cnxsoft has quit [Remote host closed the connection]
cnxsoft has joined #linux-sunxi
agraf has quit [Ping timeout: 250 seconds]
agraf_ is now known as agraf
NeuroScr has joined #linux-sunxi
cnxsoft has quit [Remote host closed the connection]
cnxsoft has joined #linux-sunxi
AneoX has joined #linux-sunxi
AneoX_ has quit [Ping timeout: 250 seconds]
dddddd has quit [Remote host closed the connection]
faisal has quit [Ping timeout: 250 seconds]
HeavyMetal has quit [Ping timeout: 240 seconds]
HeavyMetal has joined #linux-sunxi
HeavyMetal has quit [Changing host]
HeavyMetal has joined #linux-sunxi
suprothunderbolt has joined #linux-sunxi
lurchi_ is now known as lurchi__
tl_lim has quit [Read error: Connection reset by peer]
tllim has quit [Read error: Connection reset by peer]
shfil has quit [Quit: Connection closed for inactivity]
cnxsoft1 has joined #linux-sunxi
cnxsoft has quit [Ping timeout: 245 seconds]
noblock has quit [Remote host closed the connection]
NeuroScr has quit [Quit: NeuroScr]
cnxsoft1 has quit [Ping timeout: 250 seconds]
cnxsoft has joined #linux-sunxi
<suprothunderbolt>
anyone know how to force a clock to be enabled/connected to the I2S0_MCLK pin? This codec won't appear on I2C for me to look at its registers unless it's getting clock.
<suprothunderbolt>
on some other platforms there was a GPIO you could set to high to enable that behaviour
return0e_ has quit [Read error: Connection reset by peer]
return0e has joined #linux-sunxi
DonkeyHotei has quit [Ping timeout: 244 seconds]
Gerwin_J has joined #linux-sunxi
lurchi_ has joined #linux-sunxi
lurchi__ has quit [Ping timeout: 258 seconds]
Gerwin_J has quit [Ping timeout: 264 seconds]
sunilmohan has joined #linux-sunxi
sunilmohan has joined #linux-sunxi
sunilmohan has quit [Changing host]
sunilmohan has quit [Client Quit]
HeavyMetal has quit [Ping timeout: 276 seconds]
HeavyMetal has joined #linux-sunxi
HeavyMetal has joined #linux-sunxi
<suprothunderbolt>
I'm still missing how to get it to send MCLK on the MCLK pin :) The clock rate is set correctly but if I look at that pin (PB3) on the scope there's no signal
<suprothunderbolt>
if I print the clock rate in the driver for the requested pin it's correct (24.576)
<suprothunderbolt>
looks like "I2S/PCM Clock Divide Register" eanbles the MCLK, any one know how I should see what it's current value is and change it?
Mandragor has quit [Ping timeout: 276 seconds]
Mandragor has joined #linux-sunxi
mpmc has quit [Ping timeout: 245 seconds]
Gerwin_J has quit [Read error: Connection reset by peer]
<codekipper>
is everything probed?....get in a habit of posting logs and changes somewhere.
<suprothunderbolt>
I've got that in my DTS and those pins are setup: pins = "PB3";
<suprothunderbolt>
function = "i2s0";
hipboi has joined #linux-sunxi
hipboi has quit [Remote host closed the connection]
<suprothunderbolt>
the module probes correctly and I've added a few printk statement, the clocks base value wasn't correct on startup because the codec wasn't requesting any specific clock rate
<suprothunderbolt>
but currently nothing is actually output on the pin.
<suprothunderbolt>
have you run the a64 with an external DAC? Were you using the A64 as MCLK or did you have an external crystal?
<suprothunderbolt>
I might just add a crystal if it's easier...
nuuuciano has quit [Ping timeout: 258 seconds]
<suprothunderbolt>
./devmem2 0x01C20824
<suprothunderbolt>
-- 33333777 so the pin is configured correctly
<suprothunderbolt>
oh nice. Is there an example DTS for that board?
<suprothunderbolt>
is that the one in your github?
<codekipper>
yep..the above link
<suprothunderbolt>
I'm just looking through the registers to see why clock isn't enabled. This codec annoyingly doesn't respond to I2C unless it's getting an MCLC signal that makes sense
<wens>
suprothunderbolt: does the codec already have a driver? and DT bindings?
cmeerw has joined #linux-sunxi
TheSeven has quit [Ping timeout: 264 seconds]
<suprothunderbolt>
yes
return0e_ has quit [Ping timeout: 258 seconds]
<wens>
does it take a clock in the binding?
