gsilvis has quit [Read error: Connection reset by peer]
gsilvis has joined #arm-netbook
tuliom has quit [Quit: Konversation terminated!]
<lundman>
bah, can not get the NIC to come up with replacement kernel, there sure is some black magic there
<lundman>
repalce script.bin and it works
<lundman>
So, ultimately, even though allwinner A40 looks like good hardware, we would still have the situation that they don't provide linux kernel, nor drivers for anything?
Gumboot has joined #arm-netbook
avernos has quit [Read error: Connection reset by peer]
mSquare has joined #arm-netbook
<Turl>
or worse, they provide a arch/arm/mach-sun6i/built-in.o
<lundman>
just thinking that even with A40, do I want it, when it is a hassle to work with. When there are similar boards with great source support
<lundman>
well, replaced mele-v1.3 android kernel with mine which has NIC fix, and zfs
tekzilla has quit [Ping timeout: 246 seconds]
<Turl>
on this price point?
tekzilla has joined #arm-netbook
<Turl>
I mean, we could all get a pandaboard, it has source and all
<Turl>
costs 5x of an A10
<lundman>
for me at least, the prices is irrelevant if the hardware can't do what I want anyway, right? So what if it is cheap
<lundman>
odroid-x maybe
ping__ has quit [Ping timeout: 260 seconds]
<lundman>
I wonder which bit in the script.bin is it that makes the NIC from not working
<Turl>
emac_used? :P
<lundman>
only diffs in emac area are:
<lundman>
-emac_power =
<lundman>
so one doesn't have it, and one has it empty
<lundman>
mele-1.3-script.fex works with *their kernel, but my kernel does not NIC link-up
<lundman>
mele-a1000.fex works with my kernel and NIC link-up
<Turl>
lundman: does your kernel have the sun*i ugly gpio thingy enabled?
<lundman>
using default sun4i_crane_defconfig
<Turl>
lundman: then edit the script and not the fex, do the gpio_para changes on your nonworking stuff
<Turl>
I bet it works then
<lundman>
you say, change the gpio_para in mele-1.3-script ?
<Turl>
change gpio_num to 3 and nuke the line talking about the 4th
<Turl>
and I bet it works on your kernel too
<lundman>
great, good hints, will try
ping__ has joined #arm-netbook
<Turl>
these FirstSteps on the wiki look wrong
<Turl>
Note that the boot partition should start at cluster 2048 (1024MB). Also note that this partition can have the linux partition type, we will format it to vfat later. You might want to make your boot partition bigger though.
<Turl>
I doubt I need to start my partition on a 1GB offset :)
<Turl>
uboot env ain't that big :P
<Turl>
lundman: any luck?
<lundman>
will try after I check zfs works ok
<lundman>
zfs dont run without /etc/mtab which is amusing
<Turl>
/etc/what?
avernos has joined #arm-netbook
avernos has quit [Changing host]
avernos has joined #arm-netbook
cheng has joined #arm-netbook
cheng has quit [Ping timeout: 246 seconds]
<lundman>
and of course root is readonly
cheng has joined #arm-netbook
n6pfk has quit [Remote host closed the connection]
mikey_w has quit [Read error: Connection reset by peer]
ping__ has quit [Ping timeout: 240 seconds]
mikey_w has joined #arm-netbook
n6pfk has joined #arm-netbook
<lundman>
ok, I am satisfied with the zfs stability under mele android
<lundman>
heh ok, I should remove the debug prints from the kernel
ping__ has joined #arm-netbook
ping__ has quit [Ping timeout: 246 seconds]
<lundman>
After you manually mount something in android, anyone know what is needed to tell the UI that it is there?
<lundman>
must be some combination of dbus, uevent and vold-fstab
ping__ has joined #arm-netbook
wingrime has joined #arm-netbook
<Turl>
lundman: you don't just 'mount something manually'
<Turl>
lundman: you use vold to mount
<Turl>
and dbus is there just for bluetooth as far as I know
<lundman>
yeah I mean its mounted, but android UI dont know that
<lundman>
normally vold gets the uevent for new device, then mounts it, then sends the event "volume_mounted"
<lundman>
I just dont know what happens after that
<Turl>
android talks to vold
<Turl>
so when vold mounts something, it tells android 'br0 I mounted /sdcard' then android knows that
<Turl>
on ics/jb there's an overlayable xml listing the mounts and its characteristics on the framework too
<lundman>
sure, both talk over /dev/socket/vold
<lundman>
and the message is literally "volume_mounted: /mnt/extsd"
<lundman>
just not seeing any tools to make that event by hand
<Turl>
why don't you just mount it with vold? :P
<lundman>
annoyingly hard to google for android crap, always 5 million morons wondering how to see their SD card
<Turl>
you're not supposed to make random events by hand
<lundman>
vold dont really know zfs
<Turl>
make it know it? :)
<lundman>
stop making my 5 mins project that took 3 days into a 1 week project
<Turl>
why do you need android to know it's mounted anyway?
<lundman>
so the filemanager can see the pool
<Turl>
filemanagers don't care as long as the permissions are right
<Turl>
just navigate to it
<lundman>
filemanagers only show "mounted volumes"
<lundman>
which currently is "/mnt/sdcard"
<lundman>
the built in one I mean, not those smart enough to go outside that
<lundman>
yes, but when you import pools, zfs mounts the filesystems
<lundman>
so its a different way to handle it, there is no "mount" command for zfs, normall
<lundman>
+y
<lundman>
anyway, its on stackoverflow
<lundman>
right, next was hmm
<lundman>
turl: hey, can you get the apk url for bubbleupnp?
