<lundman>
study it, and you can help me work out why my zfs crypto panics :)
debianin has joined #arm-netbook
mysteryname has joined #arm-netbook
rz2k has joined #arm-netbook
rellla has joined #arm-netbook
<pucko>
sometimes I see "/dev/nandX" (where X is a-z) and other times "/dev/mtd/mtdN" (where N is 0-9).. what is the difference?
Quarx has joined #arm-netbook
<lundman>
nand is allwinners own "mtd" device
<lundman>
and the X is just the partition that it has been carved into
<lundman>
which is the same as N in mtd
mysteryname has quit [Ping timeout: 244 seconds]
popolon has joined #arm-netbook
hipboi has joined #arm-netbook
MindBeat has quit [Quit: Leaving.]
<rellla>
i can confirm, that the new armhf-cedars seem to work. without A10HWR=1 -> 14fps, with enabled stable 24fps. bogomips ~ 333, cpu ~ 50%.
<rellla>
not all formats/ encoding profiles might work that smooth yet.
<rellla>
tested on debian/unstable-armhf with latest stage/sunxi-3.0(without luc's commits), r3p0(armhf/fb)-mali-libs from github/linux-sunxi and new cedarx-armhf. just for your info ;-)
<rellla>
mnemoc: and new sata-fix works for my rev-b, too.
Thomas42 has quit [Ping timeout: 245 seconds]
mysteryname has joined #arm-netbook
steev has quit [Excess Flood]
Thomas42 has joined #arm-netbook
sspiff has joined #arm-netbook
<hno>
rellla, great! On both cedarx and sata!
Thomas42 has quit [Ping timeout: 244 seconds]
AndChat138129 has joined #arm-netbook
Thomas42 has joined #arm-netbook
mysteryname has quit [Read error: Connection reset by peer]
mysteryname has joined #arm-netbook
AndChat138129 has quit [Quit: Bye]
AndChat138129 has joined #arm-netbook
CaCtus492 has quit [Remote host closed the connection]
CaCtus491 has joined #arm-netbook
AndChat138129 has quit [Client Quit]
mSquare has quit [Ping timeout: 244 seconds]
Necrosporus has quit [Ping timeout: 260 seconds]
pawel5870 has joined #arm-netbook
AndChat138129 has joined #arm-netbook
arnd has joined #arm-netbook
mSquare has joined #arm-netbook
<mnemoc>
rellla: thanks :)
<rz2k>
probably time to make first linux-sunxi rootfs image based on something
<rz2k>
with xbmc, mali and cedar.
arnd has quit [Ping timeout: 260 seconds]
<oliv3r>
yeah, that'd be awesome :)
<mnemoc>
rellla: did you see any performance impact in using armhf/cedarx ?
<mnemoc>
rellla: compared to the armel flavour
dodger_one has joined #arm-netbook
<Maqs>
i'll try that here in a minute, i have an armel chroot with xbmc and armhf with xbmc i just have to configure it and reboot, as for some reason plugging out my usb sound device "killed" usb
<Maqs>
and by killed i mean "Unable to handle kernel NULL pointer dereference at virtual address 00000004"
<mnemoc>
:<
<Maqs>
wifi still works, everything new i plug in is ignored :)
dodger_one has quit [Client Quit]
gimli has joined #arm-netbook
<Maqs>
armhf looks fine.. between 30 and 38 fps, 1080p playback
arnd has joined #arm-netbook
<Gumboot>
Has language evolved out from under me again? What's microcode?
<phh>
cpu's firmware usually
<Gumboot>
Is there a more liberal interpretation?
<Maqs>
with 912 mhz
<Maqs>
and around 75% cpu usage
<orly_owl>
Gumboot: a small blob of code loaded onto a chip by software
<orly_owl>
as part of initialisation
<Gumboot>
In no way distinct from firmware, then?
<oliv3r>
it basically IS firmware
<oliv3r>
intel and AMD cpu's have microcode too
<phh>
firmware can be on an eeprom
<orly_owl>
yes, it is firmware really
<Gumboot>
I mean, I tried wikipedia to see how things had changed, but that still seemed pretty specific.
<oliv3r>
I wouldn't be supprised that microcode in said CPU's is volatile
<Maqs>
btw, is there a way to make dlna playback more fluent in xbmc? it stutters..
<orly_owl>
but firmware often stays on a device
<oliv3r>
and the bios supplies the CPU with current microcode
<Gumboot>
Why would the bios itself not be microcode?
<orly_owl>
because it isnt loaded by software during boot/initialisation
<oliv3r>
the bios is 'firmware' :)
<orly_owl>
it stays on a flash chip
<mnemoc>
microcode is used for smaller pieces of firmware :p
<orly_owl>
aye
<oliv3r>
but the firmware (bios) also contains a little blob (microcode) that gets uploaded into the CPU
<Gumboot>
It never used to. It used to be a performance optimisation to copy BIOS code into RAM where it would be faster.
<Maqs>
btw, is there a way to get rid of the blinking cursor? i don't know why it's still there in XBMC :D
<Gumboot>
Even these days I think it's decompressed into RAM.
<oliv3r>
that's different use of the word microcode i guess
<oliv3r>
generally speaking, microcode refers to the volatile 'firmware' in the CPU
<Gumboot>
So the BIOS _is_ microcode, then?
<oliv3r>
which the bios uploads there. the CPU works perfectly fine without the microcode
<oliv3r>
think of microcode as a 'patch' to the firmware of the CPU
pawel5870 has quit [Ping timeout: 260 seconds]
<oliv3r>
it is there so small bugs can be fixed in the CPU
<Gumboot>
So the _size_ of the code determines whether it's firmware or microcode?
<mnemoc>
and the volatility
<oliv3r>
firmware is non-volatile; micccrocode is volatile
<Gumboot>
Historically a lot of "firmware" has been held within kernel drivers and uploaded at boot time.
<Gumboot>
That's all microcode?