<suprothunderbolt>
wens: I have the codec working correctly on the AM3358 platform
<suprothunderbolt>
yes.
<suprothunderbolt>
clocks = <&ccu CLK_AC_DIG>;
<suprothunderbolt>
clock-names = "scki";
return0e has joined #linux-sunxi
TheSeven has joined #linux-sunxi
<wens>
hmm, I remembered wrong
<wens>
the i2s mclk is not exported
<suprothunderbolt>
./devmem2 0x01C22024
<suprothunderbolt>
outputs 00000000
<wens>
the i2s driver does runtime PM, so if the interface is not active, the driver shuts down the block, and you read all 0
<suprothunderbolt>
ahh okay, that makes sense.
<suprothunderbolt>
Can I disable that?
<suprothunderbolt>
I want no PM. The I2S will always be in use.
<wens>
you can, though I dont remember how right now
<KotCzarny>
if you use it, pm wont run, no?
<KotCzarny>
that's the logic of pm
<suprothunderbolt>
at least I can see why went I just tried to write values to those registers nothing happened
<suprothunderbolt>
KotCzarny, not sure how I can get the i2s to be used when I'm attempting to work out why the MCLK isn't being output.
tllim has quit [Read error: Connection reset by peer]
<suprothunderbolt>
I assume something has to set that register but I'm not sure where and how :(
<KotCzarny>
something should start using the audio?
<suprothunderbolt>
KotCzarny: the codec can't register without the MCLK being active.
<KotCzarny>
how the other codecs do it?
<suprothunderbolt>
so I'm attempting to get the MCLK to output on the pin.
<suprothunderbolt>
on other platforms there was a GPIO you could set to high to force MCLK to be output
<KotCzarny>
no, would be interesting to compare with other codecs functionality on sunxi platform
<KotCzarny>
i would assume there are working ones
<suprothunderbolt>
codekipper mentioned he had a working one but it doesn't seem to do anything cunning to enable MCLK.
<suprothunderbolt>
maybe the PM is the issue as maybe it's enabled correctly but never powers up for long enough for the codec to wake up. The codec needs to see a certain number of MLCK before it powers up.
<suprothunderbolt>
which seems like a quirk specific to this TI part.
<KotCzarny>
or it opens/initializes something that indirectly disables pm/oenables mclk?
<wens>
if the codec driver asks ASoC to drive a master clock, it will enable mclk on the i2s interface
<wens>
i.e. the codec is running in slave mode
<wens>
you also want to make sure that you muxed in the MCLK, along with the other I2S pins/signals
<suprothunderbolt>
I think I've done those things but I'll double check :) Thanks.
<suprothunderbolt>
on forces the device on, auto allows PM
<suprothunderbolt>
now I can set values through devmen2 :) Thanks for telling me about the PM issue!
<suprothunderbolt>
ahh and now when I run i2cdetect the codec is there
f0xx has joined #linux-sunxi
<sunilmohan>
jo0nas: I was looking some more into the poor performance on Rev.C. Workaround for RTL8211C was introduced to force it in master mode. Then seems to be removed later (perhaps inadvertantly).
<suprothunderbolt>
I'll attempt to get the TDM stuff working now then :) If anyone had an idea about that I'd love to know. I assume I just need to write the registers for channel mapping / slot width
clemens3 has joined #linux-sunxi
msimpson has joined #linux-sunxi
clemens3_ has joined #linux-sunxi
clemens3 has quit [Ping timeout: 246 seconds]
<jo0nas>
sunilmohan: yes, I am aware of the force-master hack - that's why I asked if you used a switch or a cross-over cable: That affects the effectiveness of that hack
chomwitt has joined #linux-sunxi
robogoat has quit [Ping timeout: 252 seconds]
robogoat has joined #linux-sunxi
<wens>
suprothunderbolt: codekipper had some patches (I believe he mentioned it the last few days) along the lines
<suprothunderbolt>
wens, alas he suggested his patches aren't for TDM.
<suprothunderbolt>
Though I've got no idea on the differences between HDMI multichannel output and normal TDM.
<suprothunderbolt>
codec I'm using is pcm3168A. I think I've got some AD codecs I could test with too.
tnovotny has joined #linux-sunxi
<jo0nas>
sunilmohan: or rephrased: Even _without_ the force-master hack the same behaviour should be achieved by forcing the other end of a cross-over connection into slave mode
<jo0nas>
...not that I have any experience with doing that - but I guess mii-tool can do it
msimpson has quit [Read error: Connection reset by peer]
<codekipper>
suprothunderbolt: I think it's possible to do what you want...with multichannel hdmi we have 4 data lines which transmit 2 channels on each. The TDM solution has 1 data line which will transmit all 8 channels.
