<Turl>
oliv3r: I also get that message on cb2, fixit :)
<oliv3r>
Turl: ahh ok; then i'll have to have a word with hno about it :)
<oliv3r>
Turl: i noticed u-boot is currently limited to 1GB
<oliv3r>
Turl: so fixiing that for the CT
granite_crusher has quit [Quit: Page closed]
<Turl>
oliv3r: CT is A20 right?
<wingrime>
Turl: YES
<oliv3r>
DRAM: -2048 MiB
<oliv3r>
Turl: yes
<oliv3r>
i got negative memory!
<oliv3r>
wingrime: yours probably may take a year
<oliv3r>
but atleast you have dhl track-trace
<oliv3r>
wingrime: did you get an e-mail?
<oliv3r>
DRAM: 2 GiB
<oliv3r>
atleast regular u-boot doesn't lie
<oliv3r>
hno: does u-boot support more then 2GB ram at all?
ganbold_ has quit [Remote host closed the connection]
<wingrime>
oliv3r: they used some 3rd guy for resend
<oliv3r>
hno: the only change in dram settings is density went to 8192 from 4096
<oliv3r>
wingrime: ours got shipped from germany
<oliv3r>
wingrime: did you get an e-mail from them?
<oliv3r>
mripard: Truncating RAM at 40000000-bfffffff to -6f7fffff (vmalloc region overlap).
<oliv3r>
mripard: what does that mean?!
<mripard>
oliv3r: that you have something wrong with the RAM ? :)
<wingrime>
oliv3r: yeax but no any track
<oliv3r>
wingrime: :(
<jemk>
:( I didn't get one. looks like i have to wait for second batch.
<oliv3r>
mripard: heh :p
<Turl>
oliv3r: probably doesn't fit under the default memory split
<Turl>
you'll need to change it or use highmem
<oliv3r>
jemk: did you sign up for the cb2 list back in the time?
<jemk>
oliv3r: no
<oliv3r>
Turl: how doesn't it fit when i have MORE ram :D
<Turl>
oliv3r: addr space is just 4GB
<oliv3r>
jemk: ah, i know they used that as template for the addresses. that said i think i did see your name pop by
<oliv3r>
Turl: i remember the pae thing etc; but i first need to know if u-boot passes the amount of ram along properly
<Turl>
I think 3/1 split (the default) lets you have 1G of addressable ram only, and the rest via highmem
<oliv3r>
ah
<oliv3r>
ok i'll rebuild then
<oliv3r>
how do you enable highmem
<Turl>
CONFIG_HIGHMEM or sth :p
<oliv3r>
ok let me check
<oliv3r>
i thought 3/1 split was really default for desktops too
hno has quit [Read error: Operation timed out]
ganbold_ has joined #linux-sunxi
hno has joined #linux-sunxi
<oliv3r>
Turl: i think we should add atleast highmem tot he sunxi_defconfig and maybe even multi_v7?
<Turl>
oliv3r: I think it is indeed
<Turl>
oliv3r: but most people don't use x86 anymore :p
<Turl>
oliv3r: 64bit processors have huge addr space, so split is mostly irrelevant I'd say
<oliv3r>
yep
<oliv3r>
but 2G socs is getting more and more common?
<oliv3r>
phones etc
<oliv3r>
and all those will be 32bit for some time
<oliv3r>
(yeah, there's new socs that are arm64bit now, but older socs will be used for a while with more ram possibly)
HeHoPMaJIeH has quit [Ping timeout: 240 seconds]
<Turl>
the issue is that you need to balance it with the apps' split
HeHoPMaJIeH has joined #linux-sunxi
<oliv3r>
highmem?
<Turl>
if you go, say, 2/2 split, the apps only get a 2G addr space to use
<wingrime>
oliv3r: I better want new ddr controller for normal speed
<oliv3r>
wingrime: for?
<wingrime>
oliv3r: than 2gb
<wingrime>
oliv3r: all system, cpu is fast , mem - slow
<Turl>
oliv3r: did mbus get bumped on uboot?
<Turl>
wingrime: not much if you tweak mbus
<Turl>
wingrime: 3GB/s fill
<wingrime>
Turl: sended patch?
<oliv3r>
ok, no i really am to confused and don't get anything
<oliv3r>
wingrime: it's still A20, so same (or almost same mem controller) all they could do, was add more memory :)
<Turl>
wingrime: no, but ssvb did I think
<oliv3r>
Turl: ah ok bump mbus, now i get it/remember, i don't know if hno has applied the patch
<oliv3r>
i can, if everybody things its a good idea
<oliv3r>
(i think it is, but some chips might not handle it or something?)
<Turl>
oliv3r: there was a thread on the ML, nobody opposed as far as I recall
<oliv3r>
ok i'll find the patch and apply it later today if i don't forget
<oliv3r>
gotta get the cb booting first :)
<ssvb>
Turl: appears that bumping the MBUS clock can cause some SD card related problems
<oliv3r>
erm CT with 2GB
<oliv3r>
that's crap then :(
<Turl>
ssvb: which kind of problems?
