<cnxsoft>
tkaiser: So different models of regulators are used during manufacturing of Orange Pi One boards? Wow how is that supposed to work? Unless regulators have the same specs...
paulk-collins has joined #linux-sunxi
<tkaiser>
No idea, unfortunately I can't read what's written on U53 and the cameras I've here are too crappy either
<KotCzarny>
not even with photo enhancing?
<tkaiser>
At least the other Armbian user was able to confirm that's not SY8113B on his board
<tkaiser>
KotCzarny: No, just a smartphone with wrong autofocus with low distances. I've large DSLR but battery is empty and charger is borrowed to a friend
<KotCzarny>
hmm
<KotCzarny>
take the photo from 20cm then or more at highest possible res
<KotCzarny>
maybe it will be enough
<cnxsoft>
Was there anything else on Baidu? Somebody just asked me about the CAD files.
<KotCzarny>
nitehawk, problem is that i cant get lircd to work
<KotCzarny>
set protocol to nec, evtest works, but cant teach lircd the keys
<hp197>
plaes: Yes, but on mux6
<GeneralStupid>
did anyone already tried an 4.x kernel on orangepi PC?
<tkaiser>
GeneralStupid: sure
<hp197>
generalstupid: yes
<jelle>
GeneralStupid: yup works fine here
<NiteHawk>
KotCzarny: i'm not too familiar with that. what i know is that handling of input between kernel and LIRC changed some time ago, the kernel meanwhile is able to process events and manage keytables on its own. see the diagram at http://www.lirc.org/html/configuration-guide.html
<NiteHawk>
evtest targets the 'linux input layer'. so maybe this applies here: "Last option is to connect the LIRC driver to the linux input layer using LIRC's --uinput option. This means that application sees the input as coming from the kernel, and LIRC's other capabilities are not available. This is not described here."
<KotCzarny>
nitehawk: i need lircd because i want audio app in background to receive the events
<NiteHawk>
ah, sorry - forget about that. seems lircd --uinput works "the other way around" and is intended to inject events into the input layer
<GeneralStupid>
jelle, tkaiser, hp197 what combination of uboot and kernel are you running?
<jelle>
GeneralStupid: using hans his branch and mainline u-boot
<GeneralStupid>
i loaded hans kernel too, but didnt gave it a try until now
<GeneralStupid>
uboot is missing for
<GeneralStupid>
me
<NiteHawk>
KotCzarny: i suggest going through http://linux-sunxi.org/LIRC for useful tips. REMOTE_DRIVER="devinput" and REMOTE_DEVICE="/dev/input/eventX" seem pretty straightforward and 'generic' enough to get lircd working
<KotCzarny>
nitehawk, been there, done that
<KotCzarny>
worked for me on legacy kernel
<KotCzarny>
but with mainline there is something missing
<KotCzarny>
part of the problem is that with kernel event interface there are no repeat events coming
<KotCzarny>
which confuses irrecord
<KotCzarny>
i can see things repeating when catting /dev/lirc0, but not using /dev/input/eventX
<KotCzarny>
when i do evtest, keys are working ok (-repeat part), but i cant get lircd to work with /dev/input/eventX
<KotCzarny>
(only ev_msd and ev_syn events)
<NiteHawk>
okay, that seems weird. but i don't have an idea why...
<jelle>
I'm trying to test the hdmi driver at the moment
<NiteHawk>
KotCzarny: one thing that puzzles me is the presence of "lirc" in /sys/class/rc/rc0/protocols - maybe lirc[d] requires its own specialized IR protocol handling?
<GeneralStupid>
NiteHawk: i would search for alternatives... lirc is very heavy for just doing input stuff
IgorPec has joined #linux-sunxi
<GeneralStupid>
NiteHawk: do you need lircs network stuff?
<NiteHawk>
GeneralStupid: i'm not really using IR on my bpi, just for testing purposes. KotCzarny is the one experiencing trouble trying to get lircd to work with recent mainline kernel
pstef has joined #linux-sunxi
<GeneralStupid>
jelle: ls /sys/class/gpio_sw/ and for example: echo 1 > /sys/class/gpio_sw/standby_led/data
<jelle>
GeneralStupid: I don't have a gpio_sw
<jelle>
might be because I am missing modules
<GeneralStupid>
NiteHawk: yes, and lirc is very special .... it is really bloated, i use my remote control as an hid and just remap the commands
<GeneralStupid>
jelle: too bad :(
<NiteHawk>
i am aware that lirc isn't really required any more if you're purely interested in the input events - however a number of applications still relies on doing things "the lirc way"
<KotCzarny>
GeneralStupid: how do you do things to apps that run in background?
