<lvrp16_>
KotCzarny: all the other blobs just say "failed to initialize EGL display"
<anarsoul>
finally :) I got it working
<anarsoul>
timings are a bit off, but it's fixable
<anarsoul>
and tcon driver doesn't report vblank properly if dsi is used
<anarsoul>
bsp driver uses dsi interrupt for that
<anarsoul>
so I'll need to hook it up
msimpson has joined #linux-sunxi
<mripard>
anarsoul: great :)
<mripard>
(sorry, I just saw your message)
vagrantc has quit [Quit: leaving]
<anarsoul>
mripard: I still have few problems though
<anarsoul>
AVDD regulator supplies several consumers
<anarsoul>
so linux may enable AVDD before DVDD
<anarsoul>
panel doesn't like it, axp overheats and board shuts down
<anarsoul>
(AVDD is DC1SW, it also powers ethernet PHY on pine64)
<anarsoul>
another problem - I need to read tcon_div from DSI driver
<anarsoul>
and I'm not sure how to properly do that
<mripard>
why?
<anarsoul>
mripard: I need it to calculate SUN6I_DSI_BURST_DRQ_REG value
<mripard>
(and I'm not sure there's something in the regulator API at the moment to deal with that)
<mripard>
do you just need the divider?
<anarsoul>
yes
<mripard>
then you can do it pretty easily by using the TCON clock rate and its parent
<anarsoul>
so I need to pass tcon0 clock to dsi driver via dts?
leviathan has joined #linux-sunxi
clemens3 has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
<rellla>
mripard, do you have any clue about the deinterlacer on the various socs? afaics on the newer ones (H3, H5...) it's separated from the display engine/ scaler.
<rellla>
and on the older ones it's part of the scaler?
<MoeIcenowy>
jernej: interestingly... in U-Boot it works but in Linux it doesn't
<MoeIcenowy>
(Pine H64 model B HDMI
popolon has joined #linux-sunxi
<mripard>
anarsoul: you can call clk_get_parent I think?
<mripard>
rellla: any clue about what?
fkluknav has joined #linux-sunxi
<rellla>
mripard: about if it is correct, what is said above.
<MoeIcenowy>
mripard: what's the current status of Request API?
<rellla>
meaning, implementing a deinterlacer for A10/A20 -> use drm, for H3/H5 -> do a separate v4l2 driver?
tom_nov has joined #linux-sunxi
fkluknav has quit [Ping timeout: 244 seconds]
LargePrime has quit [Ping timeout: 264 seconds]
martin__ has joined #linux-sunxi
martinayotte has quit [Ping timeout: 264 seconds]
LargePrime has joined #linux-sunxi
megi has joined #linux-sunxi
<mripard>
rellla: I haven't been able to find any info about a deinterlacer on the H3, where did you find it ?
clemens3 has quit [Read error: Connection reset by peer]
clemens3 has joined #linux-sunxi
kaspter1 has joined #linux-sunxi
kaspter has quit [Ping timeout: 240 seconds]
kaspter1 is now known as kaspter
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
chomwitt has joined #linux-sunxi
gnufan has quit [Ping timeout: 260 seconds]
BenG83 has joined #linux-sunxi
<pgreco>
mripard, I didn't see R40/V40 listed anywhere in the cedrus news
<pgreco>
any hardware difference I'm not seeing, or just wasn't tested?
<pgreco>
MoeIcenowy: I was thinking about using bpi-m2b for a project, but I see that some patches landed for r40 and not for v40 (both kernel and uboot)
<pgreco>
patches related to the common things
<pgreco>
do you have have a device to test until I get mine? if I bring the dts to the same level as r40?
<MoeIcenowy>
pgreco: I have a BPi M2B
<MoeIcenowy>
and you should know that R40 display is just supported yet
<pgreco>
MoeIcenowy, that is one of the thinks I was talking about
<pgreco>
also gmac on uboot landed a few days ago for r40 (and not for v40)
<pgreco>
if that looks ok, I can do a 2 patch series, 1 sync, 2 unify
<pgreco>
btw, that is just a theory, completely untested :)
SP7RT has quit [Read error: Connection reset by peer]
iamfrankenstein has joined #linux-sunxi
anarsoul|2 has joined #linux-sunxi
return0e has joined #linux-sunxi
<MoeIcenowy>
I'm doing
<MoeIcenowy>
no need for a common
BenG83 has joined #linux-sunxi
<MoeIcenowy>
jernej: last linux-next with 12/18 and 13/18 applied cannot display on BPi M2 Berry (with M2 Ultra DTB)
elros has joined #linux-sunxi
victhor has quit [Remote host closed the connection]
<MoeIcenowy>
oh sorry it works
<MoeIcenowy>
just I didn't enable VC...
iamfrankenstein has quit [Ping timeout: 256 seconds]
[LawnmowerMan] has quit [Quit: There are paranoid crossroads and shipwreck alleys.]