<wingrime>
ssvb: use divider luke
<Turl>
ssvb: I haven't noticed any (but then my linux doesn't have an mmc driver either :P)
tinti has quit [Ping timeout: 264 seconds]
<ssvb>
Turl: I'm using NFS exclusively :) but tried to install Ubuntu on the SD card today in the morning for running some mali tests (to verify the correctness of the instructions on wiki)
<ssvb>
Turl: with the overclocked MBUS, it was dying horribly
<n01>
mripard: about the RTC patch I left in the while cycles because with for cycle IMO it was a bit messier
<Turl>
ssvb: synced to PLL5 with no divisors?
<ssvb>
Turl: clocking MBUS down solved these issues, and I have not investigated it further yet
<Turl>
ssvb: are you using 480Mhz memory?
<ssvb>
Turl: yes, PLL5 with no divisors and 480MHz memory
<speakman>
rm: The TI platform description was to tell how I used to manage it.
<oliv3r>
Mem: 2077120 13116 2064004 0 0
<oliv3r>
whoot
<oliv3r>
highmem++
<Turl>
ssvb: maybe if you run ram at, say, 408 issues disappear
<oliv3r>
so the negative value is a spl printing issue
tinti has joined #linux-sunxi
<oliv3r>
nothing too serious i think
<speakman>
rm / Turl: Are there any "good pratice" how to store individual addresses on each device?
<Turl>
ssvb: and who knows, you might still get better perf than normal mbus with fast ddr clock
jlj_ has quit [Ping timeout: 256 seconds]
<speakman>
How does Allwinner Android tables do it, for example?
<Turl>
ssvb: tablets don't have ethernet
<oliv3r>
what clock is mmc running from?
<Turl>
er, speakman ^
<speakman>
Turl: lol of course :D
<oliv3r>
speakman: proper practise is to store that kind of info in an EEPROM :)
<oliv3r>
speakman: a10/a20 can in theory use the efuses for this
jlj_ has joined #linux-sunxi
ykchavan has quit [Quit: Leaving]
<oliv3r>
n01: it doesn't matter if it is in the soc or not, the dt is supposed to be called sun4i-a10-rtc, that's just how it's named, sun4i-rtc was a bug, that we can't fix i think
<mripard>
n01: hmm?
<Turl>
oliv3r: I think we have a tradition of keeping the bug though
<Turl>
for sun4i-*
<Turl>
as there's not going to be another sun4i anyway
<speakman>
oliv3r: thanks.
<jemk>
Turl, ssvb: my guess would be, mmc/sd has own dma which is directly connected to mbus. maybe cpu can handle such fast mbus, but dma can not
tinti has quit [Max SendQ exceeded]
arokux2 has joined #linux-sunxi
<n01>
mripard: I didn't address one of your comments on the previous RTC driver
tinti has joined #linux-sunxi
<mripard>
n01: oh, ok
<mripard>
if you say so :)
<mripard>
you'll probably have the same comment then
<mripard>
the only thing that I don't get with your patches, is why didn't you send them to the kernel mailing list...
<n01>
mripard: because I want a preliminary screening
<n01>
just to clean the most obvious stuff
<n01>
oliv3r: I'll change it then
<Turl>
n01: I didn't like the fact that you removed one interrupt from a node and moved it into the rtc
<Turl>
it deserves an explanation on the commit msg at least :)
<n01>
Turl: mmmmh, I had a problem with the IRQ and that was one of the modifications I did to solve the problem. but probably you are right and it is better to leave it also in timer2
<mripard>
n01: it's your second one, on your first, I already told you it was clean
<oliv3r>
Turl: you have no idea how itchy my fingers are to send a patch for sun4i- to sun4i-a10-; but i know it'll get nacked anyway :(
<mripard>
the point is not to get a driver merged at the first version. You *will* have to do other versions. So why not send them to the people that will have much more to say than "the indentation is bad here"
<oliv3r>
jemk: excellent point, well worth investigating if that means we can use high mbus
<oliv3r>
and dynamically downclock it on mmc access
<n01>
mripard: wise guy :) ok
<mripard>
the real deal of a review is done by the maintainer of the subsystem, and the only review I can give is wether it looks fine to me or not, but they are not the real issue that your driver might have
<jemk>
oliv3r: that would be a bad hack, and there are many more dma devices that need to be checked
<mripard>
and only the maintainer and the guys doing rtc stuff on a regular basis will be able to give you these feedbacks
<n01>
yes I understand
<Turl>
mripard: do you have any pending clock patches so far btw?
<mripard>
(and really, don't be afraid, no one will flame you for sending a patch if you behave normally, ie. sending your patches like you should, replying to comments, not ignoring some comments, etc.)
<mripard>
Turl: that I should send, or that I have received ?
<Turl>
mripard: that you have sent and has not been merged so far
<arokux2>
ssvb: I was told the non-working usbmon log has much faster timing. maybe mbus clock is guilty here too... how do I change mbus to normal speeds?
<Turl>
arokux2: A10 doesn't have (configurable?) mbus
<arokux2>
thanks Turl I need to find out what casing this so much quicker timing
<atsampson>
I'm not sure saving £3 on something designed to be connected to the mains is a good tradeoff ;)
<oliv3r>
yeah it is made in china quality
<oliv3r>
atsampson: absolutly, but when buying for example a bunch adapters from ebay, you don't know what you get
<arokux2>
oliv3r: you've said so much without even taking a look at the links.
