<oliv3r>
ohh the hangout! yeah i have that youtube page open somewhere as a 'todo'
focus has quit [Ping timeout: 240 seconds]
diego_r has quit [Ping timeout: 240 seconds]
Gerwin_J has joined #linux-sunxi
<oliv3r>
anybody recalls how Hansg's temperature driver reports the temperature to the kernel/user? It looks like it's a 2nd input or something, but my kernel foo is a little weak
<CaptHindsight>
is the .brd and old Orcad layout file?
avsm has quit [Quit: Leaving.]
<Gerwin_J>
i don't know
<Gerwin_J>
try it
alexst has quit [Ping timeout: 272 seconds]
<oliv3r>
wens: is there a kernel interface to get the cpu temperature? (from within the kernel), I don't see functions or calls in that regard in hwmon
<CaptHindsight>
Type: Allegro Board
<mripard>
oliv3r: there's a thermal framework iirc
<mripard>
but I never used it
<CaptHindsight>
Layers: 8
<wens>
oliv3r: i haven't looked through kernel apis
<ssvb_>
oliv3r: also first we need cpufreq support, to be able to change the clock frequency
<oliv3r>
ssvb_: yeah i was just being curious and browsing stuff :p
<oliv3r>
i'm starting with 3.4 support first anyway
<oliv3r>
i take it there's not many unknowns
<ssvb_>
oliv3r: nand is the only thing that is missing in the mainline kernel for you, right?
<ssvb_>
oliv3r: instead of the nand, which adds 10 euro to the price tag, they Olimix could probably add a second internal micro sd slot
<oliv3r>
because then you put the nand firmware into the emmc firmware :)
<oliv3r>
i'm not sure if the nand is pin compatible
markusr has joined #linux-sunxi
<oliv3r>
but an emmc chip would fit where the nand is located
alexst has joined #linux-sunxi
<wens>
depends on whether they left mmc traces or not?
focus has quit [Remote host closed the connection]
<ssvb_>
oliv3r: we just need some open source friendly sd card hardware, the firmware of many of them seems to be upgradable - http://www.bunniestudios.com/blog/?p=3554
alexst has quit [Ping timeout: 240 seconds]
<ssvb_>
but I guess everyone had seen the old news about the sd cards hacking already
<CaptHindsight>
that board file is using Allegro V16
<markusr>
I want to test mainline kernel 3.16 with ubuntu linaro distribution on cubietruck. But I struggle to find a kernel config. Where can I find one?
focus has joined #linux-sunxi
<markusr>
or do I need to make my own?
<mripard>
sunxi_defconfig
<markusr>
don't I need anything more?
<mripard>
I don't know, what do you need?
<CaptHindsight>
is Olimex planning a COM for the A20?
<wens>
sunxi_defconfig should cover all the hardware we have drivers for
<oliv3r>
ssvb_: the opensource friendly bit is going to be an issue :)
<wens>
you might want to add filesystems, debug options and other stuff
<Wizzup>
ssvb_: I pulled the sunxi-boards repo, and I used the new cubieboard2.fex to generate a new script.bin ; put it on my cubieboard2, but meminfo still says 480 Mhz
<Wizzup>
That seems off, but I am quite sure that putting on the new script.bin did seem to solve my stability problems before
<Wizzup>
Do I also need a (more) recent u-boot and/or kernel
alexst has joined #linux-sunxi
<ssvb_>
Wizzup: the dram clock speed is set in u-boot
<ssvb_>
but the dcdc3 voltage (critical to dram reliability at high clock speeds) is configured in fex
avsm has quit [Quit: Leaving.]
<ssvb_>
the newer cubieboard2.fex fixed the voltage and should resolve the reliability problems, so it looks good
alexst has quit [Ping timeout: 264 seconds]
<Wizzup>
ssvb_: right!
<Wizzup>
so 480 is OK
<nove>
no clue of h265/vp9 in the kernel driver, but the blobs headers will give the final response, but until now appears to be by gpu compute
kuldeepdhaka has quit [Quit: Leaving]
froese has joined #linux-sunxi
<Wizzup>
ssvb_: However, in the .fex I think it does state 432Mhz as freq, which is what got me confused
<ssvb_>
Wizzup: my understanding is that some Allwinner tool is used (or was used) to generate the boot0 binary for android, based on this information from fex
<ssvb_>
Wizzup: but if you are using u-boot with linux, then the dram parameters and clock speed information from fex is not used
<Wizzup>
right. so if I'd somehow really want to downclock to 432 (doubt it) then I would have to edit the u-boot config
bgal has joined #linux-sunxi
<nove>
libv: don't forget the part of "some arm customers asked to not open source"
bonbons has quit [Quit: Leaving]
tomcheng86 has joined #linux-sunxi
bonbons has joined #linux-sunxi
tomcheng76 has quit [Ping timeout: 240 seconds]
<mmarker>
Oh. What's a good way to debug uboot issues. Recent pull of uboot on my pcduino3 wasn't reading large files off mmc. Everything works via a netboot. Any way to try to debug this for a proper bug report?