<mnemoc>
yes
<oliv3r>
by that definition however, device firmware, that lives on a harddisk
<Gumboot>
And, of course, the BIOS itself being decompressed or just copied into RAM makes its executable image volatile.
<Gumboot>
And even microcode must _come_ from somewhere non-volatile or it wouldn't be accessible at start-up.
<mnemoc>
but you don't need to push it from userspace :)
<mnemoc>
still a very weak distinction
<Gumboot>
It sounds like another redundant word, now.
<mnemoc>
microcode (as /lib/firmware) is pushed from userspace, every time
<mnemoc>
and tends to be smallish
<Gumboot>
Wikipedia seems to treat it as fairly specialised, still.
<oliv3r>
actually, the only microcode i know, is the 'blob' that gets loaded into the CPU
<oliv3r>
everything else is called firmware
<oliv3r>
to make it more confusing, firmware is being executed by the CPU on the device it is being uploaded too :)
<oliv3r>
maybe that's the difference? Firmware is the 'OS' FOR the CPU
<Gumboot>
Well, except for the bit where they say that some people use them interchangeably; but it seems to be a fairly sharp switch at that point.
<oliv3r>
while microcode is firmware that the CPU requires to function?
<mnemoc>
the idea of microcode as runtime firmware patches sounds better
<oliv3r>
i guess one could think of microcode of the binary data required to program an FPGA, whereas firmware is as said the 'program' the cpu executes
<oliv3r>
maybe it's just a basterdization of words that has stuck around since the early days?
<oliv3r>
cubieboard is gonna reach 50k i'm sure
<oliv3r>
it's almost halfway, with only 2 days done
<jelly-home>
orly
<mnemoc>
oliv3r: and the 4k of the notify me list haven't been notified yet
gimli has quit [Quit: Verlassend]
gimli has joined #arm-netbook
<oliv3r>
wb
<Gumboot>
I'm on that list! Why haven't I been notified?
<Gumboot>
I demand answers. And notifications.
<orly_owl>
heh
<jelly-home>
Gumboot: tegra tegra tegra
<oliv3r>
pay me 70 USD and you will be heard!
* jelly-home
scares Gumboot
<mnemoc>
Gumboot: tom said he will send mails today. was waiting for paypal's verification... but that might take too long
<orly_owl>
do we like cubieboard?
<oliv3r>
so still no paypal?
<oliv3r>
we love cubeboard
<mnemoc>
Gumboot: big problem is how to send those 4k mails without been blocked by google apps
<orly_owl>
i see
<mnemoc>
oliv3r: right, still no response from paypal
<hno>
CPU "microcode" as seen by mortals is patches to the firmware embedded in the CPU, written in CPU generation specific microcode instructions.
<oliv3r>
mnemoc: i wonder what happens if he'd relay those mails via my mailserver :)
<hno>
Details of the microcode is only known by the CPU manufacturer.
<mnemoc>
oliv3r: mail him and suggest it
cheng has joined #arm-netbook
<oliv3r>
hno: how's that different from firmware that is only known by the device manufacturer?
<mnemoc>
oliv3r: i doubt @cubieboard.org has dkim or similar
<hno>
Device firmware is generally program code running on some embedded CPU, usually MIPS or ARM.
<oliv3r>
i'll mail him
<mnemoc>
hno: or avr, or... :)
<Gumboot>
Was thinking about sending some cubieboards to people, but then I realised anyone in the UK would find them a bit of a white elephant because the courier ransom would cost almost as much as the board.
<mnemoc>
Gumboot: they have a 10x pack with case and uart over dhl for 700
<Gumboot>
hno: Microcode is firmware written for an undocumented architecture?
<mnemoc>
Gumboot: you can import that 10x and then redistribute
pawel5870 has joined #arm-netbook
<Gumboot>
Actually, DHL aren't dicks about the fees.
<oliv3r>
mnemoc: done
<mnemoc>
thanks to parcelforce the 10x should be cheaper and faster than 10 x 1
<hno>
oliv3r, you can't write programs in microcode, not even the CPU manufacturer can. It's small very hardware dependent controls.
<ZaEarl>
Firmware executes on a processor. Microcode changes how the processor works.
<mnemoc>
still a weak definition...
<oliv3r>
ZaEarl: i think that's the best difinition we have so far :)
<hno>
Yes it's weak.
<ZaEarl>
there's always grey areas
<oliv3r>
though 'on a' should probably be 'by a'
<hno>
Not all firmware executes on a processor. Some are only data tables.
<Gumboot>
Unless you're one of these hardware vendors who uses the term as a synonym for firmware.
<oliv3r>
but one can always think of microcode then as being the CPU's OS :p and it still is executed by the 'cpu' :D
<hno>
oliv3r, yes, an OS where almost everything hardwired in the silicon (not even ROM) with some very limited room for patching.
<oliv3r>
which in turn, makes it called firmware again :p
<oliv3r>
but ZaEarl's naming was the best so far I guess
<Gumboot>
What about microcode being uploaded to something which isn't a CPU?
<oliv3r>
but the rpi, one could call it microcode, you 'patch' it to make the entire thing even work, yet its still firmware!
<Maqs>
let's just call it microware or firmcode.. :p
<lundman>
magic?
<oliv3r>
secret sauce
<hno>
ol1ver, the gpu fimware for R-Pi Is firmware. The GPU is a general purpose computing device with some hardware acceleration for common GPU operations, running it's own general purpose RTOS.
<oliv3r>
well gumboot did ask for soemthing that is NOT a cpu!
<hno>
you could join ranks with some hardware vendors and call it microcode if you like, but imho is a misrepresentation.
<hno>
ol1ver, The GPU is a CPU imho.
<Maqs>
anyone an idea what causes this? http://dl.dropbox.com/u/11244516/usb-oops.txt it happens after i unplug my usb sound device (that is not even recognized from one of the usb ports (mk802). the other port works fine. also, it won't detect any device on that one port
<oliv3r>
what if their RTOS is contained in the firmware? you patch said rtos in the firmware, is it not microcode then?