<oliv3r>
arokux2: i did actually
<oliv3r>
arokux2: but statements hold; it could be chinese 'prongs' but the description/image not properly updated
<oliv3r>
arokux2: or it could be chinese quality
<oliv3r>
the webshop doesn't state either, atsampsons blog link states the answer
<arokux2>
oliv3r: I meant olimex blog. olimex did nice comparison, the Chinese one is good enough for the money you pay.
<atsampson>
it's pretty impressive that you can make an SMPS like that for £4 -- I guess the energy-saving lightbulb and phone charger market has done a good job of driving the component prices down...
<atsampson>
there's not much to choose between the PCBs, but I'd probably go for the EU version for the better DC cable and connector (it'd cost about the difference to replace if it breaks later)
<oliv3r>
atsampson: well in the olimex shop; yes
<oliv3r>
but when buying from ebay, you have no idea what your getting. and the price is no indiciation unfortunatly :(
<oliv3r>
atsampson: what, fire, smoke and explosions?
<atsampson>
surprisingly rarely, actually ;)
<oliv3r>
anatec did a powersupply review once, that i have never seen done since
<oliv3r>
they took power supplies, and ran memtest 24 hrs
<oliv3r>
and that 3 times
<oliv3r>
then they looked at the amount of memory errors
<oliv3r>
turns out, not every powersupply was equally 'good'
<oliv3r>
some brands caused higher memory failures
<arokux2>
mripard: you said 3.12-rc2 boots fine on you board farm, that is why I asked about your board farm.
<atsampson>
the hardwaresecrets reviews are good along those lines -- they look at what the different rails can supply, how much ripple there is on each rail, etc.
<oliv3r>
yeah they did allt hat; temperature etc
<oliv3r>
but a 24hr burn in test running memtest, i have not seen on a PSU review
<oliv3r>
memtest86+ I should say i think
rellla has quit [Remote host closed the connection]
jemk has quit [Ping timeout: 248 seconds]
<mripard>
arokux2: no, there are board farms maintained by the arm maintainesr
<arokux2>
so if I take a look at digikey, PSU of 200W they already cost 60 EUR, where as Chieftec iArena 80Plus 500W will cost you 35 EUR, why...?
<mripard>
that boot sunxi hardware on master, linux-next, and last stable trees
<arokux2>
mripard: ok, I see.
<mripard>
on a more-than-daily basis
hramrach has quit [Write error: Broken pipe]
hramrach has joined #linux-sunxi
<oliv3r>
anybody around that understands the u-boot led assingments in board.cfg?
wingrime has quit [Ping timeout: 245 seconds]
notmart has quit [Remote host closed the connection]
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has joined #linux-sunxi
<Turl>
oliv3r: they are the pin nº I think
<oliv3r>
well it's decimals
<oliv3r>
i haven't related it yet :(
<oliv3r>
Turl: the CT has 4 status leds, led1-4
<oliv3r>
244 appearantly lights up L2
<mripard>
yes, from what I recall, it's the gpio number assigned to the led
<oliv3r>
but 243 or 245 doesn't do anything
<mripard>
the gpio number being bank * 32 + number
<oliv3r>
so how do I get from the decimal to the GPIO number?
<oliv3r>
so without schematic or source we can't really figure out what is what
<oliv3r>
since it runs the 3.3 android kernel
<mripard>
exactly
<oliv3r>
shit
<oliv3r>
:)
<oliv3r>
'the most open board'
<oliv3r>
*sigh*
<oliv3r>
i'll bug benn next week for the schematic
wingrime has joined #linux-sunxi
<Turl>
oliv3r: see drivers/gpio/sunxi_gpio.c
<Turl>
it's the 'pin' number
<Turl>
#define GPIO_BANK(pin) ((pin) >> 5)
<Turl>
#define GPIO_NUM(pin) ((pin) & 0x1f)
<oliv3r>
heh, i updated that for a31 so i should know that right? :)
<oliv3r>
in u-boot anyway
<oliv3r>
but that does clarify things a bit
<oliv3r>
but also means, without info which gpio is connected to what pin, it'll be a while before we know what led is connected where
<Turl>
oliv3r: script.bin not configured for leds?
eebrah_ has quit [Ping timeout: 240 seconds]
heffer has joined #linux-sunxi
heffer has quit [Changing host]
heffer has joined #linux-sunxi
eebrah has joined #linux-sunxi
eebrah is now known as Guest60148
<arokux2>
oliv3r: just do gpio s PH 03, for example, u-boot will tell you the number
<oliv3r>
so interestingly, what is led2 connected to then and what port
<Turl>
maybe it's the gpio above?
<oliv3r>
i was thinking that
<oliv3r>
let me use arokux2 gpio tip :)
<oliv3r>
gpio: pin PH03 (gpio 227) value is 1
<oliv3r>
nice
<oliv3r>
i like
<oliv3r>
and i can toggle leds on/off
<oliv3r>
no idea what PH03 is :p
<mripard>
oliv3r: oh, you have the fex file
<mripard>
you have a u-boot image working ?
