#timvideos

Title: Pastebin | IRCCloud (at www.irccloud.com)

i turn camera on and off

input1: 1280x720

Title: HDMI2USB-vmodvga-misoc/ad9984a.c at c6b961489b114526e65de70c44ccadfe9638b92e · rohit91/HDMI2USB-vmodvga-misoc · GitHub (at github.com)

oh also

i keep having to do 'output0 off'

else the board freaks out

you'll note my script does it after every input change

define "freaks out"

the HDMI out feeding a small monitor i have starts flicking on and off

er, output1 off rather

it's coming out of output0

this was the issue where having all inputs+outputs+encoder caused it to run out of DDR bandwidth

http://wiki.minnowboard.org/MinnowBoard_MAX < ctr+F i2c , got it

Title: MinnowBoard MAX - MinnowBoard Wiki (at wiki.minnowboard.org)

mine runs Debian Jessie, so those i2c-tools should run

it's actually my mail server so I must take care of not frying it

i'm not clear on what happens but if i do 'output1 off' initially, and then video_matrix toggle encoder/output0, output1 somehow gets re-enabled

which is again why my script just does 'output1 off' after every switch

seems to work around it

mithro: and no your script doesn't work for me, will figure out why

xfxf: if you are running out of DDR bandwidth you will be getting FIFO errors reported on the USB output

would i see that in dmesg on the host, or in the console of hdmi2usb?

either way, unless i do 'output1 off' after every switch then the HDMI output to my little confidence monitor starting flicking on and off

serial console HDMI2USB

which i assumed was a bug

right, i'm not getting FIFO errors there

wait, perhaps if i tried actually capturing at the same time i would, i can try

this is with the encoder on but not actually doing any capture

input0+input1+encoder+output0 on

xfxf: I need a clearer explanation of exactly what is happening and exactly your set up

yeah, sure, let me reproduce

ah, wait, i think i might have found the issue

does changing video_mode enable both outputs?

xfxf: Not sure, you can look at the code you know

ah, that fixed it

xfxf: that is what I do

i was doing 'output1 off' before setting the video_mode

the latter appears to turn it back on again

xfxf: you should only be doing video_mode once on start up

all good, not an issue

yeah i am, the INIT= part of my script

i just changed the order

would it be good behaviour to have hdmi2usb return the status on every command?

or at least output from commands fed to it?

it'd stop somebody else from falling into this same trap

xfxf: my changes added a status command

(after each change)

yeah i notice that

i mean at the firmware level

so every time you type a command you get something back from the device that tells you what happened

xfxf: This is all C code, so please feel free to hack on it

Title: HDMI2USB-misoc-firmware/ci.c at master · timvideos/HDMI2USB-misoc-firmware · GitHub (at github.com)

xfxf: you just need to do a "make lm32-firmware; make load-lm32" - don't even need to rebuild the gateware

sure, i don't mind attempting to hack on it, just verifying if that is a sane design decision

I can't guarantee that :P

I do think ever command having some type of success / error code result seems pretty reasonable

but I'm going to go home now

i was just typing the success/error code thing

nod

seems more sane, means i can parse that in my switchy thingy and handle errors properly

rather than having it do things blind

each valid command should already return a ack message no?

if you want to print the status after a command, just send the status command

xfxf: you might want to take a look at the other lm32 firmware issues on GitHub too

xfxf: I got most of the way through the custom EDID adding patch - would be helpful to finish that

it is rather wet outside

ysionneau: is wiki.minnowboard.org down?

hmmm works from here

oh, there DNSSec is broken

s/there/their/

ysionneau meet rohitksingh

ysionneau: hi! I follow you on twitter! :)

hi rohitksingh :)

nice to meet you!

I'm going to play a bit with i2c

rohitksingh: what's your twitter handle?

ysionneau: awesome!

its RohitK_Singh

and now I follow you as well :)

haha... :)

I'm going to use the i2c.c from lm32 fw, but to make a i2c slave acting like an EEPROM

so I'm going to need a master to test my code

I'm thinking of using my Minnow board Max which has i2c pins and runs Debian Jessie

ysionneau: ohkay... it should work, I haven't tried using it as slave yet...But, I can try that

any problem you are encountering?

Minnowboard Max side must be working well and good, i guess?

so far I don't have much to test

but I'll keep you posted if I have any issue or news

12:48 < rohitksingh> Minnowboard Max side must be working well and good, i guess? < it's a pretty good machine, I'm using it as my email server

ysionneau: Awesome! Pls do inform me in case of any issues!

very kind of you, thx!

what's your timezone?

I'm UTC+1

ysionneau: I had looked about Minnowboard Max, ~6 months back...neat little board....

yep

mine is GMT+5:30

very little power consumption, no fan (so : no noise), and ... x86! so all your preferred OS are working

I must try it! Just that it is little inaccessible in India at affordable rate (w/o huge import duties)

arg

I think I had lots of shipping cost as well :/

I'm generally available online everyday from now to 7 hrs from now (UTC 12:00PM to UTC 7:00PM)

yeah...shipping also is around 30 USD :/

ysionneau: are you sure your i2c slave will be fast enough? You have no restriction using bitbanging when you are the master, but when you are the slave you are not controlling the clock, so you will probably need to implement it in hardware (reuse code from edid)

yeah...thats a valid point...

if hardware implementation is possible/allowed, for slave it would be better to go for it

ok, first short between gnd+vcc , board reboot

I hope I won't kill my email server

_florent_: ah good point

that the reason why edid is done in hardware

yep I saw your implementation

i2c clock rate is either 100kHz or 400kHz

that's not mine, sb0 did it

ah ok

mithro: so that's probably too fast for a cpu that is doing others things

mithro: 10kHz also is an allowed speed IIRC

_florent_: The lm32 runs at ~50MHz, right?

yes

means 100 times the speed

but bitbanging is really slow

but maybe 100 instructions per cycle are enough

How many cycles per instruction?

(per i2c cycle)

mitro: sorry, 10KHz not defined for I2C..only allowed in SMBus...I take my words back :p

*mithro

If you have 100 instructions per clock transition seems pretty dooable

yep

_florent_: for now I think we want to be able to act as a slave while doing nothing else

so full time polling as a slave

that's for the FX2 configuration?

yes

what's the size of the emulated EEPROM?

you want to put everything in blockram :p

mithro: what's the size of the fx2 fw?

16k

I'm just going to remove the big buffer of 1280*8 lines for the encoder, so yes there will be block rams availables :)

I think kbytes but it could be kbit

that can fit in block rams

removing this buffer reduce blockram usage from 88% to less than 50%

_florent_: \o/

but blockram issue is that you need to re synthesize the bitstream to change the fx2 fw

why?

or you give access to the blockram via wishbone/etherbone from the host

yes, that's what is done for edid

having the fw in spi flash is better I think

so that when you change it, and you reboot, it stays