<ElBarto>
does mainline support HS200 for a64-emmc ?
* ElBarto
is trying to add support to it for FreeBSD but the eMMC doesn't seems to like this mode ...
BenG83 has joined #linux-sunxi
Putti has joined #linux-sunxi
razorfishsl has joined #linux-sunxi
clemens3 has joined #linux-sunxi
BenG83 has quit [Ping timeout: 260 seconds]
<mripard>
ElBarto: yes
<ElBarto>
ok that's just me doing something stupid then :)
<ElBarto>
is there any trick to do to the controller ?
bonbons has joined #linux-sunxi
<jernej>
can someone explain why some encoders (like RGB -> HDMI) are implemented as DRM bridges and others as V4L2 encoders?
<jernej>
There are even 2 adv7511 drivers, one as DRM bridge and another as a V4L2 device
fkluknav has quit [Ping timeout: 248 seconds]
f0xx has joined #linux-sunxi
razorfishsl has quit []
f0xx has quit [Read error: Connection reset by peer]
foxx_ has joined #linux-sunxi
hitech95 has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
SP7RT has joined #linux-sunxi
elros has joined #linux-sunxi
dddddd has joined #linux-sunxi
<mripard>
ElBarto: yeah, but then, I'd have to show you some GPL code :)
<mripard>
jernej: that's pretty weird to me too
<mripard>
jernej: the bottom line iirc is that there's some devices that just allow to modify a stream that is outside of Linux control (like just add an overlay on top of an existing video stream)
<mripard>
so v4l2 gained support for these kind of devices
<mripard>
and then, I guess some devices were added to v4l2 to various ranges of output
<mripard>
and since you needed to convert that output signal on some boards, some bridges driver were added as well
<mripard>
and I guess now that the framework are so far appart, noone really want to work on unifying some part of them
adj__ has quit [Remote host closed the connection]
adj__ has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
LargePrime has quit [Ping timeout: 260 seconds]
<wens>
jernej: because it can feed an hdmi decoder? :)
<jernej>
I always thought that DRM is meant for video output and V4L2 as video input
<jernej>
but it seems like V4L2 also covers output, which is a bit confusing...
<MoeIcenowy>
I think some on-chip devices are even implemted for two-way
<MoeIcenowy>
jernej: it's legal to output sth via v4l2, for example, if your board wants to behave like a UVC camera
<MoeIcenowy>
(although the direct v4l2 output intf is rarely used, and few document can be found
<jernej>
wens: another topic: how to represent TVE/TCON TV clock mux in TCON-TOP device tree node?
<MoeIcenowy>
jernej: Personally I may choose to implement a CCU
<MoeIcenowy>
(but maybe I overused it
<jernej>
I highly doubt TVE is clock source
<jernej>
since it has it's own entry in CCU
LargePrime has joined #linux-sunxi
<jernej>
and support exactly the same parents and dividers as TV TCON
<jernej>
so I'm not sure what is the point having that mux
<jernej>
but BSP always chose TVE clock source when TV encoder is used...
<MoeIcenowy>
jernej: I doubt it from the first day.
<jernej>
what do you doubt?
<MoeIcenowy>
"TVE is clock source"
<jernej>
exactly
<MoeIcenowy>
remember R40 manual
<MoeIcenowy>
there's a divider in TVE
<MoeIcenowy>
(that's you told me
<MoeIcenowy>
so I think on H3 it divides the clock and feed it to TCON1
<MoeIcenowy>
(on R40 it feed the clock to TCON-TOP mux
<jernej>
MoeIcenowy: AC200 is very interesting document to study regarding TVE
<MoeIcenowy>
yes
<jernej>
there you can see that it needs 216 MHz clock
<MoeIcenowy>
they might be the same IP core
<jernej>
it is, more or less
<MoeIcenowy>
but in my memory Allwinner chips are fabbed at TSMC
<MoeIcenowy>
not SMIC
<MoeIcenowy>
(maybe memory corruption ;-)
<MoeIcenowy>
(or maybe they changed the factory, to reduse cost
<wens>
jernej: I was thinking of port groups, but don't if that would screw up dt graph parsing or not
<wens>
* don't know *
<jernej>
wens: I meant what to do with clock selection in TCON-TOP
<jernej>
TCON-TOP register 0 (R40 manual, p. 515, chapter 7.2.5.1)
<jernej>
BSP code is prety simple, if TVE is enabled, switch all muxes to TVE
<jernej>
but that would mean that graph needs to be traversed and if TVE is found, mux is switched to TVE clock
<jernej>
but that is future job
lurchi__ is now known as lurchi_
elros has quit [Remote host closed the connection]
<wens>
I suggest just not let it switch parents, and force it to use CCU for now
<wens>
but leave a note?