<ssvb_>
BorgCuba: try to type the 'fatload' command from the console to load the kernel and check if the problem persists
<ssvb_>
BorgCuba: it might be that the recent mmc changes have introduced a regression
<ssvb_>
BorgCuba: and it might be a good idea to check if you have no filesystem corruption on the sd card and the uImage file is in fact readable (on your PC)
<ssvb_>
BorgCuba: and maybe try a different sd card, your current one could be defective ("Error reading cluster" does not look very encouraging unless it is a software bug in u-boot)
<BorgCuba>
well, mine was a little cheaper (like 25 euros)
<BorgCuba>
anyway, I just bought it for no particular reason
<BorgCuba>
btw, I cannot see cdc_ether gadget driver?
<BorgCuba>
will "ACM and mass storage" give me a console and usb drive?
alexst has joined #linux-sunxi
<BorgCuba>
the mali driver doesnt seem to compile: drivers/gpu/ion/ion.c:866:4: error: implicit declaration of function ‘__dma_page_cpu_to_dev’ [-Werror=implicit-function-declaration]
alexst has quit [Ping timeout: 272 seconds]
<ssvb_>
BorgCuba: which kernel are you trying to compile?
<BorgCuba>
Linux debian 3.4.90+ #3 PREEMPT Mon Jul 14 20:40:54 CEST 2014 armv7l GNU/Linux
<ssvb_>
the mali kernel module should be enabled by default and compile without any problems
<BorgCuba>
yes, it was the ion part
<ssvb_>
I mean the git repository
<BorgCuba>
I think I cloned it last week
<ssvb_>
how did you get ion enabled?
<BorgCuba>
I enabled it but the build failed
<ssvb_>
maybe you should not do this
<BorgCuba>
yes :-)
<BorgCuba>
i disabled it and it build okay
<ssvb_>
btw, mali is clocked at a ridiculously low clock speed on a13
<BorgCuba>
and its MP1 iirc
<BorgCuba>
I have another rk3066 hdmi stick here which is more capable but I wanted to try the a13 board because it was so cheap
<BorgCuba>
and I have a sunii device :-)
<ssvb_>
well, MP1 is a hardware limitation, but the low clock speed is a purely software thing
<ssvb_>
and I guess you need the "pmu_used = 0" to get rid of that problem
<ssvb_>
BorgCuba: you can convert it back to fex and check the settings
notmart has joined #linux-sunxi
<BorgCuba>
using fexc?
<ssvb_>
yes
astr has quit [Ping timeout: 264 seconds]
<ssvb_>
BorgCuba: and I believe that everything would have been a bit easier if you used the http://linux-sunxi.org/Manual_build_howto instead of a random internet blog ;)
<BorgCuba>
for sure
<BorgCuba>
i apologize
ricardocrudo has joined #linux-sunxi
<ssvb_>
np
<BorgCuba>
ssvb_, have you ever used the usb gadget driver for console access?
<BorgCuba>
and is the micro board 2-layers only?
<fiola>
ssvb_: Indeed. Sadly it's a common thing to hear people recommend "Google it" which ends up with people following random Internet blogs, instead of going direct to known-good technical sources. The "just google it" Internet meme is appropriate for cat pictures, less good for technical pursuits. Engineers really need to push back against it.
tm512 has quit [Ping timeout: 240 seconds]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<fiola>
BorgCuba: I'd like to know the answer to that as well. I had to disable the OTG port entirely on A10 LIME to keep load average down.
<froese>
fiola: but often there are no known-good technical sources ...
<ssvb_>
BorgCuba: and a USB ethernet dongle for network connectivity
<BorgCuba>
usb gadget would save some cables
<ssvb_>
yeah, sure
<mmarker>
Hmm. Seems I missed the early morning. So mainline doesn't adjust CPU clock speeds. Does 3.4? I may be enticed to do some forward porting. Should it be against sunxi-next?
astr has joined #linux-sunxi
<fiola>
froese: That's fair enough in the absence of anything better, as long as random Internet blogs are used with caution. It's just not the first port of call, when one knows of competent and directly relevant technical sites.
<froese>
fiola: right you are. but for the allwinner stuff there is hardly a "copmetent and directly relevant technical site". even the sunxi-wiki is more a collection of (often outdated) random snippets.