<GeneralStupid>
NiteHawk: but most applications should support keys too
<GeneralStupid>
KotCzarny: i use shell scripts for that
<KotCzarny>
GeneralStupid: now THATS heavy
<GeneralStupid>
i like shell :)
fredy has quit [Excess Flood]
<GeneralStupid>
doing those things half of my life - but i dont know lirc
<KotCzarny>
i like shell either, but it's a bit heavy for the things i want to do with remote
<GeneralStupid>
i dont know what you want to do. i use it to control mpv, start it and get a new zattoo (stream TV) Url
<KotCzarny>
GeneralStupid: i have app that plays music, it runs in a background (detached from terminal, no gui), now when i press button on remote i want my app to receive that event
<GeneralStupid>
KotCzarny: i run X, thats pretty easy then :)
fredy has joined #linux-sunxi
<KotCzarny>
well, maybe i should show you my bpi audio box
<GeneralStupid>
i dont say there is no usecase for lirc anymore
<GeneralStupid>
but you should be able to get those key types on the console, too
<GeneralStupid>
and you could use special keys (VOL_UP or smthg)
<KotCzarny>
see 'detached' part
<KotCzarny>
i have networked gui for it too, but i also like to control it via remote
cnxsoft has quit [Remote host closed the connection]
<KotCzarny>
it has integrated audio amplifier, so no external one needed. cables are connecting directly to speakers. i plan to add li-ion one day for some mobile action
<GeneralStupid>
ok , i see why you need remote control but not why it has to be done with lirc
lamer14560638874 has joined #linux-sunxi
tkaiser has quit [Ping timeout: 248 seconds]
reinforce has joined #linux-sunxi
<KotCzarny>
GeneralStupid: because lirc was invented as linux rc interface for apps?
<KotCzarny>
and because it works nicely for this purpose
<KotCzarny>
also because i've already working configuration for my audio app that works on pc too
<KotCzarny>
(and everywhere with lirc installed)
<GeneralStupid>
i coudlnt get it to work in appropriate time
<KotCzarny>
ok, found a piece, one has to write proper key table with ir-keytable
<longsleep>
apritzel: Hey, i wanted to have a look at your ATF for A64/Pine64 - did you push that somewhere?
<KotCzarny>
yup. works now
* KotCzarny
happy panda now
apritzel has joined #linux-sunxi
bgardner has joined #linux-sunxi
paulk-aldrin has joined #linux-sunxi
<KotCzarny>
yup. woooorks! time to update wiki
hipboi has joined #linux-sunxi
vagrantc has joined #linux-sunxi
apritzel has quit [Ping timeout: 248 seconds]
ganbold has quit [Remote host closed the connection]
<tkaiser>
And switching to Arch should be easy with a running Armbian image as long as you keep the contents of the /boot folder
<keh>
newest armbian is already installed on sata and it is working well as the time before upgrading yesterday
<KotCzarny>
tkaiser: i prefer to have separate /boot partition
<tkaiser>
keh: Ok, if you put the rootfs on SATA then there might be some quirks when upgrading. Don't remember exactly but it should also be in the thread.
<keh>
alright. So I just need to adjust the boot folder files / configs as already necessary in armbian. Boot partition is still an own partition (ro mostly) on the sdcard and sda1 is root.
<tkaiser>
And maybe grab the latest u-boot deb, extract the contents and overwrite the first sectors on the SD card. As already said I don't know exactly since I don't care that much (leaving the rootfs always on SD-card to be able to spin a connected HDD down when unused)
<topi`>
tkaiser: loboris' MATE image boots fine, but I could not enable eth0, dmesg is full of some gmac errors
<topi`>
is this a known bug?
<tkaiser>
topi`: Get the update_kernel.sh form zoobab's mirror and execute it.
<keh>
only mistake I made was to have the bootfiles in the basement of the boot-partition and not in an boot folder on the boot-partition. Thought it is not that important.
