rellla changed the topic of #linux-sunxi to: Allwinner/sunxi /development discussion - did you try looking at our wiki? - Don't ask to ask. Just ask and wait! - - Logs at - *only registered users can talk*
RichardG867 has quit [Ping timeout: 264 seconds]
ldevulder_ has joined #linux-sunxi
ldevulder has quit [Ping timeout: 248 seconds]
\\Mr_C\\ has quit [Quit: (Read error: Connection reset by beer)]
RichardG867_ is now known as RichardG867
vagrantc has quit [Quit: leaving]
blscoe has quit [Quit: Leaving]
NeuroScr has quit [Quit: NeuroScr]
cnxsoft has joined #linux-sunxi
_whitelogger has joined #linux-sunxi
lurchi_ is now known as lurchi__
yy has joined #linux-sunxi
megi has quit [Ping timeout: 258 seconds]
TheSeven has quit [Ping timeout: 264 seconds]
TheSeven has joined #linux-sunxi
yy has quit [Remote host closed the connection]
NeuroScr has joined #linux-sunxi
rexxster has joined #linux-sunxi
dddddd has quit [Remote host closed the connection]
smooker has quit [Remote host closed the connection]
chewitt has joined #linux-sunxi
cnxsoft has quit [Ping timeout: 245 seconds]
lurchi_ has joined #linux-sunxi
lurchi__ has quit [Ping timeout: 248 seconds]
cnxsoft has joined #linux-sunxi
default__ has joined #linux-sunxi
ldevulder_ has quit [Ping timeout: 258 seconds]
selfbg has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
freemangordon has joined #linux-sunxi
sunshavi has quit [Read error: Connection reset by peer]
chewitt has quit [Quit: Zzz..]
notme2 has joined #linux-sunxi
notme2 has quit [Client Quit]
chewitt has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
reinforce has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
TheSeven has quit [Ping timeout: 264 seconds]
TheSeven has joined #linux-sunxi
cristian__c has joined #linux-sunxi
cristian_c has quit [Ping timeout: 258 seconds]
default__ is now known as ldevulder
reinforce has quit [Quit: Leaving.]
chewitt has quit [Remote host closed the connection]
reinforce has joined #linux-sunxi
libv_ has joined #linux-sunxi
NeuroScr has quit [Quit: NeuroScr]
libv has quit [Ping timeout: 244 seconds]
libv_ has quit [Ping timeout: 245 seconds]
libv has joined #linux-sunxi
diego_r has joined #linux-sunxi
BenG83 has joined #linux-sunxi
warpme_ has joined #linux-sunxi
yann|work has quit [Ping timeout: 258 seconds]
NeuroScr has joined #linux-sunxi
tllim has quit [Read error: Connection reset by peer]
camus has joined #linux-sunxi
kaspter has quit [Ping timeout: 248 seconds]
camus is now known as kaspter
<rellla> hey jernej, what is the actually used decoding pipeline in libreelec regarding cedrus? is it just ffmpeg->v4l2-request ?
<rellla> and what codecs are covered? is interlaced mpeg2 and h264 possible already?
wigyori has quit [Remote host closed the connection]
cristian_c has joined #linux-sunxi
cristian__c has quit [Ping timeout: 244 seconds]
aalm has quit [Quit: xyz 2.3]
tnovotny has joined #linux-sunxi
wigyori has joined #linux-sunxi
yann|work has joined #linux-sunxi
msevo has joined #linux-sunxi
SopaXorzTaker has joined #linux-sunxi
yann|work is now known as yann
yann is now known as Guest5409
Guest5409 is now known as yann
warpme_ has quit [Read error: Connection reset by peer]
warpme_ has joined #linux-sunxi
lapapat has quit [Ping timeout: 260 seconds]
megi has joined #linux-sunxi
AneoX has joined #linux-sunxi
lapapat has joined #linux-sunxi
Mangy_Dog has joined #linux-sunxi
lapapat has quit [Ping timeout: 260 seconds]
Xukkot has joined #linux-sunxi
<jernej> Relll
<fALSO> yo jernej
<fALSO> i also want to know your answer
<fALSO> tried vlc, and it kinda sucks
SopaXorzTaker has quit [Ping timeout: 258 seconds]
<jernej> rellla: yes, only ffmpeg is used, which directly interacts with driver
<fALSO> how does ffmpeg output to the screen ?
<jernej> Currently supported codecs are mpeg2, h264 and h265
<rellla> jernej: so the seperate libva-v4l2-request backend isn't used at all?
<jernej> Interlacing is supported with my improvements (hacks) to the driver
<jernej> Correct
<fALSO> all the h264 that i tried in vlc with that v4l2-request
<fALSO> it played with the correct fps
<fALSO> but, the places where it moved a lot
<fALSO> had smoe blocks
<fALSO> some
Xukkot is now known as SopaXorzTaker
jorik_ has joined #linux-sunxi
<rellla> jernej: ok thanks. so it should be possible to hook it all up to any app, that uses the ffmpeg pipeline?
<jernej> fALSO ffmpeg is used only for decoding, rendering is done using DRM planes using Kodi's own code with a help of libdrm
jonkerj has quit [Ping timeout: 246 seconds]
<fALSO> ahhh so that isnt possible without kodie
<fALSO> on X or on the FB
<jernej> Drm planes gives best performance...
<rellla> jernej: ...respecting drm output of course
<fALSO> req: how to drm planes ;-D
<jernej> yes, you can use this in any app you want
<jernej> But you have to do your own rendering, you don't have to use drm planes if you don't want to
<jernej> Anyway, take a look into Kodi's DRMPRIME codec and renderer, it's all pretty generic
<fALSO> i wish i had your know how jernej
<fALSO> maybe some day
<fALSO> :-D
<fALSO> yesterday i also saw a spinning horse
<fALSO> with the LIMA driver
<fALSO> on glmark2
<fALSO> at 60fps
<fALSO> got pretty excited
<fALSO> :)
msevo has quit [Quit: Leaving]
diego_r has quit [Ping timeout: 258 seconds]
cristian_c has quit [Ping timeout: 248 seconds]
cristian_c has joined #linux-sunxi
<jernej> Yes, but for h264 it has a hack, which makes it incompatible with current mainline driver. You also have to take cedrus kernel patches
<fALSO> im gonna wait until it hits mainline
<fALSO> the changes on cedrus, or the v4l2 library
<fALSO> well the v4l2 library i built from master, so probably its kernel patches that are needed
matthias_bgg has quit [Read error: Connection reset by peer]
matthias_bgg has joined #linux-sunxi
dddddd has joined #linux-sunxi
cristian_c has quit [Ping timeout: 245 seconds]
cristian_c has joined #linux-sunxi
<fALSO> sirs, what is the correct name
<fALSO> of the connector on orange pi boards
<fALSO> to connect a camera
<fALSO> which keyword should i use on ali express to find one of those cameras?
cnxsoft has quit [Quit: cnxsoft]
<gnarface> i'm guessing "csi input connector camera"
<gnarface> based on looking at
<fALSO> thanks
<fALSO> found some that say that they are for the orange pi PC
<fALSO> gonna buy one,
ldevulder has quit [Quit: Leaving]
ldevulder has joined #linux-sunxi
<libv> fALSO: they will not work out of the box with mainline
<libv> and even the csi0 code from free-electrons is a non-starter without hacks that only mripard knows about it seems
<fALSO> i love the difficulty
<fALSO> ;-P
<fALSO> with falso bothering a million people it will probably get supported quickly
<fALSO> :-P
ldevulder has quit [Client Quit]
<libv> fALSO: you seem to be purely offloading that difficulty onto others, and i see no entries in the wiki that result from it
<fALSO> i'm just a tester
<libv> and?
<fALSO> i dont have enought C knowledge to patch stuff
<libv> and?
<fALSO> well, i've hacked a patch to make network work on the orange pi one plus
<libv> TX_DELAY?
<fALSO> but im afraid of submiting it to the mailing list , because i can get something wrong
<fALSO> libv, nah, it simply isnt detected on mainline
<libv> have you at least tried documenting the issue on the wiki?
<fALSO> its not an issue, its just missing definitions on the DTS file
ldevulder has joined #linux-sunxi
<libv> so your test result, and the fact that you have asked a bunch of questions here, is that you have a working system, somewhat, which is not reproducible, and no-one else gets to benefit from your testing?