<sunilmohan>
jo0nas: I see. I am also building with force master enabled to see if that fixes my issue.
<codekipper>
I'm just looking to see how I can change the setup for your solution....it would be easier if you concentrated on getting 2 channels working first!
<suprothunderbolt>
codekipper, will do. I'm just working out how to persist this power management change so that the codec is detected properly on start up. It now at least appears after I disable I2S PM.
<suprothunderbolt>
hah yeah. And at least it's fast enough for this. I was using it for DSI stuff and it's only 100mhz, which wasn't handy for DSI :)
megi has joined #linux-sunxi
voodoovood2 has joined #linux-sunxi
msimpson has joined #linux-sunxi
voodoovood has quit [Ping timeout: 276 seconds]
sunilmohan has joined #linux-sunxi
sunilmohan has quit [Changing host]
sunilmohan has joined #linux-sunxi
Mangy_Dog has joined #linux-sunxi
<jo0nas>
sunilmohan: welcome back (power outage?)
<sunilmohan>
jo0nas: The Only Rev.C board is my production FreedomBox which runs my Quassel server :)
<jo0nas>
sunilmohan: great that you investigate force-master - just beware that omitting the force-master hack does not necessarily trigger the problem that hack solves: Only when device is (randomly or - if possible - forced) in slave mode is the issue present (as I understand it)
<jo0nas>
ahh :-)
<sunilmohan>
I tried the patch with force master.
<sunilmohan>
Mixed results, I don't know what to make of it.
<sunilmohan>
Before the patch: connect to router - bad, connect to laptop - some good/some bad.
* jo0nas
has ordered a rev.K board so if noone provides tests for that board before then I will try do that when I get the board
<sunilmohan>
After that patch: connect to router - good, connect to laptop - some good/some bad.
<jo0nas>
see my note above about no force-master not being the opposite of force-master!
<sunilmohan>
I am thinking of ordering a rev. K board too.
<jo0nas>
Tsvetan: would you perhaps be willing to sponsor the board for sunilmohan?
<sunilmohan>
If I do `mii-tool eth0 -r` the result changes somtimes. Perhaps that is expected.
NeuroScr has joined #linux-sunxi
<jo0nas>
auto-negotiation implies randomness!
<sunilmohan>
Yeah, may be I need to do better control my tests.
<sunilmohan>
I will test again tomorrow.
<jo0nas>
I suggest to try figure out a way to probe current state of connection (not change it) using either mii-tool or ethtool
<jo0nas>
...and include that info with your tests
<jo0nas>
see my latest posts to the Debian bug
<jo0nas>
so many variables here...
<sunilmohan>
I shall be sure to collect that information.
<jo0nas>
cool!
<sunilmohan>
Another issue could be that I am testing after kernel starts. May be I should be testing with just u-boot.
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
<sunilmohan>
jo0nas: Thanks for all information so far.
MoeIcenowy has joined #linux-sunxi
<sunilmohan>
My Pioneer Edition Rev.G2 box should be arriving tomorrow :) Should help with more testing.
<jo0nas>
nice!
<jo0nas>
I only understand the information now myself - catching up in parallel to your passing more info :-)
<jo0nas>
sunilmohan: thanks for all the test data!
yann has joined #linux-sunxi
<suprothunderbolt>
mmm... so I've disabled PM in the I2S block, I can now reboot and execute ./devmem2 0x01C22024 w 0x173 and the codec is visible, but the clock doesn't seem to be enabled during the codec probe.
<suprothunderbolt>
i.e. I have to execute the register setting
kaspter has quit [Read error: Connection reset by peer]
kaspter has joined #linux-sunxi
<suprothunderbolt>
ahh so the I2S code only sends a clock after the set format is called, but codec code checks if it can find the codec before calling that, even though it doesn't power up until it gets clock.
chomwitt has quit [Quit: WeeChat 1.6]
mpmc has quit [Ping timeout: 258 seconds]
mpmc has joined #linux-sunxi
suprothunderbolt has quit [Ping timeout: 246 seconds]
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
<libv>
rellla: nope, but how would that work? multi kernel, and uboot and devicetree board specific?
<diego71>
rellla: last time I used them, they work
<diego71>
libv: they have a different uboot for board
<rellla>
libv: i have done that for my cubieboard. cat firmware.img + roots.img and you get the netinstall image. works so far.
<rellla>
my question is how to rebuild the firmware/uboot part?
<rellla>
i'd like to create some images for my devices to avoid armbian, just to have plain debian.
<rellla>
is there any place where i could find some build scripts, how the debian people automate the images' build?