<hno>
No. It does not change the operations of the CPU at instruction level.
<oliv3r>
true
<oliv3r>
ah, the firmware for said GPU, could contain microcode for the [CG]pu though :D
<hno>
unless ofcourse you abstract the GPU programming to the OpenGL ES level. In which case the firmware do become microcode in a sense.
<Gumboot>
Well there it is, then.
<oliv3r>
Gumboot: confused more now?
<Gumboot>
Because I mostly write application code, the kernel is abstract to me, and I can regard it as microcode.
<oliv3r>
microcode for a microkernel :D
<Gumboot>
That's no microkernel.
<hno>
Not really. You do not feed the kernel with instructions.
<oliv3r>
lundman: do you post on the xbmc a10 thread too?
<Gumboot>
Historically I've always subscribed to the way hno puts it, but it's becoming so common to call any old crap microcode, now, I was wondering if the word was just redundant, now.
<oliv3r>
Gumboot: depends on what kernel you use! you haven't stated so ;)
hipboi has quit [Ping timeout: 264 seconds]
<oliv3r>
Gumboot: i think to sumarize, no, microcode is a patch that changes a cpu (upto it's instructional level) whereas firmware is code the CPU executes
<Gumboot>
oliv3r: Unless you're in marketing.
<Gumboot>
Or IBM.
<oliv3r>
Gumboot: lol
<lundman>
sure
<hno>
Gumboot, there is many more levels of computing these days than what we have seen before.
<oliv3r>
lundman: ah then xbmc lundman is you :)
<lundman>
you bet
<lundman>
are you sqaying there are mor ethan one lundman?!
<oliv3r>
possibly!
<lundman>
wha.!
<lundman>
so we compiled ne wxbmc with new libs and uploaded bin to me?
<mnemoc>
Maqs: 3.0 or 3.4?
<oliv3r>
lundman: will mele a1000 live to your xbmc requuirement with the latest developments?
hipboi has joined #arm-netbook
<lundman>
i need JB, if it can only do 2ch audio, thats not ok
<oliv3r>
lundman: (yes i'm backreading about 12 pages)
<lundman>
Turl gave some weak assistance there :)
<oliv3r>
ah, so you want the android xbmc version
<lundman>
well, it is my hope that Google store will be better filled with tv and movies for rental/purchase in future
<lundman>
so, android with xbmc would be the way forwards
<oliv3r>
ah, I see
<lundman>
but a mediaplayer needs to do 5.1 at least, DTS HD etc for nice
<orly_owl>
says you
<lundman>
not sure what you are arguing
<lundman>
he asked my opinion, I gave it
<orly_owl>
k
<lundman>
*I* think a mediaplayer needs to do 1080p, and more than stereo
<lundman>
if you are happy with less, then just watch your tv on a laptop
<orly_owl>
will do!
<lundman>
sure I understand you guys have shitty internet and all that ;)
<orly_owl>
hold on
<lundman>
you stab, I stab!
* orly_owl
turns monitor upside down
<orly_owl>
ah
<orly_owl>
yes, we do
<lundman>
hmm france is half ok, but you do have some cooky new laws
<lundman>
but hell, so do we now :(
<orly_owl>
where do you think i live >_>
<lundman>
your whois said france, I assumed you were skilled enough to pick an ircd close to you :)
<orly_owl>
skilled, yes. not lazy? no
<lundman>
yeah nothing beat lazy
<orly_owl>
^^
<oliv3r>
we love lazy :)
<orly_owl>
ive given you a clue, you work out where i live :3
<lundman>
upside down, australia!
<orly_owl>
\o/
<oliv3r>
lol
<lundman>
I lived in NZ for 7 years :)
<oliv3r>
NL > all
<lundman>
oh I've been to NL, smokers bus from london
<orly_owl>
smokers bus?
<orly_owl>
how fast is NL internet
<lundman>
bus drives from london to NL, spend 12 hours there then back
<lundman>
gives everyone a chance to "stock up"
<orly_owl>
oh
<lundman>
i have mates at xs4all, so nl is good
<orly_owl>
uk must have a massive ciggie tax then
<orly_owl>
*england
<lundman>
or weed is illegal
<orly_owl>
oh lol
<orly_owl>
k
<orly_owl>
xs4all?
<lundman>
i dont remember the return journey at all!
<orly_owl>
lollllll
<lundman>
xs4all.nl? a very large isp in NL
<orly_owl>
ah
<orly_owl>
sounds like excess for all
<mnemoc>
defining .nl?
<lundman>
netherlands
<lundman>
:)
<oliv3r>
i have 100MiB fiber sync; my provider is rolling out 500/1000MiB in certain area's of eindhoven
<lundman>
i have 100M dfiber, no limits no caps
<oliv3r>
eventually i should get that too, do not know when though :)
<lundman>
but yeah, I should upgrade
<lundman>
$35/months
<mysteryname>
.. That interent is so cheap =/
<orly_owl>
where do you people live ;_;
<oliv3r>
xs4all is not a very large ISP, it was one of the first-ish ISP's that was 'awesome' it also started out by a group of hackers
<orly_owl>
even nbn plans have caps here
<mysteryname>
$70AUD/mo 30/1 mibit 300G a month.
<oliv3r>
they still run lots of bsd/linux servers and are actually knowledgable
<Gumboot>
Who makes up these cubieboard 'perks'? Is a bulk-discount-on-bare-bones perk possible, or is that cutting into a margin that doesn't have room for it?
<oliv3r>
yeah, the first 5 perks seem reasonable; the other ones are ill described
<Gumboot>
Well, I'm not critical. If you had academic use in mind you'd want those boxes.
<mnemoc>
Gumboot: the first boards were sold under cost
<Gumboot>
But I don't.
<mnemoc>
the others are normal price. $49 + shipping
<Gumboot>
Actually, I'm highly critical. What I meant to say was "I'm not criticising".