<fiola>
One benefit of using the OTG port instead of the UART is that it would let you obtain late-boot console access to a closed box. Short of making a hole for serial wires in a LIME case or installing some kind of miniature socket, one needs to open a LIME case for console access.
<Black_Horseman>
hola
<Wizzup>
fiola: protip for the LIME case: leave one screw out
<Wizzup>
use the hole to tunnel some cables in
<fiola>
froese: Yeah, but the wiki's getting better all the time. I'd certainly call it more authoritative and likely to be accurate than a random Internet blog. And the combination of linux-sunxi, eLinux and eewiki beats most other sources on the net.
<Wizzup>
I did it with the LIME+case that controls the leds in my parents house
<ssvb_>
froese: feel free to improve "a collection of (often outdated) random snippets", it's definitely up to us
<fiola>
Wizzup: good idea :-)
<Wizzup>
I managed to fit in three 'jumper cables' for serial and an audio cable (abused left and right for spi data and clock)
<froese>
ssvb_: well, when i'm looking for infos i'm most of the time not in the position to add the missing data ;-)
<BorgCuba>
ssvb_, indeed, setting pmu_used = 0 resolved that issue!
<BorgCuba>
thanks again!
<ssvb_>
froese: but after you find the solution for your problem, you are in a good position to add the missing info :)
<ssvb_>
froese: this is going to make life easier for the next guy
<fiola>
Actually, even makes it easier for ourselves on the next board. Memory isn't perfect. :P
ZetaNeta has joined #linux-sunxi
<froese>
ssvb_: if I would find info where I'm 100% sure it's corrected i would even do that, but i never got to that point - there are no definit sources in the AW-world (or i haven't found them yet).
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
ganbold_ has quit [Ping timeout: 240 seconds]
<froese>
.oO(hmm... i see no "edit" button on the wiki pages)
<fiola>
Come to think of it, a 2.5mm stereo audio socket might just about fit on the end bracket of the LIME on the HDMI side. And it would have the benefit of giving anyone who plugs in their headphones a surprise. :P
TheSeven has quit [Ping timeout: 260 seconds]
<ssvb_>
froese: you need to create an account first, anonymous edits are disabled for a good reason
<fiola>
Needs a marketing bullet point: "Compatible with RS232 headphones." :P
Andy-D has joined #linux-sunxi
formruga has joined #linux-sunxi
netlynx has quit [Remote host closed the connection]
TheSeven has joined #linux-sunxi
<fiola>
ssvb_: spam troubles in the past?
<froese>
btw, has anyone seen the a20-lime2 yet?
<ssvb_>
fiola: yes, spammers are everywhere and happily shit any place they can easily reach
<fiola>
:-(
<fiola>
froese: I think all there's been so far is Tsvetan's one-liner tease about LIME2 taking another 2 weeks or so. But it'll be ready when it's ready. :P
<fiola>
I'm looking forward to the gigabit.
<froese>
yes, that and hopefully a 32bit memory interface.
<fiola>
I would hope that 32-bit memory can be assumed, even on a LIME. Adding performance features like gigabit but retaining the main bottleneck would be questionable.
<froese>
well, the regular lime has only 16bit
<fiola>
I'd guess that the memory's being doubled for LIME2 by using two chips to give two banks for 32-bit, while allowing them to keep the same BOM stock as for the other LIMEs.
<froese>
i _guess_ the original decision was made to get away with a single ram chip. i don't know, if there are 32bit wide chips.
<froese>
what would be nice were a 2nd sdcard instead of the nand...
<fiola>
LIMEs have SATA for storage expansion.
<froese>
fiola: sure, but the boot loader has to be on sdcard or nand.
<froese>
nand is simply too much trouble, more expensive (than an sdcard-socket), and slower.
<ssvb_>
froese: a single DDR3 chip can only have up to 16-bit I/O bits according to the JEDEC spec
alexst has joined #linux-sunxi
<froese>
ssvb_: ah, thanks. good to know.
<fiola>
NAND shouldn't be slower for the same flash technology, it as an 8-bit bus vs the SD card's 4-bit.
<fiola>
s/as/has/
<fiola>
Is the NAND on LIME's very slow?
<froese>
fiola: right, "should" ;-) but has anyone got a fast nand-driver for linux?
<BorgCuba>
I am running into trouble building the otg driver
<fiola>
That I don't know. But I like numbers, so I guess it's time to measure the storage throughput on the two media.
<froese>
fiola: i don't know the lime's speed, on cubietruck it's around 8MB/s.
<fiola>
froese: Thanks, good to know.
<froese>
BorgCuba: i didn't dare to touch otg yet, but i'm interested in your results ;-)
<BorgCuba>
well, i get undefined references when I build for "otg" or "device only"