<oliv3r>
mripard: yep
<mripard>
I'll do the DT tonight if you do
<oliv3r>
committed it allready
<oliv3r>
i don't mind doing it :)
<oliv3r>
you are busy enough :)
<oliv3r>
just gotta fix an overflow thing: DRAM: -2147483648 MiB
<oliv3r>
:)
<andoma>
:)
<mripard>
ok, your call :)
<oliv3r>
mripard: and i wanna be usefull and send patches ;)
<oliv3r>
Turl: gpio: pin PH20 (gpio 244) value is 1
<oliv3r>
clear turns the led off
<oliv3r>
arokux2: awesome
<oliv3r>
Turl: awesome tip, fex file
<oliv3r>
244, 245, 231, 235 for the log readers
<Turl>
:)
<oliv3r>
blue, orange, white green and red leds :)
<Turl>
oliv3r: enable cpu trigger on two of them, and use net and mmc triggers for the other two :P
<oliv3r>
:)
<arokux2>
you'r welcome oliv3r
wingrime has quit [Ping timeout: 248 seconds]
Gerwin_J has joined #linux-sunxi
wingrime has joined #linux-sunxi
<oliv3r>
Turl: opinion: returning an int won't work any longer, overflow. so, return unsigned int is an option, but then we cna't return -1. Should I just return 0 on all errors? or change it to a uint64
<Turl>
0 could be an error value, as 0 bytes would be invalid
* Turl
checks code
<oliv3r>
i think the only -1 i see is for the first 'if'
<oliv3r>
then again, it probably needs int64; since armv8 will cause problems of the same size, with 8gb ram
<arokux2>
I find it amusing to learn how many AC adapters I have in my home: PC speakers 13.5V, 5A; display: 14V, 3A; router 5V, 2A; external HDD 12V, 1.5A and that's not all
<oliv3r>
arokux2: yep
<oliv3r>
arokux2: engineers design a device, pick an adapter that works with it
<oliv3r>
but all those voltages can be explained
<Turl>
oliv3r: get_ram_size returns long
<oliv3r>
pc speakers probably have a 12V circuit and need a rectifier, so you input 13.5; that gets converted internally to a stable 12V. router is probably a 3.3V device, so also a downstep conversion maybe 5v
<oliv3r>
Turl: yeah but dramc_init returns int; so upscaling everything to int64 iss probably best (or long I suppose is enough, but i don't like the unknown-ness of long)
<Turl>
isn't long the same as int?
<oliv3r>
arokux2: external hdd needs 12V for the motor, it internally downconverts to 5V for the logic and possibly 3.3V if its' sata
<oliv3r>
Turl: depends :)
<oliv3r>
Turl: int is supposed to be the native with of the CPU, so 32bit in our case
<Turl>
oliv3r: on 32b arm at least :p
<oliv3r>
so i'll rewrite it all to iny64 and keep the errors
<oliv3r>
unsigned int would have been a quicker 'fix' :p
<oliv3r>
but not future proof at all
<oliv3r>
but mor work :p
<Turl>
oliv3r: but it'll still overflow right?
<Turl>
as get_ram_size returns long which is 32b
<oliv3r>
yep, that's where the error happens
<oliv3r>
hence, my negative ram :)
<Turl>
you can probably cast the return to unsigned long
<Turl>
and then return as int64
<oliv3r>
might aswell just clean it all up imo :)
<oliv3r>
the board specific files is where the pains at :(
<Turl>
if you're ok with touching mainline uboot code go ahead then :)
<oliv3r>
s64*
<oliv3r>
yeah i've done plenty of patches ;)
jemk has joined #linux-sunxi
<n01>
ssvb: ping
<oliv3r>
Turl: oh wow, get_ram_size() is invoked all over the place; maybe a cast for now is ok :)
<ssvb>
hi n01
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<n01>
hi ssvb, the usual OT question about fbdev :)
<n01>
for the display controller I'm working with I need to setup the LVDS
<n01>
now, all the stuff for LVDS goes inside the fbdev?
<n01>
or is it better to define another driver just for lvds?
<ssvb>
n01: I think you should just contact the fbdev kernel folks
<n01>
I'm scared they just will say "switch to drm" ;)
<n01>
nvm, tnx anyway
<oliv3r>
one more u-boot led question; what are statusled1, 2 and 3 used for; only 0 seems to be used it seems
<Turl>
oliv3r: they're user programmable
<Turl>
oliv3r: you can do 'led all on' I think
<oliv3r>
let me rephrase, what does u-boot use them for by default
<oliv3r>
Turl: but that was good enough; as that 'works' :)
<oliv3r>
led 0 on/off
<oliv3r>
toggles the led nicely, so i'll update that
<oliv3r>
thanks for saving me tons of lookup time ;0)
<Turl>
yw :)
massi has joined #linux-sunxi
<oliv3r>
Turl: btw, fex file has typo's :p all leds are 'blue' :;p
<oliv3r>
Turl: feck; mainline code updates it is, overflows galor
naobsd has joined #linux-sunxi
<slapin_>
Hi, all!
<slapin_>
have anybody got A31 mele per chance?
<oliv3r>
slapin_: i think so, someone was hacking on one a while ago
<oliv3r>
slapin_: but a31 mele isn't supported by us really :p
<oliv3r>
slapin_: nobody has a31 hardware; and most shun powervr that's in a31
<slapin_>
mine got into some wicked state and they only provide some windows tools to generate sd card, but I have no anyaccessible windows boxes, and the tool doesn't work in wine, can anybody help with making image of that SD so I could dd it to recover box?
