<lekernel>
tuxbrain_away, opencore's ethmac is bloated
<lekernel>
milkymist's minimac should fit without problem on the sakc
<wolfspraul>
lekernel: hey I was wondering...
<wolfspraul>
you mentioned several times someone should try to reverse-enginee the xilinx binary format
<wolfspraul>
I finally warm up to this idea, see the need, etc. How hard do you think this really is?
<lekernel>
not very hard
<lekernel>
there is already a methodology, some tools and info available at ulogic.org
<lekernel>
but all of this is very dirty and the guys didn't properly finish the work
<lekernel>
the hard part, however, is to write a synthesis frontend - despite the fact all docs are available for this
<tuxbrain>
lekernel: mmm  good , maybe in parallel of SACK production we have to start a FPGA library, like opencores but more open, only need to register to upload not to download, an also offering compiled bit files ready to upload to the FPGA at least on SACK
<lekernel>
more quality as well...
<lekernel>
99% of opencores sucks technically
<tuxbrain>
I have no such criteria  to know what is good and what is bad, then this migh require levels of quality to let the newbies like me also upload some crappy developments meanwhile is tagged as such.
<lekernel>
opencores also has such a tagging system... the problem is it's self-assessed, and people tend to overestimate their design skills
<lekernel>
I mean, it's *really* a problem
<lekernel>
most opencores design aren't just bad, they're way beyond that
<tuxbrain>
I'm thinking more on expert group of moderators
<tuxbrain>
but will be hard to find to do it for free, usually those kind of experts had work enough for trhee lives
<lekernel>
yeah... maybe just a distributed review system might be fit
<lekernel>
you can attribute a note to other people's code
<lekernel>
and your note weights more if yourself has got good reviews
<tuxbrain>
not bad idea indeed
<lekernel>
seriously something has to be done about quality
<lekernel>
otherwise the whole open source logic design thing is just going to be a bunch of crap
<tuxbrain>
I agree and more if this finally is intended to do ASIC at the end
<tuxbrain>
wolfspraul: maybe a interfero as backend to make the cores colaborative friendly with a frontend  with enchanced navigation ant this evaluation system? (not for be done tomorrow but to have in mind)
<tuxbrain>
regarding mini-mac, that means than uploading this module taking care of the I/O pins I can plug a rj45 and voilá
<lekernel>
er, no
<lekernel>
well for the connection to the RJ45 yes
<lekernel>
but at the other end you have a DMA master and a CSR control interface
<lekernel>
that need to be connected into a SoC
<tuxbrain>
hehehe I expect that anwswer
<lekernel>
btw it's not connected directly to the RJ45, but through a MII PHY
<tuxbrain>
and this MII PHY can be also implemented in FPGA?
<lekernel>
no, it contains analogue electronics
<tuxbrain>
ok
<tuxbrain>
so RJ45->MII PHY->(on FPGA)mini-mac->(on FPGA)Interface with CPU->(kernel)device driver->voilá?
<lekernel>
yes
<tuxbrain>
then is "just" matter of find a MII PHY thing , develop a interface with CPU, I hope we can use some already done device driver and "that's all" to have Ethernet on SACK.... damn
<lekernel>
there isn't any already done device driver
<tuxbrain>
double damn
<lekernel>
btw minimac (as well as opencore's ethmac) is meant for using in a SoC
<lekernel>
so they expect to DMA directly into the CPU memory, simply as another bus master
<lekernel>
this might not be possible with the xburst
<lekernel>
a workaround, however, is to put a SRAM in the FPGA
<tuxbrain>
and now give me the good news
<lekernel>
make the minimac DMA into it
<lekernel>
map it into the xburst memory space
<lekernel>
and then make the drivers put all the ethernet buffers into that SRAM
<lekernel>
the good news? well
<lekernel>
milkymist can do TFTP netboot, respond to ping, and run a webserver using minimac
<lekernel>
today
<lekernel>
so the core pretty much works
<tuxbrain>
let's define the good news as is theoretically posible to do so :)
<rafa>
larsc: tuxbrain : does somebody have thought to do some boot loader? I was thinking in this idea :
<rafa>
- kernel on nand has set kexec
<rafa>
- kernel boots and starts a minimal ramfs which loads and boots another kernel in rootfs nand.
