardware ECC algorithm is on the sunxi nand controllers?
<hramrach>
hmm, sandisk emmc. that might actually be fast
<hramrach>
on mmc1?
<spv>
If that was referenced to me, then no.
<hramrach>
no
<hramrach>
spv: but regarding ecc algorithm .. the MTD driver should have that covered, right?
<spv>
The only mtd driver thus far, that I've seen, is one that uses the hardware's default ECC
<spv>
Since Allwinner rolled their own nand hardware controller, I'm skeptical about its effectiveness.
<hramrach>
like it lets the hardware do the ecc and just lets it to tell if the block is correct or not?
<spv>
yes
<hramrach>
iirc there was that issue with extra data so presumably it does read the ecc data but I did not really look at it
<hramrach>
or what else would that extra data be?
<spv>
extra data?
<spv>
You're not referring to the out-of-band data, are you?
<spv>
or spare data region?
<hramrach>
and out-of-band data would be?
<spv>
every nand page has extra bits on the side for ECC
<spv>
for example, typically 2048-byte pages of NAND also have 64-bytes of spare data apiece
<hramrach>
so it does read the ecc data
<spv>
it should ...
ZetaNeta has joined #linux-sunxi
<hramrach>
so the problem is what?
<hramrach>
not knowing how to interpret the ecc?
<spv>
By knowing the algorithm, software-based ECC is possible
<spv>
which is important for developing the driver in the first place
<spv>
Also, as per above, I'm skeptical about the nand controller hardware.
paulk-collins has quit [Quit: Ex-Chat]
<hramrach>
it defines the number of ecc bits. Isn't there like a handful of standard polynomials for each size?
<spv>
pretty much
<spv>
but there's a little more to it
<spv>
and it depends on which version of the algorithm is used
netlynx has quit [Quit: Leaving]
<hramrach>
the nand partition table uses bit-inverted crc32c checksum or something like that
<hramrach>
they sure are inventive at AW
<spv>
and this is why I'm curious
<hramrach>
you can try to match the read data to software calculated ecc using the algorithms available in Linux. The original nand driver does not seem to have a routine for software ecc verification either
<hramrach>
not sure how many bit-perfect blocks are on a typical nand, though
<oliv3r>
freedome wise, it's better then eMMC or SD cards :)
<hramrach>
speed-wise emmc is awesome but I see a problem with it wearing down and being soldered on the board
<oliv3r>
should be the same as nand no?
<hramrach>
so adding extra SD slot for mmc1 would be probably the best
<hramrach>
oliv3r: the nand is so lame I would not lament over that
<oliv3r>
i guess (e)mmc has a very basic FTL, little wear layering, little trim if at all etc
<oliv3r>
while with the NAND controller, you can do all the above
<oliv3r>
i thought f2fs was a fS to be used on those 'light' FTL's, like eMMC and MMC and USB flash (which basically is usb + mmc really)
<hramrach>
but you get 10MB/s cap with nand
<Turl>
# bootp
<Turl>
Waiting for PHY auto negotiation to complete.... TIMEOUT!
<Turl>
ENET Speed is 100 Mbps - FULL duplex connection
<oliv3r>
with our nand?
ganbold__ has quit [Remote host closed the connection]
<oliv3r>
Turl: that with mine?
<hramrach>
Turl: what board?
<Turl>
with tftp and the like it does work after the timeout, it kinda hangs for a while and then prints a T, then it works
<Turl>
hramrach: ct
<Turl>
oliv3r: no, mine
<hramrach>
what speed switch?
<Turl>
hramrach: gbit
<hramrach>
then you have old driver
<oliv3r>
Turl: i re-transferd a ct u-boot now; same url 100% ct
<oliv3r>
Turl: i can use hashes to push only specific ranges right?
<Turl>
hramrach: I'm using wens' v3 on top of sunxi uboot
<Turl>
hramrach: is there something newer?
<oliv3r>
Turl: i'll commit my local branch and remove my latest patches
<Turl>
openwrt reports the port autoneg'd to 100mbit
<oliv3r>
Turl: since you won't use my pre-compiled u-boot; pull sunxi u-boot :p
<hramrach>
Turl: I am using v2 rebased on top of some other patches
<hramrach>
rather messy
<oliv3r>
i pushed wens's stuff
<oliv3r>
hramrach: but that 10 mb/s is that because our crappy nand controller; our nand is really slow, or that's a cap by definition?
<oliv3r>
hramrach: i always thought our nand was REALLY slow; but my phone boots like 10x slower then my tablet, so i'm now going 'wow this tablet boots fast'
<Turl>
oliv3r: your precomp uboot fails the same
<hramrach>
oliv3r: I don't know. What I do know that at this speed I don't mind if the nand fails.
<Turl>
I also get Warning: failed to set MAC address
<hramrach>
Turl: want try my binary? and also does the networking work with openwrt?
<oliv3r>
Turl: env set macaddr 12:23:34:45:56:67 :p
<oliv3r>
Turl: but latest u-boot now, should work; unless something is going awefully wrong somewhere
<oliv3r>
hramrach: witha proper nand controller/fs, shouldn't the size start shrinking when it's about to die?
<Turl>
oliv3r: even after that it still complains about mac