<lundman>
actually, how do you get the apk urls
ZaEarl has joined #arm-netbook
<lundman>
cool, your FileManager goes into zfs file, and plays my person.of.interest sample smoothly
mSquare has quit [Ping timeout: 268 seconds]
pawel5870 has joined #arm-netbook
<Turl>
lundman: bubbleupnp is proprietary, that'd be warez :)
<oliv3r>
mornin'
<Turl>
night oliv3r
<Turl>
ZaEarl: ohai
<oliv3r>
sleep well turl :)
<lundman>
really? even the free ad version?
<oliv3r>
any android users here by chance? got 2 extremly android specific versions :)
<Turl>
lundman: well nobody cares for the 'free ad version' on the 'distribute apks illegally' business :P
<oliv3r>
well 1 android question :p; 1 boot question
<Turl>
oliv3r: I'm one, lundman is another :P
<lundman>
yeah but I just wanted a quick way to get the upnp app to do streaming tests :)
<Turl>
lundman: I heard skifta is better btw
<lundman>
720p/1080p are playing again anyway
<lundman>
skifta?
<lundman>
ok send me that apk :)
<oliv3r>
i have a cheap clone 7" tablet, which should pretend it's a galaxy nexus 7, phone.apk IS installed, i know with CM7 (on my phone) the phone apk lets you use SIP to make calls. The tablet has phone.apk, but no icon nowhere
<oliv3r>
but if there's no dialer installed; might aswell install 3rd party sip client indeed
<shivansps>
someone knows if there are any know problem of why i cant boot a kernel never than 3.0.36 on a A10 tablet? i hear the problem where on mk802 too
<Turl>
shivansps: what are you using to boot it?
<Turl>
you might need to enable the ram hack for old bootloaders
<lundman>
odd, the bootanimation on the mele just started again
<shivansps>
you mean ATAG_MEM option? i tryied that too
<lundman>
kernel has been up 13612 seconds
<Turl>
CONFIG_SUNXI_IGNORE_ATAG_MEM = y
mSquare has joined #arm-netbook
<lundman>
ah skifta is a upnp mediarenderer, not client
<Turl>
isn't it the same thing?
* Turl
gets puzzled at UPNP buzzwording
<lundman>
I've not tried skifta
<lundman>
but bubbleupnp is for finding mediaservers on your network, then playing that media on local device
<lundman>
eyecon, mediarenderer finds other mediaserver, and clients, then you can push videos to remote devices to play
<lundman>
although, often you can also play locally
<lundman>
so I guess its a superset
* Turl
now has brain buzzworditis
<shivansps>
mmm its enabled, but no... 3.0.42 and 3.4.12 does the same thing, they power on USB devices, but there is not signal on HDMI neither it load rootfs
<Turl>
shivansps: did you enable disp, hdmi modules?
<Turl>
have you tried the other outputs?
<Turl>
I think by default it used vga?
<Turl>
lundman: probably knows better
<Turl>
if it's powering usb devices it should be fine then
<lundman>
the script.bin file says what display to use and resolution
<shivansps>
its a tablet not the mele, but im very sure it does not load rootfs because if it wheredoing so pendrives should turn on the light because it gets mounted
<shivansps>
also wifi should be connected too
<Turl>
maybe it's your mali version?
<lundman>
bootanim is soo obnoxious
<Turl>
make sure the kernel one and the one on your android matches
<lundman>
gpio changes in script.bin did not fix it :)
<Turl>
lundman: why don't you use the good script bin then? :P remember, 5 min project! :P
<Turl>
it even has faster mali clock if I saw right
<shivansps>
actually i just need some extra modules, there is some way i can build the 3.0.36 ones?
ping__ has quit [Ping timeout: 255 seconds]
<lundman>
i did, but I changed res :) I'll add that emty power thing
<Turl>
the wemac driver doesn't use it lundman, it's #if0'd
<shivansps>
well ill like to use 3.4 kernel as well :S
<Maqs>
does anybody know whether a cp2102 with rx/tx/gnd connected works as uart-to-usb for the a10? ;) i'm not quite sure about the levels
<lundman>
i know i know.. but we also left the realm of sanity 2 hours ago
<Turl>
shivansps: you should be able to git reset to the .36 hash and build whatever you need
<shivansps>
ok ill check it tomorrow them, im too tired now lol
<Turl>
Maqs: apparently it's 3.3v so it should work I'm guessing
<Turl>
confirm with the others though :P it's 4AM over here and I need sleep
sspiff has joined #arm-netbook
<Turl>
night guys
<lundman>
cheers
leszcz has quit [Read error: Operation timed out]
sspiff has quit [Client Quit]
<Maqs>
thx turl and good night
sspiff has joined #arm-netbook
sspiff is now known as ssspiff
dfletcher_ has joined #arm-netbook
ssspiff is now known as sspiff
avernos has quit [Quit: Leaving]
<mnemoc>
there are sunxi-$version-r$rev tags in the kernel tree...
dfletcher__ has quit [Ping timeout: 244 seconds]
<mnemoc>
the latest -r$rev is just before the version jump
<mnemoc>
and -r0 right after
ssspiff has joined #arm-netbook
sspiff has quit [Remote host closed the connection]
<rellla>
with the newest kernel from git master branch i get the following error, when doing a modprobe mali:
<rellla>
mali: use config clk_div 3
<rellla>
mali: clk_div 3
<rellla>
Mali: mali clock set completed, clock is 128000000 Mhz
<rellla>
after that system freeze
<mnemoc>
uhm
<rellla>
could ther be a problem with the clock.c sata fix?
<mnemoc>
and when was the last it worked?
<mnemoc>
it shouldn't affect, but...
<mnemoc>
can you test once with that change reverted?