<rellla>
mnemoc: i'm telling myself, that armhf is quite more performant ;-) but have no significant benchmarks. i have the same system on a armel sd-card with cedarx, old sunxi-3.0 and mali-libs-r2p4 and can deliver some values, if wanted. how should i proceed, what benchmarks are useful?
<oliv3r>
rellla: what are you using to benchmark?
<mnemoc>
rellla: don't know... i'm happy with the confirmation that is feels "quite more performant" :)
<Gumboot>
mnemoc: It's not immediately clear to me how I amortise the shipping across several units without buying the plastics.
<mnemoc>
Gumboot: mail tom about a not-rich 10x, just boards and dhl
<rellla>
oliv3r: don't know, i haven't done yet. what should i use?
ssspiff has joined #arm-netbook
<oliv3r>
the big buck bunny of course :) i'll find you a link
<rellla>
oliv3r: and where do i get useful values? fps? bogomips?
<Gumboot>
mnemoc: Well, at a unit price of $49, I would expect to start at $490 and add shipping to that. I see a lot of room...
<mnemoc>
Gumboot: true :)
<mnemoc>
Gumboot: send him an email
<mnemoc>
to hipboi cubieboard org
<Gumboot>
There's a hipboi right here!
<mnemoc>
mail is probably better
<Maqs>
mnemoc: any idea what could cause the usb problem?
gimli has quit [Quit: Verlassend]
<Maqs>
even after a reboot devices on that usb port are not found
<Maqs>
surprisingly, unplugging the sound device causes that oops, although it is never detected
<oliv3r>
short circuit?
<oliv3r>
over current protection kicking in?
<Maqs>
rather unlikely, as all the devices work on the other port
<Maqs>
and the mk802 is closed, so i cannot imagine how a short circuit could be possible
<mnemoc>
Maqs: haven't looked at the related code yet, but it's clearly a driver bug
<Maqs>
ok, thanks. i'll check again with 3.0.26 later, it worked the last time
<oliv3r>
which SoC has the biggest community right now you'd say? (of the cheap media playe rkind ones) I would say linux-sunxi with a10 seems by far the biggest?
<lundman>
r^pi
<Maqs>
rpi, yes
<lundman>
but it'll never work :)
<oliv3r>
Maqs: my dad once had chewinggum 'foil' stuck in the USB connector
<oliv3r>
(on an usb stick from his pants pocket)
<oliv3r>
burned out the USB stick and disabled all usb ports until you rebooted :)
<Maqs>
:-)
<oliv3r>
but maybe the audio stick draws too much power?
<mnemoc>
Maqs: if it used to work, can you try b9b33cef810cbcf68dd198e85ecb31f455a0ede5 ? before the usb driver unification
<Maqs>
my usb mouse doesn't work with that port as well
<oliv3r>
rpi will remain closed however :p so what is the SECOND biggest community (of hackers working on kernel code etc)
<oliv3r>
mice draw some power too
<Maqs>
mnemoc: ok, but later, i gotta go now. i'll be back in 3-4 hours
<mnemoc>
hipboi: i thought it was rockchip or amlogic... interesting
<oliv3r>
oh hey hipboi :)
<mnemoc>
Maqs: test the pre-unification hash first please
<oliv3r>
hipboi: i've send you e-mail with regards to emailing the indiegogo folks
<Maqs>
mnemoc: i already tried 3.0.36, doesn't seem to work there
<mnemoc>
oliv3r: s/indiegogo/notify me/
valhalla_ is now known as valhalla
<mnemoc>
Maqs: ouch, and older did?
sspiff has joined #arm-netbook
<oliv3r>
when will allwinner release a press-release about their new chip? weeks? months?
<mnemoc>
oliv3r: still secret
<oliv3r>
months then p
ssspiff has quit [Ping timeout: 244 seconds]
<Maqs>
lsusb takes 2.5 seconds and does not list anything plugged in that port
<Maqs>
but i get an oops every time i pull the usb plug
<mnemoc>
review the corresponding section in the script.bin
<Maqs>
(after booting)
<mnemoc>
and compare to the one of the other port
<mnemoc>
it must be there
<Maqs>
there are three ports
<mnemoc>
the other host port. not otg
<mnemoc>
unless it's a hub...
<Maqs>
usbc0 is type 2, usbc1 and usbc2 are type 1
<mnemoc>
otg, host, host
<mnemoc>
so your problem is with usbc0?
<Maqs>
linux tells me "USB hub found" for hub 1-0:1.0, hub 2-0:1.0 and hub 3-0:1.0 and again for hub 3-0:1.0
cheng has quit [Quit: Leaving]
<Maqs>
wlan is bus 3, device 2
<Maqs>
the working port is bus 2
<Maqs>
oh, even three times for 3-0:1.0
<Maqs>
and again for 4-0:1.0
hipboi has quit [Ping timeout: 252 seconds]
<rellla>
a little armhf-xbmc-bechmark: 1920*1080 bigbuckbunny (MPEG2, MPEG4, AVC, VC1, VP8) play smooth at 23-25fps, bogomips around 333,96-429,37, cpu~65%.
<rellla>
while playing the avi I have artefacts, seeking forward results mostly in a xbmc-crash. audio not tested, i have no cinch-cable here.
<Maqs>
i guess it's really a problem with one of the ports
<Maqs>
rellla: what governor/cpu speed?
<oliv3r>
well avi should only be a container, right? so something in the container usage going wrong? or am I off here :)
<rellla>
Maqs: how do i see that?
<oliv3r>
hipboi left :S i'll just mail him the conversation :)
<oliv3r>
you can change it back to ondemand afterwards of course :)
<rellla>
surely echo *gr*
<oliv3r>
performance scaling should force to always use Max freq.