<fALSO> you want me to add the patch to the wh
<fALSO> wiki ?
<fALSO> or a section
<fALSO> "how to get network working on mainline" ?
<libv> that's the least one could do
<fALSO> ill update that page
<fALSO> gimme 1 second
<libv> also seems to be a support black hole.
<fALSO> i just did my first edit libv :D
<libv> cool! keep it up
<libv> now people will know that they are not just sinking their time into a black hole
<fALSO> libv, should i like to my patch, or attach it to the wiki ?
<fALSO> like -> link
<libv> send it to the linux-sunxi ml, that is not as picky as lkml, and a good staging area
<fALSO> libv, did another edit
<fALSO> my english problaby isnt the best :-X
<fALSO> libv, i think that this page should also be edited
sunshavi has joined #linux-sunxi
<fALSO> it doesnt mention anything about the arm trusted firmware that also needs to be built
<fALSO> at least for arm64 boards
<DuClare> Why even bother with the wiki if you're just going to be called a black hole in turn? :|
<fALSO> well, i've learned a lot from that wiki
<fALSO> its nice to contribute
gnarface has quit [Quit: Leaving]
AneoX has quit [Quit: Textual IRC Client:]
<libv> DuClare: ?
<fALSO> also made a user page :-)
<libv> fALSO: cool, this way things do not get lost in time, and others can keep information up to date
<fALSO> libv, im getting the courage to configure git-send-email
<fALSO> with a gmail account
<fALSO> and submit the patch to the mailing list
<libv> :)
<fALSO> as it seems that no one is giving love to the Orange Pi One Plus
<fALSO> did you saw what i said about the arm trusted firmware?
<montjoie> fALSO: I will give some love soon
<fALSO> nice
<fALSO> i got it booting, u-boot works OK
<montjoie> it will be the first h6 in kernelci
<fALSO> and got it with ethernet working
<fALSO> no hdmi output though
<fALSO> and a lot of warnings on dmesg and errors :-P
<fALSO> but it "works"
<libv> DuClare: helping users with purely proprietary boards, that are intended to stay proprietary (are you even handing your customer the changes to the source code of u-boot and the kernel?), is per definition, not something that helps the sunxi community along much
<fALSO> libv, sorry if you think im always boterhing everyone
<libv> DuClare: also, the allwinner S3 does not even have a soc page it seems, it has very few mentions on the wiki
<fALSO> i mostly talk with jernej and moeicenowy
<libv> fALSO: it just was very one-sided
<fALSO> if they think im bothering, ill stop
<fALSO> i just like to DRINK from the knowledge of people that know more than me
<libv> no, just contribute back what you find, document what you run into and document the fixes
<fALSO> i helped jernej fix a bug that only happaned to me
<DuClare> libv: Shrug, my experience with the linux-rockchip community is that many people who started with proprietary boards also became quite experienced and turned out to be very useful contributors
<fALSO> in all the orange pi boards ;-)
<fALSO> if you had a huge monitor like mine
<fALSO> 2560x1080
<fALSO> it didnt gave any output
<fALSO> now it is fixed =)
<fALSO> so i think i'm giving something back
<libv> oh, the hdmi clock
<fALSO> hehe
<fALSO> yes libv
<libv> i read that commit just yesterday :)
<fALSO> jernej is a MASTER
<fALSO> well, from what i've seen
<fALSO> there have been a lot of H6 related commits
gnarface has joined #linux-sunxi
<fALSO> the only thing missing is probably the GLUE in the dts
<fALSO> for the HW to be recognized
<fALSO> I havent tried with recent kernels, but... the DTS hasnt been changed in a while, so i guess it is still the same
<fALSO> gonna cross compile a kernel for the ONE PLUS now
<fALSO> ill report later ;-D
<libv> :)
* libv goes back to cleaning up his csi1 code for fosdem-video (rant's over, for now)
<fALSO> lololol
<fALSO> im going to buy that camera
<fALSO> even if it doesnt work
<libv> fALSO: well, i went and bought some camera modules for the banana pi, to verify csi0 code functionality, only to find that it does not work out of the box (and the claim is that it needs some ov5640 changes, but no details as to what changes are needed), i then proceeded to rewrite the register tables to real code, and posted the preliminary code here, and no-one cared
<fALSO> well, if i manage to get one working
<libv> since we have a pretty hard deadline for fosdem in february, and we needed to verify the bandwidth capabilities of the csi1 engine asap, i then went off and wrote csi1 support from scratch, and i am cleaning up the commits now, so that the other fosdem guys can go and play with this as well
<fALSO> ill put an orange pi on the balcony
<fALSO> making a timelapse of my plants grwoing
<fALSO> libv, nice
<fALSO> a friend of mine went to fosdem
<libv> so do not expect the camera to work out of the box
<fALSO> i really wanted to go, but ... being poor sucks :-P
<libv> fosdem is amazingly honest and totally grass roots, i would not miss it for the world, and am glad to be a small part of it
<fALSO> i would love to hang out with the sunxi people
<fALSO> the gentoo people
<fALSO> and the bsd people
<fALSO> =)
<fALSO> those are my "loves"
<fALSO> hehee
<DuClare> I'd just end up standing in the corner, alone
<fALSO> nah, im cool, i would try to talk with people
<fALSO> until libv told me to shut up :-PPPPPPPPPpppppppppppppppppp
<fALSO> hehe
<libv> DuClare: well, those devrooms where you would try to stand in a corner if you are interested in sunxi and related things, would put you into close bodily contact with others
<libv> and you would be able to spot some people rather quickly, and people at fosdem are usually happy to chat
<fALSO> hehehe
<fALSO> i also have some social phobia
<fALSO> but the people here seem to be cool
<fALSO> probably not many people with suits and TIES
<DuClare> I'm thinking that if casual apparel makes a convention relaxed and more approachable for people with social phobia, then perhaps we need a no-pants convention
<fALSO> ill easily go and talk with a guy with shorts and tshirt, than one with a full suit
<libv> there's probably a few people in here who are wearing underpants only
<fALSO> hehehehehehehe
* fALSO always in shorts and tshirt
<KotCzarny> isnt social phobia people afraid of all people? no matter how they are dressed
<libv> i actually am wearing a t-shirt as well as it is not warm enough ;p
<fALSO> my phobia is about people that i dont know
<willmore> fALSO, if you need English proof reading, I'm glad to do it.
<libv> fALSO: imagine 12-14k geeks/nerds all gathererd in one space
<fALSO> if i start to talk to them and they seem cool, i get more relaxed
<fALSO> libv, Awesome
<fALSO> i sometimes organize an "weekend"
<fALSO> with all my geek/nerd IT friends
<libv> then know that the devroom you are probably most interested in is the graphics/hw enablement devrooms
<fALSO> here in portugal
<libv> and only 120 people fit in there
<fALSO> yes, i read that, that the rooms are filled to the brim
<fALSO> you need to get early etc
<libv> everyone around you is interested in the same stuff, so you do not need an icebreaker
<fALSO> or maybe just dont leave the chair, even if you dont like the next presentation
<fALSO> =)
<libv> well, there are physical constraints to how many people can fit in which room
<libv> and to aleviate those issues, there's close to 100% capture of talks with close to 100% live streaming
<fALSO> maybe in the next year, im gonna try to go
<fALSO> a portuguese friend of mine that is living in the netherlands
<fALSO> has been going
<fALSO> and always asking me to go
<libv> so when i have to throw people out of the room, i tell them to get their laptops out and watch the livestream from the benches/chairs just outside the devroom
<fALSO> lol
<libv> there's only a few minutes delay
<megi> fALSO: what are your current issues with H6 Opi One Plus?
<fALSO> i wasnt complaing about anything :(
<fALSO> i was just stating that the dts still doesnt enable the ethernet
<fALSO> current issues: no hdmi output
<fALSO> probably that one is the worse
<fALSO> hehe
<megi> Opi One Plus, has a special circuit to enable DDC (to read EDID), and it's not mainline
<megi> in mainline it's not enabled
<megi> it should work in some reduced resolution mode at least, though
<fALSO> i tried sunxi-next, but my kernel config for mainline
<fALSO> didnt boot with sunxi-next
<megi> I have Opi3 which is very similar
<fALSO> there should be a repo on the wiki with BASE kernel configs for each board
<wens> my opi one plus is in its box
<fALSO> or a list of "drivers" that should be enabled to make the board boot
gaston_ has joined #linux-sunxi
<megi> that's mostly optimization. sunxi_defconfig should work for most boards, no?