[Kidult] has joined #linux-sunxi
SP7RT has joined #linux-sunxi
<pgreco>
MoeIcenowy, nice!!!, I didn't know how to /delete-node/
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Client Quit]
<pgreco>
MoeIcenowy, can you do the same for uboot?
<MoeIcenowy>
jernej: yes it has some problem... it failed to drive my father's 1920x1200 DVI monitor
<MoeIcenowy>
pgreco: for u-boot I cannot do
<MoeIcenowy>
BTW you can always just use ultra image on berry
<pgreco>
and kernel will choose the right dtb?
<pgreco>
or will it try to use the ultra one?
<MoeIcenowy>
ultra dtb also works on berry
<MoeIcenowy>
the extra LED on ultra is NC on berry
<MoeIcenowy>
usb1 is enabled both on ultra and berry, and usb2 is NC on berry
<MoeIcenowy>
mmc2 is NC on berry
<MoeIcenowy>
both usb and mmc is auto probable
<pgreco>
thanks
<anarsoul|2>
mripard: are you around?
BenG83 has quit [Ping timeout: 256 seconds]
Putti has quit [Remote host closed the connection]
<MoeIcenowy>
KotCzarny: I have a Sunvell R69 with the former owner wrongly written a SB key to it
<MoeIcenowy>
what should I do on the box?
<anarsoul|2>
jernej: or you?
<KotCzarny>
sb key?
<MoeIcenowy>
KotCzarny: secure boot key
<MoeIcenowy>
you know H3 supports it
<anarsoul|2>
MoeIcenowy: you're definitely around :)
<MoeIcenowy>
and he flashed some Chinese AliOS
<KotCzarny>
ahm, i think you know better in such case
<MoeIcenowy>
yes
<MoeIcenowy>
now it can only boot via FEL
<anarsoul|2>
MoeIcenowy: how familiar are you with sun4i drm driver?
<KotCzarny>
there was a way to still boot if it wasnt entirely locked write
<MoeIcenowy>
KotCzarny: it is
<MoeIcenowy>
anarsoul: so unfamiliar
<MoeIcenowy>
but maybe you can ask
<MoeIcenowy>
I think the log collector of mripard and jernej are all here
<MoeIcenowy>
if their persons are not
<anarsoul|2>
MoeIcenowy: mripard: jernej: so I need to plug DSI vblank interrupt somehow into tcon driver
<KotCzarny>
then unless you manage to crack the key then write your own proper uboot, there is nothing to do i'm afraid
<anarsoul|2>
I'm going to split tcon irq handler into common part and into tcon-specific part
<MoeIcenowy>
KotCzarny: yes so it entered a "鸡肋" state
<anarsoul|2>
and export global part
<MoeIcenowy>
anarsoul: maybe we should refactor vblank code
<MoeIcenowy>
not tcon int code
<anarsoul|2>
is it a right way?
<anarsoul|2>
MoeIcenowy: as far as I understand it's all in tcon irq handler
<KotCzarny>
that chinese chars translate to 'chicken rib'
<KotCzarny>
*those
<MoeIcenowy>
KotCzarny: it's a phrase
<MoeIcenowy>
which means "not so useful, but not willing to drop"
<MoeIcenowy>
or the original text "食之无味弃之可惜"
<KotCzarny>
you can try rooting the droid, then playing around then
<MoeIcenowy>
the droid is dropped...
<MoeIcenowy>
and now the only boot method is FEL sideload
<KotCzarny>
does it still boot that way?
<KotCzarny>
or brom refuses to load unsigned loader?