<oliv3r>
livesuit also works with linux; but ... not widely tested
<oliv3r>
we don't have u-boot support at all yet
<slapin_>
oliv3r: I could not make fel mode running on that box, they only advertice booting from SD
<arokux2>
oliv3r: as far as I understand only voltage of an ac adapter should be absolute equal to some value and then the more amps, the better. is that correct?
<slapin_>
oliv3r: have they released u-boot source? as I managed to get into one's prompt on that box, but I'm afraid to touch anything there...
Gerwin_J has quit [Quit: Gerwin_J]
<arokux2>
slapin_: check wiki, there is some info
<slapin_>
arokux2: depends on the supply, but generally yes. Just remember that sometimes if not loaded enough the PSUs tend to do bad things to your hardware (especially economy ones with lack of capacitors)
<oliv3r>
slapin_: FEL mode works specially on that box, check the BROM wiki page about how a31 boots differently
<oliv3r>
arokux2: well sorta, but, the higher the amps, the more heat generated by various components. By ramping up the voltage, you can lower the current. P = V * I.
<oliv3r>
slapin_: i think we only have u-boot without dram init. but what is broken exactly?
Gerwin_J has joined #linux-sunxi
<slapin_>
oliv3r: it boots into android, but android doesn't display anything, so I see only A31's splash screen
<slapin_>
oliv3r: displaying lots of something on serial, which doesn't trigger in my head with anything
<oliv3r>
slapin_: and adb doesn work?
<slapin_>
oliv3r: so basically just hangs, I'd like to reflash it
<oliv3r>
i know the latest android does print debug on serial, but doesn't offer console (unless typing 'ls' works for corse)
<oliv3r>
adb shell works though
<oliv3r>
slapin_: reflashing can be done with livesuit for linux, or if that fails, ony with livesuit for windows
<slapin_>
oliv3r: adb doesn't work
<oliv3r>
slapin_: check our wiki about livesuit
<slapin_>
oliv3r: no android gadget available
<oliv3r>
when you fix it, sell it and get a20 based device like mele m5
<oliv3r>
slapin_: i fixed the usb kernel driver a while ago, should still be ok; check our github :)
<slapin_>
oliv3r: I tried their image with livesut, it can't find USB device as I can't enter FEL mode
<oliv3r>
though the livesuit installer tries to install the usb driver
<oliv3r>
you have to enter fel mode obviously
<oliv3r>
on the serial console
<oliv3r>
when you see:
<oliv3r>
Jump to Boot1.
<oliv3r>
etc etc, smash 2 madly on your keyboard
<oliv3r>
that forces fel mode
<oliv3r>
[ 3.975] Jump to fel
popolon has joined #linux-sunxi
tinti has quit [Ping timeout: 240 seconds]
<oliv3r>
i hate typedef :S
tinti has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
Gerwin_J has joined #linux-sunxi
<n01>
oliv3r: cscope is da way
<slapin_>
oliv3r: I was hoping that I could find some SD card image made with phoenixcard, as I have gut feeling it won't come that easy...
<oliv3r>
slapin_: it won't come that easy, but try entering FEL mode via serial port; shouldn't be hard
<oliv3r>
boot0/1 support going into fel mode via the '2' key
<slapin_>
oliv3r: as I never had any success with livesuit on mk802-likes
<oliv3r>
livesuit is crap :D
<slapin_>
oliv3r: so that is why I'd vote for sdcard
<arokux2>
oliv3r: have you seen powerstrips for DC?
<slapin_>
oliv3r: btw, debian on jffs2 was successfully installed on mk802. still slow, though
<oliv3r>
arokux2: nope; though in thory, you can use any powerstrip
<oliv3r>
slapin_: with mtd?
<slapin_>
oliv3r: yes
<oliv3r>
slapin_: awe-some :D
FDCX_ has joined #linux-sunxi
<oliv3r>
slapin_: so when are you hacking more on NFC? :)
<oliv3r>
Turl: this is gonna be one huge ass patchset :S everything needs to be changed
<oliv3r>
Turl: not hard, just lots of work :S
<slapin_>
oliv3r: I want to get some time, it is hard with my lifestyle, as I get some time, I get some small success.
<oliv3r>
slapin_: :D
Tartarus has quit [Excess Flood]
<Montjoie>
hello wingrime, I dont understand your remark on the y/m option of Kconfig, all AXP stuff are bool only
<oliv3r>
slapin_: well the todo for NFC is: port boot-block code from the new lichee-3.4 kernel to sunxi-3.4; port mtd to u-boot and port mtd to the kernel
Tartarus has joined #linux-sunxi
<slapin_>
oliv3r: mtd to u-boot is almost done, but I need to have discussions with hno about it as the patch is not really ok from sanity point of view. Kernel was made ad-hoc but I'd prefer to sit and think , as I want to understand why the damn thing is slower than block driver, as it does almost the same. As for boot-block I am listenning.
<oliv3r>
Turl: mripard why does cubieboard1 have drive '1' and cubieboard2 drive '0' for the led pins?!
<oliv3r>
slapin_: i'm know where the bootblock lives (just before the partitions where boot0/boot1 are stored) but know not more, mnemoc will upload lichee-3.4 kernel to his github soon for reference.