<fALSO> never tried, probably that one doesnt
<fALSO> maybe theres a sunxi64_defconfig
<fALSO> gonna try
<fALSO> :)
<megi> maybe, I try to reduce my configs
<megi> so I don't use these defconfigs, anyway
<megi> my config for opi3
<KotCzarny> sunxi defconfig should just include all known sunxi drivers, it's up to dts to use them
<KotCzarny> but i also finx sunxi_defconfig lacking in many areas
<KotCzarny> *find
<fALSO> megi, aahh youre MEGOUS
<megi> yup
<fALSO> i've mailed you once :-D
<megi> did you?
<fALSO> yes and you responded
<fALSO> i think it was about a config also LOL
<megi> ah
<fALSO> nah, it was about boot.cmd
<fALSO> and other was that your kernel had some EXTRA firmware enabled
<fALSO> and it didnt build for me
<fALSO> =)
<megi> yes, I remember
<fALSO> heheh
lapapat has joined #linux-sunxi
<wens> h6 is probably the only soc in rapid active development right now?
<megi> I'm slowly trying to get erquired bits for Opi3 mainline
<megi> wens: seems so
<megi> ethernet still needs some testing, because of phy startup issues, that I suspect are caused by not all ephy regulators being off after reboot
<megi> I have some patch that forces them off, to try
<megi> but I need someone who actually had the issue with phy not starting after reboot, to test :)
<megi> jernej: maybe you can help ^
<megi> I remember you had this issue, too. Patch is here, if you want to try:
NeuroScr has quit [Quit: NeuroScr]
gaston_ has quit [Remote host closed the connection]
gaston_ has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: KVIrc 5.0.0 Aria]
selfbg has quit [Quit: selfbg]
selfbg has joined #linux-sunxi
<lapapat> megi: which phy does Opi3 use?
<megi> RTL8211E
lapapat has quit [Remote host closed the connection]
lapapat has joined #linux-sunxi
<lapapat> megi: (btw your ths patches for h5 are working flawlessly, thank you, with the NPN2 I've found that it's best to specify just 3 opp points 408MHz@1.1v, 816MHz@1.1v and 1296MHz@1.3v. this allows for effective cooling by switching down from 1.3v but still maintains good performance. Frequencies lower than 408 don't really provide much power benefit j
<lapapat> ust collapse performance)
<KotCzarny> is 1.3ghz stable at just 1.3v?
<KotCzarny> you might run some burn in test
<megi> lapapat: yeah, I suspect you're right. I'll probably drop those lowest OPPs
<lapapat> it looks stable, i was loading all cores with multiple dd tasks from urandom to dev/null
<lapapat> the power findings on this board are somewhat crazy. idles on mainline 4.19 with ths and dvfs enabled conservative governor at 1.1w
<lapapat> dropping from gig-e to fast-e saves 0.4 watts
<KotCzarny> and that's why folks we need switchable voltages on boards
<KotCzarny> :/
<lapapat> the rtl8211e chip is the hottest thing on the board
<megi> lapapat: I'm not sure about reducing opps, the power draw reduces lineraly with frequency, and it's often enough to step down from 1.3 to 1.1 in realistic loads
<KotCzarny> megi: power draw reduces with voltage, not freq
<megi> sure, I mean OPP wise
<KotCzarny> on boards with proper vreg it makes sense
<megi> yes
<KotCzarny> on bi-voltage ones, not really
<lapapat> exactly! if we could get down to 0.8v for 120MHz it would make a lot of sense :)
<megi> lapapat: that's quite a bit :D
<KotCzarny> well
<KotCzarny> on opipc there is minimal stable voltage
<KotCzarny> under which you cant go lower
<KotCzarny> and it's around 408mhz i think
<megi> I have the lowest one at 1.04V/648MHz
<lapapat> i've a very decent usb power meter here and it seems to support what you're saying KotCzarny it appears best to have just the fastest opp point for each of your opp volts
<z3ntu> codekipper: you there?
<KotCzarny> anyway Hxx series isnt about powersaving
<lapapat> megi: on this exact point i wanted to ask you about that rtl8211e chip, have you found any way of enabling EEE on it? could result in a substantial power saving
<lapapat> alsonheat reduction :D
<KotCzarny> lapapat: that might be a question for montjoie
<megi> ethtool?
<KotCzarny> register hacking?
<lapapat> ethtool says the feature isn't supported
<megi> hmm
<lapapat> when the datasheet says it is
<lapapat> all the mailing list chat suggests that EEE is enabled by default on rtl8211 but maybe not in this weird state when it's just being used as a phy. i've dug myself into the ground now :)
<KotCzarny> lapapat: maybe its enabled but just not reported?
<KotCzarny> do you have a way to check ?
<megi> can you read phy registers via ethtool?
<lapapat> oooh, i don't know - i'll give that a shot. have you been having to do that with your h6 hacking?
<megi> no
<lapapat> okay, well this isn't a huge priority for me rn, but if i have time to dig i'll report back. it might really help reduce the heat load on these little gig-e boards, especially when that tiny phy chip can't be easily heatsinked
<megi> you may try adding some printks to phy_init_eee in phy.c, to see why it ends up being reported as not supported, exactly
<megi> or maybe to phy_ethtool_get_eee
<lapapat> good hints, cheers boss!
Perlovka_ has quit [Ping timeout: 246 seconds]
<megi> KotCzarny: countrepoint to your voltage theory, look at A83T dtsi, it has bunch of OPPs for all kins of frequencies, and they are all on the same voltage level
<KotCzarny> a83t might be different beast, but i think point still stands, until someone measures otherwise
<megi> and I did detailed autoanmted measurements on that, and the power draw scales linearly with the freqnecy
<megi> same voltage
<KotCzarny> on h3 you barely notice the difference
<KotCzarny> it might be that Axx is better power optimized than Hxx
<megi> it may be hidden behind the noise, the diff is 450mW at most for 4x100% cpu loads
<megi> that's 100mA diff top/bottom OPP
sunshavi has quit [Read error: Connection reset by peer]
<megi> the effect may be less than the voltage effect, but it's there
<KotCzarny> would be interesting to check the same on H3/5/6
<megi> would need automated access to the current current draw
<megi> otherwise it's too laborious
<KotCzarny> yup
<KotCzarny> and automated testing suite
<megi> AXP818 luckily has battery current indicator :)
<megi> that I have
<KotCzarny> you might test A20 then
<megi> no HW
<KotCzarny> A20 is nice
<KotCzarny> and amazingly expensive
<megi> anyway, I think there's something weird about A83T OPPs, I'm not sure why they are all the same voltage
<megi> doesn't make much sense
<KotCzarny> internal regulator?
<megi> I actually didn't notice that before
pgreco_ has joined #linux-sunxi
<megi> regualtor has fine settings
<megi> maybe 10/20mV steps
<KotCzarny> i mean hidden in soc and on full auto
<megi> no
pgreco has quit [Ping timeout: 245 seconds]
<megi> it's controlled by the kernel
<megi> and external
<megi> maybe some experimentation and tunning is in order :D
gaston_ has quit [Remote host closed the connection]
<KotCzarny> btw. A10/A20 tablets are available dirt cheap on classifieds boards here
<KotCzarny> hard part is finding out it's a10/a20
<megi> interesting, what brands?
<megi> ah
<KotCzarny> local rebrands, but often the generics
gaston_ has joined #linux-sunxi
<KotCzarny> usually in 5-20usd range
<KotCzarny> which is cheaper than ordering touchscreen from china ;)
<fALSO> megi, i also had a usse
<fALSO> issue on the ONE PLUS
<fALSO> sometimes it didnt boot
<fALSO> "random"
<megi> yeah, me too
<fALSO> i had to put on the cable on the barrel two or three times
<megi> it sometimes detects 4GB
<fALSO> until it booted
<megi> and fails in u-boot
<fALSO> hum mine is just 1gb
<fALSO> never noticed that
<fALSO> (i just remembered it now, and thout that i shoudl share :))
<megi> does it fail early in u-boot?