<topi`>
nor was I able to get a Realtek usb wifi dongle working, it complained about not being able to load the firmware although I installed firmware-realtek
<KotCzarny>
just anything related to normal OS use
<tkaiser>
Ok, different use case. I use cross-compiling most of the times and simply try to avoid writes.
<keh>
Well, better to use a sata drive for the root. Because of the many ebooks I have it is better to store the library on the sdcard. And with ro boot-partition there should be no writes at all
<GeneralStupid>
keh: iam using some old 4gb card... but with that line it performs ok
<KotCzarny>
even stupid mc is writing in ~
<tkaiser>
KotCzarny: Do you really use an SBC as a desktop replacement?
<KotCzarny>
tkaiser: yes
<GeneralStupid>
sbc?
<KotCzarny>
single board computer
<keh>
in general it works, but even if 64gb cards are cheap, I'll use the sata and maybe with a connected and chargeable battery I can have a nice portable server with wlan router functionality
<KotCzarny>
just any sunxi thingie
<tkaiser>
keh: Nope, you chose the wrong board ;)
<GeneralStupid>
i like the idea but i wouldnt use any allwinner crap for that :)
<keh>
what? why tkaiser ?
<tkaiser>
Olimex or Lamobo R1 can power SATA from battery, the other A20 boards not
<tkaiser>
With Banana Pro you would have to try to use 5V from the USB ports and power a disk with that
<keh>
Oh, okay. Then I'll buy some of these boards (I think lamobo is quite new?) and find another task for the bpro
<keh>
thanks for the info
<GeneralStupid>
but afaik the SATA on that allwinner SoCs are just usb-to-sata
<tkaiser>
keh: Please have a look into the Wiki before you consider the R1 crapboard
<KotCzarny>
GeneralStupid: on a10/a20 its native sata
<tkaiser>
BTW: We have a wiki, most of this stuff is outlined there ;)
<KotCzarny>
yes, bpi-r1 is crappy
<tkaiser>
Including performance numbers, what to consider and so on
<keh>
tkaiser: I did, but so many boards... It needs a nice overview to compare. what was the other one... olimex. Let me see some info about
<KotCzarny>
but if one just wants simple board with most things included it can do the work
<KotCzarny>
keh: if you want nice board consider clearfog
<GeneralStupid>
why does cpu clock speed always directly influence IO ?
<KotCzarny>
because allwinner
<apritzel>
longsleep: yeah, that't in the "enter non-secure world in non-secure EL2" commit
<diego71>
KotCzarny: because almost every chip?
<KotCzarny>
diego71: some chips have proper dma that isnt affected by cpu clock
<tkaiser>
diego71: It depends on what the chip was made for, you can use old Marvell SoCs that are able to saturate both GbE and SATA at the same time. That's nothing most other SoCs were made for
<tkaiser>
And these Kirkwood SoC were clocked way lower than most more recent SoCs
<diego71>
tkaiser: yes, i'm quite impressed with marvell chip. They are not the fastest around in row cpu power, but it's quite good in I/O. I think there is a reason if it used often in NAS and router.
<diego71>
KotCzarny: true, but I never seen strong correlation between cpu speed and sata performance, so I don't think is a problem of dma in A20.
<keh>
clearfog looks nice (already seen, but not really available in germany as I watched a few weeks ago), but nethertheless it is too big
domidumont has quit [Quit: Leaving.]
iamfrankenstein has quit [Quit: iamfrankenstein]
<diego71>
KotCzarny: there is a big (ihmo) problem in write performance in A20, but anyway is faster than a usb-sata adapter
<tkaiser>
keh: Turris Omnia uses the same SoC and there is a Linksys router also relying on this chip for 120 bucks but with only 512 MB RAM
<tkaiser>
On the Marvell Armada 38x you can fiddle around in u-boot to convert mPCI to mSATA and use mechanical converters to connect a few normal disks
<tkaiser>
And a Solid-Run engineer already confirmed that he was able to max out 3 SATA connections in parallel with this dual core SoC: Above 1.5GB/s
<KotCzarny>
:)
<diego71>
I suspect that problem sata speed is more related with dram speed than cpu clock
<keh>
most crap is the mali400 gfx and their driver support (certanly open source, because... when I think about nvidia / closed source driver and sometimes the network card is also nvidia on some motherboards, it should be not so hard to deliver the framebuffer via network :P)
<keh>
that sounds cool tkaiser
<longsleep>
apritzel: great, will try to use your ATF then in my next image
valkyr1e has quit [Ping timeout: 264 seconds]
<longsleep>
apritzel: currently struggeling with sound, just getting Inappropriate ioctl for device when trying to aplay :/
valkyr1e has joined #linux-sunxi
tkaiser has quit [Ping timeout: 250 seconds]
<apritzel>
well, I briefly looked at it the other day and you need DMA, don't you?