<rellla>
i will build one with the old clock.c and report
<mnemoc>
otoh it's good enough for kernel purposes (enabling pins and features)
<shivansps>
but i use miniand images on my A10, because im lazy :P 3.0.36 works, but anything never, like 3.0.42 or 3.4.12 it gets stucks somewhere in the boot process... USB devices turns on but thats it, no hdmi signal and the system is stuck because no wifi or usb storage gets mounted
<mnemoc>
Maqs: please boot from nand and stop by u-boot
<mnemoc>
Maqs: to make a dump of boot1
popolon has joined #arm-netbook
<mnemoc>
shivansps: miniand has never cared to collaborate. no idea what problems they have or what changes they've done
<mnemoc>
shivansps: without a serial console it's hard to find what's going wrong
<mnemoc>
rellla: can you try with the original patch that started that thread?
<shivansps>
yeah i know, ill have to wait until my usb uart arrive, still it will not hurt to try with a never u-boot...
cat_n9 has quit [Quit: Communi 1.2.0.1 for MeeGo]
<shivansps>
i guess ill need a newer evb.bin, bah scritp.bin
<rellla>
mnemoc: i will do, but something later ....
<mnemoc>
shivansps: to use a newer u-boot you need a board definition on it
<mnemoc>
shivansps: what device you said?
<shivansps>
novo 7 aurora
<shivansps>
i think there is one for the elf araound
<mnemoc>
for u-boot needs, it would only work if the have exactly the same dram
<mnemoc>
shivansps: also, when replacing u-boot it's always useful to clear the env too
<rz2k>
rellla: what revision is your A10?
<rellla>
on the mele there is a little paper with V1.3. is this correct?
<rz2k>
nope, first letter of code printed on A10
<rz2k>
there is A, B and C revisions of silicone, code from A doesnt work on C (sata pll) and seems like vice versa for Mali
<rellla>
rz2k: B!
* rz2k
pokes mnemoc to do CONFIG_ for selection of revision
<mnemoc>
but revision is supposed to be detected programatically
dfletcher__ has joined #arm-netbook
<mnemoc>
my mele a1000 has B too, and boots fine :<
dfletcher_ has quit [Ping timeout: 240 seconds]
<rellla>
mnemoc: with mali enabled? mine boots fine, too. sata... ev is working
<rellla>
s/enabled/loaded/
<ibot>
rellla meant: mnemoc: with mali loaded? mine boots fine, too. sata... ev is working
<Maqs>
so i have to set dram_size and otherwise this mk802 script.fex can be compiled into script.bin, placed on the boot fat partition and it should work?
<mnemoc>
Maqs: u-boot doesn't read the script.bin file. it goes to the source
<mnemoc>
once you have a good and _verified_ script.fex, it would be awesome if you could add it to sunxi-boards
<mnemoc>
eventually u-boot will need the pmu info from it too
<rz2k>
rellla: dumb idea, but could you please try playing with mali_div @ script.bin so it will go around 400mhz?
leszcz has joined #arm-netbook
<rz2k>
if pll is stable enough, you can still live with that
<Maqs>
screen*_output_mode and _type differs
<rz2k>
my mele by default has divisor of 4 instead of 3 for some reason
leszcz has quit [Read error: Connection reset by peer]
<Maqs>
is there a way to dump the config from another image that is working?
<Maqs>
i guess one problem might be that there are mk802 with 512mb and 1gb of ram, so there have to be different versions
<shivansps>
one question, CPU clock is set on u-boot or in the scritp file?
<mnemoc>
Maqs: from nand's u-boot you can dump boot1
<mnemoc>
Maqs: and then we can extract the real dram_para from it
<mnemoc>
shivansps: the initial is set by u-boot, once running linux cpufreq takes control
<mnemoc>
shivansps: latest u-boot starts at 1GHz
<Maqs>
ok.. how do i dump it? there is an android image on nand
<mnemoc>
get into u-boot's console and...
<Maqs>
the working kernel says "DRAM: 1024<6>Total Detected Memory: 1024MB with 2 banks"
<mnemoc>
md.b 0x42400000 0x82d0
<Maqs>
ok, i'll try that
<rellla>
btw: i actually successfully play a 1080p hd-trailer @1280*1024vga smoothly with xbmc on mele :-) constant 24fps, cpu~60%
mSquare has quit [Ping timeout: 246 seconds]
<oliv3r>
did you get libcedarx to work properly?
<oliv3r>
i know empat0 (or who was it again) was having issues with 1080p h264 stuff
<Maqs>
mnemoc: wow, this gives me a lot of 0x00 :D
<mnemoc>
Maqs: :<
<mnemoc>
Maqs: that is BAD
<Maqs>
there's something at the very beginning
<mnemoc>
hno: ---^
<rellla>
oliv3r: i forked empat0's repo and did some min changes. sys is debian armel because of armel cedarx libs. kernel sunxi-master branch. everything natively compiled on mele/a10
<Maqs>
how do i install the new uboot on the uSD card? I patched uboot, compiled it for mk802 and copied uImage, but I think this is not enough, as the version did not change
cheng has quit [Read error: Connection reset by peer]
<andoma>
i wonder if the mk802 have CEC wired as well
<hipboi>
has anyone noticed that cubieboard.org had a new theme?
<RaYmAn>
andoma: no one seems to know how to access the CEC part of a10 though :P
tzafrir_laptop has joined #arm-netbook
<andoma>
RaYmAn: aww....
<mnemoc>
hipboi: yes, since last nice. looks nice :)
<mnemoc>
s/nice/night/
<ibot>
mnemoc meant: hipboi: yes, since last night. looks nice :)
mysteryname has joined #arm-netbook
<hipboi>
the Indian designer doesn't mess this time
<mnemoc>
hipboi: it only misses the shop to pre-order 2012-08-08, "next" and to buy power bricks and cases :)
<Maqs>
uboot now just shows "U-Boot SPL 2012.10-g0c38834-dirty (Oct 17 2012 - 10:55:39)".. i don't have a system.bin, as i have no idea what it is or where to get it :)
<mnemoc>
system.bin???