<MoeIcenowy>
yes, I can sideload a SPL and then U-Boot etc via FEL
<MoeIcenowy>
FEL doesn't load any loader
<MoeIcenowy>
it only load words
<KotCzarny>
nice to know that
<MoeIcenowy>
anarsoul: yes
<MoeIcenowy>
you should split it to two functions
<KotCzarny>
it means one can still work on such box
<MoeIcenowy>
one dealing with vblank
<anarsoul|2>
MoeIcenowy: that's what I was going to do
<MoeIcenowy>
one dealing with TCON irq
<MoeIcenowy>
KotCzarny: I used to use it to test the behavior of secure H3
<KotCzarny>
btw. those previous chinese chars translate to: "It's a pity to eat without taste"
<MoeIcenowy>
but now I have a burned BPi M2 Zero early sample
<anarsoul|2>
just need to figure out how to get pointer to tcon in dsi driver :)
<KotCzarny>
also, it means one can probably buy those alios victims cheaply
<MoeIcenowy>
KotCzarny: to be honest I "bought" the victim for free
<KotCzarny>
:)
<MoeIcenowy>
only shipment fee for <$2
<KotCzarny>
nice
<MoeIcenowy>
I think I need some sunxi-feld to boot locked devices up
<MoeIcenowy>
;-)
<anarsoul|2>
btw, DE and HDMI patches for A64 haven't landed yet, have they?
<KotCzarny>
yeah, and some cheap usb capable device in a convenient stick format
<MoeIcenowy>
KotCzarny: maybe I should try use CH554?
<MoeIcenowy>
(a cheap 8051-based USB MCU
<paulk-leonov>
anarsoul|2, on h3 DSI uses 8080 CPU mode for vblank interrupt IIRC
<paulk-leonov>
on A33*
<paulk-leonov>
sorry
<MoeIcenowy>
paulk: so from TCON view they're diferent?
<anarsoul|2>
paulk-leonov: it's enabled, but it's not signalled here
<anarsoul|2>
paulk-leonov: and BSP driver uses DSI interrupt for vblank on A64
<paulk-leonov>
anarsoul|2, heh I have the same problem with a DSI panel on A33, I'm not getting the vblank interrupt
<KotCzarny>
MoeIcenowy: but honestly, other than experimentation device, r69 isnt particularly interesting (needs a fan to work reliably, and it's h2+)
<paulk-leonov>
so it just hangs until it timesout
<MoeIcenowy>
it comes with a HS
<anarsoul|2>
paulk-leonov: how did you fix it?
<MoeIcenowy>
KotCzarny: so maybe I can drop it
<KotCzarny>
HS ?
<paulk-leonov>
anarsoul|2, I didn't, it's still broken at the moment
<MoeIcenowy>
or gift it to another one?
<MoeIcenowy>
heatsink
<KotCzarny>
ahm, still, needs fan
<anarsoul|2>
paulk-leonov: I see :) so I guess you can try my patch when it's ready
<KotCzarny>
heatsink alone isnt enough
<paulk-leonov>
anarsoul|2, for sure!
<KotCzarny>
keep it for SB experiments or send to someone who is interested in hacking secureboot
<MoeIcenowy>
KotCzarny: I have another SB device
<MoeIcenowy>
(without key
<MoeIcenowy>
and as a SBC it can have more usage (e.g. attach a JTAG to dump S-BROM
<anarsoul|2>
paulk-leonov: btw, do you have patches for DSI on A33?
<MoeIcenowy>
paulk-leonov: this problem happen only on a panel or on all panels?
<paulk-leonov>
anarsoul|2, not really, I haven't modified the driver much, just added lots of debug
<anarsoul|2>
paulk-leonov: I see
<paulk-leonov>
I think timings calculation might be wrong in a few places
<paulk-leonov>
MoeIcenowy, mripard has a panel that works for sure
<paulk-leonov>
but apparently it's the only one known to work
<paulk-leonov>
with the mainline driver
<MoeIcenowy>
ahhhhhhhhhhhhh
<anarsoul|2>
paulk-leonov: well, timings are wrong for sure for me
<anarsoul|2>
but I know where to fix it
<paulk-leonov>
since it's an android tablet I'm working on, I can just dump register from Android, where it works
<anarsoul|2>
I have one more pll-mipi fix not pushed
xyntrix has joined #linux-sunxi
<anarsoul|2>
looks like it doesn't work when it's freq < 300 MHz
<anarsoul|2>
it does work on 330MHz, but doesn't on ~290MHz (don't remember exact value)
<anarsoul|2>
I had to use my pinebook to figure that out
<anarsoul|2>
i.e. I removed all parent but pll-mipi for tcon0
<anarsoul|2>
disabled video init in u-boot to make sure that everything works from scratch in linux
<anarsoul|2>
and started trying to get it working
<paulk-leonov>
I see
kaspter has quit [Read error: Connection reset by peer]
kaspter has joined #linux-sunxi
Putti has joined #linux-sunxi
formruga has quit [Quit: Konversation terminated!]
