lekernel changed the topic of #milkymist to: Mixxeo, Migen, Milkymist-ng & other Milkymist projects :: Logs: http://en.qi-hardware.com/mmlogs
wpwrak has quit [Read error: Connection reset by peer]
wpwrak has joined #milkymist
bhamilton has joined #milkymist
<barmstrong>
holy cow lm32 has a lot of ifdefs :)
<ysionneau>
yep =)
bhamilton has left #milkymist [#milkymist]
sh4rm4 has quit [Ping timeout: 264 seconds]
sh4rm4 has joined #milkymist
lekernel has joined #milkymist
<lekernel>
finally, tested HDMI + video out
<lekernel>
fully works on one board, the other one has only the red channel working on the video out
<lekernel>
probably some broken solders
<lekernel>
only tested HDMI port 0 so far
<davidc__>
lekernel: I've been working on modding the verilog codegen to use blocking assignments when generating always @(*) blocks
<davidc__>
lekernel: would there be any interest in getting that upstream?
<davidc__>
lekernel: (as a side effect, it prohibits cyclic dependencies; even if they do settle)
<lekernel>
have you tested substantial code (eg. milkymist-ng) with your code?
<davidc__>
lekernel: I haven't tested milkymist-ng; the code I'm testing it on is a 2k LOC project at the moment. I of course would before I sent in patches (its not ready yet anyhow)
antgreen has joined #milkymist
<davidc__>
lekernel: (I can't really test if it results in correct syntehsis though, I don't have a milkymist board to run it on)
<larsc>
wondering if it was the guy who was asking for ulpi
<mumptai>
wasn't that like some weeks ago?
<lekernel>
I really wish this thing would work, USB is such a pain as you know
* mumptai
nods
<mumptai>
how did the soft-usb thingy work out?
<lekernel>
it needs more software to handle correctly the industrial quantities of pesky details in every USB protocol
<lekernel>
and no, linux won't easily solve the problem. implementing e.g. OHCI is a lot of pain.
<wpwrak>
industriously pesky ?
<larsc>
that's why they invented uhci
<larsc>
or maybe the other way around
<wpwrak>
that's why uhci, little brother of cthulhu, invented them ?
<davidc__>
USB is the biggest goddamn clusterfuck of a standard to ever be 'smeared' (and yes, smeared is the only appropriate word) out into the world
<wpwrak>
now shush and go back to write your HIDeous driver :)
<lekernel>
I suppose driving two BUFPLLs with a single PLL will fail?
aeris has quit [Ping timeout: 246 seconds]
aeris- has joined #milkymist
lekernel has quit [Quit: Leaving]
lekernel has joined #milkymist
<lekernel>
the nvidia driver is a lot of fun
<lekernel>
when starting the X server, it ignores the HPD pin and determines the port connection status by directly attempting EDID read
<lekernel>
HPD is only used *after* the X server has been started
<lekernel>
also, after a HDMI port has been shown as connected in xrandr, it keeps being reported as connected even when it's not, and until the X server is restarted (and beware of EDID used instead of HPD!)
<lekernel>
conclusion: seems the HPD pin control actually works correctly on the Mixxeo, only the nvidia driver has been playing tricks
<GitHub177>
[milkymist-ng] sbourdeauducq pushed 2 new commits to master: http://git.io/j1qbxw
<GitHub177>
milkymist-ng/master 2459d9f Sebastien Bourdeauducq: top: DVI samplers and 2-channel FB on Mixxeo only
<GitHub177>
milkymist-ng/master 8bc1620 Sebastien Bourdeauducq: dvisampler: disable EDID when HPD_EN is low