<oliv3r>
can be done runtime during any test
<Maqs>
it's possible to overclock it, but i don't know how the hardware will take this, especially for mk802, as there is no cooling at all in a plastic case with only a few holes
<oliv3r>
id glue a heatsink probably if i had a a1000
<oliv3r>
i think you can safely run it around 1.1 - 1.2GHz
<Maqs>
i have those little gpu ram heatsinks here, but they are smaller than the a10 :-)
<oliv3r>
doesnt matter, any heatsink is better then no heatsink, if heat is an issue
<oliv3r>
all a heatsink does is increase the size of the thermal surface
<oliv3r>
at the cost of contact resistance between heatsink/package
<libv>
8x mali sounds like a 450
<libv>
8x is just the number of fragment shaders
<rellla>
oliv3r: cpu usage ~ 20-25% ;-) freq stable at 1008000 while only displaying video, but still stuttering when switching gui-screens and cpu usage raises up to ~95%!
<Maqs>
i just don't know whether i'll be able to remove it once it's glued :-) i might get a bigger heatsink soon
<andoma>
anyone know if there is some effort to fix the fickering when doing alpha blending with the video mixer in 1080p modes?
<andoma>
i guess not, but it's a bit hard to try to figure out what's wrong when there is virtually no documentation of the hw around
<jinzo>
Maqs, glued? why would you do that?
<jinzo>
andoma, you are aware of the manual (won't help you a bit probably but still) that was leaked?
<Maqs>
the heat sink? O.o
<jinzo>
Maqs, yes. Just use the thermal paste
<Maqs>
it will just fall off otherwise ;)
<jinzo>
and if needed attach it to something else. Gluing it on the chip is not a good idea.
<andoma>
jinzo: i've taked a look in it but afaict the video mixer register reference there does not match the actual hardware
<Maqs>
and the heatsinks i have have a glue pad..
<jinzo>
as I said, probably doesen't help you - just tought it's better to say than not.
<jinzo>
andoma, you porting showtime to A10?
<andoma>
jinzo: yes
<andoma>
it works quite nice in 720
<jinzo>
great :)
<andoma>
yup
<libv>
andoma: is your showtime going to use the disp device directly?
<andoma>
libv: it does
<libv>
git tree is where?
<andoma>
nowhere just yet
<libv>
ok
<libv>
well, be aware that i will be restructuring everything in the back of disp, only a set of the ioctls that are in active use will be regularly checked for api/abi changes
<libv>
s/a set/the set/
<ibot>
libv meant: well, be aware that i will be restructuring everything in the back of disp, only the set of the ioctls that are in active use will be regularly checked for api/abi changes
<andoma>
no worries i'll keep track of what happens :)
<libv>
andoma: once i am done with this stupid busywork i am doing (unification, beautification -- again), i will introduce a new ioctl for version control, so people in userspace can verify their compatibility with the kernel
<andoma>
nice
<libv>
but i will still do my utmost best to keep the existing api/abi, for as long as makes sense, and only for those bits which see active use
<andoma>
it would be sweet with some resourcetracking though perhaps that can be turned on via some ioctl
<andoma>
so when the program exit/crashes all layers etc that are created gets released
<libv>
heh, so you are seeing that behaviour right now?
<libv>
bloody chinese c++ kiddies :)
<andoma>
yeah the first thing showtime needs to do when starting is to delete all layers
<libv>
value &= (bit0 & bit1 & bit2);
<lundman>
heh apple puts up apology
<andoma>
in the very unlikely event that showtime should crash :)
<libv>
'nuff said :)
<andoma>
libv: yeah
<andoma>
#define BIT1 0x00000001
<oliv3r>
rella looks like the CPU is stressed out when changin the GUI menu
<andoma>
#define BIT2 0x00000002
<andoma>
i like those as well
<lundman>
as long as you dont have
<andoma>
oh .. it was BIT0 0x00000001
<andoma>
ofcourse
<lundman>
define BIT3 0x00000003
<andoma>
heh
<andoma>
another project with chinese code i once saw had: #define TEN 10
<andoma>
gotta love those things
<mnemoc>
libv: your patches that got into stage/sunxi-3.0 are also in stage/sunxi-3.4 now
<libv>
andoma: ah, i noticed "big litter" some place
<libv>
big little would be the correcter term for it :)
<mnemoc>
"big litter" is atom :p
<andoma>
:)
<libv>
mnemoc: hold off on techn_ his stuff, i am completely redoing stuff, as there was nothing else sensible
<libv>
how can you have _two_ glaring errors in such a simple line?
<jelly-home>
are those CLK_* like inverse masks? 0xffffefff?
<libv>
during my VIA days i sometimes proclaimed to be a racist
<RaYmAn>
It's like in the android world where some custom rom authors recommend wiping the cache 3 times, then wiping everything 3 times, then wiping dalvik cache (which is part of cache) 3 times =P
<libv>
that i was quite prejudiced against the taiwanese :)
<libv>
my racism is broadening :p
<jelly-home>
better confess -- you just hate people
<RaYmAn>
It's only racist if you only dislike crappy code that comes from that region :P
<RaYmAn>
I'm sure you just dislike all crappy code :P
<libv>
jelly-home: that's not true! and i now hate you for saying so :p
<libv>
the via hw did have grammar correction though, but i never bothered hooking it up to openoffice :p
<oliv3r>
what you really hate, is undeducated baboons who write shitty unreadable and thus hard to maintainable code; it just happens to be that those uneducated baboons often come from certain regions of the world, or so ti seems
<oliv3r>
trust me however, in our western world, we have the same baboons
<oliv3r>
those just dont do much OSS work and get hired to do some internal business app that hopes to net them lots of money
<libv>
true :)
slash_random has joined #arm-netbook
<oliv3r>
ill immediatly admit i often bash stereotypes like the next guy, because its human nature :)
<oliv3r>
us against them
<oliv3r>
with them being anything
<mnemoc>
just look at the baboons writting vb.net code...
<oliv3r>
mnemoc: even better point
<oliv3r>
i should have phrased it, here the uneducated baboons write vb.net code and crap like that, not real code
<oliv3r>
that said, with uneducated baboons, i absolutly mean Batchelor or even Master degreed Computer Science students
<oliv3r>
that have no idea how to write proper code :)
<mnemoc>
many "computer scientiest" don't have any idea about programming
<ZaEarl>
hey, at least you have a copy of the crappy code!