<mnemoc>
Maqs: clean the env in the card
ZaEarl_ has joined #arm-netbook
<mnemoc>
Maqs: so u-boot can use it's default
ZaEarl has quit [Read error: Connection reset by peer]
<oliv3r>
there, they actually split the CEC signal into an input and an output
<RaYmAn>
yeah
rellla has joined #arm-netbook
<oliv3r>
so A10 could still do it that way :p
<oliv3r>
and the circuit is really really simple; so that's not the issue
<RaYmAn>
well, assuming you can access the gpio
<RaYmAn>
:P
<oliv3r>
But! if the A10 does that in hardware, and exposes 1 CEC pin, that would be better.
<RaYmAn>
I guess one could design the board so CEC pin is hooked up to something else
<RaYmAn>
definitely
<oliv3r>
RaYmAn: taht's what I ment yea :)
<lundman>
then make TCPIP over CEC!
<oliv3r>
hipboi: could give _some_ insight, as he did connect CEC pin to P23 i think
<andoma>
on the cubieboard the CEC pin on the HDMI connceter is wired to CEC_HDMI on the A10
<andoma>
so there's got to be some support in the chip
<oliv3r>
i can't find CEC_HDMI in the manual though
<RaYmAn>
yeah, there should be support, but without a clue on how to access it, it's rather hard :/
<andoma>
yeah
<lundman>
usually they are usb devices, but occasionally /dev/ nodes for ioctl
<RaYmAn>
on tegra3, it's just a /dev node with ioctls, which then very simply just puts the raw data into hdmi-cec registers
<oliv3r>
CEC buss a USB device? in the A10, i doubt it.
Maqs has quit [Ping timeout: 246 seconds]
<lundman>
merely commenting on 80% of the device support in libcec, is for usb device s :)
Quarx has quit []
<oliv3r>
lundman: that's probably because pulse-8 has an USB -> cec bridge and wrote libcec i think
Maqs has joined #arm-netbook
hubert_ has quit [Ping timeout: 245 seconds]
<lundman>
still, i agree a10 most likely doesnt have it on usb bus
<lundman>
although wifi is there
<lundman>
maybe a lsusb would quickly rule it out
<oliv3r>
i think i did that on my tablet, and didn't find anything other then wifi
<oliv3r>
which only showed up when it was active
<lundman>
tablet has hdmi?
cheng_ has quit [Remote host closed the connection]
<oliv3r>
mini-hdmi out yeah
<RaYmAn>
tbh, given it seems to be connected to a dedicated CEC pin on the SoC, it won't be USB
<RaYmAn>
like, connected straight from cable to soc
<RaYmAn>
wifi is external to the SoC
<oliv3r>
to bad there's nothing in the manual (that i've found yet)
<oliv3r>
could be named other then CEC or hdmi though
<RaYmAn>
definitely
<mnemoc>
there might be something hidden within the hdmi driver
<mnemoc>
at least there is partial edid support
<mnemoc>
which iirc also uses i2c
<oliv3r>
if the time is right, i'll dig some into that
<mnemoc>
you are the registers master ;-)
<oliv3r>
LOL, right :p
<oliv3r>
Pin 13 is CEC and Pin 15/16 is I2C
<RaYmAn>
DDC/EDID seems to be implemented as a seperate i2c controller accessed through hdmi regs - so it's quite likely cec is similar (Except not i2c obviously)
<oliv3r>
once i'll get to that section, i'll write about it
<oliv3r>
now, after aa week hyatus, it's back to documenting the CCM :p
<oliv3r>
stupid tablet cost me way to much time :)
alcides has quit [Ping timeout: 244 seconds]
<mnemoc>
oliv3r: go by importance to hackers, not document order :)
<oliv3r>
i thought you said CCM was the first crucial bit :p
Quarx has joined #arm-netbook
<oliv3r>
new cheap nexus 7 coming from google, they'll be using WM8950 (wondermedia). I don't get it. Last I heard about wondermedia, was they weren't so opensource friendly ... :S
<oliv3r>
google should have used the A10 and used their leverage for open cedarX :D
<oliv3r>
*dream*
<mnemoc>
google needs to show they love all members of the aliance
hp__ has joined #arm-netbook
mnemoc has quit [Excess Flood]
mnemoc has joined #arm-netbook
<cat1>
doing mali update is painful.. really tempted to build it outside tree..
<mnemoc>
if we don't get r3p0/p1 armel libs from hipboi by the weekend I'll also +1 that
<hipboi>
hope i can get the cedarx hf lib
<mnemoc>
that would be even better :)
<mnemoc>
we will be able to forget about armel and r2p4 entirely
<cat1>
and switch to r3p1 completely?
<mnemoc>
r3p0 first, we don't have p1 libs for android :|
<ZaEarl_>
are the libs for android and linux different?
<mnemoc>
yes
alcides has joined #arm-netbook
alcides has joined #arm-netbook
alcides has quit [Changing host]
<mnemoc>
by "linux" we don't only refer to a kernel, but also to a certain common userspace shared by mosly all those things called "linux distributions"
<mnemoc>
ZaEarl_: the libraries are built for a very specific... environment, not only a kernel api
<RaYmAn>
mnemoc: isn't that, usually GNU/Linux when userspace is included :P
<RaYmAn>
technically speaking, linux is the kernel :P
<mnemoc>
sure
<libv>
what is up with all the files in linux-sunxi/include/config ?
<libv>
can this whole directory not be removed?
<mnemoc>
RaYmAn: but when you say I want a linux laptop you don't mean android even when android uses the linux kernel :p
<libv>
all the files in there are empty anyway
<mnemoc>
libv: include/config is not part of .git, it's a build time thing
<libv>
ah, hrm, how stupid of me then :)
<mnemoc>
:)
<RaYmAn>
mnemoc: sure, but it's still wrong :P
<mnemoc>
RaYmAn: human communication relies on what people understand by the terms, not on what the terms are defined in the dictionary ;-)
<RaYmAn>
:P
<mnemoc>
and sure, it is wrong to call the whole thing "linux", and it's also wrong to call it "gnu/linux"
<mnemoc>
but well....