* rellla
is to lazy to do the sd card creation from scratch ...
<libv>
rellla: i rewrote the debian image creation bit in the wiki 2 weeks ago
<libv>
it's really not that scary
<libv>
having said that, i would love to have the image based solution documented as well
<KotCzarny>
he said 'too lazy' not 'too scared'
<KotCzarny>
so unless there is some ready to use script available..
<libv>
in sunxi style, with sunxi specific, with sunxi bits, on the sunxi wiki
<libv>
KotCzarny: i am a lazy person as well, and scared of and annoyed by pointless work
<KotCzarny>
i love scripting and automation
<libv>
which is why i invest so much time in the wiki, as i really really dread doing work twice
<libv>
i do not like automation and scripting here, as things constantly evolve, and any digressions become either impossible or become features which then vastly complicates things...
<libv>
making the scripts even less flexible
<libv>
which then means that you spend all your time fixing the scripts
<libv>
so for things sunxi, automation does not work imho
<KotCzarny>
adding sunxi bits to any arm installer/distro is usually as simple as replacing uboot, kernel and modules
<KotCzarny>
and the rest could be done via apt-get (in case of debian)
<libv>
KotCzarny: do you really need a script for that, or would running through a short wiki snippet with 3 commands that you need to do be much more effective, more general and more flexible?
<libv>
plus, with a script, you lose the wiki effect
<KotCzarny>
script might be part of the wiki page
<libv>
i do not buy that
<rellla>
libv: i agree in general, but using the prebuild-debian images is really fine. last build was done 2019-04-10, so it's recent
<libv>
i also do not like complicated multiline copy-paste bash things for the wiki
<libv>
rellla: as said, i would love to see that necessary info for making it sunxi specific added to our wiki
<libv>
with referrals to debians own info where appropriate
<libv>
KotCzarny: i believe that people should know what they are doing, so they can fix things or adjust things as needed
<libv>
and we are not exactly documenting rocket science here
<KotCzarny>
comments are usually fine for that
<KotCzarny>
but yeah, in general people should know what they are going to do
<libv>
rellla: but to reiterate, debootstrap is really not that painful, it just takes a tiny bit longer as it pulls in individual packages
<libv>
i also dislike wiki snippets depending on environment variables
<libv>
as those usually get set somewhere earlier on the page
<libv>
and when there is page churn, and blocks get moved to other pages, this stuff vanishes
<rellla>
i'd just like to see, how they do that, and if i find it comfortable, add other devices to the build scripts. it would just
<rellla>
be ~2 options regarding dtb and uboot.
<libv>
rellla: can one not point the script to a specific uboot binary and a dtb file?
<rellla>
i'm on search for that script...
<libv>
i still hope that the wiki still somewhat has a "so you have an android device with an allwinner in it, here is how you get to running a real linux"
<libv>
ok :)
<libv>
... feel to it
<rellla>
i think you know, that i debootstrapped and dd'ed enough already :)
<libv>
true
<rellla>
though i agree with you on the wiki points.
<jo0nas>
rellla: Debian developer here (but I don't maintain the debian-installer parts specifically, which those netinst images are part of)
<jo0nas>
the Debian approach involves building packages from each upstream source projects, and then building images pulling in binary parts from various packages
<rellla>
joOnas: where is the right place to catch the right people?
<jo0nas>
you mean irc? Debian developers use OFTC.net (not Freenode), and there are many many channels, most of them starting with "debian-"
<jo0nas>
you are unlikely to catch debian-installer developers on irc - there is mostly only one person working on that, who in my (possibly wrong) experience is not often on irc
<jo0nas>
...but if your interest is in covering different boards than Debian currently supports (as opposed to hacking on features for existing produced images) then _before_ messing with debian-installer framework you should make sure that all involved packages support your target device - i.e. most likely linux kernel, u-boot and flash-kernel
<jo0nas>
...but if you are lazy and your board can use a rootfs from another supported board - i.e. you only really need a different u-boot and different device-tree file for an already built Debian linux kernel, then you might instead try use an existing netinst image and just overwrite the u-boot bits with dd
<jo0nas>
...and if that works (your boot succeeds until loading initramfs) then I can help you in how to repackage the debian-installer initramfs to inject a custom device-tree file
<rellla>
though it would be interesting, how this is done automatically for the debian repo. there should be a script for the nightly builds, shouldn't it?
<jo0nas>
that approach obviously cannot be adopted as-is for Debian officially, but might be sufficient for your personal needs, and can also help guide others implement official support into debian-installer (so please do blog about your results!)