<KotCzarny> megi, if you want i can send you my a10 with broken touch
<megi> KotCzarny: thanks, but I already have too much HW :)
<KotCzarny> same :/
<KotCzarny> trying to clean the closet
<megi> and trying to make some mental space for PinePhone :)
<megi> haha
<KotCzarny> pinephone would rock if it had physical kb
<KotCzarny> :>
pgreco_ is now known as pgreco
<megi> just plug it in
<KotCzarny> full one, not just quirky only-letters
<KotCzarny> nah, it should be mobile and convenient
<KotCzarny> had foldable bt kb, while it worked, not very convenient
<megi> I meant AT 102-key mechnaical keyboard :)
<KotCzarny> :)
<megi> but anyway, that's not better either
Perlovka has joined #linux-sunxi
<megi> hmm, A83T's lowest voltage OPP is 1.2GHz and highest is 2.02GHz :)
<megi> that's where the voltage needs to rise
<megi> and mainline OPPs go only to 1.2GHz
<megi> so there's no need for dynamic voltage regulation at all below that
<KotCzarny> maybe it just doesnt go under the lowest set voltage?
<megi> that's from the fex file
<megi> maybe datasheet has other lower limit for the voltage
<megi> 840mV is probably just the lowest workable voltage
<megi> fALSO: maybe reboot your board a few times with attached serial cable, to see if you can catch the failure with more details
<fALSO> megi, i'm gonna connect it today
<fALSO> megi, ill try to report later in the day
<megi> I suspect it will be failing early in u-boot
<megi> interestingly, SW reboot never failed me
<megi> it's only on cold power on via HW switch
<megi> that it sometimes fails
<megi> and it looks like some DRAM related issue
<fALSO> i bought it because it was the cheapest h6
<megi> maybe it's probed too early, and just adding some delay to u-boot dram probe code will fix it
selfbg has quit [Remote host closed the connection]
<megi> too early after powerup
<fALSO> i wish i knew how to do those changes
<fALSO> but theres no easy way to "learn"
<fALSO> =)
<megi> look for sunxi_dram_init in dram_sun50i_h6.c in u-boot
<megi> and add msleep(200) at the beginning of that function
<megi> or maybe udelay(200*1000), I don't remember what the delay functions are in uboot
<fALSO> megi, i was looking at the default configs on the kernel
<fALSO> theres only one
<megi> it's for all the 64bit boards
<fALSO> this one "should" boot on a H6?
<megi> that I have: Opi PC 2 and Opi 3
<fALSO> gonna try it
<megi> it should boot Opi One Plus too
<fALSO> the config that i have has a ton of stuff enabled, takes a long time to build
<KotCzarny> defconfig just includes drivers in kernel
<fALSO> stuff that isnt required
<megi> yeah, this one is more minimal
<KotCzarny> it's the dts/dtb that enables them on the board
<megi> KotCzarny: yeah, but if the driver is missing, dts will not do anything
<KotCzarny> one kernel to boot them all, but the devil is in the details (DT)
<fALSO> =)
<megi> fALSO: I think my config builds in 3-4mins on my old ivy bridge
<fALSO> man
<fALSO> i have an core2quad
<fALSO> with gentoo
<megi> so make it 2x :)
<KotCzarny> buy old core i3-4130, it's nice
<fALSO> yes, i should buy one, but i dont know anything about cpus and stuff
<KotCzarny> cheap, cool and speedy enough
<fALSO> how many cores does that have kotczarny ?
<KotCzarny> 2
<KotCzarny> but they go up to 3.4ghz
<fALSO> nice
<fALSO> there are a lot of dells and stuff
<fALSO> probably buy one just for my compilations :D
<megi> KotCzarny: that sounds like mb and dram update, too going from core2
<KotCzarny> megi, yeah, just mentioning quite nice cpu price/speed wise
<megi> right
JohnDoe_71Rus has joined #linux-sunxi
<KotCzarny> and it's really cool (cold)
<KotCzarny> never hit more than 60C even under heavy load
<montjoie> for compile stuff I steal the 48 cores of my work, waiting for build is so 2018
megi has quit [Ping timeout: 245 seconds]
<fALSO> =)
<montjoie> but last time I tried distcc with enough boards, it was fast
<fALSO> lol
<fALSO> once i installled gentoo on my orange pi pc2
<fALSO> and used distcc on my core2quad
<fALSO> with cross compiler to help it
<montjoie> all my boards have gentoo, so same gcc
<fALSO> but....
<fALSO> even with that
<fALSO> there were big packages that were impossible to LINK
<fALSO> the ram wasnt enough
<montjoie> swap over nfs
<montjoie> SUCH SPEED
<fALSO> probably faster than the sd card
<fALSO> :-P
lapapat has quit [Remote host closed the connection]
sunshavi has joined #linux-sunxi
jorik_ has quit [Read error: Connection reset by peer]
<diego71> montjoie: maybe with a gigabit and a fast server, it will not be too awful...
<diego71> ofc, if you had a fast server, you probably want to use it to crosscompile
jonkerj has joined #linux-sunxi
<montjoie> diego71: even just the bpim3 (8core+gigabit) (along my desktop) made my kernel build faster
<solderfumes> NFS won't work, the filesystem can't support swap. Makes me wonder if iSCSI can be used instead
<gnarface> isn't there swap over nfs though too? i thought there was
<montjoie> solderfumes: you need an option in the kernel to use it
<montjoie> but you could swap over NFS
<solderfumes> montjoie: I'll be damned, CONFIG_NFS_SWAP
<gnarface> it is new :)
<gnarface> ... sorta
<solderfumes> noone bothered to ask why, so they just merged it, I guess
reinforce has quit [Quit: Leaving.]
<montjoie> but in my memory you need to create a big file on the NFS ans swapin
<montjoie> no real swap partition exported
<mru> better trust the server admin not to mess with the data
<solderfumes> Need more memory? Sign up for Honest 🎸solderfumes's SwapfileCloud️®️ service and get a discount over 16G Cloud Connected Swap files!
<solderfumes> give SAAS a new meaning
<diego71> you can also use a nbd (network block device) as swap
<mru> and iSCSI, I suppose
<diego71> solderfumes: some server don't have local disk, so it make sense have also the swap on remote disk
megi has joined #linux-sunxi
<Mangy_Dog> is there an easy way to set the audio out to go through the analogue audio pins of the 13 pin header rather than hdmi out? currently getting no audio
<Mangy_Dog> is pulse audio the problem?
<solderfumes> Mangy_Dog:
<solderfumes> what are you using for playback?
<mru> pulseaudio is always _a_ problem
<solderfumes> pulseaudio could be a problem, but also a solution
<mru> you're better off without it here
<megi> Mangy_Dog: use pactl to set default sink
<solderfumes> i don't want to go into pros and cons of pulse. It works if it works, and it usually just does.
<mru> until it doesn't
reinforce has joined #linux-sunxi
tllim has joined #linux-sunxi
<Mangy_Dog> sorry was afk so currently aplay-l only lists allwinnerhdmi
<Mangy_Dog> for audio devices
<Mangy_Dog> so i uninstall pulse and install pactl?
<megi> pactl is a basic pulseaudio control tool
<megi> installed along with pulseaudio
netlynx has joined #linux-sunxi
<Mangy_Dog> is that the toolbar icon thing?
<megi> you can use pavucontrol if you like GUI tools
<megi> pactl is CLI tool
<megi> it probably needs to be installed separately
<Mangy_Dog> port says analog output
<Mangy_Dog> still no audio
<megi> all this assumes that alternative output actually works on your board
<Mangy_Dog> well like i say aplay shows only allwinnerhdmi
<megi> so nothing in userspace will help
return0e has quit [Read error: Connection reset by peer]
return0e has joined #linux-sunxi
<Mangy_Dog> if i uninstall pulseua=audio what do i replace it with
<mru> nothing
<mru> for gaming you want to talk directly to the alsa hw driver
<mru> minimise latency
<Mangy_Dog> allwinnerhdmi is still the only audio being listed in aplay-l
<mru> maybe you're missing a driver
<Mangy_Dog> thought as much tbh
<Mangy_Dog> but i would have thought what ever audio hardware thats passing trhough to the hdmi would still be hardware in the analog out
<Mangy_Dog> any suggestions?