<apritzel>
longsleep: ^^^
<longsleep>
apritzel: DMA for sound?
<apritzel>
for sure you don't want to do PIO every sample in?
<longsleep>
and Firefox is indeed crashing on startup
<apritzel>
that's totally stupid, really
<apritzel>
RedHat decided to support only 64K pages with their Enterprise Linux on AArch64, but I guess they will regret it
<longsleep>
apritzel: and chrome does not seem to exist for arm64 in arch repository, so we are lacking a reasonable browser
<diego71>
apritzel: really? Have you any evidence that 64k pages waste too much memory? just curius...
<diego71>
I mean most of the programs use much more than 64k ...
<longsleep>
apritzel: anyhow, it would be nice if the mainline kernel would compile with 64k pages, the BSP kernel needs fixes i am not willing to make
<apritzel>
64K pages here means 64K granule, so that's the _smallest_ possible chunk of memory allocation
<KotCzarny>
seriously?
<KotCzarny>
even for malloc(256) ?
<diego71>
apritzel: only for segments. It's not like malloc(1) use 64kb :P
<apritzel>
it's not about malloc, really
<apritzel>
but for instance for every file the kernel touches it allocates at least 64K in the page cache
<diego71>
apritzel: 64k pages only means that elf segments must be aligned to 64Kb.
<apritzel>
no
<apritzel>
that's one side effect
<apritzel>
and the reason that compatibility to ARM32 binary breaks
<diego71>
apritzel: I suspect that (broken compatibility)
<apritzel>
I have seen data with running 64K pages on Android, and the waste was enormous
<apritzel>
for RH that's nice, because they don't care about compatibility and so they don't need to support it ;-)
<apritzel>
if an _application_ wants to use bigger pages, there is always (transparent) huge pages
<apritzel>
so really, give it a try: run the same (desktop) workload with 4K and 64K pages and look at the memory consumption
<apritzel>
maybe some specific database workload benefits, but they have other means of optimization
<apritzel>
longsleep: but that 64K pages Firefox crash report is really old (I remember Steve complaining about it ;-)
<diego71>
apritzel: I don't know now, but in the past huge pages were painful
<longsleep>
apritzel: well it is still open and i guess nobody cared :)
<apritzel>
diego71: indeed, but the applications that care cope with that
<apritzel>
and also there is now transparent huge pages
<apritzel>
which means you get it mostly for free
<apritzel>
longsleep: mmh, I will ask Steve tomorrow
<longsleep>
apritzel: That would be nice, so far Firefox is the only thing which crashes what i found.
* apritzel
is afk for a short while, but just keep on typing ;-)
<keh>
Oh, there is! (LiPo Battery connector with battery-charging capabilities) But does it support sata?
<lauri>
keh: I am guessing yes, on Cubietruck I am getting 5V power for USB ports as well
zoobab has quit [Ping timeout: 244 seconds]
zoobab has joined #linux-sunxi
<keh>
lauri: cubietruck also has battery support? But no gbit ethernet as far as I know. Still looking for some sites with more information which soc supporting battery for sata.
FDCX has quit [Remote host closed the connection]
apritzel has quit [Ping timeout: 248 seconds]
paulk-aldrin has quit [Quit: Quitte]
yann|work has quit [Ping timeout: 240 seconds]
yann|work has joined #linux-sunxi
FlorianH has quit [Ping timeout: 250 seconds]
Wizzup has quit [Ping timeout: 244 seconds]
reinforce has quit [Quit: Leaving.]
yann|work has quit [Ping timeout: 255 seconds]
apritzel has joined #linux-sunxi
bonbons has quit [Quit: Leaving]
ricardocrudo has quit [Remote host closed the connection]