<oliv3r>
slapin_: as for mtd integration into u-boot; yeah you deff. need to talk with hno what he prefers
<mripard>
oliv3r: I'm not sure it makes much difference for the LED
<mripard>
drive 1 is 20mA, 0 is 10mA
dwilkins has quit [Ping timeout: 245 seconds]
<slapin_>
oliv3r: I think it is not the preference question, but 'this is wrong, but dunno how to make it right' list of discussion topics
dwilkins has joined #linux-sunxi
<oliv3r>
mripard: well the led is controlled via a transistor on the cubietruck; so 10mA should be enough as long as it 'switches' the transistor on
<oliv3r>
slapin_: aye :( well we are jumping for mtd and excited for it :D
<oliv3r>
slapin_: btw, i'm not sure those bugs reported by r2z are any different with the stock driver. I occasionally have corroptions on my tablet running stock android too
<oliv3r>
even on the read-only partition (/system)
<oliv3r>
which shouldn't be possible afaik
<mripard>
oliv3r: then make it to 0
<slapin_>
oliv3r: there is surprise if there is no corruptions - NANDs degrade, and there is no code in driver which cope with that, so any Linux flash system will be safer than AW's.
<slapin_>
oliv3r: this is the price for cheapness of MLC NAND - you have to handle all that really carefully, not only detect bad data, but also handle it properly.
<oliv3r>
mripard: i did; just noticed the differenc e :)
<oliv3r>
hmm kernel crashes
<oliv3r>
oops :)
<slapin_>
oliv3r: the problem with mtd+jffs is different - the flash is large, the erase block size is large, and page size is large too. and usable window in OOB area is tiny (all used by hardware ECC), and jffs2 was just not designed for all that.
<slapin_>
oliv3r: but if you're persistant enough (like me) that you can ignore this all facts, then you can run jffs2 with mtd in such configuration, and it is a good test for driver quality. And yes, it is hacky, too, as I put jffs2 OOB metadata into data part of page, which might be frowned upon, but who cares?
<slapin_>
mripard: ^
<slapin_>
oliv3r: can you please gimme links for rz2k ravings re corruptions?
FDCX_ has quit [Remote host closed the connection]
<slapin_>
I see far less people here than before, strange...
<Montjoie>
Hello what is the difference between linux-sunxi@googlegroups.com and dev@linux-sunxi.org ?
<Montjoie>
does I need for a patch to send to both ?
ykchavan has joined #linux-sunxi
<slapin_>
oliv3r: ^
gogi has joined #linux-sunxi
gogi has left #linux-sunxi [#linux-sunxi]
<slapin_>
Montjoie: sending to both won't harm anybody's feeling, I think. And patches are tasty.
* slapin_
likes patches very much
HeHoPMaJIeH has quit [Remote host closed the connection]
Gerwin_J has quit [Quit: Gerwin_J]
jemk has quit [Read error: Operation timed out]
granite_crusher has joined #linux-sunxi
<granite_crusher>
arokux2:hi, are still working on usb, or did you started AXP209 hacking?
<arokux2>
granite_crusher: hi, USB is kind of ready, still no idea why on-board wifi doesn't work though.
<arokux2>
granite_crusher: currently I'm trying to figure out what cables, sd cards and PSU I need to buy
<granite_crusher>
is that for this A13?
<arokux2>
granite_crusher: I've got 2 board recently: a13-micro-wifi and radxa
<arokux2>
granite_crusher: I should get cubietruck soon too.
<granite_crusher>
arokux2: oh, you got micro.. 5V? less flexible than A13
<arokux2>
granite_crusher: I do not quite like how wifi module sticks out
<arokux2>
granite_crusher: turns out it needs 6V-16V power supply and micro sd
<granite_crusher>
arokux2: when I had to choose I chosed without it, because I also didn't liked the idea of wifi soldered on...
<granite_crusher>
arokux2: It have NAND
<arokux2>
granite_crusher: no NAND in mainline yet
<granite_crusher>
You can do things without SD
massi has quit [Quit: Sto andando via]
<granite_crusher>
arokux2: but anyway You boot from net?
<arokux2>
granite_crusher: u-boot needs to be on sd
<granite_crusher>
arokux2: it do not work from nand?
<arokux2>
granite_crusher: is it ours u-boot that is installed on nand?
<arokux2>
granite_crusher: ours u-boot doesn't have NAND support, so I think no
<granite_crusher>
arokux2: I don't know I have no nand... what about that lichee?
<arokux2>
granite_crusher: I do not want to touch it :)
<arokux2>
granite_crusher: hm.. its a pity that a13 doesn't have ethernet
<arokux2>
granite_crusher: I might need usb2ethernet.. ok adding it to my list
<arokux2>
granite_crusher: what about you, have you added rtc to dts?
<granite_crusher>
arokux2: so Olimex sent you just board without LCD, touchscreen, battery, AC-DC adapter, USB-UART debuging cable USB-ethernet?
<arokux2>
granite_crusher: yeah, just a board
<granite_crusher>
arokux2: I got it working
<arokux2>
granite_crusher: cool :)
<granite_crusher>
but I start to doubt that things should be patched into device tree... rather it would be cool if there would be way to teach how to creat one (something as fex guide)...