<Mangy_Dog> is there a sudo apt-get install thisaudiodriver?
msimpson has joined #linux-sunxi
yann has quit [Ping timeout: 245 seconds]
<Mangy_Dog> anything?
<megi> what board is this?
<Mangy_Dog> orange pi zero plus 2 h3
<megi> it doesn't have codec enabled in dts
<megi> you'll need something like this
<fALSO> =)
<Mangy_Dog> how do i do that?
<Mangy_Dog> and where?
<megi> sun8i-h3-orangepi-zero-plus2.dts
<Mangy_Dog> btw since removing pulse i now get no audio devices listed in mixer
<megi> that's not surprising :)
<Mangy_Dog> i mean in alsa mixer
<fALSO> lol pulse
<Mangy_Dog> so what am i meant to do? btw reinstalled pulse
<megi> patch sun8i-h3-orangepi-zero-plus2.dts
<Mangy_Dog> with the line you showed?
<megi> add those lines, or copy them from sun8i-h3-orangepi-pc.dts
<Mangy_Dog> if so where is the file?
<megi> linux kernel source code
<Mangy_Dog> ...
<Mangy_Dog> im not going to build an entire linux kernal
<Mangy_Dog> you kidding
<megi> so decompile dtb add the lines and recompile it
<Mangy_Dog> are you really trying to tell me there is no way to get analoge out of thie board without patching the kernal and recompiling the thing?
<Mangy_Dog> you say its like a flippant easy thing to do
<megi> you're on the sunxi kernel development channel
<Mangy_Dog> im not a linux developer i came here for help with these boards
<Mangy_Dog> is it really too much to ask to have things work as expected?
<megi> yes :)
<Mangy_Dog> without having to reinvent teh wheel liturally
<megi> the simplest thing you can do is the suggested decompilation of dtb file
<megi> you'd avoid kernel rebuild that way
<Mangy_Dog> thats not a simple thing to do
<DuClare> Assuming you have the dtb, it's just one command to decompile it
<Mangy_Dog> i dont know do i? im running latest armbian image, (one kernal away from the latest)
<DuClare> I assume it has dtbs under /boot
<DuClare> But I don't use armbian, what do I know
<megi> here's how to install device tree compiler that you'll need
<Mangy_Dog> sun8i-h3-orangepi-zeroplus2.dts found as .dtb
<megi> yes, that's the one
<megi> just make backup of the original one, in case you make your board unbootable :)
<megi> if you decompile you'll have to do things differently, becase codec node already exists somewhere in the decompiled output with status disabled
<Mangy_Dog> trying now, using a denovo mini keyboard its hard to type :p
<Mangy_Dog> hmm
<Mangy_Dog> saying no such file or directory
<Mangy_Dog> but ive tripple checked the spelling
<Mangy_Dog> also just remembered i could ssh into this rather than teminal with a shitty bluetooth keyboard
<Mangy_Dog> found spelling error D:
<Mangy_Dog> ok done
<Mangy_Dog> but i got a bunch of node has unit name, but no reg property
<Mangy_Dog> warnings
<megi> those are probably fine
<Mangy_Dog> had to install gedit :p
<Mangy_Dog> right in the file now
<Mangy_Dog> so do i just add the {} line you posted
<Mangy_Dog> because theres nothing like that line in the file
<Mangy_Dog> i have a sound headding that lists simple audio card and in that class allwinner,hdmi
<Mangy_Dog> bellow that is another codec class @ 1c22c00
<megi> the one below is status disabled?
<Mangy_Dog> yes
<megi> make it enabled
<Mangy_Dog> so make codec@1c22c00 enabled?
<megi> you can just delete the status="disabled" line
<Mangy_Dog> enabled or "okay"?
<megi> and add the allwinner,audio-routing lines
<megi> okay, or delete the line
<Mangy_Dog> add the allwinner,audio-routing lines??
<megi> nodes are enabled unless disabled explicitly
<Mangy_Dog> ok
<Mangy_Dog> add it just bellow or at the end of the file?
<megi> inside {} of the node you've just enabled
<Mangy_Dog> ok
<megi> just below the status line, for example
<megi> and that's it
<Mangy_Dog> just bellow is phandle=<0x58>;
<megi> it just has to be inside {}
<megi> putting it just in front of } is the safest choice
matthias_bgg has quit [Ping timeout: 272 seconds]
<Mangy_Dog> ok file edited
<Mangy_Dog> i guess do the compile it back part of the guide
Gerwin_J has quit [Quit: Gerwin_J]
<Mangy_Dog> syntax error
vagrantc has joined #linux-sunxi
<KotCzarny> swap over nfs is OOOOOLD
<KotCzarny> i remember having it at uni, some 15 years ago
<fALSO> lol
<fALSO> nice
<fALSO> i have to try it
<KotCzarny> dont.
<fALSO> is it worse than the SD card?
<KotCzarny> its only if you have cheap terminals
<Mangy_Dog> ive copied that code as you image pasted it
warpme_ has quit [Quit: warpme_]
<Mangy_Dog> still getting syntax error
<vagrantc> swap over NBD works
<KotCzarny> i dont think nbd was there 15 years ago
* vagrantc doubts
<Mangy_Dog> megi are you sure the &codec is meant to be inside the same penrentese ?
<Mangy_Dog> perentithes
<Mangy_Dog> what ever its called
<KotCzarny> mangy_dog: idea for you:
<Mangy_Dog> eeesh
<Mangy_Dog> though is that running nes emulator on that tiny screen? :p
<Mangy_Dog> cute
<KotCzarny> :)
<KotCzarny> that's the whole point
<Mangy_Dog> heh
<Mangy_Dog> Error: /tmp/temp.dts:838.4-10 syntax error
<Mangy_Dog> FATAL ERROR: Unable to parse input tree
<Mangy_Dog> line 838
<Mangy_Dog> is &codec {
<Mangy_Dog> no idea what the 4-10 is
<KotCzarny> character at that line?
<fALSO> check other DTS
<fALSO> that have that part
<fALSO> and see how it is
<Mangy_Dog> assuming they have it? considdering this one didnt
<fALSO> which board is it
<Mangy_Dog> orange pi zero plus 2 h3
<fALSO> ok ill try to help
<fALSO> one sec
<Mangy_Dog> read up and see what megi has been posting
<Mangy_Dog> as thats what i was following before they seemed to go afk
<fALSO> this is yours, right?
<fALSO> this is what you want to add
<fALSO> check if you have it correctly typed
<fALSO> and on the "correct" place
<fALSO> probably add it before the &de... stuff
<fALSO> hum
<fALSO> why are you adding it there?
<Mangy_Dog> megi told me to?
<fALSO> ahh OK
<fALSO> it doesnt seem right
<fALSO> check as the other one that i pasted is
<megi> iu did not
<fALSO> you probably just need to copy those lines &codec...
<fALSO> and paste it before the &de...
<fALSO> at least that would be what i would try :-P
<megi> he doesn't have dts from the kernel
<Mangy_Dog> <Mangy_Dog> add it just bellow or at the end of the file?
<Mangy_Dog> <megi> inside {} of the node you've just enabled
<Mangy_Dog> <Mangy_Dog> ok
<megi> just the decompiled one
<Mangy_Dog> you did
<megi> you must not add &codec ...
<megi> just those exact changes I described
<fALSO> AHH he is not using mainline
<fALSO> ok..
* fALSO shuts up
<Mangy_Dog> i am
<Mangy_Dog> it is mainline
<Mangy_Dog> its one version behind latest kernal
<Mangy_Dog> latest armbian build
<fALSO> better to listen to megi , he knows better than me
<megi> just remove status=... and add allwinner,... as I wrote before and recompile dtb
<megi> nothing else
<Mangy_Dog> so remove the &codec {} but keep its contense?