<cat1>
mnemoc: can p0 libs for android work with p1 kernel counterpart?
<mnemoc>
cat1: i don't think so
<mnemoc>
cat1: but baby steps ;-)
<cat1>
mnemoc: i can't do it, i am gigantic jumper :)
<mnemoc>
"D
<mnemoc>
:D
<cat1>
mnemoc: can we keep 3.4 branch somewhat in sync with mali official releases? it should be safe enough as none is using it for graphics/video atm. Or I am completely wrong?
* cat1
is constantly loosing strategic line..
<mnemoc>
cat1: i might agree on that. I already removed allwinner boot hacks from 3.4 and that kicked out everyone booting from nand :|
<RaYmAn>
imo that's OK - goal is to make it work with open u-boot on nand, isn't it?
<mnemoc>
yes
<RaYmAn>
ffs, I hate corporate proxies *sigh*
<libv>
cat1: not this again
leszcz has quit [Read error: Connection reset by peer]
avernos has joined #arm-netbook
avernos has quit [Changing host]
avernos has joined #arm-netbook
ping__ has quit [Ping timeout: 256 seconds]
<cat1>
libv: why, i am simply seeking for the baseline from which i can start doing something. sunxi-3.4 is a good candidate. Moreover, in my ideal world i dream of working process like this: patch -- review/test -- ack/nack -- push/pull into main branch, w/o having branch multiplications.. Well, sorry for being idealistic :)
<mnemoc>
cat1: there are very little limitation on what you can do on 3.4
<libv>
cat1: with binary blobs, the world is not that simplistic
<mnemoc>
and there is tons of work to do on the rest of the kernel too, not only mali
<cat1>
libv: right, but afaiu we consider binary blobs being temporary while progressing with kernel not.
<cat1>
mnemoc: exactly
<rellla>
mnemoc: still have a freeze with minor patch
<oliv3r>
what is AHB and APB in the A10? dont' quickly see it described
<mnemoc>
rellla: please reply that one too, commenting the issue
<RaYmAn>
oliv3r: it's a standar ARM thing - different bus'es of some sort (I'm not really entirely sure how it works, but it should give a hint I guess)
<mnemoc>
cat1: work on mali is obviously slower, can need to keep userspace working.
<RaYmAn>
Advanced High-performance Bus and Advanced Peripheral Bus
<oliv3r>
snap, I just figured that out whilst not looking at IRC
<RaYmAn>
:)
<oliv3r>
though I thought it was Allwinner Host bus and Allwinner Peripheral bus
<oliv3r>
close enough
<mnemoc>
cat1: on the other edge we have the testing limitation, rellllla now has troubles with a clock change which was supposed to be tested my a large group of people
<oliv3r>
They should have done some marketing spin on that and named it as such :p
<RaYmAn>
wouldn't have been too surprising ;) But it does seem to be fairly standard :P
<mnemoc>
cat1: and know we have a monster patch (usb unification) which also needs review/testing. and I can't do it myself because of lack of time.
<oliv3r>
RaYmAn: anyway, that makes perfect sense :)
<mnemoc>
cat1: please try to target your energy where it helps more, and not were it only helps to waste time discussing things again and again :<
ping__ has joined #arm-netbook
<libv>
cat1: binary blobs are not as temporary as you'd think
<libv>
cat1: reverse engineered graphis drivers do not fall out of the sky
<lundman>
are you sure?
<cat1>
mnemoc: i am trying but desperately, too many branches too many things to follow..
<mnemoc>
cat1: no, only two main branches, the stable sunxi-3.0, and the development sunxi-3.4
<cat1>
mnemoc: what about fixes that are in next_mali?
<mnemoc>
fixes in next_mali aren't relevant to either sunxi-3.0 or sunxi-3.4
<cat1>
mnemoc: how come, fb does not work if i do not apply patch from there
<mnemoc>
which patch?
<cat1>
video: sunxi: disp-ump: Add support for buf2
<cat1>
well, not fb, but rather x
<mnemoc>
r2p4 doesn't use buf2
<cat1>
mnemoc: this is what i am talking about.. for 3.4 r2p4 is not relevant but the former is kinda broken because of that.
<mnemoc>
cat1: please
<cat1>
mnemoc: ok, ok, i'd rather stop nagging and shut up :)
<oliv3r>
Shall i start holding my hands out for those drivers falling? :D
<lundman>
wow, uh, interesting reply from github
<lundman>
"I suggest you pull in upstream changes first, then create a new topic branch from the default branch, do your change, push, and create the pull request from that topic branch instead of the commit directly."
<lundman>
what?!
<oliv3r>
huh, why?
<jinzo>
that's preatty common?
<lundman>
what, you guysunderstand that? thats unfair:)
<oliv3r>
oh, yes, after re-reading yeah
<lundman>
to me, it sopunds like it is not worth sending a 1 line fix for allwinner :)
<oliv3r>
clone, branch, pull, merge, push (i push from branch to remote/master though) pull-request!
<oliv3r>
in that case, just send the patch to the ML
<oliv3r>
someone will put it in their branch, credit you and push it
<lundman>
meh, nobody cares about making android kernels for mele here
<oliv3r>
if it's a fix to one of the linux-sunxi branches, it's a fix nevertheless :p
wingrime has joined #arm-netbook
mnemoc has quit [Excess Flood]
mnemoc has joined #arm-netbook
mnemoc has quit [Excess Flood]
mnemoc has joined #arm-netbook
<mnemoc>
znc is getting on my nerves :<
ZaEarl_ has quit [Ping timeout: 240 seconds]
mysteryname has joined #arm-netbook
<oliv3r>
does the wiki support fancy math stuff?
<oliv3r>
latex-style etc?
<lundman>
both addition AND subtraction
<jelly>
how does what-if.xkcd.com do their fancy math stuff?