<arokux2>
granite_crusher: what? (there is a device tree wiki)
<arokux2>
granite_crusher: and a lot of examples in kernel src
<granite_crusher>
arokux2: especially for olimex boards which are devel boards, and people have different needs, one can need UART3, while other SPI1 on the same pins...
ganbold_ has quit [Remote host closed the connection]
<granite_crusher>
arokux2: in linux-sunxi?
<arokux2>
granite_crusher: ah, I see what you mean now.
<granite_crusher>
arokux2: and with that device tree I see some nightmare with LCDS coming up...
<arokux2>
oliv3r: which kind of SD card slot is there in CT?
<arokux2>
oliv3r: I see now - micro sd
<arokux2>
granite_crusher: so what you'll do next?
<granite_crusher>
arokux2: I see linux-sunxi stable 3.4 is going to 3.10... I am a bit lost what I could do useful.. I can test things, and solder....
<granite_crusher>
arokux2: actually I remember in irc there was mentioned that way boot of SPI-NOR should exist, and this idea haunts me...
<arokux2>
granite_crusher: 3.4 -> 3.10: I'm not the fan of it. i'll see how it develops.
<arokux2>
granite_crusher: we could work on AXP together
<arokux2>
granite_crusher: where did you buy you battery btw?
<granite_crusher>
arokux2: good idea
<granite_crusher>
arokux2: I bought from olimex, but I saw similar in few kinds of shops around me
<granite_crusher>
the most important thing is it should be 3.7V
<granite_crusher>
and one cell
<arokux2>
granite_crusher: I couldn't find anything that cheap around me. the battery is the only thing I cannot find in germany!
<granite_crusher>
I saw such in electronic shops (shops with electrical parts as FETs, cables, pins...)
<granite_crusher>
arokux2: did you tried modeler shops (these with plane models)?
<arokux2>
granite_crusher: they have very few mAh
<arokux2>
granite_crusher: how much do you have?
<granite_crusher>
arokux2: I have 1400mAh (smallest from olimex)
<granite_crusher>
arokux2: I would'nt go with expensive ones at the begining, and smaller ones explodes less intensively :P
<arokux2>
:)
<arokux2>
it is impossible to find all the stuff in one shop :(
<wingrime>
what....
<wingrime>
[PATCH V2 Resend 43/92] cpufreq: cris: don't initialize part of policy that is set by core too
<wingrime>
92 omg
<granite_crusher>
arokux2: I think it even would be better to have small one, because it would be esier to test AXP, You shouldn't wait half day, but just minutes or few hours to see if discharging or recharging works corectlly..
<granite_crusher>
*easier
<arokux2>
granite_crusher: yeah, this is a good idea.
<granite_crusher>
arokux2: mine battery (from olimex) have some tiny PCB attached to it directtly, and as I understand AXP comunnicates with battery throught 1-wire interface
<granite_crusher>
it is probably fuel gauge(?) and protection(?) of overcharege or switched +/-
<granite_crusher>
overcharge*
<arokux2>
granite_crusher: sorry, I know nothing about it yet..
gzamboni has quit [Read error: Connection reset by peer]
<oliv3r>
hno: mripard: Turl: mnemoc: I named the CT cubietruck now, but I think it should be named cubieboard3 as that's what's in smaller letters on the box
<granite_crusher>
arokux2: I wonder why they are not doing diy?..
n01 has quit [Ping timeout: 246 seconds]
<granite_crusher>
arokux2: anyway, if you would find something similar, you should look for battery with 2wires (I think there should be some comunity of modelers, so should be any shop, maybe they have no internet page?), and PCM or PCB (protective circuit module/board)
<arokux2>
granite_crusher: i'll just buy at olimex..
eebrah|noPC has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
wingrime has quit [Quit: Lost terminal]
<oliv3r>
slapin_: it was on the ML; i could try and find it :)
_BJFreeman is now known as BJfreeman
granite_crusher has quit [Quit: Page closed]
geecko_ has joined #linux-sunxi
eebrah|noPC has quit [Remote host closed the connection]
\\Mr_C\\ has quit []
granite_crusher has joined #linux-sunxi
<oliv3r>
hno: i broke u-boot :D (my own copy)
<oliv3r>
hno: I need to pick your brain on how u-boot passes memory size to the kernel
balage has joined #linux-sunxi
<balage>
hello everyone, i am back. i am still having problem with sunxi-3.4 on hackberry a10, namely some usb buses are not initialized or something like that, so no on-board wifi, no usb keyword is working. on miniand's forum I found two other people of having the same issue. details can be read here. I'd like to ask some help from you if it's possible. https://www.miniand.com/forums/forums/development--3/topics/custom-kernel-powering-usb-devices
popolon has quit [Quit: Quitte]
<arokux2>
balage: tgaz and jukivili are working on usb otg
<balage>
arokux2, is it related to otg? I've tried to disable otg alltogether but it wasn't even compile
<balage>
arokux2, but it's good news anyway :)
<arokux2>
balage: it is not, but we call this port like this, which is incorrect of course.
<arokux2>
balage: this port isn't an ordinary usb host, it has dual mode.