<jernej>
yep, that's what I was thinking
<jernej>
MoeIcenowy: The reason why TVE and TCON need 216 MHz rate is because they use 8 bit parallel interface, which needs 4 cycles for one transfer
<MoeIcenowy>
jernej: ok
hanni76 has joined #linux-sunxi
<wens>
216 is a multiple of what dotclock?
<MoeIcenowy>
I think it's 43.2
<jernej>
27 MHz
<MoeIcenowy>
oh memory corruption again
<jernej>
anyway, TVE needs for PAL and NTSC resolution 216 MHz clock
<jernej>
this is 8 times more
<jernej>
this TVE thing is very confusing
foxx_ has quit [Ping timeout: 264 seconds]
LargePrime has quit [Remote host closed the connection]
<wens>
8 bits per dot? so it has to queue up 8 bits to convert to analog signal
lurchi_ is now known as lurchi__
imcsk8 has quit [Read error: Connection reset by peer]
imcsk8 has joined #linux-sunxi
reinforce has joined #linux-sunxi
<hitech95>
got my simple aplifier driver to manage speaker AMP gpio WORKING WITH SUNXI CODEC. I have added some more DAPM on sunxi81-codec and seems ok. Still lot's of rework is needed to support TDM and multiple AIFs and the clock
<KotCzarny>
:)
<hitech95>
it toggle the supply on and off (if not set as dummy regulator) and also the gpio pin.
<KotCzarny>
try making your driver easy to port to other socs too
<hitech95>
KotCzarny, it is a genreic one, it should work with all. I tested as an auz for the simple card of the sunxi codec
<jernej>
wens: it's actually 10 bit per dot, check CCIR mode on LCD TCON if you're interested
<wens>
:/
<wens>
hitech95: if you have some stuff already working, you can send it
<wens>
hitech95: as for generic amp driver, I think I asked Mark about this, and IIRC there currently isn't one, but he's not against it either
<hitech95>
wens, Yup is working but all the TDM stuff and others AIF are not. I'm also tring to test the capture (wich looks like it is still not working)
<wens>
and frankly I'd like one too
<wens>
hitech95: I suggest sending off any cleanup patches first
<wens>
hitech95: then possibly capture
<wens>
it's going to be hard to find people to test the other AIFs and TDM
<hitech95>
the generic amp is called "simple-amp" like the simple-card. the idea is to push it on mainline for all those devices that have an power amp on board.
<hitech95>
About tdm and AIFs i'm still tring to understand how the codec to codec is managed on Asoc.
<wens>
IIRC there's DPCM that does this
<wens>
I guess you could ask the maintainer about the details. I don't think ASoC is well documented
<jernej>
wens: I'm not sure anymore how many bits per dot is, 8 parallel lines suggest 8, but DAC in TVE is 10 bit
<MoeIcenowy>
wens: amp?
<MoeIcenowy>
amplifier?
<wens>
MoeIcenowy: yes. Some boards have external amplifiers or buffers
<wens>
like tablets
<MoeIcenowy>
ok I know it
<hitech95>
MoeIcenowy, I made a dummy driver to support a possible regulator and a gpio for mute.
<MoeIcenowy>
P.S. I think Allwinner tablets usually use a power amplifier added to the headphone route to enable speaker
IgorPec has quit [Ping timeout: 240 seconds]
<MoeIcenowy>
hitech95: yes I think it's really important for A33 to get external power amplifier support
<MoeIcenowy>
currently I usually enable it via /sys/class/gpio ;-)
<hitech95>
so dapm can turn the amplifier on and off only when needed. According to the bsp of a33 codec it support aplifier via gpio. it should be connected to the headphone output
<MoeIcenowy>
hitech95: this is what I said just before
<MoeIcenowy>
a speaker after the pa
<MoeIcenowy>
(and when a headphone jack is inserted, the pa is shutdown to prevent speaker from playing
rexxster has joined #linux-sunxi
<hitech95>
MoeIcenowy, Yup i just wantend to confirm that :)
<MoeIcenowy>
try to make it more generic
<MoeIcenowy>
as it's surely not Allwinner only
<hitech95>
MoeIcenowy, already done :)
<MoeIcenowy>
good ;-)
<hitech95>
but i teste donly on allwinner HW
<MoeIcenowy>
no need to test on other HW
<MoeIcenowy>
it's ok
<MoeIcenowy>
it's usual for not-so-professional developers to have no enough HW to test ;-)
GrimKriegor has quit [Read error: Connection reset by peer]
<hitech95>
Now I have to leard ho to sand patches to mainline :)
jbrown has joined #linux-sunxi
GrimKriegor has joined #linux-sunxi
aalm has quit [Ping timeout: 248 seconds]
<MoeIcenowy>
hitech95: this is far more rich than my first commit ;l-)
<MoeIcenowy>
;-)
<MoeIcenowy>
ls
<MoeIcenowy>
oops wrong window
<MoeIcenowy>
hitech95: I suggest you to use amplifier rather than "AMP" as many as possible
<MoeIcenowy>
and for the regulator name "vcc" is better than "VCC"
<hitech95>
LOL, yesterday I tiped my password bit I realized in time. (Multi monitor setup). About that name I know that it sould be lowercase but the DAPM ar all uppercase and unfortunatly DAPM is the one in charge to manage the supply. I checked onthers codec and those supplies are uppercase.