<Mangy_Dog> putting it into the other codec@ function/class
<megi> there should ne no &codec in decompiled dts
<Mangy_Dog> ok it was in yur image post of the code you said to copy
<megi> that was before we switched from "recdompiling kernel" to "recopmpiling dtb"
<fALSO> :-)
<fALSO> i didnt even knew that was possible
<Mangy_Dog> ok think that compiled
<Mangy_Dog> rebooting now
<fALSO> lets cross our fingers
<Mangy_Dog> well
<Mangy_Dog> its working
<Mangy_Dog> sort of
<fALSO> =)
<Mangy_Dog> its just insainly quite
<Mangy_Dog> and i dont know if thats my audio amp missbehaving or the system :p
<Mangy_Dog> the reason i wanted to get audio working was to test the audio on the board
<fALSO> sorry for the dumb question, but did you try to change the volume on the mixer?
<Mangy_Dog> its at 199%
<Mangy_Dog> 100
<fALSO> ok
<fALSO> i had to ask :-D
<Mangy_Dog> well my headphone switching is working too
<Mangy_Dog> but just very very quite
<Mangy_Dog> im not convinced though its my amp
<Mangy_Dog> because i was getting system noise on the speaker before
<fALSO> i never tried the headphones jack on mine
<fALSO> ill try mangy_dog
<fALSO> its also an H3
<fALSO> one sec
<megi> aren't those 4 pin jacks?
<megi> incl. with tv signal
<Mangy_Dog> mines setup just for audiop
<fALSO> mangy_dog,
<megi> I used audio out, but I added opamp on the output as a buffer
<fALSO> mangy_dog,
<fALSO> i had to change LINE OUT value
<fALSO> and also DAC
<fALSO> and if i mute unmute that DAC reverse that also *changes* the sound
<fALSO> but with those at 100 its quite high the volume
<fALSO> on big headphones
<fALSO> press f5 on alsamixer
<fALSO> to show all the mixers
<Mangy_Dog> im not sure its that getting some strange readings on my sillyscope
<fALSO> try to mute the MIC1
<Mangy_Dog> im getting a modulating square wave off my speaker line
<fALSO> :(
<fALSO> dont know what else you could try
<Mangy_Dog> with the duty cycle changing with the audio
<fALSO> i also found out that i can only use mpg123 as root
<Mangy_Dog> but ill take a look at the mixer
<fALSO> probably the user doesnt have write acess to the device
<fALSO> have to check that out ;-)
<Mangy_Dog> wow
<Mangy_Dog> something very wrong
<Mangy_Dog> i wasnt sure at first
<Mangy_Dog> but i got a little shock off the usb plug i have soldered onto the board
<Mangy_Dog> just tried to plug in the hub i have and power died
<fALSO> lol
sunshavi has quit [Read error: Connection reset by peer]
<fALSO> damn
<fALSO> lets hope you didnt fry anything :-X
<Mangy_Dog> restarted right away
<Mangy_Dog> right ive had enough for now
<Mangy_Dog> going to do supper
<Mangy_Dog> might look into this again later this evening prolly leave it to tomorrow :p
<Mangy_Dog> at least the audio out is working
<Mangy_Dog> sort of
<Mangy_Dog> its progress
<fALSO> awesome
<KotCzarny> audio mixer toggles are quirky
<KotCzarny> there is magical set of them to make the audio routed right
<KotCzarny> (and probably easier to understand with schematics in other hand)
<fALSO> kotczarny, nice
<fALSO> i love when things are magical :-D
<KotCzarny> difference between magic and tech is documentation
<mru> any sufficiently advanced technology is indistinguishable from magic
<KotCzarny> unless you study it
<mru> then it is no longer sufficiently advanced
<KotCzarny> and start tearing it apart, then putting blocks back
sunshavi has joined #linux-sunxi
warpme_ has joined #linux-sunxi
BenG83 has quit [Quit: Leaving]
vagrantc has quit [Quit: leaving]
Putti has quit [Quit: Leaving]
Putti has joined #linux-sunxi
<megi> Mangy_Dog: if you use the wrong jack and short aduio_l to cvbs out, there's 50 Ohm resistor load on cvbs out, and that's additional load to the headphones, and will halve the power output
<megi> if you have 50Ohm headphones that is
JohnDoe_71Rus has quit [Quit: KVIrc 5.0.0 Aria]
<megi> but i guess using 3 intead of 4 wire jack, will probably short the closest pin to ground
netlynx has quit [Quit: Ex-Chat]
<megi> it would make sense, because that would groud the mic input when using headphones without the mic, and connecting them to the Orange Pi. Pretty weird that Xunlong decided to put CVBS *output* on the mic pin
<megi> because putting normal headphones there would short CVBS out to ground
<megi> eh, kinky HW design
<mru> using a trs plug in a trrs jack shorts the pin next to ground
<mru> which might be mic in or video out
<megi> CVBS is there
<megi> yup
<mru> shorting it to ground should be harmless
<mru> otherwise the design is flawed
<mru> since someone will inevitably use such a plug
<megi> there's no saying whether it is harmless actually, it's connected over some inductor dirtectly to the SoC
<megi> it's probably harmless unless CVBS is enabled, normally the pin may be in hi-Z state
<mru> even when enabled, I doubt there's enough drive strength there to damage anything
<megi> it may just consume more power
<mru> that's a given
tnovotny has quit [Ping timeout: 246 seconds]
<Mangy_Dog> i dont think my audio is shoting to ground...
<Mangy_Dog> but tomorrow im going to try one of the modules i have of the amp im using
<Mangy_Dog> i cloned the circuit in my design
<Mangy_Dog> that will rule out the pi or my audio circuit
<Mangy_Dog> actually looking over my circuit design again
<Mangy_Dog> im more convinced its the pi letting out a weak signal... as it was quiet on my headphones, which is a direct line to the audio pins
<Mangy_Dog> im using a pj 311d socket its internally switching
<fALSO> ok
<fALSO> just powered on my orange pi one plus ;-)
<fALSO> still works
<fALSO> and booted at the first try
<fALSO> gonna try to upgrade the kernel
<Mangy_Dog> just went into the mixer and turned up the dac
<Mangy_Dog> seems i have volume now :)
<Mangy_Dog> though its a little distorted
<Mangy_Dog> i dont know if its audio amo to speaker distortion or dac processing distortion
<fALSO> mangy_dog, try muting/unmuting the "DAC reverse"
<fALSO> mangy_dog, that somehow "changes" the sound
msimpson has quit [Ping timeout: 245 seconds]
<Mangy_Dog> oh didnt try that
<Mangy_Dog> however
<fALSO> trying to boot my one plus now
<Mangy_Dog> i did turn the dac and the other line out up into the white which did bring up the volume
<Mangy_Dog> however it distors
<Mangy_Dog> distorts
<Mangy_Dog> even in my headphones
<fALSO> Linux orangepioneplus 5.2.0-10808-g9637d517347e-dirty #14 SMP PREEMPT Tue Jul 16 17:56:09 WEST 2019 aarch64 GNU/Linux
<fALSO> :-D
<Mangy_Dog> and its no even that loud in my headphones so its not the diaphrams distorting its something in the dac processing
<Mangy_Dog> and when at middle volume the sound quality is....
<Mangy_Dog> extreamly poor
<Mangy_Dog> very very mushy
<fALSO> :(
warpme_ has quit [Quit: warpme_]
<mru> I wouldn't expect it to be great
yann has joined #linux-sunxi
<Mangy_Dog> i would have expected it to be better than what i was getting
<Mangy_Dog> like mp3 player from 1999 or something
<Mangy_Dog> that level
<Mangy_Dog> not super hifi but passable
<mru> connect the output to your shiny scope and play a 1 kHz sine wave
<mru> see if it looks distorted
<Mangy_Dog> nods will do tomorrow
<Mangy_Dog> but my suspision is something is still not set right in the audio hardware
<mru> quite possible
<mru> sometimes looking at the waveform can give a clue as to what is wrong
<Mangy_Dog> to me it was like turning up in the asla mixer its like theres still a disconnect between the audio being processed and the output volue
<Mangy_Dog> so the DAC was distortion before the amp and speakers would
<Mangy_Dog> nods
<Mangy_Dog> but tomorrow its now late evening and i want to play games
<Mangy_Dog> :p
<mru> don't you have to finish the console before you can play?