<mnemoc>
ZaEarl: indeed
<oliv3r>
crappy code may very much be better then no code and no docs :)
<mnemoc>
(information technologists are even worse)
<ZaEarl>
it's virtually impossible to get code from the various solutions companies.
<Turl>
the main issue with crappy code is that it's full of bugs, but it works
<Turl>
so you don't know if bugs make it work
<Turl>
and maybe fixing them breaks the code :P
<mnemoc>
you never know the hacks used to compensate that buggy &= (foo & bar)
<oliv3r>
so fix a bug left, introduce a bug right that was to work around crappy code left; yeah
<mnemoc>
any hint why, if kernel tells "Machine: sun4i" machine_is_sun4i() returns 0?
ZaEarl has quit [Ping timeout: 246 seconds]
<oliv3r>
mnemoc: you dont have a sun4i? hur hur
<Turl>
mnemoc: CONFIG_MACH_SUN4I set?
<mnemoc>
of course
<mnemoc>
it's a normal sun4i build
<mnemoc>
but just realized machine_is_sun4i() hates us
<Turl>
mnemoc: my defconfigs don't have CONFIG_MACH_SUN4I for example
<mnemoc>
o_O
<rellla>
i'm looking for a sata-ssd fitting in my mele. ~16GB capacity for hosting rootfs/ (rootfs++). does every 1,8" or 2,5" fit? any recommendation?
<rellla>
does 2,5"-sata-hd work without external power adapter? 3.5" does not.
<Turl>
if a specific machine is not selected, then the macro is 'hardcoded' to 0 and the compiler optimizes if blocks out
<mnemoc>
oh
<mnemoc>
missed the ARCH vs MACH diff
<Turl>
ARCH ("omap2") vs MACH (nokia thingy, omap xyz dev board)
<Turl>
they'll get more use when we get device trees :P
<lundman>
youwant MACH 3 at least, before you can scramjet
<Turl>
lundman: it's over 9000
<lundman>
MACH 881742 would be speed oflight not sure what 9000 would be
<oliv3r>
_generated_ i see, then it wont show up in my tree :)
<oliv3r>
9000 is a gaming reference
<oliv3r>
i think
<lundman>
The day I am serious is the day you read my gravestone!
<oliv3r>
i see
<lundman>
well, I can't help here, so I'll go play same games
<lundman>
oh hey, new minecraft. bbl..
<rellla>
mnemoc: hopefully at idle?
slapin has quit [Quit: Lost terminal]
<mnemoc>
rellla: no, at max
<mnemoc>
rellla: the whole device gets 5V*2A, a hungry driver will eat your mele
<mnemoc>
or not works at all
<oliv3r>
I assume you mean drive, as in harddisk drive?
<rellla>
mnemoc: ok thanks. so a "Samsung 830series 2,5" 64 GB" seems suitable...
<mnemoc>
yes, drive
<mnemoc>
my fingers tend to type whatever they like
<oliv3r>
LOL yeah
<oliv3r>
i know the feeling
bfree has joined #arm-netbook
rellla has quit [Quit: rellla]
<Turl>
hno: re " There is a hardware bug where the CPU can draw sufficient power from the UART I/O pins to start booting, but in that half-powered mode there is no power supplied to the SD port."
<Turl>
hno: I so hate that on my mele :p
<Maqs>
is it possible to toggle usb0 between host and gadget mode?
slapin has joined #arm-netbook
<oliv3r>
that cpu bug on a specific hardware revision only? or all of them have that bug?
<Marex>
Maqs: EHCI does support that ;-)
<mnemoc>
isn't that an electrical thing?
<mnemoc>
5th pin shorted vs nc
slapin has quit [Quit: Lost terminal]
Almamuetya7 has joined #arm-netbook
empat0 has joined #arm-netbook
drachensun has joined #arm-netbook
hipboi has left #arm-netbook ["Leaving"]
<oliv3r>
oh didnt notice he rejoined
<empat0>
turl: im playing with cm9. i followed the sunxi wiki, everything comes up, but i do not have a mouse pointer (also mouse events are coming). any hints for me?
mSquare has quit [Quit: Leaving.]
<Turl>
you should have one
<Turl>
empat0: can you run getevent on adb shell and see what name your mouse got?
<Turl>
mnemoc: might be, I have one of the uart cables not connected
<oliv3r>
empat0: why not try CM10?
<empat0>
turl: could not get driver version for /dev/input/mice, Not a typewriter
<empat0>
add device 2: /dev/input/event2
<empat0>
could not get driver version for /dev/input/mouse0, Not a typewriter
<empat0>
name: "RCMCU iPazzPort"
<empat0>
add device 3: /dev/input/event1
<empat0>
name: "RCMCU iPazzPort"
<mnemoc>
Turl: eh? lost the context, what do are you refering too? uart should always be tx,rx,gnd only, no vcc.
<empat0>
oliv3r: just getting started with cm. there is a sunxi wiki page on this, refering to cm9.
debianin has quit [Ping timeout: 246 seconds]
<drachensun>
whenever I try to modprobe mali I'm getting segmentation fault
<drachensun>
I just built the latest sunxi-3.0 kernel
<drachensun>
is this because of the u-boot and the memory settings?
<mnemoc>
might be
<mnemoc>
mali needs a very special memory block cutted off
<drachensun>
empat0 when you run xbmc with the latest kernel do you have a recent u-boot or and older one before the ram changes?
<mnemoc>
drachensun: check begining of dmesg
<drachensun>
wow, I'm so used to working over the serial terminal I forgot now I wont see kernel errors
<drachensun>
ok, getting now
<empat0>
drachensun: i was lazy, i'm still using a very old bootloader
<drachensun>
empat0: ok, because I'm getting seg faults in xbmc but I think is probably because I don't have the mali driver loaded
<Turl>
mnemoc: ctx = booting from nand if uart connected before powering device
<Turl>
empat0: I assume that's your mouse then
<Turl>
if you move it do you see the events?