<mnemoc>
iirc he use to work for nasa...
<jelly>
they seem to use something called MathJax
<mnemoc>
used*
<mnemoc>
i think he draws the stuff himself
<oliv3r>
wikipedia, or rather, mediawiki, uses something math/latex-y
<mnemoc>
oliv3r: need it for linux-sunxi.org
<mnemoc>
?
<oliv3r>
well the PLL's use some simple math
<oliv3r>
and looks prettier with math stuffs :)
<oliv3r>
i'll make do with regular ascii math
<mnemoc>
choose an extension with few deps and I can install it
<RaYmAn>
do we really need latex? It's uhm, usually not recommended for not encessarily being very safe :P
<RaYmAn>
but it might have improved
<oliv3r>
well wikipedia uses it :p
<rz2k>
<mnemoc> [16:08:17] r2p4 doesn't use buf2
<mnemoc>
rz2k: it does?
<rz2k>
I believe techn_ wrote patch that way that it is compatible with r2p4
sspiff has joined #arm-netbook
ssspiff has quit [Ping timeout: 264 seconds]
<rz2k>
ioctl for standard secure id is buf1
<oliv3r>
mnemoc: texvc should be part of the extension: math
<mnemoc>
rz2k: but needed before merging r3p0?
<rz2k>
buf2 is just a int counter and multiply of framebuffer
<oliv3r>
mnemoc: deps should be ocaml, gnu make; latex and dvips, imagemagick and ghostscript
<rz2k>
could someone with android test buf2 patch on r2p4 libs/kernel ? Turl?
<rz2k>
mnemoc: if buf2 patch will be in kernel, we can compile r3 externally for now
<mnemoc>
rz2k: fair enough. if it doesn't harm and all agree it's good. give me the hash I can push it
<rz2k>
lets test first
<rz2k>
android guys can use it
<mnemoc>
ok
hipboi has quit [Quit: hipboi]
wingrime has quit [Quit: leaving]
pawel5870 has quit [Read error: Connection reset by peer]
wingrime has joined #arm-netbook
<mnemoc>
oliv3r: will poke you after the ext. is installed
sspiff has quit [Ping timeout: 246 seconds]
sspiff has joined #arm-netbook
hipboi has joined #arm-netbook
<wingrime>
what status have 3.4 kernel?
<rellla>
mnemoc: problem fixed for me. send patch to ml.
<oliv3r>
mnemoc: rgr
drachensun has joined #arm-netbook
<oliv3r>
mnemoc: did a save for now so my data should be in the db :)
pawel5870 has joined #arm-netbook
<oliv3r>
i have to take my VM offline for a short while, i'm not editing anything atm and have saved everything.
<oliv3r>
will check back later
ping__ has quit [Quit: Leaving]
mysteryname has quit [Ping timeout: 240 seconds]
ol1ver has quit [Quit: ol1ver]
oliv3r has quit [Quit: Leaving.]
<rellla>
is there a reason, why the following lines are in cubieboard.fex but not in my mele_a2000.fex:
<rellla>
[target]
<rellla>
pll4_freq = 960
<rellla>
pll6_freq = 960
<rellla>
power_start = 0
<RaYmAn>
a guess would be that one is 'decoded' from script.bin (mele) while the other is the original "source" file (cubieboard)
<RaYmAn>
We already know that some values are coded into e.g. boot0/boot1 instead of written to script.bin
pawel5870 has quit [Quit: Leaving.]
<rellla>
ok. so could there be a relation for my issue with the mentioned sata fix?
pawel5870 has joined #arm-netbook
<wingrime>
mnemoc: I send fist patch
arokux has quit [Remote host closed the connection]
<mnemoc>
wingrime: thanks!
arokux has joined #arm-netbook
arokux has quit [Remote host closed the connection]
<libv>
so there is no chinese speaker with time to burn here?
arokux has joined #arm-netbook
<mnemoc>
try #chinalug
<mnemoc>
libv: ----^
hipboi has quit [Remote host closed the connection]
<libv>
that'll be a tough sell :)
<mnemoc>
long text?
rz2k has quit [Ping timeout: 245 seconds]
<mnemoc>
google translate does a pretty decent job for single sentences
sspiff has quit [Remote host closed the connection]
arokux has quit [Remote host closed the connection]
pawel5870 has quit [Ping timeout: 246 seconds]
rz2k has joined #arm-netbook
mSquare has quit [Ping timeout: 256 seconds]
arokux has joined #arm-netbook
arokux has quit [Remote host closed the connection]
<libv>
just many bits of it
drachensun has quit [Quit: Leaving]
<wingrime>
how about to make own SPL (boot.axf) ?
<wingrime>
GLP based ofcouse
<L84Supper>
libv: is it all highly technical?
<libv>
L84Supper: how would i know, i do not speak chinese :p
<libv>
L84Supper: but i guess so, yes, but i am going to be working with $chinese_volunteer on making sense from it
<L84Supper>
libv: ah I thought you needed it translated into Mandarin vs out of
<wingrime>
I have dram_clk = 360 , if I set 480 will it work ?
<techn_>
rz2k: you got xbmc running? :D
<rz2k>
techn_: yep
<techn_>
what was the trick.. or is there some bug fixed to r3p1
<rz2k>
with libUMP recompiled with your fixes
<techn_>
I'm using r3p0 and it wont start :(
<techn_>
chrashes with that callstack I paster yesterday
<techn_>
*pasted
<rz2k>
I have unity and plasma-active crashing though
<rz2k>
so there is some missing stuff there
<techn_>
oh so you are using other version of *ubuntu
<techn_>
I'm using linaros ALIP version
<techn_>
Or do you?
<rz2k>
no, linaro-alip
<techn_>
and my ALIP is from july
<rz2k>
I've just installed plasma-active & untiy from repo
<rz2k>
mine is updated to latest packages
<techn_>
will it update with upgrade/dist-upgrade?