<Mangy_Dog> but at this stage im 90% confidant my sound amp setup on my board is fine
<Mangy_Dog> :D
<Mangy_Dog> i have a pc too :p
<mru> I currently have a logic analyser hooked up to the usb and i2c lines on a commercial dac
<Mangy_Dog> I must admit though, no offence to you guys as you have been really helpful to me thank you all... But I am starting to feel... Maybe i should have just got a raspberry pi compute module :p
<mru> wouldn't you be having the same kind of problems with that?
<mru> like flickering backlights
<Mangy_Dog> all the retropi builds on raspberry is pretty solid and more prooven
<Mangy_Dog> oh the flickering backlight yeah
<Mangy_Dog> thats sorted now
<Mangy_Dog> design change
<mru> oh?
<Mangy_Dog> using a simpler more fool proof circuit
<Mangy_Dog> one also recommended by the display manfacturer
<Mangy_Dog> its like 4 componants... so i cant fuck it up
<Mangy_Dog> :p
<mru> for every foolproof design there is a new and improved fool
<Mangy_Dog> indeed
<Mangy_Dog> im fool 328.4
<mru> what does that circuit look like?
<Mangy_Dog> theres going to be a rev2 board now... Im slowly editing the pcb to acomodate the tweaks... thats why i wanted to test audio to be sure it works
<mru> yeah, test everything you can on each rev
<Mangy_Dog> nods
<Mangy_Dog> coincidentially it uses the same resistor value to set the mA rating
<mru> isn't that essentially the same?
<mru> a boost regulator
<mru> hmm, there's no capacitor on the output
<mru> that doesn't seem right
<Mangy_Dog> no need to according to the example design
* mru is sceptical
<mru> the "typical application" in the pt4110 datasheet has a cap
<Mangy_Dog> dso it does
<Mangy_Dog> was just finding the datasheet to double chec k:D
<Mangy_Dog> thanks >.<
<Mangy_Dog> ahhh
<Mangy_Dog> page 6
<Mangy_Dog> was looking there
<Mangy_Dog> 2.2uf though
<mru> read page 5, "selecting the output capacitor"
<Mangy_Dog> yeah
<Mangy_Dog> though 2.2uf seems like a low value
<mru> the adafruit circuit had a 1u cap
<mru> at least put a footprint on the pcb
<Mangy_Dog> thought it was 10
reinforce has quit [Quit: Leaving.]
<mru> 10u on the input side
<Mangy_Dog> anything you think i can do to reducce system noise?
<Mangy_Dog> yep 10 on the in
<mru> the pt4110 has a 1.2 MHz switching frequency
<mru> with a 40 mA current, a 2.2u cap will give a ripple of 15 mV or so
<mru> perfectly acceptable
<Mangy_Dog> was worried you were going to tell me that that it was terrible need to redo it all
<mru> we're talking about 20 V or so here
<mru> the LEDs don't care
<Mangy_Dog> nods
<fALSO> man
<fALSO> i think theres probablu something wrong
<fALSO> with the red led on the orange pi one plus
<fALSO> lol
<Mangy_Dog> over volt bright?
<Mangy_Dog> or just a bright
<mru> some LEDs are crazy bright
<fALSO> bright that a guy with glasses cant look at the board
<Mangy_Dog> the ones on the this pi zero are rather dim
<mru> "do not stare into laser with remaining eye"
<fALSO> makes a nice LENS FLARE on the glasses
<fALSO> ;-P
<mru> get anti-reflective glasses
<mru> I don't understand how anyone can stand glasses without that coating
<fALSO> ill use anything.....
<fALSO> :-P
<fALSO> just need to be able to see
<mru> I wear glasses
<fALSO> me too, since 18 years old
<fALSO> :/
<mru> and I gladly pay a little extra for anti-reflective coating
<Mangy_Dog> same
<fALSO> lol, i got my glasses from CHINA
<Mangy_Dog> specsavers
<fALSO> went to a store here in portugal
<fALSO> made the *tests*
<mru> vision express, not that it matters
<fALSO> got the paper with the values, and ordered online
<fALSO> the difference was like 100€
<Mangy_Dog> tbh i want to try vision express but im poor
<mru> they were the nearest opticians when I moved here
<fALSO> for 700€ glasses in portugal
<Mangy_Dog> but specsavers i think slightly screwed something up
<fALSO> glasses in portugal are very expensive
<Mangy_Dog> my glasses give me horrid migrains
<fALSO> bleh
<Mangy_Dog> holycow
<fALSO> i even sleep with them
<Mangy_Dog> 700euros
<fALSO> ;-D
<fALSO> mangy_dog, ya man
<Mangy_Dog> thats insane
<fALSO> they come with ESSILOR GLASS
<fALSO> etc
<fALSO> they bullshit you a lot
<fALSO> anti reflex -> 100€
<fALSO> computer "protection" -> 100€
<Mangy_Dog> specsafers arent subsedised nor are vision express... only time any sobsadese come in are if you get tests and sometimes can get nhs help for some things
<mru> the whole glasses business is a racket
<Mangy_Dog> but most of the time you pay the full price
<fALSO> i wish that some of those shops opened in portugal
<fALSO> it would ruin the business for the estabbilsiedh ones
<fALSO> estabilished
<fALSO> the "expensive" ones
<fALSO> U-Boot SPL 2019.07-00316-g6674dc77c2-dirty (Jul 16 2019 - 15:37:54 +0100)
<fALSO> DRAM: 1024 MiB
<fALSO> Trying to boot from MMC1
<mru> almost all the brands are owned by luxottica
<Mangy_Dog> my sunglasses paire are fcuk
<Mangy_Dog> osiris is my specs
<fALSO> i like those big glasses
<fALSO> 70's pedophile style
<fALSO> the ones with the bridge on top
<Mangy_Dog> D:
<fALSO> hipster-falso
SopaXorzTaker has quit [Remote host closed the connection]
<Mangy_Dog> oh very dear
<Mangy_Dog> !
<fALSO> latest u-boot doesnt seem to work on the orange pi one plus
<karlp> did you clean first?
<megi> fALSO: yup, not even a beep
<karlp> libv failed without a clean first too.
<fALSO> megi, here just shows u-boot SPL bla bla
<fALSO> i pastes it earlier
<fALSO> hangs on: Trying to boot from MMC1
<megi> interesting
<fALSO> i should have made a backup of the working uboot spl :-X
<megi> I didn't even try to debug, becase no output is not really welcomming
<fALSO> lol, i have output via serial, but no boot
<libv> in my case, nothing showed up
<fALSO> lets wait for a few days, and try again :-P
<fALSO> or maybe try megous uboot fork?
<libv> spl text base address needed to be set
<megi> my doesn't work
<libv> CONFIG_SPL_TEXT_BASE=0x20060 is set for orangepi_one_plus_defconfig
<megi> is that wrong?
<libv> i don't think so
<megi> seems like this is set to various values on various sunxi boards
<libv> my issue was that i blindly reused the old hacked lime2 defconfig for the fosdem daughterboard, and expected it to work
<fALSO> well i noticed that the arm trusted firmware that i'm using
<megi> ah
<fALSO> also had some changes since the last time i bcompiled
<fALSO> dont know if that would make it not boot
<libv> so i needed to adopt the lime2 changes for it, and then all was a-ok
<libv> and the uart showed absolutely nothing
<libv> so this is not the issue that fALSO is having
<fALSO> nop. ours is difference
<fALSO> nop. ours is differente
<fALSO> im trying on a "Orange Pi One Plus" H6 cpu
<fALSO> had an older uboot version that worked... but i always want the latest and greatest :-PPPPPPPPPPPppppppppppppp
<megi> for u-boot, this mindset is a bit less useful than for Linux, unless you really need something that's in newer u-boot
<fALSO> megi, i upgraded because
<fALSO> i had "that problem"
<fALSO> that the board just booted randomly
<fALSO> i had to pull and push the power cable a few times
<fALSO> until it booted
<fALSO> i tought that it probably was a uboot bug, thats why i also upgraded :D
<megi> hmm, right :)
<fALSO> got worse ;-D
<megi> happens :)
<fALSO> no problem. ill give love to my PI PC
<megi> there's dm conversion work that's probably causing regressions
<megi> u-boot moving various drivers to a new driver model
<fALSO> ill try again in a few days
<karlp> or just go back in time,
<fALSO> i was looking if any of the boards based on H6 have any sound support
<karlp> like megi said, you don't reaally need to hhave the latest uboot
<karlp> and certainly for things like fiddling with sound
<fALSO> but then i looked at the board
<fALSO> and it doesnt have an headphone jack
<fALSO> just probably hdmi aduio
<fALSO> ;-D
* fALSO is dumb
<jernej> well, recent U-Boot is needed for H6 DRAM improvements
<jernej> with 2019.04 I don't have any issues anymore
<karlp> well, they already had aworking one, so no need to udpate then :)
<fALSO> U-Boot SPL 2019.07-00316-g6674dc77c2-dirty (Jul 16 2019 - 15:37:54 +0100)
<fALSO> i tried the latest :-P
<fALSO> probably better to go back to 2019.04
<jernej> I'll test 2019.07 soon, but I don't really expect much troubles
<jernej> last few updates were smooth
<megi> jernej: I don't see any dram improvements there since 2019.04
<jernej> they were before that
<jernej> in 2019.10 there will be DDR3 and hopefully half DQ support (for this one I have to send a patch)
<megi> I have a hunch, there's an issue with dram being probed too early.