<empat0>
drachensun: without mali xbmc will not start
mysteryname has quit [Read error: Connection reset by peer]
mysteryname has joined #arm-netbook
<empat0>
turl: yes. i can even select items.
<mnemoc>
BROM chip-id: aw1623C (A10) or BROM chip-id: aw1623-revC (A10) or ..... aw1623 (A10 revC) or .... something else?
<mnemoc>
brom only tells 1623, rev is taken from a different part
<mnemoc>
----^ RFC
<Turl>
empat0: maybe the kernel driver for that RCMCU iPazzPort thingy has incorrect flags then, you should be able to override it with an idc file
<Turl>
mnemoc: chip-id: AW1623 (revision C)
<mnemoc>
Turl: I don't boot from nand... ever :|
<mnemoc>
Turl: no "A10" friendly name?
<drachensun>
pc is at mali_kernel_memory_mmu_interrupt_handler_upper_half when it crashes
<Turl>
mnemoc: my mele likes to boot from nand after a kpanic or power unplug/plug if I leave uart connected
<Turl>
mnemoc: AW1623 (A10 revision C)
<mnemoc>
i never unplug the uart...
<Turl>
or even better, chip id: A10 (AW1623 revision C)
<Turl>
as A10 is marketing, and the revision isn't
<empat0>
turl: any pointer for idc files?
<mnemoc>
Turl: ok
<drachensun>
I guess I will break out the old u-boot later and try it that way
<Turl>
empat0: you'll need a text file named RCMCU_iPazzPort.idc
<oliv3r>
there, actually put some effort into today. pll4 is now up :p
<Turl>
empat0: put this line inside
<Turl>
touch.deviceType = pointer
<Turl>
and drop the file in /system/usr/idc/
debianin has joined #arm-netbook
<oliv3r>
if anybody sees hipboi, tell him ive mailed him e-mail details; hell know what to do with them
<oliv3r>
hometime for me :)
<oliv3r>
PLL5 i leave for monday, and probably try to also do PLL6
<empat0>
turl: still no pointer
<oliv3r>
have a good weekend all!
<Turl>
empat0: check logcat when you plug your device, what do you see?
<Turl>
empat0: settings-dev options also has a setting you can enable to show touches/pointer, you can try that
<techn_>
3rd day and 20k raised :p
<empat0>
turl: i can not find this option. where is it in detail?
<techn_>
Tom will propably need to make more boards than 1k :)
<mnemoc>
techn_: that was the point in making the campaign flexible, fixed stops at the goal
<Turl>
empat0: I'm running jelly bean but it should be somewhere in settings-developer options
<empat0>
turl: i think i will try jelly bean, same instructions as on wiki, just use -b jellybean?
<techn_>
so you need to pay for free delivery? or what you are talking about?
<mnemoc>
techn_: Turl's post? it's about the fees charged by DHL to deliver your package
<Turl>
mnemoc: 1GB ram'd mele :O
<Turl>
empat0: you need a new local_manifest
<empat0>
turl: where can i get this?
<mnemoc>
vi local_manifest.xml
<Turl>
it doesn't exist yet :P let me make one
<bfree>
techn_: I think you are in the EU so you should be ok ... it was less then €22 so it is exempt from duty or VAT (if VAT, or duty, is due the courier likes to add their own "admin fee" on top)
<drachensun>
empat0: Is there an output log or debug mode or anything to see error messages from xbmc?
<drachensun>
empat0: I've got the mali driver now but I'm still getting "ERROR: unable to create GUI. Exiting" seg fault
<Turl>
consider it an invitation to make a device tree for it :p
<empat0>
turl: will do my best
<drachensun>
empat0: I was following relllas build instructions which said to remove libegl1-mesa libegl1-mesa-dev libegl1-mesa-drivers libgles2-mesa libgles2-mesa-dev
<empat0>
drachensun: are you building armel or armhf?
<drachensun>
empat0: armhf
<empat0>
drachensun: you need the stage/sunxi-3.0 kernel for this, have you installed it?
<empat0>
drachensun: (or wip/*/next_mali)
<Maqs>
sunxi-3.0 won't work for hardware decoding?
<mnemoc>
stage is preferred
<mnemoc>
Maqs: yes. diff is in mali
<Maqs>
btw, is there a way to make sure it's using hardware decoding?
<mnemoc>
and some other disp fixes
<Maqs>
and how do i get rid of the blinking cursor in xbmc? :)
<empat0>
Maqs: you can check the xbmc logfile. (hopefully) all A10 messages are prefixed with A10.
<Maqs>
FB_SUNXI/FB_SUNXI_HDMI? something else?
<drachensun>
oh ok, I missed th stage/ part
<mnemoc>
empat0: btw, this also affects A13 and A10s. not only A10 ;-)
<Maqs>
16:47:22 T:1335301184 NOTICE: A10: using software rendering.
<Maqs>
ok, that does not look like hardware rendering :-)
<hno>
Maqs, you need to tell fbcon to hide when starting xmbc.
<drachensun>
I will try it now
<Maqs>
i guess there was some solution for the screen going black for a few seconds sometimes, too.. but i don't remember where i found it
<empat0>
Maqs: be carefull, it gets initialized for software rendering also when building the tumbnails.
<Maqs>
DEBUG: A10: cedar open.
<Maqs>
looks better.. i guess .-)
<empat0>
Maqs: now you are using the vpu for decoding. for hardware rendering you should see A10: using hardware rendering.
<Maqs>
oh, okay
<Maqs>
empat0: could you maybe put your kernel .config online somewhere?
<drachensun>
empat0: ok, I build the stage/sunxi-3.0 kernel and I'm getting the same error, is there a list of the drivers I need loaded?