<techn_>
Atleast for ubuntu there is some hackery needed for to get newer version
<rz2k>
yep
<rz2k>
it will update
<techn_>
oh.. then we must have same ALIP then.. only difference is that you are using r3p1 and I r3p0?
<rz2k>
probably
<rz2k>
also there was a thing that I needed to do: somehow before I've killed all mesa/mesa-egl swrasts including OpenGL (not ES) one. in this situation xbmc dies with libGL not found, reinstalled mesa and now it works
<rz2k>
also I have debugging off in my kernel
<rz2k>
with ARM's awesome libraries and code this might be mandatory.
<Turl>
000014bc g DF .text000001a0 ump_secure_id_get
<rz2k>
Turl: main problem is that we dont know if ump_secure_id_get is exactly same as ump_secure_id_buf1, they should be same, but it needs a test.
popolon has quit [Quit: Quitte]
<traeak>
thanks sorry
shivansps has quit [Quit: Page closed]
ceo16 has quit [Ping timeout: 245 seconds]
mnemoc has quit [Excess Flood]
mnemoc has joined #arm-netbook
<mnemoc>
any suggested alternative for znc that doesn't die when you have too many open channels?
<techn_>
I have been using screen+irssi
<mnemoc>
screen+irssi kind of sucks from the phone :|
<rz2k>
have a wide channel, up to date znc ? mine dies only when server crashes. I have 3 networks added with 10 channels per network.
<mnemoc>
0.206-1 from ubuntu
<jinzo>
mnemoc, znc is rock solid for me
<rz2k>
I'm sure there is ton of updates from version that is in ubuntu
<jinzo>
currently 5 networks and ~15 channels - no problems.
<rz2k>
check their github
<mnemoc>
206 is the latest release, but 7M old
<jinzo>
and it's easy to compile.
<mnemoc>
rz2k: what do you mean by "wide channel"?
gimli has joined #arm-netbook
<rz2k>
I mean that znc can possibly fail when you have bad connection with many packets dropped
<mnemoc>
nah, it's by "flood"
<rz2k>
oh, new znc in archlinux testing by the way :p
<Turl>
mnemoc: rock solid znc over here too
<Turl>
mnemoc: 3 networks for a total of 14 channels autojoined
<mnemoc>
my problem is with freenode in particular, 18 channels.
<Turl>
mnemoc: maybe lower MaxJoins config?
<Turl>
mine is set to 5
<mnemoc>
mine is 2 :<
<Turl>
mnemoc: do you have ctcpflood enabled?
<mnemoc>
no module at all... 1m
<Turl>
maybe znc is sending you the ctcp queries and your client feels like answering all of them and flooding :<
<mnemoc>
the flood is not on the client side, but on znc itself
<mnemoc>
znc gets disconnected, not me
<cat1>
mnemoc: can we have some place where kernel+modules can be uploaded for testing? it would be good to supply each patch series with binaries for those who does not compile stuff themselves.
<mnemoc>
cat1: dl.linux-sunxi.org/something ?
<cat1>
mnemoc: why not
<mnemoc>
give me your pub key
<cat1>
mnemoc: like testing/sunxi-3.0 testing/sunxi-3.4
<techn_>
static struct cpufreq_frequency_table *sun4i_freq_tbl = NULL; .. should this be combined to static struct cpufreq_frequency_table *sunxi_freq_tbl = sunxi_frequency_table(); ??
pawel5870 has joined #arm-netbook
<Turl>
techn_: static requires a constant initializer
<techn_>
ah.. you have used it some other place..
<techn_>
but it's not used everywhere
<hno>
Turl, have not seen those clock ranges before.
<mnemoc>
afaik wiki values come from the datasheet/manual
<mnemoc>
which not necesarily match reality :|
<hno>
don't copy stuff from the manual please.
n6pfk has quit [Read error: Operation timed out]
mikey_w has quit [Ping timeout: 246 seconds]
tmerle has quit [Quit: Leaving.]
tmerle has joined #arm-netbook
functionbeforefo has quit [Ping timeout: 260 seconds]
functionbeforefo has joined #arm-netbook
mikey_w has joined #arm-netbook
n6pfk has joined #arm-netbook
cat1 has quit [Quit: Leaving.]
<techn_>
Turl: How I'll see effect of your patch?
<Turl>
techn_: if you see any effect, it's a bug
<Turl>
make sure you have cpufreq enabled
lkcl has joined #arm-netbook
<techn_>
bogomisp while idle 60, bogomips while top/htop 60 (cpu 30%), bogomips while es2gears 1000 (cpu 85%)
<Turl>
techn_: this doesn't change anything regarding the cpufreq logic
<Turl>
just the magic to switch between them
<techn_>
so it should be more responsive?
<Turl>
no
<Turl>
it should be the same
<Turl>
the logic that makes it check load, notice it's high, therefore go 60->1000 or whatever is unchanged
<Turl>
my patch changes the stuff that actually does the switch
<Turl>
it's lower level
<mnemoc>
techn_: it's a cleanup
<Turl>
that pretty much summarizes it :P
<mnemoc>
everything should work just as always, but with an implementation simpler to maintain, extend and improve in future
cat1 has joined #arm-netbook
<techn_>
ok.. no difference seen.. yet :)
cat_n9 has joined #arm-netbook
popolon has joined #arm-netbook
<hno>
techn_, that's good.
ooo_ has quit [Quit: Page closed]
gimli has quit [Quit: Verlassend]
<mnemoc>
cat_n9: does your patchset consider the commit from a13-sdk we had to merge?
alcides` has joined #arm-netbook
alcides` has quit [Remote host closed the connection]
<mnemoc>
cat_n9: or that goes after the refactoring?
<Turl>
mnemoc: I think he said he thought it was better after cleaning up the current stuff
gzamboni has joined #arm-netbook
<mnemoc>
Turl: thanks :)
<libv>
f me.