<jernej>
anarsoul|2: I suggest you take a look at BSP for minimal acceptable clock rate since datasheets tend to be unreliable about that (at least for H3)
<anarsoul|2>
jernej: I couldn't find minimal rate for it in BSP driver
<jernej>
let me check
<anarsoul|2>
anyway, it's pretty confusing - reg says that PLL is locked, but display shows nothing
<jernej>
which pll? mipi?
<anarsoul|2>
yes
formruga has joined #linux-sunxi
<jernej>
ah, there is only maximum, 1440 MHz
<jernej>
but I guess you already check BSP algorithm for calculating PLL MIPI factors
JohnDoe_71Rus has quit [Quit: KVIrc KVIrc Aria 5.0.0, revision: git-7326-85ef2059c, build type: debug, sources date: 20160102, built on: 2018-06-01 10:18:08 UTC git-7326-85ef2059c http://www.kvirc.net/]
SP7RT has quit [Ping timeout: 256 seconds]
msev- has joined #linux-sunxi
SP7RT has joined #linux-sunxi
dc13ff has joined #linux-sunxi
<willmore>
anarsoul|2, Okay, looking at the datasheet some more, what are you needing the clock to be <300MHz for? Driving a display at a lower resolution? Then I assume you're using that to clock the TCONx. TCON0 at least, seems to have some further divisors in there.
<willmore>
Table 6.2.5.1?
<willmore>
No, sorry, 6.2.5.11
<willmore>
6.2.5.9 seems to have some division as well. I don't understand the clocking for all of that, so I can't really dig in much more.
elros has quit [Remote host closed the connection]
xerpi has joined #linux-sunxi
montjoie has quit [Quit: leaving]
<anarsoul|2>
willmore: that's what driver tcon driver chooses
<anarsoul|2>
willmore: basically it selects ~290MHz and tcon_div = 4 to get 72.5MHz for dot clock
anarsoul|3 has joined #linux-sunxi
anarsoul|2 has quit [Ping timeout: 256 seconds]
aalm has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
f0xx has quit [Ping timeout: 256 seconds]
<miasma>
hm, interesting decision :-/ https://github.com/Fourdee/DietPi/issues/1550 "We will not support a device which lacks a good standard build quality, and, basic level of software support from the manufacture."
<miasma>
apparently they require official vendor support, relatively good mainline support isn't enough
<anarsoul|3>
:)
reinforce has quit [Quit: Leaving.]
return0e has quit [Read error: Connection reset by peer]
return0e has joined #linux-sunxi
Putti has quit [Ping timeout: 256 seconds]
afaerber has quit [Quit: Leaving]
dc13ff has quit [Quit: Connection closed for inactivity]
afaerber has joined #linux-sunxi
<willmore>
miasma, guess that rules out Rpi
<willmore>
anarsoul|3, any reason not to crank it up to 362.5 and a tcon_div of 5?
<anarsoul|3>
willmore: nope, but driver doesn't do it
<willmore>
The driver doesn't have a concept of a minimum input clock speed? Or are you saying 4 is the max divider?
<anarsoul|3>
willmore: are you talking about clock driver or tcon driver?
<willmore>
tcon
<anarsoul|3>
I can set minimum clock for pll-mipi, and that's what I'm going to do
<willmore>
Okay.
<willmore>
Then the tcon will pick a sane value over that?
<anarsoul|3>
tcon just tries all possible combinations and chooses one that has smallest error
<willmore>
Ahh, okay. Then it just needs one more sanity check on its choices.
lurchi__ is now known as lurchi_
SP7RT has quit [Ping timeout: 256 seconds]
<miasma>
willmore: dunno. i just noticed that dietpi didn't provide the images anymore. i think we used them in some project
<willmore>
I assume dietpi is some stripped down distro? I'm not familiar with it.
<miasma>
something like that
<willmore>
Do they actually build for a modern processor? ;)
<miasma>
i'm not that familiar with that. i think the main reason i heard about was that it had a pretty large selection of board support some time ago
<miasma>
but now they've decided that orange pi boards are not worth supporting ¯\_(ツ)_/¯