<ElBarto>
mripard: yeah fair enough :) anyway I'm not sure this is the driver since it fail to switch during CMD6 so clocks are still at 400Khz etc ...
leviathan has joined #linux-sunxi
f0xx has joined #linux-sunxi
cnxsoft has quit [Quit: cnxsoft]
nuuuciano has joined #linux-sunxi
<willmore>
buZz, you were talking about getting greyscale on a LED display. Instead of PWM per pixel which needs N updates for 1/N level control, you can use Log2(N) updates. You just do them with variable spacing. For the high bit you do T, but the next bit you do T/2 then T/4, etc. Until you have the brightness resolution you want.
<micken>
ElBarto: ok, but isn't bch for endian setting?
<micken>
ElBarto: there is that config thing
<micken>
hmm
rexxster has quit [Remote host closed the connection]
paulliu has quit [Quit: Leaving.]
<micken>
ok
<micken>
so it affects image booting
<micken>
with "go" it just jumps to the address and nothing else, so I think my hack is neccesary
<micken>
just passing the address to the transition function
sunshavi has quit [Ping timeout: 264 seconds]
IgorPec has joined #linux-sunxi
liqsquid has joined #linux-sunxi
dddddd has quit [Remote host closed the connection]
<MoeIcenowy>
micken: bch seems to be just netbsd building fix
<MoeIcenowy>
micken: try to pack the thing with mkimage?
<MoeIcenowy>
hitech95: you can send it as a RFC (request for comments) now
<MoeIcenowy>
it may raise a long discussion
<hitech95>
yup, but i still don't understand how mailing list works :D I used PR on github for OpenWrt/Lede or I sent directly the patches via mail to the right guys
<jernej>
hitech95: you have to split dt bindings changes and driver in two patches. First has to be dt binding patch.
<hitech95>
jernej, there is no dt bindings inthe patch of the amplifier. Only the documentation
<MoeIcenowy>
jernej: do you have any try on implementing H6 HDMI in U-Boot?
<hitech95>
I have a dummy question U guys use a IRC proxy to be always connected, right?
<jernej>
hitech95: Documentation/devicetree/bindings/sound/simple-amplifier.txt <- this qualifies as dt binding
<KotCzarny>
i just use console client running inside screen session
<hitech95>
So the docs must be separated?
<jernej>
yes
<KotCzarny>
but there are irc bouncers (proxies) you can install use
<jernej>
I use bouncer and I imagine many others
<hitech95>
right now 'm just using hexchat. But i have some always on machine that I could use for a proxy :) I just wanted to see if there are other solutions :D
<jernej>
MoeIcenowy: Of course, that was first thing I did on my PineH64
<jernej>
didn't you find it on my github?
<KotCzarny>
you can connect to irc bouncer from any irc client
<MoeIcenowy>
oops
<MoeIcenowy>
oooooooooooooooops
<jernej>
but it's hacky
<MoeIcenowy>
overlooked
<MoeIcenowy>
no problem as I forgot to send H6 support to U-Boot mainline
<MoeIcenowy>
;-)
<KotCzarny>
MoeIcenowy: laziness has it's positive sides sometimes ;)
<MoeIcenowy>
ah it has more reason
<jernej>
hitech95: two comments on your patch
sunshavi has joined #linux-sunxi
<jernej>
1. Use full name along email in copyright notice
<jernej>
2. use SPDX identifier instead full license text
<hitech95>
1) Opps it got removed at some point, thx for that :)
<hitech95>
Whay tou mean by SPDX?
<MoeIcenowy>
hitech95: it's a tag about the source file's open source license