<jo0nas>
there are scripts that grab binary packages of u-boot and linux kernel and flash-disk
<jo0nas>
those scripts are useless if the binary packages grabbed lack the support you want
<rellla>
so, mainly i search for THE git repo for that, which includes the scripts like armbian, libreelec, buildroot have it. if such repo or even sripts exist.
<jo0nas>
there is no single git repo for everything
<jo0nas>
Debian builds things in multiple stages, involving 50.000+ source trees, some but not all tracked in git
<rellla>
ok thanks. maybe i just debootstrap again :) ^libv
<jo0nas>
I recommend that you look into hacking the Debian package for u-boot - and when you have a u-boot package that can boot on your board then I can point you to the commands I use to repackage netinst images for injecting custom stuff in there
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
MoeIcenowy has joined #linux-sunxi
<rellla>
joOnas: isn't this just some loop-mounting the img and add the bits? and adapting initramfs maybe?
<jo0nas>
not sure what you are asking - sure debian-installer is a huge pile of "just" executing scripts and commands in the right order...
<jo0nas>
please note that those build notes are for an A64 board - routines are different (simpler!) for A20 boards
<rellla>
... i'm not sure i want to dig into d-i anymore :) but thanks for the hints. i have to read some bits first :p
<jo0nas>
fair enough :-)
NeuroScr has quit [Quit: NeuroScr]
lurchi_ is now known as lurchi__
megi has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
steev has quit [Ping timeout: 252 seconds]
narmstrong has quit [Ping timeout: 268 seconds]
narmstrong has joined #linux-sunxi
steev has joined #linux-sunxi
jernej has quit [Read error: Connection reset by peer]
jernej has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
Rafael1980 has joined #linux-sunxi
msimpson_ has joined #linux-sunxi
msimpson has quit [Ping timeout: 246 seconds]
Putti has joined #linux-sunxi
msimpson_ has quit [Remote host closed the connection]
msimpson_ has joined #linux-sunxi
msimpson_ has quit [Ping timeout: 258 seconds]
megi has quit [Ping timeout: 246 seconds]
msimpson_ has joined #linux-sunxi
dddddd has joined #linux-sunxi
nashpa has quit [Ping timeout: 245 seconds]
msimpson_ has quit [Ping timeout: 246 seconds]
nashpa has joined #linux-sunxi
msimpson_ has joined #linux-sunxi
msimpson_ has quit [Ping timeout: 258 seconds]
msimpson has joined #linux-sunxi
afaerber has quit [Quit: Leaving]
cnxsoft has quit [Quit: cnxsoft]
megi has joined #linux-sunxi
jstefanop has joined #linux-sunxi
libv_ has joined #linux-sunxi
lennard has quit [Ping timeout: 252 seconds]
tllim has joined #linux-sunxi
libv has quit [Ping timeout: 276 seconds]
JohnDoe_71Rus has joined #linux-sunxi
afaerber has joined #linux-sunxi
Putti has quit [Ping timeout: 250 seconds]
jstefanop has quit [Remote host closed the connection]
jstefanop has joined #linux-sunxi
jstefanop has quit [Ping timeout: 258 seconds]
clemens3_ has quit [Ping timeout: 245 seconds]
msimpson has quit [Quit: Leaving]
victhor has joined #linux-sunxi
voodoovood has joined #linux-sunxi
voodoovood2 has quit [Ping timeout: 258 seconds]
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
libv_ is now known as libv
tnovotny has quit [Quit: Leaving]
lennard has joined #linux-sunxi
<Guest32482>
I've been working with my Orange Pi Win Plus running an upgraded Xubuntu image (19.04 now) with A64 a lot over the past few weeks and been able to use it as a desktop even playing YouTube videos in the Luakit browser and streams in VLC and I was wondering if there are still performance improvements that could be made, even after I replace the current image with one containing the new mainline 5.2 kernel and Lima
<Guest32482>
driver
<KotCzarny>
this is tv box, not desktop computer
<KotCzarny>
you might try libreelec
wens has quit [Ping timeout: 255 seconds]
wens has joined #linux-sunxi
camh has quit [Ping timeout: 255 seconds]
camh has joined #linux-sunxi
<Guest32482>
well, I've been using it as a poor man's desktop with compilers and IDEs such as GNOME Builder installed for almost 2 years until I get something better like Odroid N2
<Guest32482>
but I will definitely have a look at that
kaspter has quit [Read error: Connection reset by peer]
kaspter has joined #linux-sunxi
vagrantc has joined #linux-sunxi
_0x5eb_ has joined #linux-sunxi
libv_ has joined #linux-sunxi
libv has quit [Read error: Connection reset by peer]