<rafa>
This idea could be useful to (perhaps) to update the rootfs distribution without the usb mode thing which is a little ugly to do.
<rafa>
well, at least I think that we will try this idea in Jlime (if kristoffer agrees with me to try it ;) )
<rafa>
(i am talking about ben nn BTW :) )
<larsc>
rafa: whats wrong with uboot?
<rafa>
larsc: nothing is wrong I think. I am thinking some way to avoid Boot your NanoNote into USB BOOT mode to update ben nanonote software
<larsc>
i would like to use dfu-util.
<larsc>
mirko wants to use a simple webinterface though
<larsc>
but yeah, kexec would be another option
<rafa>
larsc: okey, but I would try a pure software way with the current nn configuration/software. So we could update the rootfs just booting nn and before to kexec the current rootfs distro.
<rafa>
larsc: so... you like the idea ? (I would like your opinion to know if it is a waste of time :) )
<larsc>
still undecided. but i think it's worth looking at it.
<xakh>
So, I have a friend that wrote a game originally for J2ME phones, meant to run at SVGA resolution
<xakh>
well, originally for QVGA
<xakh>
but can be patched
<xakh>
so, if he could get it ported, and running nicely, you think we could have it included in the main image, as one of the programs to showcase the nano?
<wolfspraul>
xakh: you are often in irc when not many people are around :-)
<wolfspraul>
are you in the US?
<xakh>
yup.
<xakh>
but I try to stay up late to get in touch with you guys
<wolfspraul>
west coast?
<xakh>
for instance it's 2230 here
<xakh>
no, I'm in West Virginia
<wolfspraul>
I'm in China, so you often catch me, but even though I would like to do more hacking, in the end all my time is wasted running aroud meeting people, vendors, etc.
<xakh>
haha
<wolfspraul>
yeah
<wolfspraul>
horrible
<xakh>
ouch
<wolfspraul>
I can barely catch up building stuff for my Nano.
<wolfspraul>
too many things going on too fast
<xakh>
well, I have a tiny hack right now
<wolfspraul>
still haven't even tried Debian! argh...
<wolfspraul>
you have something you would like to feedback/commit to the openwrt builds?
<xakh>
I can't find the adduser system
<xakh>
Well, I have a friend who made games designed for J2ME phones
<xakh>
and he's working to port a free software shooter
<xakh>
it uses code to actively turn SVG files into meshes, making it able to run on 16mb of RAM
<xakh>
but anyway, right now I'm trying to hack in a new user
<wolfspraul>
understood, but don't know the answer
<xakh>
dang.
<xakh>
Well, you think we could include a compact adduser in the next release?
<FrankBlues>
waves
<xakh>
hallo
<wolfspraul>
xakh: sure but I don't know enough about openwrt
<xakh>
dang
<FrankBlues>
Hey! How goes?
<wolfspraul>
it's an openwrt question
<wolfspraul>
FrankBlues: goes well, thanks. how about you?
<FrankBlues>
VICE keeps complaining my hardware isn't fast enough...
<xakh>
pretty good myself
<FrankBlues>
But other than that I'm doing well, thanks!
<xakh>
hm
<xakh>
I'm having an annoying problem.
<FrankBlues>
Oh?
<xakh>
yeah
<xakh>
tar -x seems to just hang
<xakh>
I'm trying to compile something
<xakh>
hey
<xakh>
does this thing not have gcc?
<xakh>
wolfspraul: do you know if the NanoNote has gcc?
<xakh>
because this file says I don't have it
<wolfspraul>
xakh: I'm pretty sure I heard someone saying they have gcc running on the NanoNote, but forgot the details
<wolfspraul>
let me check whether it's in the packages we are building...