<oliv3r>
balage: but your problem appears to be related to the regular USB port
<oliv3r>
balage: one common pitfall here is to have an incorrect FEX file that doesn't supply power to the ports
<arokux2>
oliv3r: he's using otg in host mode afaik
<oliv3r>
so no on-board wifi,
<oliv3r>
arokux2: doubt it ;)
<arokux2>
oliv3r: I've talked to him already.
<arokux2>
hm.. but yes, no on-board wifi is strange
<oliv3r>
but you have to enable vbus via fex file
<oliv3r>
and people often overlook that
<oliv3r>
you can always use a powererd hub to double check if that'st he problem
<balage>
arokux2, let me tell you that I am using the exact same fex file (script.bin) for 3.0 and 3.4, and 3.0 works without problems.
<oliv3r>
mripard: looks like the PHY on the ct isn't backwards compatible with the emac driver or something; it crashes the kernel
<arokux2>
balage: what is your device?
<oliv3r>
well the emac driver is
<oliv3r>
libphy: 1c0b080.mdio-mii:00 - Link is Up - 100/Full
<oliv3r>
IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
<balage>
arokux2, yet it is, the only change is that I hard wired the MAC for ethernet, 'diff' shows no other changes
<oliv3r>
'hardwired'?
<balage>
oliv3r, [dynamic]
<balage>
-MAC = "000000000000"
<balage>
+MAC = "9a8246b2992c"
<balage>
oliv3r, that's all
<oliv3r>
balage: ohh mac address :) got it; that's fine
<arokux2>
balage: I have no ideas. do you use the most recent kernel?
<arokux2>
balage: maybe you could add a few printk in the kernel?
<balage>
arokux2, yes, commit 9ee9fc5f0988d is the latest on branch sunxi-3.4 (not staging)
<oliv3r>
balage: does staging work for you?
<arokux2>
oliv3r: staging has usb changes?
<balage>
oliv3r, haven't tried staging yet, does it worth to try?
<arokux2>
balage: no
panda84kde has quit [Ping timeout: 246 seconds]
<balage>
arokux2, okay but where? I'm not really skilled in 'C'
<arokux2>
balage: no problem, I'll tell you where.
<oliv3r>
arokux2: i don't recall the last syn cfrom stage to normal
<oliv3r>
mripard: ok reading the RTL8211E datasheet; it seems it may not be backwards compatible with emac, only compatible with gmac, the gmac however should be compatible with both kinds of PHY
<oliv3r>
so only the question stays whether to name it cb3 ro ct; i'm acking more towards cb3 actually
<arokux2>
oliv3r: why should it have a different name as product?!
<oliv3r>
arokux2: it's named both really
<oliv3r>
arokux2: it's the cubieboard3; the CubieTruck
<arokux2>
oliv3r: show me it on the official website
<arokux2>
balage: you messed up something. does usb works now?
<arokux2>
balage: controllers appeared now, but there was some oops
<arokux2>
balage: you decided to use in kernel driver for wifi?
<balage>
arokux2, now i'm confused :) really lsusb shows them, but I haven't changed the config (old config: 8192cu was loaded automatically)
<arokux2>
balage: you've messed up something...
<arokux2>
balage: backup your config, clean the kernel tree, configure and build again
<granite_crusher>
arokux2: sorry, I was away, I have absolutely no experience in low level programing I guess I could be usefull with electronics and share findings I expermented with older implimmentation (which do not work fully)
<arokux2>
granite_crusher: ok, no problem
<arokux2>
granite_crusher: I'll let you know once I have some idea of axp, it may take a while
_whitelogger has joined #linux-sunxi
granite_crusher has quit [Quit: Page closed]
notmart has quit [Quit: notmart terminated!]
ykchavan has quit [Quit: Leaving]
zetaneta has joined #linux-sunxi
tzafrir_laptop has quit [Ping timeout: 245 seconds]
ykchavan has joined #linux-sunxi
n01 has quit [Ping timeout: 252 seconds]
Black_Horseman has joined #linux-sunxi
balage has quit [Quit: leaving]
balage has joined #linux-sunxi
<balage>
arokux2, okay, i mrproper'd, then removed 8192cu and recompiled, booted again: http://sprunge.us/PVLe (i know realtek fw is missing) usb buses are online, tomorrow I'll try without debug again and see
<balage>
arokux2, thanks a lot for your help! cu
<arokux2>
balage: wait
<balage>
arokux2, yes
<arokux2>
balage: is usb stick or smth working?
balage has quit [Read error: Connection reset by peer]
balage has joined #linux-sunxi
FR^2 has joined #linux-sunxi
<balage>
arokux2, i've just tried an usb stick bit it doesn't get recognized at all (no dmesg output, neither lsusb)
<arokux2>
balage: hm, ok. let's continue tomorrow.
<arokux2>
balage: prepare a hub
<arokux2>
balage: if you have one..
<balage>
arokux2, yes I have one, then cu tomorrow, thanks for your help!
<arokux2>
balage: you are welcome!
balage has left #linux-sunxi [#linux-sunxi]
popolon has joined #linux-sunxi
wingrime has joined #linux-sunxi
FR^2 has quit [Quit: und weg...]
akaizen has joined #linux-sunxi
tzafrir_laptop has joined #linux-sunxi
<arokux2>
o_O
<arokux2>
Turl: you there?
<ykchavan>
I am getting error booting F19 from sdcard using hans's image. Error is "spl: not an uImage at 1600"