<libv>
i just pulled one of the OSAL files through google translate.
<Turl>
OSAL?
<libv>
"NO SHIT SHERLOCK" is the phrase that popped up in my mind every few seconds
<mnemoc>
also, there is a patch to fix the sata fix to not break mali in A10-revB
<libv>
this will make for a time and nerve wrecking commit, which every one will have a really nice laugh about
<mnemoc>
libv: OSAL removal deserves a monster commit
<libv>
what is OSAL anyway?
<libv>
ah
<libv>
os abstraction layer
<libv>
f me.
<libv>
no wonder anyone translated this crap yet :)
<mnemoc>
to use their stuff in their rtos, on bare metal, on u-boot, on linux, on wince, ...
ibrah has joined #arm-netbook
<libv>
yeah
<Turl>
they have a RTOS? o.O
<libv>
it hit me as i saw my "what is OSAL anyway?" line hit the channel
<mnemoc>
:)
<mnemoc>
Turl: yes, they have their own closed rtos based on another well known rtos.... it's mentioned in the mail tom sent to the u-boot list
<mnemoc>
don't remember the names
<libv>
this is just bad busywork...
<Turl>
VxWorks
<Turl>
?
<Turl>
(is that an RTOS?)
<libv>
Turl: well... it might be realtime certified.
<libv>
and the mother company might have been one of the companies defining the certification protocol
leszcz has joined #arm-netbook
<libv>
don't go near automotive companies... you lose the will to live.
* mnemoc
looks at his copy of misra c
<libv>
mnemoc: osal removal does not get a monster commit, it still needs to be reviewable.
<lkcl>
osal comes up in the cedarx library. one of the files
<lkcl>
the c source is available for it.
<mnemoc>
there are copies of osal on most drivers
<Turl>
probably legacy from sun3i :)
<libv>
but there will be a big "all these functions only return 0 anway" commit
<mnemoc>
lkcl: we are talking about removing it :)
tmerle has quit [Ping timeout: 246 seconds]
<hno>
OSAL is operating system abstraction layer or similar acronym.
<lkcl>
joy!
<lkcl>
mnemoc: that wouldn't impact libcedarx from working, would it?
<mnemoc>
and about the comments on them
<lkcl>
libcedarx uses osal to initialise the cedarx hardware (via ioctls)
<mnemoc>
lkcl: libcedarx has it's own copy. not affected by removing abstractions layer within the drivers
<lkcl>
s/it's/its
<libv>
lkcl: will just be about libc wrapping i guess
<jelly-home>
s#s/it's/its#s/it's/its/# </pedant>
<mnemoc>
ok. sorry for the typo
<lkcl>
i actually saw someone use "its" where "it's" was supposed to be used. first time, last week, i fell over backwards
<lkcl>
first time in 16 years of being on the internet
<lkcl>
anyway
<Turl>
wha?
<Turl>
first time? :|
<libv>
mnemoc: if only the osal authors knew as much english as you do :)
<lkcl>
Turl: yeah. most people get it wrong the other way round.
<libv>
wait, then they would've been clued and they wouldn't have written osal
<lkcl>
libv: :)
<Turl>
libv: it's probably a 'we have this old sun3i code, need to ship sun4i the cheapest, what do we do? REUSE ALL THE THINGS!11'
<libv>
Turl: they just did what ATI does for every new radeon revision...
<Turl>
cedarx for example, it's hacked into android as a media player
<libv>
copy tree of closest revision, hack it, throw it on the pile, start again
<Turl>
when the right, future proof way is OMX
<mnemoc>
Turl: that's the normal way of "the industry"
<Turl>
intel rewrites its drivers yearly pretty much :P
<libv>
Turl: no, intel is different
<libv>
Turl: they have dozens of teams competing against eachother, each with a slightly different focus and absolutely no visibility of what the rest of the company is doing
<libv>
there is no big pile
<libv>
just tons of small piles that are broken in different ways, and noone can help you fix them
<libv>
maybe osal is not the worst thing on the planet :)
mysteryname has joined #arm-netbook
<hno>
the previous os (which is still running in boot.axf) was appraently uCOS based.
ibrah has quit [Ping timeout: 245 seconds]
<hno>
and boot.axf, u-boot and kernel all share the same drivers mostly.
<hno>
boot1 also.
<mnemoc>
libv: not sure which copy of osal it was, but there is an osal_malloc() that always return NULL
<Turl>
lol
<Turl>
nice implementation of it
* Turl
just noticed cpufreq commit date
<Turl>
5 months ago :O
<mnemoc>
check authoriship date
<mnemoc>
5 months ago I stoped rebasing the 3.0 branch
<Turl>
mnemoc: Date: Thu, 31 May 2012 17:57:20 +0000
* mnemoc
scratches his head
<Turl>
time flies
<mnemoc>
it's october and the first public proto of the cubieboard is from august
* mnemoc
doesn't remember when the eoma68 thing started
<mnemoc>
2y ago?
ibrah has joined #arm-netbook
functionbeforefo has quit [Read error: Connection reset by peer]
functionbeforefo has joined #arm-netbook
ibrah has quit [Ping timeout: 246 seconds]
revident has joined #arm-netbook
functionbeforefo has quit [Ping timeout: 245 seconds]
hg_5 has quit [Ping timeout: 246 seconds]
<leszcz>
i tried to build u-boot with : make sun4i CROSS_COMPILE=armv7a-hardfp-linux-gnueabi-
<leszcz>
but it failed : error: /usr/lib/gcc/armv7a-hardfp-linux-gnueabi/4.7.2/libgcc.a(bpabi.o) uses VFP register arguments, u-boot does not
<leszcz>
after : git checkout 10664cbea6575bf77259f4e4db5bb566540b1961
<leszcz>
compilation went successfully but there is no sunxi-spl.bin in spl directory