<drachensun>
wait you said something about permission to the devices, seems like I remember that being something that had to be done manually even if running as root
<jelly-home>
RaYmAn: at least kickstarter doesn't take your money when the goal is not reached
<RaYmAn>
yeah
<techn_>
mnemoc: for me it wont pass Q= variable to linux makefiles
<mnemoc>
techn_: any variable set in the parent Makefile will win against those defined in the child makefile
<mnemoc>
not as argument
<mnemoc>
as part of the env
<mnemoc>
as evil magic
<mnemoc>
make linux Q= 2>&1 | less and look the output
<mnemoc>
techn_: but, anything against the diff pasted above?
<techn_>
mnemoc: I tried to move include chosen_board.mk to beginning.. it helped :)
Almamuetya8 has quit [Ping timeout: 276 seconds]
<mnemoc>
techn_: let me commit my change first
<techn_>
ok.. If you'll pass argument as command line it will be passed to child makefile.. but inside makefile it should pass :/
<techn_>
*shouldn't
gimli has joined #arm-netbook
<techn_>
that's why I dont have that problem since I change that in top makefile
revident has joined #arm-netbook
<techn_>
.. I dont pass it via command line
<mnemoc>
adjusting Makefiles manually on the working tree doesn't play nice with git. and i would even claim it a bad practice, but I won't argue. np :)
<mnemoc>
what really disturbs me is your `cp foo/* bar -rf` .... options at the end are AWFUL
<techn_>
yep.. I was thinking to have better solution for that
<mnemoc>
cp -a is a common way to do recursive cp
<mnemoc>
but anyway, I need to return to $work$
<mnemoc>
having the u-boot and linux targets fixed I'm happy
<techn_>
I was thinking to make archivemount to get rid of extra file handling
<techn_>
or do we need those dirs elsewhere than hwpack creation?
<mnemoc>
what dirs?
<techn_>
output/$(BOARD)
arnd has joined #arm-netbook
<mnemoc>
lets assume we don't until we do
<mnemoc>
for other final targets, and a livesuit image or an sd card
<mnemoc>
a/and/like/
<techn_>
plus.. after this is done we need to figure out different rootfs thingies
<mnemoc>
those are better done is scripts imo
<mnemoc>
s/is/with/
<ibot>
mnemoc meant: those are better done with scripts imo
<techn_>
what are supported rootfs's.. linaro's, android,.. ?
<techn_>
handbuild debian..
<mnemoc>
can't really think about it right now. my brain is crippled and need to return to work
cat_x301 has quit [Ping timeout: 265 seconds]
<techn_>
:)
<mnemoc>
for those where we can download a rootfs tarball I wouldn't care to build
<mnemoc>
just make an script with the url/path to the tarball
cat_x301 has joined #arm-netbook
<mnemoc>
and let me make the card, or the livesuit image, or whatever
<mnemoc>
doing all those things in Makefiles is awful
<mnemoc>
but i might be wrong.... can't really think straight atm
<techn_>
damn.. this starts to look great :)
<mnemoc>
my migrane?
ZaEarl has joined #arm-netbook
<mnemoc>
yes, kind of
<mnemoc>
try delegating long shell outs to scripts/*.sh helpers
<mnemoc>
scripts/mk_hwpack.sh for example
<mnemoc>
you can . chosen_board.mk, it's shell friendly
revident has quit [Quit: Combustible lemons? Bah, I bring you weaponized asparagus!]
Undertasker has joined #arm-netbook
MMlosh has quit [Quit: Bye...]
merbzt has joined #arm-netbook
andrew__ has joined #arm-netbook
gsilvis has quit [Read error: Connection reset by peer]
gsilvis has joined #arm-netbook
popolon has joined #arm-netbook
Almamuetya8 has joined #arm-netbook
arnd has quit [Ping timeout: 244 seconds]
arnd has joined #arm-netbook
<libv>
i cannot help but feel satisfied: 129 files changed, 21000 insertions(+), 42000 deletions(-)
<libv>
flatting sunxi/disp after unification :)
<libv>
flattening even
<libv>
but i hope i never have to type CONFIG_ARCH_SUN[45]I again in my life.
* libv
now heads off to hollow out OSAL again
<mnemoc>
libv: stage/sunxi-3.0 has machine_is_sun[45]i() support now :p
<libv>
mnemoc: in some promising future, much of those preprocessorred defined bits would just fold together
<libv>
at least a large portion of it
<mnemoc>
:)
<libv>
but for this step, preserving code is important
<mnemoc>
yes
<mnemoc>
it's the "documentation" :|
<libv>
yup
eebrah has joined #arm-netbook
<steev>
good news everybody! my 10" MID970 appears to have broken wifi
<specing>
My 7" one too.
<specing>
It worked before
<steev>
mnemoc: fyi, mainline wants to get rid of the machine_is_ macros
<techn_>
mnemoc: take a look
<steev>
specing: ditto, but this means i can take it apart
<specing>
Heh
<specing>
Here it stopped working after I took it apart
<steev>
so i can finally get back to a10 hacking
<steev>
ah, no, i haven't taken it apart yet, in fact, i dunno why wifis stopped working
<steev>
it was starting to have issues with connecting/downloading updates, and then it just went kaput
<rz2k>
welcome to realtek
<specing>
steev is becoming german ;)
<specing>
I like realtek's crabs
<mnemoc>
steev: we are way way away of what mainline wants. we first need unification, and that fills the code with #ifdefs
gimli has quit [Quit: Verlassend]
<mnemoc>
steev: machine_is_ turns many of the #ifdefs within functions into something readable :<
<steev>
i know :(
<mnemoc>
eventually we might reach the point where DT rules our world to... but that won't happen any time soon :|
<steev>
i'm still on the fence about DT
<steev>
i mean i did the work for the efika, and it just... really so not worth it
<steev>
went from decently working to, well, at least we have serial
<mnemoc>
:(
MMlosh has joined #arm-netbook
Undertasker has quit [Read error: Connection reset by peer]
<steev>
at least the imx-drm stuff made it into staging
Almamuetya8 has quit [Read error: Connection reset by peer]
arnd has quit [Ping timeout: 252 seconds]
freakazoid0223 has quit [Remote host closed the connection]