<megi> it probes unreliably on hard power on, but always probes fine after SW reset
<jernej> anyway, I didn't have any DRAM related issues on H6 boards for a few months now, so it's kind of strange for me that you have it
<megi> every 5-10 power ons it detects 4GB, or doesn't preobe at all
<megi> while I have 2GB
<jernej> I think latest H6 DRAM commit from MoeIcenowy solved that issue for me, but that fix is already few months into U-Boot
msimpson has joined #linux-sunxi
<megi> i suspect it needs a little more time to setttle after board power on / before DRAM probe, I'll try adding some delay
lurchi_ has quit [Quit: Konversation terminated!]
<megi> i have all the patches
lurchi_ has joined #linux-sunxi
<megi> to H6 DRAM code that are in u-boot master
<jernej> idk, I can say that I went over H6 DRAM code hundred times now and I don't think there's anything wrong with it. I'm also not sure why would delay be needed on some boards and not on the others...
<jernej> unless defconfig for your board is overoptimistic and sets too high DRAM frequency
<fALSO> i still have here a patch that maybe jernej or moeicenowy gave me for uboot
<fALSO> for uboot to "lookup" the ethernet mac address i think
lurchi_ has quit [Quit: Konversation terminated!]
lurchi_ has joined #linux-sunxi
<jernej> you mean to generate it? MAC addresses are not stored anywhere
lurchi_ is now known as lurchi__
<fALSO> one second, ill show you, my building gentoo is booting :-)
msimpson has quit [Ping timeout: 258 seconds]
<fALSO> maybe its not needed anymore.
<jernej> fALSO: that's not needed IF you have ethernet alias in both, U-Boot and Linux DTs (if you use separate sources)
<fALSO> im using mainline uboot and mainline linux
<fALSO> going to try 2019.04 to see if it boots again :-)
<jernej> yeah, but DT in U-Boot usually lags behind a bit
<jernej> so I already have the case that U-Boot DT was too old and didn't have emac node with alias, so MAC address was still random
<jernej> s/have/had
<fALSO> ahh yes, that was the problem. now i remember
<fALSO> each time that i booted, it had a different mac address
<fALSO> and got a differnt ip in the router
<jernej> libv: that SPL_TEXT_BASE is now moved out of defconfig again:
<jernej> I guess it missed 2019.07 release
<libv> ah, cool, yeah, it felt seriously out of place
<libv> jernej: thanks for the heads up
<jernej> megi: DDR3 patch series also contains some reliability improvements. Do you have those patches too?
<jernej> I think it was pushed yesterday or today in u-boot-sunxi repo
<megi> no
<karlp> anyone got one of these?
TEKrantz has joined #linux-sunxi
<jernej> karlp: I have Tanix TX6 (without mini), which is very similar. Mini version uses only half of data lines for DRAM (half DQ) for which I plan to send patch to U-Boot ML in following days.
<megi> jernej: tried those reliability patches now, and it still fails occasionally with misdetection of dram size
<jernej> megi: try using lower frequency?
<megi> but why would that matter?
<megi> it never fails on reset only on power on, and u-boot initializes dram in both cases
<jernej> which board is that?
<megi> anyway, I'll try that too, just to see
<megi> Orange Pi 3
gaston_ has quit [Quit: Konversation terminated!]
<megi> 2GB version
<jernej> hm... I have 1 GB version but I didn't noticed any U-Boot issue
<jernej> however, Linux freezes sometimes
<megi> hmm, for me Linux never freezes, but u-boot sometimes fails :)
<jernej> but that will probably improve with temperature driver and frequency scaling
<jernej> it never frezes if GPU is run at low frequency ~250 MHz
<megi> I have both, so you may be right
<megi> hmm, now the misdetection also happened on reboot
NeuroScr has joined #linux-sunxi
<karlp> jernej: thanks. have a very old ?x86 tv box that is just starting to ha ve some issues with newest movie files, and two newer rockchip boxes, but they've both been headaches to bring online usefully. starting to use allwinner a lot more now, and already happier with it, thought an allwinner tv box might be nice too :)
<jernej> I have test version of LibreELEC for that box if you want to check it out
<jernej> already got confirmation that it runs on TX6 mini
<karlp> good to know, thanks
<jernej> but it has dreaded XR819 wifi :)
<karlp> fuck that
<karlp> so over that
<karlp> I'd have it on ethernet, but so over that.
freemangordon has quit [Ping timeout: 248 seconds]
<megi> jernej: maybe I know what's wrong :)
<megi> mctl_mem_matches
<jernej> what wrong with that exactly?
<megi> the issue happens more frequently the more frequently I power cycle, so I think that DRAM still contains values from this test at the fixed addresses that are tested from previous boot leading to misdetection
<megi> not exactly sure how, because I don't understand the code very well, but it may not account for the fact that 0xaa55aa55 may be there without being written there
<megi> maybe clearing the value after test may help
<megi> I'll try to patch it, to see if it changes anything
<megi> it would explain why I did not see issues on reboot before, because I run the board for days between reboots typically
<megi> so mem gets overwritten eventually
<megi> but power cycling quickly, it happens almost on every power cycle
<megi> misdetection for 4GB that is
return0e_ has joined #linux-sunxi
<jernej> I guess I have to test PineH64 with 4 GiB of RAM more often
<megi> how much time would you guess for DRAM to clear after power off?
return0e has quit [Ping timeout: 272 seconds]
<jernej> there's some research on this topic, but I don't remember numbers anymore
<jernej> about a minute, just to be sure?
<megi> I was thinking that if the issue is data retention, waiting after power off, before powering the board back on, would eliminate this issue too
<megi> if the theory is right
<jernej> uh, I just remembered, do you have UART always connected?
<jernej> I don't have it most of the time
<megi> yes
<jernej> it may deliver just enough power to keep some content or cause any other issue
<megi> but it also happens on quick reboots, so probably no
<Mangy_Dog> What i think is currently happening the orange pi, is producing a really lowaudio signal... in the DAC. and when i raise the DAC and line out levels in alsamixer its only boosting that low level signal, so its not really using its full amplatude rage that it should so like amping a really weak signal in analogue but this is happening within the DAC processor
<Mangy_Dog> whish is why it sounds really mushy
<megi> so it's not mctl_mem_matches :)
<Mangy_Dog> i know the soc does, but does the o pi zero plus 2 h3 have a pcm audio pin by any chance?
<jernej> megi: do reboots from U-Boot directly also cause same issue?
<megi> yes
<megi> sometimes
ganbold has quit [Ping timeout: 258 seconds]
ganbold has joined #linux-sunxi
<megi> lowering freq to 640MHz did not help either
<jernej> also doesn't matter how much RAM board has?
<megi> I only have a 2GB version, so I don't know
<jernej> what about other H6 boards? you have any?
<megi> but fALSO reported that he also had unreliable boot, when power cycling on orange pi one+, which has 1GB
<megi> but it's not completely clear if it's this issue
<jernej> ok, I'll do some tests regarding that, but I don't have time before next week
<megi> thanks for help, I'll also try instrumenting dram init code, to see what exactly happens during failure
Mangy_Dog has quit [Ping timeout: 244 seconds]